The impact of home-based HIV counseling and testing on care-seeking and incidence of common infectious disease syndromes in rural western Kenya

After three decades of the HIV epidemic in sub-Saharan Africa, in many parts of the continent many people still do not know their HIV status [1–5]. This is due to multiple factors including low perception of risk, distance to HIV testing centers, costs incurred in getting tested, and ongoing stigma of being HIV-infected [6, 7]. Two national AIDS indicator surveys have been done in Kenya. In the first survey in 2007, only 34% of adults nationwide had been tested before, and among those found to be HIV-infected, only 16% correctly reported their HIV status. These proportions increased to 72% and 47% respectively in the second survey in 2012 [1, 2, 4]. Without knowledge of their HIV status HIV-infected persons do not access HIV care and treatment services, including cotrimoxazole prophylaxis for opportunistic infections and antiretroviral therapy (ART), nor are they likely to receive adequate treatment for acute illness episodes [8–11].

Because the majority of HIV-infected persons not in care are those that do not know their HIV status, mass initiatives to test more persons for HIV are needed [12]. One strategy for increasing the number of individuals who know their HIV status is home-based HIV counseling and testing (HBCT) [4, 13, 14]. HBCT can overcome barriers of cost, time-expenditure and stigma that prevent many persons in rural Africa from seeking voluntary counseling and testing (VCT) at fixed facilities [13–15]. While it has been shown that HBCT can increase HIV testing and the number of people who know their HIV status, it is not clear whether knowledge of HIV status influences care seeking which in turn impacts incidence of common infectious illnesses such as acute respiratory illness, febrile illness and diarrhea whose burden is high in several parts of Africa [16]. We hypothesize that as more people know their HIV status, enroll into HIV care and treatment, and receive highly-active antiretroviral treatment and cotrimoxazole prophylaxis that the incidence of infectious disease morbidity will decrease in the community. We further hypothesize that as people newly learn their HIV-status they will be more likely to seek care for common infectious disease syndromes, either because their access to care has improved or their health awareness and concern has increased due to learning their HIV status. We describe the impact of a large-scale HBCT program among newly diagnosed HIV-infected residents in the year after HBCT in a high HIV prevalence population under ongoing surveillance for common infectious disease syndromes in rural western Kenya.

Among the 698 HIV-infected first-time testers, 219 (31%) had enrolled in HIV care by 3 months after HBCT, 50% had enrolled at one year and 58% by 24 months (Figure 2). Over half (56%) who enrolled in care enrolled at Lwak Hospital. There were no age or sex differences in the proportions of HIV-infected people enrolled at PSCs versus those HIV-infected but not enrolled (data not shown). CD4 counts were available from 635 (91%) of the first-time testers. No significant differences were observed in the median CD4 counts for those who enrolled for HIV care versus those who did not (402 cells/mm³, inter-quartile range 240 – 616 cells/mm³ versus 485 cells/mm³, inter-quartile range 347 – 583 cells/mm³). Of those enrolled at the PSCs according to national guidelines at the time, 92% were initiated on cotrimoxazole prophylaxis and 37% of people with CD4 counts ≤250 cells/mm3 started taking antiretroviral drugs (ARVs) (Figure 3).

Among the group enrolled at PSCs, the overall hospitalization rate in the post-HBCT period was 2.5 times higher (RR 2.57, 95% CI 1.59 – 4.16) than the pre-HBCT period (Table 4). No changes were observed in the hospitalization rates of those HIV-infected not enrolled at PSCs or those who were HIV-uninfected.

Our study was unique in showing that a mass campaign of HBCT led to improvements in health-care seeking and decreases in incidence of some common infectious diseases among newly diagnosed HIV-infected persons. HIV counseling and testing is considered the entry point to HIV care, including access to cotrimoxazole prophylaxis and antiretroviral treatment [9, 12, 24–26]. For those persons who did access a PSC, the majority received appropriate treatment for HIV. The greater than 90% use of cotrimoxazole in those enrolled for HIV care in our study area was commendable and was in line with the Ministry of Health’s guidelines for universal use of cotrimoxazole irrespective of CD4 levels [27]. We found that 37% of newly-diagnosed HIV-infected persons who qualified for ARV (CD4 count <250 cells/mm³) had started this treatment.

