Absolute Lymphocyte Count in the Emergency Department Predicts a Low CD4 Count in Admitted HIV-positive Patients

Authors

  • Anthony M. Napoli MD,

    1. From the Department of Emergency Medicine, Brown University Medical School (AMN), Providence, RI; the Department of Emergency Medicine, Beth Israel Deaconess Medical Center (CMF), Boston, MA; the Department of Emergency Medicine (JMP, MG, DM), George Washington University School of Medicine; the Department of Health Policy, George Washington University School of Public Health and Health Sciences (JMP), Washington, DC; and the Georgetown University School of Medicine (HS), Washington, DC.
    Search for more papers by this author
  • Christopher M. Fischer MD,

    1. From the Department of Emergency Medicine, Brown University Medical School (AMN), Providence, RI; the Department of Emergency Medicine, Beth Israel Deaconess Medical Center (CMF), Boston, MA; the Department of Emergency Medicine (JMP, MG, DM), George Washington University School of Medicine; the Department of Health Policy, George Washington University School of Public Health and Health Sciences (JMP), Washington, DC; and the Georgetown University School of Medicine (HS), Washington, DC.
    Search for more papers by this author
  • Jesse M. Pines MD, MBA, MSCE,

    1. From the Department of Emergency Medicine, Brown University Medical School (AMN), Providence, RI; the Department of Emergency Medicine, Beth Israel Deaconess Medical Center (CMF), Boston, MA; the Department of Emergency Medicine (JMP, MG, DM), George Washington University School of Medicine; the Department of Health Policy, George Washington University School of Public Health and Health Sciences (JMP), Washington, DC; and the Georgetown University School of Medicine (HS), Washington, DC.
    Search for more papers by this author
  • Hahn Soe-lin,

    1. From the Department of Emergency Medicine, Brown University Medical School (AMN), Providence, RI; the Department of Emergency Medicine, Beth Israel Deaconess Medical Center (CMF), Boston, MA; the Department of Emergency Medicine (JMP, MG, DM), George Washington University School of Medicine; the Department of Health Policy, George Washington University School of Public Health and Health Sciences (JMP), Washington, DC; and the Georgetown University School of Medicine (HS), Washington, DC.
    Search for more papers by this author
  • Munish Goyal MD,

    1. From the Department of Emergency Medicine, Brown University Medical School (AMN), Providence, RI; the Department of Emergency Medicine, Beth Israel Deaconess Medical Center (CMF), Boston, MA; the Department of Emergency Medicine (JMP, MG, DM), George Washington University School of Medicine; the Department of Health Policy, George Washington University School of Public Health and Health Sciences (JMP), Washington, DC; and the Georgetown University School of Medicine (HS), Washington, DC.
    Search for more papers by this author
  • David Milzman MD

    1. From the Department of Emergency Medicine, Brown University Medical School (AMN), Providence, RI; the Department of Emergency Medicine, Beth Israel Deaconess Medical Center (CMF), Boston, MA; the Department of Emergency Medicine (JMP, MG, DM), George Washington University School of Medicine; the Department of Health Policy, George Washington University School of Public Health and Health Sciences (JMP), Washington, DC; and the Georgetown University School of Medicine (HS), Washington, DC.
    Search for more papers by this author

Errata

This article is corrected by:

  1. Errata: Erratum Volume 18, Issue 5, 565, Article first published online: 13 May 2011

  • Presented at the Society for Academic Emergency Medicine annual meeting, Phoenix, AZ, June 2010.

  • The authors have no relevant financial information or potential conflicts of interest to disclose.

  • Supervising Editor: Richard E. Rothman, MD, PhD.

Address for correspondence and reprints: Anthony M. Napoli, MD; e-mail: anapoli@lifespan.org.

Abstract

ACADEMIC EMERGENCY MEDICINE 2011; 18:385–389 © 2011 by the Society for Academic Emergency Medicine

Abstract

Objectives:  This study sought to determine if the automated absolute lymphocyte count (ALC) predicts a “low” (<200 × 106 cells/μL) CD4 count in patients with known human immunodeficiency virus (HIV+) who are admitted to the hospital from the emergency department (ED).

