This study examined co-morbidity patterns in a cohort of patients receiving at least one prescription for one of 4 diseases over a 9-month period, and has a number of major findings. Firstly, a quartet of chronic diseases accounted for 45% of all consultations in a community health centre providing primary care services for people of low socioeconomic status in Cape Town. Secondly, we found a 22.6% prevalence of MM among chronic disease patients and an associative pattern of MM, with HPT and DM often co-existing. Furthermore HPT was the most common co-morbidity in both HIV/ART and DM patients. Thirdly, we demonstrated that while HIV/ART was the most common co-morbidity among TB patients, 37% and 12% of multi-morbid TB patients were also on treatment for HPT and DM respectively. Fourthly, we found a high prevalence of MM in younger patients on ART (26% and 30% in 18–35 year and 36–45 year age groups respectively). Lastly, we showed that among these younger HIV/ART patients with MM, HPT and DM prevalence was significantly higher than patients in the same age groups who were not on ART.
Our results reveal a very high prevalence of HPT among chronic disease patients. This is congruent with national data that shows a high overall prevalence of HPT in the general population and that HPT is the commonest reason for attendance of primary health clinics in South Africa [14]. A national, population-based study of persons 50 years and older in South Africa reported HPT prevalence of 77.3% [15]. It is noteworthy that HPT was the most common co-morbidity in both DM and HIV/ART MM patients. It was not surprising that 88% of all DM MM patients were also on treatment for HPT; DM patients are routinely screened for HPT in this setting. On the other hand, blood pressure screening is not currently routine in the management of HIV/ART and TB patients. Further the finding that 75% of HPT patients with MM were on treatment for DM suggesting that HPT patients should also routinely be screened for DM, which is not current local practice. Our results are comparable to data from high-income settings. A recent study from Ireland examined co-morbidity patterns (including hypertension, heart disease, arthritis, depression, and chronic lung disease) and reported that 90% of a cohort of DM patients had another morbidity and 66% had HPT [16]. However, 90% of study participants in this study were >50 years old and HIV status was not documented.
Among TB patients with other chronic disease co-morbidities, the prevalence of HIV co-infection was high. This was an expected result as the HIV/TB co-infection rate in this local setting has previously been estimated as 67% [17]. Our data demonstrated that in addition to HIV, TB patients also have a significant prevalence of HPT and DM. A study in Brazil examined chronic disease multimorbidity in TB patients. Whilst they reported a lower prevalence of MM than we found (1.14%; higher in older age groups), their results highlighted the importance of MM as they demonstrated that death from causes other than TB was higher, and cure rates lower in TB-MM patients [18]. While these conditions may co-exist, some interact, through either shared risk factors or pathophysiology; or one disease influencing susceptibility and outcomes of the other. For example, DM is associated with a 2–3 fold higher risk of TB [19]. Further research is therefore required to evaluate the proportion of TB cases attributable to DM in this high HIV/TB/DM setting.
MM was lower overall in HIV/ART patients compared to patients not on ART with unknown HIV status. However when stratified by age, we noted that in the younger age groups (18–35 and 36–45 years), MM was higher in HIV/ART patients, in particular, there was a higher prevalence of HPT, DM and TB. One possible reason for this difference is the previously reported association between HIV/ART and premature and accelerated ageing [20]. This could also be due to increased awareness of NCD among HIV/ART patients, and possibly increased access to NCD screening in ART clinics. Obesity in HIV-infected patients is an emerging issue in South Africa; with some antiretroviral drugs, such as non-nucleoside reverse transcriptase inhibitors currently in use in South Africa, contributing to lipodystrophy and truncal obesity, increasing the risk of DM, HPT, and metabolic syndrome [10, 21]. A study of MM in HIV-infected patients in the United States found a prevalence of MM of 65%, with prevalence increasing with increasing body mass index (BMI) [22]. In the 46–55 age group, while HPT prevalence was similar between groups, DM prevalence was higher in the HIV/ART group; possibly highlighting a previously reported association between an increased risk of dysglycaemia in HIV-infected patients on ART [21].
Multimorbidity results in complex disease patterns that may have multiplicative, and not merely additive, consequences on health outcomes; and could diminish patients’ ability to manage their condition and enact behavior changes that may be required to improve health. This increasing complexity impacts on both health services, through more intensive health care requirements, and on health providers, with an increased requirement for integrated generalist care at the primary care level. This changing pattern of disease will therefore require health policy and interventions that differ from traditional vertical approaches and single disease management such as integrated management of chronic disease patients considered to be stable [23]. Although there is data paucity on the cost benefit of integrated systems, existing data suggests that integrating chronic disease services into existing HIV care may improve cost effectiveness [24]. For example, established systems for the delivery of ART and TB medications could be adapted to include essential drugs required for NCD management in order to streamline healthcare delivery and potentially improve adherence to chronic medications [25]. Similarly, funding mechanisms that usually fund NCD research could facilitate further research into integrated management building on existing HIV care infrastructure established through HIV funding mechanisms such as the President’s Emergency Plan for AIDS Relief (PEPFAR) [26].
Furthermore the associative patterns of MM described in this study suggest active bidirectional and targeted screening for these conditions should be implemented. Routine active screening is likely to result in an even higher burden of diagnosed co-morbidities in the short term, but diagnosis and intervention at an earlier stage may ultimately result in reduced overall cost. The impact of active screening on the health care system should therefore be evaluated. Beyond these direct interactions, MM and the associated increased complexity could also influence the psychological state or patients’ beliefs and values, influencing decision-making and acceptability of treatment options, and adherence to treatment [27].
Strengths and limitations
A significant strength of this study is the use of the unique patient identifier number which enabled any treatment prescribed within the public health system in the Western Cape province, even if outside the primary care clinic, to be identified and included in the study. The availability of linked records in a public sector primary care setting is rare both within South Africa and sub-Saharan Africa. This study utilized data from routine databases of prescribed drugs. Diagnoses could therefore not be verified. As a result, only patients with diagnosed chronic diseases receiving treatment for the selected diseases were identified and included in this study. This could underestimate the prevalence of the individual conditions and MM. Patients in chronic care for ART, HPT and DM are plausibly more likely than TB patients to have blood pressure and urine glucose measured over time due to regular clinic visits where routine observations include these measurements; while HIV testing is routinely performed in TB but not HPT and DM clinics. Therefore ascertainment bias with the potential for under ascertainment among TB patients could be a factor. Similarly, in the context of high HIV prevalence in this setting, as emphasized in the methodology, it is important to interpret the HIV stratification within the context of ART versus non-ART as the ‘non-ART/HIV unknown’ subset are likely to include HIV-infected patients who have not been diagnosed or who are not on ART. Another potential limitation was the use of prescribed drugs as a proxy for diagnosis. Whilst medications for DM, HIV, and TB are relatively specific to these diseases, the prescription of hydrochlorthiazide, enalapril or amlodipine may be prescribed for cardiovascular diseases and may not be specific to HPT. However, given the high prevalence of HPT of all chronic diagnoses in primary care [14], and the prescription patterns of doctors at the primary care level, we are confident that this proxy is a valid estimate of HPT. Prior to this study, to confirm prescription patterns in a primary care setting, we conducted a folder review of 100 patients attending another primary care clinic in Khayelitsha and found that these 3 drugs accounted for all HPT patients reviewed.
From a health system perspective, despite these limitations, this study highlights the significant burden of MM among patients receiving chronic disease care at the primary health care level.