Joe G. Kesterson, BS

Patients and Protocol

The institutional review board of Indiana University, Purdue University, Indianapolis, approved this study. We included in the study all adult patients (those at least 18 years old) admitted to Wishard Hospital between January 1, 1993, and December 31, 2002, except for prisoners (for IRB reasons) and obstetric patients (because their 30‐day mortality is very close to zero0.07% at our institution). Wishard Hospital is a large urban hospital that serves a diverse but predominantly inner‐city population in Indianapolis. If a patient was admitted more than once during the 10 years of observation, we included only the first admission in the analysis in order to assure statistical independence of the observations. We extracted data from the Regenstrief Medical Record System (RMRS), a comprehensive medical records system that has demographic data, vital signs, diagnoses, results of clinical tests, and pharmacy information on all inpatient, emergency department, and outpatient encounter sites.18

We obtained the admission and discharge ICD9 and DRG codes to assess the disease patterns associated with individual CBC abnormalities. We obtained these codes from routine hospital case abstractions performed by Wishard Hospital's medical records department using NCoder+ and Quadramed. Patients assigned DRG codes 370‐384 were identified as obstetric and therefore excluded. Using the ICD9 and CPT codes according to the Charlson algorithm, we calculated a Charlson Comorbidity Index value19 for each patient as a marker of coexisting conditions.

Outcomes

The primary outcome was 30‐day mortality counted from the date of admission. We used information from the hospital record (inpatient deaths) and the Indiana state death tapes to determine the dates of death of all patients. Patients were matched to the Indiana death tapes by an algorithm using name, social security number, date of birth, and sex.20

Hemogram and Differential Count Test Methods

The hemogram, differential counts, and blood smear exam results included in this study all came from Wishard Hospital's laboratory. During this study, the hospital used only 2 cell counters, the Coulter STK‐S and the Gen‐S automated blood analyzer (Beckman Coulter, Brea, California), to produce hemogram and automated blood differential counts. Both instruments provided automated differential counts of 5 cell types: neutrophils, lymphocytes, monocytes, basophils, and eosinophils. The latter machine also produced platelet counts and reticulocyte counts, but during the study period these counts were not routinely reported to physicians unless ordered specifically, so we did not include them in the analyses. The laboratory reflexively performed 100‐cell manual differential counts and blood smear exams when abnormalities as defined by College of American Pathologists (CAP) criteria were observed in the automated measures. Both automated blood analyzers used the same automated CAP criteria to decide when to add a manual differential count and blood smear analysis, and these criteria were constant throughout the study. This protocol predicts manual differential abnormalities with high sensitivity, missing less than 1% of important findings in a manual differential.21 When the CAP criteria did not require a manual differential count and blood smear exam, we assumed that those counts unique to a manual count, for example, blast cell count, were zero and that there were no abnormalities in blood smear morphology.

Laboratories may report white blood cells as absolute counts (eg, number of cells/mm³) and/or as percentages. We converted all counts reported as percentages to absolute numbers (eg, WBC count 1000 cell type percent/100). For absolute counts that have both high and low ranges, such as white blood cell (WBC) count, we constructed two binary variables. WBC‐low was 1 when the WBC was below the lower limit of normal; otherwise it was 0. WBC‐high was 1 if the WBC was above the upper limit of normal; otherwise it was 0. For continuous variables such as NRBCs or blasts where any presence on the manual differential count is abnormal, we constructed binary variables with 0 indicating absence of the cell type and 1 indicating a cell count was at least 1.

Measurements of many cell types in the manual differential count and smear assessment (eg, burr cells) are reported in qualitative terms such as occasional, few, increased, or present, if observed, or none seen, unremarkable, or no mention, if not observed. We dichotomized all such results as present or absent for analysis purposes.

Statistical Analysis

For all the original variables, we plotted cell counts against 30‐day mortality to graphically show this univariate association. To screen the effects of these 45 binary CBC variables univariately, we used each as the sole independent variable in a logistic regression model with 30‐day mortality as the dependent variable.