Methods:  This retrospective cohort study over an 8-year period was performed in a single, urban academic tertiary care hospital with over 85,000 annual ED visits. Included were patients who were known to be HIV+ and admitted from the ED, who had an ALC measured in the ED and a CD4 count measured within 24 hours of admission. Back-translated means and confidence intervals (CIs) were used to describe CD4 and ALC levels. The primary outcome was to determine the utility of an ALC threshold for predicting a CD4 count of <200 × 106 cells/μL by assessing the strength of association between log-transformed ALC and CD4 counts using a Pearson correlation coefficient. In addition, area under the receiver operator curve (AUC) and a decision plot analysis were used to calculate the sensitivity, specificity, and the positive and negative likelihood ratios to identify prespecified optimal clinical thresholds of a likelihood ratio of <0.1 and >10.

Results:  A total of 866 patients (mean age 42 years, 40% female) met inclusion criteria. The transformed means (95% CIs) for CD4 and ALC were 34 (31–38) and 654 (618–691), respectively. There was a significant relationship between the two measures, r = 0.74 (95% CI = 0.71 to 0.77, p < 0.01). The AUC was 0.92 (95% CI = 0.90 to 0.94, p < 0.001). An ALC of <1700 × 106 cells/μL had a sensitivity of 95% (95% CI = 93% to 96%), specificity of 52% (95% CI = 43% to 62%), and negative likelihood ratio of 0.09 (95% CI = 0.05 to 0.2) for a CD4 count of <200 × 106 cells/μL. An ALC of <950 × 106 cells/μL has a sensitivity of 76% (95% CI = 73% to 79%), specificity of 93% (95% CI = 87% to 96%), and positive likelihood ratio of 10.1 (95% CI = 8.2 to 14) for a CD4 count of <200 × 106 cells/μL.

Conclusions:  Absolute lymphocyte count was predictive of a CD4 count of <200 × 106 cells/μL in HIV+ patients who present to the ED, necessitating hospital admission. A CD4 count of <200 × 106 cells/μL is very likely if the ED ALC is <950 × 106 cells/μL and less likely if the ALC is >1,700 × 106 cells/μL. Depending on pretest probability, clinical use of this relationship may help emergency physicians predict the likelihood of susceptibility to opportunistic infections and may help identify patients who should receive definitive CD4 testing.

The estimated prevalence of known human immunodeficiency virus (HIV) is as much as 3% in some urban environments, and the prevalence of undiagnosed HIV in the emergency department (ED) ranges from 0.7% to 16%.1–4 The CD4 count is a useful predictor of the susceptibility of HIV patients to certain types of illnesses; a CD4 count of <200 × 106 cells/μL signifies a greater risk for opportunistic infections.5 Emergency care for patients with HIV can be challenging because the differential diagnoses and treatment for acute illness may vary based on the level of immune suppression.

The CD4 count is often unknown in the ED because testing is not rapidly available and patient recollection or medical record reviews are sometimes not helpful. The inability to accurately assess immunosuppression in HIV patients may lead emergency physicians to either overuse or underuse resources aimed at specific opportunistic infections, such as antibiotics for Pneumocystis jiroveci. Several studies have shown that the absolute lymphocyte count (ALC), a commonly available ED test, is a reliable predictor of a low CD4 count.6–10 However, these studies have included U.S. clinics,6,9 underserved regions of the world,7 or a mixture of inpatient, outpatient, and ED patients.10 To our knowledge, no study has assessed the relationship between ALC and CD4 count in a cohort of only ED patients.

We sought to determine the utility of the ALC in predicting CD4 counts of HIV patients presenting to the ED. Our primary hypothesis was that the ED ALC would predict the CD4 count in HIV+ ED patients being admitted to the hospital. Our secondary hypothesis was that this relationship would be strong enough such that one could use the ALC as a surrogate for the CD4 in clinical decision-making.

Methods

Study Design

This was a retrospective, cohort study of consecutive HIV+ patients admitted through the ED. The institutional review board approved this study with waiver of written informed consent.

Study Setting and Population

The study was conducted at an urban academic Level I trauma center (annual ED census >85,000) between November 30, 2001, and November 30, 2009. Patients were included if they 1) were admitted to the hospital through the ED, 2) had a known diagnosis of HIV by International Classification of Diseases (ICD-9) code, 3) had an ALC measured in the ED, and 4) had a CD4 count measured within 24 hours of the ALC. The main outcome was a “low” CD4 count, defined as <200 × 106 cells/μL.5

Study Protocol

The chart review of demographic, laboratory, and clinical data used an electronic medical record (Amalga, Redmond, WA). The primary hypothesis, inclusion and exclusion criteria, and desired data variables were all defined prior to data abstraction. A trained chart abstractor with extensive experience using the electronic medical record, who was blinded to the hypothesis of the study, extracted the data based on the predefined inclusion and exclusion criteria. Inter-rater reliability was not assessed because all data extracted were objective data and did not require any interpretation other than application of the search filter within the electronic medical record. One of the authors of the study (not blinded to the hypothesis) confirmed that the correct filter had been applied prior to data analysis. In the study hospital, the ALC was measured using a FACSCalibur (Becton Dickson Immunocytometry Systems, San Jose, CA). CD4 lymphocyte counts were analyzed utilizing a Sysmex XT-1800 (Sysmex, Inc., Mundelein, IL). Data were extracted into a standard Excel database (Microsoft Corp., Redmond, WA).

