CMPH

Background

The risk of overweight/obesity has increased globally particularly in developing countries over the last two decades. Some experts have attributed that to the rapid urbanization [1] that has taken away families from homemade to higher calorie diet in fast-food restaurants. Likewise, urbanization has reduced sedentary lifestyle, forcing people to eat anything anywhere while on the move and thus, promoting jumble eating habit with all the consequences that comes with it, i.e., risks of overweight/obesity and related comorbidities [1-3]. However, the key to the baffling epidemics of overweight/obesity and related comorbidities cannot be so simplistic. Looking at existing diversification in sociodemographic, economic, cultural and biomedical factors amongst individuals and communities, as well as the positive/negative effects of each of such factors, there is a pressing need to address these issues. Highly active antiretroviral treatment (HAART) drugs, such as protease and integrase inhibitors, i.e., dolutegravir, and NRTI, particularly tenofovir alafenamide (TAF) have been pointed out liable for overweight/obesity and for related comorbidities among HAART recipients [4,5]. Indeed, since the advent of ART, new profile of people living with HIV (PLHIV) has emerged. Reason why, PLHIV formally characterized by wasting syndrome [6], are now subjugated by overweight/obesity [7,8]. Though experts have reported comparable risks of overweight/obesity as well as associated comorbidities among ART recipients versus general population [9], questions of comparability of risk magnitude in developing these undesirable health events among ART-recipients versus, non-recipients, still remains ill understood.

Studies conducted in Botswana estimated the prevalence of overweight/obesity in the country to be 41% [8]. With an adult HIV prevalence of 20.3%, Botswana ranks the fourth in the world amongst the countries most affected by HIV/AIDS, leaving behind only South Africa, Swaziland and Lesotho [10]; this clearly pinpoint PLHIV as a non-negligible subpopulation. Therefore, the need for equitable and prompt attention; since ± 84% of this subgroup, is on ART [10]. Hence, it is not uncommon to see higher level of overweight/obesity and related comorbidities among PLHIV compared to the general population. Works disapproving this observation seem to be limited; nevertheless, a recent report from four Gaborone general clinics showed comparable rates of overweight/obesity between HIV-positive versus HIV-negative patients; paradoxically in the same report, rates of comorbidities were higher among HIV-negative compared to HIV-positive patients [11]; an unexpected situation from the authors. Since, to date no such data has been reported in the country. Nonetheless, in the neighboring South Africa this situation was already seen before. Madogoro and colleagues investigated 11,083 households during the South African national demographic health survey in 2016 [12], they reported lower prevalence of hypertension (HTN), diabetes mellitus (DM) or Coronary Heart Diseases (CHD) among HIV-positive patients with overweight/obesity compared to HIVnegative patients with overweight/obesity. Likewise, Tripathi et al [13] studied 6816 HIV-positive patients; over 80% of them on ART had lower DM incidence rates than HIV-negative patients (11.4 vs 13.6). These unexpected results in fact reflect a complex interplay among factors within individuals and/or communities, example overweight/obese-HIV-positive patients and related comorbidities versus overweight/obese-HIV-negative patients and related comorbidities. The need to systematically delineate interplaying factors within and/or between different components leading to expected/unexpected outcomes is immense in order to alleviate existing controversies for an unequivocal base for decision making.

Thus, we sought in this study to: (i) identify and compare risk factors of overweight/obesity among HIV-positive versus HIVnegative patients attending same general clinics the same time period and, (ii) compare risk factors of comorbidities commonly associated with overweight/obesity among HIV-positive versus HIV-negative patients by magnitude and direction, (that is whether they have predictive or protective effect).

Materials and Methods

Study design

This was a cross-sectional study using analytical approaches to address the study objectives.

Study site and settings

The study was conducted in Gaborone, the capital city of Botswana. Its population was estimated to be 233,135 (12%) based on the 2011 Botswana Population and Housing census.

The study was conducted in 4 different Gaborone clinics: Julia Molefe clinic, BH3 clinic, Extension 2 clinic and Gaborone West clinic as they serve the highest number of patients per day compared to all other clinics. All the clinics in Gaborone have an outpatient department (OPD) and Infectious Diseases Control Centre (IDCC) where HIV positive clients follow up for treatment and routine monitoring, IDCCs are linked online through an Integrated Patient Management System (IPMS), while general outpatient clinics have either a standalone network installed or Patient Information Management System (PIMS).