The simultaneous effects of the 45 CBC measures on mortality were investigated using multiple logistic regression models, always controlling for patient age (in years, as a continuous variable) and sex (as a dichotomous variable). Two approaches were taken to handle the large number of predictors in the model. First, we formed subgroups of predictors based on clinical judgment (eg, the subgroup of bands, Dohle bodies, and toxic granules associated with infections) and ran logistic regressions of each subgroup to choose the significant predictors of these subgroups to fit them into an overall prediction model of 30‐day mortality. The results were verified using a second approach that did not depend on subjective judgment. Both backward and forward stepwise variable selection procedures were used to choose the subset of significant predictors (P < .005) of 30‐day mortality in logistic regression, again controlling for age and sex. To be sure that the predictive power of the models was not decreased by converting continuous variables into categorical variables, we also ran models that included the continuous variables as potential predictors. We used the c statistic as a measure of the goodness‐of‐fit of the models. We included the Charlson Index and the 10 most common admission diagnoses in our model to control for comorbidities and prime reason for admission, respectively.

We performed the analysis using SAS software, version 8.02 (SAS Institute, Inc., Cary, NC).

Chart Review

For each independent predictor of 30‐day mortality that was both statistically significant and had a very high relative risk (>2.5), one author (A.K.) took a random sample of 100200 patients with positive values for this predictor and reviewed the dictated discharge summaries in order to asses the clinical correlates of these findings.

Predictors of 30‐Day Mortality

We examined the univariate effect of age, sex, and the 45 CBC variables (Table 2) on 30‐day mortality. Most of these variables showed a significant (P < .0001) effect on mortality. Only a few abnormalities, for example, a low WBC (< 5000/L), basophilia (>200/L), and eosinophilia (>450/L), were unrelated to 30‐day mortality. Increasing age and male sex were associated with increased mortality. Of the 45 CBC variables, 29 were strong (P < .0001) univariate predictors of mortality and had odds ratios (ORs) greater than 2.5. Eight variables had univariate ORs greater than 4: toxic granules, Dohle bodies, smudge cells, promyelocytes, myelocytes, metamyelocytes, NRBCs, and burr cells. All but 2 of these are white blood cell observations.

All the statistical approaches produced essentially the same model for predicting mortality. Table 3 shows that age, sex, and 13 of the CBC variables were retained in the final model of dichotomous variables using backward and forward selection. Lymphocytosis, burr cells, and NRBCs were the greatest independent predictors of mortality, with odds ratios greater than 2.5. Only 1 variable, sickle cells, predicted reduced mortality (with an odds ratio well below 1).

The c statistic (the ratio of the area under the ROC curve to the whole area, which reflects the overall predictive power of the final model), was about 0.80 by any approach, which compared favorably with previous prediction models.3, 4 Using continuous measures of CBC in the model did not increase the predictive power. Inclusion of the Charlson Index and the top 10 admission diagnoses did not significantly change the prediction model, although 2 admission diagnoses, chest pain and acute but ill‐defined cerebrovascular disease, emerged as independent predictors of 30‐day mortality, with odds ratios of 0.314 and 2.033, respectively, at P < .0001.

Chart Review

Of the 200 cases with NRBCs, the leading probable causes for this finding were severe hypoxia (average A‐a gradient = 326 mm Hg), acute anemia (average hgb = 6.1 gm/dL), and sickle‐cell anemia. Other diseases associated with NRBCs were infection/sepsis, HIV, solid tumors (breast/lung/colon/prostate), and leukemia or multiple myeloma. Having even a single NRBC at admission correlated with a 25.5% mortality rate. Of note, 30%40% of patients with sickle‐cell disease had NRBCs and moderate anemia (hgb = 8.7 gm/dL) on admission to the hospital, but there was no excess risk of mortality. Indeed, the 49 patients with sickle‐cell disease who had NRBCs at admission had a 30‐day mortality of 0%.

Most of the patients with NRBCs reviewed exhibited overt signs of severe disease, for example, shock, respiratory failure, or severe trauma, in addition to having NRBCs. However, in 2 patients the NRBCs were the only strong signal of disease severity. Both had NRBCs on the day of discharge and were readmitted within 3 days in extremis and died. One was readmitted in fulminant septic shock, likely from a bacterial peritonitis or urinary tract infection, and the other was readmitted in shock, likely from decompensated heart failure.

