Predictors of Mortality among Adult People Living with HIV and Its Implications for Appointment Spacing Model Approach Care in Southwest Ethiopia.

A retrospective cohort study was conducted among 676 adult PLWHA who enrolled to ART clinic from September 1st, 2012 - August 30, 2016. Multivariable Cox Regression analysis was used to identify predictors of mortality using 95% condence interval (CI) and at P value ≤ 0.05 cut of point. Poor nutritional status at baseline, advanced HIV disease, occurrence of treatment failure, female sex, substance use disorders, lack of social support, immunological failures, clinical failures, and younger age, low level of education and poor physical access to healthcare facility were found to be important predictors of mortality. Intervening, those factors as routine and part of “appointment spacing model care” can improve survival of PLWHA.


Introduction
Human Immunode ciency virus (HIV) is a global epidemic, which affects individuals of all ages, sexes, races and income status (1). Globally, between 1980 and 2015, about 78 million people were infected and half of them have died of the disease (2). According to the 2016 report of Global Burden of Diseases (GBD 2016), worldwide 38.8 million people had HIV infection; 2.1 million people had been newly infected and 1.2 million died due to the virus in 2015 (3). The Africa continent contributes about 11% of the global population (2). Yet, the highest proportions of HIV infection have been recorded in sub-Saharan Africa (SSA) and Southern Africa countries in particular (4).
Ethiopia is among African countries hard hit by the HIV pandemic (5). The overall prevalence of the disease among adults, pregnant women and, women who had multi sexual partners was reported to be 0.9%, 0.4%, and more than 6%, respectively (6). The prevalence among women (1.2%) was twice that of men (0.6%), seven times in urban residence (2.9%) compared to rural settings (0.4%). The burden of HIV varies from region to region, ranging from < 0.1% in Somali Region to 4.8% in Gambela Region (ref?). In 2016, a total of 19,743 people were died of AIDS and about 792,840 children lost either or both of their parents due to AIDS (7,9).
To alleviate the impact of HIV/AIDS, Ethiopia introduced ART in 2003 for the rst time at cost of patients, and two years later ART started to be provided for free (8, 10, and 11). Following the 3 by 5 program, 535,069 PLWHA who had 'ever started' ART and 375,811 PLWHA who did not yet started ART were put on ART in more than 1000 health facilities in 2014/5 (8). Subsequently, report of 2017 revealed that HIV incidence, AIDS related mortality and overall HIV prevalence were fallen by 95%, 73% and 29%, respectively (12).
Ethiopia has also been implementing prevention, treatment and care interventions. The prevention activities include voluntary counseling and testing (VCT), prevention of mother to child transmission (PMTCT), behavioral change communication and community mobilization. The treatment and care theme encompasses ART, laboratory monitoring, provider initiated testing and counseling (PITC), nutritional support, palliative care and home-based care activities (8,13). Despite the efforts devoted to limit the impact of HIV, still numerous gaps and challenges remain unsolved. Among others, timely initiation of ART, retention in care and prevention of HIV related mortality, narrowing signi cant inequities in ART coverage, which varies between children (23%) versus adults (60%), females (54%) versus male (69%), and disparities among regions ranging from 5.6-93% are worth mentioned (9,14).
To help HIV patients fully bene t from ART, attain national goals and use resources most effectively,

