Original Research

Mycosis fungoides (MF), the most common cutaneous T-cell lymphoma (CTCL), is characterized by clonal proliferation of predominantly CD4⁺ T cells with localization to the skin.¹ Mycosis fungoides typically affects older adults with a male to female ratio of 2:1 but also can occur in children and younger adults.^2,3 Known as the great imitator, the manifestations of MF can be variable with considerable clinical and pathologic overlap with benign inflammatory skin diseases, rendering definitive diagnosis challenging.^4-7 The early stages of classic MF manifest as pruritic erythematous patches and plaques with variable scaling that can progress in later stages to ulceration and tumors.⁸ Histopathologically, classic MF is characterized by epidermotropic proliferation of small- to intermediate-sized pleomorphic lymphocytes with cerebriform nuclei and a haloed appearance; intraepidermal nests of atypical lymphocytes known as Pautrier microabscesses occasionally are observed.⁵ Mycosis fungoides typically follows an indolent clinical course, with advanced-stage MF portending a poor prognosis.^9,10 Current treatment is focused on halting disease progression, with topical therapies, phototherapy, and radiation therapy as the standard therapies for early-stage MF.^11-13 For advanced-stage MF, treatment may include systemic therapies such as interferon alfa and oral retinoids along with chemotherapies for more refractive cases.¹⁴ Allogenic hematopoietic cell transplantation is the only curative treatment.¹¹

Current staging guidelines for MF do not address anatomic location as there is little known about its impact on patient outcomes.^11,15 Due to the indolent nature of MF leading to diagnostic challenges, the exact frequency of each primary disease site for MF also remains unclear, though the suggested incidence of MF of the head and neck ranges from 30% to 70%.^16,17 Involvement of the head and neck^16,18 or external ear and external auditory canal¹⁹ is associated with worse prognosis. The purpose of this study was to examine the impact of anatomic location of primary disease site on survival in MF.

Methods

The National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) database includes patient records from 18 registries and encompasses approximately 48% of the US population.²⁰ Using SEER*STAT software (version 8.4.0.1), we conducted a search of patients diagnosed with MF (International Classification of Diseases for Oncology, Third Edition [ICD-O-3] histologic code 9700/3 [mycosis fungoides]) between 2000 and 2019. For inclusion in the study, patients were required to have a known age, specified primary site, and a known cause of death (if applicable). Patients with known Sézary syndrome (SS)—an aggressive form of CTCL that is characterized by the presence of clonally related neoplastic T cells in the skin, lymph nodes, and peripheral blood—were not included because the World Health Organization/European Organisation for Research and Treatment of Cancer considers SS and MF to be separate entities^1,15; SS does not necessarily arise from preexisting MF and is associated with markedly poorer survival. This study was exempt from institutional review board approval because the data were publicly available and anonymized.

Data Collection—For age at diagnosis, patients were divided into the following categories: younger than 40 years, 40 to 59 years, 60 to 79 years, and 80 years and older. Demographics, tumor characteristics, and surgical management (if applicable) were obtained for each patient. The designations of chemotherapy and radiation treatment in the SEER database are not reliable and prone to false negatives. As such, these were excluded from analysis.

The primary outcomes of interest were overall survival (OS) and disease-specific survival (DSS), which were calculated as time from MF diagnosis to death. Although OS included all patients who died of any cause, DSS only included patients who died of MF.

Statistical Analysis—Demographics (age, sex, race, ethnicity), tumor characteristics (tumor size, primary site, T stage, lymph node involvement, metastasis), and surgical management (if applicable) were summarized. Overall survival and DSS were calculated using Kaplan-Meier analysis. Univariate and multivariable Cox proportional hazards regression models were generated to determine which prognostic factors for MF were associated with poorer OS and DSS. Only statistically significant variables in the univariate analysis were used to construct the multivariable analysis. Hazard ratios (HRs) and their associated 95% CIs were reported. Incidence rates were calculated and age adjusted to the 2000 US standard population. The SEER JoinPoint Regression program was used to determine the annual percent change (APC)—change in incidence rate over time. P<.05 was considered statistically significant. All statistical analyses were conducted with R version 4.0.2.

Patient Demographics and Tumor Characteristics—There were 4265 patients diagnosed with MF from 2000 to 2019. The overall incidence of MF was 2.55 per million (95% CI, 2.48-2.63) when age adjusted to the 2000 US standard population, which increased with time (mean APC, 0.97% per year; P=.01). The mean age at diagnosis was 56.4 years with a male to female ratio of 1.2:1. Males (3.07 per million; 95% CI, 2.94-3.20) had a higher incidence of MF than females (2.16 per million; 95% CI, 2.06-2.26), with incidence in females increasing over time (mean APC, 1.52% per year; P=.02) while incidence in males remained stable (mean APC, 1.09%; P=.37). Patients predominantly self-identified as White (73.08%). Patients with MF of the head and neck were more likely to have smaller tumors (P=.02), a more advanced T stage (P<.001), and lymph node involvement (P=.01) at the time of diagnosis. Additional demographics and tumor characteristics are summarized in eTable 1.

CT113004177_eTable1_part1.jpg

CT113004177_eTable1_part2.jpg

Survival Outcomes—The mean follow-up time was 86.9 months. The 5- and 10-year OS rates were 85.4% (95% CI, 84.2%-86.6%) and 75.0% (95% CI, 73.4%-76.7%), respectively (Figure 1)(Table). The 5- and 10-year DSS rates were 93.3% (95% CI, 92.4%-94.1%) and 89.5% (95% CI, 88.3%-90.6%), respectively. For OS, univariate analysis indicated that significant prognostic factors included increasing age (P<.001), female sex (P<.001), self-identifying as Asian or Pacific Islander (P<.001), self-identifying as Hispanic Latino (P<.001), primary tumor sites of either the head and neck or upper limb (P<.001), T3 or T4 staging (P=.001), lymph node involvement at the time of diagnosis (P<.001), and metastasis (P<.001).

CT113004177_Fig1.jpg

CT113004177_Table.jpg

For DSS, univariate analysis had similar risk factors with self-identifying as Black being an additional risk factor (P=.02), though self-identifying as Asian/Pacific Islander or Hispanic Latino were not significant nor was location on the lower limb. For recorded tumor size, the HR increased by 1.001 per each 1-mm increase in size (eTable 2).

CT113004177_eTable2_part1.jpg

CT113004177_eTable2_part2.jpg

Multivariate analysis showed age at diagnosis (60–79 years: HR, 23.11 [95% CI, 3.03-176.32]; P=.002; ≥80 years: HR, 92.41 [95% CI, 11.78-724.75]; P<.001), T3 staging (HR, 2.37 [95% CI, 1.32-4.27]; P=.004), and metastasis (HR, 40.14 [95% CI, 4.14-389.50]; P=.001) significantly influenced OS. For DSS, multivariate analysis indicated the significant prognostic factors were age at diagnosis (60–79 years: HR, 8.94 [95% CI, 1.16-69.23]; P=.04];≥80 years: HR, 26.71; [95% CI, 3.26-218.99]; P=.002), tumor size (HR, 1.001 [95% CI, 1.000-1.002]; P=.04), T3 staging (HR, 3.71 [95% CI, 1.58-8.67]; P=.003), lymph node involvement (HR, 3.87 [95% CI, 1.11-13.50]; P=.03) and metastasis (HR, 49.76 [95% CI, 4.03-615.00]; P=.002)(Figure 2). When controlling for the aforementioned factors, the primary disease site was not significant (eTable 3).

CT113004177_Fig2_ABCD.jpg

CT113004177_eTable3.jpg

Comment

Although the prognostic significance of primary disease sites on various types of CTCLs has been examined, limited research exists on MF due to its rarity. For the 4265 patients with MF included in our study, statistically significant prognostic factors on multivariate analysis for DSS included age at diagnosis, tumor size, T staging, lymph node involvement, and presence of metastasis. For OS, only age at diagnosis, T staging, and presence of metastasis were statistically significant predictors. Although initially statistically significant on univariate analysis for both OS and DSS, tumor location was not significant when controlling for confounders.

Our population-based analysis found that 5- and 10-year OS for patients with head and neck MF were 85.4% and 75.0%, respectively, and 5- and 10-year DSS were 93.3% and 89.5%, respectively. Our 10-year OS survival rate of 75.0% was slightly worse than the 81.6% reported by Jung et al¹⁶ in a study of 39 cases of MF of the head and neck from the Asan Medical Center database. The difference in survival rate may not only be due to differences in sample size but also because the Asan Medical Center database had a higher proportion of Asian patients as a Korean registry. In our univariate analysis, Asian/Pacific Islander race was shown to be a statistically significant predictor of worse prognosis for OS (P<.001). When comparing survival in patients with head and neck MF vs all primary tumor sites, our OS rate for head and neck MF was more favorable than the 5-year OS of 75% reported by Agar et al²¹ in their analysis of 1502 patients with MF of all locations, though their cohort also included patients with SS, which is known to have a poorer prognosis. Additionally, our 10-year OS rate of 75.0% for patients with MF with a primary tumor site of the head and neck was slightly less favorable than the 81.0% reported by a prior analysis of the SEER database for MF of all locations,²² which initially may be suggestive of worse outcomes associated with MF originating from the head and neck.

Although MF originating in the head and neck region was found to be a statistically significant prognostic factor under univariate analysis (P<.001), tumor location was not significant upon adjusting for confounders in the multivariate analysis. These results are consistent with those reported in a multivariable analysis conducted by Jung et al,¹⁶ which compared 39 cases of head and neck MF to 85 cases without head and neck involvement. The investigators found that the head and neck as the primary site was a significant prognostic factor associated with worsened rates of OS when patients had stages IA to IIA (P=.009) and T2 stage tumors (P=.012) but not in either T1 stage or advanced stage IIB to IVB tumors.¹⁶ In contrast, a study by Su et al¹⁸ evaluating patients with MF from the National Cancer Database found that patients with MF originating in the head and neck region had similar survival compared with MF originating in the lower limbs after pairwise propensity matching. It previously has been postulated that primary MF lesions originating in the head and neck region have relatively higher frequencies of biological markers believed to be associated with more aggressive tumor behavior and poorer prognosis, such as histopathologic folliculotropism, T-cell receptor gene rearrangements, and large-cell transformations.¹⁶ However, MF typically is an indolent disease with advanced-stage MF following an aggressive disease course that often is refractory to treatment. A review from a single academic center noted that 5-year DSS was 97.3% for T1a but only 37.5% for T4.²³ Similarly, a meta-analysis evaluating survival in patients with MF noted the 5-year OS for stage IB was 85.8% while for stage IVB it was only 23.3%.²⁴ As such, having advanced-stage MF influences survival to a far greater extent than the presence of head and neck involvement alone. Accordingly, the significantly higher prevalence of advanced T stage disease and increased likelihood of lymph node involvement in MF lesions originating in the head and neck region (both P<.001) may explain why previous studies noted a poorer survival rate with head and neck involvement, as they did not have the sample size to adjust for these factors. Controlling for the above factors likely explains the nonsignificance of this region as a prognostic indicator in our multivariate analysis of OS and DSS.

Similar to MF originating in the head and neck region, the upper limb as a primary tumor site initially was found to be a significant predictor of both OS and DSS on univariate analysis but not on multivariate analysis. By contrast, Su et al¹⁸ found survival outcomes were worse for patients diagnosed with MF with the upper limb as the primary tumor site compared with the lower limb on multivariate Cox proportional hazards analysis but not on pairwise propensity score matching. The difference in our results compared with Su et al¹⁸ may be because the National Cancer Database only reports OS, while DSS may be more useful in determining prognostic factors associated with poorer survival, especially in an older patient population with greater comorbidities. Furthermore, the nonsignificance of the upper limb as a primary tumor site on further multivariate analysis may be due to similar reasonings as for the head and neck, including more advanced T staging and an anatomic location close to lymph nodes.

Study Limitations—The SEER database is a national registry, which lends itself to potential data heterogeneity in recording and miscoding. Additionally, there may be higher rates of unconfirmed or missing information given the retrospective nature of the SEER database; the database also does not delineate facility type, insurance status, or Charlson/Deyo comorbidity index as demographic factors, which could influence the multivariable analysis. Finally, the SEER database does not further demarcate subtypes of MF, such as the aggressive folliculotropic variant commonly seen in head and neck MF lesions, which precludes independent analysis of disease course by subtype.

Conclusion

Our study evaluated primary disease site as a prognostic factor for OS and DSS in patients with MF. Although head and neck and upper limb as primary disease sites were found to be significant on univariate analysis, they were found to be an insignificant prognostic variable for OS or DSS in our multivariable analysis, potentially due to the aggressive nature of advanced-stage MF and localization close to lymph nodes. Further research including a deeper dive into MF of all stages and subtypes is needed to fully investigate primary disease site as a prognostic indicator. Older age, larger tumor size, a higher T stage, lymph node involvement, and presence of metastasis were associated with worse DSS, and patients with these attributes should be counseled regarding expected disease course and prognosis.

Mycosis fungoides (MF), the most common cutaneous T-cell lymphoma (CTCL), is characterized by clonal proliferation of predominantly CD4⁺ T cells with localization to the skin.¹ Mycosis fungoides typically affects older adults with a male to female ratio of 2:1 but also can occur in children and younger adults.^2,3 Known as the great imitator, the manifestations of MF can be variable with considerable clinical and pathologic overlap with benign inflammatory skin diseases, rendering definitive diagnosis challenging.^4-7 The early stages of classic MF manifest as pruritic erythematous patches and plaques with variable scaling that can progress in later stages to ulceration and tumors.⁸ Histopathologically, classic MF is characterized by epidermotropic proliferation of small- to intermediate-sized pleomorphic lymphocytes with cerebriform nuclei and a haloed appearance; intraepidermal nests of atypical lymphocytes known as Pautrier microabscesses occasionally are observed.⁵ Mycosis fungoides typically follows an indolent clinical course, with advanced-stage MF portending a poor prognosis.^9,10 Current treatment is focused on halting disease progression, with topical therapies, phototherapy, and radiation therapy as the standard therapies for early-stage MF.^11-13 For advanced-stage MF, treatment may include systemic therapies such as interferon alfa and oral retinoids along with chemotherapies for more refractive cases.¹⁴ Allogenic hematopoietic cell transplantation is the only curative treatment.¹¹

Current staging guidelines for MF do not address anatomic location as there is little known about its impact on patient outcomes.^11,15 Due to the indolent nature of MF leading to diagnostic challenges, the exact frequency of each primary disease site for MF also remains unclear, though the suggested incidence of MF of the head and neck ranges from 30% to 70%.^16,17 Involvement of the head and neck^16,18 or external ear and external auditory canal¹⁹ is associated with worse prognosis. The purpose of this study was to examine the impact of anatomic location of primary disease site on survival in MF.

Methods

The National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) database includes patient records from 18 registries and encompasses approximately 48% of the US population.²⁰ Using SEER*STAT software (version 8.4.0.1), we conducted a search of patients diagnosed with MF (International Classification of Diseases for Oncology, Third Edition [ICD-O-3] histologic code 9700/3 [mycosis fungoides]) between 2000 and 2019. For inclusion in the study, patients were required to have a known age, specified primary site, and a known cause of death (if applicable). Patients with known Sézary syndrome (SS)—an aggressive form of CTCL that is characterized by the presence of clonally related neoplastic T cells in the skin, lymph nodes, and peripheral blood—were not included because the World Health Organization/European Organisation for Research and Treatment of Cancer considers SS and MF to be separate entities^1,15; SS does not necessarily arise from preexisting MF and is associated with markedly poorer survival. This study was exempt from institutional review board approval because the data were publicly available and anonymized.

Data Collection—For age at diagnosis, patients were divided into the following categories: younger than 40 years, 40 to 59 years, 60 to 79 years, and 80 years and older. Demographics, tumor characteristics, and surgical management (if applicable) were obtained for each patient. The designations of chemotherapy and radiation treatment in the SEER database are not reliable and prone to false negatives. As such, these were excluded from analysis.

The primary outcomes of interest were overall survival (OS) and disease-specific survival (DSS), which were calculated as time from MF diagnosis to death. Although OS included all patients who died of any cause, DSS only included patients who died of MF.

Statistical Analysis—Demographics (age, sex, race, ethnicity), tumor characteristics (tumor size, primary site, T stage, lymph node involvement, metastasis), and surgical management (if applicable) were summarized. Overall survival and DSS were calculated using Kaplan-Meier analysis. Univariate and multivariable Cox proportional hazards regression models were generated to determine which prognostic factors for MF were associated with poorer OS and DSS. Only statistically significant variables in the univariate analysis were used to construct the multivariable analysis. Hazard ratios (HRs) and their associated 95% CIs were reported. Incidence rates were calculated and age adjusted to the 2000 US standard population. The SEER JoinPoint Regression program was used to determine the annual percent change (APC)—change in incidence rate over time. P<.05 was considered statistically significant. All statistical analyses were conducted with R version 4.0.2.

Patient Demographics and Tumor Characteristics—There were 4265 patients diagnosed with MF from 2000 to 2019. The overall incidence of MF was 2.55 per million (95% CI, 2.48-2.63) when age adjusted to the 2000 US standard population, which increased with time (mean APC, 0.97% per year; P=.01). The mean age at diagnosis was 56.4 years with a male to female ratio of 1.2:1. Males (3.07 per million; 95% CI, 2.94-3.20) had a higher incidence of MF than females (2.16 per million; 95% CI, 2.06-2.26), with incidence in females increasing over time (mean APC, 1.52% per year; P=.02) while incidence in males remained stable (mean APC, 1.09%; P=.37). Patients predominantly self-identified as White (73.08%). Patients with MF of the head and neck were more likely to have smaller tumors (P=.02), a more advanced T stage (P<.001), and lymph node involvement (P=.01) at the time of diagnosis. Additional demographics and tumor characteristics are summarized in eTable 1.