In univariate analysis, burr cells at admission correlated with a mortality rate of 27.3%. A review of 100 randomly chosen patients with burr cells revealed a pattern of associated diseases, that is, acute renal failure, liver failure, and congestive heart failure, different from that of patients with NRBCs. There was little overlap in the presence of burr cells and NRBCs, but the 12% who had burr cells and NRBCs had a high mortality rate (57%).

Absolute lymphocytosis was associated with a mortality rate of 8.6%. Although univariate analysis showed that the risk with lymphocytosis was not as high as that for patients with NRBCs or burr cells at admission, lymphocytosis was much more common (8.5%), and within the logistic model its presence explained more of the chi‐square statistic than any other variable except age. Indeed, lymphocytosis was a stronger predictor of 30‐day mortality than was high WBCs or anemia. Chart review of 200 patients with lymphocytosis showed a preponderance of them had large physiologic stressors, for example, traumatic tissue injury (surgery) or cerebrovascular injury. In one subset, half the patients (50.9% of 53 patients) who underwent craniotomy for trauma and had absolute lymphocytosis at admission died, compared with 20.8% of 101 patients admitted for the same diagnosis without absolute lymphocytosis.

Patients and Protocol

The institutional review board of Indiana University, Purdue University, Indianapolis, approved this study. We included in the study all adult patients (those at least 18 years old) admitted to Wishard Hospital between January 1, 1993, and December 31, 2002, except for prisoners (for IRB reasons) and obstetric patients (because their 30‐day mortality is very close to zero0.07% at our institution). Wishard Hospital is a large urban hospital that serves a diverse but predominantly inner‐city population in Indianapolis. If a patient was admitted more than once during the 10 years of observation, we included only the first admission in the analysis in order to assure statistical independence of the observations. We extracted data from the Regenstrief Medical Record System (RMRS), a comprehensive medical records system that has demographic data, vital signs, diagnoses, results of clinical tests, and pharmacy information on all inpatient, emergency department, and outpatient encounter sites.18

We obtained the admission and discharge ICD9 and DRG codes to assess the disease patterns associated with individual CBC abnormalities. We obtained these codes from routine hospital case abstractions performed by Wishard Hospital's medical records department using NCoder+ and Quadramed. Patients assigned DRG codes 370‐384 were identified as obstetric and therefore excluded. Using the ICD9 and CPT codes according to the Charlson algorithm, we calculated a Charlson Comorbidity Index value19 for each patient as a marker of coexisting conditions.

Outcomes

The primary outcome was 30‐day mortality counted from the date of admission. We used information from the hospital record (inpatient deaths) and the Indiana state death tapes to determine the dates of death of all patients. Patients were matched to the Indiana death tapes by an algorithm using name, social security number, date of birth, and sex.20

Hemogram and Differential Count Test Methods

The hemogram, differential counts, and blood smear exam results included in this study all came from Wishard Hospital's laboratory. During this study, the hospital used only 2 cell counters, the Coulter STK‐S and the Gen‐S automated blood analyzer (Beckman Coulter, Brea, California), to produce hemogram and automated blood differential counts. Both instruments provided automated differential counts of 5 cell types: neutrophils, lymphocytes, monocytes, basophils, and eosinophils. The latter machine also produced platelet counts and reticulocyte counts, but during the study period these counts were not routinely reported to physicians unless ordered specifically, so we did not include them in the analyses. The laboratory reflexively performed 100‐cell manual differential counts and blood smear exams when abnormalities as defined by College of American Pathologists (CAP) criteria were observed in the automated measures. Both automated blood analyzers used the same automated CAP criteria to decide when to add a manual differential count and blood smear analysis, and these criteria were constant throughout the study. This protocol predicts manual differential abnormalities with high sensitivity, missing less than 1% of important findings in a manual differential.21 When the CAP criteria did not require a manual differential count and blood smear exam, we assumed that those counts unique to a manual count, for example, blast cell count, were zero and that there were no abnormalities in blood smear morphology.

