Uptake of same-day initiation of HIV treatment in Malawi, South Africa, and Zambia as reported in routinely collected data: the SPRINT retrospective cohort study

Background: Since 2017 global guidelines have recommended “same-day initiation” (SDI) of antiretroviral treatment (ART) for patients considered ready for treatment on the day of HIV diagnosis. Many countries have incorporated a SDI option into national guidelines, but SDI uptake is not well documented. We estimated average time to ART initiation at 12 public healthcare facilities in Malawi, five in South Africa, and 12 in Zambia. Methods: We identified patients eligible to start ART between January 2018 and June 2019 from facility testing registers and reviewed their medical records from HIV diagnosis to the earlier date of treatment initiation or 6 months. We estimated the proportion of patients initiating ART on the same day or within 7, 14, 30, or 180 days of baseline. Results: We enrolled 825 patients in Malawi, 534 in South Africa, and 1,984 in Zambia. Overall, 88% of patients in Malawi, 57% in South Africa, and 91% in Zambia received SDI. In Malawi, most who did not receive SDI had not initiated ART ≤6 months. In South Africa, an additional 13% initiated ≤1 week, but 21% had no record of initiation ≤6 months. Among those who did initiate within 6 months in Zambia, most started ≤1 week. There were no major differences by sex. WHO Stage III/IV and tuberculosis symptoms were associated with delays in ART initiation; clinic size and having a CD4 count done were associated with an increased likelihood of SDI. Conclusions: As of 2020, SDI of ART was widespread, if not nearly universal, in Malawi and Zambia but considerably less common in South Africa. Limitations of the study include pre-COVID-19 data that do not reflect pandemic adaptations and potentially missing data for Zambia. South Africa may be able to increase overall ART coverage by reducing numbers of patients who do not initiate ≤6 months.


Introduction
In 2017, the World Health Organization (WHO) began recommending "same-day initiation" (SDI) of treatment for patients with HIV considered clinically and personally ready for antiretroviral therapy (ART) 1 . This recommendation was based on a series of randomized trials and observational studies that had demonstrated improved retention in care and viral suppression rates for patients offered SDI, compared to those offered what was then the standard of care, which typically required multiple clinic visits before a patient received an initial supply of antiretroviral medications (ARVs) 2 . Although determining a patient's "readiness" for SDI ultimately relies on provider and patient discretion, reasons for delaying initiation include indications or suspicion of opportunistic infections such as tuberculosis or cryptococcal disease and a patient's expressed desire for more time to accept the HIV diagnosis or disclose it to others 2 .
In the clinical trials, most of the improvement in outcomes arose from a reduction in loss to follow up between testing positive for HIV and initiating ART. Observational studies have reported that among patients actually starting ART on the same day under routine care, loss to follow up during the first year after initiation has tended to be higher than in the pre-same day initiation period 3,4 , but these studies have generally compared outcomes among patients who started on the same day to outcomes among those known to have started within a specified number of days, ignoring patients who never started ART at all.
In response to the WHO recommendation, many countries adopted SDI into their national guidelines. Instructions to healthcare facilities as to exactly how to implement SDI and to which patients it should be offered, however, were not precise, allowing the interval between a patient's first clinic visit and ART initiation to be determined by provider judgment and/or patient preference. Time intervals from a patient's initial presentation (HIV diagnosis or first HIV-related healthcare interaction) to initiation of ART are not routinely reported in electronic medical record (EMR) systems, which typically only create a record for a patient starting at the time of ART initiation. Little evidence is available on the proportion of patients actually starting ART on the same day or the characteristics of those who do or do not, including age, sex, and other characteristics. Without this information, it is difficult to understand observed differences or identify opportunities for improving care.
This paper reports quantitative results of the SPRINT study (Survey of Procedures and Resources for Initiating Treatment of HIV in Africa). It documents uptake of SDI, time to ART initiation for those not receiving SDI, and predictors of SDI for adult men and women in three high HIV burden countries in sub-Saharan Africa. As this was a retrospective review of medical records and no identifiers were collected, a waiver of informed consent was provided. Data transcription from patient records began on May 27, 2020 in Malawi, September 28, 2020 in Zambia, and November 12, 2020 in South Africa.

Ethics
IRB approval from the University of the Witwatersrand was sought to allow individuals on the team who were based in South Africa to access Malawi data for analysis purposes. There was no need for University of the Witwatersrand approval in order to collect the data, as that was only under the purview of the IRBs of Boston University and the National Health Sciences Research Committee in Malawi, whose approval was secured prior to the start of data collection. All required approvals were in place before data collection began in all three countries. Team members at the University of the Witwatersrand did not access the data from Malawi prior to their own institutional IRB approval.

