The economic benefits of high CD4 counts among people living with HIV/AIDS in Zambia

Abstract

Background

The economic effects of poor immunologic recovery among HIV-infected patients receiving antiretroviral therapy (ART) in sub-Saharan Africa are not well understood. We examined the relationship between the CD4 counts of patients on long-term ART and employment outcomes in HIV-affected households in Lusaka, Zambia.

Methods

Administrative data and a household survey captured information on the clinical records, demographics and employment outcomes of the ART-treated adults and their adult family members (n = 311). Multivariable regression analyses were used to assess relationships between CD4 counts of ART-treated adults and household employment outcomes.

Results

Patients with a CD4 count of at least 350 cells/µl were 22 percentage points more likely to be engaged in the labor force (P < 0.05) and worked ∼6 more days per month (P < 0.05) and 9 more hours per week (P = 0.05) compared with patients with a CD4 count <350 cells/µl. Non-patient adults in the HIV-affected household had significantly higher labor participation if the patient's CD4 count was ≥500 compared with <500 cells/µl (P < 0.05), but this was not significant for a CD4 ≥350 versus <350.

Conclusion

These findings suggest that interventions to improve or maintain robust immune recovery during ART may confer economic benefits for both HIV-infected individuals and HIV-affected households.

ART, CD4 count, employment, household, HIV/AIDS

Introduction

In addition to a profound toll in morbidity and mortality, the HIV epidemic in sub-Saharan Africa has had severe economic consequences for HIV-infected individuals, their families and national economies.^1–3 Progressive debilitation from untreated HIV disease and the associated burden of caring for chronically ill family members reduce the capacity for labor force participation and economic productivity among working age adults, which negatively impacts the household through reduced income.^2,4–6 However, with the rapid scale up of access to antiretroviral therapy (ART) in sub-Saharan Africa over the past decade, patients are now surviving far longer on treatment.

Prior studies report increased labor force participation, income and other economic benefits among HIV-infected individuals starting treatment, and in South Africa, a large longitudinal analysis found that individuals on ART for 5 or more years returned to the level of employment observed prior to the onset of advanced HIV disease.^7–10 While these studies indicate that ART has a profound effect on the capacity for labor, the effect of immune status after treatment initiation on economic outcomes at the individual and household level is not as well described.

A recent study in Uganda of employment among HIV-infected adults (37% on ART) found that those with a CD4 cell count over 350 cells/µl worked an average of 6 more days per month compared with patients with a CD4 count under 200 cells/µl.¹¹ A similar study, also from Uganda, found lower labor force participation among HIV-infected adults not yet on ART with a CD4 count <200 cells/µl compared with patients above that value.¹² However, after 1 year of ART, the labor participation of those with a pretreatment CD4 count below 200 cells/µl had reached parity with those starting at a higher CD4 count.

The employment status of a HIV-infected individual may not be representative of the economic welfare of their HIV-affected household. Prior studies have assessed the relationship between CD4 count and patient employment outcomes prior to or shortly after ART initiation, but there are no similar assessments for patients on long-term ART and other adult household members. At present, it is unclear whether poor immune recovery in an ART-treated, HIV-infected individual reduces the labor force participation of family members, possibly due to a higher care burden, or conversely requires family members to work more to maintain the household economic welfare. Using clinical and economic data collected from HIV-infected patients on long-term ART and their households, we assessed whether patient CD4 count was associated with labor force participation in HIV-affected households in Lusaka, Zambia.

Methods

Participant recruitment

Participant enrollment occurred over the month of August 2009. The collection of clinical and economic data for this analysis was performed as part of an impact assessment for a World Food Program-Zambia (WFP) initiative to provide food baskets to HIV-affected individuals and households in Lusaka, Zambia.¹³ The patients in this analysis comprise the control cohort at clinics not involved in the food distribution program. Research field staff were assigned to the four study clinics located in the Bauleni, Chipata, Matero and Chilenje neighborhoods of Lusaka and randomly selected the medical records of adult (≥18 years of age) patients who presented for a clinic visit and had documentation of ART start date prior to January 2009 (i.e. on treatment for over 6 months). Patients were approached and asked to participate in the study, and the refusal rate was <5%. Recruitment at each clinic concluded when 50 patients were enrolled, which comprised ∼7% of the active ART patient population at the clinics.