Laboratories may report white blood cells as absolute counts (eg, number of cells/mm³) and/or as percentages. We converted all counts reported as percentages to absolute numbers (eg, WBC count 1000 cell type percent/100). For absolute counts that have both high and low ranges, such as white blood cell (WBC) count, we constructed two binary variables. WBC‐low was 1 when the WBC was below the lower limit of normal; otherwise it was 0. WBC‐high was 1 if the WBC was above the upper limit of normal; otherwise it was 0. For continuous variables such as NRBCs or blasts where any presence on the manual differential count is abnormal, we constructed binary variables with 0 indicating absence of the cell type and 1 indicating a cell count was at least 1.

Measurements of many cell types in the manual differential count and smear assessment (eg, burr cells) are reported in qualitative terms such as occasional, few, increased, or present, if observed, or none seen, unremarkable, or no mention, if not observed. We dichotomized all such results as present or absent for analysis purposes.

Statistical Analysis

For all the original variables, we plotted cell counts against 30‐day mortality to graphically show this univariate association. To screen the effects of these 45 binary CBC variables univariately, we used each as the sole independent variable in a logistic regression model with 30‐day mortality as the dependent variable.

The simultaneous effects of the 45 CBC measures on mortality were investigated using multiple logistic regression models, always controlling for patient age (in years, as a continuous variable) and sex (as a dichotomous variable). Two approaches were taken to handle the large number of predictors in the model. First, we formed subgroups of predictors based on clinical judgment (eg, the subgroup of bands, Dohle bodies, and toxic granules associated with infections) and ran logistic regressions of each subgroup to choose the significant predictors of these subgroups to fit them into an overall prediction model of 30‐day mortality. The results were verified using a second approach that did not depend on subjective judgment. Both backward and forward stepwise variable selection procedures were used to choose the subset of significant predictors (P < .005) of 30‐day mortality in logistic regression, again controlling for age and sex. To be sure that the predictive power of the models was not decreased by converting continuous variables into categorical variables, we also ran models that included the continuous variables as potential predictors. We used the c statistic as a measure of the goodness‐of‐fit of the models. We included the Charlson Index and the 10 most common admission diagnoses in our model to control for comorbidities and prime reason for admission, respectively.

We performed the analysis using SAS software, version 8.02 (SAS Institute, Inc., Cary, NC).

Chart Review

For each independent predictor of 30‐day mortality that was both statistically significant and had a very high relative risk (>2.5), one author (A.K.) took a random sample of 100200 patients with positive values for this predictor and reviewed the dictated discharge summaries in order to asses the clinical correlates of these findings.

Predictors of 30‐Day Mortality

We examined the univariate effect of age, sex, and the 45 CBC variables (Table 2) on 30‐day mortality. Most of these variables showed a significant (P < .0001) effect on mortality. Only a few abnormalities, for example, a low WBC (< 5000/L), basophilia (>200/L), and eosinophilia (>450/L), were unrelated to 30‐day mortality. Increasing age and male sex were associated with increased mortality. Of the 45 CBC variables, 29 were strong (P < .0001) univariate predictors of mortality and had odds ratios (ORs) greater than 2.5. Eight variables had univariate ORs greater than 4: toxic granules, Dohle bodies, smudge cells, promyelocytes, myelocytes, metamyelocytes, NRBCs, and burr cells. All but 2 of these are white blood cell observations.

All the statistical approaches produced essentially the same model for predicting mortality. Table 3 shows that age, sex, and 13 of the CBC variables were retained in the final model of dichotomous variables using backward and forward selection. Lymphocytosis, burr cells, and NRBCs were the greatest independent predictors of mortality, with odds ratios greater than 2.5. Only 1 variable, sickle cells, predicted reduced mortality (with an odds ratio well below 1).