Study sites
SPRINT was a retrospective record review conducted in Malawi, South Africa, and Zambia. In each country, in collaboration with national health officials, we first purposively selected provinces and districts that provided diversity of setting (rural/urban) and of nongovernmental supporting partner organizations. We then identified a convenience sample of public sector primary healthcare clinics within these provinces or districts and conducted visits to these sites to assess facility-level and HIV testing and initiation-specific aggregate indicators. The final sites selected provided diversity in size and geographic setting, had relatively strong routine data collection, and were welcoming of the research team. Selected study sites are described in Table 1. All of the study sites offered universal treatment and followed national guidelines for

Amendments from Version 1
In response to reviewer comments and suggestions, we have provided further details about data collection, added more variables to the adjusted analysis, and strengthened the discussion of the results. In addition, we have posted a supplementary table with site-level data.
Any further responses from the reviewers can be found at the end of the article REVISED ART treatment eligibility 5-7 ; eligibility criteria included confirmed HIV-positive status and provider and patient concurrence that the patient was ready to start treatment, based on clinical and personal characteristics. The SPRINT studies are registered with Clinicaltrials.gov (NCT04468399, July 13, 2020, Malawi; NCT04170374, Nov 20, 2019, South Africa; NCT04470011, July 14, 2020, Zambia).
At the time of study enrollment, guidelines for ART initiation in the three study countries encouraged but did not require that SDI be offered to patients whom providers determined to be eligible, without waiting for CD4 count results. In all countries, guidelines encouraged starting HIV treatment as soon as possible 5-7 but also noted the potential need for delay among patients who indicated that they were not yet ready to start. The decision on whether a patient was eligible for same-day initiation was made by the authorized clinician, typically a clinical officer or trained ART nurse in Malawi and Zambia and a public health (senior) nurse in South Africa, each of whom was also authorized to prescribe antiretroviral medications once the decision to initiate was made. manually at all facilities, but for 92 of 534 participants (17%) distributed across all the sites, paper files could not be found. For these records, EMR data were extracted instead.
Records were reviewed from the point of HIV treatment eligibility (HIV-positive diagnosis) to the point of treatment initiation or for a 6-month period, whichever was shorter. By starting with HIV diagnosis, rather than with treatment initiation, the study captured information about patients who failed to initiate treatment within 6 months (dropped out of care before being dispensed medications) and allowed us to observe services received and days elapsed before initiation. We note that the HIV test used to identify study participants may not have been any individual participant's first positive test anywhere, but only their first positive test at the study site during the study period. Because repeat testing is common 8 , we assume that many participants had previously tested positive at another location but failed to initiate treatment at that time or place.

Results
We identified a total 3,374 participants whose ART initiation dates were between 01 January to 31 December 2018 in South Africa or between 01 July 2018 and 30 June 2019 in Malawi and Zambia. After excluding 31 participants who were found to be ineligible (under 18 years old or initiated treatment outside the study period), the final analytic sample included 3,343 participants (825 patients in Malawi, 534 in South Africa, and 1984 in Zambia). We did not meet sample size targets due to resource limitations and COVID-19 restrictions that shortened the data collection period. CD4 counts and TB status were not available for patients in Malawi. There were no important differences in patient characteristics among the three countries, as shown in Table 2.
Time to ART initiation Time to ART initiation is presented in Table 3. Same-day initiation was the norm in Malawi (734/825, 89%) and Zambia (1,813/1,984, 91%), where a vast majority of patients did start ART on the day of diagnosis or first HIV-related clinic interaction. It was much less common in South Africa (304/534, 57%), where a fifth of patients (116, 22%) did not have a record of initiating ART at all within the 6-month study observation period. Among patients who did not initiate on the same day, varying proportions proceeded to initiate within 1, 2, 4, and 24 weeks, as shown in Table 3. Those who did initiate within 6 months, but not on the same day, made a median of 1 additional clinic visit (2 in total) between diagnosis and initiation (inclusive.) Across all three countries, as shown in Table 3 and Table 4, we identified a total of 181 participants who did not initiate treatment within 6 months of HIV diagnosis or their first recorded HIV-related clinic visit (non-initiators). These patients comprised 8% (n=64) of the cohort in Malawi, 21% (n=114) in South Africa, and just 0.2% (n=3) in Zambia. We note that while these proportions are consistent with expectations in Malawi and South Africa, it is likely that non-initiators in Zambia who should have been included in our sample were not. Although we made multiple attempts to confirm the completeness of our cohort in Zambia, we believe it is likely that more than 0.2% of patients in Zambia who were eligible to initiate ART did not start within 6 months. Among those who did not initiate within 6 months, in Malawi most were either lost to follow-up or remained alive and in care, but not on ART, while in South Africa most were either lost to follow-up or were simply missing follow up records (Table 4), a category which likely includes both unreported (silent) transfer to other healthcare facilities and loss to care.
When we stratified results by country, clinic, and TB symptom status and ranked sites according to ART patient volume (Supplemental Table 1), our results suggested that patients presenting with TB symptoms in South Africa were less likely to initiate ART on the same day than were those with TB symptoms. In rural clinics in South Africa, the likelihood of SDI tended to decline as patient volume increased. We did not see these associations in Zambia or Malawi.  Median (IQR) number of clinic visits up to ART initiation for non-same day initiates (>day 0, ≤month 6 † ) 2 (2-3.5) 2 (2-2) 2 (2-3) 2 (2-2) 2 (2-2) 2 (2-2) 2 (2-2) 2 (2-2) 2 (2-2) *As explained in text, the very small number of patients reported as not having initiated ART in Zambia may reflect missing data. ART, antiretroviral treatment. † Includes visit at which ART was initiated.

