Background

Although 1.9 million patients are undergoing renal replacement therapy (RRT) worldwide, yielding usage of 316 per million population, and annual initiation of 73 per million population, only about one-third (648,000) of the patients reside in developing regions, which contribute 85% of the world’s population [1]. There is also a high mortality rate of patients on hemodialysis, especially in the first 3 months after initiation of dialysis with annual mortality around 9% per year and with a 5-year survival of 40–50% [2].

In many developing countries, there is a shortage of renal replacement services which causes an estimated 2.3–7.1 million premature deaths [3]. African patients with ESRD have the lowest access to RRT with only 9–16% being treated; in central and eastern Africa, the treatment rate is estimated to be as low as 1–3% [3, 4].

Despite the establishment of renal registries and publication of data about RRT, there is a lack of countrywide publication in many African countries due to sustainability issues [5]. Consequently, less is known about initial survival or the risk factors which cause early mortality in dialysis patients. However, mortality during the first 90 days remains high, and modifications might be made during this critical period to impact not only early survival but also possibly long-term survival of patients on dialysis [6, 7].

Ethiopia like other developing countries struggles with a double burden of communicable and non-communicable diseases that contributes to the growing burden of ESRD, so there is a limitation of resources to address RRT to all patients who require it [8,9,10]. Since the first private dialysis center was opened in the early 2000s and the first public unit in 2013 dialysis is considered to be a service for the rich [10, 11]. The current Ethiopian health care system is structured into a three-tier system namely primary, secondary and tertiary level health care. Hemodialysis service is given at tertiary level health care and the main source to cover the expenses of the service is out of pocket money of the patients [10].

Inaccessibility of dialysis centers and financial constraints were the main reasons for delayed initiation of hemodialysis and frequent discontinuation in the country [10, 12]. Frequent breakdown of machines, electric power fluctuation, and lack of filtered water were also the main reasons for reduced sessions per week and reduced duration of hemodialysis per session from the standard [12]. In addition to the limited access to the service, quality is also compromised due to the limitation of duration and frequency of treatment, and the restriction of publicly funded dialysis for patients eligible for transplantation [10]. The cost of hemodialysis treatment among end-stage renal disease patients at the tertiary hospitals of Addis Ababa city and the Amhara region shows that the annual cost of the hemodialysis treatment was ETB 121,089.27 ($4466.59) per patient per year [13].

In Ethiopia, many prevalence studies on CKD have been done and show the increasing burden of the disease but only a few studies have been done to understand survival rates and predictors of mortality among CKD patients on hemodialysis [14,15,16,17]. The previous studies were conducted using relatively small sample size and single dialysis unit. In Amhara regional state there were no previous studies done on the same topic to date.

There are two governmental and one non-governmental hemodialysis center in the region namely; Felege Hiwot, the university of Gondar teaching referral hospitals, and Gambi teaching hospital. The Hospitals give service to nearly 800 patients. Felege Hiwot comprehensive specialized hospital and Gambi private teaching hospitals are found in Bahir Dar city. Gambi teaching hospital started hemodialysis service in 2014 while Felege Hiwot comprehensive specialized hospital started in 2015. University of Gondar teaching hospital is found in Gondar town, northwestern Ethiopia. The hospital provides inpatient and outpatient medical services in several departments. The hospital started hemodialysis service in 2017 [18].

The study was done to fill the identified gaps using multiple dialysis units to effectively assess the survival status and predictors of mortality among CKD patients on hemodialysis in Amhara region, Ethiopia.

Methods and materials

An institution-based retrospective follow-up study was conducted among chronic kidney disease patients who undergo hemodialysis from March 1, 2016, to March 1, 2021, in the region. The medical record numbers of the patients from Felege Hiwot (385 medical records), Ghambi (200 medical records), and Gonder hospitals (215 medical records) were retrieved. Then by merging the registered medical numbers to create a sampling frame of 800 medical records, the 436 computer-generated random records were selected.