The c statistic (the ratio of the area under the ROC curve to the whole area, which reflects the overall predictive power of the final model), was about 0.80 by any approach, which compared favorably with previous prediction models.3, 4 Using continuous measures of CBC in the model did not increase the predictive power. Inclusion of the Charlson Index and the top 10 admission diagnoses did not significantly change the prediction model, although 2 admission diagnoses, chest pain and acute but ill‐defined cerebrovascular disease, emerged as independent predictors of 30‐day mortality, with odds ratios of 0.314 and 2.033, respectively, at P < .0001.

Chart Review

Of the 200 cases with NRBCs, the leading probable causes for this finding were severe hypoxia (average A‐a gradient = 326 mm Hg), acute anemia (average hgb = 6.1 gm/dL), and sickle‐cell anemia. Other diseases associated with NRBCs were infection/sepsis, HIV, solid tumors (breast/lung/colon/prostate), and leukemia or multiple myeloma. Having even a single NRBC at admission correlated with a 25.5% mortality rate. Of note, 30%40% of patients with sickle‐cell disease had NRBCs and moderate anemia (hgb = 8.7 gm/dL) on admission to the hospital, but there was no excess risk of mortality. Indeed, the 49 patients with sickle‐cell disease who had NRBCs at admission had a 30‐day mortality of 0%.

Most of the patients with NRBCs reviewed exhibited overt signs of severe disease, for example, shock, respiratory failure, or severe trauma, in addition to having NRBCs. However, in 2 patients the NRBCs were the only strong signal of disease severity. Both had NRBCs on the day of discharge and were readmitted within 3 days in extremis and died. One was readmitted in fulminant septic shock, likely from a bacterial peritonitis or urinary tract infection, and the other was readmitted in shock, likely from decompensated heart failure.

In univariate analysis, burr cells at admission correlated with a mortality rate of 27.3%. A review of 100 randomly chosen patients with burr cells revealed a pattern of associated diseases, that is, acute renal failure, liver failure, and congestive heart failure, different from that of patients with NRBCs. There was little overlap in the presence of burr cells and NRBCs, but the 12% who had burr cells and NRBCs had a high mortality rate (57%).

Absolute lymphocytosis was associated with a mortality rate of 8.6%. Although univariate analysis showed that the risk with lymphocytosis was not as high as that for patients with NRBCs or burr cells at admission, lymphocytosis was much more common (8.5%), and within the logistic model its presence explained more of the chi‐square statistic than any other variable except age. Indeed, lymphocytosis was a stronger predictor of 30‐day mortality than was high WBCs or anemia. Chart review of 200 patients with lymphocytosis showed a preponderance of them had large physiologic stressors, for example, traumatic tissue injury (surgery) or cerebrovascular injury. In one subset, half the patients (50.9% of 53 patients) who underwent craniotomy for trauma and had absolute lymphocytosis at admission died, compared with 20.8% of 101 patients admitted for the same diagnosis without absolute lymphocytosis.

Patients and Protocol

The institutional review board of Indiana University, Purdue University, Indianapolis, approved this study. We included in the study all adult patients (those at least 18 years old) admitted to Wishard Hospital between January 1, 1993, and December 31, 2002, except for prisoners (for IRB reasons) and obstetric patients (because their 30‐day mortality is very close to zero0.07% at our institution). Wishard Hospital is a large urban hospital that serves a diverse but predominantly inner‐city population in Indianapolis. If a patient was admitted more than once during the 10 years of observation, we included only the first admission in the analysis in order to assure statistical independence of the observations. We extracted data from the Regenstrief Medical Record System (RMRS), a comprehensive medical records system that has demographic data, vital signs, diagnoses, results of clinical tests, and pharmacy information on all inpatient, emergency department, and outpatient encounter sites.18

We obtained the admission and discharge ICD9 and DRG codes to assess the disease patterns associated with individual CBC abnormalities. We obtained these codes from routine hospital case abstractions performed by Wishard Hospital's medical records department using NCoder+ and Quadramed. Patients assigned DRG codes 370‐384 were identified as obstetric and therefore excluded. Using the ICD9 and CPT codes according to the Charlson algorithm, we calculated a Charlson Comorbidity Index value19 for each patient as a marker of coexisting conditions.