Ethiopia uses the HIV continuum care (HCC) framework initially incepted by the Centers for Disease
Control and prevention (CDC) in Atlanta, USA, in 2013 (16). The HIV continuum care is a series of stages from the time a person is diagnosed with HIV through assessment for ART eligibility, retention in care, and immunologic success and virological suppression via treatment adherence towards reduction of mortality and ART care success out comes (17,18,19). Many activities have been attempted to address negative ART care outcomes (10). Nevertheless, there have been challenges at every stage of the HCC. These include late HIV care presentation (LP) (20,21), discontinuation of ART (22,23), poor ART adherence (24,25), immunologic (26,27), clinical (28), treatment (23,28) and virologyical failures (29,30).
Currently, Ethiopia is implementing a new care approach known as "appointment spacing model care (ASM)", which mainly focuses on a few visits of hospitals (two times per year) and comprehensive care at a time. However, there are still massive challenges pertaining to ART program regardless of the efforts being devoted by both governmental and non-governmental organizations (31). Consequently, patients usually experience different unintended health care outcomes such as loss of many from the ART schedules (32), failure to recover their immunity (33) and deaths from the disease (34,35).
The overall picture shows that research is lacking to identify key factors that lead to negative ART care outcomes, including mortality. Furthermore, studies conducted in Africa have revealed that factors affecting ART care are beyond individual level. The factors are related to health care institutions such as lack of ART trained health professionals (31)(32)(33)(34), and lack of quality care (35)(36)(37)(38); community level factors such as stigma (39,40) and traditional healing (41)(42)(43)(44); and program level factors such as political commitment (45,46) and lack of coordinated HIV care activities (47). Nevertheless, comprehensive study has not been carried out in Ethiopia to adequately address the challenges of ART care from the viewpoints of stakeholders. Therefore, the aim of this study was to assess predictors of mortality among adult PLWHA in Jimma Zone Public Hospitals.

Cohort Description
Study Setting and period The study was conducted in the Jimma zone ve public hospitals (Jimma Medical Center, Shenen Gibe, Agaro, Seka and Limmu-Genet hospitals). Jimma zone is one of the 20 administrative zones in Oromia Regional State located in southwest Ethiopia. Jimma Medical Center is the only tertiary hospital in southwest Ethiopia. The other four hospitals are primary hospitals with catchment population of about a million each. At the time of the study, a total of 11,186 adults and 2,683 children were in chronic HIV care at the hospitals. The data was collected from February 15/2018 to March 16/2018 and the study involved patients on chronic HIV care at the hospitals during period of 01 September 2012 to 30 August 2016.

Patient And Public Involvement
A ve-year retrospective open cohort study was conducted to assess the predictors of mortality and the implication of appointment spacing model approach care among PLWHA in Jimma zone public hospitals. This retrospective cohort study has assessed whether the risk of expected outcome was different by the exposure of interest using existing data (48).
Hence, two groups who were known to be either exposed or not exposed to the factor of interest in a speci c time frame were compared on the basis of outcomes of interest. The hazard risk is used as measures of estimation of the relative risk (50).
The research question of this study is in line with the Ethiopian Public Health research plan developed through multiple consultative processes involving broader stakeholders including patient representatives. We plan to disseminate the research ndings through the national HIV, TB and Nutrition research conferences involving researchers, policymakers, stakeholders and affected communities.

Enrolment Procedures
A total of 676 adult PLWHA was randomly selected from All adult (≥ 18 years old) who were on chronic care follow up. The Sample size was estimated using two population proportions formula. Based on BMI measurement, those PLWHA, who had undernourishments at the initiation of ART were considered to be exposed (n 1 ) whereas those who had normal BMI were taken to be non-exposed group (n 2 ). Accordingly, 281exposed and 395 non-exposed samples were estimated, implying total of 676 samples of PLWHA for the study.
The study population was all adult PLWHA enrolled in ART care in the selected hospitals. Patients with unknown nutritional status at ART initiation, those who defaulted, those with incomplete information and those transferred out to other facilities were excluded from the study.
Recruitment and informed consent procedures take place within the Ethiopian HIV Program clinics; any client's clinical records who present to hospital are approached by a Doctors/nurse or another allied healthcare provider and asked whether they are willing to speak to a research assistant about participating in a research project.
If a client is agreeable, a research assistant meets with them to explain the purpose, context and methods for this cohort study, and reviews the informed consent form to determine whether the person is interested, willing and able to participate. The results were also maintained using aggregated report rather than individual characteristics in any aspect of the data throughout all steps and manuscript preparation. These data were not identi able; therefore informed consent was not required. Using appropriate case reporting format prepared in English. The data were collected by 10 BSc professional nurses using data collection format. All the ve hospitals have an electronic patient database called Comprehensive Care Centre Patient Application Database (C-PAD). C-PAD is Electronic Medical Records (EMR) system database that contains patients' both clinical and non-clinical information. This was the main source of data in this study. Data were extracted using a data extraction check list from the database. When data were incomplete, we tried to refer the patients' cards, registration and log books using patient medical record number and ART registration number.

