Pregnancy and Other Factors Associated With Higher CD4+ T-Cell Counts at HIV Diagnosis in Southeast Michigan, 1992–2002

Abstract and Introduction

Abstract

Background

Early diagnosis of HIV infection provides the opportunity for treatment to prevent progression to AIDS and for intervention to prevent further transmission. The impact of routine screening of pregnant women and other factors on the stage of HIV disease at diagnosis were evaluated.

Methods

Data were collected in 1992–2002 from the medical records of persons presenting for HIV-related care at 2 major medical centers in Detroit, Michigan. Patients were included in the analysis if they had a CD4+ T-cell count recorded within 6 months of their first positive HIV test (N = 1858).

Results

Half of the patients (49%) had a first CD4+ T-cell count of < 200 cells/mm³ and 19% had an AIDS-defining illness at the time of HIV diagnosis. In the multivariate model, pregnant women were less likely than nonpregnant women to enter care with a CD4+ T-cell count of < 200 cells/mm³ (odds ratio, .24; 95% confidence interval, .14–.41). Even after adjusting for pregnancy, female sex was protective, as was age < 30 years. HIV-transmission risk factors, race, and time period of HIV diagnosis were not significantly associated with first CD4+ T-cell counts of < 200 cells/mm³.

Conclusion

Routinely offering HIV testing in prenatal care, as required by Michigan law, resulted in earlier diagnoses of HIV in pregnant women, as indicated by their higher CD4+ T-cell counts. Increasing routine HIV testing of all persons seeking medical care may increase the overall proportion of HIV diagnoses that are made early in the disease process.

Introduction

Increasing the proportions of HIV-positive persons who know their status and receive care is a strategic goal of the US Centers for Disease Control and Prevention (CDC).[1] As one of the tactics for achieving this goal, the CDC's recommendation on routinely offering HIV testing has recently been expanded to include all persons seeking healthcare in all clinical settings where the prevalence of HIV in the population is 1% or more.[2] Increasing the proportion of patients screened is expected to result in a larger proportion of HIV diagnoses being made at early stages of the disease. Early diagnosis of HIV benefits the patient because it provides the opportunity for treatment before the CD4+ T-cell counts drop below critical levels.[3–8] Early diagnosis of HIV also benefits the population because it provides the opportunity for intervention to decrease the risk of further transmission.[1,2]

Pregnant women in Michigan provided an opportunity to demonstrate the impact of routinely offering HIV testing. The Michigan law that requires healthcare providers to offer HIV testing to their pregnant patients at their first prenatal care visit, or at labor and delivery, became effective in 1989.[9,10] A high rate of compliance with this law was shown in 1998.[11] Approximately 80% of women who delivered live infants at a major medical center in Detroit reported being tested for HIV.[11] The data on pregnant women for the analysis presented here were collected at this same medical center, over a longer period, 1992–2002. The stage at which HIV infection was diagnosed in pregnant women was compared with other persons, for whom there was not a legal requirement for routinely offering HIV testing. This analysis also tested for associations of the year of first positive HIV test, sex, age, race, and HIV-transmission risk categories with the stage of HIV disease at diagnosis.

Methods

Population

Two observational databases were used for this analysis.

First, persons included in this analysis were selected from those enrolled in the Adult and Adolescent Spectrum of Disease (ASD) project. ASD is a national multisite, facility-based, longitudinal review of the medical records of HIV-infected persons, which was initiated in 1990.[12] Data for Michigan ASD were collected in Detroit at 2 large, urban medical centers, which provide primary and subspecialty care in outpatient clinics, emergency facilities, and hospitals, for a population of which 97% reside in Southeast Michigan. Prenatal care for 96% of the HIV-infected pregnant women in this analysis was provided at the same site used in the report on compliance with the Michigan statute on HIV testing of pregnant women.[11]

A stratified sample of all the persons who presented for HIV-related care at these sites were enrolled in ASD: all females age 13 and over, all males age 13–19 years, and a systematic random sample (40%) of new patients who were men aged >/= 20 years. The ASD database includes the date of the first positive HIV test recorded in the medical record, all CD4+ T-cell counts, and dates of pregnancy. Approval for ASD in Michigan was obtained from the institutional review boards of the Michigan Department of Community Health and the 2 medical centers where data were collected.