Data collection

All participants completed a household survey to assess demographics, employment outcomes, expenditures, income and asset wealth. Clinical data on duration of ART treatment and CD4 counts were obtained from the SmartCare electronic database of the Zambian national HIV treatment program. The window period for CD4 counts was prespecified as the value closest to 1 August 2009 within 60 days before or after this date, an approach used by previous similar studies.^7,14

Key variables

The survey had a module to capture information on formal or informal income-generating work performed by household members over age 18. Examples included farm work, owning a business of any size and salaried work. Three variables were used to measure employment outcomes. One variable was binary, and it measured participation in any income-generating work at the time of the survey. Two were continuous variables, the number of hours worked in the previous week and the number of days worked during the previous month.

Statistical analyses

We used a binary classification of CD4 ≥ 350 cells/µl and CD4 < 350 cells/µl, which was based on the 2010 and 2013 WHO recommendations for ART treatment initiation,^15,16 clinical trials and observational studies showing worse health outcomes among patients with a CD4 count of <350 cells/µl at ART initiation^17–19 and to permit comparisons with prior studies of economic outcomes.¹¹

Ordinary least squares regressions were used to determine the association between the CD4 count of ART-treated adult patients and the continuous outcome variables (number of hours worked in previous week and number of days worked in previous month). A probit model was used to examine the association between the CD4 count of ART-treated adult patients and the binary outcome variable (i.e. whether the patient participates in any income-generating work). A probit regression was utilized to express the effect size in percentage points or marginal effects, which provides a more straightforward interpretation of labor force participation rates, as opposed to odds ratios. All regression models were adjusted for age, sex, an interaction between age and sex, duration of ART and education attainment (i.e. some formal education versus none). We utilized the same models to assess the relationship between the ART patient's CD4 count and employment outcomes among non-patient adults in the HIV-affected household.

The regression models were also used to obtain the predicted values of the employment outcomes. Non-parametric regression (kernel-weighted local polynomial regression with a zero degree polynomial) examined the association between the duration of ART and the predicted values of the three employment outcomes by comparing individuals with CD4 ≥ 350 and CD4 < 350. These analyses assessed whether the duration of ART mediates the influence of CD4 count on employment. All statistical analyses were performed using Stata/SE 13.1 (StataCorp, College Station, TX, USA).

Ethical approval

Ethical approval was obtained from the University of Zambia Research Ethics Committee and the Ministry of Health of the Republic of Zambia. All participants provided written informed consent.

Results

The analytical sample is composed of 311 individuals, comprising 112 adult patients receiving ART and 199 non-patient adults. Of the 201 participants enrolled based on the clinic paper medical record, 179 were found to meet inclusion criteria after the electronic medical record was reviewed. The 22 exclusions were due to age <18 years (n = 1), medical record number from clinic chart not found in the national electronic database (n = 11), clinic visit data present but incomplete in the electronic database (n = 9) or the participant was concurrently enrolled in the WFP food supplementation program (n = 1). Of the remaining 179 patients, 112 (63%) had CD4 count information within the window period and, together with 199 non-patient adult family members, formed the basis of the household analytical sample. Compared with patients included in the analysis, those missing a CD4 count had comparable age, education attainment and gender, but a longer median of duration on ART (973 versus 1415 days, P < 0.05). Table 1 describes the socio-demographic characteristics of the cohort. About 70% of the patients were female compared with nearly 47% of the non-patient adults. On average, patients were older than non-patients by 10 years. The marriage and educational attainment rates of the patients and non-patients are similar. Among the patients receiving ART, the median CD4 count was 349 cells/µl, and the median duration of ART was 973 days.

