Steven A. Dosh, MD

Antibiotics are widely prescribed for acute respiratory infections (ARIs) of viral origin.^1-3 ARIs include the clinical syndromes of the common cold (upper respiratory infection [URI]), acute bronchitis, pneumonia, pharyngitis, sinusitis, and otitis media. Some cases of pneumonia, pharyngitis (streptococcal), sinusitis, and otitis media are bacterial infections that may improve more rapidly with antibiotic treatment, but viruses cause most of these infections.

Because of the predominantly viral etiology of most ARIs, antibiotics are of little or no benefit,^4-7 and the current widespread use of antibiotics for ARIs has significant costs. In addition to the substantial monetary costs,¹ overuse of antimicrobials for ARIs has contributed to the emergence and spread of resistant bacteria. Data from Iceland suggest that community-wide consumption of antimicrobials is positively related to the nasopharyngeal carriage of penicillin-resistant pneumococci.⁸ And data from Finland indicate that after a nationwide reduction in the use of macrolide antibiotics for outpatient therapy, there was a significant decline in the prevalence of erythromycin-resistant group A streptococci.⁹

Why do physicians persist in prescribing antibiotics for ARIs of likely viral origin despite a lack of evidence of effectiveness? Two explanations are plausible. First, because of the overlapping constellations of signs and symptoms of different ARIs, physicians may have some difficulty distinguishing viral respiratory infections from those that may improve more rapidly with antibiotic treatment, such as acute sinusitis or acute bronchitis. However, there is, at best, equivocal evidence from randomized trials for a modest treatment effect of antibiotics for sinusitis and a minimal effect for bronchitis.^7,10-12 Second, many patients expect an antibiotic. Most clinicians believe antibiotics present minimal risk and do not want to risk patient loyalty, so they accede to patient requests, even when an antibiotic is not indicated.¹³ There may be other reasons unrelated to the clinical diagnosis or patient expectation, such as physician and patient concern that the patient’s condition may worsen without an antibiotic.

Most clinicians, researchers, and public health officials agree that antibiotics are overprescribed for ARIs. However, changing clinician and patient behaviors and beliefs is difficult. Admonitions and exhortations to stop prescribing are likely to be minimally effective. Educational interventions linked to performance assessment may be most effective.¹⁴ The interventions should be targeted to change specific beliefs and to correct specific knowledge gaps of both clinicians and patients. However, the factors that trigger antibiotic prescribing are not well understood. If it is patient expectation, what factors underlie that expectation? If it is physician knowledge, what are the specific gaps? Are there other factors that trigger antibiotic prescribing?

Objective

The purpose of our study was to determine the factors associated with antibiotic prescribing for ARIs in a prospective observational study. Our hypothesis was that antibiotic prescribing could be predicted from a limited number of signs and symptoms and that patients’ expectations and clinicians’ beliefs about patient expectations for antibiotics would be among the predictors of antibiotic prescribing. Our research question was: Among patients presenting to outpatient settings with symptoms compatible with ARIs, what factors-including symptoms, signs, patient characteristics, diagnoses, patient expectations, and so forth-are associated with antibiotic prescribing?

Methods

Setting

All 15 practices of the Upper Peninsula Research Network (UPRNet), a rural family practice research network in Northern Michigan, participated in our study. In January and February 1998 we asked each clinician in an UPRNet practice to enroll at least 20 consecutive patients 4 years and older who presented to the office with symptoms of an ARI. Patients were identified and invited to participate by 31 family physicians, 3 pediatricians, 5 internists, 16 physician assistants, and 3 nurse practitioners. We instructed the clinicians and their staffs not to enroll patients with a chief complaint of pharyngitis or ear pain. We excluded from the analyses patients who had a primary or secondary diagnosis of pneumonia, pharyngitis, or otitis media, because antibiotics are currently thought to be indicated for many of those cases. We included patients with the clinical diagnoses of nonspecific URI, bronchitis, and sinusitis because of the large overlap in signs and symptoms and the lack of precision in clinical diagnosis for those 3 conditions. Also, assigning a diagnosis of bronchitis or sinusitis may legitimize the use of an antibiotic for patients whose symptoms are otherwise identical to those of patients assigned the diagnosis of URI. Patients reporting symptoms for more than 30 days were excluded.

Interventions and Main Measures

Each patient (or parent) participating in the study was asked to complete a questionnaire detailing the patient’s symptoms, duration of symptoms, expected treatment for the current illness, history of previous respiratory illness and prior treatment, and smoking history. The nurse caring for the patient noted the patient’s temperature, location of visit (regular office or urgent care), and the type of clinician (physician, physician assistant, or nurse practitioner) who saw the patient. Finally, the clinician completed a questionnaire identifying the patient’s chief complaint, history of lung disease, physical findings, treatment, the clinician’s belief about the patient’s expectations, diagnosis, and secondary factors affecting the decision to give antibiotics, if given. Secondary factors were reasons for prescribing an antibiotic that were not directly related to the diagnosis. The list of options for secondary factors included: patient expected an antibiotic, patient requested an antibiotic, patient leaving town, patient not improving, patient getting worse, patient sick too long, patient smokes, patient has chronic lung disease, patient has comorbidity, and patient is extremely ill. Finally, we asked the clinicians, “What is the likelihood of an adverse outcome if this patient does not receive an antibiotic today?” Response options were: very unlikely, unlikely, moderately likely, and very likely.

Our study was approved by the Michigan State University Committee on Research Involving Human Subjects, and we obtained informed consent from all subjects.

All analyses were done using SPSS 7.0 (SPSS Inc, Chicago, Ill). We calculated frequency distributions of individual variables and used chi-square analysis to assess univariate associations between each variable and both antibiotic prescribing and patient expectation for an antibiotic prescription. Continuous variables were converted to categorical data for the univariate analyses but treated as continuous data for the logistic regression. Because we measured secondary factors only in encounters in which an antibiotic was prescribed, that analysis was descriptive only and was not included in the logistic regression models.

We developed 2 logistic regression models, 1 to identify the factors independently associated with antibiotic prescribing (antibiotic prescribing model) and 1 to identify the factors independently associated with patients’ expectations of receiving an antibiotic prescription (patient expectation model).

For the antibiotic prescribing model, dichotomous variables that had a statistically significant univariate association with prescriptions for antibiotics and all continuous variables were included in the first step of the logistic regression modeling. We used the following method in an attempt to avoid type 1 errors due to the many variables we tested: All variables with a P value of .05 or less on univariate analysis were included in the initial logistic regression model, then removed sequentially, in order of ascending statistical significance, if they did not affect the odds ratio of the remaining model by at least 20%. Only main effects were assessed. We calculated odds ratios and their 95% confidence intervals from the final antibiotic prescribing model.