Data Analysis

Patient age was normally distributed and is reported as mean and standard deviation (SD). To reduce the effects of skewness in the data distribution, data were log-transformed for analysis. Back-transformed means and 95% confidence intervals (CIs) were reported for the CD4 and ALC. Because some CD4 counts were zero, log transformation of the CD4 was done after adding one to each CD4 count to avoid undefined numbers. A p-value of <0.05 was considered significant. A Pearson correlation coefficient was used to assess if a significant relationship exists between ALC and CD4, the primary study hypothesis. Using an expected correlation of r = 0.75 and a null correlation of r = 0.7 (α = 0.05, β = 0.9), the estimated necessary sample size was approximately 944 patients. Based on a single-year sampling, and excluding repeated analyses of a single patient, we estimated this would require 8 years of record review. To test our secondary hypothesis, the area under the receiver operator curve (AUC) was calculated using the Wilcoxon method. Using decision plot analysis, we calculated the sensitivity, specificity, and positive and negative likelihood ratios across all ALC for a CD4 count of <200 × 106 cells/μL. Bayesian theory suggests that a positive likelihood ratio of >10 and negative likelihood ratio of <0.1 for a CD4 count <200 × 106 cells/μL would significantly affect clinical decision-making, so we identified ALC cutoffs that corresponded to these values from the decision plot and present the most salient points as proportions.

Results

A total of 1,546 patients were identified over an 8-year period who were treated in the ED with an ICD-9–coded diagnosis of HIV. A portion of those were admitted (82.4%) leaving a total of 1,274 records in 936 unique patients. The first ED admission was included in the case of duplicate patient admissions. Of the 936 unique patients, 14 left against medical advice, and two expired prior to obtaining a CD4 count. Fifty-four patients were excluded because an ALC or CD4 count was not drawn. The remaining 866 unique patients fully met inclusion criteria during the study period. The mean (±SD) age was 42 (±10) years, 40% were female, and 91% were African American. The transformed mean for CD4 was 34 (95% CI = 31–38) and for ALC was 654 (95% CI = 618–691); the raw ranges were 0–1116 and 12–7725, respectively. A CD4 count of <200 × 106 cells/μL was present in 84.6% (95% CI = 82.2% to 87.0%) of the patients.

There was a significant relationship between the ALC and CD4, r = 0.74 (95% CI = 0.71 to 0.77). A linear regression through the log-transformed data demonstrates a linear relationship; this is apparent in the scatter plot in Figure 1. The AUC was 0.92 (95% CI = 0.90 to 0.94, p < 0.001; Figure 2). The sensitivity, specificity, positive likelihood ratio, and negative likelihood ratios at certain ALC are shown in Table 1. An ALC of <1,700 × 106 cells/μL has a sensitivity of 95% (95% CI = 93% to 96%), specificity of 52% (95% CI = 43% to 62%), and negative likelihood ratio of 0.09 (95% CI = 0.05 to 0.2) for a CD4 count of <200 × 106 cells/μL. Of all patients with a CD4 count <200 × 106 cells/μL, 95% (95% CI = 93% to 96%) had an ALC of <1,700 × 106 cells/μL (Table 2). An ALC <950 × 106 cells/μL had a sensitivity of 76% (95% CI = 73% to 79%) and specificity of 93% (95% CI = 87% to 96%), with a positive likelihood ratio of 10.1 (95% CI = 8.2% to 14%) for a CD4 count of <200 × 106 cells/μL. Of all patients with a CD4 count of >200 × 106 cells/μL, 93% (95% CI = 87% to 96%) had an ALC of >950 × 106 cells/μL (Table 3).

Figure 1.

 Scatterplot of CD4 count as a function of ALC. ALC = absolute lymphocyte count.

Figure 2.

 The ROC for ALC as a predictor of a CD4 count less than 200 × 106 cells/μL (ROC = 0.92, 95% CI = 0.9 to 0.94, p ≤ 0.0001). ALC = absolute lymphocyte count; ROC = receiver operating curve.