Data processing and analysis
Cleaned and coded data were entered to EPI-data version 3.14, and then exported to SPSS version 20.0 for analysis. Kaplan Meier (KM) survival function and Log rank test was used to test the statistical difference in the KM curves. Bivariate Cox regression analysis was done to estimate the unadjusted Hazard Ratios (HRs). Independent variables with P ≤ 0.25 at bivariate analysis level were entered into the multivariable Cox regression model to control potential confounders. Finally, variables, which have P ≤ 0.05 and non-null values within respective con dence intervals (CIs) at multivariable analysis were considered to be independent predictors of mortality among the PLWHA.

Results:
Socio demographic Characteristic of Adult PLWHA in Jimma zone Public Hospitals In this study, records of 382 (56.5%) females and 294 (43.4%) males were reviewed. The mean age of the study participants at the time of ART initiation was 30.4 years (± SD 7.4); 487 (72%) were between 25 and 44 years of age. More than half (55.6%) of them were married and 132 (19.5%) of them were divorced ones. Most of the participants (79.3%) reported to have attended at least a primary school. About 61% of the participants (412) were urban residents and 128 (18.9%) were living outside the catchment area of the respective hospitals. Nearly two-third (66.3%) of the study participants reported to have had persons taking care of them. Disclosure of their HIV diagnosis to family member was reported by 68.9% of the participants. About 80% (540) and 58.6% of the study participants reported to have used Khat and alcohol respectively (Table 1).   (Table 3). Cumulative Incidence and Incidence rate of mortality Ninety-eight (14.5%) patients were died during the ve-year of ART follow up period. The cumulative incidence among females was 16.8% (64/382). The total follow up time was 28,209 person-months when an overall incidence rate of 11 deaths per 1000 person-months was observed. The death rates among those PLWHA who had severe and moderate malnutrition were 5.7 and 2.9 deaths per 1000 personmonths, respectively. The overall survival probability of the adult PLWHA was 86.6%. The overall estimated mean (± SD) survival time of adult PLWHA on ART follow up was 42.8 (SD = 17.8) months. In the ve-year time interval, the highest mortality was observed during 18 to 24 months of follow up, followed by the months of 30 to 36 and 24 to 30, respectively. The lowest mortality was observed during the follow up period from 48 to 54 months followed by 54-60 months of follow up, followed and during the period 6 to 12 months (Table 4). year treatment follow up period compared to their counterparts. The risk of death was also found to be high among PLWHA living outside catchment area of the hospital (AHR: 3.6; 95% CI: 1.5, 5.4) compared to their counterparts (Table 5). Discussion ART Program has created a signi cant change in improving both health status and life expectance of PLWHA. However, rst-line drug resistance and death from HIV related causes are quiet formidable challenges in Sub-Saharan Africa, including Ethiopia. In this study, 14.9% cumulative incidence of HIV mortality was reported in patient on ART. This is lower than nding of a study conducted in Debre-Markos, west Ethiopia (49). The variation could be due to the fact that in the earlier study, participants were recruited both from health centers and hospitals and the service from health Center is not comprehensive as Hospital.
The study revealed that mortality was found to be higher among (65.3%). This nding is in line with the national EDHS data (2016) in which the prevalence of HIV infection among females was two times higher than males (50). Studies from African countries also support this nding. The prevalence of HIV related death among female and male was reported to be 6.5% and 4.7% in Kenya (51), and 7.5% and 4.3% in Uganda (52), respectively. The fact that rates of HIV infection and death among women are higher imply not only gender difference but also gender inequality, leaving women more vulnerable to its impact, which implies the need for strengthening HIV care and support more among females to curb the challenge.
In this study, the highest number of deaths was noticed in the 2nd and 3rd years of follow up. This could be attributed to eligibility criteria, change of guideline for ART initiation and project driven approach care as well as using test and treat immediately, regardless of CD4 count and WHO clinical staging (53,54). The other reason may be because of Ethiopia doesn't implement strict follow up care (55) for patients with good adherence at 1st year and 2nd the trust in and awareness of modern medicine was poor, and conversely high in traditional medicine, which patients could consider as an alternative option (56,57), and the presents of nutrition by prescription program and accessibility of ART was deprived during the period as compared to today. On the other hand, mortality declined in years 2015 and 2016 as compared with earlier years. This might be because of improvement, accessibility and awareness of ART users and service providers by the Federal Ministry of Health (58). Marital status was signi cantly associated with the survival status of ART users. Those who were not married were more affected by HIV related death compared to married ones. This nding is similar to nding of a study done in developing countries (59,65). This might be due to the fact that those adult PLWHA who were out of marriage were at higher risk of exposure to unhealthy lifestyles, as well as they might have been challenged with stigma and discrimination emanated from the community, which can in turn lead to poor compliance to treatment.  (66)(67)(68)(69). This similarity could be explained because of the same ART care system and similar sociocultural conditions in study population. In addition this similarity could be termed through the document, these PLWHA are defenseless for advanced stage of disease escorted by numerous health problem outcomes, like underprivileged for bad prediction and following mortality as end outcome.