Second, data from the HIVAIDS Reporting System (HARS) were used to supplement the information in ASD, so that the earliest known positive HIV test date and the earliest known CD4+ T-cell count were used, even if these data were reported from a non-ASD site. HARS is the HIV/AIDS surveillance database in Michigan.[13] It includes the date of each person's first reported positive HIV test, the city and county of residence at the time of the first positive HIV test, and all reported CD4+ T-cell counts.

Inclusion Criteria

This analysis included the non-Hispanic black and white persons in ASD whose first reported positive HIV test occurred in 1992–2002, and whose first reported CD4+ T-cell count was within 6 months of their first reported positive HIV test. Persons in race categories other than black and white were excluded, because they constituted less than 5% of the ASD population and their numbers were too small to provide meaningful results.

Outcomes

First CD4+ T-cell count < 200 cells/mm, first CD4+ T-cell count < 50 cells/mm³, and definitive or presumptive diagnosis of any AIDS-defining illness[14] within a month before or after the first positive HIV test were used as indicators of advanced HIV disease. Any of these outcomes constitutes a diagnosis of AIDS, according to the CDC 1993 surveillance case definition.[14] In multivariate modeling, only the first CD4+ T-cell count of < 200 cells/mm³ was used as an outcome because of the comparatively small number of diagnoses of AIDS-defining illnesses.

Predictor Variables

Associations of stage of HIV disease with year of first positive HIV test, sex, race, age, pregnancy at the time of first CD4+ T-cell count, and mode of HIV transmission were studied. Age was divided into 2 groups, < 30 years and >/= 30 years, because in a preliminary analysis with 4 age categories, < 30, 30–39, 40–49, and >/= 50 years, the proportion of person with advanced HIV disease in the < 30 age group was significantly different from the others, whereas there were no significant differences among the 3 older age groups. The CDC-defined categories for mode of HIV transmission[15] were used with 1 modification. The heterosexual risk category included both the CDC-defined high-risk heterosexual category and the persons without an identified risk who had heterosexual contact.[16]

Data Analysis

Differences by year of first positive HIV test in the proportions of persons with CD4+ T-cell counts < 200 cells/mm³ within 6 months of their first positive HIV test, and in the proportions with diagnoses of any AIDS-defining illness[14] within 1 month of their first positive HIV test, were tested in logistic regressions without other covariates.

In univariate analyses, unadjusted odds ratios were calculated for first CD4+ T-cell counts < 200 cells/mm³ and for diagnoses of AIDS-defining illness, by sex, race, age group, pregnancy, and modes of HIV transmission.

Multivariate logistic regression models of first CD4+ T-cell counts < 200/mm³ over the predictor variables, sex, race, age group, pregnancy, HIV-transmission risk factors, and time period of first reported HIV test, were evaluated. The period of first reported HIV test was dichotomous, 1992–1995 vs 1996–2002, to compare outcomes before highly active combination antiretroviral therapy (HAART) was available with outcomes when HAART was available. Model fit was assessed with the −2 log likelihood method, and the most parsimonious model was selected. Interaction terms were not significant predictor variables and did not improve the fit of the model.

Software

Data analyses were performed with SAS/STAT Software (SAS Institute Inc., Cary, North Carolina) and Epi Info (CDC).

Results

The 2 medical centers where Michigan ASD data were collected provided HIV-related primary and subspecialty care for 69% of the 8307 residents of Southeast Michigan whose positive HIV status was reported in 1992–2002. Of the 2705 non-Hispanic persons with black or white race who were enrolled in ASD, 1858 met the additional inclusion criterion that their first CD4+ T-cell counts were within 6 months of their first positive HIV tests. The non-Hispanic persons with black or white race who did not meet this criterion were not significantly different from those included, with respect to sex, race, and age group (Table 1). However, injection drug users and persons with their first CD4+ T-cell counts >/= 200 cells/mm³ were more likely to be excluded than persons who did not have a history of injection drug use or whose first CD4+ T-cell counts were < 200 cells/mm³, respectively.

Table 1.