The patient expectation model was developed in a similar fashion. For the logistic regression, continuous variables with clinically implausible outlying values (0%) or missing values (7%) were replaced with the mean value if normally distributed and with the median value if skewed. We excluded from analysis records missing dichotomous variables necessary for logistic regression analysis (1%).

Results

Nine hundred twenty-eight patients completed questionnaires; 11 declined participation. We eliminated from analysis 222 patients with primary diagnoses of otitis media, pneumonia, pharyngitis, allergy, or other. An additional 113 were eliminated because their questionnaires were lacking essential information, they were out of the age range, or they did not give informed consent. The remaining 593 had a main diagnosis of bronchitis, sinusitis, or nonspecific URI. There were 482 records entered in the analysis: 80 were eliminated because of a secondary diagnosis of otitis media (38), pharyngitis (33), or pneumonia (9), and 31 were eliminated because the patient reported being sick longer than 30 days. Mean temperature and mean days sick were assigned respectively to 17 records (3%) missing temperature and 21 records (4%) missing number of days sick.

The subjects were primarily adult women (67%) from a predominantly white (95%) population. Physicians saw most of the patients (67%), and most visits were in a physician’s office (81%) rather than an urgent care facility. Most patients were being seen for the current illness for the first time (91%) and had been sick for no more than 14 days (86%).

The principal diagnoses assigned by the clinicians were: sinusitis for 176 patients (37%), nonspecific URI in 167 patients (34%), and bronchitis in 139 patients (29%). Antibiotics were prescribed for 319 (66%) of all diagnoses, 111 (80%) of bronchitis diagnoses, 173 (98%) of sinusitis diagnoses, and 35 (21%) of nonspecific URI diagnoses.

Antibiotic Prescribing Model

Univariate analysis revealed that those patients 18 years and older (odds ratio [OR] =3.4; 95% confidence interval [CI], 2.0-5.7), seen in urgent care clinics (OR=1.7; 95% CI, 1.0-2.9), evaluated by a nurse practitioner (OR=5.9; 95% CI, 1.8-19.9), and sick for more than 14 days (OR=4.3; 95% CI=2.0-9.3) were more likely to be treated with an antibiotic Table 1. Antibiotic prescribing was much more frequent for those patients judged by the clinician to be moderately or very likely to have an adverse outcome compared with those judged unlikely or very unlikely to have an adverse outcome if not treated with an antibiotic (OR=61.8; 95% CI, 24.6-155). Table 2 and Table 3 summarize symptoms and signs associated with antibiotic prescribing. Antibiotic prescribing was modestly associated with a number of symptoms and signs Table 4 and much more common for patients with the physical findings of discolored nasal discharge (OR=20.4), sinus tenderness (OR=15.9), a wet cough (OR=8.6), and rales or rhonchi (OR=7.7). After controlling for confounding with logistic regression, 5 variables were independently associated with antibiotic prescribing: sinus tenderness (OR=20.0), rales/ronchi (OR=19.9), yellow/green nasal discharge (OR=11.7), and postnasal drainage (OR=3.1), and negatively associated with clear nasal discharge (OR=0.3).

Patient Expectations Model

An antibiotic was expected by 290 (60%) of the patients with a URI, bronchitis, or sinusitis. A majority of the patients (68%) had received an antibiotic for a similar illness in the past, and 299 (86%) of these patients believed the antibiotic helped. Patients expecting an antibiotic were more likely to receive one Table 5. Patients were more likely to expect an antibiotic who reported green or yellow nasal discharge (OR=2.0; 95% CI, 1.2-3.2), cough productive of yellow or green phlegm (OR=1.7; 95% CI, 1.1-2.5), pain over the eye or cheek (OR=1.6; 95% CI, 1.1-2.4), or who were helped by an antibiotic for similar symptoms in the past (OR=4.5; 95% CI, 2.9-6.9). However, only “helped by an antibiotic for similar symptoms in the past” remained significantly and independently associated with patient expectation after controlling for confounding by logistic regression.

Clinicians believed 298 patients (62%) expected an antibiotic, and they were significantly more likely to prescribe an antibiotic for those patients (OR=4.7; 95% CI, 3.2-7.1) on univariate analysis. The clinicians’ accuracy in predicting which patients expected antibiotics Table 6 was only fair (Cohen’s k=0.21). After controlling for confounding with logistic regression, patient expectation and clinician belief that the patient expected an antibiotic were not independent predictors of antibiotic prescribing. There were no significant differences in the demographic, symptomatic, and diagnostic variables between those records included in the analysis and those records excluded because of incomplete data, implausible data, or lack of informed consent. An analysis restricted to those records with every field completed and with plausible values did not change the final logistic regression models.

Secondary Factors

Clinicians reported that secondary factors contributed to the decision to prescribe an antibiotic for 95% of the patients. The most common reasons were: patient not improving (21%), patient getting worse (19%), and patient has been sick too long (19%). Other secondary reasons included: patient smokes (11%), patient has chronic lung disease (7%), patient expected an antibiotic (6%), patient was quite ill (5%), patient had a comorbid condition other than chronic lung disease (4%), patient was leaving town (2%), and patient requested an antibiotic (2%).

Discussion

Our study suggests that antibiotic prescribing for patients with the clinical diagnoses of URI, bronchitis, or sinusitis is determined by 2 general sets of factors: a small number of physical findings and clinicians’ perceptions of the clinical course of the illness. The physical findings are rales, rhonchi, sinus tenderness, postnasal drainage, purulent nasal discharge, and clear nasal discharge (a negative association). These findings were associated with antibiotic prescribing regardless of the diagnosis the clinician assigned. Notably absent from this list is cough, which was reported by more than 70% of patients assigned any of the 3 diagnoses. Apparently cough alone, which is the chief complaint in many cases of acute respiratory infection, is often not sufficient for clinicians to prescribe antibiotics. In our study, clinicians readily admitted that factors other than the diagnoses contributed to their decisions to prescribe antibiotics. In 95% of the cases, they listed a secondary reason. The majority of these reasons (59%) had to do with the clinical course of the illness: patient not improving (21%), patient getting worse (19%), and patient being sick too long (19%). For patients with the diagnosis of nonspecific URI, clinicians listed one of these reasons for 83% of cases. Whether any of these are legitimate reasons for prescribing antibiotics for acute respiratory infections has not been investigated. Similarly, when clinicians believed that patients were likely to have an adverse outcome if not treated with an antibiotic, they prescribed antibiotics in 98% of cases. This most likely reflects clinicians’ belief that acute sinusitis and acute bronchitis do not improve without antibiotics, a belief largely unsupported by the medical literature.

