HIV and AIDS-Sci Forschen

Full Text

Research Article
Factors Associated with ART Outcomes among Adult (15+) Persons Living with HIV in Zimbabwe in the Multi-Month Scripting (MMS) Regime

  Hamfrey Sanhokwe1      Patrick Shabangu2      Fastel Chipepa1*   

1Department of Applied Mathematics and Statistics, Midlands State University, Zimbabwe
2Country Director, Institute for Health Measurement, Mbabane, Swaziland

*Corresponding author: Fastel Chipepa, Department of Applied Mathematics and Statistics, Midlands State University, Zimbabwe, Tel: +2687 76027521; E-mail:


Acquired Immuno Deficiency Syndrome (AIDS) continue to be a major global public health concern. There are an estimated 1.3 million people living with in Zimbabwe and 1,100,000 million are estimated to be on Antiretroviral Therapy (ART) by 2018, which translates to over 80% ART coverage [1]. This study explores the factors associated with ART outcomes among adult (15+) persons living with HIV in Zimbabwe.

A synopsis from literature on the subject

Below is a summary of the factors from prior studies across the globe.

Factors associated with treatment failure: Several studies have explored the factors associated with ART outcomes across the globe. The United States’ Department of Health and Human Services (2013) reported that treatment failure may occur in PLHIV due to various risk factors. These include poor adherence to treatment, poor absorption of ARVs, previous treatment failure, drug resistance, comorbidities, drug toxicity and drug interactions, poor health prior to initiation of ART, as well as substance abuse (e.g. tobacco smoking or excessive alcohol consumption) leading to poor adherence. Other studies [2-4] also seem to suggest that other factors such as advanced HIV disease, gender, ARV regimens low baseline CD4, age, and long periods on ART are strongly associated with treatment failure.

A study by Haile, et al. [5] investigated the predictors of treatment failure among adult ART clients in the Bale Zone Hospitals, South Eastern Ethiopia. The study used a retrospective cohort study from four hospitals of Bale zone. The study results showed that male ART clients were more likely to experience treatment failure as compared to females [AHR=4.49; 95% CI: (2.61 ± 7.73)]. In addition, lower CD4 count (<100 m3 /dl) at initiation of ART was found to be significantly associated with higher odds of treatment failure [AHR=3.79; 95% CI: (2.46 ± 5.84)]. Similarly, bedridden [AHR=5.02; 95% CI: (1.98 ± 12.73)] and ambulatory [AHR=2.12; 95% CI: (1.08 ± 4.07)] patients were more likely to experience treatment failure as compared to patients with working functional status. As expected, TB co-infected clients also had higher odds of experiencing treatment failure [AHR=3.06; 95% CI: (1.72 ± 5.44)]. The study further showed that patients who developed TB after ART initiation had higher odds to experience treatment failure as compared to their counter parts [AHR=4.35; 95% CI: (1.99 ± 9.54]. Having other opportunistic infections during ART initiation was also found to be associated with higher odds of experiencing treatment failure [AHR=7.0, 95% CI: (3.19 ± 15.37)]. Having fair [AHR=4.99 95% CI: (1.90 ± 13.13)] and poor drug adherence [AHR=2.56; 95% CI: (1.12 ± 5.86)] were significantly associated with higher odds of treatment failure as compared to clients with good adherence.

Overall, the study by Haile, et al. [5] shows two sets of variables: one with narrower confidence intervals (implying less variability) e.g., having a lower CD4 count, ambulatory status, TB co-infection and poor drug adherence. The other set of variables have wider confidence intervals (implying higher variability) e.g., bed ridden, developing TB after initiation, other OIs and fair drug resistance. This study explores some of the major factors being investigated in this study and provides critical insights into what to potentially expect. That it is a recent study adds value to what is being investigated in this study. However, the study report does not show whether these findings are reflective of a post multi-month scripting era or not. Nonetheless, other reviews done by this researcher show that Ethiopia has adopted the differentiated models of care.

Factors associated with survival: A retrospective study conducted by Ram Bajpai, et al. [6] assessed the survival rates and factors associated with survival among adult PLHIV in Andhra Pradesh, India. This research piece used data from 139 679 PLHIV aged ≥ 15 years on ART, registered between 2007 and 2011. These were followed up through December 2013. The outcome of interest was death of the client. The Kaplan-Meier was used to estimate survival, while the Cox-regression models were used to explore the factors associated with survival.

