Incidence and determinants of mortality among adult HIV infected patients on second-line antiretroviral treatment in Amhara region, Ethiopia: a retrospective follow up study

Introduction Mortality of adult patients who are on antiretroviral therapy (ART) is higher in low-income than in high-income countries. After the failure of standard first-line treatment, patients switch to second-line regimens. However, there are limited data about the outcome of patients after switching to a second-line regimen in the study area. This study aimed to measure the rate of mortality and its determinants among HIV patients on second-line ART regimens. Methods Multicenter institution based retrospective follow up study was conducted among 1192 adult patients who started second-line ART between 2008 and 2016 in eight selected hospitals of Amhara region. Patients who started second-line treatment after the failure of first-line treatment were included. Patient medical records, registration books, and computer database were used to collect the data. Time to death after a switch to second-line ART was the primary outcome of interest. Cox proportional hazard model was fitted to identify determinant factors of mortality. Results Among 1192 patients who were on second-line ART, 136 (11.4%) died with 3,157 person-years of follow up. Over the study period, the mortality rate was 4.33 per 100 person-years. Not taking isoniazid preventive therapy (IPT) (Adjusted Hazard Ratio (AHR): 6.6; 95% CI: 2.9, 15.0), did not make modification on second-line regimen (AHR: 4.4; 95% CI: 2.8, 6.8), poor clinical adherence (AHR: 2.5; 95% CI: 1.4, 4.5), functional status of bedridden (AHR: 2.7; 95% CI: 1.5, 4.8), and having attained a tertiary level of education (AHR: 0.4; 95% CI: 0.2, 0.8) were independent determinants of mortality. Conclusion The incidence rate of mortality was high and most of the deaths occurred within 12 months after switching to second-line ART. Higher mortality among adult HIV-infected patients was associated with poor adherence, no formal education, not taking IPT, being bedridden at the time of the switch, and not modifying second-line treatment. Improving treatment adherence of patients by providing consistent adherence counseling, providing INH prophylaxis and monitoring patient's regimen more closely during the first twelve months after switch could decrease mortality of HIV patients on a second-line regimen.


Introduction
Acquired Immune Deficiency Syndrome (AIDS), which is caused by the Human Immunodeficiency Virus (HIV), has been the major public health problem worldwide [1]. The majority of HIV infected individuals live in sub-Saharan Africa (SSA) [2]. In these countries, AIDS-related morbidity and mortality remain the highest in the world because of limited access to HIV diagnosis and treatment. According to the 2012 UNAIDS report, around 1.7 million people died from AIDSrelated causes worldwide, and 70% occurred in SSA [3]. Ethiopia is one of the hardest hit sub-Saharan African countries by the HIV pandemic with an estimated death of 52,405 by 2014 [4]. Since the introduction of Zidovudine (ZDV) in 1987 as a first antiretroviral drug, there have been significant advancements in the antiretroviral treatment [5]. Currently, there are six classes of antiretroviral drugs such as Nucleoside reverse transcriptase inhibitors (NRTIs), Nonnucleoside reverse transcriptase inhibitors (NNRTIs), Protease

inhibitors (PIs), Integrase inhibitors (IIs), Fusion inhibitors (FIs), and
Chemokine receptor antagonists (CRAs) [5]. For a new patient who is going to start ART, a combination of three drugs is provided. The drug combination contains two NRTIs and one NNRTIs [6]. After highly active antiretroviral treatment (HAART) introduced in 1995-96, HIV infection turns from inevitable fatal condition into chronic manageable disease [7]. In recent years, efforts have been made to expand access to ART in low-income countries, and it is showing encouraging results [8][9][10][11]. The benefit of ART in restoring immune function and reducing HIV related morbidity and mortality is lost when a patient develops treatment failure [12,13]. This happens when a patient has poor adherence, drug resistance, high baseline plasma viral load, and low baseline CD4 count [14][15][16][17][18][19]. Following a failure of the first-line regimen, patients switch to a second-line regimen containing two new NRTIs and one PI [6,20,21]. Since third-line regimens are costly and not readily available in resource-limited countries, second-line regimens are often the last therapeutic option available for patients in these settings [22]. Not all patients who initiate antiretroviral therapy respond well. Some patients may not respond due to poor adherence, suboptimal antiretroviral treatment potency, and genetic mutation of HIV strains [23]; this could be followed by death. Since living longer with HIV is one of the global strategies, provision of ART by itself is not enough to control the problems of HIV treatment; rather its therapeutic effect on improving survival needs thorough monitoring and evaluation through scientific research. Though some studies based in Ethiopia investigate the outcomes of first-line HIV treatment; little has been done regarding second-line treatment. Therefore this study aimed to measure the incidence and determinants of mortality after a switch to second-line ART regimens.

