District-level strategies to control the HIV epidemic in Zimbabwe: a practical example of precision public health

We conducted a descriptive cross-sectional study using survey and programme data to assess district-level performance along the HIV care cascade (HIV testing target achievement, linkage to ART and ART coverage) in order to formulate district-specific recommendations, taking into consideration prevalence and yield of testing. Data from 60 districts were analysed. Forty-eight districts (80.0%) surpassed 90% of their 2018 HIV testing targets. Linkage to ART was less than 90% in 40 districts (83.3%). Thirty districts (50.0%) had ART coverage above 90%. Of the 30 districts with suboptimal (< 90%) ART coverage, 18 districts had achieved high HIV testing target but with suboptimal linkage to ART, 6 had achieved high HIV testing targets and high linkage to ART, 4 had both suboptimal HIV testing target achievement and linkage to ART and 2 had suboptimal HIV testing target achievement and high linkage to ART. Priority should be given to districts with suboptimal ART coverage. Remediation strategies should be tailored to address the poorly performing stage of the cascade in each of the districts.

Most countries show national data, without evidence on whether this mirrors the data at sub-national levels. Hence, most national programmes, including those in Zimbabwe, do not provide recommendations adapted for district-level indicators. At the same time, donors call for targeted and cost-effective strategies, due to dwindling funds [4,5]. Advances in information technology provide an opportunity for data-driven health interventions. In our study, we combined mapping data on HIV prevalence [6] with data from the Zimbabwe National ART Program 2018. We studied the district-level performance along the HIV care cascade (HIV testing target achievement, linkage to ART and ART coverage) and formulated districtlevel recommendations considering HIV prevalence and HIV testing yield.

Methods
We conducted a descriptive cross-sectional study and retrieved the district-level HIV prevalence from the mapping study [6].
For each district, the number of people tested and the number tested positive and initiated on ART in 2018, and the total number of clients on ART at the end of 2018 were obtained from the Ministry of Health and Child Care (MoHCC), District Health Information System 2 (DHIS 2) [7]. Permission was obtained from the relevant authorities.
To estimate the district-level HIV testing achievement in 2018, we first calculated the district-level targets. We calculated the percentage of PLHIV in every district by dividing the estimated number of PLHIV in a district with the estimated total number of PLHIV in the country in 2018. We then use this percentage to distribute the overall national HIV testing target for 2018 to the districts [8,9]. Second, we calculated the testing coverage by dividing the number of tests conducted with the district target (number of tests proposed by each district) [10]. The HIV testing yield was calculated by dividing the number of positive tests with the number of tests conducted per district.
Linkage to ART was estimated by dividing the number of patients that were started on ART in 2018 with the number of positive tests in the same year (with the assumption that under the World Health Organization's HIV "Treat All" recommendations, all those tested positive had been put on ART [11]). The ART coverage was estimated by dividing the number of patients reported as active on ART at the end of 2018 in a district with the total number of PLHIV in that district.
HIV testing target achievement, linkage to ART and ART coverage were categorised into low (< 70%), medium (70-90%) and high (> 90%) and the districtlevel categorisation was visualized on maps using geographic information systems (GIS). We also described district-level performance along the HIV cascade for those districts with suboptimal (< 90%) ART coverage. These were grouped as follows: 1. Suboptimal HIV testing target achievement/suboptimal linkage to ART, 2. Suboptimal HIV testing target achievement/high linkage to ART, 3. High HIV testing target achievement/suboptimal linkage to ART and 4. High HIV testing target achievement/high linkage to ART.

Results
We included all the districts of Zimbabwe in the analysis. District-level categorization of HIV prevalence, HIV testing target achievement, linkage to ART and ART coverage are shown in Fig. 1.

Discussion
In Zimbabwe, half of the districts had high (> 90%) ART coverage, and thus had achieved the second target of the UNAIDS 90-90-90 targets [2]. However, substantial within-country variations in terms of HIV prevalence, HIV testing coverage, HIV testing yield, linkage to ART and ART coverage were observed. We have proposed four categories of districts according to gaps along the HIV cascade, for which district-specific recommendations can be formulated.

Suboptimal HIV testing target achievement and suboptimal linkage to ART
In the four districts with both suboptimal HIV testing achievement and linkage to ART, both the HIV prevalence and the HIV testing yield were above average, highlighting an important unmet need for HIV testing services. Therefore, HIV testing services should become more widely available. It has been seen that widespread door-to-door HIV testing substantially increases the acceptance of HIV testing and should be considered as a priority [12]. In addition, strategies linking those who tested positive to ART should be strengthened (elaborated below).

High testing target achievement and suboptimal linkage to ART
Eighteen districts with suboptimal ART coverage that achieved their HIV testing target but with suboptimal linkage to ART should assess if strategies known to improve linkage to care and ART initiation are in place. These strategies include same-day ART initiation, community (home-based) ART initiation, decentralisation of ART services to the primary health care level and integration of HIV care in other health care services [13][14][15][16][17]. Linkage to ART should be regularly assessed at the health facility level. Those who are diagnosed with HIV but have not been started on ART should be tracked. Consent for tracking should be incorporated into the testing strategy [18]. Patients diagnosed at higher-level referral health facilities should be initiated on ART before down referral to lower-level health facilities for follow up, with tracking of arrival after referral [19]. Some districts may have specific challenges. Poor linkage to ART around Kwekwe and the surrounding districts might be explained by the presence of illegal artisanal miners in the region. Illegal artisanal miners are highly mobile, and strategies to link and retain these highly mobile populations should be identified [20,21]. Also, in districts bordering Zambia and Mozambique (Centenary, Mudzi and Nyanga), the poor linkage may be due to patients crossing the border as a result of the economic challenges in Zimbabwe. The Southern African Development Community (SADC) HIV and AIDS Cross Border Initiative should be fully implemented to enable the provision of care and tracking of such patients [22][23][24].

High testing target achievement and high linkage to ART
Six districts with suboptimal ART coverage achieved their HIV testing targets and had high linkage to ART. In such a scenario, other indicators may assist in the formulation of a district-specific strategy. If the prevalence and HIV testing yield are lower than average, HIV testing may need to be delivered in a more targeted manner. Health facility-based strategies in combination with community testing in high-risk groups may be most efficient. Health facility-based strategies that have worked elsewhere include index case testing, targeting sexual partners and HIV-exposed infants and intensified provider-initiated testing and counselling (IPITC) [25][26][27][28][29][30][31]. Community testing should prioritise subgroups with a higher prevalence. Key populations and hotspots identified from a mapping exercise conducted in Zimbabwe may guide programming [32]. Social network testing, using peer educators, can be a useful tool to reach some of the key populations [30].

Suboptimal HIV testing target achievement and high linkage to ART
There were two districts with suboptimal ART coverage that had suboptimal HIV testing target achievement but with high linkage to ART. In both, the HIV prevalence and testing yield were higher than average. Hence, HIV testing strategies recommended for high prevalence settings, discussed above, should be considered.
In conclusion, there is substantial within-country variation in terms of HIV prevalence, HIV testing target achievement, HIV testing yield, linkage to ART and ART
Additional file 1: Figure S1. Yield per district among HIV tests performed in 2018 in Zimbabwe.