The study results show that approximately 13% of those newly initiated on ART died during follow-up with 56% of all deaths occurring within the first three months. From the study, the CD4 count (adjusted hazard ratio of 4.88; 95% confidence interval of 4.36 to 5.46 for <100 cells/mm3 vs. >350 cells/mm3 ); functional status (adjusted hazard ratio of 3.05; 95% confidence interval of 2.82 to 3.30 for bed ridden vs. normal), and body weight (adjusted hazard ratio of 3.69; 95% confidence interval of 3.42 to 3.97 for <45 kg vs. >60 kg) were strongly associated with the survival of HIV patients. This study by Ram Bajpai, et al. [6] shares a lot of methodological similarities with the study undertaken by this researcher. This is exemplified by the following: a retrospective design, a relatively long follow up (which however is only half of what my study is looking at), and significantly larger sample size. It is also worth recognizing the narrower confidence intervals in this study (reflective of less variability owing to a large sample, among other factors). However, this study explores only a limited set of outcomes (primarily survival) in comparison to what this study is investigating. Nonetheless, it provides a rich repository for the current study.

Factors associated with immunological failure: Virologic failure remains one of the major ART outcomes for PLHIV on ART. Evidence shows that there is a myriad of factors associated virologic failure. These include, but not limited to the following: hazardous drinking of alcohol which affect adherence to ARVs [7]; opportunistic infections during ART [8]; previous exposure to ARVs before initiation of ART; advanced HIV disease (WHO clinical stage 4) [9]; change of ARVs due to toxicity; and baseline haemoglobin level less than 10g/dl [2]. On the other hand, other studies indicated that sexual orientation of the patient (comparing heterosexual and non-heterosexual); marital and employment status [2]; patient’s residence (whether urban or rural), pre-ART opportunistic infections, co-infection with Hepatitis B Virus (HBV) and Hepatitis C Virus (HCV) [3].

In addition, other studies have also shown that non disclosure of HIV status, history of Tuberculosis (TB), socioeconomic status/class, history of smoking at time of viral load testing, herbal medicine use at the time of viral load testing [2] and co-morbidities are associated with virologic failure.

While the HIV epidemic remains a “gendered one”, studies done in Massachusetts General Hospital out-patient HIV clinic in the USA; in Clinicas de Porto of Brazil and in Chiang Mai University Hospital in Thailand showed that male gender was not associated with virologic failure [3]. Similarly, a study done in Nigeria [2] showed that the male gender was not associated with virologic failure. However, a study done in the Americas (Mexico) showed different results i.e., the results showed that the female gender was marginally associated with virologic failure [4].

Study design and data sources

This is a retrospective cohort analysis of treatment outcomes. Data were abstracted from the OI/ART patient care booklets for clients initiated on ART between October 2012 and March 2013.

Study population

Site selection: Data were collected from all five MOHCC facilities in Chitungwiza; namely Chitungwiza Central Hospital, Seke North Clinic, Seke South Clinic, St Mary’s Clinic and Zengeza Clinic.

Patient inclusion criteria: All HIV positive clients 15 years and older, who were initiated on ART between the October 2012 and March 2013, at the five ART sites in Chitungwiza, regardless of treatment outcome, were included in the study.

Patient exclusion criteria: Patients initiated on ART after March 2013 was excluded from the study. Patients without a documented ART initiation date were excluded from the study.

Sample size

The following formula was used to come up with the sample size:

\[n = \left[ {p\left( {100 - p} \right)/\Delta {\rm{\^}}2 \times f\left( {1 - \propto } \right)} \right]\]

n = computed sample size

p = estimate of the proportion

∆= the desired width of the confidence interval

1-∝= confidence level

This implied that the study needed to sample a minimum of 310 OI/ART Patient Care Booklets to generate 95% confidence intervals with +/- 2.5% bounds around the proportion of interest. The sample is distributed as follows, per site, using probability proportional to size (as per their ART volume the five health facilities in June 2013) (Table 1).

Name of Health Facility Proportion Sample size
Chitungwiza Central Hospital 0.44337 137 (F=82; M=55)
Seke North Clinic 0.05673 18 (F=11; M=7)
Seke South Clinic 0.19602 60 (F=36; M=24)
St Mary’s Clinic 0.12711 39 (F=24; M=15)
Zengeza Clinic 0.17679 56 (F=33; M=23)
Total 1 310 (F=186; 124)

Table 1: Sample size per health facility.

