Clinical outcome of admitted HIV/AIDS patients in Ethiopian tertiary care settings: A prospective cohort study

Background Acquired ImmunoDeficiency Syndrome (AIDS) related illnesses are the leading cause of death in the developing world. However; there is limited evidence regarding the incidence of mortality among admitted HIV patients in Ethiopia. Objective To determine the incidence of mortality and its predictors among admitted HIV/AIDS patients in selected tertiary care hospitals in Ethiopia. Methods A prospective cohort study involving 136 admitted HIV/AIDS patients from April 1 to August 31, 2018 was conducted in selected tertiary care hospitals in Ethiopia. Data were collected on socio-demographic, clinical characteristics, and drug related variables. Kaplan-Meier and Cox regression were used to compare survival experience of the patients and identify independent predictors of mortality. Hazard ratio was used as a measure of strength of association and p-value of <0.05 was considered to declare statistical significance. Results Of 136 patients, 80 (58.8%) were females. The overall in-hospital incidence of mortality was 2.83 per 1000 person-years. The incidences of mortality due to AIDS and non-AIDS related admissions were 6.1 [3.95, 8.67] and 5.3 [3.35, 8.23] per 1000 person-years respectively. The mean ± SD survival times among patients with AIDS and non-AIDS related illnesses were 32 ± 3.1 and 34 ± 3.3 days respectively (log rank p = 0.599). Being on non-invasive ventilation (AHR: 2.99, 95%CI; [1.24, 7.28]; p = 0.015) and having baseline body mass index (BMI) of less than 18.5 (AHR: 2.6, 95%CI; [1.03, 6.45]; p = 0.04) were independent predictors of mortality. Conclusion The study found high incidence of in-hospital mortality among admitted HIV/AIDS patients in Ethiopian tertiary care hospitals. Being on non-invasive ventilation and body mass index (BMI) of less than 18.5 were found to be independent predictors of mortality.

and specialized hospital in the southwest region of Ethiopia. The hospital serves as a referral site and provides specialized care for southwest Ethiopia with a catchment population of about 15 million. The available diagnostic tools used for this study in the settings includes; Human immunodeficiency virus (HIV) test was performed using the HIV 1/2 STAT-PAK1 RDT (Chembio Diagnostics, Medford, NY, USA). Complete blood counts were performed using hematologic analyzers: XT-1800i (Sysmex, Japan), KX-21 N™ (Sysmex, Japan), and Cell-Dyn 18001 (Abbot, USA). Mycobacterium tuberculosis diagnosis was made using Cepheid Xpert MTB/RIF 1 ). Test for Hepatitis B surface antigen (HBsAg) was done using the rapid diagnostic tests Determine™ (Alere, Waltham, MA, USA), and Bio-Dignos HBsAg cassette (Nantong Diagnose, China); hepatitis C virus (HCV) antibody (anti-HCV) test was done using the rapid diagnostic test EGENS HCV Test cassette (Jiangsu, China), and SD BIOLINE HCV (Standard Diagnostics, Yongin-si, Republic of Korea). Stroke and heart failure diagnosis was made using computed tomography CT-Scan, S192 from Optima CT/GE Healthcare and Echocardiography (HUMANSCAN Co Ltd) respectively. Weight and height were measured using WS703 from Narang medical limited.

Study population and variables
We included all adult (age>18years) HIV/AIDS patients, admitted to medical, intensive care unit (ICU) and surgical wards for at least 24 hours. Signed informed consent was taken from all patients included in the study. Patients with readmission, incomplete medical records, pregnant and obstetric admissions were excluded. Data were collected on socio-demographic characteristics (sex, age, ethnicity, marital status, area of residence, educational, and occupational status, monthly income, body mass index (BMI)), Diseases Related Factors (reason for admission, time of HIV diagnosis, history of OI, current CD4 count, WHO clinical stage, co-morbidities), Treatment Related Factors (OI prophylaxis (Cotrimoxazole prevention therapy (CPT)), non-invasive ventilation, cART status, duration on cART, type of cART regimen) and clinical outcome (mortality). Monthly income of the patients was measured using Ethiopian birr (ETB) and it's classified based on Ethiopian Civil Service monthly salary scale for civil servants.