Discussion
In this study, we found that by 2020, uptake of SDI was extensive in Zambia and Malawi, reaching 80-90% of patients initiating or re-initiating ART. It was much less common in South Africa, where just 57% of patients initiated on the same day. We did not find significant differences in time to initiation by sex; men were equally likely to start on the same day as women. Among patients who did not initiate same-day, roughly one third in Malawi, half in South Africa, and nearly all in Zambia went on to initiate within 6 months; the rest had no record of treatment initiation by the end of the 6-month follow-up period. Patients who initiated ART within 6 months but not on the day of diagnosis had a median of just two visits, including the initiation visit, suggesting that the initiation process has accelerated even outside SDI, with only one "pre-ART" visit occurring for most who do not receive SDI.
We found several other publications that reported the rate of SDI in various countries in sub-Saharan Africa, mostly at earlier time periods than our study, which was conducted between 2018 and 2020. A multi-country cohort study using data from The International epidemiology Databases to Evaluate AIDS (IeDEA) Network through 2018, including South Africa but not Malawi or Zambia, reported an overall rate of SDI of 64% 10 . Three studies from South Africa were all completed roughly one year earlier than our study, in 2017-2018. The first, in Johannesburg, reported 20% uptake of SDI 3 . The second, in Johannesburg and Limpopo Province, reported that 40% of patients received SDI, with an increase in uptake from 30.3% at the start of the study period to 54.2% at the end 6 . The third, with data from four provinces, reported 54% SDI uptake in 2018, an increase from 18% in 2016 11 . The difference between these findings and our result of 57% SDI in South Africa in 2018-2020 likely reflects the impact of an additional year of facility experience in implementing the SDI guideline.
In Zambia, uptake of SDI increased from 42% to 75% of newly initiating patients between the beginning of 2016 and the beginning of 2018, a phenomenon attributed largely to the adoption of universal treatment eligibility in 2017 12 . By the time our data were collected in 2018-2020, Zambia's SDI rate had increased to more than 91% (Table 3). We did not find any published data on SDI uptake in Malawi to compare to our estimate of 88%. Reports from neighboring countries showed SDI rates of 65% in Zimbabwe 7 and 52% in Mozambique in 2017 13 ; in Botswana uptake increased from 28% to 59% between 2018 and 2019 14 . SPRINT adds both geographic breadth and, importantly, more recent experience to these prior publications.
Reasons for uptake of SDI or of having delays in ART initiation in our study are not clear. In our data from South Africa and Zambia, only the presence of TB symptoms or a WHO stage of III or IV predicted a delay in initiation. This result likely reflects concerns about starting ART while patients may be experiencing co-infections, such as TB or cryptococcal meningitis. Global guidelines continue to regard these conditions as justification for delaying ART initiation 15 , but recent studies suggest that many patients with TB symptoms can and should start ART immediately, as the risk of becoming lost to follow up if ART initiation is delayed may exceed the very low risk of experiencing TB immune reconstitution inflammatory syndrome 16 . We note, however, that the timing of ART initiation in the presence of undiagnosed TB remains an area of active debate 17 . Other reasons for delaying initiation may include clinicians' individual views on SDI, patient preferences (e.g., concerns about stigma, disclosure, or side effects; feeling healthy or not in need of treatment 18 ), and/or lack of resolve or resources to adopt new SDI procedures at some facilities.
At the same time, valuable as SDI may be as a strategy for minimizing loss to follow up before starting ART treatment, it is likely that some patients should indeed delay initiation, as they require more urgent medical care before starting ART or have personal concerns about or barriers to treatment that cannot be addressed in a single visit 19,20 . The proportion of patients falling into this latter category is unclear, and it likely varies across different populations. Based on findings of the SLATE II trial in South Africa, a reasonable guess may be that 10-15% of patients are not good candidates for SDI 16 . If this is correct, then both Zambia and Malawi have achieved near-universal uptake of SDI. South Africa still has some distance to go, however, before it has saturated demand for SDI.
The advent of COVID-19 as a major health risk in all three study countries as of the second quarter of 2020 is likely to have changed both patient and provider behavior with regard to ART initiation. On one hand, COVID-19 morbidity, fear of transmission, and societal restrictions kept many potential patients away from the clinic. In South Africa, for example HIV initiations are known to have fallen by 28% in the first year of the pandemic 21 . Conversely, the desire to minimize personal interaction and clinic visits to avoid COVID-19 transmission could have favored SDI for patients who did make it to the clinic to start ART. It is also not clear whether provider and/or patient behavior will revert to earlier norms as pandemic restrictions wane or will sustain any changes that were made. A new round of data collection will be needed to answer those questions.
The SPRINT study had a number of limitations. Our final enrolled sample size was smaller than intended, due to resource limitations that shortened the data collection period. We relied entirely on routinely collected data, which resulted in many missing values for key variables. It is possible that the data sources we used for enrollment, such as HIV testing logs, missed some individuals who tested positive for HIV but did not register for HIV treatment. Despite every effort to enroll a representative sample of patients eligible for ART initiation, we are not confident that all patients who did not initiate ART during the study period were captured in Zambia, potentially leading to overestimates of total initiation in Zambia. Outcomes reported as "missing" in Table 4 may reflect unrecorded transfers to other healthcare facilities, rather than failure to initiate ART entirely, particularly in South Africa. Other country-and sitelevel differences in data completeness may also have contributed to the variations observed. Finally, we cannot discern from our data exactly why certain patients faced a delay in starting ART-whether it was a result of patient or provider preference, patient condition, social pressures, or some other barrier. We also cannot determine whether those who did not start on the same day were offered SDI and declined it or were simply not offered it at all. To achieve the full potential of SDI in increasing treatment uptake for those who are eligible and willing-and not creating risks for or coercing those who are not-a better understanding of how decisions are made at healthcare facilities would be of value.