Demographic Characteristics and HIV-Transmission Risk Factors for Included vs Excluded Persons, Southeast Michigan, 1992–2002

		Included* (%) (n = 1858)	Excluded†† (%) (N = 847)	OR for Exclusion (95% CI)‡‡
Sex	Man	1111 (60)	489 (58)	Referent
	Woman	747 (40)	358 (42)	1.09 (.92–1.28)
Race	White	348 (19)	148 (17)	Referent
	Black	1510 (81)	699 (83)	1.09 (.88–1.34)
Age	< 30	451 (24)	183 (22)	Referent
	>/= 30	1407 (76)	664 (78)	1.16 (.96–1.41)
HIV-transmission risk factors§	IDU	558 (29)	335 (40)	Referent
	MSM	701 (38)	279 (33)	.65 (.54–.79)
	All heterosexual	381 (21)	149 (18)	.64 (.51–.81)
	Others	228 (12)	84 (10)	.60 (.45–.80)
First CD4+ T-cell count	>/= 200 cells/mm3	945 (51)	385 (56)	Referent
	< 200 cells/mm3	913 (49)	297 (44)	.80 (.67–.95)

^*Included, first CD4+ T-cell count within 6 months of first reported positive HIV test

^††Excluded, first CD4+ T-cell count missing (165 persons) or more than 6 months after first reported positive HIV test (682 persons)

^‡‡OR = odds ratio. The ratio of the odds of being excluded to the odds of being excluded in the referent group; 95% CI, 95% confidence interval. 95% CI's for sex, race, age, and first CD4+ T-cell count are from chi-square tests, and those for the transmission risk categories are from a logistic regression analysis of the transmission risk categories over the outcome. Results that are significant at the 95% level are in bold type.

^§IDU = injection drug users (including men who have sex with men as well as inject drugs); MSM = men who have sex with men; all heterosexual, both high-risk heterosexual (persons who have heterosexual contact with persons known to be HIV-positive or to have a risk factor for HIV infection) and presumed heterosexual (persons who have heterosexual contact and do not have any other known risk factors); others, persons with blood contact or no identified risk

Overall, 51% of the included persons had advanced HIV disease at the time of their first reported positive HIV tests, as indicated by one or more factors. Forty-nine percent had their first CD4+ T-cell counts < 200 cells/mm³; 27% had their first CD4+ T-cell counts < 50 cells/mm³; and 19% had diagnoses of an AIDS-defining condition within 1 month of the first reported positive HIV test.

These proportions did not change significantly over the period of 1992–2002. In any single year, 1993–2002, there were no significant differences compared with 1992 (for the various years, .05 < P < .69), and there were no significant linear trends over the time period (in chi-square tests for linear trend, .15 < P < .63).

A smaller proportion of women than men had advanced HIV disease at the time of diagnosis, and the proportion was even smaller among pregnant women than among their nonpregnant counterparts (Table 2). In the frequency distributions of the first CD4+ T-cell counts, it was evident that the counts for pregnant women were skewed toward higher values compared with nonpregnant women and men (Figure). For example, less than 3% of pregnant women had first CD4+ T-cell counts < 50 cells/mm³, in contrast to 25% of nonpregnant women and 31% of men (P < .0001). The mean first CD4+ T-cell count for pregnant women was the highest (475 ± 23.9 cells/mm³, mean ± standard error of the mean), followed by nonpregnant women (304 ± 11.9 cells/mm³), and then men (241 ± 7.5 cells/mm³).

Table 2.

Stage of HIV Disease at Time of First Positive HIV Test, Southeast Michigan, 1992–2002

		n	Proportion with CD4 < 200*	Unadjusted Odds Ratios for CD4 < 200 (95% CI)††	Proportion with O. I.‡‡	Unadjusted Odds Ratios for O. I. (95% CI)
Sex	Man (referent)	1111	54%	.59 (.49–.71)	23%	.52 (.40–.66)
Woman	747	41%	13%
Race	White (referent)	348	49%	1.03 (.81–1.30)	19%	.98 (.73–1.32)
Black	1510	49%	19%
Age	< 30 (referent)	451	35%	2.09 (1.68–2.61)	15%	1.46 (1.09–1.95)
>/= 30	1407	54%	20%
Pregnancy	Nonpregnant woman (referent)	611	47%	.19 (.11–.33)	16%	–§
Pregnant woman	120	14%	0%
HIV-transmission risk factors|	IDU	548	52%	Referent¶	18%	Referent¶
	MSM	701	53%	1.05 (.84–1.32)	23%	1.39 (1.05–1.85)
	All heterosexual	381	39%	.60 (.46–.79)	11%	.58 (.40–.86)
Others	228	47%	.84 (.61–1.14)	24%	1.50 (1.03–2.18)
Overall		1858	49%		19%

^*CD4 < 200, first CD4+ T-cell count < 200 cells/mm³

^††CI = confidence interval. 95% CI's are from chi-square tests; results significant at this level are in bold type.