Sample size determination and sampling technique
The sample size was calculated using a single population proportion formula, No¼nð1À PÞPZ 2 d 2 where, No = minimal sample size required, z = 1.96 (normal deviate corresponding to 95% confidence interval), d = 0.05 (degree of precision of 5%), p = 0.445 (a mortality rate of 44.5%; from study conducted in Addis Ababa [17]). Thus, No¼0:445�1:96 2 ð1À 0:445Þ 0:05 2 ¼ 379 patients. Since the target population is less than 10,000 using correction formula nf = N o /1+N o /N. Where nf is the corrected sample size and N is obtained from the last 5 months admission of patients in the two hospitals, which was 185. The corrected sample size is nf = 379/1+379/185 = 124. By adding 5% contingency value, the final sample size was 131 patients. Based on the previous admissions, this number is proportionally divided for both hospitals in the ratio of 1:1.2. Accordingly, 71 and 60 patients were allocated for JUMC and TASH, respectively.

Data collection procedure and quality assurance
Clinical data were collected prospectively from active patient follow up chart using English version checklists. The data collection tool was carefully prepared after reviewing relevant literatures to enable the data collectors to gather all the information required to address the study objectives. Pre-test was conducted on 5% of patients. A 2-day training was provided on the data collection tool and the general procedures for data collectors i.e. 3 pharmacists (B.Pharm) and 3 nurses, and 2 medical interns who were acting as supervisor.
Patient follow-up and outcome measurement. The follow-up started at hospital admission and ended if patients were discharged with improvement or developed an outcome. Inhospital incidence of mortality was the clinical outcome of this study. The underlying cause of death was extracted from the death summary (possible cause of death) as determined by the attending physicians. Subjects were considered as censored if: they do not develop the event (death) during study period or lost on follow up.

Ethical consideration
The study was approved by Institutional Review Board (IRB) of Jimma University. It has designated with an IRB number of IHRPGB/219/2018. The principal investigator or data collectors briefed the aim of the study to the patients and signed informed consent was taken from all participants or their care givers prior to data collection. During data collection, confidentiality was ensured and for this reason, name and address of the patient was not recorded in the data collection check list.

Statistical analysis
Data were entered into EpiData 3.2 and exported to statistical package for social sciences (SPSS) Version 21.0 for cleaning and analysis. Descriptive analysis was performed and results were presented by text, tables and figure. Kaplan-Meier (log rank test) was used to compare survival experience of the patients. Chi-square test was performed to check adequacy of cells before performing Cox regression. Bivariate Cox regression was performed to identify candidate variables for multivariable Cox regressions. Variables with p-value � 0.25 in bivariate regression were considered as candidates for multivariable regression. Multivariable Cox regression was performed to identify independent predictors of in-hospital mortality. Hazard ratio was used as a measure of strength of association. Predictors with p-value < 0.05 were considered to declare a statistical significance.

Overview of the study participants
During the study period, 144 HIV-positive patients were admitted to JUMC and TASH. Of these, 8 patients were excluded (age <18 = 6, readmitted = 2). During the follow up period six patients were lost to follow up, basically due to loss of hope in modern medicine and strong belief in spiritual healing. Finally, 136 patients were included in the data analysis. Majority, 76 (55.9%) of the patients were from JUMC. The overall analysis time under risk was 74.9 months or 6.24 years.
Socio-demographic and behavioural characteristics. Of 136 eligible patients, 80 (58.8%) of them were females. The mean ± SD age of the patients was 36.8 ±11.3 years. More than half, 73 (53.7%) of the patients, had baseline body mass index (BMI) of greater than 18.5 kg/m2. Majority, 99 (72.8%) of the patients were from urban areas. Fifty one (37.5%) of them had attended primary education and only 36 (26.5%) of them had higher education. A higher proportion of the patients, 69 (50.7%) were unemployed. Regarding the marital status, 58 (42.64%) of the patients were married while, 22 (16.2%) of them were divorced. Fifty nine (43.4%) of the patients were from Oromo ethnic group followed by Amhara 43 (31.6%). With respect to social drug use habit, 34 (25%) and 14 (10.3%) of the patients reported alcohol consumption and cigarette smoking respectively. About 56 (41.2%) participants had average monthly income of less than 1500 Ethiopian Birr (ETB). Socio-demographic and behavioral characteristics are depicted in Table 1.