Strengths And Limitations
This study has some limitations. First, it is a one zone public hospitals study which limits its national representativeness. Second, it has been used patient records, which limits the number of variables included into the models. Thirdly, the data we used are secondary data and cannot incorporate other possible predictors of HIV-related mortality such as drug resistance, substance abuse, unspeci ed OIs, House hold food security status, Livelihood status and other possible confounders As strength of this study only focusing on cost and stigma and discrimination doesn't give sense unless, the health system is working on the crucial change of quality health care, life expectancy of clients, quality of life, understanding that all patient didn't have same livelihood in Economical states, cultural, residence, support, ART retention, Access to health care and Information, level of understanding to HIV care, and fear of inadvertent disclosure due to having a large volume of medication and concerns regarding safety and storage of medication for prolonged periods at home.

Conclusions
In conclusion, the incidence of HIV mortality was 11 deaths per 1000 person-months, whereas most deaths (20%) occurred during the 18-24 month follow up period interval. Females are at higher probability of dying from AIDS related causes. Marital status, functional status, education level, disclosure of HIV status, having care givers/support, abnormal test of liver function, presence of opportunistic infection, malnutrition were found to be independent predictors of AIDS related mortality.
These ndings imply that applying an action/intervention towards these leading cases of mortality as routine and as part of the new approach care "appointment spacing model care (ASMC)" can reduce mortality and can be a sustainable clinical care other than project driven intervention of HIV care. Except children less than 14 years old, pregnant women and clients other than good adherence (not stable per World Health Organization criteria) were enrolled in the ASMC approach care in Ethiopia, which minimizes the direct and indirect cost of frequent visit, stigma and discrimination of clients.
But, from this study only focusing on cost and stigma and discrimination doesn't give sense unless, the health system is working on the crucial change of quality health care, life expectancy of clients, quality of life, understanding that all patient didn't have same livelihood in Economical states, cultural, residence, support, ART retention, Access to health care and Information, level of understanding to HIV care, and fear of inadvertent disclosure due to having a large volume of medication and concerns regarding safety and storage of medication for prolonged periods at home.
The models also require ongoing evaluation and further adaptation in order to address gaps, such as reaching clients who are at high risk of loss to follow up. Improving access to health services, including ART, and support for retention in care is enhanced by decongestion of high-volume clinics. This takes place through decentralization of services to, for example, community service outlets, and through task shifting beyond hospital/Health center-initiated management of ART (HIMART) by deploying community cadres such as community health workers, peer educators and PLWHA. We extracted an anonymised data from the record and no participant was taken part in the study. The data and collected information was kept and locked in a ling cabinet with the key only accessible to principal investigator (PI) and the computer les were protected with passwords that only the PI knows. We couldn't use any individual-level data from inception to this manuscript. The results were also maintained using aggregated report rather than individual characteristics in any aspect of the data throughout all steps and manuscript preparation. These data were not identi able; therefore informed consent was not required.

Consent for publication:
Not required Availability of data and materials: all data reviled to the study are included in the article or the data sets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests:
The authors declare that they do not have competing interest.
Author Information