Table 1

Description of the participant cohort

Characteristics	Adult patients on ART (n = 112)	Non-patient adults in the same household (n = 199)
Female, %	71	47
No education, %	15	15
Married, %	46	46
Age, mean, years	40.9	30.6
CD4 count, cells/µl
≥350, %	49	NA
Mean (SD)	373 (190)
Median (IQR)	349 (245–481)
Duration of ART, days
Mean (SD)	978 (505)	NA
Median (IQR)	973 (523–1434)

IQR, inter-quartile range; ART, antiretroviral therapy.

Descriptive statistics of employment outcomes

Table 2 shows the means of each employment outcome for patients and non-patients. About 47% of patients with CD4 ≥ 350 cells/µl participated in income-generating work compared with around 32% of patients with CD4 < 350. The mean number of days worked in the past months were higher for patients with CD4 ≥ 350 (16.4 days) compared with those with CD4 < 350 (12.0 days).

Table 2

Comparison of household demographics and employment outcomes according to HIV-infected patient CD4 counts

Households
Variables	CD4 ≥ 350 cells/μl (n = 55)	CD4 < 350 cells/μl (n = 57)	P-value
Female, %
Patients	84	58	<0.001
Adults living in same household	45	49	0.54
No education, %
Patients	9	18	0.19
Adults living in same household	11	21	0.05
Married, %
Patients	42	49	0.44
Adults living in same household	38	56	0.05
Age, mean years
Patients	40.9	41.0	0.94
Adults living in same household	31.5	29.4	0.24
Participated in income-generating work, %
Patients	47	32	0.09
Adults living in same household	38	30	0.21
Days worked in previous month (mean)
Patients	16.4	12.0	0.09
Adults living in same household	12.2	10.8	0.59
Hours worked in previous week (mean)
Patients	18.0	13.0	0.26
Adults living in same household	10.2	12.2	0.51

Association between CD4 count and employment outcomes of HIV patients and adult family members

Patients with CD4 ≥ 350 cells/µl were 22 percentage points more likely to be in the labor force than those with CD4 < 350 (95% confidence interval [CI]: 0.02, 0.42, P < 0.05) (Table 3). Given the average labor force participation rates for patients with CD4 < 350 (32%), this result implies that labor participation rates of patients with CD4 ≥ 350 are ∼69% higher. Patients with CD4 ≥ 350 worked 5.97 more days in the previous month than those with CD4 < 350 and 9.06 more hours in the past week than those with CD4 < 350 (P < 0.05 for both). This implies that patients with CD4 ≥ 350 worked ∼50% more days and nearly 70% more hours. When CD4 was treated as a linear variable, each 100 cells/µl increase was associated with a 10 percentage point greater likelihood of labor force participation by patients (P < 0.05) and 4 more hours worked in the previous week (P < 0.01, data not shown).

Table 3

Relationship between HIV-infected patient's CD4 count and household employment outcomes