Baseline clinical, laboratory and drug related characteristics
A CD4 T lymphocytes (CD4 cells) count was done in 133 (97.8%) patients, and 80 (60.1%) of them had less than 200cells/μL. The mean ± SD hemoglobin (g/dl) of the patients was 10.3 ± 3.9. The serologic test showed that, 13 (9.6%) and 9 (6.6%) of the patients had positive hepatitis B virus (HBV) and hepatitis C virus (HCV) surface antigen respectively. Majority, 107 (78.7%) of the patients were already on cART. Of patients on cART, 93 (86.9%) of them were on treatment for more than 6 months. Majority, 87 (81.3%) of the patients were on first line cART regimen and a combination of Tenofovir, Lamivudine with Efavirenz (TDF+3TC +EFV) was the most, 77 (56.6%) prescribed cART regimen. Regarding WHO clinical stage, 60 (44.1%) of the patients were at WHO clinical stage four. In-hospital survival was found to have a significant association with WHO clinical stage of the patients (p = 0.04). The mean (±SD) since the diagnosis of HIV was 4.5± 4.1 and 6.4± 5.2 years for survivors and non-survivors respectively (p = 0.01). Majority, 108 (79.4%) of the patients were on cotrimoxazole preventive therapy (CPT) at admission. About, 59 (43.4%) of the patients had previous history of opportunistic infections. Baseline clinical, laboratory and drug related characteristics of the study subjects are shown in Table 2.

Clinical outcome
Mortality. The overall in-hospital incidence of mortality was found to be 2.83 per 1000 person-years. The incidences of mortality were 6 AIDS and Non-AIDS illness was made based on annex "A" by the caring physician who is blind to this study. Twenty (28.6%) patients were died from JUMC, while nineteen (28.8%) died from TASH (p = 0.49). The diagnosis of HIV was made in 111 (81.6%) patients before current hospital admission and there was no baseline difference in case-fatality between known HIV/AIDS patients and those newly diagnosed at admission (p = 0.16). A difference in the incidence of mortality was observed between those on cART and non-cART at baseline (p = 0.030). In-hospital incidence of mortality was also different at baseline between patients who were on CPT and non-CPT (p = 0.02). A higher proportion of patients who were on non-invasive ventilation were died (p<0.001). The mean ±SD, length of hospital stay was 16.5±10 and 16.7± 12.9 days for in-hospital survivors and non-survivors respectively (p = 0.01). Follow up outcome of the study subjects are shown in Table 3.
Gastrointestinal bleeding, 7 (37.8%), non-recurrent bacterial infections, 6 (36%), and Non-AIDS defining cancers (NADC) were the major contributors for case fatality among non-AIDS related admissions. Cause of death among patients admitted with AIDS and non-AIDS related diseases are depicted in Table 4.
The mean ± SD survival times among patients with AIDS and non-AIDS related diseases were 32 ± 3.1and 34± 3.3 days respectively (log rank p = 0.599) (Fig 1).

