Previous title: HIV index testing services in urban Lusaka: a review of medical records

Background: As the proportions of people living with HIV (PLHIV) who do not know their HIV infection status decrease, reaching the remaining few who are asymptomatic and not in contact with the health care system becomes a critical challenge. Therefore, reaching the first 90 of the UNAIDS 90-90-90 targets will require effective and efficient HIV testing approaches. The number of PLHIV who know their HIV status and who receive antiretroviral therapy could increase by the expansion of index testing services. Methods: This project was a retrospective study looking at medical records of HIV positive clients who were elicited for index testing between October and December 2019. It was conducted in three high volume health facilities in Matero Urban sub-district 3 in Lusaka, Zambia. Results: The HIV test outcomes for index contacts were as follows: 452 index contacts (53.5%) tested HIV negative, 113 index contacts (13.4%) tested HIV positive, 108 index contacts (12.8%) were known HIV positive, and 172 index contacts (20.4%) were not yet tested for HIV. Of the 113 contacts who tested HIV positive, 90 index contacts started anti-retroviral therapy within 7 days (79.6%). The total number of 845 contacts were elicited from 604 index clients, giving a low elicitation ratio of 1:1.4. There was not much difference between gender for elicited contacts (423 men and 422 women). A total number of 565 index contacts were eligible for HIV test. 113 of them tested HIV positive, representing a positivity yield of 20%. Pearson Chi-Square test value was 0.498 and the p value was 0.481. This result is not significant since p value (0.481) is greater than the designated alpha level (0.05). Conclusions: HIV programs need to explore and address barriers to HIV partner testing services to avoid over-testing and maximize HIV case identification (thus, improving HIV testing positivity yield). Open Peer Review


Introduction
According to the 2018 UNAIDS Global AIDS Update 1 , there are an estimated 36.9 million people living with HIV (PLHIV). Recently, marked progress on HIV test and treat strategy has been achieved by countries' commitment to achieve the UNAIDS 90-90-90 targets by 2020 1  The cornerstone for achieving the UNAIDS 90-90-90 targets by 2020 begins with PLHIV knowing their status. As the proportions of those living with HIV who do not know their HIV infection status decrease, reaching the remaining few who are asymptomatic and not in contact with the health care system becomes a critical challenge. Therefore, reaching the first 90 goal will require effective and efficient HIV testing approaches. In Zambia, about 25% of adult men living with HIV didn't know their HIV status in 2018. A study conducted in Zambia revealed that index testing and targeted community-based HTS are effective strategies to identify men living with HIV 5 . Men and young people have shown limited uptake of HIV testing services under standard facility-and community-based services. Approaches such as HIV self-testing delivered at scale using several different models reached a high proportion of men, young people and first-time testers in Malawi, Zambia and Zimbabwe 6 .
The number of PLHIV who know their HIV status and who receive antiretroviral therapy (ART) could increase by the expansion of index testing services. This will result in the reduction of the number of people who can transmit the virus, and subsequently in reduced new HIV infections. In another qualitative study conducted in Malawi and Zambia, most participants considered different approaches to partner HIV testing to be acceptable. However, there are concerns about each and implementation challenges need to be addressed 7 .
The objective of this study was to review existing medical files and registers in Matero subdistrict of Zambia in order to describe existing information on index testing and propose better ways to improve HIV index testing positivity yield.

Study design
This was a retrospective study looking at index registers of clients who tested HIV positive and were elicited for index testing between October and December 2019. The study was conducted between January and February 2020 in three high volume health facilities in Matero sub-district 3 of Lusaka district in Zambia. The study facilities included Matero First Level Hospital, Matero Main Clinic, and George Health Centre. The overview results of the study, which looked at the effectiveness of HIV index testing, were described. The analysis examined index clients' identification, elicitations of index contacts, and testing of index contacts. The main quantitative outcome of interest for this analysis was the success of index testing to improve yield for HIV Testing Services (HTS) among female and male, and across ages among index clients; and secondly ART initiation for positive index contacts.

Sampling
This retrospective study used a total sample enumeration technique.
The study population comprised all index clients (males and females at the study facilities) who had been diagnosed with HIV, gave informed consent and were elicited for HIV index contact testing during the study period.