Data availability
Underlying data Data used in this study, which were abstracted from routinely collected medical records at public sector healthcare facilities, are owned by the Ministry of Health or Department of Health in each study country and cannot be shared by the authors.

Open Peer Review
Thank you for asking me to review this interesting and important paper, which presents routinely collected data on SDI in Malawi, South Africa, and Zambia. A total of 3344 records from 29 health facilities were recorded, which is lower than the target sample size of 6400 due to resource limitations. Overall, 91% of non-pregnant adults in Zambia, 88% in Malawi, and 57% in South Africa initiated SDI. Predictors of not initiating SDI were clinical Stage 3 or 4 disease, and TB symptoms.
I have read the paper, and also the comments from 3 other reviewers. I agree with all comments by the other reviewers. This manuscript is clear and well written, and is an important contribution to the evidence base around SDI initiation. However, it's difficult to know if differences in how data are captured and recorded could be contributing to the country-level differences which were found.
I have included some additional details below, with the attempt at avoiding comments which have already been stated by the other reviewers: Abstract: Methods: I suggest substituting another phrase for "sequentially enrolled patients eligible to start ART" because this is a retrospective record review.

1.
Please state the eligibility criteria to start ART. 2.

Results:
Please describe the different study periods from South Africa, Malawi and Zambia. As it is currently written, it appears that this is 18 months in each country, but in the Methods section it clarifies that this is 12 months in each country, with different start and stopping periods. If possible, it would also be helpful to include the actual dates of inclusion, since it appears that these periods were shorter than 12 months (i.e. started at the last date in the 12-month period, and sequentially moved backward by day.

1.
Please add numbers as well as percentages for those who were offered and accepted SDI. Do you also have the number of patients who were offered SDI and didn't accept it? 2.
For clarity, please give the numbers of patients who had not started SDI, but started within 7, 14, 30, or 180 days by country.

3.
Please add p-values and confidence intervals for the predictors of SDI. 4.
Introduction: The last sentence states that this paper documents "update of SDI, time to ART initiation for those not accepting SDI, and predictors of SDI …." However, it appears that it is not possible to distinguish those who were not offered SDI from those who did not accept SDI. Please clarify this distinction.
Methods: Enrollment and data collection. I recommend changing the wording "patients were eligible for enrollment in the study if…" and "we sequentially enrolled up to 200 patients per site" as this is a retrospective chart review.

Results:
I suggest changing the wording "we enrolled a total of 3374 participants" as this was a retrospective chart review, with a waiver of informed consent.

1.
Time to ART Initiation: The first paragraph states that a third of patients in South Africa had 2.
not initiated ART within 6 months, but according to Table 3, this is 21.2%. I would also suggest deleting this sentence, because in Table 4, it indicates that many of these noninitiators actually had missing follow-up data.
I agree with the review by Burke, in which she clearly the concerns about data quality, particularly for Zambia. I agree with comments and suggested approach in this regard. This is the most important critique of the paper.

If applicable, is the statistical analysis and its interpretation appropriate? Partly
Are all the source data underlying the results available to ensure full reproducibility? No source data required

Are the conclusions drawn adequately supported by the results? Yes
Competing Interests: No competing interests were disclosed.

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
Author Response 14 Apr 2023

Sydney Rosen
Thank you for asking me to review this interesting and important paper, which presents routinely collected data on SDI in Malawi, South Africa, and Zambia. A total of 3344 records from 29 health facilities were recorded, which is lower than the target sample size of 6400 due to resource limitations. Overall, 91% of non-pregnant adults in Zambia, 88% in Malawi, and 57% in South Africa initiated SDI. Predictors of not initiating SDI were clinical Stage 3 or 4 disease, and TB symptoms.
Response: We thank the reviewer for these comments.
I have read the paper, and also the comments from 3 other reviewers. I agree with all comments by the other reviewers. This manuscript is clear and well written, and is an important contribution to the evidence base around SDI initiation. However, it's difficult to know if differences in how data are captured and recorded could be contributing to the country-level differences which were found.
○ Response: We agree that it is hard to distinguish true differences from data discrepancies. Like virtually every other observational analysis of routine records in this region, our study almost certainly includes a mix of both. We believe, however, that it is still worth reporting the observed findings and discussing this problem in the manuscript. We have noted this in the limitations of the study.
I have included some additional details below, with the attempt at avoiding comments which have already been stated by the other reviewers: Abstract: Methods: I suggest substituting another phrase for "sequentially enrolled patients eligible to start ART" because this is a retrospective record review.
○ Response: We have revised this phrase to "retrospectively identified." This is the language suggested in https://journals.plos.org/plosmedicine/article/info:doi/10.1371/journal.pmed.0040297, the instructions for reporting cohort studies. Please state the eligibility criteria to start ART. ○ Response: Thank you for this suggestion. There is a limit of 300 words for the abstract, and we cannot add this information without exceeding that limit or replacing something that may be more important. We have therefore not included the eligibility criteria to start ART in the abstract but have confirmed that this information is included in the manuscript text with the description of the study sites.