The dependent variable of the study is event i.e., death at time t (event = 1 and censored = 0) and the independent variables include socio-demographic factors, clinical factors, possible etiology of CKD, and presence of comorbidity.

The information available on the eligible patients’ medical records was collected using a data extraction tool prepared by adapting from different works of literature [14,15,16,17]. The selected charts of CKD patients on hemodialysis between March 01, 2016, to March 01, 2021, at Felege Hiwot, Gondar, and Gambi hospitals were retrieved and then reviewed. Baseline data of the patients were taken from the charts. Death certificates were extracted from the hospital registries by the patient’s medical record number. For those patients whose outcomes were not registered, it was confirmed using a phone call to their close family members.

In the current study censored means patients whose status was unknown (lost to follow-up), patients who did not develop the outcome of interest (death) until March 01/2021, those patients transferred or referred to other health institutions, and patients who died with an accident (e.g., car accidents). The event is the death of CKD patients on hemodialysis. The follow-up period was the time from the beginning of the study (March 01/ 2016) to an event, the end of the study (March 01/2021), or loss of contact or withdrawal from the study.

Data processing and analysis

After the data collection, the collected data were entered into Epi data version 3.1 and then exported to STATA™ 14.1 for data cleaning and analysis. The life table was used to estimate probabilities of survival after initiation of hemodialysis at different time intervals and the cumulative probability of survival for each year. Kaplan Meier survival curves were used to compare the survival differences among categories of the variables.

Before running the Cox Proportional hazard regression model, the proportional hazard assumption was checked using global goodness of fit based on Schoenfeld residual, and variables having a P-value > 0.05 were considered as fulfilling the assumption. The p-value for the global test was 0.5661, and each predictor variable had a p-value greater than 0.05.

To identify potential predictors of mortality of ESRD patients on hemodialysis bi-variable Cox proportional regression model was fitted for each explanatory variable. Those having a P value less than 0.25 in the bi-variable analysis were included in the multivariable Cox proportional hazard regression model in the multivariable analysis. Variables having a P value less than 0.05 with 95% CI was considered significant predictors of time to death and an adjusted hazard ratio was used to show the strength of association between each predictor variable and the outcome variable. Finally, the result of the study was presented using text, tables, and graphs.

Results

Socio-demographic characteristics of the patients

Among 436 patients who participate in the study, the median age was 45 years (IQR 55 − 35 years). The number of male participants was 268 (61.5%). More than half of the participants live in urban residences 282 (64.7%). Most of the patients 365 (83.7%) had no family history of CKD (Table 1).

Table 1 Sociodemographic characteristics of CKD patients on hemodialysis in Amhara regional state, Ethiopia, 2016-21 (n = 436)
Full size table

Etiology of end-stage renal diseases

Among 436 ESRD patients who were on hemodialysis, the primary causes of chronic kidney disease were 225(51.6%) hypertension, 130(29.8%) diabetes, 137(31.4%) glomerulonephritis, 31(7.1%) polycystic kidney disease, 47(10.8%) HIV and 35(8.0%) of the patients due to unknown causes as shown in Table 2.

Table 2 Primary causes of end stage renal disease among patients on hemodialysis Amhara regional state, Ethiopia,2016-21 (n = 436)
Full size table

Presence of comorbidity

Among the 436 patients, 396(90.8%) had comorbidity. The leading comorbidity was hypertension 313(80.3%) followed by anemia 245(62.8%) and diabetes 112(28.7%) (Table 3).

Table 3 Types of comorbidities among ESRD patients on hemodialysis in Amhara regional state, Ethiopia, 2016-21 (n = 396)
Full size table

Clinical characteristics of the patients

Among the 436 patients, 221(50.7%) used arteriovenous fistula, the other 171 (39.2%) and 44(10.1%) use a central venous catheter and arteriovenous graft respectively as their vascular access. About 331(75.9%) patients undergo hemodialysis twice, 73(16.7%) three times, and 32(7.3%) once per week. Regarding the duration of hemodialysis per session 378(86.7%) spend 4 h, 35(8.0%) for three and half hours and 23(5.3%) spend 3 h on hemodialysis per session (Table 4).