Outcome	(1)		(2)		(3)
	Labor force participation		Days worked in previous month		Hours worked in previous week
	ME	95% CI	β	95% CI	β	95% CI
Patients
CD4 ≥350 cells/µl	0.22**	0.02,0.42	5.97**	0.47,11.47	9.06*	−0.02,18.14
Age, per year	−0.012	−0.03,0.01	−0.67**	−1.24,−0.10	−0.75	−1.71,0.22
Female	−0.71**	−1.34,−0.08	−38.7***	−67.86,−9.46	−56.8**	−105.98,−7.61
Female ×age	0.02	−0.01,0.04	0.82**	0.15,1.49	0.95*	−0.17,2.08
Days on ART	2.44E − 06	−0.0002,0.0002	0.001	−0.004,0.01	0.003	−0.01,0.01
No education	−0.19	−0.42,0.04	−4.33	−11.26,2.60	−3.59	−15.17,8.00
Observations	112		106		112
Non-patient adults
CD4 ≥ 350 cells/µl	0.05	−0.10,0.20	1.39	−4.12,6.90	−1.07	−7.55,5.41
Age	0.01**	0.00,0.02	0.17	−0.11,0.45	0.42**	0.08,0.76
Female	−0.19	−0.54,0.16	0.64	−12.61,13.89	−1.51	−17.09,14.06
Female × age	−0.0006	−0.01,0.01	−0.12	−0.51,0.28	−0.15	−0.63,0.32
Days on ART	−0.00004	−0.00,0.00	−0.003	−0.01,0.00	−0.006*	−0.01,0.00
No education	−0.18**	−0.34,−0.01	−6.34*	−13.48,0.79	−1.74	−9.98,6.50
Observations	198		182		198

No education refers to whether individual was never formally educated. Female × age is an interaction of female and age.

ME, denotes marginal effects.

*P < 0.10, **P < 0.05, ***P < 0.01.

Table 3 also presents the results for the non-patient adult members living in the households with ART-treated adult patients. The results show that living with a patient with CD4 ≥ 350 was associated with a greater likelihood of labor force participation and 1.39 more days worked in the previous month. However, these effects were not statistically significant. In a sensitivity analysis, we adjusted for the residential community of the households, to control for any unobserved location-specific confounders, but the results for both patients and non-patients did not change.

Non-parametric regressions were used to assess whether the association between a high CD4 count and predicted employment outcomes varies by the length of time on ART (Fig. 1). At all points along the distribution of days receiving ART, patients with CD4 ≥ 350 consistently had higher probabilities (above 50%) of labor force participation, and more days and hours worked than those with CD4 < 350. The gap between patients with CD4 ≥ 350 and those with CD4 < 350 is largely similar at both the low and high end of the distribution of ART duration, suggesting time on treatment was not a major confounder.

Fig. 1

Role of the duration of ART on patient employment: non-parametric regression estimates. Results from kernel-weighted local polynomial regressions (zero degree polynomial) with width of 200 days around each point and estimated locally at 50 points. Regressions compare patients’ high CD4 count (CD4 ≥ 350 cells/µl = 1) and those with low CD4 count (CD4 < 350 cells/µl = 1). Prob, probability.

We conducted sensitivity analyses to assess the effects of a higher CD4 count of 500 cells/µl on employment outcomes and the effect of adjusting for the patient's medication possession ratio (MPR). The MPR is a measure of adherence (the number of pills consumed compared with the number prescribed) based on pharmacy refill data, which is associated with the likelihood of HIV-1 virus suppression in the blood.^20,21

ART-treated patients with CD4 ≥ 500 were 26 percentage points more likely to be in the labor force than those with CD4 < 500 (81% more, P < 0.05; Supplementary data, Table). Patients with CD4 ≥ 500 worked 13.2 more hours in the past week than those with CD4 < 500 (101% more, P < 0.05 both). Non-patient adults living with a patient with CD4 ≥ 500 were also 27 percentage points more likely to be in the labor force than those with living with a patient with CD4 < 500 (90% more, P < 0.01). They also worked 4.3 more days in the previous month and ∼0.8 more hours in the previous week, but these differences were not statistically significant. When MPR was included in the models assessing a CD4 ≥ 350 versus CD4 < 350, the results were similar to those from the primary analysis in Table 3.

In an analysis which assessed the collective household rather than per capita employment outcomes of non-patient adults in each household, we found that a CD4 ≥ 500 cells/µl was accompanied by an increase in collective labor participation of 40 percentage points (P < 0.1), but no statistically significant increase in total hours and days worked. Furthermore, a CD4 ≥ 350 was associated with a statistically insignificant increase in collective employment outcomes of non-patient adults. While the results for the collective household analyses found many of the same directional relationships as the per capita analyses, the failure to detect similar statistical associations may be affected by the smaller sample size.