Results:
Please describe the different study periods from South Africa, Malawi and Zambia. As it is currently written, it appears that this is 18 months in each country, but in the Methods section it clarifies that this is 12 months in each country, with different start and stopping periods. If possible, it would also be helpful to include the actual dates of inclusion, since it appears that these periods were shorter than 12 months (i.e. ○ started at the last date in the 12-month period, and sequentially moved backward by day. Response: We have added the actual dates of inclusion for each country. We assume that the reviewer refers to the period from January 2018 and June 2019 mentioned in the abstract as the 18 month study period. This is correct as the overall interval of ART initiation for participants in the study. The details for each country are provided in the text.
Please add numbers as well as percentages for those who were offered and accepted SDI. Do you also have the number of patients who were offered SDI and didn't accept it?
○ Response: We have added numbers where there were previously only percentages. We have changed "offered and accepted" to "received," as we do not know for certain that SDI was offered to those who did not start on the same day. It may have been offered but not accepted, or it may not have been offered. For clarity, please give the numbers of patients who had not started SDI, but started within 7, 14, 30, or 180 days by country. ) we would therefore prefer not to report p-values for this work. Introduction: The last sentence states that this paper documents "update of SDI, time to ART initiation for those not accepting SDI, and predictors of SDI …." However, it appears that it is not possible to distinguish those who were not offered SDI from those who did not accept SDI. Please clarify this distinction.

○
Response: Thank you for this comment. We agree, we cannot make this distinction. We have thus changed the sentence from "accepting" to "receiving" SDI.
Methods: Enrollment and data collection. I recommend changing the wording "patients were eligible for enrollment in the study if…" and "we sequentially enrolled up to 200 patients per site" as this is a retrospective chart review. Results: I suggest changing the wording "we enrolled a total of 3374 participants" as this was a retrospective chart review, with a waiver of informed consent.
○ Response: We have revised this sentence. Time to ART Initiation: The first paragraph states that a third of patients in South Africa had not initiated ART within 6 months, but according to Table 3, this is 21.2%. I would also suggest deleting this sentence, because in Table 4, it indicates that many of these non-initiators actually had missing follow-up data. Response: Thank you for catching the error in the first paragraph-it should have read "a fifth," not a third. We have adjusted the language in the sentence to note that the result is based on records found. I agree with the review by Burke, in which she clearly the concerns about data quality, particularly for Zambia. I agree with comments and suggested approach in this regard. This is the most important critique of the paper. Beyond that, we believe it is a judgment call, as some evidence supports and other evidence argues against this approach. Our statement "many patients can and should start ART immediately" appears to be correct for the setting in which our study took place, based on the reference cited. We did not say, and certainly do not believe, that all patients with TB symptoms should start ART immediately-only that many can and should, particularly in light of the high loss to follow up experienced between HIV testing and ART initiation at many of our sites. We would therefore like to keep this sentence in the manuscript. We have added the citation suggested, so that readers may pursue it if they wish.
participants from all clinics? Or were some clinics under-represented? I see the discussion says results limitations, but it might be good if this were a little more detailed.
I agree with the other reviewers that it might be helpful to describe the data extraction process in a bit more detail by country and possibly by clinic. In what clinics were EMR data used vs. paper records? What happened if EMR data and paper data disagreed? (For example, same day ART recorded on one but not the other)?
What was the process of ascertaining HIV testing or first contact with systems? You say in the introduction often people are only entered into database at the point of starting ART (rather than point of first contact with system)? How were you able to get "one step" back from what is routinely reported?
There is presumably an issue with under-ascertaining people who tested positive but didn't ever start ART was a problem in Zambia (as the authors have said they don't think their estimate of % who started ART is accurate), describing the data sources might help with understanding this.
Was there a process of reconciling paper HIV testing registers with ART registers? How were similar-but-not-identical names or ages etc. dealt with? Do you think data collection processes at different clinics was similar?
This information might be useful, particularly for understanding the Zambia context and how data on people who had a positive HIV test but didn't initate might be lost.
I do understand that this is all observational data from routine records, and some of these items might be very difficult to ascertain or not available at all. That is very understandable and doesn't invalidate the research. Nonetheless, I think it would be good to describe a little further about the process of gathering/extracting data.

2.
I agree with other reviewers that it might be helpful to see data broken down by clinic to see how much variation there is by clinic.

3.
In table 5 from the model predicting SDI, is this comparing SDI vs. initiated ART (but not on the same day)? Or is it SDI vs. initiated not same day and not initiated at all? From the description of variables being collected at the point of ART initiation, it sounds like it is the former but it would be good to clarify.

4.
In general, I agree with the comments from the other reviewers.
Minor comment I am not familiar with Zambia and South Africa ART guidelines, but the authors are correct in saying that in Malawi 2022 guidelines the timing of ART in people with diagnosed TB (or TB symptoms but not diagnosed TB) is not specified in detail.

Is the work clearly and accurately presented and does it cite the current literature?
Yes

If applicable, is the statistical analysis and its interpretation appropriate? Partly
Are all the source data underlying the results available to ensure full reproducibility?
No source data required

Are the conclusions drawn adequately supported by the results? Yes
Competing Interests: No competing interests were disclosed.