CT113004177_eTable1_part1.jpg

CT113004177_eTable1_part2.jpg

Survival Outcomes—The mean follow-up time was 86.9 months. The 5- and 10-year OS rates were 85.4% (95% CI, 84.2%-86.6%) and 75.0% (95% CI, 73.4%-76.7%), respectively (Figure 1)(Table). The 5- and 10-year DSS rates were 93.3% (95% CI, 92.4%-94.1%) and 89.5% (95% CI, 88.3%-90.6%), respectively. For OS, univariate analysis indicated that significant prognostic factors included increasing age (P<.001), female sex (P<.001), self-identifying as Asian or Pacific Islander (P<.001), self-identifying as Hispanic Latino (P<.001), primary tumor sites of either the head and neck or upper limb (P<.001), T3 or T4 staging (P=.001), lymph node involvement at the time of diagnosis (P<.001), and metastasis (P<.001).

CT113004177_Fig1.jpg

CT113004177_Table.jpg

For DSS, univariate analysis had similar risk factors with self-identifying as Black being an additional risk factor (P=.02), though self-identifying as Asian/Pacific Islander or Hispanic Latino were not significant nor was location on the lower limb. For recorded tumor size, the HR increased by 1.001 per each 1-mm increase in size (eTable 2).

CT113004177_eTable2_part1.jpg

CT113004177_eTable2_part2.jpg

Multivariate analysis showed age at diagnosis (60–79 years: HR, 23.11 [95% CI, 3.03-176.32]; P=.002; ≥80 years: HR, 92.41 [95% CI, 11.78-724.75]; P<.001), T3 staging (HR, 2.37 [95% CI, 1.32-4.27]; P=.004), and metastasis (HR, 40.14 [95% CI, 4.14-389.50]; P=.001) significantly influenced OS. For DSS, multivariate analysis indicated the significant prognostic factors were age at diagnosis (60–79 years: HR, 8.94 [95% CI, 1.16-69.23]; P=.04];≥80 years: HR, 26.71; [95% CI, 3.26-218.99]; P=.002), tumor size (HR, 1.001 [95% CI, 1.000-1.002]; P=.04), T3 staging (HR, 3.71 [95% CI, 1.58-8.67]; P=.003), lymph node involvement (HR, 3.87 [95% CI, 1.11-13.50]; P=.03) and metastasis (HR, 49.76 [95% CI, 4.03-615.00]; P=.002)(Figure 2). When controlling for the aforementioned factors, the primary disease site was not significant (eTable 3).

CT113004177_Fig2_ABCD.jpg

CT113004177_eTable3.jpg

Comment

Although the prognostic significance of primary disease sites on various types of CTCLs has been examined, limited research exists on MF due to its rarity. For the 4265 patients with MF included in our study, statistically significant prognostic factors on multivariate analysis for DSS included age at diagnosis, tumor size, T staging, lymph node involvement, and presence of metastasis. For OS, only age at diagnosis, T staging, and presence of metastasis were statistically significant predictors. Although initially statistically significant on univariate analysis for both OS and DSS, tumor location was not significant when controlling for confounders.

Our population-based analysis found that 5- and 10-year OS for patients with head and neck MF were 85.4% and 75.0%, respectively, and 5- and 10-year DSS were 93.3% and 89.5%, respectively. Our 10-year OS survival rate of 75.0% was slightly worse than the 81.6% reported by Jung et al¹⁶ in a study of 39 cases of MF of the head and neck from the Asan Medical Center database. The difference in survival rate may not only be due to differences in sample size but also because the Asan Medical Center database had a higher proportion of Asian patients as a Korean registry. In our univariate analysis, Asian/Pacific Islander race was shown to be a statistically significant predictor of worse prognosis for OS (P<.001). When comparing survival in patients with head and neck MF vs all primary tumor sites, our OS rate for head and neck MF was more favorable than the 5-year OS of 75% reported by Agar et al²¹ in their analysis of 1502 patients with MF of all locations, though their cohort also included patients with SS, which is known to have a poorer prognosis. Additionally, our 10-year OS rate of 75.0% for patients with MF with a primary tumor site of the head and neck was slightly less favorable than the 81.0% reported by a prior analysis of the SEER database for MF of all locations,²² which initially may be suggestive of worse outcomes associated with MF originating from the head and neck.

Although MF originating in the head and neck region was found to be a statistically significant prognostic factor under univariate analysis (P<.001), tumor location was not significant upon adjusting for confounders in the multivariate analysis. These results are consistent with those reported in a multivariable analysis conducted by Jung et al,¹⁶ which compared 39 cases of head and neck MF to 85 cases without head and neck involvement. The investigators found that the head and neck as the primary site was a significant prognostic factor associated with worsened rates of OS when patients had stages IA to IIA (P=.009) and T2 stage tumors (P=.012) but not in either T1 stage or advanced stage IIB to IVB tumors.¹⁶ In contrast, a study by Su et al¹⁸ evaluating patients with MF from the National Cancer Database found that patients with MF originating in the head and neck region had similar survival compared with MF originating in the lower limbs after pairwise propensity matching. It previously has been postulated that primary MF lesions originating in the head and neck region have relatively higher frequencies of biological markers believed to be associated with more aggressive tumor behavior and poorer prognosis, such as histopathologic folliculotropism, T-cell receptor gene rearrangements, and large-cell transformations.¹⁶ However, MF typically is an indolent disease with advanced-stage MF following an aggressive disease course that often is refractory to treatment. A review from a single academic center noted that 5-year DSS was 97.3% for T1a but only 37.5% for T4.²³ Similarly, a meta-analysis evaluating survival in patients with MF noted the 5-year OS for stage IB was 85.8% while for stage IVB it was only 23.3%.²⁴ As such, having advanced-stage MF influences survival to a far greater extent than the presence of head and neck involvement alone. Accordingly, the significantly higher prevalence of advanced T stage disease and increased likelihood of lymph node involvement in MF lesions originating in the head and neck region (both P<.001) may explain why previous studies noted a poorer survival rate with head and neck involvement, as they did not have the sample size to adjust for these factors. Controlling for the above factors likely explains the nonsignificance of this region as a prognostic indicator in our multivariate analysis of OS and DSS.

Similar to MF originating in the head and neck region, the upper limb as a primary tumor site initially was found to be a significant predictor of both OS and DSS on univariate analysis but not on multivariate analysis. By contrast, Su et al¹⁸ found survival outcomes were worse for patients diagnosed with MF with the upper limb as the primary tumor site compared with the lower limb on multivariate Cox proportional hazards analysis but not on pairwise propensity score matching. The difference in our results compared with Su et al¹⁸ may be because the National Cancer Database only reports OS, while DSS may be more useful in determining prognostic factors associated with poorer survival, especially in an older patient population with greater comorbidities. Furthermore, the nonsignificance of the upper limb as a primary tumor site on further multivariate analysis may be due to similar reasonings as for the head and neck, including more advanced T staging and an anatomic location close to lymph nodes.

Study Limitations—The SEER database is a national registry, which lends itself to potential data heterogeneity in recording and miscoding. Additionally, there may be higher rates of unconfirmed or missing information given the retrospective nature of the SEER database; the database also does not delineate facility type, insurance status, or Charlson/Deyo comorbidity index as demographic factors, which could influence the multivariable analysis. Finally, the SEER database does not further demarcate subtypes of MF, such as the aggressive folliculotropic variant commonly seen in head and neck MF lesions, which precludes independent analysis of disease course by subtype.

Conclusion

Our study evaluated primary disease site as a prognostic factor for OS and DSS in patients with MF. Although head and neck and upper limb as primary disease sites were found to be significant on univariate analysis, they were found to be an insignificant prognostic variable for OS or DSS in our multivariable analysis, potentially due to the aggressive nature of advanced-stage MF and localization close to lymph nodes. Further research including a deeper dive into MF of all stages and subtypes is needed to fully investigate primary disease site as a prognostic indicator. Older age, larger tumor size, a higher T stage, lymph node involvement, and presence of metastasis were associated with worse DSS, and patients with these attributes should be counseled regarding expected disease course and prognosis.

Mycosis fungoides (MF), the most common cutaneous T-cell lymphoma (CTCL), is characterized by clonal proliferation of predominantly CD4⁺ T cells with localization to the skin.¹ Mycosis fungoides typically affects older adults with a male to female ratio of 2:1 but also can occur in children and younger adults.^2,3 Known as the great imitator, the manifestations of MF can be variable with considerable clinical and pathologic overlap with benign inflammatory skin diseases, rendering definitive diagnosis challenging.^4-7 The early stages of classic MF manifest as pruritic erythematous patches and plaques with variable scaling that can progress in later stages to ulceration and tumors.⁸ Histopathologically, classic MF is characterized by epidermotropic proliferation of small- to intermediate-sized pleomorphic lymphocytes with cerebriform nuclei and a haloed appearance; intraepidermal nests of atypical lymphocytes known as Pautrier microabscesses occasionally are observed.⁵ Mycosis fungoides typically follows an indolent clinical course, with advanced-stage MF portending a poor prognosis.^9,10 Current treatment is focused on halting disease progression, with topical therapies, phototherapy, and radiation therapy as the standard therapies for early-stage MF.^11-13 For advanced-stage MF, treatment may include systemic therapies such as interferon alfa and oral retinoids along with chemotherapies for more refractive cases.¹⁴ Allogenic hematopoietic cell transplantation is the only curative treatment.¹¹

Current staging guidelines for MF do not address anatomic location as there is little known about its impact on patient outcomes.^11,15 Due to the indolent nature of MF leading to diagnostic challenges, the exact frequency of each primary disease site for MF also remains unclear, though the suggested incidence of MF of the head and neck ranges from 30% to 70%.^16,17 Involvement of the head and neck^16,18 or external ear and external auditory canal¹⁹ is associated with worse prognosis. The purpose of this study was to examine the impact of anatomic location of primary disease site on survival in MF.

Methods

The National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) database includes patient records from 18 registries and encompasses approximately 48% of the US population.²⁰ Using SEER*STAT software (version 8.4.0.1), we conducted a search of patients diagnosed with MF (International Classification of Diseases for Oncology, Third Edition [ICD-O-3] histologic code 9700/3 [mycosis fungoides]) between 2000 and 2019. For inclusion in the study, patients were required to have a known age, specified primary site, and a known cause of death (if applicable). Patients with known Sézary syndrome (SS)—an aggressive form of CTCL that is characterized by the presence of clonally related neoplastic T cells in the skin, lymph nodes, and peripheral blood—were not included because the World Health Organization/European Organisation for Research and Treatment of Cancer considers SS and MF to be separate entities^1,15; SS does not necessarily arise from preexisting MF and is associated with markedly poorer survival. This study was exempt from institutional review board approval because the data were publicly available and anonymized.

Data Collection—For age at diagnosis, patients were divided into the following categories: younger than 40 years, 40 to 59 years, 60 to 79 years, and 80 years and older. Demographics, tumor characteristics, and surgical management (if applicable) were obtained for each patient. The designations of chemotherapy and radiation treatment in the SEER database are not reliable and prone to false negatives. As such, these were excluded from analysis.

The primary outcomes of interest were overall survival (OS) and disease-specific survival (DSS), which were calculated as time from MF diagnosis to death. Although OS included all patients who died of any cause, DSS only included patients who died of MF.

Statistical Analysis—Demographics (age, sex, race, ethnicity), tumor characteristics (tumor size, primary site, T stage, lymph node involvement, metastasis), and surgical management (if applicable) were summarized. Overall survival and DSS were calculated using Kaplan-Meier analysis. Univariate and multivariable Cox proportional hazards regression models were generated to determine which prognostic factors for MF were associated with poorer OS and DSS. Only statistically significant variables in the univariate analysis were used to construct the multivariable analysis. Hazard ratios (HRs) and their associated 95% CIs were reported. Incidence rates were calculated and age adjusted to the 2000 US standard population. The SEER JoinPoint Regression program was used to determine the annual percent change (APC)—change in incidence rate over time. P<.05 was considered statistically significant. All statistical analyses were conducted with R version 4.0.2.

Patient Demographics and Tumor Characteristics—There were 4265 patients diagnosed with MF from 2000 to 2019. The overall incidence of MF was 2.55 per million (95% CI, 2.48-2.63) when age adjusted to the 2000 US standard population, which increased with time (mean APC, 0.97% per year; P=.01). The mean age at diagnosis was 56.4 years with a male to female ratio of 1.2:1. Males (3.07 per million; 95% CI, 2.94-3.20) had a higher incidence of MF than females (2.16 per million; 95% CI, 2.06-2.26), with incidence in females increasing over time (mean APC, 1.52% per year; P=.02) while incidence in males remained stable (mean APC, 1.09%; P=.37). Patients predominantly self-identified as White (73.08%). Patients with MF of the head and neck were more likely to have smaller tumors (P=.02), a more advanced T stage (P<.001), and lymph node involvement (P=.01) at the time of diagnosis. Additional demographics and tumor characteristics are summarized in eTable 1.

CT113004177_eTable1_part1.jpg

CT113004177_eTable1_part2.jpg

Survival Outcomes—The mean follow-up time was 86.9 months. The 5- and 10-year OS rates were 85.4% (95% CI, 84.2%-86.6%) and 75.0% (95% CI, 73.4%-76.7%), respectively (Figure 1)(Table). The 5- and 10-year DSS rates were 93.3% (95% CI, 92.4%-94.1%) and 89.5% (95% CI, 88.3%-90.6%), respectively. For OS, univariate analysis indicated that significant prognostic factors included increasing age (P<.001), female sex (P<.001), self-identifying as Asian or Pacific Islander (P<.001), self-identifying as Hispanic Latino (P<.001), primary tumor sites of either the head and neck or upper limb (P<.001), T3 or T4 staging (P=.001), lymph node involvement at the time of diagnosis (P<.001), and metastasis (P<.001).

CT113004177_Fig1.jpg

CT113004177_Table.jpg

For DSS, univariate analysis had similar risk factors with self-identifying as Black being an additional risk factor (P=.02), though self-identifying as Asian/Pacific Islander or Hispanic Latino were not significant nor was location on the lower limb. For recorded tumor size, the HR increased by 1.001 per each 1-mm increase in size (eTable 2).

CT113004177_eTable2_part1.jpg

CT113004177_eTable2_part2.jpg

Multivariate analysis showed age at diagnosis (60–79 years: HR, 23.11 [95% CI, 3.03-176.32]; P=.002; ≥80 years: HR, 92.41 [95% CI, 11.78-724.75]; P<.001), T3 staging (HR, 2.37 [95% CI, 1.32-4.27]; P=.004), and metastasis (HR, 40.14 [95% CI, 4.14-389.50]; P=.001) significantly influenced OS. For DSS, multivariate analysis indicated the significant prognostic factors were age at diagnosis (60–79 years: HR, 8.94 [95% CI, 1.16-69.23]; P=.04];≥80 years: HR, 26.71; [95% CI, 3.26-218.99]; P=.002), tumor size (HR, 1.001 [95% CI, 1.000-1.002]; P=.04), T3 staging (HR, 3.71 [95% CI, 1.58-8.67]; P=.003), lymph node involvement (HR, 3.87 [95% CI, 1.11-13.50]; P=.03) and metastasis (HR, 49.76 [95% CI, 4.03-615.00]; P=.002)(Figure 2). When controlling for the aforementioned factors, the primary disease site was not significant (eTable 3).

CT113004177_Fig2_ABCD.jpg

CT113004177_eTable3.jpg

Comment

Although the prognostic significance of primary disease sites on various types of CTCLs has been examined, limited research exists on MF due to its rarity. For the 4265 patients with MF included in our study, statistically significant prognostic factors on multivariate analysis for DSS included age at diagnosis, tumor size, T staging, lymph node involvement, and presence of metastasis. For OS, only age at diagnosis, T staging, and presence of metastasis were statistically significant predictors. Although initially statistically significant on univariate analysis for both OS and DSS, tumor location was not significant when controlling for confounders.

Our population-based analysis found that 5- and 10-year OS for patients with head and neck MF were 85.4% and 75.0%, respectively, and 5- and 10-year DSS were 93.3% and 89.5%, respectively. Our 10-year OS survival rate of 75.0% was slightly worse than the 81.6% reported by Jung et al¹⁶ in a study of 39 cases of MF of the head and neck from the Asan Medical Center database. The difference in survival rate may not only be due to differences in sample size but also because the Asan Medical Center database had a higher proportion of Asian patients as a Korean registry. In our univariate analysis, Asian/Pacific Islander race was shown to be a statistically significant predictor of worse prognosis for OS (P<.001). When comparing survival in patients with head and neck MF vs all primary tumor sites, our OS rate for head and neck MF was more favorable than the 5-year OS of 75% reported by Agar et al²¹ in their analysis of 1502 patients with MF of all locations, though their cohort also included patients with SS, which is known to have a poorer prognosis. Additionally, our 10-year OS rate of 75.0% for patients with MF with a primary tumor site of the head and neck was slightly less favorable than the 81.0% reported by a prior analysis of the SEER database for MF of all locations,²² which initially may be suggestive of worse outcomes associated with MF originating from the head and neck.

