The study set out to describe the prevalence of comorbidity and multi-morbidity in a population of adults living with HIV infection and receiving care at a public sector facility in Zimbabwe. This was an initial step in understanding the emergent challenges of NCD comorbidity in HIV infection. The occurrence of these comorbidities and multi-morbidities was further described according to patient demographic and clinical characteristics. It is anticipated that any observed associations with these characteristics may help in the formulation of risk stratification algorithms.
In our study, we had more female than male participants, with females presenting at a younger age and with less severe and/or less advanced HIV disease compared to their male counterparts. These findings are in keeping with sub-Saharan Africa's HIV epidemic, where transmission is largely heterosexual coupled with increased infection vulnerability of women [18], on one hand, and their better health-seeking behavior [20], on the other. We found that women were more likely to have both comorbidities and multi-morbidities, compared to men. Similarly, this pattern was observed for individual NCDs, notably more women had both hypertension and asthma, compared to men. Women also had a higher prevalence of T2DM than men though this relationship did not reach statistical significance. Though pregnant women were excluded from the study, the possible contribution of previous pregnancies to prevalent hypertension or diabetes mellitus was not investigated.
Meaningful comparisons across studies are precluded by differences in case ascertainment and the actual definitions of comorbidity. Other than findings reported from cohorts treated in high income countries [10, 11], the study by Oni et al. [21] is, to our knowledge, the only one from SSA that sought to address a similar research question as ours. In a primary care facility-based study in a peri-urban South African setting, they reported gender- and age-stratified prevalence rates of comorbidity in HIV. Amongst males, comorbidity rates were 17 % (18–34 years), 28 % (36–45 years), 34 % (46–55 years) and 20 % (>55 years). The corresponding rates among females were 28, 30, 30 and 10 %, respectively; meaning that only in the younger subjects (18–45 years) was comorbidity higher in females than males.
Within each respective gender group, our study did not further stratify comorbidity according to age groups because of very limited events. Though this limits comparison with the South African findings, our study undoubtedly suggests lower overall prevalence of HIV-related comorbidity. Noteworthy is the fact Oni et al. included only hypertension, diabetes mellitus and TB in their definition of comorbidity. They reported hypertension as the commonest comorbid condition. We had similar findings, with overall prevalence of hypertension in our study at 10.2 % (95 % CI 8.4–12.3).
To date, reported hypertension prevalence rates among HIV positive persons in SSA range from 11 % in Kenya [22] to 28 % in Uganda [23]. A recent systematic review [24] on hypertension prevalence in HIV infection and with a broader population focus reported rates ranging from 8.7 to 45.9 % in low- and middle-income countries (LMICs). Although we did not include an HIV negative comparison group in our study, our reported prevalence of 10.2 % is less than the previously reported Zimbabwean national average rate of 17.9 % in the general population [16]. Hypertension is largely asymptomatic and coupled with the clinic's lack of active screening for hypertension, it may be speculated that our reported low prevalence, relative to the general population, may represent under-diagnosis.
The literature on hypertension in HIV infected persons in SSA is sparse and methodological differences aside, findings are equivocal: some studies report higher hypertension rates in the HIV positive versus the uninfected [25] while others suggest lower prevalence or no differences [26, 27]. Evidence from a systematic review including predominantly non-African subjects concluded that the risk of hypertension among HV positive persons was higher in the ART experienced versus the ART-naïve. The review did not compare hypertension to the general population [28]. Interestingly, and in contrast to our findings, men have been reported as being disproportionately affected [21].
Our study also found type 2 diabetes mellitus (T2DM) prevalence of 2.1 % with a higher but statistically insignificant rate in females (2.2 vs. 1.9 %; p = 0.731) compared to males. We could not ascertain how the diagnosis of T2DM was made in each patient. Rates of T2DM in HIV higher than what we found have been reported elsewhere in SSA: 16 % in Kenya [29] and 23.5 % [30] in South Africa. A nationally representative survey of the general adult Zimbabwean population found T2DM prevalence of 3.7 % in 2005 [16]. This estimate (reported in 2005) is most likely outdated and unrepresentative of the current situation. Unpublished findings from an urban general population community-based survey (2012) found combined rates of self-reported prevalent T2DM and previously undiagnosed hyperglycemia (random venous blood glucose ≥11.1 mmol/l) of nearly 9 %. A more recent estimate from the WHO (2014) places T2DM (fasting glucose ≥7.0 mmol/l or on medication) prevalence at 7.8 % (95 % CI 3.0–15.9) among adults in the general population [31].
The relationship between HIV/ART and diabetes is the subject of ongoing controversy. Observations of both increased risk and no difference have been noted in high income countries [32–34]. Elevated risks are linked to, among other factors, duration of and exposure to certain ARV drugs like stavudine, efavirenz and protease inhibitors [30, 34, 35]. A systemic review (2013) of the association between HIV/ART and development of T2DM in SSA subjects implied protective effects as judged by reduced HbA1c levels [12]. Notwithstanding lack of an HIV negative comparator group within our study, and taking into consideration sample variation, our finding of a 2.1 % T2DM prevalence maybe difficult to reconcile with, firstly, the much higher background T2DM prevalence in the general population, and secondly, the high frequency of stavudine (60.4 %), efavirenz (>60 %) and protease inhibitor (4.7 %) use among study participants.