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
Author Response 14 Apr 2023

Sydney Rosen
Many thanks for asking me to review. This paper presents information gathered from routine data sources on Same Day ART initiation in three Southern African countries. It is well reported and clear. These are important data for understanding programmatic ART implementation. I have read the paper and also the comments from the other two reviewers. I don't have much to add that the other reviewers haven't already mentioned.
Response: We thank the reviewer for these comments.
To echo Dr Lessells, it would be helpful to know a bit more about why the sample size wasn't reached in Malawi and Zambia? Did this introduce bias? Was this a smaller number of participants from all clinics? Or were some clinics under-represented? I see the discussion says results limitations, but it might be good if this were a little more detailed.
○ Response: We have added mention of this at the start of the results section. We do not believe that this limitation introduced bias, as it resulted simply from slower data collection than expected (due in part to COVID-19 restrictions), which reduced the time available for the overall study. We enrolled sequentially from registers, which should produce a representative sample, even if smaller than we originally desired. (Supplemental Table 1, which we added in response to reviewers' comments, indicates enrollment numbers for each study site.) I agree with the other reviewers that it might be helpful to describe the data extraction process in a bit more detail by country and possibly by clinic. In what clinics were EMR data used vs. paper records? What happened if EMR data and paper data disagreed? (For example, same day ART recorded on one but not the other)?
○ Response: We have added further details about the data extraction process to the Methods section. We have clarified that we did not use both EMR and paper record data for any single patient, only one or the other. After reviewing the overall state of the EMR data for Malawi and Zambia, we decided to use manual data extraction from paper files only in these two countries. In South Africa, as explained above, we relied on EMR data only where paper files could not be found (lost or misfiled). What was the process of ascertaining HIV testing or first contact with systems? You say in the introduction often people are only entered into database at the point of starting ART (rather than point of first contact with system)? How were you able to get "one step" back from what is routinely reported?
There is presumably an issue with under-ascertaining people who tested positive but didn't ever start ART was a problem in Zambia (as the authors have said they don't think their estimate of % who started ART is accurate), describing the data sources might help with understanding this.
Was there a process of reconciling paper HIV testing registers with ART registers? How were similar-but-not-identical names or ages etc. dealt with? Do you think data collection processes at different clinics was similar?
This information might be useful, particularly for understanding the Zambia context and how data on people who had a positive HIV test but didn't imitate might be lost.
I do understand that this is all observational data from routine records, and some of these items might be very difficult to ascertain or not available at all. That is very understandable and doesn't invalidate the research. Nonetheless, I think it would be good to describe a little further about the process of gathering/extracting data. ○ Response: We thank the reviewer for understanding the challenges in using routine records. These are all really good questions. We don't think it's really possible to get "one step back" from routinely reported data in a retrospective, observational study. That said, as explained above, by starting with HIV testing, we went farther back than do most other studies of ART uptake. We believe that the HIV testing registers are relatively accurate and complete, though probably not perfect. In the end we did not enroll anyone who did not have a positive HIV test recorded in the HIV testing registers, which meant that all participants were enrolled as of the date of their HIV test. (We note that this might not have been a participant's first positive HIV test ever, but it was the first at the study facility during the study period.) Matching HIV register patients to those in the ART registers was not a major concern, as the time period of interest for the ART registers was limited. Most patients initiated ART on the date of their HIV test, which allowed easy matching by date. For those who did not, we searched the ART registers and EMR records and matched by name and date of birth. We believe that for the sample enrolled this was likely an accurate reflection.
The study had an established protocol, SOPs, and data collection forms that were used by all three countries and at all study sites. Where we had concerns about inconsistency, data collectors re-visited sites and re-checked registers and data. We do not know what happened in Zambia, or even if our speculation that some of those who tested positive and should have been enrolled were not captured. We just think it likely in view of the very high proportion of same-day initiation observed.
In sum, we believe that the study has done everything possible to collect a complete data set for a representative sample of those eligible for same-day initiation. It is not perfect but we think it is better than has been done previously.
I agree with other reviewers that it might be helpful to see data broken down by clinic to see how much variation there is by clinic.

Response:
We have now reported results by facility in Supplemental Table 1. We h have also mentioned site-level variations in the Results section. In table 5 from the model predicting SDI, is this comparing SDI vs. initiated ART (but not on the same day)? Or is it SDI vs. initiated not same day and not initiated at all? From the description of variables being collected at the point of ART initiation, it sounds like it is the former but it would be good to clarify. In general, I agree with the comments from the other reviewers.
○ Response: The model compares SDI vs initiated not same day and not initiated at all (the latter option above). We created outcome variables of SDI = 1, anything else =0. This makes sense to us since the purpose of our analysis is identifying predictors of same-day initiation specifically, and the outcomes of those who did not initiate at all in our data set are unknown. As mentioned in the manuscript (now), some may have initiated at another facility, within 6 months but not same-day from our study site's perspective.
Minor comment I am not familiar with Zambia and South Africa ART guidelines, but the authors are correct in saying that in Malawi 2022 guidelines the timing of ART in people with diagnosed TB (or TB symptoms but not diagnosed TB) is not specified in detail.