Data collection

All analyses were performed using STATA 13 software. Univariate and multivariate regression modelling was used to identify the factors associated with treatment response of ART patients. Data were faced checked for completeness and consistency. Missing data were estimated using imputation techniques. Logistic regression was done to identify the factors associated with patient retention and linear regression was done to identify factors associated with weight gain. Poisson regression model was used to identify factors associated with immunological response. Univariate Cox proportional hazards model was run for covariate and factor variables. Variables for the multivariate Cox proportional hazards model were selected using the Lemeshow Hosmer statistic, all variables with p-value <0.25 from univariate analysis were considered for inclusion in the multivariate model. The proportional hazards assumption was checked using Kaplan Meir curves and verified using the log-rank test statistic. Post estimation diagnosis was also performed to test the adequacy of all the models.

Data analysis

All analyses were performed using STATA 13 software. Data management was performed, checking the data for completeness and consistency. Variables were managed using recode, encode, generate, destring, and tabstat commands in STATA 13 software. Univariate analysis was conducted to come with descriptive statistics and pictorial representations. The Wilcoxon matched-pairs signed-ranks test was applied to test for median difference between baseline CD4 and CD4 follow up, and baseline weight and follow-up weight, respectively. The Kaplan Meier and Nelson-Aalen methods were used to model survivorship function curves for retention and survival time, stratified by selected independent variables. The log-rank test was performed to test the significance of the difference in retention and survival for selected categorical variables.

Ethical Considerations

Clearance was sought from the MOHCC Head Office, the Chitungwiza Central Hospital CEO, the Superintendent at CITIMED Chitungwiza Hospital and the Chitungwiza City Health Department. To ensure confidentiality, no personally identifiable information relating to clients, such as patient name or clinic registration, number were collected during chart extraction. All the data was kept by the principal investigator on a personal computer with a passwordprotected login screen.

Demographic characteristics

Three hundred and five (305) respondents were considered in the study. Sixty percent (60%) of the respondents were females. The majority (71%) of the clients attained a secondary level of education. Only one percent of both males and females reached tertiary level. In addition, only one percent did not have any level of education. Sixtyfour percent (64%) of the clients were married, 18% were widowed, 10% were divorced and 7% were single. Table 2 shows other demographic characteristics of the respondents. Fifty-eight percent (58%) of the respondents were enrolled through Voluntary Counseling and Testing (VCT). Thirty-seven percent (37%) of the sampled clients were in WHO clinical stage III, see table 3 for more details. Of the total sample, 23% did not have a CD4 count done (42/183 women and 29/122 men). In 2012/13, where point of care CD4 counts was done, ART initiations were restricted to those with a CD4 cell count of 350 cells/µL or less. The exceptions to this rule were pregnant women as well as those who were TB-HIV co-infected regardless of sex. The proportion of those with/without a documented CD4 result was the same for both males and females. The average CD4 count at initiation was 334 cells/µL for females and 289 cells/µL for males.

Demographic Variable Female ART Clients Male ART Clients Overall Sample
Marital Status Number Percent Number Percent Number Percent
Divorced 24 13% 5 4% 29 10%
Married 99 54% 96 79% 195 64%
Single 13 7% 9 7% 22 7%
Widowed 47 26% 9 7% 56 18%
0 0% 3 2% 3 1%
Total 183 100% 122 100% 305 100%
Level of Education
None 2 1% 1 1% 3 1%
Primary 27 15% 10 8% 37 12%
Secondary 123 67% 95 78% 218 71%
Tertiary 2 1% 1 1% 3 1%
29 16% 15 12% 44 14%
Total 183 100% 122 10% 305 100%

Table 2: Marital status and level of education of respondents.

Referral source for HIV care and Treatment Female ART Clients   Male ART Clients   Overall Sample
VCT 89 49% 87   176 58%
TB Clinics 5 3% 10 8% 15 5%
PMTCT 13 7% 0 0% 13 4%
Obstetrics Unit 10 5% 0 0% 10 3%
Hospitalization 62 34% 23 19% 85 28%
Home 1 1% 0 0% 1 0%
Other 3 2% 2 2% 5 2%
Total 183 100% 122 100% 305 100%
WHO Stage at
Number Percent Number Percent Number Percent
Stage I 55 30% 20 16% 75 25%
Stage II 73 40% 40 33% 113 37%
Stage III 49 27% 59 48% 108 35%
Stage IV 6 3% 3 2% 9 3%
Total 183 100% 122 100% 305 100%
CD4+ Cell Count Done
Yes 42 23% 29 24% 71 23%
No 141 77% 93 76% 234 77%
Total 183 100% 122 100% 305 100%
Pre-ART Exposure
HAART 39 21% 21 17% 60 20%
PMTCT 7 4% 0 0% 7 2%
SD NVP 4 2% 1 1% 5 2%
None 133 73% 100 82% 233 76%
Total 183 100% 122 100% 305 100%
Exposure to OI prior to ART initiation
TB 9 5% 21 17% 30 10%
Other OI 36 20% 16 13% 52 17%
None 138 75% 85 70% 223 73%
Total 183 100% 122 100% 305 100%

Table 3: Clinical characteristics of the sampled male and female clients.