Inclusion criteria:
HIV positive clients (index clients or index cases) and their sexual contacts (sexual partners of index clients who have been elicited and offered HIV index testing services). The study participants included: • HIV positive clients identified through either voluntary counseling and testing (VCT) or provider-initiated counseling and testing (PICT)

Amendments from Version 1
The title has been reviewed and improved to "HIV index testing services in urban Lusaka: a retrospective review of medical records" to be in line with the narratives. The aim of the study was refined specifying objective to review of existing medical files/registers instead of medical records in Matero Subdistrict of Zambia for more clarity. We have also conducted a more detailed literature review on index testing in the introduction section to support the importance of the study. Under the methods section, the term "phase" (under sampling) was deleted since this particular study was not conducted in phases. The study population was specified to males and females. The exclusion criteria were clarified, excluding index clients identified through other service entry points other than VCT and PICT such as MCH and VMMC. The variables tracked were also aligned in accordance with the study objective. The article now mentions the details of informed consent taken from the PLHIV while approaching their contacts for HIV testing. Additional analytical framework detailing total sample, and the details of contact tracing in cases where sexual partners were not the only contacts has been provided. We used the STROBE cross sectional reporting guidelines to ensure the study meets international standards for peer reviewed articles. We summarized the results in the narrative for better presentation. The discussion section of the study has been rewritten including recent references. The conclusion has also been refined in line with the study findings.
The revised version has also acknowledged various contributors to index testing in Lusaka.

REVISED
• Being documented in HIV index registers • Having elicited at least one sexual partner

STROBE cross sectional guidelines
We used the STROBE cross sectional reporting guidelines to ensure the study meets international standards for peer reviewed articles 8 . A checklist was completed by entering the page numbers from the manuscript where readers can easily find each of the listed items. Where the article didn't currently address all the items on the checklist, the text was modified to include the missing information. Where certain that an item does not apply, we wrote "n/a" and provided a short explanation.

Results
The total number of index clients included in the study was 604. Matero First Level Hospital leads the participation per facility with 292 participants, followed by George Health Centre and Matero Main Clinic with 164 and 148 participants, respectively. The total number of female participants was 314 (representing 52%) and male participants was 290 (representing 48%) ( Table 1).
The age of participating index clients ranged from 16 to 78 years, with mean age calculated at 34 years (SD = 9.1). Out of the total number of 604 participants, 514 clients (85.1%) were married, 85 clients (14.1%) were unmarried, 3 clients were widowed, and 2 clients were divorced.
Concerning the time spent from HIV test to the initiation of ART for index cases: 595 index clients started ART within 7 days (98.5%), 1 index client started ART within a month (0.2%), 1 index client started ART after 1 month (0.2%), and there was no evidence of starting ART for 7 clients (1.2%).
The mean age of elicited contacts was calculated at 33 years (range, 17-80 years SD = 9.4). From the total number of 845 The total number of 565 index contacts were tested for HIV and 172 index contacts had not yet been tested for HIV (Table 3). The Pearson Chi-Square test value was calculated at 0.498 and the p value was 0.481.
The overall index testing cascade is represented in Figure 1 below.
The total number of HIV positive cases reported during the study period were 617, out of whom 13 did not accept index testing (had no recorded elicitation). We excluded 57 cases because of elicitation of children or siblings only as contacts (without any sexual partner elicited).

Discussion
The overall key findings of our study are shown in the result section. The 98 percent acceptability rate for index testing services in Lusaka is at an agreeable level. Though, continuous efforts are needed to cover every who is eligible for the service. The index cases to index contacts elicitation ratio of 1 to 1.4 is below the documented ratio of 1 to 1.8 in Zambia 5 . There was not much difference between gender for elicited contacts (423 males and 422 females). A total number of 565 index contacts were eligible for HIV test. 113 of them tested HIV positive, representing a positivity yield of 20%. This index testing positivity yield is below the expected yield of above 25% as reported by several other studies 9-17 . It is not in keeping with many studies that have shown that index partner testing has the potential to increase HIV testing services (HTS) uptake; identify and diagnose HIV infected partners (yield ranging from 35% to 62% without reported intimate partner violence (IPV) 9 . The current study has also revealed that only 80 percent of named contacts were reached with index testing services. Limitations to index testing such as relationship conflict have been documented 7 . For partner notification, additional barriers included women losing letters, being fearful to give partners letters, being unable to read and men refusing to come to the clinic, lack of privacy or confidentiality and stigma 7 . Other implementation challenges in personnel, resources or space have also been noted 7 . Specific barriers to index testing in Zambia need to be explored and addressed for optimal index testing positivity yield. The current linkage rate for positive contacts is 79.6%. Most index clients (98.5%) had documented evidence of starting ART within 7 days of HIV diagnosis. This demonstrates strongly that the test and start

Open Peer Review
study is not in line with the findings and the analysis done. I also recommend that the title be reviewed and be improved to be in line with the narrative.

Introduction and Method:
This study did not have phases so delete term phase under-sampling.

○
The study population is non-specific -when do we consider someone to be a man or woman vs a boy or a girl? Best to use age categories, as this study reached out to 16-year-olds who are girls and boys. 1.
The article needs to mention details of informed consent taken from the PLHIV while approaching their contacts for testing for HIV. 2.