In the general population hypertension and T2DM have similar risk factors: tobacco use, unhealthy diets, physical inactivity and harmful use of alcohol. These have all been documented to be highly prevalent in some HIV positive populations [36], though they were not measured in this study. ARV drug toxicities including lipodystrophy and dyslipidemias, chronic systemic inflammation, endothelial dysfunction, coagulation disorders and the virus itself are additional risk factors peculiar to HIV disease [37].
Another finding of note was the relatively high occurrence of asthma in our study population. We report a prevalence of 4.3 % (95 % CI 3.1–5.8 %). The asthma burden in the Zimbabwean general population is estimated at 2.3 % [38]. Female study participants had higher rates than their male peers (5.3 vs. 2.2 %; p = 0.026). Though not within the scope of the study to ascertain each included diagnosis, review of some patient case notes indicated that the diagnosis of asthma in some patients was based on adult-onset wheezing, effort dyspnoea and persistent coughing. HIV trained nurses provided the bulk of on-going care including initial patient assessments. Pulmonary function testing was not available either.
Consequently, argument can be made to refer to this class of condition broadly as obstructive lung disease (OLD) [39], which includes chronic bronchitis, fixed airway obstruction, bronchial hyper-responsiveness as well as asthma. OLD has recently come to the fore as the number of persons ageing with HIV continues to grow. Known risk factors for OLD include HIV infection, pulmonary TB, tobacco smoking and indoor pollution from biomass fuel exposure [40–42]. The latter disproportionately affects women in LMICs, especially SSA. Our study only assessed history of pulmonary TB, which stood at 61 % among participants, and we did not measure the burden of the other risk factors. Asthma prevalence among HIV-infected persons is reported to be increased with possibility of a different risk factor profile from the uninfected population [43].
The study's primary outcome of HIV with any comorbid NCD has some drawbacks. From an epidemiological or health systems perspective, where the concern might be burden of care and health resources allocation, the study's end-point has currency. However, from a pathogenesis-clinical standpoint in the given contest of HIV and ART exposure, the prominent limitation is that this end-point is potentially heterogeneous. This is because whereas cardiometabolic diseases like diabetes mellitus and hypertension may have common risk factors, other NCDs like asthma arguably have a different risk factor profile, as outlined above.
Other co-variables of study interest were ART exposure, age and HIV disease severity. Overall, the study population had considerable ART experience with at least half the participants having been on treatment for at least 5 years. However, we found no significant relationships between either ARV drug class or duration of exposure with risk of comorbidity. This is in contrast with the accumulated evidence linking, for example, PIs with diabetes and cardiovascular events, or NRTIs with cardiac disease [12, 27, 28, 30]. CD4+ cell counts (baseline and latest) and baseline WHO stage had no demonstrated significant associations with comorbidity risk. Similarly, duration of (known) HIV sero-positive state was insignificant after multi-variate analysis. All these factors have been previously linked with comorbidity development [44]. Our study may have been underpowered to detect these effects because of its relatively small sample size.
Lastly, the HIV epidemic in SAA is maturing with ever growing numbers of adults ageing with HIV, notwithstanding the observation from this study of a relatively small proportion (13 %) being aged 55 years or more. In the general population, NCD risks are known to be considerably increased in this age (50+ years old) group. Indeed, our findings point to age as being an important risk factor for both comorbidity and multi-morbidity in the HIV infected population. In multi-variate analysis adjusting for gender, duration of HIV sero-positive state, ART regimens received, and baseline WHO HIV disease stage and CD4 + cell counts, each additional year of age was associated with a 6 % increased risk of comorbidity (adjusted OR 1.06; 95 % CI 1.04–1.08).
Limitations
A number of limitations with our study warrant cautious interpretation of its findings. Firstly, as a health-facility based study, our inferences about the burden of comorbidity in the wider HIV-infected Zimbabwean adult population can only be guarded. A community-based study would have been ideal. Related to this is the second limitation: selection bias. The study site is an HIV out-patient service of a tertiary care and teaching hospital, which also runs specialist medical out-patient clinics. It is likely that HIV positive patients with more severe comorbid NCDs were preferentially reviewed in the medical clinics inadvertently reducing the total numbers of NCD/HIV patients seen in the HIV clinic. In similar fashion, it is arguable that those with asymptomatic diseases may have been missed due to the lack of a policy of active NCDs screening. Both situations would underestimate comorbidity prevalence.
Disease misclassification is a possibility considering that the bulk of diagnoses and on-going care were nurse-driven. Our study did not seek to validate clinical diagnoses. We merely enumerated the diagnosed NCD cases. Non-infectious chronic lung diseases, for example, may have been misdiagnosed and treated as pulmonary TB. Lastly, we used routine data which is not usually collected with the rigour required for research purposes. Though the data were sufficiently complete allowing us to answer our research question, more detailed data would have enabled additional analyses, case ascertainment and more robust conclusions. Lastly, inclusion of an HIV uninfected age- and gender-matched comparison group would have afforded insights into the putative roles of HIV and ART as drivers of NCDs.