Table 4 Types of vascular access among ESRD patients on hemodialysis in Amhara regional state, Ethiopia, 2016-21 (n = 436)
Full size table

Survival outcomes of ESRD patients on hemodialysis

Among 436 patients who were included in the study 153 (35.1%) (95%CI (30.7, 39.7)) had died during the 5 years follow-up time with a median survival time of 345 days (95%CI (278, 419)). The 436 patients contributed 80,774 person-days to the follow up making the incidence rate of death 1.89 (95%CI (1.61, 2.21)) per 1000 person-days.

The survival probability of the patients at three months was 85.46% (95%CI (81.61, 88.57)), and at 1 year it was 49.58% (95%CI (42.13, 56.58)) as shown in Table 5.

Table 5 Survival probability of ESRD patients on hemodialysis with 90 days interval in Amhara regional state, Ethiopia, 2016-21
Full size table

The overall Kaplan-Meier survival curve also shows a steep decrease in the survival rate of CKD patients on hemodialysis in the first 540 days (where the highest number of deaths occurs), then the curve becomes steady until 720 days and it steadily declines afterward as shown in Fig. 1.

Fig. 1
figure 1

Overall Kaplan-Meier survival curve with 95% confidence interval showing the survival time of chronic kidney disease patients on hemodialysis in Amhara region, Ethiopia, 2021

Full size image

Predictors of mortality among end-stage renal disease patients on hemodialysis

Variables that had a P-value < 0.25 in the bivariable cox regression were residence, CKD secondary to hypertension, CKD secondary to glomerulonephritis, CKD secondary to HIV, vascular access, frequency of dialysis per week, duration of dialysis per session, adequacy of dialysis, and age of the patients.

The proportional hazard assumption was checked statistically using the global goodness-of-fit test using Schoenfeld residuals. The proportional hazard assumption was fulfilled with a global test P-value of 0.5661. The assumption was also checked for each predictor with minimum and maximum p values of 0.0825 and 0.8873 respectively (Table 6).

Table 6 Proportional hazard assumption output table of the variables included in the multivariable analysis among ESRD patients on hemodialysis in Amhara regional state, Ethiopia, 2016-21
Full size table

Among the variables which were included in the multivariable Cox proportional hazard model five variables, residence, hypertension as a cause of CKD, HIV as a cause of CKD, central venous catheter, and 4 h of duration of dialysis per session were significantly (P-value < 0.05) associated with mortality among ESRD patients on hemodialysis after adjusting for the other variables.

The risk of mortality among patients who live in rural residences was 1.48 times (95%CI (1.04, 2.12)) higher than patients who live in urban residences. The risk of death was 1.49 times (95%CI (1.01, 2.23)) higher among patients with ESRD secondary to hypertension than patients with other causes. Those patients with ESRD secondary to HIV had 2.22 times (95%CI (1.41, 3.51)) higher risk of mortality than patients with other causes of ESRD. Patients who used central venous catheters had 3.15 times (95%CI (2.08, 4.77)) higher risk of death than those patients who use arteriovenous fistula as their vascular access.

The risk of mortality among patients who stayed 4 h on hemodialysis decreased by 57% (95%CI (0.23, 0.80)) than those patients who stay 3 h on hemodialysis per session (Table 7).

Table 7 Results from bivariable and multivariable Cox proportional hazard model analysis of ESRD patients on hemodialysis in Amhara regional state, Ethiopia, 2016-21 (n = 436)
Full size table

Discussion

Overall, this study shows that the overall survival proportion of ESRD patients on hemodialysis was 64.9% which was lower than that of the studies done in Ethiopia Adama, Saint Gabrial, and Ayider [15,16,17]. This result might be due to the inclusion of ESRD patients who were on hemodialysis in the first three months of hemodialysis in the current study [19].