Reductions in morbidity, mortality, HIV incidence and transmission, are the ultimate goal of large-scale efforts at counseling and testing as people may likely enroll for HIV care if they know they are HIV-infected [12, 13]. Our study was one of the first to demonstrate a likely causal sequence between HBCT, accessing HIV care and treatment, and reduction in infectious disease morbidity. Among those newly diagnosed persons who accessed a PSC, we observed a 16% decline in incidences of diarrhea and acute febrile illnesses. The decline likely resulted, at least in part, from cotrimoxazole prophylaxis, which is known to reduce infectious disease morbidity [9, 14, 28–33]. Cotrimoxazole prophylaxis has been shown to reduce diarrhea in several previous studies in Africa, particularly from bacterial and parasitic causes [31–33]. Cotrimoxazole also has been demonstrated to be a potent prophylactic against clinical malaria, with reductions of more than 70% [31]. The Lwak area is a malaria holoendemic area, where both children and adults suffer from several clinical malaria episodes per year [19]. Cotrimoxazole also likely reduces the occurrence of other causes of AFI besides malaria, such as mycobacterial infections, bacteremia, and fevers of unknown etiology [32, 33]. In addition to cotrimoxazole, initiation of ART has been shown to decrease both diarrheal illness and undiagnosed acute febrile illness in the first year after initiation of treatment. Additionally, ART leads to reduction in the incidence of known causes of AFI, such as bacteremia. We did not see a significant decrease in respiratory illness among those newly tested HIV-positive compared to the other two groups. This may be due to the fact that much of ARI is caused by viruses, which are generally not affected by cotrimoxazole prophylaxis. Severe ARI is a rarer syndrome and we might not have had the statistical power to detect a significant decrease in our sample. Moreover, there was a secular trend towards decreased SARI in the population, as witnessed in the HIV-negative group, which might have masked any specific impact of HIV treatment on SARI rates.

Care-seeking for illnesses increased more among the HIV-infected group in the post-HBCT period than the uninfected group. Care-seeking increased more at Lwak hospital than in the other three area clinics, which are run by the Ministry of Health (MOH). The greater increase in care-seeking at Lwak hospital might have been driven by several factors. First, the PBIDS field staff making the biweekly home visits encouraged study participants to go to Lwak hospital when ill. Second, Lwak hospital provides free high-quality care to enrolled study participants for all potentially infectious disease syndromes, including free hospitalization if needed, which is not available in other area MOH clinics [34]. Since access to free care was available at Lwak Hospital in pre-and post-HBCT periods, a key difference in motivation may, therefore, have resulted from an increased knowledge of one’s HIV positive status. Indeed, similar increases in attendance were not observed for the HIV-uninfected group who served as a comparison group to identify secular trends in health-care seeking in the pre versus post-HBCT period.

We believe that the observed increase in hospitalizations among those with HIV enrolled at PSCs was unlikely to be reflective of a true increase in severe disease among this group. As discussed, there was a general increase in healthcare seeking among HIV-infected persons, and therefore sicker patients were more likely to be hospitalized. This was particularly true at Lwak Hospital, because inpatient care was free for PBIDS participants and standardized clinical criteria were used for decisions on admission [16].

Approximately half of newly-diagnosed HIV-infected persons attended PSCs in the follow-up period. Many entered HIV care when their CD4 count was still relatively high (median 402 cells/mm³), which enhanced their ability to start prophylaxis before their disease had progressed. The reasons why half of those testing positive for the first time did not access HIV care are not clear. One possibility is that we missed some newly diagnosed persons who actually did attend a PSC. We only followed up persons in their homes one month after testing and then could only rely upon linking the names and residences of persons from HBCT with those in the registers of the area PSCs. This effort is limited by the fact that often persons use alternate names and change residences frequently [7]. It is also possible that individuals sought care in a clinic outside of the study area for reasons of anonymity. Second, there is likely still a barrier between HIV-testing and seeking care [7, 35–37]. While doing HIV-testing in the home clearly overcame some of the obstacles of stigma, as well as the actual and opportunity costs of traveling to a fixed VCT center, the same obstacles would still exist for those seeking care at a PSC [6, 37]. Further strategies are needed to improve the uptake of HIV care services after HBCT.

Our study had several limitations. Our data was limited to only one year of pre- and post-HBCT surveillance. It is possible that decreases in disease incidence among HIV-infected persons was due to annual variations in disease burden, which might not have been observed with more years of pre- and post-HBCT surveillance. However, the fact that we found differences in the reduction in disease between HIV-infected persons on treatment versus HIV-infected persons not on treatment and HIV-uninfected persons suggests that the reduction in disease was specific to receipt of HIV care and treatment. Second, the limited time periods of surveillance did not allow us to monitor the impact of care seeking on mortality as this is still a relatively rare event. Moreover, temporal factors might have influenced morbidity and mortality in this time period, particularly post-election violence that occurred in Kenya in 2008 [18]. Third, we assumed that the HIV status at the time of HBCT was the same for that individual in the year before and after testing. While we believe misclassification of HIV status was rare at the population level (1.5% annual incidence), it might have introduced slight bias in the morbidity findings due to acute illness symptoms with acute HIV seroconversion [38]. Lastly, we only followed-up HIV-positive clients at home one month after HBCT, which would not have documented those who sought care later; although we did search records for local PSCs later to try to capture those who might have registered later.

In summary, we demonstrated that HBCT led to a substantial number (approximately 700 or 11% of first-time testers) of persons learning their HIV status for the first time. Newly diagnosed HIV-infected persons had a reduction in common infectious disease syndromes, likely through increased access to HIV care and treatment. While we did not demonstrate it, other studies suggest that HBCT, by improving access to HIV care and treatment, could lead to reductions in mortality and HIV transmission [12, 24, 25, 30]. We believe such impact of HBCT on morbidity can be realized in other African populations in which many persons do not know their HIV status. However, we do caution that in order to realize the full benefits of HBCT enhanced efforts are needed to ensure those who test positive successfully enter HIV care and treatment.