Factors associated with the observed treatment outcomes

The section below highlights the critical factors associated with the observed outcomes.

Predictors of retention: Table 4 below provides a synopsis of factors associated with retention. As shown in the table 4, females were twice likely to be retained on ART than their male counterparts. Similarly, the odds of being retained in care among clients in WHO Stage II were 3.02 times that of clients in Stage IV. The odds of being retained in care among clients who received Cotrimoxazole were 1.8 times that of clients who did not receive it. Clients under Multi-Month Scripting (MMS) regime were 28% less likely to be retained than those who were not on MMS. Similarly, clients with a university level of education were 73% less likely to be retained in care compared to those with primary education. See table 4 for additional outputs.


SE P-Value 95% CI
Female 2.01 0.84 0.09 0.89 4.55
Age group 0.38 0.07


0.26 0.55
WHO Stage
II 3.02 0.68


1.94 4.68
Yes 1.8 0.17


1.49 2.17
ART Regimen
Lamivudine+ Stavudine+
0.26 0.14 0.01 0.09 0.73
Yes 0.72 0.28


0.34 1.54
Level of education
Secondary 1.48 0.25 0.02 1.06 2.05
University 0.27 0.31 0.26 0.03 2.65
None 3.15 1.59 0.02 1.17 8.49

Table 4: Predictors of retention.

Predictors of weight gain: Table 5 gives a synopsis of the key factors associated with weight gain (the clinical outcome of interest) using a linear regression model. Female patients have extra weight gain compared to male patients, holding other factors constant. Females were 3.72 times likely to gain weight than their male counterparts (1.313 kg; range of 0.347; 2.279 kg). Widows were 96% less likely to gain weight than those who were married. Similarly, PLHIV with no education were 96% less likely to gain weight compared to PLWHA who attended at least primary school.

Covariates Odds
SE P-value   95% CI
Female 3.72 1.37 0.02 1.41 9.76
(25-29) 0.01 0.02 0.02 0 0.28
(30-34) 0.01 0.01 0.01 0 0.21
Marital status
Widowed 0.04 0.06 0.09 0 2.03
Level of education
None 0.18 0.11 0.04 0.04 0.85
ART Regimen
Lamivudine+ Stavudine+ Nevirapine   0.28   0.12   0.04   0.09   0.87

Table 5: Predictors of weight gain (clinical outcome of interest).

Predictors of immunological outcomes (CD4): Table 6 below shows the factors associated with positive changes in CD4 count. The IRRs are the incidence rate ratios for the Poisson model which are obtained by exponentiating the Poisson regression coefficient. The IRR is the estimated rate ratio for one-unit increase in CD4 holding other variables constant in this model. If a widowed person living with HIV were to increase their CD4 by one point, his/her rate ratio for CD4 is expected to decrease by a factor of 0.68. This is statistically significant (CI: 0.57-0.81). The picture is the same for single persons (IRR=0.54; CI: 0.36-0.80); received Cotrimoxazole (IRR=0.69; CI: 0.63-0.76); on MMS (IRR=0.77; CI: 0.64-0.92). Clients in WHO stage IV are expected to have a rate of 1.71 times greater for loss in CD4 than clients in stage 1, while holding other variables constant (IRR=1.71; CI:1.48-1.96).

Covariates IRR SE P-Value 95% CI
Marital status
Widowed 0.68 0.06 0 0.57 0.81
Single 0.54 0.11 0 0.36 0.8
Level of education
Secondary 0.00 0 0 0 0
WHO stage
IV 1.71 0.12 0 1.48 1.96
Yes 0.69 0.03 0 0.63 0.76
Yes 0.77 0.07 0.01 0.64 0.92
ART regimen
Lamivudine+ Tenefovir + Efavirenz   1.59   0.31   0.02   1.08   2.33
Zidovudine+ Lamivudine+ Nevirapine   1.52   0.12   0   1.31   1.76

Table 6: Predictors of positive changes in CD4.

Predictors of survival: Table 7 below details the factors associated with survival. Using the Weibull regression for predictors, the data shows that women were 96% more likely to survive than male (HR=0.04; CI:0.00-0.28). Similarly, those who received Cotrimoxazole were 93% more likely to survive than those who did not (HR=0.07; CI: 0.02-0.24) as shown in table 7 below. PLHIV with no education were dying at a rate that was 60% more than those with some education (HR=0.4; CI: 0.03-5.72), while clients in WHO stage I were dying at a rate that was 96% lower than those in Stage IV (HR=0.04; CI: 0.01- 0.15). The hazard of dying was significantly higher in the 25-29 year olds compared to the 15-19 year olds (HR=11.64; CI: 1.67-81.15), albeit with a wider confidence interval. Other important factors are detailed in the same table.