Discussion
This study assessed the clinical outcome of hospitalized HIV/AIDS patients treated at two selected Ethiopian tertiary care hospitals. The (mean ± SD) age of the patients was (36.8± 11.34) years. About 13 (9.6%) and 9 (6.6%) of them had HBV and HCV co-infection respectively. The overall in-hospital incidence of mortality was 2.83 per 1000 person-years. The incidences of mortality among patients with AIDS and non-AIDS related illnesses were 6. Most of the deaths were attributed to cryptococcal meningitis, 5 (25%), followed by disseminated tuberculosis, 4 (20%), and cerebral toxoplasmosis, 2 (10%). The mean ± SD survival times of the patients with AIDS and non-AIDS related diseases were 32 ± 3.1 and 34 ± 3.3 days respectively (log rank p = 0.599). Baseline body mass index (BMI) of less than 18.5 (p = 0.04) and being on non-invasive ventilation p = 0.015) were independent predictors of mortality. The (mean ± SD) age of the admitted patients was (36.8± 11.34) years. This is similar with the study conducted in Uganda [18], but less than reported figures in the United States [19].
This finding indicated that HIV/AIDS is affecting the most productive age group of the society, having a significant impact on their family and country as a whole. About 13 (9.6%) and 9 (6.6%) of the patients had HBV and HCV co-infection respectively. This finding is comparable for HBV with CDC report, that 10% of HIV infected Americans are co-infected with HBV but lower than reported figure for HCV in which 25% of them are co-infected with HCV [20].
In this study, the over all in-hospital incidence of mortality was 2.83 per 1000 person-years. This finding is similar with the report from Nigeria (28.6%) [21]. However, in a finding from Ghana, higher (40.6%) in-hospital mortality was reported [22]. The deviation might be due to the study design (retrospective), larger sample size (547 vs. 136) and study setting. Slightly higher proportion of death was reported among patients with AIDS related illnesses, 51.3% versus 48.7% (p = 0.68). This finding is in line with the study conducted in Spain, in which 53% of deaths were attributed to AIDS defining illnesses [23]. Similar study from Brazil reported that in-hospital mortality was almost two times higher in AIDS-related hospitalizations than in non-AIDS-related hospitalizations [24]. However, several studies from developed countries reported non-AIDS-related illnesses as being the most common cause of death [25][26][27]. In a Swiss cohort, majority (84%) of deaths were caused by AIDS unrelated cases [26]. This difference could be because of better socio-economic, education and patient care in developed settings. Cryptococcal meningitis (25%), disseminated tuberculosis (20%) and cerebral toxoplasmosis (10%) were the leading cause of death among AIDS-related admissions. This finding is comparable with the study conducted in West Africa [14]. The possible reason that cryptococcal meningitis was a leading cause of death is because of the fact that, majority of patients admitted with AIDS-related illnesses had a CD4 cell count of <200 cells/μL. In contrary to our finding, tuberculosis is the leading cause of morbidity and mortality among people living with HIV worldwide [20]. Clinical outcome of admitted HIV/AIDS patients in Ethiopian tertiary care settings Among patients admitted with non-AIDS related illnesses, gastrointestinal bleeding (36.8%), non-recurrent bacterial infections (31.6%), and non-AIDS defining cancers (NADC) (26.3%) were the major cause of death. However, studies conducted in New York [28] and West Africa [14] reported sepsis as a major cause of death in non-AIDS related admissions. The mean ± SD survival time between those admitted with AIDS and non-AIDS related illnesses was not statistically different (log rank p = 0.599).
In this study, patients with low body mass index (<18.5kg/m2) were found to be at increased risk of death (AHR: 2.6[1.03, 6.45]). This finding is supported by the study conducted in Rwanda [29]. According to Anjali Sharma et.al [30], AIDS related admissions had doubled the hazard of death in underweight (BMI<18.5 kg/m2) patients (AHR: 2.04, 95% CI [1.03, 4.04]). Even though the underlying mechanisms placing underweight HIV-infected patients at risk of death are unclear, several factors which could have facilitated progression to death were more prevalent among our study participants. The most important once include, majority of patients at advanced stage of the disease and low CD4 count. Furthermore, malnutrition (typically defined as a body mass index (BMI) of <18.5 kg/m2) aggravates the underlying immunosuppression enhancing their susceptibility to various infections and HIV disease progression [31][32].
Being on non-invasive ventilation (NIV) was also associated with higher hazard of mortality (AHR: 2.99 [1.24, 7.28]). This finding is in line with studies conducted in Uganda [13], Mexico  [33] and Brazil [34]. The possible explanation could be that most of patients in this study were at an advanced stage of the disease and they tended to be critically ill with multiorgan failure during admission. Our study has several limitations. Inclusion of small number of patients; loss of follow-up during the study period; inclusion of only two tertiary hospitals and lack of diagnostic equipment, such as viral load measurement, were the limitations of the present study. We recommend a large scale, multicenter prospective cohort study to be conducted, including other tertiary care hospitals in the country.

Conclusion
This study found high incidence of in-hospital mortality among admitted HIV/AIDS patients in Ethiopian tertiary care settings. The overall incidence of mortality in our study is still higher than middle and high income countries. Furthermore, this study found being on non-invasive ventilation and low body mass index as independent predictors of mortality. This target group's needs special attention as they carried higher hazards of mortality.