Study population, inclusion and exclusion criteria

HIV-positive and HIV-negative-participants attending local health facilities between June and July 2019 in Gaborone was the study population. Outpatient clients aged 18 years and above, non-HIV patients, HIV-patients receiving ART even for pre/post-exposure prophylaxis, or as therapy for any other disease, i.e. chronic hepatitis B virus infection, as well as HIV-positive patients not yet on ART, all met the inclusion criteria. Excluded from the study were patients below 18 years of age, patients of unknown HIV status, pregnant women and women attending clinics for a 6 weeks post-natal visit, patients that were not mobile or were wheelchair bound, not capable of standing without support and patients with decreased mental capacity.

Sample size and Sampling strategy

The sample size was estimated as described by Ariya and colleagues [14], using 1.96 (Z-score) as the value of the standard normal distribution corresponding to a level of alpha that was significant (<0.05) for a two-side test or margin of error of 5%; where n is the sample size, P = the expected prevalence, d = the allowable error.

Assuming that P and d, are decimal values but would hold correct if they are percentages, except that the term (1-P) will become 100-P in the numerator. P was set to 0.97 for common conditions, and d = 0.015; thus, we estimated the sample size as:

Considering respondents who will not meet inclusion criteria or withdraw from the study, this sample was increased by 17% to be 581 patients.

Selection and enrolment of both HIV-negative and HIV-positive patients were conducted in selected sites where both HIVnegative and HIV-positive were attending. Since the majorities of patients were drop-ins and did not have doctors’ appointments, it was difficult to produce a sampling frame from which potential participants could have been drawn. Nevertheless, using a systematic, every 3rd patient was selected and approached when he/she finished with his/her clinic visit of the day. Those who consented were enrolled in the study. In case of non-consent, the latter was automatically replaced by the next patient. Recruitment continued until the estimated sample size was reached. The number of patients recruited in each of the four settings was proportional to the size of patients that the clinic catered for. The study was conducted in all four clinics at the same time period.

Data collection

Data was collected by trained data collectors using a pre-tested data collection tool between, June 15 and July 31, 2019.

Data analysis

Data was analysed using IBM SPSS version 27 (Chicago, IL). Patients’ BMI were estimated as presented by Tshikuka et al., [12]; the same population was used in the current study but this time not to study the prevalence but to investigate whether what prompted outcomes were different/same between the 2 study groups and pointed to the same direction with the same magnitude. Thus, the primary outcome of the study was overweight/obesity and the secondary outcome was comorbidity (comorbidities: HTN, DM and CHD). Being HIV-positive on ART and HIV-negative were the main exposure variables. Other exposure variables are provided in Tables 1-3.

For simplicity of analysis, overweight and obesity were aggregated as a single variable “overweight/obesity”; likewise, HTN, DM and CHD were aggregated as “comorbidity”. We estimated the duration of exposure to ART among HIV-positive patients by subtracting the date of ART initiation from the date of data collection expressed in years. We used major mode of commuting as the indicator of socioeconomic status (SES) [11]. Thus, participants whose major mode of commuting was walking were considered as having a low socioeconomic status (LSES), those who used public transportation had a middle-income socioeconomic status (MSES) and those who used private cars as mode of transportation were categorized as having a high socioeconomic status (HSES). Patients with HIV/AIDS and comorbidities were identified based on the attending physician diagnosis and in line with International Classification of Diseases, tenth revision (ICD-10). The risk of overweight/obesity and of comorbidity was estimated using odds ratio as a proxy. To identify and compare the risk of overweight/ obesity between HIV-positive and HIV-negative patients, potential exposure factors were investigated in series of bivariate analyses using simultaneously overweight/obesity and comorbidity as dependent variables. Unadjusted odds ratios (UOR) and their 95% CI were estimated for each covariate. Those that achieved UOR with a p ≤ 0.90 were further investigated in multivariate logistic sub-models.

Four logistic regression models were run, two examined the risk of overweight/obesity among HIV-positive versus HIV-negative patients and, two other explored the risk of comorbidity (this is because overweight/obesity and comorbidities associated to them may be prompted by different factors) among HIV-positive versus HIV-negative patients using the following as independent factors: (i) patients’ sociodemographic, economic and cultural factors, (ii) patient and parent biomedical factors and, (iii) patient lifestyle related factors or type of activities practiced. The model-A included only HIV-positive-patients or ART recipients; overweight/obesity was the model-A dependent variable. Independent variables were: the duration of exposure to ART, expressed as a continued variable in years and as categorical variable inclosing patients exposed to ART for >4 years and, patients exposed to ART for ≤ 4 years. Other independent variables are as presented in Tables 1-3. Model-B included only HIV-negative-patients or those who never been exposed to ART, the dependent variable was “overweight/ obesity”. Independent variables were as presented in Tables 1-3, while omitting the duration of exposure, as none of the patients directly/indirectly was on ART. Model-C included only HIVpositive patients; likewise, model-D included only HIV-negative patients. Both models had Comorbidity as dependent variable. Independent variables were as presented in Tables 1-3.