Although MF originating in the head and neck region was found to be a statistically significant prognostic factor under univariate analysis (P<.001), tumor location was not significant upon adjusting for confounders in the multivariate analysis. These results are consistent with those reported in a multivariable analysis conducted by Jung et al,¹⁶ which compared 39 cases of head and neck MF to 85 cases without head and neck involvement. The investigators found that the head and neck as the primary site was a significant prognostic factor associated with worsened rates of OS when patients had stages IA to IIA (P=.009) and T2 stage tumors (P=.012) but not in either T1 stage or advanced stage IIB to IVB tumors.¹⁶ In contrast, a study by Su et al¹⁸ evaluating patients with MF from the National Cancer Database found that patients with MF originating in the head and neck region had similar survival compared with MF originating in the lower limbs after pairwise propensity matching. It previously has been postulated that primary MF lesions originating in the head and neck region have relatively higher frequencies of biological markers believed to be associated with more aggressive tumor behavior and poorer prognosis, such as histopathologic folliculotropism, T-cell receptor gene rearrangements, and large-cell transformations.¹⁶ However, MF typically is an indolent disease with advanced-stage MF following an aggressive disease course that often is refractory to treatment. A review from a single academic center noted that 5-year DSS was 97.3% for T1a but only 37.5% for T4.²³ Similarly, a meta-analysis evaluating survival in patients with MF noted the 5-year OS for stage IB was 85.8% while for stage IVB it was only 23.3%.²⁴ As such, having advanced-stage MF influences survival to a far greater extent than the presence of head and neck involvement alone. Accordingly, the significantly higher prevalence of advanced T stage disease and increased likelihood of lymph node involvement in MF lesions originating in the head and neck region (both P<.001) may explain why previous studies noted a poorer survival rate with head and neck involvement, as they did not have the sample size to adjust for these factors. Controlling for the above factors likely explains the nonsignificance of this region as a prognostic indicator in our multivariate analysis of OS and DSS.

Similar to MF originating in the head and neck region, the upper limb as a primary tumor site initially was found to be a significant predictor of both OS and DSS on univariate analysis but not on multivariate analysis. By contrast, Su et al¹⁸ found survival outcomes were worse for patients diagnosed with MF with the upper limb as the primary tumor site compared with the lower limb on multivariate Cox proportional hazards analysis but not on pairwise propensity score matching. The difference in our results compared with Su et al¹⁸ may be because the National Cancer Database only reports OS, while DSS may be more useful in determining prognostic factors associated with poorer survival, especially in an older patient population with greater comorbidities. Furthermore, the nonsignificance of the upper limb as a primary tumor site on further multivariate analysis may be due to similar reasonings as for the head and neck, including more advanced T staging and an anatomic location close to lymph nodes.

Study Limitations—The SEER database is a national registry, which lends itself to potential data heterogeneity in recording and miscoding. Additionally, there may be higher rates of unconfirmed or missing information given the retrospective nature of the SEER database; the database also does not delineate facility type, insurance status, or Charlson/Deyo comorbidity index as demographic factors, which could influence the multivariable analysis. Finally, the SEER database does not further demarcate subtypes of MF, such as the aggressive folliculotropic variant commonly seen in head and neck MF lesions, which precludes independent analysis of disease course by subtype.

Conclusion

Our study evaluated primary disease site as a prognostic factor for OS and DSS in patients with MF. Although head and neck and upper limb as primary disease sites were found to be significant on univariate analysis, they were found to be an insignificant prognostic variable for OS or DSS in our multivariable analysis, potentially due to the aggressive nature of advanced-stage MF and localization close to lymph nodes. Further research including a deeper dive into MF of all stages and subtypes is needed to fully investigate primary disease site as a prognostic indicator. Older age, larger tumor size, a higher T stage, lymph node involvement, and presence of metastasis were associated with worse DSS, and patients with these attributes should be counseled regarding expected disease course and prognosis.

To the Editor:

Phototherapy regularly is utilized in the outpatient setting to address various skin pathologies, including atopic dermatitis, psoriasis, pruritus, vitiligo, and mycosis fungoides.^1,2 Phototherapy is broadly defined by the measured administration of nonionizing radiation within the UV range including wavelengths within the UVA (eg, psoralen sensitizer plus UVA-1) and UVB (eg, broadband UVB, narrowband UVB) spectrums.^1,3 Generally, the mechanism of action is derived from effects on inflammatory components of cutaneous disorders and the induction of apoptosis, both precipitating numerous downstream events.⁴

From 2015 to 2018, there were more than 1.3 million outpatient phototherapy visits in the United States, with the most common procedural indications being dermatitis not otherwise specified, atopic dermatitis, and pruritus.⁵ From 2000 to 2015, the quantity of phototherapy services billed to Medicare trended upwards by an average of 5% per year, increasing from 334,670 in the year 2000 to 692,093 in 2015.⁶ Therefore, an illustration of associated costs would be beneficial. Additionally, because total cost and physician reimbursement fluctuate from year to year, studies demonstrating overall trends can inform both US policymakers and physicians. There is a paucity of research on geographical trends for procedural reimbursements in dermatology for phototherapy. Understanding geographic trends of reimbursement could duly serve to optimize dermatologist practice patterns involving access to viable and quality care for patients seeking treatment as well as draw health policymakers’ attention to striking adjustments in physician fees. Therefore, in this study we aimed to illustrate the most recent regional payment trends in phototherapy procedures for Medicare B patients.

We queried the Centers for Medicare & Medicaid Services Medicare Physician Fee Schedule (MPFS) database (https://www.cms.gov/medicare/payment/fee-schedules/physician/lookup-tool) for the years 2010 to 2023 for Current Procedural Terminology (CPT) codes common to phototherapy procedures: actinotherapy (96900); photochemotherapy by Goeckerman treatment or using petrolatum and UVB (96910); photochemotherapy using psoralen plus UVA (96912); and photochemotherapy of severe dermatoses requiring a minimum of 4 hours of care under direct physician supervision (96913). Nonfacility prices for these procedures were analyzed. For 2010, due to midyear alterations to Medicare reimbursement (owed to bills HR 3962 and HR 4872), the mean price data of MPFS files 2010A and 2010B were used. All dollar values were converted to January 2023 US dollars using corresponding consumer price index inflation data. The Medicare Administrative Contractors were used to group state pricing information by region in accordance with established US Census Bureau subdivisions (https://www.census.gov/programs-surveys/economic-census/guidance-geographies/levels.html). Weighted percentage change in reimbursement rate was calculated using physician (MD or DO) utilization (procedure volume) data available in the 2020 Physician and Other Practitioners Public Use File (https://data.cms.gov/provider-summary-by-type-of-service/medicare-physician-other-practitioners/medicare-physician-other-practitioners-by-provider-and-service). All descriptive statistics and visualization were generated using R software (v4.2.2)(R Development Core Team).

Table 1 provides physician utilization data and the corresponding number of Part B beneficiaries for phototherapy procedures in 2020. There were 65,045 services of actinotherapy provided to a total of 6855 unique Part B beneficiaries, 173,979 services of photochemotherapy by Goeckerman treatment or using petrolatum and UVB provided to 13,122 unique Part B beneficiaries, 2524 services of photochemotherapy using psoralen plus UVA provided to a total of 357 unique Part B beneficiaries, and 37 services of photochemotherapy of severe dermatoses requiring a minimum of 4 hours of care under direct physician supervision provided to a total of 27 unique Part B beneficiaries.

CT113001039_e_Table1.jpg

On average (unweighted), phototherapy reimbursement rates in the North increased by 0.68% between 2010 and 2023 (Table 2). After weighting for 2020 physician utilization, the average change in reimbursement rate was +19.37%. During this time period, CPT code 96910 reported the greatest adjusted increase in reimbursement (+31.45%)($98.12 to $128.98; compound annual growth rate [CAGR], +0.0213), and CPT code 96912 reported the greatest adjusted decrease in reimbursement (−12.76%)($126.09 to $109.97; CAGR, −0.0105). For CPT code 96900, the reported adjusted decrease in reimbursement was −11.68% ($30.21 to $26.68; CAGR, −0.0095), and for CPT code 96913, the reported adjusted decrease in reimbursement was −4.27% ($174.03 to $166.60; CAGR, −0.0034).

CT113001039_e_Table2.jpg

On average (unweighted), phototherapy reimbursement rates in the Midwest increased by 8.40% between 2010 and 2023 (Table 3). After weighting for 2020 physician utilization, the average change in reimbursement rate was +28.53%. During this time period, CPT code 96910 reported the greatest adjusted change in reimbursement (+41.48%)($80.42 to $113.78; CAGR, +0.0270), and CPT code 96912 reported the greatest adjusted decrease in reimbursement (−6.14%)($103.28 to $97.03; CAGR, −0.0049). For CPT code 96900, the reported adjusted decrease in reimbursement was −4.73% ($24.69 to $23.52; CAGR, −0.0037), and for CPT code 96913, the reported adjusted increase in reimbursement was +2.99% ($142.72 to $146.99; CAGR, +0.0023).

CT113001039_e_Table3.jpg

On average (unweighted), phototherapy reimbursement rates in the South decreased by 2.62% between 2010 and 2023 (Table 4). After weighting for 2020 physician utilization, the average change in reimbursement rate was +15.41%. During this time period, CPT code 96910 reported the greatest adjusted change in reimbursement (+27.26%)($90.40 to $115.04 USD; CAGR, +0.0187), and CPT code 96912 reported the greatest adjusted decrease in reimbursement (−15.50%)($116.08 to $98.09; CAGR, −0.0129). For CPT code 96900, the reported adjusted decrease in reimbursement was −15.06% ($28.02 to $23.80; CAGR, −0.0125), and for CPT code 96913, the reported adjusted decrease in reimbursement was −7.19% ($160.11 to $148.61; CAGR, −0.0057).

CT113001039_e_Table4.jpg

On average (unweighted), phototherapy reimbursement rates in the West increased by 27.53% between 2010 and 2023 (Table 5). After weighting for 2020 physician utilization, the average change in reimbursement rate was +51.16%. Reimbursement for all analyzed procedures increased in the western United States. During this time period, CPT code 96910 reported the greatest adjusted increase in reimbursement (+66.56%)($80.84 to $134.65; CAGR, +0.0400), and CPT code 96912 reported the lowest adjusted increase in reimbursement (+10.64%)($103.88 to $114.93; CAGR, +0.0078). For CPT code 96900, the reported adjusted increase in reimbursement was 11.54% ($24.88 to $27.75; CAGR, +0.0084), and for CPT code 96913, the reported adjusted increase in reimbursement was 21.38% ($143.39 to $174.04; CAGR, +0.0150).

CT113001039_e_Table5.jpg

In this study evaluating geographical payment trends for phototherapy from 2010 to 2023, we demonstrated regional inconsistency in mean inflation-adjusted Medicare reimbursement rates. We found that all phototherapy procedures had increased reimbursement in the western United States, whereas all other regions reported cuts in reimbursement rates for at least half of the analyzed procedures. After adjusting for procedure utilization by physicians, weighted mean reimbursement for phototherapy increased in all US regions.

In a cross-sectional study that explored trends in the geographic distribution of dermatologists from 2012 to 2017, dermatologists in the northeastern and western United States were more likely to be located in higher-income zip codes, whereas dermatologists in the southern United States were more likely to be located in lower-income zip codes,⁷ suggesting that payment rate changes are not concordant with cost of living. Additionally, Lauck and colleagues⁸ observed that 75% of the top 20 most common procedures performed by dermatologists had decreased reimbursement (mean change, −10.8%) from 2011 to 2021. Other studies on Medicare reimbursement trends over the last 2 decades have reported major decreases within other specialties, suggesting that declining Medicare reimbursements are not unique to dermatology.^9,10 It is critical to monitor these developments, as the Centers for Medicare & Medicaid Services emphasized health care policy changes aimed at increasing reimbursements for evaluation and management services with compensatory payment cuts in billing for procedural services.¹¹

Mazmudar et al¹² previously reported a mean reimbursement decrease of −6.6% for laser/phototherapy procedures between 2007 and 2021, but these data did not include the heavily utilized Goeckerman treatment. Changes in reimbursement pose major ramifications for dermatologists—for practice size, scope, and longevity—as rates influence changes in commercial insurance reimbursements.¹³ Medicare plays a major role in the US health care system as the second largest expenditure¹⁴; indeed, between 2000 and 2015, Part B billing volume for phototherapy procedures increased 5% annually. However, phototherapy remains inaccessible in many locations due to unequal regional distribution of phototherapy clinics.⁶ Moreover, home phototherapy units are not yet widely utilized because of safety and efficacy concerns, lack of physician oversight, and difficulty obtaining insurance coverage.¹⁵ Acknowledgment and consideration of these geographical trends may persuasively allow policymakers, hospitals, and physicians to facilitate cost-effective phototherapy reimbursements that ensure continued access to quality and sustainable dermatologic care in the United States that tailor to regional needs.

In sum, this analysis reveals regional trends in Part B physician reimbursement for phototherapy procedures, with all US regions reporting a mean increase in phototherapy reimbursement after adjusting for utilization, albeit to varying degrees. Mean reimbursement for photochemotherapy by Goeckerman treatment or using petrolatum and UVB increased most among phototherapy procedures. Mean reimbursement for both actinotherapy and photochemotherapy using psoralen plus UVA decreased in all regions except the western United States.

Limitations include the restriction to Part B MPFS and the reliance on single-year (2020) physician utilization data to compute weighted changes in average reimbursement across a multiyear range, effectively restricting sweeping conclusions. Still, this study puts forth actionable insights for dermatologists and policymakers alike to appreciate and consider.

To the Editor:

Phototherapy regularly is utilized in the outpatient setting to address various skin pathologies, including atopic dermatitis, psoriasis, pruritus, vitiligo, and mycosis fungoides.^1,2 Phototherapy is broadly defined by the measured administration of nonionizing radiation within the UV range including wavelengths within the UVA (eg, psoralen sensitizer plus UVA-1) and UVB (eg, broadband UVB, narrowband UVB) spectrums.^1,3 Generally, the mechanism of action is derived from effects on inflammatory components of cutaneous disorders and the induction of apoptosis, both precipitating numerous downstream events.⁴

From 2015 to 2018, there were more than 1.3 million outpatient phototherapy visits in the United States, with the most common procedural indications being dermatitis not otherwise specified, atopic dermatitis, and pruritus.⁵ From 2000 to 2015, the quantity of phototherapy services billed to Medicare trended upwards by an average of 5% per year, increasing from 334,670 in the year 2000 to 692,093 in 2015.⁶ Therefore, an illustration of associated costs would be beneficial. Additionally, because total cost and physician reimbursement fluctuate from year to year, studies demonstrating overall trends can inform both US policymakers and physicians. There is a paucity of research on geographical trends for procedural reimbursements in dermatology for phototherapy. Understanding geographic trends of reimbursement could duly serve to optimize dermatologist practice patterns involving access to viable and quality care for patients seeking treatment as well as draw health policymakers’ attention to striking adjustments in physician fees. Therefore, in this study we aimed to illustrate the most recent regional payment trends in phototherapy procedures for Medicare B patients.

We queried the Centers for Medicare & Medicaid Services Medicare Physician Fee Schedule (MPFS) database (https://www.cms.gov/medicare/payment/fee-schedules/physician/lookup-tool) for the years 2010 to 2023 for Current Procedural Terminology (CPT) codes common to phototherapy procedures: actinotherapy (96900); photochemotherapy by Goeckerman treatment or using petrolatum and UVB (96910); photochemotherapy using psoralen plus UVA (96912); and photochemotherapy of severe dermatoses requiring a minimum of 4 hours of care under direct physician supervision (96913). Nonfacility prices for these procedures were analyzed. For 2010, due to midyear alterations to Medicare reimbursement (owed to bills HR 3962 and HR 4872), the mean price data of MPFS files 2010A and 2010B were used. All dollar values were converted to January 2023 US dollars using corresponding consumer price index inflation data. The Medicare Administrative Contractors were used to group state pricing information by region in accordance with established US Census Bureau subdivisions (https://www.census.gov/programs-surveys/economic-census/guidance-geographies/levels.html). Weighted percentage change in reimbursement rate was calculated using physician (MD or DO) utilization (procedure volume) data available in the 2020 Physician and Other Practitioners Public Use File (https://data.cms.gov/provider-summary-by-type-of-service/medicare-physician-other-practitioners/medicare-physician-other-practitioners-by-provider-and-service). All descriptive statistics and visualization were generated using R software (v4.2.2)(R Development Core Team).

Table 1 provides physician utilization data and the corresponding number of Part B beneficiaries for phototherapy procedures in 2020. There were 65,045 services of actinotherapy provided to a total of 6855 unique Part B beneficiaries, 173,979 services of photochemotherapy by Goeckerman treatment or using petrolatum and UVB provided to 13,122 unique Part B beneficiaries, 2524 services of photochemotherapy using psoralen plus UVA provided to a total of 357 unique Part B beneficiaries, and 37 services of photochemotherapy of severe dermatoses requiring a minimum of 4 hours of care under direct physician supervision provided to a total of 27 unique Part B beneficiaries.

CT113001039_e_Table1.jpg

On average (unweighted), phototherapy reimbursement rates in the North increased by 0.68% between 2010 and 2023 (Table 2). After weighting for 2020 physician utilization, the average change in reimbursement rate was +19.37%. During this time period, CPT code 96910 reported the greatest adjusted increase in reimbursement (+31.45%)($98.12 to $128.98; compound annual growth rate [CAGR], +0.0213), and CPT code 96912 reported the greatest adjusted decrease in reimbursement (−12.76%)($126.09 to $109.97; CAGR, −0.0105). For CPT code 96900, the reported adjusted decrease in reimbursement was −11.68% ($30.21 to $26.68; CAGR, −0.0095), and for CPT code 96913, the reported adjusted decrease in reimbursement was −4.27% ($174.03 to $166.60; CAGR, −0.0034).

CT113001039_e_Table2.jpg

On average (unweighted), phototherapy reimbursement rates in the Midwest increased by 8.40% between 2010 and 2023 (Table 3). After weighting for 2020 physician utilization, the average change in reimbursement rate was +28.53%. During this time period, CPT code 96910 reported the greatest adjusted change in reimbursement (+41.48%)($80.42 to $113.78; CAGR, +0.0270), and CPT code 96912 reported the greatest adjusted decrease in reimbursement (−6.14%)($103.28 to $97.03; CAGR, −0.0049). For CPT code 96900, the reported adjusted decrease in reimbursement was −4.73% ($24.69 to $23.52; CAGR, −0.0037), and for CPT code 96913, the reported adjusted increase in reimbursement was +2.99% ($142.72 to $146.99; CAGR, +0.0023).