Table 1. 
Sensitivity, Specificity, and Likelihood Ratios (LR) for ALC Predicting a CD4 Count of <200 × 106 cells/μL
ALC (×106 Cells/μL)Sensitivity (%)Specificity (%)Positive LRNegative LR
  1. Sensitivity = true positives/(true positives + false negatives); specificity = true negatives/(true negatives + false positives); positive likelihood ratio = sensitivity/(1 – specificity); negative likelihood ratio = (1 – sensitivity)/specificity.

  2. ALC = absolute lymphocyte count.

 750639728.30.37
 950769310.10.26
100078907.40.24
125087753.50.16
150093622.40.12
170095522.00.09
200097361.50.07
Table 2. 
2 × 2 Table for Diagnostic Test Results for a Cutoff Likelihood ratio < 0.1
 CD4 < 200CD4 ≥ 200Total
  1. ALC = absolute lymphocyte count.

ALC < 170069664760
ALC ≥ 17003769106
Total733133866
Table 3. 
2 × 2 Table for Diagnostic Test Results for a Cutoff Likelihood Ratio >10
 CD4 < 200CD4 ≥ 200Total
  1. ALC = absolute lymphocyte count.

ALC < 95055710567
ALC ≥ 950176123299
Total733133866

Discussion

Rapid diagnostic testing for the presence of HIV in ED patients is becoming common; however, rapid assessment of the risk of opportunistic infections by measuring CD4 counts in the ED is limited because the test can take hours to run, and most laboratories do not have the resources to immediately run it. Early and appropriate treatment of opportunistic infections is a key aspect of the care of the HIV patient in the ED. We have found a simple, readily available measure to estimate a CD4 count in HIV+ ED patients based on the ALC. This relationship has been reported previously, although not in a cohort of only ED patients. Prior studies have demonstrated this relationship in cohorts without acute illness or in mixed populations. Multiple factors, particularly the acuity of illness and acute comorbidities, may affect the lymphocyte count and the accuracy of this relationship. Thus, demonstrating a clinically useful relationship in ED patients with acute illness is important before using certain thresholds in clinical decision-making. Because clinicians cannot currently obtain a CD4 count, and patients may often be unaware, it is important to validate this relationship in a select group of patients where clinical use of the relationship may be important.

By only studying the relationship between the ALC and CD4 count in this group of patients, we hoped to target the spectrum of patients in which clinician uncertainty led to ordering the CD4 within the first 24 hours. Discharged patients, or admitted HIV+ patients with a recent CD4 count, would not likely undergo testing of the CD4 count early in hospitalization. Augmentation of clinical treatment of acutely ill patients requiring admission who have an unknown CD4 count is likely a more important population to study because knowledge of the risk of opportunistic infections may alter the early course of hospital care.

Our results demonstrate a strong relationship between ALC and CD4 in HIV+ patients, with a higher correlation (r = 0.74) than prior studies.7,8,10 Our cutoffs are similar to prior studies where ALC cutoffs of approximately 1,100 × 106 to 1,300 × 106 and 2,000 × 106 cells/μL have been proposed to rule in and rule out a “low” CD4, although the results should be viewed as part of a continuum.6,7,10 These results can be used as a tool in conjunction with Bayesian decision-making and consideration of pretest probability. It is generally recognized that likelihood ratios of >10 or <0.1 may significantly affect clinical decision-making, particularly with intermediate pretest probability.11 For example, a likelihood ratio of 0.1 means an ALC greater than that value would be 10 times less likely to come from an individual with a CD4 count <200 × 106 cells/μL. An ALC of >1,700 × 106 cells/μL has a negative likelihood ratio of 0.1, and an ALC of <950 × 106 cells/μL has a positive likelihood ratio of 10.1. Because of the high prevalence of a low CD4 in our cohort, one cannot use a cutoff of ALC >1,700 × 106 cells/μL to rule out disease despite its high sensitivity; the posttest probability would still be 34%. An ALC of <950 × 106 cells/μL, however, would have a 98% posttest probability of disease and confirm a low CD4 when suspected.