Richard Lessells
KwaZulu-Natal Research Innovation & Sequencing Platform, University of KwaZulu-Natal, Durban, KwaZulu-Natal, South Africa This paper reports the main quantitative results from the SPRINT study, an observational study using routinely collected health data in Malawi, South Africa and Zambia. This study evaluated same-day antiretroviral therapy initiation for people with HIV in routine care settings. A total of 3344 records from 29 health facilities were analyzed (lower than target sample size of 6400 records due to resource limitations). Overall, the proportion of non-pregnant adults receiving same-day ART initiation was 91% in Zambia, 88% in Malawi and 57% in South Africa; and the proportion initiating ART within 6 months was 100% in Zambia, 92% in Malawi and 79% in South Africa. Individuals with clinical stage 3 or 4 disease and individuals with TB symptoms were less likely to receive same-day ART.
This is an important contribution to the evidence base around same-day ART initiation, and adds to our understanding of how this is being implemented in HIV programmes in Africa. Despite not meeting the target sample size, the study still provides useful information that will be of interest to policy makers, programme managers and health care workers involved in HIV care delivery. Important differences between the three countries were found, with far fewer people receiving same-day ART at the South African sites. However, it's difficult to know the extent to which this reflects true differences in practice, or whether differences in how data are captured/recorded could also contribute to these findings.
The manuscript is clear, concise and well-written. I have a few comments that the authors might want to consider, mainly highlighting areas where I think additional information or clarity would be helpful for the reader.
Major comments: As this is an observational study using routinely collected health data, I would recommend that reporting is in line with the RECORD statement and that the completed checklist is added as a supplementary file (https://www.equator-network.org/reportingguidelines/record/).

1.
In the study registrations on clinicaltrials.gov, there are two primary outcome measures listed -Average number of visits required to start ART and Time to ART initiation. In this article, only Time to ART initiation has been reported. Please clarify if the protocol and primary outcomes were amended at some stage, or explain why the average number of visits required to start ART has not been reported.

2.
For me, it's critical to see more details of the data sources (e.g. electronic, register, files), and particularly what variables were used at each site/in each country to calculate the time to ART initiation. This seems key to understanding whether differences in data sources and data collection procedures might underlie any of the difference between sites/countries. Essentially it would be helpful to know how reliably the date of HIV testing or date of first engagement in care is captured, so that we know how confident we can be that if there is a true delay until ART initiation this will be reliably ascertained from the data source.

3.
Minor comments: recommend that reporting is in line with the RECORD statement and that the completed checklist is added as a supplementary file (https://www.equatornetwork.org/reporting-guidelines/record/). Response: The RECORD checklist has been completed and will be posted to the Open BU repository (www.open.bu.edu), as this journal does not accept supplementary files. (Editor: when the final page numbering is available, we will update the checklist and post it to the repository.) In the study registrations on clinicaltrials.gov, there are two primary outcome measures listed -Average number of visits required to start ART and Time to ART initiation. In this article, only Time to ART initiation has been reported. Please clarify if the protocol and primary outcomes were amended at some stage, or explain why the average number of visits required to start ART has not been reported.
○ Response: For this manuscript, we originally chose to report only one of the two primary outcomes in order to focus on the uptake of same-day initiation, with the intention to report visit numbers elsewhere. In response to the reviewer's request, however, we now report visit numbers for those not initiating on the same day, as a final row in Table 3. The median for clients not initiating on the same day was two visits, including the visit at which ART was initiated.
For me, it's critical to see more details of the data sources (e.g. electronic, register, files), and particularly what variables were used at each site/in each country to calculate the time to ART initiation. This seems key to understanding whether differences in data sources and data collection procedures might underlie any of the difference between sites/countries. Essentially it would be helpful to know how reliably the date of HIV testing or date of first engagement in care is captured, so that we know how confident we can be that if there is a true delay until ART initiation this will be reliably ascertained from the data source.
○ Response: Dates of HIV testing were collected from clinic registers and records, as mentioned. We cannot say how reliable these data are, but in our experience, dates of discrete events like HIV tests are usually accurate. We did not mean to imply, however, that these dates necessarily reflect a participant's first HIV test, or even first positive test, only the test associated with starting ART now. We have clarified this in the Methods section.
Minor comments: Abstract, Methods: 'We sequentially enrolled patients…' sounds like this was a prospective cohort study where participants were enrolled. I would suggest rewording this to make clear that this was retrospective analysis of prospectively collected routine health data.

Peter MacPherson
University of Glasgow, Glasgow, UK Many thanks for asking me to review this manuscript. Great to see these data, and very clearly reported. I have only a few comments that the authors should address.
In the introduction, it might be worth adding a sentence or two explaining some of the reasons why same day ART initiation may not always be possible or recommended.

1.
Page 3, in the paragraph describing country guidelines for SDI. Did any country guidelines have recommendations for ART initiation in people with TB symptoms who were not able to produce sputum samples, or did not receive results? This is likely a common reason for ART initiation delay.

2.
Methods: good to provide assurance that, in each of the study clinics, people were systematically recorded in registers/records at the point of eligibility assessment. i.e. is it possible that some people who attended the clinic, but felt to be not eligible for ART initiation by staff, did not have their attendance recorded? If this was a possibility, did practice vary by clinic/country, and what biasing effect may it have had on estimates of same-day initiation? 3.
Similarly, in ascertaining outcomes (ART initiation as defined by dispensing of an initial supply of medication), did recording practices differ between sites/countries, and what methods were put in place to ensure ascertainment bias was minimised? 4.
In Table 1, would it be possible to add a row to provide the number of clinics per country from which electronic data records were available for extraction?