Our findings regarding patient expectations for antibiotics were surprising. On univariate analysis, we found that patient expectation for an antibiotic was a predictor of antibiotic prescribing. However, contrary to widely held beliefs, patient expectation of an antibiotic prescription was not an independent predictor of antibiotic prescribing in our study. After adjusting for other related factors in the logistic regression model, antibiotics were prescribed as frequently for those not expecting one as for those expecting one. We believe that patient expectation may be a surrogate marker for “having received an antibiotic in the past for a similar illness and had good results.” Thus, patient expectation is driven by past physician behavior in prescribing antibiotics. Other researchers have noted an association between patient expectation and antibiotic prescribing,^15,16 but they did not adjust for other factors associated with both patient expectation and antibiotic prescribing.

Also, in the absence of the 5 physical findings noted above (rales or rhonchi, sinus tenderness, postnasal drainage, purulent nasal discharge, and clear nasal discharge), patient expectation of an antibiotic and receiving an antibiotic in the past for a similar illness does not appear to affect clinicians’ decisions to prescribe an antibiotic. Are any of these physical findings good reason for prescribing an antibiotic? Other than rales, which may be associated with pneumonia, there is no good empiric evidence that purulent nasal discharge, postnasal discharge, rhonchi, or sinus tenderness distinguish between viral and bacterial illnesses.¹² Purulent nasal discharge is commonly thought to indicate bacterial infection, but this is not supported by clinical or laboratory research.⁶ Purulent nasal discharge does increase the odds of sinusitis, but it is unclear whether uncomplicated sinusitis diagnosed on clinical grounds improves more rapidly with antibiotic treatment.^17-20 The Centers for Disease Control and Prevention guidelines for treatment of sinusitis in children recommend no antibiotic unless symptoms are severe or have been present for more than 10 days.²¹

Inadequate Physician Knowledge

It has been suggested that inadequate knowledge of the nature of viral illnesses accounts for inappropriate antibiotic prescribing practices.⁵ Our findings support this assertion. The high percentage of cases of sinusitis (37%) and relatively low percentage of antibiotic prescriptions for nonspecific URI (21%) suggest that clinicians may have assigned the diagnoses of sinusitis to justify an antibiotic prescription, as has been suggested by Vinson and Lutz¹⁶ for bronchitis. However, this would suggest that clinicians intentionally falsified physical findings to justify antibiotic prescribing to satisfy patient demand or for other unidentified reasons. It is more likely that sinus tenderness was the basis for a diagnosis of sinusitis, rales or rhonchi were the basis for a diagnosis of bronchitis, and these diagnoses were the justification for prescribing an antibiotic.

Before an all-out war on antibiotics is waged, however, a very large randomized controlled trial of antibiotic treatment of acute respiratory infections that would allow subgroup analysis by the factors we have described may be necessary to determine whether some cases of ARI are antibiotic responsive. Although current evidence does not strongly support use of antibiotics for URIs, acute bronchitis, or acute sinusitis, sufficient evidence does not exist to refute a clinically significant effect in certain subgroups of these patients.

We found that the clinicians in our network depend on the presence of discolored nasal drainage, postnasal drainage, rales or rhonchi, sinus tenderness, and clear nasal drainage to decide whether to prescribe an antibiotic for patients with a URI, acute bronchitis, and acute sinusitis. In making their decisions they also considered the clinical course of the illness and a subjective assessment of the likelihood a patient will get worse without an antibiotic. We found that patients expect an antibiotic if they have had a similar illness in the past and had improved with an antibiotic; yet patient expectation alone does not appear to be a sufficient reason for clinicians to prescribe antibiotics for acute respiratory infections. These new findings may be of benefit to those designing interventions to reduce inappropriate use of antibiotics for acute respiratory infections.

Acknowledgments

We express our gratitude to Teresa L. Ettenhofer, Renee Lauscher, and Meghana Kasetty, MD, of Escanaba and the clinicians and office staffs of the following Upper Peninsula Research Network practices: Alcona Health Center in Lincoln, Alpena Medical Arts in Alpena, Burns Clinic in St. Ignace, Doctors Park Family Physicians in Escanaba, East Jordan Family Health Center in East Jordan, Ewen Medical Center in Ewen, OSF Medical Group in Escanaba and Gladstone, Gwinn Medical Center in Gwinn, Mackinac Straits Primary Care Clinic in St. Ignace, Marquette Internal Medicine Associates in Marquette, Marquette Medical Clinic in Iron River, and Northern Michigan Health Services in Houghton Lake.

Antibiotics are widely prescribed for acute respiratory infections (ARIs) of viral origin.^1-3 ARIs include the clinical syndromes of the common cold (upper respiratory infection [URI]), acute bronchitis, pneumonia, pharyngitis, sinusitis, and otitis media. Some cases of pneumonia, pharyngitis (streptococcal), sinusitis, and otitis media are bacterial infections that may improve more rapidly with antibiotic treatment, but viruses cause most of these infections.

Because of the predominantly viral etiology of most ARIs, antibiotics are of little or no benefit,^4-7 and the current widespread use of antibiotics for ARIs has significant costs. In addition to the substantial monetary costs,¹ overuse of antimicrobials for ARIs has contributed to the emergence and spread of resistant bacteria. Data from Iceland suggest that community-wide consumption of antimicrobials is positively related to the nasopharyngeal carriage of penicillin-resistant pneumococci.⁸ And data from Finland indicate that after a nationwide reduction in the use of macrolide antibiotics for outpatient therapy, there was a significant decline in the prevalence of erythromycin-resistant group A streptococci.⁹

Why do physicians persist in prescribing antibiotics for ARIs of likely viral origin despite a lack of evidence of effectiveness? Two explanations are plausible. First, because of the overlapping constellations of signs and symptoms of different ARIs, physicians may have some difficulty distinguishing viral respiratory infections from those that may improve more rapidly with antibiotic treatment, such as acute sinusitis or acute bronchitis. However, there is, at best, equivocal evidence from randomized trials for a modest treatment effect of antibiotics for sinusitis and a minimal effect for bronchitis.^7,10-12 Second, many patients expect an antibiotic. Most clinicians believe antibiotics present minimal risk and do not want to risk patient loyalty, so they accede to patient requests, even when an antibiotic is not indicated.¹³ There may be other reasons unrelated to the clinical diagnosis or patient expectation, such as physician and patient concern that the patient’s condition may worsen without an antibiotic.