Discussion

Main findings of the study

To our knowledge, this is the first study to assess the effect of CD4 counts on labor force participation among patients on long-term ART. We found working-age adults on long-term ART in urban Lusaka, Zambia, with a higher CD4 count had improved employment outcomes. After accounting for the duration of ART treatment and other factors, we found that ART-treated patients with CD4 counts above 350 cells/µl were 22 percentage points more likely to be in the labor force, and they worked ∼6 more days per month and 9 more hours per week than patients with a CD4 count below this threshold. Sensitivity analyses showed that a CD4 count over 500 cells/µl was also associated with similar work rates for the patients.

We also assessed the relationship between HIV patient CD4 count and the employment outcomes of non-patient adults in the same household. We observed a positive relationship between patient CD4 count (above 350 cells/µl) and labor force participation of non-patient adults, but the results were not statistically significant. However, living with a patient with a CD4 count over 500 cells/µl, rather than over 350 cells/µl, was associated with higher and statistically significant labor force participation for non-patient adults in the household. Lastly, an analysis that used collective household employment outcomes of non-patient adults found a CD4 count above 500 cells/µl was associated with an increase in labor force participation. These findings may indicate that as the HIV-infected family member's immune status improves above 350 cells/µl, other adults shift from providing care to income-generating work.

What is already known on this topic?

Our findings are similar to a recent study of CD4 count and labor force participation in 168 HIV-infected Ugandan adults, the majority of whom (63%) had not yet started ART.¹¹ In that study, patients with a CD4 count over 350 cells/µl worked an average of 6 more days per month than patients with a CD4 count less than 200 cells/µl, and those with a CD4 count over 500 cells/µl worked 44% more hours per week compared with those less than 200 cells/µl.¹¹ A second study in 505 HIV-infected but ART-naïve Ugandan adults found lower labor force participation among those with a CD4 count less than 200 cells/µl compared with patients above that value.¹² However, after 1 year of ART, those who started treatment below 200 cells/µl had reached a similar level of employment as those starting at a higher CD4 count, and both groups maintained similar trajectories in labor participation going forward.

Studies of untreated HIV infection found that functional status and labor force participation declined rapidly in the last 2 years before an employee stopped working due to AIDS.^2,4 At the household level, the presence of a family member with advanced HIV reduced yearly income by 30–35% in one study⁶ and resulted in markedly lower agricultural output in farming families.^22,23 At the health system level, patients with more advanced immunosuppression at ART initiation generate greater overall costs.^24,25 However, a large longitudinal study of socioeconomic survey data from South Africa, an upper middle income country by World Bank criteria, found that while HIV-infected individuals experienced a sharp decline in employment in the year prior to ART initiation, an individual's employment level 5–8 years after ART initiation was not significantly different from the period 3–5 years prior to ART initiation.¹⁰

What this study adds?

Unlike prior studies that focused only on ART patients, our study found that family members of HIV-infected members had better employment outcomes as the ART patient's CD4 count rose above 500 cells/µl. This implies that disease severity in the ART patient can have broader intra-household economic effects through changes in employment among other adult family members, and that interventions to promote robust immune recovery, such as routine virologic monitoring, may have household economic benefits.

Current guidelines for the initiation of ART in Zambia are a CD4 count less than 350 cells/µl or Stage 3 or 4 HIV disease, but most HIV-infected individuals present for care in Zambia with CD4 counts below this level, and up to a third have malnutrition and/or advanced immunosuppression at ART initiation.^26,27 Because our study did not enroll patients in the pre-ART period, we cannot conclude that initiating ART at a CD4 count over 350 cells/µl (the current guidelines in Zambia) would have beneficial economic effects. However, our results suggest that treatment of HIV prior to advanced immunosuppression confers an economic benefit, and further studies in this area are warranted.