From each model, adjusted odds ratio (AOR) and their 95% confidence interval (95% CI) were computed. Using backward elimination, variables kept in the models were considered as having independent effects on overweight/obesity when they achieved p <0.05 or if they had established recognition within the relationship under study; how well the models fitted the data was estimated using Cox and Snell R2 tests. Similarity between model-A (for HIV-positive) and model-B (for HIV-negativepatients) was established when the two models shared the same covariates independently of their contribution to the outcome of interest. A covariate was considered to have a predictive effect on the outcome when it had a positive β and a non-decimal 95% CI; it had a protective effect on the outcome when it had a negative β and a decimal 95% CI. The magnitude of covariate effects on the outcome, was function of the size of the AOR at p <0.05.

Results

Of the 581 participants enrolled in the study, 294 (51.63%) were

HIV negative and 287 (48.36%) were HIV positive. Of those, 136 (47.4 %) were on first line ART regimen (TDF-FTC-EFV), 61 (21.3%) were on second line (TDF-3TC-DTG), 44 (15.3%) on third line ART regimen (TDF-FTC-DTG) and 46 (16.0%) on other ART combinations. More details on HIV positive patients by ART regimen are described in Tshikuka et al [11]; Thirty-nine (7,0%) of the participants were underweight, two hundred seventy-three 273 (47%) had normal BMI. Hundred and ten patients, 110 (18.9%) were obese and 269 (46%) were overweight/obese.

Details on overweight/obesity by sociodemographic, economic and cultural factors as well as by biomedical and physical activities among HIV-positive and HIV-negative patients are provided in Table 1-3.

The bivariate analysis of overweight/obesity by sociodemographic and cultural factors revealed that of all examined covariate, only SES did not meet our arbitrary requirement of (^¥P ≤ 0.90) for re-evaluation in multivariate models. But given the established association of SES with overweight/obesity and other health conditions it was kept for multivariate analysis. More details are provided in Table 1.

Table 1: Bivariate analysis of overweight/obesity by sociodemographic, economic and cultural factors among HIV-positive versus HIVnegative patients attending general clinics in Gaborone, Botswana in 2019.

Bivariate analysis of overweight/obesity by biomedical factors revealed that HIV-positive patients whose parents had HTN were more likely (p<0.05) to be overweight/obese compared to those whose parents were free of HTN. Similarly, HIV-negative patients whose parents had HTN were more likely to be overweight/obese compared to HIV-negative whose parents did not have HTN. HIV-negative patients who had HTN were more likely (p<0.05) to be overweight/obese compared HIV-negative patient free of HTN. HIV-negative patients who were free of HTN were not associated with overweight/obesity. More details are given bellow in Table 2.

Table 2: Bivariate analysis of overweight/obesity by biomedical factors among HIV-positive versus HIV-negative patients attending general clinics in Gaborone, Botswana in 2019.

Bivariate analysis of overweight/obesity by life style habits in Table-3 showed that smoking was strongly associated (P<0.05) with overweight/obesity for both HIV-positive and negative patients while lifestyle factor was significantly associated with overweight/ obesity among HIV-negatives patients. More details are provided bellow in Table 3.

Table 3: Bivariate analysis of overweight/obesity by life style habits among HIV-positive versus HIV-negative patients attending general clinics in Gaborone, Botswana in 2019.

The risks of overweight/obesity among HIV-positive versus HIV-negative patients attending outpatient department of general clinics in Gaborone are provided in Table 4. (i) Results indicate that the covariate gender entered with significance (p<0.05) in both sub-model A and B; female HIV-positive patients were 5.12 (AOR: 5.12, 95% CI:2.73-9.85) times more at risk of overweight/ obesity than their counterparts HIV-positive male in Sub-model A; likewise, in model B female HIV-negative patients had 4.15 (AOR: 4.15, 95% CI: 2.30-7.48) times more risk of overweight/ obesity compared to their male counterparts ; (ii) aside from HIVpositive patients aged 25-34, the rest of age groups entered both model A and B with significance (p<0.05), HIV-positive patients aged 35-44 and those aged ≥ 45 years old had respectively 4.0 (AOR: 4, 95% CI:1.31 – 12.21), and 5.42 (AOR: 5.42, 95% CI:

1.69-17.40) times the risk of overweight/obesity compared to HIV-positive patients aged 18-24 years; model B, HIV-negativepatients aged 25 to 34, 35 to 44 years and those aged ≥ 45 years old were respectively 2.38, (AOR: 2.38, 95% CI: 1.16-4.88), 5.18 (AOR: 5.18, 95% CI: 2.07-12.95) and 3.19 (AOR: 3.19, 95% CI:

1.10-9.21) times at risk of overweight/obesity compared to those aged 18-24 years. (iii) Alcohol consumption is strongly associated with overweight/obesity in both models; HIV-positive patients had

1.78 (AOR:1.78, CI: 0.98-3.23) and, HIV-negative patients had 1.79 (AOR:1.79, CI: 1.01-1.31) the risk of overweight/obesity compared to non-alcohol drinkers, HIV-positive-patients residing in semi-urban area had 3.10 (AOR: 3.10, 95% CI: 1.17 – 8.21) times the risk of overweight/obesity compared to those who resided in the rural area, more details are in Table 4; (iv) Patients with parental history of HTN, entered the model-A for HIV-positive with significance (p<0.05); they had 1.85 (AOR: 1.85, 95%, CI: 1.03 – 3.34,) more the risk of overweight/obesity compared to HIV-positive patients with no parental history of HTN; patients with HTN had 3.67 (AOR: 3,67, 95% CI: 1.18-11.42) more risk of being overweight/obese compared to HIV-negative patients with no HTN; patients with other types of comorbidities in model-A for HIV-positive and for the model-B for HIV-negative respectively did not contribute to the outcome of interest. (vi); in model-B, HIV-negative patients who did not practiced physical activities had 1.97 (AOR: 1.97, 95% CI: 1.04-3.74) times more risk of being overweight/obese compared to those who practiced physical

activities, (ix) in model-A, HIV-positive patients exposed to ART for ≤4 years had 2.50 (AOR: 2.50, 95% CI: 1.45 – 4.31) times more risk of being overweight/obesity compared HIV-positive exposed to ART for >4 years (Table 4).

Table 4: Adjusted correlates of overweight/obesity among HIV-positive versus HIV negative patients attending clinics in Gaborone, Botswana in 2019.

The risk of comorbidity (HTN+DM+CHD) among HIV-positive versus HIV-negative patients attending same outpatient department of general clinics in Gaborone are presented in Table 5 model-C: (i) HIV-positive-patients aged ≤ 45 years old had 34% (AOR 0.34, 95% CI: 0.14-0.62) less risk of having comorbidities compared to HIV-positive patients aged > 45 years old; yet in Sub-model D, HIVnegative patients aged ≤ 45 years had only 12% (AOR; 0.12, 95% CI: 0.07 – 0.19) less risk of having comorbidity compared to patients aged > 45 years. (ii) in model C, gender among HIV-positive patients did not enter in the other; while in the model-D, females (AOR: 1.84, 95% CI: 1.055-3.193) were 84% more at risk of comorbidity compared to male HIV-negative patients ; (iii) Overweight/obese HIV-positive-patients had 29% (AOR: 1.29,95% CI: 1.14-1.62) more risk of developing comorbidity compared to HIV-positive-patients who were not overweight/obese; This covariate did not enter the model-C for HIV-negative patients; (iv) in the model-C HIV-positive patients who practiced physical activities had 36% (AOR: 0.36, 95% CI: 0.13-0.044) less risk of having comorbidity than HIV-positive patients who did not practice physical activity. More details in this covariate is in Table 4 model D (v) HIV-positive patients who smoked had 34% (AOR 0.34, 95% CI: 0.17- 0.66) less risk of having comorbidity compared to HIV-positive patients who were not smokers; again, more details on this covariate and the rest are in Table 4.

Table 5: Adjusted correlates of Comorbidity (HTN+DM+CHD) among HIV-positive and HIV-negative patients attending clinics in Gaborone, Botswana in 2019.

Discussion

The current epidemic of overweight/obesity among ART recipients has prompted polemics among scientists as indicated herein by the number of conflicting reports [9, 11-13,16-21]. Some experts claiming the epidemic as an aftermath of ART [17-20] while others are maintaining that it is part of the current overweight/obesity epidemic in the general population [11, 15]. On the other-hand there is literature that even pinpoint specific ARTs, i.e., protease [15] and integrase inhibitors [18,22-24], as factors explaining most variabilities of overweight/obesity among the recipients. A narrative review of critical appraisal of articles published until year 2024 in the field, by Konstatinos Markakis and colleagues [25] indicate challenges of assessing the role of HAART on weight gain as an interplay of different factors needing cautious investigation.