Outcomes

The primary outcome was 30‐day mortality counted from the date of admission. We used information from the hospital record (inpatient deaths) and the Indiana state death tapes to determine the dates of death of all patients. Patients were matched to the Indiana death tapes by an algorithm using name, social security number, date of birth, and sex.20

Hemogram and Differential Count Test Methods

The hemogram, differential counts, and blood smear exam results included in this study all came from Wishard Hospital's laboratory. During this study, the hospital used only 2 cell counters, the Coulter STK‐S and the Gen‐S automated blood analyzer (Beckman Coulter, Brea, California), to produce hemogram and automated blood differential counts. Both instruments provided automated differential counts of 5 cell types: neutrophils, lymphocytes, monocytes, basophils, and eosinophils. The latter machine also produced platelet counts and reticulocyte counts, but during the study period these counts were not routinely reported to physicians unless ordered specifically, so we did not include them in the analyses. The laboratory reflexively performed 100‐cell manual differential counts and blood smear exams when abnormalities as defined by College of American Pathologists (CAP) criteria were observed in the automated measures. Both automated blood analyzers used the same automated CAP criteria to decide when to add a manual differential count and blood smear analysis, and these criteria were constant throughout the study. This protocol predicts manual differential abnormalities with high sensitivity, missing less than 1% of important findings in a manual differential.21 When the CAP criteria did not require a manual differential count and blood smear exam, we assumed that those counts unique to a manual count, for example, blast cell count, were zero and that there were no abnormalities in blood smear morphology.

Laboratories may report white blood cells as absolute counts (eg, number of cells/mm³) and/or as percentages. We converted all counts reported as percentages to absolute numbers (eg, WBC count 1000 cell type percent/100). For absolute counts that have both high and low ranges, such as white blood cell (WBC) count, we constructed two binary variables. WBC‐low was 1 when the WBC was below the lower limit of normal; otherwise it was 0. WBC‐high was 1 if the WBC was above the upper limit of normal; otherwise it was 0. For continuous variables such as NRBCs or blasts where any presence on the manual differential count is abnormal, we constructed binary variables with 0 indicating absence of the cell type and 1 indicating a cell count was at least 1.

Measurements of many cell types in the manual differential count and smear assessment (eg, burr cells) are reported in qualitative terms such as occasional, few, increased, or present, if observed, or none seen, unremarkable, or no mention, if not observed. We dichotomized all such results as present or absent for analysis purposes.

Statistical Analysis

For all the original variables, we plotted cell counts against 30‐day mortality to graphically show this univariate association. To screen the effects of these 45 binary CBC variables univariately, we used each as the sole independent variable in a logistic regression model with 30‐day mortality as the dependent variable.

The simultaneous effects of the 45 CBC measures on mortality were investigated using multiple logistic regression models, always controlling for patient age (in years, as a continuous variable) and sex (as a dichotomous variable). Two approaches were taken to handle the large number of predictors in the model. First, we formed subgroups of predictors based on clinical judgment (eg, the subgroup of bands, Dohle bodies, and toxic granules associated with infections) and ran logistic regressions of each subgroup to choose the significant predictors of these subgroups to fit them into an overall prediction model of 30‐day mortality. The results were verified using a second approach that did not depend on subjective judgment. Both backward and forward stepwise variable selection procedures were used to choose the subset of significant predictors (P < .005) of 30‐day mortality in logistic regression, again controlling for age and sex. To be sure that the predictive power of the models was not decreased by converting continuous variables into categorical variables, we also ran models that included the continuous variables as potential predictors. We used the c statistic as a measure of the goodness‐of‐fit of the models. We included the Charlson Index and the 10 most common admission diagnoses in our model to control for comorbidities and prime reason for admission, respectively.

We performed the analysis using SAS software, version 8.02 (SAS Institute, Inc., Cary, NC).

Chart Review

For each independent predictor of 30‐day mortality that was both statistically significant and had a very high relative risk (>2.5), one author (A.K.) took a random sample of 100200 patients with positive values for this predictor and reviewed the dictated discharge summaries in order to asses the clinical correlates of these findings.