CT113001039_e_Table3.jpg

On average (unweighted), phototherapy reimbursement rates in the South decreased by 2.62% between 2010 and 2023 (Table 4). After weighting for 2020 physician utilization, the average change in reimbursement rate was +15.41%. During this time period, CPT code 96910 reported the greatest adjusted change in reimbursement (+27.26%)($90.40 to $115.04 USD; CAGR, +0.0187), and CPT code 96912 reported the greatest adjusted decrease in reimbursement (−15.50%)($116.08 to $98.09; CAGR, −0.0129). For CPT code 96900, the reported adjusted decrease in reimbursement was −15.06% ($28.02 to $23.80; CAGR, −0.0125), and for CPT code 96913, the reported adjusted decrease in reimbursement was −7.19% ($160.11 to $148.61; CAGR, −0.0057).

CT113001039_e_Table4.jpg

On average (unweighted), phototherapy reimbursement rates in the West increased by 27.53% between 2010 and 2023 (Table 5). After weighting for 2020 physician utilization, the average change in reimbursement rate was +51.16%. Reimbursement for all analyzed procedures increased in the western United States. During this time period, CPT code 96910 reported the greatest adjusted increase in reimbursement (+66.56%)($80.84 to $134.65; CAGR, +0.0400), and CPT code 96912 reported the lowest adjusted increase in reimbursement (+10.64%)($103.88 to $114.93; CAGR, +0.0078). For CPT code 96900, the reported adjusted increase in reimbursement was 11.54% ($24.88 to $27.75; CAGR, +0.0084), and for CPT code 96913, the reported adjusted increase in reimbursement was 21.38% ($143.39 to $174.04; CAGR, +0.0150).

CT113001039_e_Table5.jpg

In this study evaluating geographical payment trends for phototherapy from 2010 to 2023, we demonstrated regional inconsistency in mean inflation-adjusted Medicare reimbursement rates. We found that all phototherapy procedures had increased reimbursement in the western United States, whereas all other regions reported cuts in reimbursement rates for at least half of the analyzed procedures. After adjusting for procedure utilization by physicians, weighted mean reimbursement for phototherapy increased in all US regions.

In a cross-sectional study that explored trends in the geographic distribution of dermatologists from 2012 to 2017, dermatologists in the northeastern and western United States were more likely to be located in higher-income zip codes, whereas dermatologists in the southern United States were more likely to be located in lower-income zip codes,⁷ suggesting that payment rate changes are not concordant with cost of living. Additionally, Lauck and colleagues⁸ observed that 75% of the top 20 most common procedures performed by dermatologists had decreased reimbursement (mean change, −10.8%) from 2011 to 2021. Other studies on Medicare reimbursement trends over the last 2 decades have reported major decreases within other specialties, suggesting that declining Medicare reimbursements are not unique to dermatology.^9,10 It is critical to monitor these developments, as the Centers for Medicare & Medicaid Services emphasized health care policy changes aimed at increasing reimbursements for evaluation and management services with compensatory payment cuts in billing for procedural services.¹¹

Mazmudar et al¹² previously reported a mean reimbursement decrease of −6.6% for laser/phototherapy procedures between 2007 and 2021, but these data did not include the heavily utilized Goeckerman treatment. Changes in reimbursement pose major ramifications for dermatologists—for practice size, scope, and longevity—as rates influence changes in commercial insurance reimbursements.¹³ Medicare plays a major role in the US health care system as the second largest expenditure¹⁴; indeed, between 2000 and 2015, Part B billing volume for phototherapy procedures increased 5% annually. However, phototherapy remains inaccessible in many locations due to unequal regional distribution of phototherapy clinics.⁶ Moreover, home phototherapy units are not yet widely utilized because of safety and efficacy concerns, lack of physician oversight, and difficulty obtaining insurance coverage.¹⁵ Acknowledgment and consideration of these geographical trends may persuasively allow policymakers, hospitals, and physicians to facilitate cost-effective phototherapy reimbursements that ensure continued access to quality and sustainable dermatologic care in the United States that tailor to regional needs.

In sum, this analysis reveals regional trends in Part B physician reimbursement for phototherapy procedures, with all US regions reporting a mean increase in phototherapy reimbursement after adjusting for utilization, albeit to varying degrees. Mean reimbursement for photochemotherapy by Goeckerman treatment or using petrolatum and UVB increased most among phototherapy procedures. Mean reimbursement for both actinotherapy and photochemotherapy using psoralen plus UVA decreased in all regions except the western United States.

Limitations include the restriction to Part B MPFS and the reliance on single-year (2020) physician utilization data to compute weighted changes in average reimbursement across a multiyear range, effectively restricting sweeping conclusions. Still, this study puts forth actionable insights for dermatologists and policymakers alike to appreciate and consider.

To the Editor:

Phototherapy regularly is utilized in the outpatient setting to address various skin pathologies, including atopic dermatitis, psoriasis, pruritus, vitiligo, and mycosis fungoides.^1,2 Phototherapy is broadly defined by the measured administration of nonionizing radiation within the UV range including wavelengths within the UVA (eg, psoralen sensitizer plus UVA-1) and UVB (eg, broadband UVB, narrowband UVB) spectrums.^1,3 Generally, the mechanism of action is derived from effects on inflammatory components of cutaneous disorders and the induction of apoptosis, both precipitating numerous downstream events.⁴

From 2015 to 2018, there were more than 1.3 million outpatient phototherapy visits in the United States, with the most common procedural indications being dermatitis not otherwise specified, atopic dermatitis, and pruritus.⁵ From 2000 to 2015, the quantity of phototherapy services billed to Medicare trended upwards by an average of 5% per year, increasing from 334,670 in the year 2000 to 692,093 in 2015.⁶ Therefore, an illustration of associated costs would be beneficial. Additionally, because total cost and physician reimbursement fluctuate from year to year, studies demonstrating overall trends can inform both US policymakers and physicians. There is a paucity of research on geographical trends for procedural reimbursements in dermatology for phototherapy. Understanding geographic trends of reimbursement could duly serve to optimize dermatologist practice patterns involving access to viable and quality care for patients seeking treatment as well as draw health policymakers’ attention to striking adjustments in physician fees. Therefore, in this study we aimed to illustrate the most recent regional payment trends in phototherapy procedures for Medicare B patients.

We queried the Centers for Medicare & Medicaid Services Medicare Physician Fee Schedule (MPFS) database (https://www.cms.gov/medicare/payment/fee-schedules/physician/lookup-tool) for the years 2010 to 2023 for Current Procedural Terminology (CPT) codes common to phototherapy procedures: actinotherapy (96900); photochemotherapy by Goeckerman treatment or using petrolatum and UVB (96910); photochemotherapy using psoralen plus UVA (96912); and photochemotherapy of severe dermatoses requiring a minimum of 4 hours of care under direct physician supervision (96913). Nonfacility prices for these procedures were analyzed. For 2010, due to midyear alterations to Medicare reimbursement (owed to bills HR 3962 and HR 4872), the mean price data of MPFS files 2010A and 2010B were used. All dollar values were converted to January 2023 US dollars using corresponding consumer price index inflation data. The Medicare Administrative Contractors were used to group state pricing information by region in accordance with established US Census Bureau subdivisions (https://www.census.gov/programs-surveys/economic-census/guidance-geographies/levels.html). Weighted percentage change in reimbursement rate was calculated using physician (MD or DO) utilization (procedure volume) data available in the 2020 Physician and Other Practitioners Public Use File (https://data.cms.gov/provider-summary-by-type-of-service/medicare-physician-other-practitioners/medicare-physician-other-practitioners-by-provider-and-service). All descriptive statistics and visualization were generated using R software (v4.2.2)(R Development Core Team).

Table 1 provides physician utilization data and the corresponding number of Part B beneficiaries for phototherapy procedures in 2020. There were 65,045 services of actinotherapy provided to a total of 6855 unique Part B beneficiaries, 173,979 services of photochemotherapy by Goeckerman treatment or using petrolatum and UVB provided to 13,122 unique Part B beneficiaries, 2524 services of photochemotherapy using psoralen plus UVA provided to a total of 357 unique Part B beneficiaries, and 37 services of photochemotherapy of severe dermatoses requiring a minimum of 4 hours of care under direct physician supervision provided to a total of 27 unique Part B beneficiaries.

CT113001039_e_Table1.jpg

On average (unweighted), phototherapy reimbursement rates in the North increased by 0.68% between 2010 and 2023 (Table 2). After weighting for 2020 physician utilization, the average change in reimbursement rate was +19.37%. During this time period, CPT code 96910 reported the greatest adjusted increase in reimbursement (+31.45%)($98.12 to $128.98; compound annual growth rate [CAGR], +0.0213), and CPT code 96912 reported the greatest adjusted decrease in reimbursement (−12.76%)($126.09 to $109.97; CAGR, −0.0105). For CPT code 96900, the reported adjusted decrease in reimbursement was −11.68% ($30.21 to $26.68; CAGR, −0.0095), and for CPT code 96913, the reported adjusted decrease in reimbursement was −4.27% ($174.03 to $166.60; CAGR, −0.0034).

CT113001039_e_Table2.jpg

On average (unweighted), phototherapy reimbursement rates in the Midwest increased by 8.40% between 2010 and 2023 (Table 3). After weighting for 2020 physician utilization, the average change in reimbursement rate was +28.53%. During this time period, CPT code 96910 reported the greatest adjusted change in reimbursement (+41.48%)($80.42 to $113.78; CAGR, +0.0270), and CPT code 96912 reported the greatest adjusted decrease in reimbursement (−6.14%)($103.28 to $97.03; CAGR, −0.0049). For CPT code 96900, the reported adjusted decrease in reimbursement was −4.73% ($24.69 to $23.52; CAGR, −0.0037), and for CPT code 96913, the reported adjusted increase in reimbursement was +2.99% ($142.72 to $146.99; CAGR, +0.0023).

CT113001039_e_Table3.jpg

On average (unweighted), phototherapy reimbursement rates in the South decreased by 2.62% between 2010 and 2023 (Table 4). After weighting for 2020 physician utilization, the average change in reimbursement rate was +15.41%. During this time period, CPT code 96910 reported the greatest adjusted change in reimbursement (+27.26%)($90.40 to $115.04 USD; CAGR, +0.0187), and CPT code 96912 reported the greatest adjusted decrease in reimbursement (−15.50%)($116.08 to $98.09; CAGR, −0.0129). For CPT code 96900, the reported adjusted decrease in reimbursement was −15.06% ($28.02 to $23.80; CAGR, −0.0125), and for CPT code 96913, the reported adjusted decrease in reimbursement was −7.19% ($160.11 to $148.61; CAGR, −0.0057).

CT113001039_e_Table4.jpg

On average (unweighted), phototherapy reimbursement rates in the West increased by 27.53% between 2010 and 2023 (Table 5). After weighting for 2020 physician utilization, the average change in reimbursement rate was +51.16%. Reimbursement for all analyzed procedures increased in the western United States. During this time period, CPT code 96910 reported the greatest adjusted increase in reimbursement (+66.56%)($80.84 to $134.65; CAGR, +0.0400), and CPT code 96912 reported the lowest adjusted increase in reimbursement (+10.64%)($103.88 to $114.93; CAGR, +0.0078). For CPT code 96900, the reported adjusted increase in reimbursement was 11.54% ($24.88 to $27.75; CAGR, +0.0084), and for CPT code 96913, the reported adjusted increase in reimbursement was 21.38% ($143.39 to $174.04; CAGR, +0.0150).

CT113001039_e_Table5.jpg

In this study evaluating geographical payment trends for phototherapy from 2010 to 2023, we demonstrated regional inconsistency in mean inflation-adjusted Medicare reimbursement rates. We found that all phototherapy procedures had increased reimbursement in the western United States, whereas all other regions reported cuts in reimbursement rates for at least half of the analyzed procedures. After adjusting for procedure utilization by physicians, weighted mean reimbursement for phototherapy increased in all US regions.

In a cross-sectional study that explored trends in the geographic distribution of dermatologists from 2012 to 2017, dermatologists in the northeastern and western United States were more likely to be located in higher-income zip codes, whereas dermatologists in the southern United States were more likely to be located in lower-income zip codes,⁷ suggesting that payment rate changes are not concordant with cost of living. Additionally, Lauck and colleagues⁸ observed that 75% of the top 20 most common procedures performed by dermatologists had decreased reimbursement (mean change, −10.8%) from 2011 to 2021. Other studies on Medicare reimbursement trends over the last 2 decades have reported major decreases within other specialties, suggesting that declining Medicare reimbursements are not unique to dermatology.^9,10 It is critical to monitor these developments, as the Centers for Medicare & Medicaid Services emphasized health care policy changes aimed at increasing reimbursements for evaluation and management services with compensatory payment cuts in billing for procedural services.¹¹

Mazmudar et al¹² previously reported a mean reimbursement decrease of −6.6% for laser/phototherapy procedures between 2007 and 2021, but these data did not include the heavily utilized Goeckerman treatment. Changes in reimbursement pose major ramifications for dermatologists—for practice size, scope, and longevity—as rates influence changes in commercial insurance reimbursements.¹³ Medicare plays a major role in the US health care system as the second largest expenditure¹⁴; indeed, between 2000 and 2015, Part B billing volume for phototherapy procedures increased 5% annually. However, phototherapy remains inaccessible in many locations due to unequal regional distribution of phototherapy clinics.⁶ Moreover, home phototherapy units are not yet widely utilized because of safety and efficacy concerns, lack of physician oversight, and difficulty obtaining insurance coverage.¹⁵ Acknowledgment and consideration of these geographical trends may persuasively allow policymakers, hospitals, and physicians to facilitate cost-effective phototherapy reimbursements that ensure continued access to quality and sustainable dermatologic care in the United States that tailor to regional needs.

In sum, this analysis reveals regional trends in Part B physician reimbursement for phototherapy procedures, with all US regions reporting a mean increase in phototherapy reimbursement after adjusting for utilization, albeit to varying degrees. Mean reimbursement for photochemotherapy by Goeckerman treatment or using petrolatum and UVB increased most among phototherapy procedures. Mean reimbursement for both actinotherapy and photochemotherapy using psoralen plus UVA decreased in all regions except the western United States.

Limitations include the restriction to Part B MPFS and the reliance on single-year (2020) physician utilization data to compute weighted changes in average reimbursement across a multiyear range, effectively restricting sweeping conclusions. Still, this study puts forth actionable insights for dermatologists and policymakers alike to appreciate and consider.

Competency in paracentesis is an important procedural skill for medical practitioners caring for patients with decompensated liver cirrhosis. Paracentesis is performed to drain ascitic fluid for both diagnosis and/or therapeutic purposes.¹ While this procedure can be performed without the use of ultrasound, it is preferable to use ultrasound to identify an area of fluid that is away from dangerous anatomy including bowel loops, the liver, and spleen. After prepping the area, lidocaine is administered locally. A catheter is then inserted until fluid begins flowing freely. The catheter is connected to a suction canister or collection kit, and the patient is monitored until the flow ceases. Samples can be sent for analysis to determine the etiology of ascites, identify concerns for infection, and more.

Paracentesis is a very common procedure. Barsuk and colleagues noted that between 2010 and 2012, 97,577 procedures were performed across 120 academic medical centers and 290 affiliated hospitals.² Patients undergo paracentesis in a variety of settings including the emergency department, inpatient hospitalizations, and clinics. Some patients may require only 1 paracentesis procedure while others may require it regularly.

Due to the rising need for paracentesis in the Central Texas Veterans Affairs Hospital (CTVAH) in Temple, a paracentesis clinic was started in February 2018. The goal of the paracentesis clinic was multifocal—to reduce hospital admissions, improve access to regularly scheduled procedures, decrease wait times, and increase patient satisfaction.³ Through the CTVAH affiliation with the Texas A&M internal medicine residency program, the paracentesis clinic started involving and training residents on this procedure. Up to 3 residents are on weekly rotation and can perform up to 6 paracentesis procedures in a week. The purpose of this article was to evaluate resident competency in paracentesis after completion of the paracentesis clinic.

Methods

The paracentesis clinic schedules up to 3 patients on Tuesdays and Thursdays between 8 am and noon. All the necessary equipment is readily available and includes the paracentesis kit, lidocaine, sterile gloves, ultrasound, and albumin if needed. Because this procedure is performed at the hospital, direct access to the emergency department is available. Residents are scheduled weekly. Up to 2 residents are scheduled for the paracentesis clinic during their dedicated clinic week. They are expected to practice obtaining consent, performing the procedure, and documenting the encounter under staff supervision. Additionally, 1 or 2 residents participate in the paracentesis clinic as part of an ultrasound elective twice per year. In this elective, they practice ultrasound skills using a simulation and translate that information in the paracentesis clinic while identifying anatomy and performing the paracentesis procedure under staff supervision.

table.png

A survey was sent via email to all categorical internal medicine residents across all 3 program years at the time of data collection. Competency for paracentesis sign-off was defined as completing and logging 5 procedures supervised by a competent physician who confirmed that all portions of the procedure were performed correctly. Residents were also asked to answer questions on a scale from 1 to 10, with 1 representing no confidence and 10 representing strong confidence to practice independently (Table).

We also evaluated the number of procedures performed by internal medicine residents 3 years before the clinic was started in 2015 up to the completion of 2022. The numbers were obtained by examining procedural log data for each year for all internal medicine residents.