The result was also lower than studies done in England, Scotland, the United Kingdom, India, and South Africa [20,21,22,23]. The low survival rate of the study is likely attributed to late presentation, frequent dialysis discontinuation, suboptimum dialysis quality, poor quality of care before referral, financial constraints, and the lack of national screening and management programs for ESRD in most of sub-Saharan Africa including Ethiopia [24, 25].

Even though, the median survival time of the study was 345 days which was consistent with the study conducted in Ayider, the incidence of death in the current study was 1.89 per 1000 person-days which was higher than studies conducted in Ethiopia, Ayider comprehensive specialized hospital and Ghana [7, 15]. This result might be explained by the relatively longer follow-up time of the current study [19, 21].

On the other hand, the median survival time was higher than in a study conducted at Black lion hospital [14]. This result might be due to the time difference between the studies and improved hemodialysis service quality and increased access to hemodialysis in Ethiopia throughout the years [18].

The risk of mortality among patients who live in rural residences was 1.48 times higher than their counterparts which were in agreement with a study conducted in the United States [26]. The high mortality among rural residents might be attributed to distance from the dialysis centers, less affordability, low accessibility, low knowledge about ESRD management due to decreased access to health education, and high susceptibility to infections from their day to day activities among patients from rural residents when compared with patients who live in urban residents [27].

The risk of death was 1.49 times higher among patients with ESRD secondary to hypertension than patients with no hypertension. The result was in line with studies conducted in China, Korea, Cameron, and the United States [28,29,30,31]. Reduced compliance, side effects, and financial costs among hypertensive patients might impact hemodialysis effectiveness, systolic blood pressure, and diastolic blood pressure along with traditional risk factors for cardiovascular disease, which are associated with end-organ damage, including vascular stiffness and mortality of patients on dialysis [32].

Patients whose primary cause of nephropathy was HIV had 2.22 times increased risk of mortality than those patients with a primary cause other than HIV. This result was consistent with a study conducted in Ethiopia, Ayider hospital [15]. This similarity might be attributed to the compromised immune status of HIV-positive patients and the use of vascular access different from AVF, since they are easily susceptible to vascular site infection, bacteremia, and opportunistic infections [33, 34].

Patients who used central venous catheters had 3.15 times more risk of death than those patients who use arteriovenous fistula as their vascular access. The result was supported by studies done in Tehran, Japan, Madrid, and Birmingham [35,36,37,38]. This result might implicate that patients with vascular access to central venous catheters were commonly prone to prolonged hospitalizations, and complications due to serious invasive infections [39].

This study shows that, compared with patients who stay 3 h on hemodialysis per session, staying 4 h on hemodialysis decreases the risk of mortality by 57%, which is consistent with studies conducted in Washington, Australia, New Zealand, and Saint Gebrial [17, 40,41,42]. This might be due to a shorter duration of hemodialysis being more likely to result in complications, increased hospitalizations, and increased mortality [43].

The main limitation of the study was that it does not address socioeconomic factors (such as the income of the patients) that might affect the survival of the patients and that might also be a possible confounder. Since the factors that affect the mortality of patients who were on hemodialysis were not fully included because of the incompleteness of the records, the associations observed in the study might be potentially confounded by underlying factors.

Conclusions

There was a significantly high mortality rate and relatively lower median survival time of chronic kidney disease patients on hemodialysis when compared to middle- and high-income countries. The highest mortality occurred during the first three months after the initiation of hemodialysis. Hence Ethiopian ministry of health shall work towards addressing hemodialysis services for patients who live in rural residences and has low socioeconomic status.

Researchers should consider conducting a prospective follow-up study triangulated with qualitative data to address the behavioral characteristics, socioeconomic status, and body mass index of the patients.