Most clinicians, researchers, and public health officials agree that antibiotics are overprescribed for ARIs. However, changing clinician and patient behaviors and beliefs is difficult. Admonitions and exhortations to stop prescribing are likely to be minimally effective. Educational interventions linked to performance assessment may be most effective.¹⁴ The interventions should be targeted to change specific beliefs and to correct specific knowledge gaps of both clinicians and patients. However, the factors that trigger antibiotic prescribing are not well understood. If it is patient expectation, what factors underlie that expectation? If it is physician knowledge, what are the specific gaps? Are there other factors that trigger antibiotic prescribing?

Objective

The purpose of our study was to determine the factors associated with antibiotic prescribing for ARIs in a prospective observational study. Our hypothesis was that antibiotic prescribing could be predicted from a limited number of signs and symptoms and that patients’ expectations and clinicians’ beliefs about patient expectations for antibiotics would be among the predictors of antibiotic prescribing. Our research question was: Among patients presenting to outpatient settings with symptoms compatible with ARIs, what factors-including symptoms, signs, patient characteristics, diagnoses, patient expectations, and so forth-are associated with antibiotic prescribing?

Methods

Setting

All 15 practices of the Upper Peninsula Research Network (UPRNet), a rural family practice research network in Northern Michigan, participated in our study. In January and February 1998 we asked each clinician in an UPRNet practice to enroll at least 20 consecutive patients 4 years and older who presented to the office with symptoms of an ARI. Patients were identified and invited to participate by 31 family physicians, 3 pediatricians, 5 internists, 16 physician assistants, and 3 nurse practitioners. We instructed the clinicians and their staffs not to enroll patients with a chief complaint of pharyngitis or ear pain. We excluded from the analyses patients who had a primary or secondary diagnosis of pneumonia, pharyngitis, or otitis media, because antibiotics are currently thought to be indicated for many of those cases. We included patients with the clinical diagnoses of nonspecific URI, bronchitis, and sinusitis because of the large overlap in signs and symptoms and the lack of precision in clinical diagnosis for those 3 conditions. Also, assigning a diagnosis of bronchitis or sinusitis may legitimize the use of an antibiotic for patients whose symptoms are otherwise identical to those of patients assigned the diagnosis of URI. Patients reporting symptoms for more than 30 days were excluded.

Interventions and Main Measures

Each patient (or parent) participating in the study was asked to complete a questionnaire detailing the patient’s symptoms, duration of symptoms, expected treatment for the current illness, history of previous respiratory illness and prior treatment, and smoking history. The nurse caring for the patient noted the patient’s temperature, location of visit (regular office or urgent care), and the type of clinician (physician, physician assistant, or nurse practitioner) who saw the patient. Finally, the clinician completed a questionnaire identifying the patient’s chief complaint, history of lung disease, physical findings, treatment, the clinician’s belief about the patient’s expectations, diagnosis, and secondary factors affecting the decision to give antibiotics, if given. Secondary factors were reasons for prescribing an antibiotic that were not directly related to the diagnosis. The list of options for secondary factors included: patient expected an antibiotic, patient requested an antibiotic, patient leaving town, patient not improving, patient getting worse, patient sick too long, patient smokes, patient has chronic lung disease, patient has comorbidity, and patient is extremely ill. Finally, we asked the clinicians, “What is the likelihood of an adverse outcome if this patient does not receive an antibiotic today?” Response options were: very unlikely, unlikely, moderately likely, and very likely.

Our study was approved by the Michigan State University Committee on Research Involving Human Subjects, and we obtained informed consent from all subjects.

All analyses were done using SPSS 7.0 (SPSS Inc, Chicago, Ill). We calculated frequency distributions of individual variables and used chi-square analysis to assess univariate associations between each variable and both antibiotic prescribing and patient expectation for an antibiotic prescription. Continuous variables were converted to categorical data for the univariate analyses but treated as continuous data for the logistic regression. Because we measured secondary factors only in encounters in which an antibiotic was prescribed, that analysis was descriptive only and was not included in the logistic regression models.

We developed 2 logistic regression models, 1 to identify the factors independently associated with antibiotic prescribing (antibiotic prescribing model) and 1 to identify the factors independently associated with patients’ expectations of receiving an antibiotic prescription (patient expectation model).

For the antibiotic prescribing model, dichotomous variables that had a statistically significant univariate association with prescriptions for antibiotics and all continuous variables were included in the first step of the logistic regression modeling. We used the following method in an attempt to avoid type 1 errors due to the many variables we tested: All variables with a P value of .05 or less on univariate analysis were included in the initial logistic regression model, then removed sequentially, in order of ascending statistical significance, if they did not affect the odds ratio of the remaining model by at least 20%. Only main effects were assessed. We calculated odds ratios and their 95% confidence intervals from the final antibiotic prescribing model.

The patient expectation model was developed in a similar fashion. For the logistic regression, continuous variables with clinically implausible outlying values (0%) or missing values (7%) were replaced with the mean value if normally distributed and with the median value if skewed. We excluded from analysis records missing dichotomous variables necessary for logistic regression analysis (1%).

Results

Nine hundred twenty-eight patients completed questionnaires; 11 declined participation. We eliminated from analysis 222 patients with primary diagnoses of otitis media, pneumonia, pharyngitis, allergy, or other. An additional 113 were eliminated because their questionnaires were lacking essential information, they were out of the age range, or they did not give informed consent. The remaining 593 had a main diagnosis of bronchitis, sinusitis, or nonspecific URI. There were 482 records entered in the analysis: 80 were eliminated because of a secondary diagnosis of otitis media (38), pharyngitis (33), or pneumonia (9), and 31 were eliminated because the patient reported being sick longer than 30 days. Mean temperature and mean days sick were assigned respectively to 17 records (3%) missing temperature and 21 records (4%) missing number of days sick.

The subjects were primarily adult women (67%) from a predominantly white (95%) population. Physicians saw most of the patients (67%), and most visits were in a physician’s office (81%) rather than an urgent care facility. Most patients were being seen for the current illness for the first time (91%) and had been sick for no more than 14 days (86%).

The principal diagnoses assigned by the clinicians were: sinusitis for 176 patients (37%), nonspecific URI in 167 patients (34%), and bronchitis in 139 patients (29%). Antibiotics were prescribed for 319 (66%) of all diagnoses, 111 (80%) of bronchitis diagnoses, 173 (98%) of sinusitis diagnoses, and 35 (21%) of nonspecific URI diagnoses.