Results

Thirty-three residents completed the survey: 10 first-year internal medicine residents (PGY1), 12 second-year residents (PGY2), and 11 third-year residents (PGY3). The mean participation was 4.8 paracentesis sessions per person for the duration of the study. The range of paracentesis procedures performed varied based on PGY year: PGY1s performed 1 to > 10 procedures, PGY2s performed 2 to > 10 procedures, and PGY3s performed 5 to > 10 procedures. Thirty-six percent of residents completed > 10 procedures in the paracentesis clinic; 82% of PGY3s had completed > 10 procedures by December of their third year. Twenty-six residents (79%) were credentialed to perform paracentesis procedures independently after performing > 5 procedures, and 7 residents were not yet cleared for procedural independence.

In the survey, residents rated their comfort with performing paracentesis procedures independently at a mean of 7.9. The mean comfort reported by PGY1s was 7.2, PGY2s was 7.3, and PGY3s was 9.3. Residents also rated their opinion on whether or not the paracentesis clinic adequately prepared them for paracentesis procedural independence; the mean was 8.9 across all residents.

The total number of procedures performed by residents at CTVAH also increased. Starting in 2015, 3 years before the clinic was started, 38 procedures were performed by internal medicine residents, followed by 72 procedures in 2016; 76 in 2017; 58 in 2018; 94 in 2019; 88 in 2020; 136 in 2021; and 188 in 2022.

Paracentesis is a simple but invasive procedure to relieve ascites, often relieving patients’ symptoms, preventing hospital admission, and increasing patient satisfaction.⁴ The CTVAH does not have the capacity to perform outpatient paracentesis effectively in its emergency or radiology departments. Furthermore, the use of the emergency or radiology departments for routine paracentesis may not be feasible due to the acuity of care being provided, as these procedures can be time consuming and can draw away critical resources and time from patients that need emergent care. The paracentesis clinic was then formed to provide veterans access to the procedural care they need, while also preparing residents to ably and confidently perform the procedure independently.

Based on our study, most residents were cleared to independently perform paracentesis procedures across all 3 years, with 79% of residents having completed the required 5 supervised procedures to independently practice. A study assessing unsupervised practice standards showed that paracentesis skill declines as soon as 3 months after training. However, retraining was shown to potentially interrupt this skill decline.⁵ Studies have shown that procedure-driven electives or services significantly improved paracentesis certification rates and total logged procedures, with minimal funding or scheduling changes required.⁶ Our clinic showed a significant increase in the number of procedures logged starting with the minimum of 38 procedures in 2015 and ending with 188 procedures logged at the end of 2022.

By allowing residents to routinely return to the paracentesis clinic across all 3 years, residents were more likely to feel comfortable independently performing the procedure, with residents reporting a mean comfort score of 7.9. The spaced repetition and ability to work with the clinic during elective time allows regular opportunities to undergo supervised training in a controlled environment and created scheduled retraining opportunities. Future studies should evaluate residents prior to each paracentesis clinic to ascertain if skill decline is occurring at a slower rate.

The inpatient effect of the clinic is also multifocal. Pham and colleagues showed that integrating paracentesis into timely training can reduce paracentesis delay and delays in care.⁷ By increasing the volume of procedures each resident performs and creating a sense of confidence amongst residents, the clinic increases the number of residents able and willing to perform inpatient procedures, thus reducing the number of unnecessary consultations and hospital resources. One of the reasons the paracentesis clinic was started was to allow patients to have scheduled times to remove fluid from their abdomen, thus cutting down on emergency department procedures and unnecessary admissions. Additionally, the benefits of early paracentesis procedural performance by residents and internal medicine physicians have been demonstrated in the literature. A study by Gaetano and colleagues noted that patients undergoing early paracentesis had reduced mortality of 5.5% vs 7.5% in those undergoing late paracentesis.⁸ This study also showed the in-hospital mortality rate was decreased with paracentesis (6.3%) vs without paracentesis (8.9%).⁸ By offering residents a chance to participate in the clinic, we have shown that regular opportunities to perform paracentesis may increase the number of physicians capable of independently practicing, improve procedural competency, and improve patient access to this procedure.

Limitations

Our study was not free of bias and has potential weaknesses. The survey was sent to all current residents who have participated in the paracentesis clinic, but not every resident filled out the survey (55% of all residents across 3 years completed the survey, 68.7% who had done clinic that year completed the survey). There is a possibility that those not signed off avoided doing the survey, but we are unable to confirm this. The survey also depended on resident recall of the number of paracenteses completed or looking at their procedure log. It is possible that some procedures were not documented, changing the true number. Additionally, rating comfortability with procedures is subjective, which may also create a source of potential weakness. Future projects should include a baseline survey for residents, followed by a repeat survey a year later to show changes from baseline competency.

Conclusions

A dedicated paracentesis clinic with internal medicine resident involvement may increase resident paracentesis procedural independence, the number of procedures available and performed, and procedural comfort level.

Competency in paracentesis is an important procedural skill for medical practitioners caring for patients with decompensated liver cirrhosis. Paracentesis is performed to drain ascitic fluid for both diagnosis and/or therapeutic purposes.¹ While this procedure can be performed without the use of ultrasound, it is preferable to use ultrasound to identify an area of fluid that is away from dangerous anatomy including bowel loops, the liver, and spleen. After prepping the area, lidocaine is administered locally. A catheter is then inserted until fluid begins flowing freely. The catheter is connected to a suction canister or collection kit, and the patient is monitored until the flow ceases. Samples can be sent for analysis to determine the etiology of ascites, identify concerns for infection, and more.

Paracentesis is a very common procedure. Barsuk and colleagues noted that between 2010 and 2012, 97,577 procedures were performed across 120 academic medical centers and 290 affiliated hospitals.² Patients undergo paracentesis in a variety of settings including the emergency department, inpatient hospitalizations, and clinics. Some patients may require only 1 paracentesis procedure while others may require it regularly.

Due to the rising need for paracentesis in the Central Texas Veterans Affairs Hospital (CTVAH) in Temple, a paracentesis clinic was started in February 2018. The goal of the paracentesis clinic was multifocal—to reduce hospital admissions, improve access to regularly scheduled procedures, decrease wait times, and increase patient satisfaction.³ Through the CTVAH affiliation with the Texas A&M internal medicine residency program, the paracentesis clinic started involving and training residents on this procedure. Up to 3 residents are on weekly rotation and can perform up to 6 paracentesis procedures in a week. The purpose of this article was to evaluate resident competency in paracentesis after completion of the paracentesis clinic.

Methods

The paracentesis clinic schedules up to 3 patients on Tuesdays and Thursdays between 8 am and noon. All the necessary equipment is readily available and includes the paracentesis kit, lidocaine, sterile gloves, ultrasound, and albumin if needed. Because this procedure is performed at the hospital, direct access to the emergency department is available. Residents are scheduled weekly. Up to 2 residents are scheduled for the paracentesis clinic during their dedicated clinic week. They are expected to practice obtaining consent, performing the procedure, and documenting the encounter under staff supervision. Additionally, 1 or 2 residents participate in the paracentesis clinic as part of an ultrasound elective twice per year. In this elective, they practice ultrasound skills using a simulation and translate that information in the paracentesis clinic while identifying anatomy and performing the paracentesis procedure under staff supervision.

table.png

A survey was sent via email to all categorical internal medicine residents across all 3 program years at the time of data collection. Competency for paracentesis sign-off was defined as completing and logging 5 procedures supervised by a competent physician who confirmed that all portions of the procedure were performed correctly. Residents were also asked to answer questions on a scale from 1 to 10, with 1 representing no confidence and 10 representing strong confidence to practice independently (Table).

We also evaluated the number of procedures performed by internal medicine residents 3 years before the clinic was started in 2015 up to the completion of 2022. The numbers were obtained by examining procedural log data for each year for all internal medicine residents.

Results

Thirty-three residents completed the survey: 10 first-year internal medicine residents (PGY1), 12 second-year residents (PGY2), and 11 third-year residents (PGY3). The mean participation was 4.8 paracentesis sessions per person for the duration of the study. The range of paracentesis procedures performed varied based on PGY year: PGY1s performed 1 to > 10 procedures, PGY2s performed 2 to > 10 procedures, and PGY3s performed 5 to > 10 procedures. Thirty-six percent of residents completed > 10 procedures in the paracentesis clinic; 82% of PGY3s had completed > 10 procedures by December of their third year. Twenty-six residents (79%) were credentialed to perform paracentesis procedures independently after performing > 5 procedures, and 7 residents were not yet cleared for procedural independence.

In the survey, residents rated their comfort with performing paracentesis procedures independently at a mean of 7.9. The mean comfort reported by PGY1s was 7.2, PGY2s was 7.3, and PGY3s was 9.3. Residents also rated their opinion on whether or not the paracentesis clinic adequately prepared them for paracentesis procedural independence; the mean was 8.9 across all residents.

The total number of procedures performed by residents at CTVAH also increased. Starting in 2015, 3 years before the clinic was started, 38 procedures were performed by internal medicine residents, followed by 72 procedures in 2016; 76 in 2017; 58 in 2018; 94 in 2019; 88 in 2020; 136 in 2021; and 188 in 2022.

Paracentesis is a simple but invasive procedure to relieve ascites, often relieving patients’ symptoms, preventing hospital admission, and increasing patient satisfaction.⁴ The CTVAH does not have the capacity to perform outpatient paracentesis effectively in its emergency or radiology departments. Furthermore, the use of the emergency or radiology departments for routine paracentesis may not be feasible due to the acuity of care being provided, as these procedures can be time consuming and can draw away critical resources and time from patients that need emergent care. The paracentesis clinic was then formed to provide veterans access to the procedural care they need, while also preparing residents to ably and confidently perform the procedure independently.

Based on our study, most residents were cleared to independently perform paracentesis procedures across all 3 years, with 79% of residents having completed the required 5 supervised procedures to independently practice. A study assessing unsupervised practice standards showed that paracentesis skill declines as soon as 3 months after training. However, retraining was shown to potentially interrupt this skill decline.⁵ Studies have shown that procedure-driven electives or services significantly improved paracentesis certification rates and total logged procedures, with minimal funding or scheduling changes required.⁶ Our clinic showed a significant increase in the number of procedures logged starting with the minimum of 38 procedures in 2015 and ending with 188 procedures logged at the end of 2022.

By allowing residents to routinely return to the paracentesis clinic across all 3 years, residents were more likely to feel comfortable independently performing the procedure, with residents reporting a mean comfort score of 7.9. The spaced repetition and ability to work with the clinic during elective time allows regular opportunities to undergo supervised training in a controlled environment and created scheduled retraining opportunities. Future studies should evaluate residents prior to each paracentesis clinic to ascertain if skill decline is occurring at a slower rate.

The inpatient effect of the clinic is also multifocal. Pham and colleagues showed that integrating paracentesis into timely training can reduce paracentesis delay and delays in care.⁷ By increasing the volume of procedures each resident performs and creating a sense of confidence amongst residents, the clinic increases the number of residents able and willing to perform inpatient procedures, thus reducing the number of unnecessary consultations and hospital resources. One of the reasons the paracentesis clinic was started was to allow patients to have scheduled times to remove fluid from their abdomen, thus cutting down on emergency department procedures and unnecessary admissions. Additionally, the benefits of early paracentesis procedural performance by residents and internal medicine physicians have been demonstrated in the literature. A study by Gaetano and colleagues noted that patients undergoing early paracentesis had reduced mortality of 5.5% vs 7.5% in those undergoing late paracentesis.⁸ This study also showed the in-hospital mortality rate was decreased with paracentesis (6.3%) vs without paracentesis (8.9%).⁸ By offering residents a chance to participate in the clinic, we have shown that regular opportunities to perform paracentesis may increase the number of physicians capable of independently practicing, improve procedural competency, and improve patient access to this procedure.

Limitations

Our study was not free of bias and has potential weaknesses. The survey was sent to all current residents who have participated in the paracentesis clinic, but not every resident filled out the survey (55% of all residents across 3 years completed the survey, 68.7% who had done clinic that year completed the survey). There is a possibility that those not signed off avoided doing the survey, but we are unable to confirm this. The survey also depended on resident recall of the number of paracenteses completed or looking at their procedure log. It is possible that some procedures were not documented, changing the true number. Additionally, rating comfortability with procedures is subjective, which may also create a source of potential weakness. Future projects should include a baseline survey for residents, followed by a repeat survey a year later to show changes from baseline competency.

Conclusions

A dedicated paracentesis clinic with internal medicine resident involvement may increase resident paracentesis procedural independence, the number of procedures available and performed, and procedural comfort level.

Competency in paracentesis is an important procedural skill for medical practitioners caring for patients with decompensated liver cirrhosis. Paracentesis is performed to drain ascitic fluid for both diagnosis and/or therapeutic purposes.¹ While this procedure can be performed without the use of ultrasound, it is preferable to use ultrasound to identify an area of fluid that is away from dangerous anatomy including bowel loops, the liver, and spleen. After prepping the area, lidocaine is administered locally. A catheter is then inserted until fluid begins flowing freely. The catheter is connected to a suction canister or collection kit, and the patient is monitored until the flow ceases. Samples can be sent for analysis to determine the etiology of ascites, identify concerns for infection, and more.

Paracentesis is a very common procedure. Barsuk and colleagues noted that between 2010 and 2012, 97,577 procedures were performed across 120 academic medical centers and 290 affiliated hospitals.² Patients undergo paracentesis in a variety of settings including the emergency department, inpatient hospitalizations, and clinics. Some patients may require only 1 paracentesis procedure while others may require it regularly.

Due to the rising need for paracentesis in the Central Texas Veterans Affairs Hospital (CTVAH) in Temple, a paracentesis clinic was started in February 2018. The goal of the paracentesis clinic was multifocal—to reduce hospital admissions, improve access to regularly scheduled procedures, decrease wait times, and increase patient satisfaction.³ Through the CTVAH affiliation with the Texas A&M internal medicine residency program, the paracentesis clinic started involving and training residents on this procedure. Up to 3 residents are on weekly rotation and can perform up to 6 paracentesis procedures in a week. The purpose of this article was to evaluate resident competency in paracentesis after completion of the paracentesis clinic.

Methods

The paracentesis clinic schedules up to 3 patients on Tuesdays and Thursdays between 8 am and noon. All the necessary equipment is readily available and includes the paracentesis kit, lidocaine, sterile gloves, ultrasound, and albumin if needed. Because this procedure is performed at the hospital, direct access to the emergency department is available. Residents are scheduled weekly. Up to 2 residents are scheduled for the paracentesis clinic during their dedicated clinic week. They are expected to practice obtaining consent, performing the procedure, and documenting the encounter under staff supervision. Additionally, 1 or 2 residents participate in the paracentesis clinic as part of an ultrasound elective twice per year. In this elective, they practice ultrasound skills using a simulation and translate that information in the paracentesis clinic while identifying anatomy and performing the paracentesis procedure under staff supervision.

table.png

A survey was sent via email to all categorical internal medicine residents across all 3 program years at the time of data collection. Competency for paracentesis sign-off was defined as completing and logging 5 procedures supervised by a competent physician who confirmed that all portions of the procedure were performed correctly. Residents were also asked to answer questions on a scale from 1 to 10, with 1 representing no confidence and 10 representing strong confidence to practice independently (Table).

We also evaluated the number of procedures performed by internal medicine residents 3 years before the clinic was started in 2015 up to the completion of 2022. The numbers were obtained by examining procedural log data for each year for all internal medicine residents.

Results

Thirty-three residents completed the survey: 10 first-year internal medicine residents (PGY1), 12 second-year residents (PGY2), and 11 third-year residents (PGY3). The mean participation was 4.8 paracentesis sessions per person for the duration of the study. The range of paracentesis procedures performed varied based on PGY year: PGY1s performed 1 to > 10 procedures, PGY2s performed 2 to > 10 procedures, and PGY3s performed 5 to > 10 procedures. Thirty-six percent of residents completed > 10 procedures in the paracentesis clinic; 82% of PGY3s had completed > 10 procedures by December of their third year. Twenty-six residents (79%) were credentialed to perform paracentesis procedures independently after performing > 5 procedures, and 7 residents were not yet cleared for procedural independence.

In the survey, residents rated their comfort with performing paracentesis procedures independently at a mean of 7.9. The mean comfort reported by PGY1s was 7.2, PGY2s was 7.3, and PGY3s was 9.3. Residents also rated their opinion on whether or not the paracentesis clinic adequately prepared them for paracentesis procedural independence; the mean was 8.9 across all residents.

The total number of procedures performed by residents at CTVAH also increased. Starting in 2015, 3 years before the clinic was started, 38 procedures were performed by internal medicine residents, followed by 72 procedures in 2016; 76 in 2017; 58 in 2018; 94 in 2019; 88 in 2020; 136 in 2021; and 188 in 2022.

Paracentesis is a simple but invasive procedure to relieve ascites, often relieving patients’ symptoms, preventing hospital admission, and increasing patient satisfaction.⁴ The CTVAH does not have the capacity to perform outpatient paracentesis effectively in its emergency or radiology departments. Furthermore, the use of the emergency or radiology departments for routine paracentesis may not be feasible due to the acuity of care being provided, as these procedures can be time consuming and can draw away critical resources and time from patients that need emergent care. The paracentesis clinic was then formed to provide veterans access to the procedural care they need, while also preparing residents to ably and confidently perform the procedure independently.

Based on our study, most residents were cleared to independently perform paracentesis procedures across all 3 years, with 79% of residents having completed the required 5 supervised procedures to independently practice. A study assessing unsupervised practice standards showed that paracentesis skill declines as soon as 3 months after training. However, retraining was shown to potentially interrupt this skill decline.⁵ Studies have shown that procedure-driven electives or services significantly improved paracentesis certification rates and total logged procedures, with minimal funding or scheduling changes required.⁶ Our clinic showed a significant increase in the number of procedures logged starting with the minimum of 38 procedures in 2015 and ending with 188 procedures logged at the end of 2022.