Predictors HR SE P-value 95% CI
Female 0.04 0.04 0 0 0.28
Yes 0.07 0.04 0 0.02 0.24
(25-29) 11.64 11.53 0.01 1.67 81.15
(30-34) 0 0 0 0 0
(35+) 0 0 0 0 0
Level of education
Secondary 0.05 0.03 0 0.01 0.2
University 0 0 0 0 0
None 0.4 0.54 0.5 0.03 5.72
Other 0 0 0 0 0
WHO stage
I 0.04 0.03 0 0.01 0.15
II 0 0 0 0 0
ART regimen
Zidovudine+ Lamivudine+ Efavirenz 0 0 0 0 0
Zidovudine+ Lamivudine+Nevirapine 0 0 0 0 0
/ln_p -0.72 0.2 0 -1.11 -0.33
p 0.49 0.1   0.33 0.72
1/p 2.05 0.41   1.39 3.04

Table 7: Predictors of survival.


The study confirms some of the findings from earlier studies in Zimbabwe, albeit with differences on some of the factors deemed significant. This study contributes to the existing body of knowledge on critical factors affecting ART outcomes in Zimbabwe, albeit the need to explore the subject further given the transition to differentiated service delivery models as well as changes in ART regimens.


  1. . UNAIDS (2018) Global Aids Response Progress Report 2018. National AIDS Council, Ministry of Health and Child Care. Zimbabwe. [Ref.]
  2. Anude CJ, Eze E, Onyegbutulem HC, Charurat M, Etiebet MA, et al. (2013) Immuno-virologic outcomes and immuno-virologic discordance among adults alive and on anti-retroviral therapy at 12 months in Nigeria. BMC Infectious Diseases 13: 113. [Ref.]
  3. Khienprasit N, Chaiwarith R, Sirisanthana T, Supparatpinyo K (2011) Incidence and risk factors of antiretroviral treatment failure in treatment-naïve HIV-infected patients at Chiang Mai University Hospital, Thailand. AIDS Res Ther 8: 42. [Ref.]
  4. Ramírez BC, Keever AV, Fraga AG, del Río C, Madero JS (2010) Effectiveness of Highly Active Antiretroviral Therapy (HAART) Among HIV-Infected Patients in Mexico. AIDS Res Hum Retroviruses 26: 373-378. [Ref.]
  5. Haile D, Takele A, Gashaw K, Demelash H, Nigatu D (2016) Predictors of Treatment Failure among Adult Antiretroviral Treatment (ART) Clients in Bale Zone Hospitals, South Eastern Ethiopia. PLoS One 11: e0164299. [Ref.]
  6. Bajpai R, Chaturvedi H, Jayaseelan L, Harvey P, Seguy N, et al. (2016) Effects of Antiretroviral Therapy on the Survival of Human Immunodeficiency Virus-positive Adult Patients in Andhra Pradesh, India: A Retrospective Cohort Study, 2007-2013. J Prev Med Public Health 49: 394-405. [Ref.]
  7. Chander G, Lau B, Moore RD (2006) Hazardous Alcohol Use: A Risk Factor for Non-Adherence and Lack of Suppression in HIV Infection. J Acquir Immune Defic Syndr 43: 411-417. [Ref.]
  8. Alave J, Paz J, González E, Campos M, Rodríguez M, et al. (2013) Risk factors associated with virologic failure in HIV- infected patients receiving antiretroviral therapy at a public hospital in Peru. Rev Chilena Infectol 30: 42-48. [Ref.]
  9. Huong DTM, Bannister W, Phong PY (2011) Factors associated with HIV-1 virological failure in an outpatient clinic for HIV-infected people in Haiphong, Vietnam. Int Jour of STD & AIDS 22: 659-664. [Ref.]

Download Provisional PDF Here


Article Information

Article Type: Research Article

Citation: Sanhokwe H, Shabangu P, Chipepa F (2019) Factors Associated with ART Outcomes Among Adult (15+) Persons Living with HIV in Zimbabwe in the Multi-Month Scripting (MMS) Regime. J HIV AIDS 5(2):

Copyright: © 2019 Sanhokwe H, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Publication history: 

  • Received date: 19 Dec, 2018

  • Accepted date: 12 Mar, 2019

  • Published date: 19 Mar, 2019