Antibiotic Prescribing Model

Univariate analysis revealed that those patients 18 years and older (odds ratio [OR] =3.4; 95% confidence interval [CI], 2.0-5.7), seen in urgent care clinics (OR=1.7; 95% CI, 1.0-2.9), evaluated by a nurse practitioner (OR=5.9; 95% CI, 1.8-19.9), and sick for more than 14 days (OR=4.3; 95% CI=2.0-9.3) were more likely to be treated with an antibiotic Table 1. Antibiotic prescribing was much more frequent for those patients judged by the clinician to be moderately or very likely to have an adverse outcome compared with those judged unlikely or very unlikely to have an adverse outcome if not treated with an antibiotic (OR=61.8; 95% CI, 24.6-155). Table 2 and Table 3 summarize symptoms and signs associated with antibiotic prescribing. Antibiotic prescribing was modestly associated with a number of symptoms and signs Table 4 and much more common for patients with the physical findings of discolored nasal discharge (OR=20.4), sinus tenderness (OR=15.9), a wet cough (OR=8.6), and rales or rhonchi (OR=7.7). After controlling for confounding with logistic regression, 5 variables were independently associated with antibiotic prescribing: sinus tenderness (OR=20.0), rales/ronchi (OR=19.9), yellow/green nasal discharge (OR=11.7), and postnasal drainage (OR=3.1), and negatively associated with clear nasal discharge (OR=0.3).

Patient Expectations Model

An antibiotic was expected by 290 (60%) of the patients with a URI, bronchitis, or sinusitis. A majority of the patients (68%) had received an antibiotic for a similar illness in the past, and 299 (86%) of these patients believed the antibiotic helped. Patients expecting an antibiotic were more likely to receive one Table 5. Patients were more likely to expect an antibiotic who reported green or yellow nasal discharge (OR=2.0; 95% CI, 1.2-3.2), cough productive of yellow or green phlegm (OR=1.7; 95% CI, 1.1-2.5), pain over the eye or cheek (OR=1.6; 95% CI, 1.1-2.4), or who were helped by an antibiotic for similar symptoms in the past (OR=4.5; 95% CI, 2.9-6.9). However, only “helped by an antibiotic for similar symptoms in the past” remained significantly and independently associated with patient expectation after controlling for confounding by logistic regression.

Clinicians believed 298 patients (62%) expected an antibiotic, and they were significantly more likely to prescribe an antibiotic for those patients (OR=4.7; 95% CI, 3.2-7.1) on univariate analysis. The clinicians’ accuracy in predicting which patients expected antibiotics Table 6 was only fair (Cohen’s k=0.21). After controlling for confounding with logistic regression, patient expectation and clinician belief that the patient expected an antibiotic were not independent predictors of antibiotic prescribing. There were no significant differences in the demographic, symptomatic, and diagnostic variables between those records included in the analysis and those records excluded because of incomplete data, implausible data, or lack of informed consent. An analysis restricted to those records with every field completed and with plausible values did not change the final logistic regression models.

Secondary Factors

Clinicians reported that secondary factors contributed to the decision to prescribe an antibiotic for 95% of the patients. The most common reasons were: patient not improving (21%), patient getting worse (19%), and patient has been sick too long (19%). Other secondary reasons included: patient smokes (11%), patient has chronic lung disease (7%), patient expected an antibiotic (6%), patient was quite ill (5%), patient had a comorbid condition other than chronic lung disease (4%), patient was leaving town (2%), and patient requested an antibiotic (2%).

Discussion

Our study suggests that antibiotic prescribing for patients with the clinical diagnoses of URI, bronchitis, or sinusitis is determined by 2 general sets of factors: a small number of physical findings and clinicians’ perceptions of the clinical course of the illness. The physical findings are rales, rhonchi, sinus tenderness, postnasal drainage, purulent nasal discharge, and clear nasal discharge (a negative association). These findings were associated with antibiotic prescribing regardless of the diagnosis the clinician assigned. Notably absent from this list is cough, which was reported by more than 70% of patients assigned any of the 3 diagnoses. Apparently cough alone, which is the chief complaint in many cases of acute respiratory infection, is often not sufficient for clinicians to prescribe antibiotics. In our study, clinicians readily admitted that factors other than the diagnoses contributed to their decisions to prescribe antibiotics. In 95% of the cases, they listed a secondary reason. The majority of these reasons (59%) had to do with the clinical course of the illness: patient not improving (21%), patient getting worse (19%), and patient being sick too long (19%). For patients with the diagnosis of nonspecific URI, clinicians listed one of these reasons for 83% of cases. Whether any of these are legitimate reasons for prescribing antibiotics for acute respiratory infections has not been investigated. Similarly, when clinicians believed that patients were likely to have an adverse outcome if not treated with an antibiotic, they prescribed antibiotics in 98% of cases. This most likely reflects clinicians’ belief that acute sinusitis and acute bronchitis do not improve without antibiotics, a belief largely unsupported by the medical literature.

Our findings regarding patient expectations for antibiotics were surprising. On univariate analysis, we found that patient expectation for an antibiotic was a predictor of antibiotic prescribing. However, contrary to widely held beliefs, patient expectation of an antibiotic prescription was not an independent predictor of antibiotic prescribing in our study. After adjusting for other related factors in the logistic regression model, antibiotics were prescribed as frequently for those not expecting one as for those expecting one. We believe that patient expectation may be a surrogate marker for “having received an antibiotic in the past for a similar illness and had good results.” Thus, patient expectation is driven by past physician behavior in prescribing antibiotics. Other researchers have noted an association between patient expectation and antibiotic prescribing,^15,16 but they did not adjust for other factors associated with both patient expectation and antibiotic prescribing.

Also, in the absence of the 5 physical findings noted above (rales or rhonchi, sinus tenderness, postnasal drainage, purulent nasal discharge, and clear nasal discharge), patient expectation of an antibiotic and receiving an antibiotic in the past for a similar illness does not appear to affect clinicians’ decisions to prescribe an antibiotic. Are any of these physical findings good reason for prescribing an antibiotic? Other than rales, which may be associated with pneumonia, there is no good empiric evidence that purulent nasal discharge, postnasal discharge, rhonchi, or sinus tenderness distinguish between viral and bacterial illnesses.¹² Purulent nasal discharge is commonly thought to indicate bacterial infection, but this is not supported by clinical or laboratory research.⁶ Purulent nasal discharge does increase the odds of sinusitis, but it is unclear whether uncomplicated sinusitis diagnosed on clinical grounds improves more rapidly with antibiotic treatment.^17-20 The Centers for Disease Control and Prevention guidelines for treatment of sinusitis in children recommend no antibiotic unless symptoms are severe or have been present for more than 10 days.²¹

Inadequate Physician Knowledge

It has been suggested that inadequate knowledge of the nature of viral illnesses accounts for inappropriate antibiotic prescribing practices.⁵ Our findings support this assertion. The high percentage of cases of sinusitis (37%) and relatively low percentage of antibiotic prescriptions for nonspecific URI (21%) suggest that clinicians may have assigned the diagnoses of sinusitis to justify an antibiotic prescription, as has been suggested by Vinson and Lutz¹⁶ for bronchitis. However, this would suggest that clinicians intentionally falsified physical findings to justify antibiotic prescribing to satisfy patient demand or for other unidentified reasons. It is more likely that sinus tenderness was the basis for a diagnosis of sinusitis, rales or rhonchi were the basis for a diagnosis of bronchitis, and these diagnoses were the justification for prescribing an antibiotic.