By allowing residents to routinely return to the paracentesis clinic across all 3 years, residents were more likely to feel comfortable independently performing the procedure, with residents reporting a mean comfort score of 7.9. The spaced repetition and ability to work with the clinic during elective time allows regular opportunities to undergo supervised training in a controlled environment and created scheduled retraining opportunities. Future studies should evaluate residents prior to each paracentesis clinic to ascertain if skill decline is occurring at a slower rate.

The inpatient effect of the clinic is also multifocal. Pham and colleagues showed that integrating paracentesis into timely training can reduce paracentesis delay and delays in care.⁷ By increasing the volume of procedures each resident performs and creating a sense of confidence amongst residents, the clinic increases the number of residents able and willing to perform inpatient procedures, thus reducing the number of unnecessary consultations and hospital resources. One of the reasons the paracentesis clinic was started was to allow patients to have scheduled times to remove fluid from their abdomen, thus cutting down on emergency department procedures and unnecessary admissions. Additionally, the benefits of early paracentesis procedural performance by residents and internal medicine physicians have been demonstrated in the literature. A study by Gaetano and colleagues noted that patients undergoing early paracentesis had reduced mortality of 5.5% vs 7.5% in those undergoing late paracentesis.⁸ This study also showed the in-hospital mortality rate was decreased with paracentesis (6.3%) vs without paracentesis (8.9%).⁸ By offering residents a chance to participate in the clinic, we have shown that regular opportunities to perform paracentesis may increase the number of physicians capable of independently practicing, improve procedural competency, and improve patient access to this procedure.

Limitations

Our study was not free of bias and has potential weaknesses. The survey was sent to all current residents who have participated in the paracentesis clinic, but not every resident filled out the survey (55% of all residents across 3 years completed the survey, 68.7% who had done clinic that year completed the survey). There is a possibility that those not signed off avoided doing the survey, but we are unable to confirm this. The survey also depended on resident recall of the number of paracenteses completed or looking at their procedure log. It is possible that some procedures were not documented, changing the true number. Additionally, rating comfortability with procedures is subjective, which may also create a source of potential weakness. Future projects should include a baseline survey for residents, followed by a repeat survey a year later to show changes from baseline competency.

Conclusions

A dedicated paracentesis clinic with internal medicine resident involvement may increase resident paracentesis procedural independence, the number of procedures available and performed, and procedural comfort level.

Piperacillin/tazobactam (PTZ) is a combination IV antibiotic comprised of the semisynthetic antipseudomonal β-lactam, piperacillin sodium, and the β-lactamase inhibitor, tazobactam sodium.¹ PTZ is extensively prescribed in the hospital setting for a multitude of infections including but not limited to the US Food and Drug Administration–approved indications: intra-abdominal infection, skin and skin structure infection (SSTI), urinary tract infection (UTI), and pneumonia. Given its broad spectrum of activity and relative safety profile, PTZ is a mainstay of many empiric IV antibiotic regimens. The primary elimination pathway for PTZ is renal excretion, and dosage adjustments are recommended with reduced creatinine clearance. Additionally, PTZ use has been associated with acute renal injury and delayed renal recovery.^1-3

There are various mechanisms through which medications can contribute to acute decomopensated heart failure (ADHF).⁴ These mechanisms include direct cardiotoxicity; negative inotropic, lusitropic, or chronotropic effects; exacerbating hypertension; sodium loading; and drug-drug interactions that limit the benefits of heart failure (HF) medications. One potentially overlooked constituent of PTZ is the sodium content, with the standard formulation containing 65 mg of sodium per gram of piperacillin.^1-3 Furthermore, PTZ must be diluted in 50 to 150 mL of diluent, commonly 0.9% sodium chloride, which can contribute an additional 177 to 531 mg of sodium per dose. PTZ prescribing information advises caution for use in patients with decreased renal, hepatic, and/or cardiac function and notes that geriatric patients, particularly with HF, may be at risk of impaired natriuresis in the setting of large sodium doses.

It is estimated that roughly 6.2 million adults in the United States have HF and prevalence continues to rise.^5,6 Mortality rates after hospitalization due to HF are 20% to 25% at 1 year. Health care expenditures for the management of HF surpass $30 billion per year in the US, with most of this cost attributed to hospitalizations. Consequently, it is important to continue to identify and practice preventative strategies when managing patients with HF.

Methods

This single-center, retrospective, cohort study was conducted at James H. Quillen Veterans Affairs Medical Center (JHQVAMC) in Mountain Home, Tennessee, a 174-bed tertiary medical center. The purpose of this study was to compare the incidence of ADHF in patients who received PTZ vs cefepime (CFP). This project was reviewed by the JHQVAMC Institutional Review Board and deemed exempt as a clinical process improvement operations activity.

The antimicrobial stewardship team at JHQVAMC reviewed the use of PTZ in veterans between January 1, 2018, to December 31, 2019, and compared baseline demographics, history of HF, and outcomes in patients receiving analogous broad-spectrum empiric antibiotic therapy with CFP. Patients were included if they received at least 24 hours of PTZ or CFP. Patients were excluded if they were diagnosed with ADHF before initiation of antibiotic therapy. Patients with ADHF were identified by clinical diagnosis of ADHF documented by the treating clinician and reaffirmed by the study clinician during retrospective chart review. Clinical information used to determine ADHF included clinical presentation, imaging (ie, chest X-ray, echocardiograms), and laboratory parameters, such as B-type natriuretic peptide. The primary endpoint of this study was the incidence of ADHF during the current hospitalization. Secondary endpoints included the length of hospital stay, hospital readmission, and overall mortality. Patient chart reviews were performed using the JHQVAMC Computerized Patient Record System (CPRS).

Statistical Analysis

Analysis was conducted with R Software. Pearson χ² and t tests were used to compare baseline demographics, length of stay, readmission, and mortality. Significance used was α = .05.

table_1.png

A retrospective chart review was performed on 389 veterans. Of the 389, 204 patients received at least 24 hours of PTZ, and 185 patients received CFP. The mean age in both groups was 75 years. Patients in the PTZ group were more likely to have been admitted with the diagnosis of pneumonia (105 vs 49, P < .001). However, 29 patients (15.7%) in the CFP group were admitted with a UTI diagnosis compared with 6 patients (2.9%) in the PTZ group (P < .001) and 62 patients (33.5%) in the CFP group were admitted with a SSTI diagnosis compared with 48 patients (23.5%) in the PTZ group (P = .03). Otherwise, there were no differences between other admitting diagnoses. Additionally, there was no difference in prior history of HF between groups (Table 1).

Twenty-five patients (12.3%) in the PTZ group and 4 patients (2.2%) in the CFP group were subsequently diagnosed with ADHF (P < .001). Hospital readmissions due to HF were higher in the PTZ group compared with the CFP group (11 vs 2, P = .02). Hospital readmission due to other causes was not significantly different between groups. Hospital readmission due to infection occurred in 18 patients who received PTZ and 25 who received CFP (8.8% vs 13.5%, P = .14). Hospital readmission due to any other indication occurred in 24 patients who received PTZ and 24 who received CFP (11.8% vs 13.0%, P = .72). There was no statistically significant difference in all-cause mortality during the associated admission or within 6 months of discharge between groups, with 59 total deaths in the PTZ group and 50 in the CFP group (28.9% vs 27.0%, P = .63).

table_2.png

There was no difference in length of stay outcomes between patients receiving PTZ compared with CFP. Twenty-eight patients in the PTZ group and 20 in the CFP group had a length of stay duration of < 3 days (13.7% vs 10.8%, P = .46). Seventy-three patients in the PTZ group and 76 in the CFP group had a length of stay duration of 4 to 6 days (36.3% vs 41.1%, P = .28). One hundred three patients in the PTZ group and 89 in the CFP group had a length of stay duration ≥ 7 days (50.5% vs 48.1%, P = .78). Table 2 includes a complete overview of primary and secondary endpoint results.

Discussion

The American Heart Association (AHA) lists PTZ as a medication that may cause or exacerbate HF, though no studies have identified a clear association between PTZ use and ADHF.⁴ Sodium restriction is consistently recommended as an important strategy for the prevention of ADHF. Accordingly, PTZ prescribing information and the AHA advise careful consideration with PTZ use in this patient population.^1,4

The specific mechanism responsible for the association of PTZ with cardiac-related adverse outcomes is unclear. It is easy to presume that the sodium content of PTZ is solely responsible; however, other antibiotic regimens not included as agents of concern by the AHA, such as meropenem, can approach similar overall daily sodium amounts.^4,7 Additionally, total sodium and volume can also be contributed by various IV medications and fluids. This study did not evaluate total sodium intake from all sources, but it is notable that this study identified a possible trend toward the risk of ADHF with PTZ use in a routine practice environment. It is reasonable to postulate additional intrinsic properties of PTZ may be contributing to the development of ADHF, such as its association with renal injury possibly resulting in increased fluid retainment and subsequent fluid volume overload.^1,2,4 Other hypothesized mechanisms may include those previously described, such as direct myocardial toxicity; negative inotropic, lusitropic, or chronotropic effects; exacerbating hypertension; and drug-drug interactions that limit the benefits of HF medications, although these have not been overtly associated with PTZ in the literature to date.^4,8

ADHF can present similarly to other acute pulmonary conditions, including pneumonia.^9,10 It is important to acknowledge the challenge this creates for diagnosticians to differentiate between these conditions rapidly and precisely. As a result, this patient population is likely at increased risk of IV antibiotic exposure. Other studies have identified worse outcomes in patients who receive potentially unwarranted IV antibiotics in patients with ADHF.^9,10 The results of this study further emphasize the importance of careful considerate antibiotic selection and overall avoidance of unnecessary antibiotic exposure to limit potential adverse outcomes.

Limitations

There are various limitations to this study. Firstly, no women were included due to the predominantly male population within the Veterans Health Administration system. Secondly, this study was retrospective in design and was therefore limited to the completeness and accuracy of the available data collected. Additionally, this study evaluated any ADHF episode during the associated hospitalization as the primary endpoint. The time to diagnosis of ADHF in relation to PTZ initiation was not evaluated, which may have helped better elucidate this possible association. Furthermore, while a significant statistical difference was identified, the smaller sample size may have limited the ability to accurately identify differences in lower event rate outcomes.

Conclusions

This study identifies an association between PTZ use and significant cardiac-related adverse outcomes, including increased incidence of ADHF and readmission due to HF exacerbation. While more research is needed to define the exact mechanisms by which PTZ may precipitate acute decompensation in patients with HF, it is judicious to consider close monitoring or the avoidance of PTZ when appropriate antibiotic alternatives are available in patients with a known history of HF.

Piperacillin/tazobactam (PTZ) is a combination IV antibiotic comprised of the semisynthetic antipseudomonal β-lactam, piperacillin sodium, and the β-lactamase inhibitor, tazobactam sodium.¹ PTZ is extensively prescribed in the hospital setting for a multitude of infections including but not limited to the US Food and Drug Administration–approved indications: intra-abdominal infection, skin and skin structure infection (SSTI), urinary tract infection (UTI), and pneumonia. Given its broad spectrum of activity and relative safety profile, PTZ is a mainstay of many empiric IV antibiotic regimens. The primary elimination pathway for PTZ is renal excretion, and dosage adjustments are recommended with reduced creatinine clearance. Additionally, PTZ use has been associated with acute renal injury and delayed renal recovery.^1-3

There are various mechanisms through which medications can contribute to acute decomopensated heart failure (ADHF).⁴ These mechanisms include direct cardiotoxicity; negative inotropic, lusitropic, or chronotropic effects; exacerbating hypertension; sodium loading; and drug-drug interactions that limit the benefits of heart failure (HF) medications. One potentially overlooked constituent of PTZ is the sodium content, with the standard formulation containing 65 mg of sodium per gram of piperacillin.^1-3 Furthermore, PTZ must be diluted in 50 to 150 mL of diluent, commonly 0.9% sodium chloride, which can contribute an additional 177 to 531 mg of sodium per dose. PTZ prescribing information advises caution for use in patients with decreased renal, hepatic, and/or cardiac function and notes that geriatric patients, particularly with HF, may be at risk of impaired natriuresis in the setting of large sodium doses.

It is estimated that roughly 6.2 million adults in the United States have HF and prevalence continues to rise.^5,6 Mortality rates after hospitalization due to HF are 20% to 25% at 1 year. Health care expenditures for the management of HF surpass $30 billion per year in the US, with most of this cost attributed to hospitalizations. Consequently, it is important to continue to identify and practice preventative strategies when managing patients with HF.

Methods

This single-center, retrospective, cohort study was conducted at James H. Quillen Veterans Affairs Medical Center (JHQVAMC) in Mountain Home, Tennessee, a 174-bed tertiary medical center. The purpose of this study was to compare the incidence of ADHF in patients who received PTZ vs cefepime (CFP). This project was reviewed by the JHQVAMC Institutional Review Board and deemed exempt as a clinical process improvement operations activity.

The antimicrobial stewardship team at JHQVAMC reviewed the use of PTZ in veterans between January 1, 2018, to December 31, 2019, and compared baseline demographics, history of HF, and outcomes in patients receiving analogous broad-spectrum empiric antibiotic therapy with CFP. Patients were included if they received at least 24 hours of PTZ or CFP. Patients were excluded if they were diagnosed with ADHF before initiation of antibiotic therapy. Patients with ADHF were identified by clinical diagnosis of ADHF documented by the treating clinician and reaffirmed by the study clinician during retrospective chart review. Clinical information used to determine ADHF included clinical presentation, imaging (ie, chest X-ray, echocardiograms), and laboratory parameters, such as B-type natriuretic peptide. The primary endpoint of this study was the incidence of ADHF during the current hospitalization. Secondary endpoints included the length of hospital stay, hospital readmission, and overall mortality. Patient chart reviews were performed using the JHQVAMC Computerized Patient Record System (CPRS).

Statistical Analysis

Analysis was conducted with R Software. Pearson χ² and t tests were used to compare baseline demographics, length of stay, readmission, and mortality. Significance used was α = .05.

table_1.png

A retrospective chart review was performed on 389 veterans. Of the 389, 204 patients received at least 24 hours of PTZ, and 185 patients received CFP. The mean age in both groups was 75 years. Patients in the PTZ group were more likely to have been admitted with the diagnosis of pneumonia (105 vs 49, P < .001). However, 29 patients (15.7%) in the CFP group were admitted with a UTI diagnosis compared with 6 patients (2.9%) in the PTZ group (P < .001) and 62 patients (33.5%) in the CFP group were admitted with a SSTI diagnosis compared with 48 patients (23.5%) in the PTZ group (P = .03). Otherwise, there were no differences between other admitting diagnoses. Additionally, there was no difference in prior history of HF between groups (Table 1).

Twenty-five patients (12.3%) in the PTZ group and 4 patients (2.2%) in the CFP group were subsequently diagnosed with ADHF (P < .001). Hospital readmissions due to HF were higher in the PTZ group compared with the CFP group (11 vs 2, P = .02). Hospital readmission due to other causes was not significantly different between groups. Hospital readmission due to infection occurred in 18 patients who received PTZ and 25 who received CFP (8.8% vs 13.5%, P = .14). Hospital readmission due to any other indication occurred in 24 patients who received PTZ and 24 who received CFP (11.8% vs 13.0%, P = .72). There was no statistically significant difference in all-cause mortality during the associated admission or within 6 months of discharge between groups, with 59 total deaths in the PTZ group and 50 in the CFP group (28.9% vs 27.0%, P = .63).

table_2.png

There was no difference in length of stay outcomes between patients receiving PTZ compared with CFP. Twenty-eight patients in the PTZ group and 20 in the CFP group had a length of stay duration of < 3 days (13.7% vs 10.8%, P = .46). Seventy-three patients in the PTZ group and 76 in the CFP group had a length of stay duration of 4 to 6 days (36.3% vs 41.1%, P = .28). One hundred three patients in the PTZ group and 89 in the CFP group had a length of stay duration ≥ 7 days (50.5% vs 48.1%, P = .78). Table 2 includes a complete overview of primary and secondary endpoint results.

Discussion

The American Heart Association (AHA) lists PTZ as a medication that may cause or exacerbate HF, though no studies have identified a clear association between PTZ use and ADHF.⁴ Sodium restriction is consistently recommended as an important strategy for the prevention of ADHF. Accordingly, PTZ prescribing information and the AHA advise careful consideration with PTZ use in this patient population.^1,4

The specific mechanism responsible for the association of PTZ with cardiac-related adverse outcomes is unclear. It is easy to presume that the sodium content of PTZ is solely responsible; however, other antibiotic regimens not included as agents of concern by the AHA, such as meropenem, can approach similar overall daily sodium amounts.^4,7 Additionally, total sodium and volume can also be contributed by various IV medications and fluids. This study did not evaluate total sodium intake from all sources, but it is notable that this study identified a possible trend toward the risk of ADHF with PTZ use in a routine practice environment. It is reasonable to postulate additional intrinsic properties of PTZ may be contributing to the development of ADHF, such as its association with renal injury possibly resulting in increased fluid retainment and subsequent fluid volume overload.^1,2,4 Other hypothesized mechanisms may include those previously described, such as direct myocardial toxicity; negative inotropic, lusitropic, or chronotropic effects; exacerbating hypertension; and drug-drug interactions that limit the benefits of HF medications, although these have not been overtly associated with PTZ in the literature to date.^4,8

ADHF can present similarly to other acute pulmonary conditions, including pneumonia.^9,10 It is important to acknowledge the challenge this creates for diagnosticians to differentiate between these conditions rapidly and precisely. As a result, this patient population is likely at increased risk of IV antibiotic exposure. Other studies have identified worse outcomes in patients who receive potentially unwarranted IV antibiotics in patients with ADHF.^9,10 The results of this study further emphasize the importance of careful considerate antibiotic selection and overall avoidance of unnecessary antibiotic exposure to limit potential adverse outcomes.

