Starting antiretroviral therapy within seven days of a positive HIV test increased the risk of loss to follow up in a primary healthcare clinic: a retrospective cohort study in Masaka, Uganda

Background Retention of patients initiated on antiretroviral therapy (ART) and good adherence remain cornerstones to long-term viral suppression. In this era of test and treat (T&T), ensuring that patients initiated on ART remain connected to HIV clinics will be key to the achievement of the UNAIDS 90-90-90 targets. Currently, limited studies have evaluated the effect instant ART initiation has on loss to follow up in a typical service healthcare setting. We studied the cumulative incidence, incidence rate of loss to follow up (LTFU), and factors associated with loss to follow up (LTFU) in a primary healthcare clinic that has practiced test and treat since 2012. Methods We retrospectively drew routine program data of patients initiated on ART from January 2012 to December 2016. We defined LTFU as failure of a patient to return to the HIV clinic for at least 90 days from the date of their last appointment. We calculated cumulative incidence, incidence rate and fitted a multivariable Cox proportion hazards regression model to determine factors associated with LTFU. Results Of the 8,136 patients included in our sample, 3,606 (44.3%) started ART within seven days of HIV diagnosis. Females were 62.3%, median (interquartile range) age at start of ART was 30 (25, 37) years, 50.1% had access to a mobile phone, 54.0% had a baseline CD4 cell count of <350 cells/ml, 14.8% were in either WHO stage 3 or 4 at baseline and 75.9% had a normal body mass index (BMI). There were 1,207 cases of LTFU observed over 15953.0 person years at risk. The overall incidence rate (IR) of LTFU was 7.6 (95% CI=7.2-8.0) per 100 person years of observation (pyo). Cumulative incidence of LTFU increased with duration of follow up from 8.8% (95% CI=8.2-9.4%) and 12.0% (95% CI=11.2-12.7%) at 6 and 12 months, to 17.9% (95% CI=16.9-18.9%) and 20.1% (95% CI=18.9-21.3%) at 36, and 48 months respectively. Predictors of elevated risk of LTFU were; starting ART within 7 days of a positive diagnosis ((aHR) =1.39, 95% CI, 1.13-1.71), lack of access to a telephone set (aHR=1.60, 95% CI, 1.29-1.99) and baseline WHO clinical stage 3 or 4 (aHR =1.53, 95% CI, 1.11-2.11). Factors associated with a reduced risk of LTFU were; baseline age ≥25years, and having a BMI ≥ 30 (aHR =0.28, 95% CI, 0.15-0.51). Conclusion Initiation of ART within 7 days of an HIV diagnosis was associated with an elevated risk of loss to follow up. Steep ART initiation needs to be backed by enhanced adherence and retention counseling to reach the 2020 UNAIDS goals and beyond.