5.
Table 2: interesting to see that the prevalence of reported TB symptoms was lower in the 6.
In the introduction, it might be worth adding a sentence or two explaining some of the reasons why same day ART initiation may not always be possible or recommended. ○ Response: Thank you for this suggestion. We have added a sentence about this to the introduction.
Page 3, in the paragraph describing country guidelines for SDI. Did any country guidelines have recommendations for ART initiation in people with TB symptoms who were not able to produce sputum samples, or did not receive results? This is likely a common reason for ART initiation delay. ○ Response: This is a good point. We agree that concerns about undiagnosed TB have been a major reason for delaying ART initiation. South Africa's guidelines explicitly exclude patients with a positive TB symptom screen from the recommendation for same-day initiation. Malawi's guidelines do require TB symptom screening but do not explicitly recommend delaying ART initiation on the basis for the screening results. Zambia's guidelines recommend deferring initiation if there is suspicion of TB but do not specify what should be considered for "suspicion." We have added this information to the Methods section where the guidelines are described.
Methods: good to provide assurance that, in each of the study clinics, people were systematically recorded in registers/records at the point of eligibility assessment. i.e. is it possible that some people who attended the clinic, but felt to be not eligible for ART initiation by staff, did not have their attendance recorded? If this was a possibility, did practice vary by clinic/country, and what biasing effect may it have had on estimates of same-day initiation ○ Response: Another good point. It is true that recording of patients not eligible for ART initiation is poor in many clinics in the study countries. This is the reason for enrolling SPRINT participants at the point of testing positive for HIV, which is the very first step in the ART eligibility determination process. Even patients who arrive with documentation of a positive HIV test from somewhere else are re-tested so that the facility has confirmation of their status. SPRINT enrolled on the basis of HIV testing registers and then followed patients forward in time. We did not enroll anyone who was not tested at the study clinic, though there was no requirement that the test used for study enrollment was the individual's first positive HIV test-a prior test at any time in the past was certainly possible. While we cannot be certain that every individual who tested positive was included in the registers, we believe that the study captured as many of them as is possible in a retrospective analysis. We have clarified the wording about this in the methods section and, since missing some positive testers is a possibility, we added a note about this to the study limitations section. Similarly, in ascertaining outcomes (ART initiation as defined by dispensing of an initial supply of medication), did recording practices differ between sites/countries, and what methods were put in place to ensure ascertainment bias was minimized? ○ Response: Ascertainment bias is always a possibility when relying on routinely collected data, as the reviewer suggests. Prescribing medications is usually one of the best-recorded indicators in routinely kept medical records, however, so we are not too concerned about bias in this regard. A more serious concern arising from our reliance on routinely collected records within unlinked EMRs is that patients who appeared not to have initiated ART at all during the six-month follow up period (last row of Table 4) could actually have transferred informally ("silent transfers") to other healthcare facilities. The number of such participants is very low in Zambia and Malawi but this may explain the larger number of missing outcomes in South Africa, where mobility is high. We have added this explanation to the results section and the limitations section in the discussion. In Table 1, would it be possible to add a row to provide the number of clinics per country from which electronic data records were available for extraction? ○ Response: Since Table 1 describes the study sites, we assume that this question pertains to the study sites and not the study countries as a whole. All study sites had electronic data records available for extraction, but none of the study sites reliably entered into their EMRs patients who did not start ART. We therefore relied entirely on paper record data extraction in Zambia and Malawi, where the EMRs were of greater concern. In South Africa, we used EMR data to supplement paper records for all 5 clinics, as we were not able to find all paper files (EMR used for 92 of 534 participants, 17%). In these cases, the paper records were likely misfiled or misplaced after original EMR data entry by the clinic. We have added these details to the manuscript. Table 2: interesting to see that the prevalence of reported TB symptoms was lower in the South African clinics compared to the Zambian clinics. Were there any differences in definitions of "TB symptoms" or recording between countries? ○ Response: We agree that this was a bit surprising, given what we think we know about the prevalence of TB in the two countries, but we have no way of answering the reviewer's questions. Both countries use the standard WHO four-symptom screening tool. It is possible that Zambian clinic staff are more diligent about recording results, but we have no evidence of this. Since we don't know, and cannot even really guess, we'd prefer not to speculate in the manuscript. Table 4 -do you have data on the timing of death? ○ Response: Unfortunately, date of death is rarely recorded in routine ART clinic records, so we do not.
Great to see that clinic-level random intercepts were included in the modified Poisson regression models, if I am understanding correctly. Are you able to quantify the variation in percentage achieving same-day initiation between clinics (e.g. standard deviation of the random intercepts), and any particular patterns that might help explain why clinics had higher or lower rates of same-day initiation (e.g. urban vs. rural clinics, size of clinic)? Investigating this may help untangle some of the potential reasons for delay rightly highlighted by the authors as a remaining knowledge gap in the Discussion.
○ Response: We thank the reviewer for this suggestion. We only clustered by clinic, so we are unable to quantify the variation in percentage achieving SDI between clinics. We felt it would be informative to stratify uptake of SDI by country, site location (urban/peri-urban and rural) and tuberculosis symptoms present and to rank them by size of clinic based on the total volume of patients on ART at each site. We have added Supplemental Table 1 below with the facility-level results for these variables to the manuscript. The following text into the results section in reference to any variation by site, "When we stratified results by country, clinic, and TB symptom status and ranked sites according to ART patient volume (Supplemental Table 1), our results suggested that patients presenting with TB symptoms in South Africa were less likely to initiate ART on the