Limitations

There are various limitations to this study. Firstly, no women were included due to the predominantly male population within the Veterans Health Administration system. Secondly, this study was retrospective in design and was therefore limited to the completeness and accuracy of the available data collected. Additionally, this study evaluated any ADHF episode during the associated hospitalization as the primary endpoint. The time to diagnosis of ADHF in relation to PTZ initiation was not evaluated, which may have helped better elucidate this possible association. Furthermore, while a significant statistical difference was identified, the smaller sample size may have limited the ability to accurately identify differences in lower event rate outcomes.

Conclusions

This study identifies an association between PTZ use and significant cardiac-related adverse outcomes, including increased incidence of ADHF and readmission due to HF exacerbation. While more research is needed to define the exact mechanisms by which PTZ may precipitate acute decompensation in patients with HF, it is judicious to consider close monitoring or the avoidance of PTZ when appropriate antibiotic alternatives are available in patients with a known history of HF.

Piperacillin/tazobactam (PTZ) is a combination IV antibiotic comprised of the semisynthetic antipseudomonal β-lactam, piperacillin sodium, and the β-lactamase inhibitor, tazobactam sodium.¹ PTZ is extensively prescribed in the hospital setting for a multitude of infections including but not limited to the US Food and Drug Administration–approved indications: intra-abdominal infection, skin and skin structure infection (SSTI), urinary tract infection (UTI), and pneumonia. Given its broad spectrum of activity and relative safety profile, PTZ is a mainstay of many empiric IV antibiotic regimens. The primary elimination pathway for PTZ is renal excretion, and dosage adjustments are recommended with reduced creatinine clearance. Additionally, PTZ use has been associated with acute renal injury and delayed renal recovery.^1-3

There are various mechanisms through which medications can contribute to acute decomopensated heart failure (ADHF).⁴ These mechanisms include direct cardiotoxicity; negative inotropic, lusitropic, or chronotropic effects; exacerbating hypertension; sodium loading; and drug-drug interactions that limit the benefits of heart failure (HF) medications. One potentially overlooked constituent of PTZ is the sodium content, with the standard formulation containing 65 mg of sodium per gram of piperacillin.^1-3 Furthermore, PTZ must be diluted in 50 to 150 mL of diluent, commonly 0.9% sodium chloride, which can contribute an additional 177 to 531 mg of sodium per dose. PTZ prescribing information advises caution for use in patients with decreased renal, hepatic, and/or cardiac function and notes that geriatric patients, particularly with HF, may be at risk of impaired natriuresis in the setting of large sodium doses.

It is estimated that roughly 6.2 million adults in the United States have HF and prevalence continues to rise.^5,6 Mortality rates after hospitalization due to HF are 20% to 25% at 1 year. Health care expenditures for the management of HF surpass $30 billion per year in the US, with most of this cost attributed to hospitalizations. Consequently, it is important to continue to identify and practice preventative strategies when managing patients with HF.

Methods

This single-center, retrospective, cohort study was conducted at James H. Quillen Veterans Affairs Medical Center (JHQVAMC) in Mountain Home, Tennessee, a 174-bed tertiary medical center. The purpose of this study was to compare the incidence of ADHF in patients who received PTZ vs cefepime (CFP). This project was reviewed by the JHQVAMC Institutional Review Board and deemed exempt as a clinical process improvement operations activity.

The antimicrobial stewardship team at JHQVAMC reviewed the use of PTZ in veterans between January 1, 2018, to December 31, 2019, and compared baseline demographics, history of HF, and outcomes in patients receiving analogous broad-spectrum empiric antibiotic therapy with CFP. Patients were included if they received at least 24 hours of PTZ or CFP. Patients were excluded if they were diagnosed with ADHF before initiation of antibiotic therapy. Patients with ADHF were identified by clinical diagnosis of ADHF documented by the treating clinician and reaffirmed by the study clinician during retrospective chart review. Clinical information used to determine ADHF included clinical presentation, imaging (ie, chest X-ray, echocardiograms), and laboratory parameters, such as B-type natriuretic peptide. The primary endpoint of this study was the incidence of ADHF during the current hospitalization. Secondary endpoints included the length of hospital stay, hospital readmission, and overall mortality. Patient chart reviews were performed using the JHQVAMC Computerized Patient Record System (CPRS).

Statistical Analysis

Analysis was conducted with R Software. Pearson χ² and t tests were used to compare baseline demographics, length of stay, readmission, and mortality. Significance used was α = .05.

table_1.png

A retrospective chart review was performed on 389 veterans. Of the 389, 204 patients received at least 24 hours of PTZ, and 185 patients received CFP. The mean age in both groups was 75 years. Patients in the PTZ group were more likely to have been admitted with the diagnosis of pneumonia (105 vs 49, P < .001). However, 29 patients (15.7%) in the CFP group were admitted with a UTI diagnosis compared with 6 patients (2.9%) in the PTZ group (P < .001) and 62 patients (33.5%) in the CFP group were admitted with a SSTI diagnosis compared with 48 patients (23.5%) in the PTZ group (P = .03). Otherwise, there were no differences between other admitting diagnoses. Additionally, there was no difference in prior history of HF between groups (Table 1).

Twenty-five patients (12.3%) in the PTZ group and 4 patients (2.2%) in the CFP group were subsequently diagnosed with ADHF (P < .001). Hospital readmissions due to HF were higher in the PTZ group compared with the CFP group (11 vs 2, P = .02). Hospital readmission due to other causes was not significantly different between groups. Hospital readmission due to infection occurred in 18 patients who received PTZ and 25 who received CFP (8.8% vs 13.5%, P = .14). Hospital readmission due to any other indication occurred in 24 patients who received PTZ and 24 who received CFP (11.8% vs 13.0%, P = .72). There was no statistically significant difference in all-cause mortality during the associated admission or within 6 months of discharge between groups, with 59 total deaths in the PTZ group and 50 in the CFP group (28.9% vs 27.0%, P = .63).

table_2.png

There was no difference in length of stay outcomes between patients receiving PTZ compared with CFP. Twenty-eight patients in the PTZ group and 20 in the CFP group had a length of stay duration of < 3 days (13.7% vs 10.8%, P = .46). Seventy-three patients in the PTZ group and 76 in the CFP group had a length of stay duration of 4 to 6 days (36.3% vs 41.1%, P = .28). One hundred three patients in the PTZ group and 89 in the CFP group had a length of stay duration ≥ 7 days (50.5% vs 48.1%, P = .78). Table 2 includes a complete overview of primary and secondary endpoint results.

Discussion

The American Heart Association (AHA) lists PTZ as a medication that may cause or exacerbate HF, though no studies have identified a clear association between PTZ use and ADHF.⁴ Sodium restriction is consistently recommended as an important strategy for the prevention of ADHF. Accordingly, PTZ prescribing information and the AHA advise careful consideration with PTZ use in this patient population.^1,4

The specific mechanism responsible for the association of PTZ with cardiac-related adverse outcomes is unclear. It is easy to presume that the sodium content of PTZ is solely responsible; however, other antibiotic regimens not included as agents of concern by the AHA, such as meropenem, can approach similar overall daily sodium amounts.^4,7 Additionally, total sodium and volume can also be contributed by various IV medications and fluids. This study did not evaluate total sodium intake from all sources, but it is notable that this study identified a possible trend toward the risk of ADHF with PTZ use in a routine practice environment. It is reasonable to postulate additional intrinsic properties of PTZ may be contributing to the development of ADHF, such as its association with renal injury possibly resulting in increased fluid retainment and subsequent fluid volume overload.^1,2,4 Other hypothesized mechanisms may include those previously described, such as direct myocardial toxicity; negative inotropic, lusitropic, or chronotropic effects; exacerbating hypertension; and drug-drug interactions that limit the benefits of HF medications, although these have not been overtly associated with PTZ in the literature to date.^4,8

ADHF can present similarly to other acute pulmonary conditions, including pneumonia.^9,10 It is important to acknowledge the challenge this creates for diagnosticians to differentiate between these conditions rapidly and precisely. As a result, this patient population is likely at increased risk of IV antibiotic exposure. Other studies have identified worse outcomes in patients who receive potentially unwarranted IV antibiotics in patients with ADHF.^9,10 The results of this study further emphasize the importance of careful considerate antibiotic selection and overall avoidance of unnecessary antibiotic exposure to limit potential adverse outcomes.

Limitations

There are various limitations to this study. Firstly, no women were included due to the predominantly male population within the Veterans Health Administration system. Secondly, this study was retrospective in design and was therefore limited to the completeness and accuracy of the available data collected. Additionally, this study evaluated any ADHF episode during the associated hospitalization as the primary endpoint. The time to diagnosis of ADHF in relation to PTZ initiation was not evaluated, which may have helped better elucidate this possible association. Furthermore, while a significant statistical difference was identified, the smaller sample size may have limited the ability to accurately identify differences in lower event rate outcomes.

Conclusions

This study identifies an association between PTZ use and significant cardiac-related adverse outcomes, including increased incidence of ADHF and readmission due to HF exacerbation. While more research is needed to define the exact mechanisms by which PTZ may precipitate acute decompensation in patients with HF, it is judicious to consider close monitoring or the avoidance of PTZ when appropriate antibiotic alternatives are available in patients with a known history of HF.

To the Editor:

Female pattern hair loss (FPHL), or androgenetic alopecia (AGA), is the most common form of alopecia worldwide and is characterized by a reduction of hair follicles spent in the anagen phase of growth as well as progressive terminal hair loss.¹ It is caused by an excessive response to androgens and leads to the characteristic distribution of hair loss in both sexes. Studies have shown a notable association between AGA and markers of metabolic syndrome such as dyslipidemia, insulin resistance, and obesity in age- and sex-matched controls.^2,3 However, research describing the relationship between AGA severity and these markers is scarce.

To understand the relationship between FPHL severity and abnormal cholesterol levels, we performed a retrospective chart review of patients diagnosed with FPHL at a specialty alopecia clinic from June 2022 to December 2022. Patient age and age at onset of FPHL were collected. The severity of FPHL was measured using the Sinclair scale (score range, 1–5) and unidentifiable patient photographs. Laboratory values were collected; abnormal cholesterol was defined by the American Heart Association as having a low-density lipoprotein cholesterol (LDL-C) level of 100 mg/dL or higher.⁴ Finally, data on medication use were noted to understand patient treatment status (Table).

CT113001033_e_Table.jpg

We identified 54 female patients with FPHL with an average age of 59 years (range, 34–80 years). Thirty-three females (61.11%) had a normal LDL-C level and 21 (38.89%) had an abnormal level. The mean (SD) LDL-C level was 66.02 (15.20) mg/dL (range, 29–92 mg/dL) in the group with normal levels and 138.81 (29.90) mg/dL (range, 100–193 mg/dL) in the group with abnormal levels. Patients with abnormal LDL-C had significantly higher Sinclair scale scores compared to those with normal levels (2.43 vs 1.91; P=.01). There were no significant differences in patient age (58.71 vs 59.70 years; P=.39), age at onset of AGA (47.75 vs 47.65 years; P=.49), history of polycystic ovary syndrome (9.52% vs 6.06%; P=.64), or statin use (38.09% vs 36.36%; P=.89) between patients with abnormal and normal LDL-C levels, respectively. There also were no significant differences in ferritin (96.42 vs 91.54 ng/mL; P=.40), vitamin D (42.35 vs 48.96 ng/mL; P=.09), or hemoglobin A_1c levels (5.60 ng/mL vs 5.38 ng/mL; P=.06)—variables that could have confounded this relationship. Triglycerides were within reference range in both groups (121.36 vs 116.16 mg/dL; P=.32), while total cholesterol was mildly elevated in both groups but not significantly different (213.19 vs 201.21 mg/dL; P=.13). Use of hair loss treatments such as topical minoxidil (14.29% vs 21.21%; P=.53), oral low-dose minoxidil (57.14% vs 66.67%; P=.48), oral spironolactone (47.62% vs 57.58%; P=.47), and platelet-rich plasma injections (47.62% vs 27.27%; P=.90) were not significantly different across both groups.

The data suggest a significant (P<.05) association between abnormal LDL-C and hair loss severity in FPHL patients. Our study was limited by its small sample size and lack of causality; however, it coincides with and reiterates the findings established in the literature. The mechanism of the association between hyperlipidemia and AGA is not well understood but is thought to stem from the homology between cholesterol and androgens. Increased cholesterol release from dermal adipocytes and subsequent absorption into hair follicle cell populations may increase hair follicle steroidogenesis, thereby accelerating the anagen-catagen transition and inducing AGA. Alternatively, impaired cholesterol homeostasis may disrupt normal hair follicle cycling by interrupting signaling pathways in follicle proliferation and differentiation.⁵ Adequate control and monitoring of LDL-C levels may be important, particularly in patients with more severe FPHL.

To the Editor:

Female pattern hair loss (FPHL), or androgenetic alopecia (AGA), is the most common form of alopecia worldwide and is characterized by a reduction of hair follicles spent in the anagen phase of growth as well as progressive terminal hair loss.¹ It is caused by an excessive response to androgens and leads to the characteristic distribution of hair loss in both sexes. Studies have shown a notable association between AGA and markers of metabolic syndrome such as dyslipidemia, insulin resistance, and obesity in age- and sex-matched controls.^2,3 However, research describing the relationship between AGA severity and these markers is scarce.

To understand the relationship between FPHL severity and abnormal cholesterol levels, we performed a retrospective chart review of patients diagnosed with FPHL at a specialty alopecia clinic from June 2022 to December 2022. Patient age and age at onset of FPHL were collected. The severity of FPHL was measured using the Sinclair scale (score range, 1–5) and unidentifiable patient photographs. Laboratory values were collected; abnormal cholesterol was defined by the American Heart Association as having a low-density lipoprotein cholesterol (LDL-C) level of 100 mg/dL or higher.⁴ Finally, data on medication use were noted to understand patient treatment status (Table).

CT113001033_e_Table.jpg

We identified 54 female patients with FPHL with an average age of 59 years (range, 34–80 years). Thirty-three females (61.11%) had a normal LDL-C level and 21 (38.89%) had an abnormal level. The mean (SD) LDL-C level was 66.02 (15.20) mg/dL (range, 29–92 mg/dL) in the group with normal levels and 138.81 (29.90) mg/dL (range, 100–193 mg/dL) in the group with abnormal levels. Patients with abnormal LDL-C had significantly higher Sinclair scale scores compared to those with normal levels (2.43 vs 1.91; P=.01). There were no significant differences in patient age (58.71 vs 59.70 years; P=.39), age at onset of AGA (47.75 vs 47.65 years; P=.49), history of polycystic ovary syndrome (9.52% vs 6.06%; P=.64), or statin use (38.09% vs 36.36%; P=.89) between patients with abnormal and normal LDL-C levels, respectively. There also were no significant differences in ferritin (96.42 vs 91.54 ng/mL; P=.40), vitamin D (42.35 vs 48.96 ng/mL; P=.09), or hemoglobin A_1c levels (5.60 ng/mL vs 5.38 ng/mL; P=.06)—variables that could have confounded this relationship. Triglycerides were within reference range in both groups (121.36 vs 116.16 mg/dL; P=.32), while total cholesterol was mildly elevated in both groups but not significantly different (213.19 vs 201.21 mg/dL; P=.13). Use of hair loss treatments such as topical minoxidil (14.29% vs 21.21%; P=.53), oral low-dose minoxidil (57.14% vs 66.67%; P=.48), oral spironolactone (47.62% vs 57.58%; P=.47), and platelet-rich plasma injections (47.62% vs 27.27%; P=.90) were not significantly different across both groups.

The data suggest a significant (P<.05) association between abnormal LDL-C and hair loss severity in FPHL patients. Our study was limited by its small sample size and lack of causality; however, it coincides with and reiterates the findings established in the literature. The mechanism of the association between hyperlipidemia and AGA is not well understood but is thought to stem from the homology between cholesterol and androgens. Increased cholesterol release from dermal adipocytes and subsequent absorption into hair follicle cell populations may increase hair follicle steroidogenesis, thereby accelerating the anagen-catagen transition and inducing AGA. Alternatively, impaired cholesterol homeostasis may disrupt normal hair follicle cycling by interrupting signaling pathways in follicle proliferation and differentiation.⁵ Adequate control and monitoring of LDL-C levels may be important, particularly in patients with more severe FPHL.

To the Editor:

Female pattern hair loss (FPHL), or androgenetic alopecia (AGA), is the most common form of alopecia worldwide and is characterized by a reduction of hair follicles spent in the anagen phase of growth as well as progressive terminal hair loss.¹ It is caused by an excessive response to androgens and leads to the characteristic distribution of hair loss in both sexes. Studies have shown a notable association between AGA and markers of metabolic syndrome such as dyslipidemia, insulin resistance, and obesity in age- and sex-matched controls.^2,3 However, research describing the relationship between AGA severity and these markers is scarce.

To understand the relationship between FPHL severity and abnormal cholesterol levels, we performed a retrospective chart review of patients diagnosed with FPHL at a specialty alopecia clinic from June 2022 to December 2022. Patient age and age at onset of FPHL were collected. The severity of FPHL was measured using the Sinclair scale (score range, 1–5) and unidentifiable patient photographs. Laboratory values were collected; abnormal cholesterol was defined by the American Heart Association as having a low-density lipoprotein cholesterol (LDL-C) level of 100 mg/dL or higher.⁴ Finally, data on medication use were noted to understand patient treatment status (Table).