^‡‡O. I. = opportunistic illness, ie, diagnosis within a month of first positive HIV test of any AIDS-defining illness according to the 1993 US Centers for Disease Control and Prevention (CDC) surveillance case definition of AIDS.[13]

^§No pregnant women had a history of AIDS-defining illness.

^|IDU = injection drug users (including men who have sex with men as well as inject drugs); MSM = men who have sex with men; all heterosexual, both high-risk heterosexual (persons who have heterosexual contact with persons known to be HIV-positive or to have a risk factor for HIV infection) and presumed heterosexual (persons who have heterosexual contact and do not have any other known risk factors); others, persons with blood contact or no identified risk

^¶Odds ratios and confidence intervals from the logistic regression comparing each transmission-risk category with the referent, IDU, without other covariates

Figure.

Frequency distributions of the first CD4+ T-cell counts of HIV-positive men, pregnant women, and nonpregnant women who had their first CD4+ T-cell counts within 6 months of their first reported positive HIV tests, Southeast Michigan, 1992–2002 (N = 1858). The CD4+ T-cell count categories shown in the legend are in cells/mm³.

Age < 30 years and heterosexual HIV-transmission risk were associated with earlier stages of HIV disease at diagnosis in the univariate analyses (Table 2). The unadjusted odds ratio for race was not significant at the 95% level.

Multivariate modeling of the outcome, CD4+ T-cell count < 200 cells/mm³, provided odds ratios for sex, race, age, and pregnancy that were adjusted simultaneously for all 4 of these predictor variables (Table 3). In the final model, female sex, age < 30 years, and pregnancy were associated with lower odds of having first CD4+ T-cell counts < 200 cells/mm³. Persons of white race had lower odds than persons of black race, although the odds ratios were not statistically significant at the 95% level. HIV-transmission risk factors and time period of first positive HIV test (1992–1995 vs 1996–2002) were not significantly associated with first CD4+ T-cell counts < 200 cells/mm³ in the multivariate model, and omitting them did not alter the fit of the model. Collinearity between pregnancy and age or sex did not affect the adjusted odds ratios or their confidence intervals, as shown in analyses of subgroups by sex and pregnancy (results not shown).

Table 3.

Multivariate Logistic Regression Model of First CD4+ T-Cell Count < 200 cells/mm³, Southeast Michigan, 1992–2002 (N = 1858)

Predictor Variable	Value (Referent)	Adjusted Odds Ratio (95% CI)*
Sex	Woman (referent, Man)	.69 (.56–.85)
Race	Black (referent, White)	1.25 (.98–1.60)
Age	>/= 30 (referent, < 30)	1.81 (1.44–2.27)
Pregnancy	Pregnant (referent, not pregnant)	.24 (.14–.41)

CI = confidence interval

^*95% C.I., 95% confidence interval. Results that are significant at the 95% level are in bold type.

Discussion

The utility of routine, voluntary screening in diagnosing HIV at an early stage was demonstrated in HIV-infected pregnant women presenting for testing and care. Their higher CD4+ T-cell counts compared with nonpregnant women and men were interpreted as indicating an earlier stage of HIV disease, not an effect of pregnancy on CD4+ T-cell counts, because pregnancy in HIV-infected women has not been shown to be associated with consistent changes in CD4+ T-cell counts compared with prepregnancy values.[17–24] The early diagnosis of HIV infection in pregnant women was consistent with previous reports that HIV-infected women with early, asymptomatic HIV disease were more likely to be pregnant than those who had advanced disease.[25–27] Because pregnant HIV-infected women were less likely to have advanced HIV disease, routine screening provided the opportunity to diagnose their HIV infections at an early stage.

The earlier diagnoses among pregnant women partially accounted for the earlier diagnoses among women compared with men, overall. Even after adjusting for pregnancy, however, the odds of having a first CD4+ T-cell count < 200 cells/mm³ were lower among women. This result was consistent with earlier reports from a variety of populations, times, and places, although these studies did not adjust for pregnancy.[28–35] One factor that contributed to this difference was that women in general have higher CD4+ T-cell counts than men. However, the 13% difference in mean CD4+ T-cell counts between HIV-negative women and men[36] did not entirely account for the 28% difference between the mean CD4+ T-cell counts of HIV-positive women and men in this analysis. Further study is needed to identify additional factors that may contribute to the higher CD4+ T-cell counts of HIV-infected women presenting for testing and care.