Before an all-out war on antibiotics is waged, however, a very large randomized controlled trial of antibiotic treatment of acute respiratory infections that would allow subgroup analysis by the factors we have described may be necessary to determine whether some cases of ARI are antibiotic responsive. Although current evidence does not strongly support use of antibiotics for URIs, acute bronchitis, or acute sinusitis, sufficient evidence does not exist to refute a clinically significant effect in certain subgroups of these patients.

We found that the clinicians in our network depend on the presence of discolored nasal drainage, postnasal drainage, rales or rhonchi, sinus tenderness, and clear nasal drainage to decide whether to prescribe an antibiotic for patients with a URI, acute bronchitis, and acute sinusitis. In making their decisions they also considered the clinical course of the illness and a subjective assessment of the likelihood a patient will get worse without an antibiotic. We found that patients expect an antibiotic if they have had a similar illness in the past and had improved with an antibiotic; yet patient expectation alone does not appear to be a sufficient reason for clinicians to prescribe antibiotics for acute respiratory infections. These new findings may be of benefit to those designing interventions to reduce inappropriate use of antibiotics for acute respiratory infections.

Acknowledgments

We express our gratitude to Teresa L. Ettenhofer, Renee Lauscher, and Meghana Kasetty, MD, of Escanaba and the clinicians and office staffs of the following Upper Peninsula Research Network practices: Alcona Health Center in Lincoln, Alpena Medical Arts in Alpena, Burns Clinic in St. Ignace, Doctors Park Family Physicians in Escanaba, East Jordan Family Health Center in East Jordan, Ewen Medical Center in Ewen, OSF Medical Group in Escanaba and Gladstone, Gwinn Medical Center in Gwinn, Mackinac Straits Primary Care Clinic in St. Ignace, Marquette Internal Medicine Associates in Marquette, Marquette Medical Clinic in Iron River, and Northern Michigan Health Services in Houghton Lake.

Antibiotics are widely prescribed for acute respiratory infections (ARIs) of viral origin.^1-3 ARIs include the clinical syndromes of the common cold (upper respiratory infection [URI]), acute bronchitis, pneumonia, pharyngitis, sinusitis, and otitis media. Some cases of pneumonia, pharyngitis (streptococcal), sinusitis, and otitis media are bacterial infections that may improve more rapidly with antibiotic treatment, but viruses cause most of these infections.

Because of the predominantly viral etiology of most ARIs, antibiotics are of little or no benefit,^4-7 and the current widespread use of antibiotics for ARIs has significant costs. In addition to the substantial monetary costs,¹ overuse of antimicrobials for ARIs has contributed to the emergence and spread of resistant bacteria. Data from Iceland suggest that community-wide consumption of antimicrobials is positively related to the nasopharyngeal carriage of penicillin-resistant pneumococci.⁸ And data from Finland indicate that after a nationwide reduction in the use of macrolide antibiotics for outpatient therapy, there was a significant decline in the prevalence of erythromycin-resistant group A streptococci.⁹

Why do physicians persist in prescribing antibiotics for ARIs of likely viral origin despite a lack of evidence of effectiveness? Two explanations are plausible. First, because of the overlapping constellations of signs and symptoms of different ARIs, physicians may have some difficulty distinguishing viral respiratory infections from those that may improve more rapidly with antibiotic treatment, such as acute sinusitis or acute bronchitis. However, there is, at best, equivocal evidence from randomized trials for a modest treatment effect of antibiotics for sinusitis and a minimal effect for bronchitis.^7,10-12 Second, many patients expect an antibiotic. Most clinicians believe antibiotics present minimal risk and do not want to risk patient loyalty, so they accede to patient requests, even when an antibiotic is not indicated.¹³ There may be other reasons unrelated to the clinical diagnosis or patient expectation, such as physician and patient concern that the patient’s condition may worsen without an antibiotic.

Most clinicians, researchers, and public health officials agree that antibiotics are overprescribed for ARIs. However, changing clinician and patient behaviors and beliefs is difficult. Admonitions and exhortations to stop prescribing are likely to be minimally effective. Educational interventions linked to performance assessment may be most effective.¹⁴ The interventions should be targeted to change specific beliefs and to correct specific knowledge gaps of both clinicians and patients. However, the factors that trigger antibiotic prescribing are not well understood. If it is patient expectation, what factors underlie that expectation? If it is physician knowledge, what are the specific gaps? Are there other factors that trigger antibiotic prescribing?

Objective

The purpose of our study was to determine the factors associated with antibiotic prescribing for ARIs in a prospective observational study. Our hypothesis was that antibiotic prescribing could be predicted from a limited number of signs and symptoms and that patients’ expectations and clinicians’ beliefs about patient expectations for antibiotics would be among the predictors of antibiotic prescribing. Our research question was: Among patients presenting to outpatient settings with symptoms compatible with ARIs, what factors-including symptoms, signs, patient characteristics, diagnoses, patient expectations, and so forth-are associated with antibiotic prescribing?

Methods

Setting

All 15 practices of the Upper Peninsula Research Network (UPRNet), a rural family practice research network in Northern Michigan, participated in our study. In January and February 1998 we asked each clinician in an UPRNet practice to enroll at least 20 consecutive patients 4 years and older who presented to the office with symptoms of an ARI. Patients were identified and invited to participate by 31 family physicians, 3 pediatricians, 5 internists, 16 physician assistants, and 3 nurse practitioners. We instructed the clinicians and their staffs not to enroll patients with a chief complaint of pharyngitis or ear pain. We excluded from the analyses patients who had a primary or secondary diagnosis of pneumonia, pharyngitis, or otitis media, because antibiotics are currently thought to be indicated for many of those cases. We included patients with the clinical diagnoses of nonspecific URI, bronchitis, and sinusitis because of the large overlap in signs and symptoms and the lack of precision in clinical diagnosis for those 3 conditions. Also, assigning a diagnosis of bronchitis or sinusitis may legitimize the use of an antibiotic for patients whose symptoms are otherwise identical to those of patients assigned the diagnosis of URI. Patients reporting symptoms for more than 30 days were excluded.