timeliness in ART initiation, ensuring continuous engagement of patients with the health care system 66 for periodic drug refills and running monitoring tests are critical to the success of this rapid ART scale-67 up. In spite of all this, loss to follow up of patients after ART initiation remains a great challenge. 68 Systematic reviews of studies on the rapidly expanding ART programs in SSA illustrated that about 69 60-65% of patients were retained in HIV care at 2 to 3 years after starting ART [15,16]. In settings 70 where patients start ART instantly after a positive HIV test, there is a possibility of offsetting the 71 benefits associated with the immediate initiation when patients do not return to the HIV clinics. 72 Patients need to perceive the clinical benefits of treatment continuation without interruptions. 73 Previously, late presentation was common and HIV care and treatment guidelines stipulated that, only 74 those presenting with WHO clinical stage III or IV of HIV/AIDS or had their CD4s decline to certain 75 levels qualified for ART initiation. However, majority of patients currently diagnosed with HIV 76 present in the early stages (WHO stage 1 or 2 or with CD4 cell count >500cell/ml) and initiate ART 77 right away or shortly thereafter. Experiences from prevention of mother to child transmission 78 (PMTCT) of HIV programs where instant ART initiation has widely been practiced indicate sub-79 optimal levels of ART adherence and retention of mothers in care [17 ,18]. LTFU is associated with 80 drug resistance, and comparatively poor long-term treatment outcomes, including mortality [19]. In 81 resource constrained SSA countries, enhanced ART initiation will benefit from data characterizing 82 retention of patients in typical clinical settings. To date however, a few studies have explored loss to 83 follow up and associated factors in a typical HIV clinic practicing test and treat. Most of the data 84 currently available are derived from implementation of test-and-treat in research settings. In this study, 85 we set out to study the cumulative incidence and incidence rate of loss to follow up, and factors 86 associated with loss to follow up in a primary healthcare clinic that has practiced test and treat since  90 This was a retrospective cohort study utilizing data collected on patients who were diagnosed with 91 HIV and enrolled into HIV care from January 2012 to December 2016 at Masaka regional referral 92 hospital, -Uganda Cares' clinic. A patient's ART initiation date defined the beginning of follow up 115 diagnosed with HIV at the outreach sites were referred to this focal desk by trained counselors, to 116 further aid and expedite ART initiation. Although the clinic begun piloting a T&T strategy at the 117 beginning of 2012, it is important to note that this strategy wasn't a true manifestation of T&T as 118 illustrated by the treatment as prevention (TasP) group, but rather a process where the ART initiation 119 process was expedited, with the preparatory counseling phase taking a maximum of one week. Under 120 this T&T strategy, point of care CD4 cell count and TB assessment were done within a week to further 121 determine ART eligibility. However, patients who declined ART instantly or within seven days were 122 initiated on ART at a time of their convenience with a similar array of services as their counterparts 123 who started ART instantly or shortly after.  130 We included all patients aged ≥18 years, tested and initiated on ART from 01 st January 2012 to 31 st 131 December 2016 regardless of whether or not they tested at Masaka regional referral hospital. We 132 excluded patients who transferred in from other HIV clinics because we were not able to confirm their 133 HIV test and ART initiation dates with certainty, and patients with prior ART history (for example 134 those that had ever used PEP since we could not ascertain the period when they were on medication).

135
In addition, patients <18 years were excluded because they are children according to the Ugandan 136 policy, but also because they do not decide on health service delivery options on their own but rather 137 through their parents/care givers or guardians. 139 The primary outcome was loss to follow up defined as failure of the client to show up at the Masaka 140 clinic for at least 90 days from the date of their last scheduled appointment [37,39]  Research Committee. We also sought approval from the management of the HIV clinic, to allow us 171 access to patients' data. Program data routinely collected and entered into an electronic records 172 management system (OpenMRS) was extracted without patients' direct identifying information.      Table 2 indicates the IR of loss to follow up by patients' characteristics. The incidence rate of 214 LTFU was 10.9/100 pyo in the T&T group compared to 5.7/100 pyo in the delayed ART group.

258
In this retrospective observation study of a primary healthcare clinic practicing test and treat, we  Our study had limitations that should be taken into account while interpreting these findings. First, 345 we utilized already collected data used for routine patient management. Such data presents with lots follow up in our study. There is however a dedicated team at the facility that does contact tracing/case 355 navigation for clients who miss clinic appointments, and we think this might minimize on this 356 misclassification. Lastly, the nature of data collected was limited. Some of the many health system 357 (human resource, waiting time, distance) and socio-economic factors (type of work, socio contacts,

358
HIV disclosure) known to affect loss to follow up were not studied. The effect of such under the 359 current study settings remained unknown. We anticipate that, examining the effect of these under a 360 test and treat setting could better inform ART programing and policies towards boosting retention. This is beneficial to HIV service providers to recognize patients that require enhanced support in this 368 era of test and treat.

369
Acknowledgements 370 We extend our sincere gratitude to Mathew Ssemakadde (Data manager at Masaka Uganda Cares 371 clinic) for his tireless efforts getting us the data from OpenMRS. We are indebted to Dr Cecilia 372 Nattembo, for her insightful reviews and corrections. We also wish to thank the management of 373 Masaka Uganda cares clinic for allowing us use their routine program data without hesitation. We 374 are especially grateful to all the patients from whom big volumes of data are collected routinely.