The results for age, race, and HIV-transmission risk factors were consistent with some previous reports but not others. The association of age >/= 30 years with higher odds of first CD4+ T-cell counts < 200 cells/mm³ has been reported,[28,32,33,37] although it has not been found in all studies.[38] The marginally significant association of black race with first CD4+ T-cell counts < 200 cells/mm³ in the multivariate model suggested that this population may be similar in this respect to the population studied at the University of Nebraska at Omaha, 1996–2001,[38] where a similarly marginal disparity was found in a multivariate analysis. In contrast, the unadjusted odds ratios did not show an association of race with later stages of HIV disease at diagnosis, in this analysis or in previous reports.[32,35] HIV-transmission risk factors not having statistically significant associations with CD4+ T-cell counts < 200 cells/mm³ in the multivariate model suggested that the significant unadjusted odds ratios were due to confounding, for example, by sex. Previous reports have varied in their conclusions about whether persons with different HIV-transmission risk factors presented for testing and care at different stages of disease.[32,33,37]

Limitations and Generalizability

ASD was representative of the HIV-infected population in Southeast Michigan (HARS) with respect to sex, race, age, and transmission risk factors (unpublished data). However, this analysis may have preferentially included persons who were at more advanced stages of HIV disease at the time of their first reported positive HIV tests than other HIV-infected persons in care in Southeast Michigan. One source of this bias was the preferential inclusion of persons with first CD4+ T-cell counts < 200 cells/mm³ that resulted from the requirement for a CD4+ T-cell count within 6 months of first reported positive HIV tests. This bias may have been compounded by preferential inclusion of persons with more advanced disease in the patient populations of the centers where ASD data were collected, because they provided specialized HIV care as well as primary care. In addition, a person's first positive HIV test might not have been reported, and they might have been inadvertently included in the analysis at a later stage of disease on the basis of their first positive HIV tests that were reported and followed by CD4+ T-cell counts within 6 months. However, the potential effects of these selection biases were limited because ASD represented such a large proportion of residents of Southeast Michigan whose positive HIV tests were reported to HARS.

The results of this analysis of Detroit data may not be generalizable to the rest of the state. Although compliance with the Michigan law requiring that HIV testing be routinely offered in prenatal care or in labor and delivery has been high in Detroit,[11] the same may not be true in other parts of the state.

Clinical and Public Health Significance

The incidence of AIDS and the numbers of HIV-related deaths have reached virtual plateaus since 1998, in contrast to the rapid declines that followed the introduction of HAART in 1996.[39,40] These plateaus can be partially accounted for by the lack of change over time in the proportion of HIV-infected persons who already had advanced disease when they were tested and entered care. This lack of change was found in previous reports from the United States, England, and Wales,[28,32] as well as in this analysis.

The substantial proportion of HIV-infected persons who presented for testing and care with CD4+ T-cell counts < 50 cells/mm³ and the lack of change in this proportion over time are of particular concern. Individuals with lower pretreatment CD4+ T-cell counts have been shown to have poorer responses to antiretroviral treatment.[4–8] These patients are less likely to attain a CD4+ T-cell counts above 200 cells/mm³ and, therefore, may be at prolonged risk of opportunistic diseases.

The proportion of HIV diagnoses made at an early stage of disease is expected to increase as the recent CDC recommendation to increase routine HIV screening[1,2] is implemented and the stigma attached to having an HIV test is decreased.[41] An increase in the proportion of persons tested when HIV testing was routinely offered has already been reported.[42,43] The additional persons tested, such as the pregnant women in this analysis, are expected to be at lower risk of having advanced HIV disease. Routine screening will be especially beneficial for persons over age 30, because they currently present for testing and care at later stages of HIV disease than younger persons.

The impact of earlier diagnosis on HIV-associated morbidity and mortality will depend on the rate at which the persons with newly diagnosed HIV infection enter care. The pregnant women in this analysis might have been highly motivated to enter care because of concern for their unborn children, and new approaches may be needed to increase the care-seeking motivation of other persons with newly diagnosed HIV infections.

Funding Information

The collection of ASD and HARS data and this analysis were supported by cooperative agreement number CCU 506 228 from the US Centers for Disease Control and Prevention (CDC), Atlanta, Georgia.