CT113001033_e_Table.jpg

We identified 54 female patients with FPHL with an average age of 59 years (range, 34–80 years). Thirty-three females (61.11%) had a normal LDL-C level and 21 (38.89%) had an abnormal level. The mean (SD) LDL-C level was 66.02 (15.20) mg/dL (range, 29–92 mg/dL) in the group with normal levels and 138.81 (29.90) mg/dL (range, 100–193 mg/dL) in the group with abnormal levels. Patients with abnormal LDL-C had significantly higher Sinclair scale scores compared to those with normal levels (2.43 vs 1.91; P=.01). There were no significant differences in patient age (58.71 vs 59.70 years; P=.39), age at onset of AGA (47.75 vs 47.65 years; P=.49), history of polycystic ovary syndrome (9.52% vs 6.06%; P=.64), or statin use (38.09% vs 36.36%; P=.89) between patients with abnormal and normal LDL-C levels, respectively. There also were no significant differences in ferritin (96.42 vs 91.54 ng/mL; P=.40), vitamin D (42.35 vs 48.96 ng/mL; P=.09), or hemoglobin A_1c levels (5.60 ng/mL vs 5.38 ng/mL; P=.06)—variables that could have confounded this relationship. Triglycerides were within reference range in both groups (121.36 vs 116.16 mg/dL; P=.32), while total cholesterol was mildly elevated in both groups but not significantly different (213.19 vs 201.21 mg/dL; P=.13). Use of hair loss treatments such as topical minoxidil (14.29% vs 21.21%; P=.53), oral low-dose minoxidil (57.14% vs 66.67%; P=.48), oral spironolactone (47.62% vs 57.58%; P=.47), and platelet-rich plasma injections (47.62% vs 27.27%; P=.90) were not significantly different across both groups.

The data suggest a significant (P<.05) association between abnormal LDL-C and hair loss severity in FPHL patients. Our study was limited by its small sample size and lack of causality; however, it coincides with and reiterates the findings established in the literature. The mechanism of the association between hyperlipidemia and AGA is not well understood but is thought to stem from the homology between cholesterol and androgens. Increased cholesterol release from dermal adipocytes and subsequent absorption into hair follicle cell populations may increase hair follicle steroidogenesis, thereby accelerating the anagen-catagen transition and inducing AGA. Alternatively, impaired cholesterol homeostasis may disrupt normal hair follicle cycling by interrupting signaling pathways in follicle proliferation and differentiation.⁵ Adequate control and monitoring of LDL-C levels may be important, particularly in patients with more severe FPHL.

To the Editor:

Partial or total nail plate excisions commonly are used for the treatment of onychocryptosis and nail spicules. Procedures involving the nail unit require advanced technical skills to achieve optimal functional and aesthetic outcomes, avoid complications, and minimize health care costs. Data on the frequency of nail plate excisions performed by dermatologists and their relative frequency compared to other medical providers are limited. The objective of our study was to analyze trends in nail excision practice patterns among medical providers in the United States.

A retrospective analysis on nail excisions using the Current Procedural Terminology (CPT) code 11750 (excision of nail and nail matrix, partial or complete [eg, ingrown or deformed nail] for permanent removal), which is distinct from code 11755 (biopsy of nail unit [eg, plate, bed, matrix, hyponychium, proximal and lateral nail folds][separate procedure]), was performed using data from the Medicare Provider Utilization and Payment Database 2012-2017.^1,2 This file also is used by Peck et al³ in an article submitted to the Journal of the American Podiatric Medical Association and currently under consideration for publication. Procedures were recorded by year and provider type—dermatologist, podiatrist, physician assistant (PA)/nurse practitioner (NP), nondermatologist physician—and subcategorized by provider specialty (Table). Practice locations subcategorized by provider type were mapped using Tableau Software (Salesforce)(Figure). Descriptive statistics including number of providers, mean and median excisions per provider, and minimum/maximum nail excisions were calculated (Table). Practice types of PAs/NPs and specialization of nondermatologist physicians were determined using provider name, identification number, and practice address. This study did not require institutional review board review, as only publicly available data were utilized in our analysis.

CT113001022_e_Table.jpg

A total of 6936 podiatrists, 58 nondermatologist physicians, 25 PAs/NPs, and 4 dermatologists performed 10 or more nail excisions annually under CPT code 11750 from January 2012 to December 2017 with annual means of 31, 31, 25, and 34, respectively (Table). No PAs/NPs included in the dataset worked in dermatology practices during the study period. Physician assistants and NPs most often practiced in podiatry and family medicine (FM) settings (both 40% [10/25]). Nondermatologist physicians most often specialized in FM (40% [23/58])(Table). The greatest number of providers practiced in 3 of the 4 most-populous states: California, Texas, and Florida; the fewest number practiced in 3 of the 10 least-populous states: Alaska, Hawaii, and Vermont. Vermont, Wyoming, and North Dakota—3 of the 5 least-populous states—had the fewest practitioners among the contiguous United States (Figure).

Hill_nail_figure.jpg

Our study showed that from January 2012 to December 2017, fewer dermatologists performed nail excisions than any other provider type (0.06%, 4 dermatologists of 7023 total providers), and dermatologists performed 1734-fold fewer nail excisions than podiatrists (99%, 6936 podiatrists of 7023 total providers). Only dermatologists practicing in California, Georgia, Indiana, and Oklahoma performed nail excisions. Conversely, podiatrists were more geographically distributed across the United States and other territories, with representation in all 50 states as well as the District of Columbia, Puerto Rico, and Guam.

Reasons for these large discrepancies in practice between dermatologists and other providers likely are multifactorial, encompassing a lack of emphasis on nail procedures in dermatology training, patient perception of the scope of dermatologic practice, and nail excision reimbursement patterns. Most dermatologists likely lack experience in performing nail procedures. The Accreditation Council for Graduate Medical Education requirements mandate that dermatology residents observe or perform 3 nail procedures over 3 years of residency, including 1 that may be performed on a human cadaver.⁴ In contrast, podiatry trainees must gain competency in toenail avulsion (both partial and complete), participate in anesthesia workshops, and become proficient in administering lower extremity blocks by the end of their training.⁵ Therefore, incorporating aspects of podiatric surgical training into dermatology residency requirements may increase the competency and comfort of dermatologists in performing nail excisions and practicing as nail experts as attending physicians.

It is likely that US patients do not perceive dermatologists as nail specialists and instead primarily consult podiatrists or FM and/or internal medicine physicians for treatment; for example, nail disease was one of the least common reasons for consulting a dermatologist (5%) in a German nationwide survey-based study (N=1015).⁶ Therefore, increased efforts are needed to educate the general public about the expertise of dermatologists in the diagnosis and management of nail conditions.

Reimbursement also may be a barrier to dermatologists performing nail procedures as part of their scope of practice; for example, in a retrospective study of nail biopsies using the Medicare Provider Utilization and Payment Database, there was no statistically significant difference in reimbursements for nail biopsies vs skin biopsies from 2012 to 2017 (P=0.69).⁷ Similar to nail biopsies, nail excisions typically are much more time consuming and technically demanding than skin biopsies, which may discourage dermatologists from routinely performing nail excision procedures.

Our study is subject to a number of limitations. The data reflected only US-based practice patterns and may not be applicable to nail procedures globally. There also is the potential for miscoding of procedures in the Medicare database. The data included only Part B Medicare fee-for-service and excludes non-Medicare insured, uninsured, and self-pay patients, as well as aggregated records from 10 or fewer Medicare beneficiaries.

Dermatologists rarely perform nail excisions and perform fewer nail excisions than any other provider type in the United States. There currently is an unmet need for comprehensive nail surgery education in US-based dermatology residency programs. We hope that our study fosters interdisciplinary collegiality and training between podiatrists and dermatologists and promotes expanded access to care across the United States to serve patients with nail disorders.

To the Editor:

Partial or total nail plate excisions commonly are used for the treatment of onychocryptosis and nail spicules. Procedures involving the nail unit require advanced technical skills to achieve optimal functional and aesthetic outcomes, avoid complications, and minimize health care costs. Data on the frequency of nail plate excisions performed by dermatologists and their relative frequency compared to other medical providers are limited. The objective of our study was to analyze trends in nail excision practice patterns among medical providers in the United States.

A retrospective analysis on nail excisions using the Current Procedural Terminology (CPT) code 11750 (excision of nail and nail matrix, partial or complete [eg, ingrown or deformed nail] for permanent removal), which is distinct from code 11755 (biopsy of nail unit [eg, plate, bed, matrix, hyponychium, proximal and lateral nail folds][separate procedure]), was performed using data from the Medicare Provider Utilization and Payment Database 2012-2017.^1,2 This file also is used by Peck et al³ in an article submitted to the Journal of the American Podiatric Medical Association and currently under consideration for publication. Procedures were recorded by year and provider type—dermatologist, podiatrist, physician assistant (PA)/nurse practitioner (NP), nondermatologist physician—and subcategorized by provider specialty (Table). Practice locations subcategorized by provider type were mapped using Tableau Software (Salesforce)(Figure). Descriptive statistics including number of providers, mean and median excisions per provider, and minimum/maximum nail excisions were calculated (Table). Practice types of PAs/NPs and specialization of nondermatologist physicians were determined using provider name, identification number, and practice address. This study did not require institutional review board review, as only publicly available data were utilized in our analysis.

CT113001022_e_Table.jpg

A total of 6936 podiatrists, 58 nondermatologist physicians, 25 PAs/NPs, and 4 dermatologists performed 10 or more nail excisions annually under CPT code 11750 from January 2012 to December 2017 with annual means of 31, 31, 25, and 34, respectively (Table). No PAs/NPs included in the dataset worked in dermatology practices during the study period. Physician assistants and NPs most often practiced in podiatry and family medicine (FM) settings (both 40% [10/25]). Nondermatologist physicians most often specialized in FM (40% [23/58])(Table). The greatest number of providers practiced in 3 of the 4 most-populous states: California, Texas, and Florida; the fewest number practiced in 3 of the 10 least-populous states: Alaska, Hawaii, and Vermont. Vermont, Wyoming, and North Dakota—3 of the 5 least-populous states—had the fewest practitioners among the contiguous United States (Figure).

Hill_nail_figure.jpg

Our study showed that from January 2012 to December 2017, fewer dermatologists performed nail excisions than any other provider type (0.06%, 4 dermatologists of 7023 total providers), and dermatologists performed 1734-fold fewer nail excisions than podiatrists (99%, 6936 podiatrists of 7023 total providers). Only dermatologists practicing in California, Georgia, Indiana, and Oklahoma performed nail excisions. Conversely, podiatrists were more geographically distributed across the United States and other territories, with representation in all 50 states as well as the District of Columbia, Puerto Rico, and Guam.

Reasons for these large discrepancies in practice between dermatologists and other providers likely are multifactorial, encompassing a lack of emphasis on nail procedures in dermatology training, patient perception of the scope of dermatologic practice, and nail excision reimbursement patterns. Most dermatologists likely lack experience in performing nail procedures. The Accreditation Council for Graduate Medical Education requirements mandate that dermatology residents observe or perform 3 nail procedures over 3 years of residency, including 1 that may be performed on a human cadaver.⁴ In contrast, podiatry trainees must gain competency in toenail avulsion (both partial and complete), participate in anesthesia workshops, and become proficient in administering lower extremity blocks by the end of their training.⁵ Therefore, incorporating aspects of podiatric surgical training into dermatology residency requirements may increase the competency and comfort of dermatologists in performing nail excisions and practicing as nail experts as attending physicians.

It is likely that US patients do not perceive dermatologists as nail specialists and instead primarily consult podiatrists or FM and/or internal medicine physicians for treatment; for example, nail disease was one of the least common reasons for consulting a dermatologist (5%) in a German nationwide survey-based study (N=1015).⁶ Therefore, increased efforts are needed to educate the general public about the expertise of dermatologists in the diagnosis and management of nail conditions.

Reimbursement also may be a barrier to dermatologists performing nail procedures as part of their scope of practice; for example, in a retrospective study of nail biopsies using the Medicare Provider Utilization and Payment Database, there was no statistically significant difference in reimbursements for nail biopsies vs skin biopsies from 2012 to 2017 (P=0.69).⁷ Similar to nail biopsies, nail excisions typically are much more time consuming and technically demanding than skin biopsies, which may discourage dermatologists from routinely performing nail excision procedures.

Our study is subject to a number of limitations. The data reflected only US-based practice patterns and may not be applicable to nail procedures globally. There also is the potential for miscoding of procedures in the Medicare database. The data included only Part B Medicare fee-for-service and excludes non-Medicare insured, uninsured, and self-pay patients, as well as aggregated records from 10 or fewer Medicare beneficiaries.

Dermatologists rarely perform nail excisions and perform fewer nail excisions than any other provider type in the United States. There currently is an unmet need for comprehensive nail surgery education in US-based dermatology residency programs. We hope that our study fosters interdisciplinary collegiality and training between podiatrists and dermatologists and promotes expanded access to care across the United States to serve patients with nail disorders.

To the Editor:

Partial or total nail plate excisions commonly are used for the treatment of onychocryptosis and nail spicules. Procedures involving the nail unit require advanced technical skills to achieve optimal functional and aesthetic outcomes, avoid complications, and minimize health care costs. Data on the frequency of nail plate excisions performed by dermatologists and their relative frequency compared to other medical providers are limited. The objective of our study was to analyze trends in nail excision practice patterns among medical providers in the United States.

A retrospective analysis on nail excisions using the Current Procedural Terminology (CPT) code 11750 (excision of nail and nail matrix, partial or complete [eg, ingrown or deformed nail] for permanent removal), which is distinct from code 11755 (biopsy of nail unit [eg, plate, bed, matrix, hyponychium, proximal and lateral nail folds][separate procedure]), was performed using data from the Medicare Provider Utilization and Payment Database 2012-2017.^1,2 This file also is used by Peck et al³ in an article submitted to the Journal of the American Podiatric Medical Association and currently under consideration for publication. Procedures were recorded by year and provider type—dermatologist, podiatrist, physician assistant (PA)/nurse practitioner (NP), nondermatologist physician—and subcategorized by provider specialty (Table). Practice locations subcategorized by provider type were mapped using Tableau Software (Salesforce)(Figure). Descriptive statistics including number of providers, mean and median excisions per provider, and minimum/maximum nail excisions were calculated (Table). Practice types of PAs/NPs and specialization of nondermatologist physicians were determined using provider name, identification number, and practice address. This study did not require institutional review board review, as only publicly available data were utilized in our analysis.

CT113001022_e_Table.jpg

A total of 6936 podiatrists, 58 nondermatologist physicians, 25 PAs/NPs, and 4 dermatologists performed 10 or more nail excisions annually under CPT code 11750 from January 2012 to December 2017 with annual means of 31, 31, 25, and 34, respectively (Table). No PAs/NPs included in the dataset worked in dermatology practices during the study period. Physician assistants and NPs most often practiced in podiatry and family medicine (FM) settings (both 40% [10/25]). Nondermatologist physicians most often specialized in FM (40% [23/58])(Table). The greatest number of providers practiced in 3 of the 4 most-populous states: California, Texas, and Florida; the fewest number practiced in 3 of the 10 least-populous states: Alaska, Hawaii, and Vermont. Vermont, Wyoming, and North Dakota—3 of the 5 least-populous states—had the fewest practitioners among the contiguous United States (Figure).

Hill_nail_figure.jpg

Our study showed that from January 2012 to December 2017, fewer dermatologists performed nail excisions than any other provider type (0.06%, 4 dermatologists of 7023 total providers), and dermatologists performed 1734-fold fewer nail excisions than podiatrists (99%, 6936 podiatrists of 7023 total providers). Only dermatologists practicing in California, Georgia, Indiana, and Oklahoma performed nail excisions. Conversely, podiatrists were more geographically distributed across the United States and other territories, with representation in all 50 states as well as the District of Columbia, Puerto Rico, and Guam.

Reasons for these large discrepancies in practice between dermatologists and other providers likely are multifactorial, encompassing a lack of emphasis on nail procedures in dermatology training, patient perception of the scope of dermatologic practice, and nail excision reimbursement patterns. Most dermatologists likely lack experience in performing nail procedures. The Accreditation Council for Graduate Medical Education requirements mandate that dermatology residents observe or perform 3 nail procedures over 3 years of residency, including 1 that may be performed on a human cadaver.⁴ In contrast, podiatry trainees must gain competency in toenail avulsion (both partial and complete), participate in anesthesia workshops, and become proficient in administering lower extremity blocks by the end of their training.⁵ Therefore, incorporating aspects of podiatric surgical training into dermatology residency requirements may increase the competency and comfort of dermatologists in performing nail excisions and practicing as nail experts as attending physicians.

It is likely that US patients do not perceive dermatologists as nail specialists and instead primarily consult podiatrists or FM and/or internal medicine physicians for treatment; for example, nail disease was one of the least common reasons for consulting a dermatologist (5%) in a German nationwide survey-based study (N=1015).⁶ Therefore, increased efforts are needed to educate the general public about the expertise of dermatologists in the diagnosis and management of nail conditions.

Reimbursement also may be a barrier to dermatologists performing nail procedures as part of their scope of practice; for example, in a retrospective study of nail biopsies using the Medicare Provider Utilization and Payment Database, there was no statistically significant difference in reimbursements for nail biopsies vs skin biopsies from 2012 to 2017 (P=0.69).⁷ Similar to nail biopsies, nail excisions typically are much more time consuming and technically demanding than skin biopsies, which may discourage dermatologists from routinely performing nail excision procedures.

Our study is subject to a number of limitations. The data reflected only US-based practice patterns and may not be applicable to nail procedures globally. There also is the potential for miscoding of procedures in the Medicare database. The data included only Part B Medicare fee-for-service and excludes non-Medicare insured, uninsured, and self-pay patients, as well as aggregated records from 10 or fewer Medicare beneficiaries.

Dermatologists rarely perform nail excisions and perform fewer nail excisions than any other provider type in the United States. There currently is an unmet need for comprehensive nail surgery education in US-based dermatology residency programs. We hope that our study fosters interdisciplinary collegiality and training between podiatrists and dermatologists and promotes expanded access to care across the United States to serve patients with nail disorders.