Interventions and Main Measures

Each patient (or parent) participating in the study was asked to complete a questionnaire detailing the patient’s symptoms, duration of symptoms, expected treatment for the current illness, history of previous respiratory illness and prior treatment, and smoking history. The nurse caring for the patient noted the patient’s temperature, location of visit (regular office or urgent care), and the type of clinician (physician, physician assistant, or nurse practitioner) who saw the patient. Finally, the clinician completed a questionnaire identifying the patient’s chief complaint, history of lung disease, physical findings, treatment, the clinician’s belief about the patient’s expectations, diagnosis, and secondary factors affecting the decision to give antibiotics, if given. Secondary factors were reasons for prescribing an antibiotic that were not directly related to the diagnosis. The list of options for secondary factors included: patient expected an antibiotic, patient requested an antibiotic, patient leaving town, patient not improving, patient getting worse, patient sick too long, patient smokes, patient has chronic lung disease, patient has comorbidity, and patient is extremely ill. Finally, we asked the clinicians, “What is the likelihood of an adverse outcome if this patient does not receive an antibiotic today?” Response options were: very unlikely, unlikely, moderately likely, and very likely.

Our study was approved by the Michigan State University Committee on Research Involving Human Subjects, and we obtained informed consent from all subjects.

All analyses were done using SPSS 7.0 (SPSS Inc, Chicago, Ill). We calculated frequency distributions of individual variables and used chi-square analysis to assess univariate associations between each variable and both antibiotic prescribing and patient expectation for an antibiotic prescription. Continuous variables were converted to categorical data for the univariate analyses but treated as continuous data for the logistic regression. Because we measured secondary factors only in encounters in which an antibiotic was prescribed, that analysis was descriptive only and was not included in the logistic regression models.

We developed 2 logistic regression models, 1 to identify the factors independently associated with antibiotic prescribing (antibiotic prescribing model) and 1 to identify the factors independently associated with patients’ expectations of receiving an antibiotic prescription (patient expectation model).

For the antibiotic prescribing model, dichotomous variables that had a statistically significant univariate association with prescriptions for antibiotics and all continuous variables were included in the first step of the logistic regression modeling. We used the following method in an attempt to avoid type 1 errors due to the many variables we tested: All variables with a P value of .05 or less on univariate analysis were included in the initial logistic regression model, then removed sequentially, in order of ascending statistical significance, if they did not affect the odds ratio of the remaining model by at least 20%. Only main effects were assessed. We calculated odds ratios and their 95% confidence intervals from the final antibiotic prescribing model.

The patient expectation model was developed in a similar fashion. For the logistic regression, continuous variables with clinically implausible outlying values (0%) or missing values (7%) were replaced with the mean value if normally distributed and with the median value if skewed. We excluded from analysis records missing dichotomous variables necessary for logistic regression analysis (1%).

Results

Nine hundred twenty-eight patients completed questionnaires; 11 declined participation. We eliminated from analysis 222 patients with primary diagnoses of otitis media, pneumonia, pharyngitis, allergy, or other. An additional 113 were eliminated because their questionnaires were lacking essential information, they were out of the age range, or they did not give informed consent. The remaining 593 had a main diagnosis of bronchitis, sinusitis, or nonspecific URI. There were 482 records entered in the analysis: 80 were eliminated because of a secondary diagnosis of otitis media (38), pharyngitis (33), or pneumonia (9), and 31 were eliminated because the patient reported being sick longer than 30 days. Mean temperature and mean days sick were assigned respectively to 17 records (3%) missing temperature and 21 records (4%) missing number of days sick.

The subjects were primarily adult women (67%) from a predominantly white (95%) population. Physicians saw most of the patients (67%), and most visits were in a physician’s office (81%) rather than an urgent care facility. Most patients were being seen for the current illness for the first time (91%) and had been sick for no more than 14 days (86%).

The principal diagnoses assigned by the clinicians were: sinusitis for 176 patients (37%), nonspecific URI in 167 patients (34%), and bronchitis in 139 patients (29%). Antibiotics were prescribed for 319 (66%) of all diagnoses, 111 (80%) of bronchitis diagnoses, 173 (98%) of sinusitis diagnoses, and 35 (21%) of nonspecific URI diagnoses.

Antibiotic Prescribing Model

Univariate analysis revealed that those patients 18 years and older (odds ratio [OR] =3.4; 95% confidence interval [CI], 2.0-5.7), seen in urgent care clinics (OR=1.7; 95% CI, 1.0-2.9), evaluated by a nurse practitioner (OR=5.9; 95% CI, 1.8-19.9), and sick for more than 14 days (OR=4.3; 95% CI=2.0-9.3) were more likely to be treated with an antibiotic Table 1. Antibiotic prescribing was much more frequent for those patients judged by the clinician to be moderately or very likely to have an adverse outcome compared with those judged unlikely or very unlikely to have an adverse outcome if not treated with an antibiotic (OR=61.8; 95% CI, 24.6-155). Table 2 and Table 3 summarize symptoms and signs associated with antibiotic prescribing. Antibiotic prescribing was modestly associated with a number of symptoms and signs Table 4 and much more common for patients with the physical findings of discolored nasal discharge (OR=20.4), sinus tenderness (OR=15.9), a wet cough (OR=8.6), and rales or rhonchi (OR=7.7). After controlling for confounding with logistic regression, 5 variables were independently associated with antibiotic prescribing: sinus tenderness (OR=20.0), rales/ronchi (OR=19.9), yellow/green nasal discharge (OR=11.7), and postnasal drainage (OR=3.1), and negatively associated with clear nasal discharge (OR=0.3).

Patient Expectations Model

An antibiotic was expected by 290 (60%) of the patients with a URI, bronchitis, or sinusitis. A majority of the patients (68%) had received an antibiotic for a similar illness in the past, and 299 (86%) of these patients believed the antibiotic helped. Patients expecting an antibiotic were more likely to receive one Table 5. Patients were more likely to expect an antibiotic who reported green or yellow nasal discharge (OR=2.0; 95% CI, 1.2-3.2), cough productive of yellow or green phlegm (OR=1.7; 95% CI, 1.1-2.5), pain over the eye or cheek (OR=1.6; 95% CI, 1.1-2.4), or who were helped by an antibiotic for similar symptoms in the past (OR=4.5; 95% CI, 2.9-6.9). However, only “helped by an antibiotic for similar symptoms in the past” remained significantly and independently associated with patient expectation after controlling for confounding by logistic regression.

Clinicians believed 298 patients (62%) expected an antibiotic, and they were significantly more likely to prescribe an antibiotic for those patients (OR=4.7; 95% CI, 3.2-7.1) on univariate analysis. The clinicians’ accuracy in predicting which patients expected antibiotics Table 6 was only fair (Cohen’s k=0.21). After controlling for confounding with logistic regression, patient expectation and clinician belief that the patient expected an antibiotic were not independent predictors of antibiotic prescribing. There were no significant differences in the demographic, symptomatic, and diagnostic variables between those records included in the analysis and those records excluded because of incomplete data, implausible data, or lack of informed consent. An analysis restricted to those records with every field completed and with plausible values did not change the final logistic regression models.