Our finding of an association between immune recovery and HIV-affected household economic welfare in Zambia raises additional questions for future research. In particular, the role of pre-ART health and economic factors in post-ART outcomes, and the possible social, behavioral and physiologic linkages between immune recovery and labor productivity should be explored further. Both pre-ART health status and economic status could have implications for post-ART CD4 recovery and employment outcomes. For example, undiagnosed mycobacterial infections of the gastrointestinal tract in advanced AIDS may cause a persistent diarrhea that reduces both the absorption and tolerability of ART medications, and could prevent the nutritional recovery and improvement in functional status necessary for labor productivity. On the other hand, patients with greater economic resources and skills pre-ART may be more able to attend clinic appointments, understand medication schedules and maintain adherence, purchase sufficient food to recover from nutritional deficits and utilize skills and networks developed pre-ART to find employment after treatment initiation.

Limitations of this study

Our study had several limitations. The cross-sectional design precluded the assessment of causal relationships between CD4 count and labor force participation. The lack of data on pre-ART employment and health status precluded an assessment of how these factors affected the trajectory of CD4 recovery on ART and economic outcomes. Plasma HIV-1 RNA measurements were not a routine component of care in Zambia at the time of the study, and therefore, we were unable to ascertain whether poor CD4 cell recovery despite a relatively long duration of ART was due to immunologic non-response or inadequate viral suppression. Our sensitivity analysis adjusted for the MPR, but this metric is a programmatic tool based on pharmacy refill data that cannot capture individual patient behavior. The sharing of medications, loss of tablets or inability to take tablets on schedule would all affect virologic suppression status, and therefore CD4 recovery, but might not alter the MPR. Poor immune recovery and persistent immune activation despite the suppression of circulating virus are associated with functional impairment and frailty in HIV-infected adults, though the directionality of this relationship is unclear.^28,29 Future studies should assess whether measurements of virologic suppression and functional capacity after ART initiation (e.g. grip strength or lean mass recovery) can predict patient and household employment outcomes better than CD4 count.³⁰

Among non-patient adults living in the HIV-affected household, we did not assess HIV status due to concerns regarding privacy, response bias and the potential for undiagnosed HIV infections in our setting. Lastly, our study cohort was primarily female and drawn from an urban population in Zambia, and the results may not be representative of men, rural populations or other countries in the region.

Conclusion

We observed a strong association between a CD4 count above 350 cells/µl and the labor force participation of patients on long-term ART in Zambia, and a similar positive relationship for adults living in the HIV-affected household that was only significant for a CD4 count over 500 cells/µl. These findings have implications for testing and linkage to care of HIV-infected persons in sub-Saharan Africa. Treatment initiation at higher CD4 counts could have economic benefits at the household level, but additional longitudinal studies are needed to understand how HIV disease severity, immune recovery, functional status and comprehensive measurements of socioeconomic outcomes interact across the continuum from early HIV infection through long-term ART treatment.

Authors' contributions

N.T. and J.R.K. were responsible for the study design. N.T. was responsible for the survey data collection activities while J.R.K. was responsible for obtaining clinical data. N.T. performed the statistical analysis, with input from J.R.K. N.T. and J.R.K. drafted the article and approved the final version.

Supplementary data

Supplementary data are available at PUBMED online.

Funding

This work was supported by the NIH Fogarty International Center (grant number R24-TW007988), the UNAIDS Programme Acceleration Fund, the World Health Organization, the Ford Foundation and the Poverty, Equity and Growth Network. The funding bodies were not involved in the study design, data collection, analysis, interpretation or manuscript preparation.

Acknowledgements

We thank the Zambian Ministry of Health for their consistent support of operational HIV research. We also thank the World Food Program Regional and Zambia offices, the Programme for Urban Self Help and the medical staff at the public clinics and the HIV-infected adult patients and their families for participating in this study.