Methods
Study design and setting: a multi-center institution-based retrospective follow-up study was conducted in eight selected governmental hospital found in Amhara regional state, Ethiopia. Five  (LTFU) was defined as a patient who did not receive ART refills for a period of three months or longer and is not yet classified as "dead" or "transferred-out". Patients were considered censored if they transferred out, lost to follow up or remain on follow up at the end of the study. The clinical adherence of participants was assessed and it was defined as a regular attendance of patients according to a given appointment. If the patients were coming regularly by the given appointment dates in >85% of the times, they were considered as having good clinical adherence, and <85% attendance as poor clinical adherence [25,26]. Schoenfeld residuals test (both global and scaled) and graphical methods were used to check Cox proportional hazards assumption.
The P-value for the global test was 0.1 (not significant), which indicates the model was fit. Cox proportional hazards model was used to identify determinants of mortality. Variables having a p-value of 0.2 or less in the bi-variable analysis were fitted into the multivariable model. Ninety-five percent confidence interval of hazard ratio (HR) was computed and variables having a p-value less than 0.05 in the multivariable Cox proportional hazards model were considered as statistically significant.
Ethics consideration: ethical clearance was obtained from the institutional review board (IRB) of the University of Gondar. A letter of support and a permission letter were obtained from the Amhara Regional State Health Bureau and the hospital's administration respectively. All patient records were de-identified and kept locked.
Since we were reviewed patient records, informed consent was waived. 95%CI: (0.2, 0.9)) ( Table 3). The second-line regimen modification, tertiary education, functional status of working, good adherence and using INH prophylaxis had protective effects on mortality of patients.

Results
The risk of death was 4.8 times higher for those patients with secondline regimen was not modified as compared to those with second-line regimen modified. The rate of mortality was 3 times higher among those who were bedridden at the switch to a second-line regimen compared with those who were working. The risk of death was 2.6 times higher for patients who had poor clinical adherence compared with those with good clinical adherence. Patients who had tertiary education were 60% less likely to die than those patients who had no formal education. Patients who did not take INH were 6.9 times at higher risk of death compared with those who did take INH.

Discussion
This study aimed to measure mortality and its determinants among HIV infected adult patients on second-line antiretroviral therapy. The incidence of mortality was 4.33 per 100 person-years; which is consistent with studies done in India, in resource-limited countries and in sub-Saharan African countries [27][28][29]. However, it is higher than studies done in Zambia [30] and Switzerland [31]. The higher rate of mortality in this study might be due to the fact that most patients had advanced disease at baseline (CD4 count ≤200 cells/mm 3 ); that could lead them to have bad outcomes [29]. The delay to switch to second-line regimens, that usually happens in resource-limited settings, might explain the higher mortality [32]. not. This is similar to a study done in resource-limited countries [28].
Patients who had modified their regimen might get better follow up and get their problems managed early. In addition, since deferring a drug modification while it is required might have a negative implication of accumulation of drug resistance [33], this group of patients might be benefited by the regiment modification which could help them to overcome an emerging drug resistance that could be a potential cause for lower immunity and death. Having good clinical adherence was protective of mortality. This evidence is supported by previously done researches [34,35]. Having good clinical adherence, patients wouldn't discontinue their drugs, have close follow up and counseling form the health care providers, and they might be also psychologically stable to take their treatment. Due to these reasons, bad outcomes are less likely to happen among these patients. Taking INH had a protective effect on mortality of patients. This is similar to a study done in resource-limited countries, and Ethiopia [28,35]. As already known, TB is the commonest opportunistic infection (OI) among HIV patients and it is a potential cause of death. If an HIV patient takes INH prophylaxis, it would prevent the morbidity and mortality due to TB. Patients who were bedridden at switch were at higher risk of death. This is consistent with a finding from eastern Ethiopia [36]. Functional status of patients has a correlation with their clinical and immunological status. Usually, patients in an advanced stage of the disease are bedridden and the chance of death is higher among those people [34]. Patients who were illiterate had a high risk of mortality compared to those with tertiary education. This is similar to a study done in Addis Ababa, Somalia and Tigray regions Ethiopia [35,37,38]. This could be due to the reason that educated people follow their treatment appropriately and may have better health-seeking behavior for different OIs and other related diseases.
The main strength of this study is its representativeness in terms of covering multiple treatment centers, and the length of follow up time which was long enough to estimate mortality. However, it has its own limitations. Considering all death as HIV related could overestimate the incidence of mortality. On the contrary, the lost to follow up patients might include individuals who died at home without being reported and that could underestimate the incidence of mortality. In addition, residual confounding could present by the absence of viral load record, under-reporting of clinical conditions, and missing laboratory results. The data were collected in a retrospective fashion using secondary sources, with resulting incompleteness.

Conclusion
The incidence rate of mortality was high and most of the deaths

Competing interests
The authors declare no competing interests.

Authors' contributions
Adino Tesfahun Tsegaye and Wagaye Alemu conceptualized and designed the study and performed the data analysis and wrote the draft and final manuscripts. Tadesse Awoke Ayele provided technical research guidance. All authors designed the data collection tool and wrote and approved the final manuscript.

Acknowledgements
We would like to thank the hospitals' administrative bodies, data clerks and card room workers for their cooperation and permission to conduct the study. We are grateful to the University of Gondar for financial support as well as the data collectors who participated in this study for their commitment.       1.1(0.6-2.1) 0.9(0.5-1.9) Event=death-censored=transfer out +lost to follow up+ alive (on treatment at the end of the study) * Significant factors