Secondary Factors

Clinicians reported that secondary factors contributed to the decision to prescribe an antibiotic for 95% of the patients. The most common reasons were: patient not improving (21%), patient getting worse (19%), and patient has been sick too long (19%). Other secondary reasons included: patient smokes (11%), patient has chronic lung disease (7%), patient expected an antibiotic (6%), patient was quite ill (5%), patient had a comorbid condition other than chronic lung disease (4%), patient was leaving town (2%), and patient requested an antibiotic (2%).

Discussion

Our study suggests that antibiotic prescribing for patients with the clinical diagnoses of URI, bronchitis, or sinusitis is determined by 2 general sets of factors: a small number of physical findings and clinicians’ perceptions of the clinical course of the illness. The physical findings are rales, rhonchi, sinus tenderness, postnasal drainage, purulent nasal discharge, and clear nasal discharge (a negative association). These findings were associated with antibiotic prescribing regardless of the diagnosis the clinician assigned. Notably absent from this list is cough, which was reported by more than 70% of patients assigned any of the 3 diagnoses. Apparently cough alone, which is the chief complaint in many cases of acute respiratory infection, is often not sufficient for clinicians to prescribe antibiotics. In our study, clinicians readily admitted that factors other than the diagnoses contributed to their decisions to prescribe antibiotics. In 95% of the cases, they listed a secondary reason. The majority of these reasons (59%) had to do with the clinical course of the illness: patient not improving (21%), patient getting worse (19%), and patient being sick too long (19%). For patients with the diagnosis of nonspecific URI, clinicians listed one of these reasons for 83% of cases. Whether any of these are legitimate reasons for prescribing antibiotics for acute respiratory infections has not been investigated. Similarly, when clinicians believed that patients were likely to have an adverse outcome if not treated with an antibiotic, they prescribed antibiotics in 98% of cases. This most likely reflects clinicians’ belief that acute sinusitis and acute bronchitis do not improve without antibiotics, a belief largely unsupported by the medical literature.

Our findings regarding patient expectations for antibiotics were surprising. On univariate analysis, we found that patient expectation for an antibiotic was a predictor of antibiotic prescribing. However, contrary to widely held beliefs, patient expectation of an antibiotic prescription was not an independent predictor of antibiotic prescribing in our study. After adjusting for other related factors in the logistic regression model, antibiotics were prescribed as frequently for those not expecting one as for those expecting one. We believe that patient expectation may be a surrogate marker for “having received an antibiotic in the past for a similar illness and had good results.” Thus, patient expectation is driven by past physician behavior in prescribing antibiotics. Other researchers have noted an association between patient expectation and antibiotic prescribing,^15,16 but they did not adjust for other factors associated with both patient expectation and antibiotic prescribing.

Also, in the absence of the 5 physical findings noted above (rales or rhonchi, sinus tenderness, postnasal drainage, purulent nasal discharge, and clear nasal discharge), patient expectation of an antibiotic and receiving an antibiotic in the past for a similar illness does not appear to affect clinicians’ decisions to prescribe an antibiotic. Are any of these physical findings good reason for prescribing an antibiotic? Other than rales, which may be associated with pneumonia, there is no good empiric evidence that purulent nasal discharge, postnasal discharge, rhonchi, or sinus tenderness distinguish between viral and bacterial illnesses.¹² Purulent nasal discharge is commonly thought to indicate bacterial infection, but this is not supported by clinical or laboratory research.⁶ Purulent nasal discharge does increase the odds of sinusitis, but it is unclear whether uncomplicated sinusitis diagnosed on clinical grounds improves more rapidly with antibiotic treatment.^17-20 The Centers for Disease Control and Prevention guidelines for treatment of sinusitis in children recommend no antibiotic unless symptoms are severe or have been present for more than 10 days.²¹

Inadequate Physician Knowledge

It has been suggested that inadequate knowledge of the nature of viral illnesses accounts for inappropriate antibiotic prescribing practices.⁵ Our findings support this assertion. The high percentage of cases of sinusitis (37%) and relatively low percentage of antibiotic prescriptions for nonspecific URI (21%) suggest that clinicians may have assigned the diagnoses of sinusitis to justify an antibiotic prescription, as has been suggested by Vinson and Lutz¹⁶ for bronchitis. However, this would suggest that clinicians intentionally falsified physical findings to justify antibiotic prescribing to satisfy patient demand or for other unidentified reasons. It is more likely that sinus tenderness was the basis for a diagnosis of sinusitis, rales or rhonchi were the basis for a diagnosis of bronchitis, and these diagnoses were the justification for prescribing an antibiotic.

Before an all-out war on antibiotics is waged, however, a very large randomized controlled trial of antibiotic treatment of acute respiratory infections that would allow subgroup analysis by the factors we have described may be necessary to determine whether some cases of ARI are antibiotic responsive. Although current evidence does not strongly support use of antibiotics for URIs, acute bronchitis, or acute sinusitis, sufficient evidence does not exist to refute a clinically significant effect in certain subgroups of these patients.

We found that the clinicians in our network depend on the presence of discolored nasal drainage, postnasal drainage, rales or rhonchi, sinus tenderness, and clear nasal drainage to decide whether to prescribe an antibiotic for patients with a URI, acute bronchitis, and acute sinusitis. In making their decisions they also considered the clinical course of the illness and a subjective assessment of the likelihood a patient will get worse without an antibiotic. We found that patients expect an antibiotic if they have had a similar illness in the past and had improved with an antibiotic; yet patient expectation alone does not appear to be a sufficient reason for clinicians to prescribe antibiotics for acute respiratory infections. These new findings may be of benefit to those designing interventions to reduce inappropriate use of antibiotics for acute respiratory infections.

Acknowledgments

We express our gratitude to Teresa L. Ettenhofer, Renee Lauscher, and Meghana Kasetty, MD, of Escanaba and the clinicians and office staffs of the following Upper Peninsula Research Network practices: Alcona Health Center in Lincoln, Alpena Medical Arts in Alpena, Burns Clinic in St. Ignace, Doctors Park Family Physicians in Escanaba, East Jordan Family Health Center in East Jordan, Ewen Medical Center in Ewen, OSF Medical Group in Escanaba and Gladstone, Gwinn Medical Center in Gwinn, Mackinac Straits Primary Care Clinic in St. Ignace, Marquette Internal Medicine Associates in Marquette, Marquette Medical Clinic in Iron River, and Northern Michigan Health Services in Houghton Lake.