Prior studies have documented various aspects of the relationship between the ALC and the CD4 count.6–10 The reported prevalence of a CD4 count <200 × 106 cells/μL in outpatient or mixed patient populations ranges from 16% to 40%.6,10 The similar cutoffs across studies suggest that there is little spectrum bias. Thus, in ED cohorts where the prevalence of a low CD4 count is <40%, an ALC of >1,700 × 106 cells/μL would reduce the probability of a CD4 count of <200 × 106 cells/μL to 5%. It remains to be seen what the general prevalence of a low CD4 in HIV+ patients is across multiple EDs. If the prevalence is as high as our cohort of admitted patients, then the ALC would likely be useful to confirm, but not exclude a low CD4.

The 2 × 2 tables allow for reporting of the sensitivity and specificity for the two chosen cutoffs in a more simplified proportional approach. The sensitivity represents the proportion of patients with a CD4 count <200 × 106 cells/μL (sum of first column) who also had an ALC of <1,700 × 106 cells/μL. In Table 2, this is 696 of 733, or 95%. The specificity represents the proportion of patients with CD4 count of ≥200 (sum of second column) who also had ALC of ≥1,700 × 106 cells/μL. In Table 2, this is 69 of 133, or 52%. This same process can be applied at a different cutoff as in Table 3, where one finds that the proportion of patients with a CD4 count of >200 × 106 cells/μL (sum of second column) that had an ALC of >950 × 106 cells/μL would be 123 of 133, or 93%. Taken as a whole this approach suggests it is unlikely to have a CD4 count of <200 × 106 cells/μL if the ALC is greater than 1,700 × 106 cells/μL, but likely if it is less than 950 × 106 cells/μL.

Clinicians may choose to use this information in conjunction with the clinical scenario to empirically treat for opportunistic infections in HIV+ patients. One should not interpret these findings to mean that patients with a low ALC in the ED should be empirically treated for an opportunistic infection, unless there is a high suspicion of HIV, since all studies done to date, including this one, have been in known HIV patients. Studies in undifferentiated ED patients with a low ALC would be helpful to assess the value of ALC in predicting a patient with HIV and a low CD4 count. However, these results may be useful in guiding the treatment of patients with suspected HIV when rapid testing is available.

Limitations

This study is limited by its retrospective design, potential for selection bias, and single-center nature. Studying only known HIV+ admitted ED patients with CD4 counts could lead to substantial selection bias. We envision three potential selection biases. First, discharged patients were excluded because 24-hour CD4 counts were unavailable. Second, we also excluded known HIV+ admitted patients in whom a CD4 count was not ordered, presumably because of previous knowledge on the part of the clinician or patient of a recent CD4 count. Third, patients with an unknown or unsuspected HIV+ serostatus at the time of admission could not be assessed using this study design and were therefore excluded. However, we believe known HIV+ patients with unknown CD4 counts represent the patients for whom the clinician would most likely use the ALC and CD4 count relationship in decision-making. The fact that our results are similar to those reported by other studies suggests selection bias is not a major factor influencing our results. Broad inclusion of all HIV+ patients, regardless of their disposition or knowledge of a recent CD4 count, would likely lead to spectrum bias by including these patients in determining a clinical threshold that would only be used in situations of clinical uncertainty.

Although the CD4 count was measured within 24 hours of presentation, it is unclear whether it may be partly affected by the acuity of illness, the type of illness, or the therapy that ensued in the first 24 hours. Demographic factors (age, sex, weight, presence of highly active antiretroviral therapy) have been shown to correlate with CD4 counts and may affect the accuracy of this prediction instrument.7,9 Because our population consisted primarily of African Americans, this may limit extension of our results to other ethnicities. However, previous work has demonstrated these covariates had a very modest affect on the ability to predict a “low” CD4.9 We believe that these differences were minimal and chose to evaluate this relationship independent of these factors, as a clinician would at the bedside. Last, generalizing these results to other EDs may be limited, as 85% of the patients had a CD4 of <200 × 106 cells/μL, although this may be partially reflective of the acuity of illness in an admitted HIV+ ED population.

Conclusions

Acute lymphocyte count was predictive of a CD4 count of <200 × 106 cells/μL in human immunodeficiency virus–positive patients who present to the ED necessitating hospital admission. A CD4 count of <200 × 106 cells/μL is very likely if the ED absolute lymphocyte count is <950 × 106 cells/μL and less likely if the absolute lymphocyte count is >1,700 × 106 cells/μL. Depending on pretest probability, clinical use of this relationship may help emergency physicians predict the likelihood of susceptibility to opportunistic infections and may help identify patients who should receive definitive CD4 testing.

Acknowledgments

The authors are indebted to Dr. Mark Smith for his development of the software necessary for completion of this work. His vision and continued support made this work possible.

Ancillary