Predictors of 30‐Day Mortality

We examined the univariate effect of age, sex, and the 45 CBC variables (Table 2) on 30‐day mortality. Most of these variables showed a significant (P < .0001) effect on mortality. Only a few abnormalities, for example, a low WBC (< 5000/L), basophilia (>200/L), and eosinophilia (>450/L), were unrelated to 30‐day mortality. Increasing age and male sex were associated with increased mortality. Of the 45 CBC variables, 29 were strong (P < .0001) univariate predictors of mortality and had odds ratios (ORs) greater than 2.5. Eight variables had univariate ORs greater than 4: toxic granules, Dohle bodies, smudge cells, promyelocytes, myelocytes, metamyelocytes, NRBCs, and burr cells. All but 2 of these are white blood cell observations.

All the statistical approaches produced essentially the same model for predicting mortality. Table 3 shows that age, sex, and 13 of the CBC variables were retained in the final model of dichotomous variables using backward and forward selection. Lymphocytosis, burr cells, and NRBCs were the greatest independent predictors of mortality, with odds ratios greater than 2.5. Only 1 variable, sickle cells, predicted reduced mortality (with an odds ratio well below 1).

The c statistic (the ratio of the area under the ROC curve to the whole area, which reflects the overall predictive power of the final model), was about 0.80 by any approach, which compared favorably with previous prediction models.3, 4 Using continuous measures of CBC in the model did not increase the predictive power. Inclusion of the Charlson Index and the top 10 admission diagnoses did not significantly change the prediction model, although 2 admission diagnoses, chest pain and acute but ill‐defined cerebrovascular disease, emerged as independent predictors of 30‐day mortality, with odds ratios of 0.314 and 2.033, respectively, at P < .0001.

Chart Review

Of the 200 cases with NRBCs, the leading probable causes for this finding were severe hypoxia (average A‐a gradient = 326 mm Hg), acute anemia (average hgb = 6.1 gm/dL), and sickle‐cell anemia. Other diseases associated with NRBCs were infection/sepsis, HIV, solid tumors (breast/lung/colon/prostate), and leukemia or multiple myeloma. Having even a single NRBC at admission correlated with a 25.5% mortality rate. Of note, 30%40% of patients with sickle‐cell disease had NRBCs and moderate anemia (hgb = 8.7 gm/dL) on admission to the hospital, but there was no excess risk of mortality. Indeed, the 49 patients with sickle‐cell disease who had NRBCs at admission had a 30‐day mortality of 0%.

Most of the patients with NRBCs reviewed exhibited overt signs of severe disease, for example, shock, respiratory failure, or severe trauma, in addition to having NRBCs. However, in 2 patients the NRBCs were the only strong signal of disease severity. Both had NRBCs on the day of discharge and were readmitted within 3 days in extremis and died. One was readmitted in fulminant septic shock, likely from a bacterial peritonitis or urinary tract infection, and the other was readmitted in shock, likely from decompensated heart failure.

In univariate analysis, burr cells at admission correlated with a mortality rate of 27.3%. A review of 100 randomly chosen patients with burr cells revealed a pattern of associated diseases, that is, acute renal failure, liver failure, and congestive heart failure, different from that of patients with NRBCs. There was little overlap in the presence of burr cells and NRBCs, but the 12% who had burr cells and NRBCs had a high mortality rate (57%).

Absolute lymphocytosis was associated with a mortality rate of 8.6%. Although univariate analysis showed that the risk with lymphocytosis was not as high as that for patients with NRBCs or burr cells at admission, lymphocytosis was much more common (8.5%), and within the logistic model its presence explained more of the chi‐square statistic than any other variable except age. Indeed, lymphocytosis was a stronger predictor of 30‐day mortality than was high WBCs or anemia. Chart review of 200 patients with lymphocytosis showed a preponderance of them had large physiologic stressors, for example, traumatic tissue injury (surgery) or cerebrovascular injury. In one subset, half the patients (50.9% of 53 patients) who underwent craniotomy for trauma and had absolute lymphocytosis at admission died, compared with 20.8% of 101 patients admitted for the same diagnosis without absolute lymphocytosis.