Introduction

The term chronic kidney disease refers to reduced kidney function that lasts for at least three months and a glomerular filtration rate (GFR) of less than 60 mL/min per 1.73 m2, as well as the existence of signs of kidney damage [1].

In chronic kidney disease, there is a problem with the kidney filtering out and performing its activity. Due to this, dialysis can be used, whether in the form of hemodialysis or peritoneal dialysis. Hemodialysis is a type of dialysis modality used for about 90% of patients globally and the mainstay therapy for chronic kidney patients who cannot undergo renal transplantation, whereby blood is put through a filter outside the body, cleaned, and then returned to the body to perform its functions [2].

Approximately 2 million patients worldwide undergo regular renal replacement therapy in the form of dialysis to maintain their lives, with around 80% receiving hemodialysis and 15% receiving peritoneal dialysis [3]. In Ethiopia, the first dialysis was started in 1980 at Tikure Anbessa specialized hospital; later, maintenance dialysis started in a private center in 2001 at St. Gabriel Hospital; and the first transplant occurred in September 2015 at St. Paul’s Hospital Millennium Medical College [4].

Currently, there are sustainable development goals (SDGs) that work toward decreasing non communicable disease and also enhancing renal disease prevention and treatment for all age groups [5].Ethiopia is also one of the countries that is striving to achieve sustainable development goals aimed at reducing non communicable-related deaths up to 2030, even if more than one-third (39%) of all annual deaths are caused by NCD [6].

According to the global burden of illness, chronic renal disease is the 12th most common cause of mortality and is responsible for 1.1 million deaths globally [7]. Additionally, it was placed 27th in the list of factors contributing to all deaths globally in 1990 but rose to 18th in 2010 [8]. Besides, it is expected to be the fifth-leading cause of death by 2040 around the globe [9].

Even if dialysis is used as a treatment modality in Ethiopia, access to it is quite limited and unaffordable to low-socioeconomic societies [10].

As far as my concerned, there is only a limited amount of information available about time to death and its predictors of mortality in our country. Additionally, evidence from previous studies indicated that there is an increasing number of hemodialysis patients. That's why studying time to death and predictors of mortality is crucial to increasing the survival time of hemodialysis patients. Furthermore, most studies are confined to a single dialysis unit, but this study tried to include both public and private dialysis units and also attempted to include some predictors. This study aimed to investigate the time to death and its predictors of mortality among hemodialysis patients in selected dialysis centers in Addis Ababa, Ethiopia.

Methods and material

Study area and period

The study was conducted in dialysis unit in Addis Ababa, the capital of Ethiopia. The total population Addis Ababa has 3,273,000 from this 1,722,000 female and 1,551,000 male [11].

There are 35 hemodialysis units found in Ethiopia from those only eleven units are government-run hospitals the rest are privately owned [12]. The study was carried out from April 1st -May 20, 2023.

Study design and population

An institution-based retrospective cohort study design was used. All patients with chronic kidney disease undergoing hemodialysis who attended dialysis units in Addis Ababa were source population and all patients with chronic kidney disease undergoing hemodialysis at randomly selected dialysis units in Addis Ababa during the period from January 1st, 2017 to December 30, 2022 were study population. On the other way study unit were those individual patient with chronic kidney patients undergoing hemodialysis attended dialysis units, who had been sampled and fulfilled the inclusion criteria, those adult patients who had hemodialysis for chronic kidney disease during the specified period and aged 18 and above were included in the study whereas Patients who received hemodialysis for acute renal failure and were transferred from other dialysis facilities or whose time from the start of hemodialysis to the end of hemodialysis was shorter than 30 days and lost medical records were excluded from the study.

Sample size determination and sampling procedures

The minimum required sample size was calculated by using a survival analysis formula. Using STATATM Version 15 statistical software to calculate the sample size by considering the following assumptions: the level of significance (α) = 5%, Zα /2 (value at 95% confidence interval =1.96), power of 80%, and hazard ratio and probability of death Table 1.

Table 1 Sample size calculation for those hemodialysis patients who were on hemodialysis from January 1st, 2017 to December 30th, 2022, at selected hospitals giving dialysis Service, Addis Ababa, Ethiopia (n = 370), 2023
Full size table

Therefore: The larger sample size was, n = 310, and after adding a 10% for incomplete records the final sample size will be about n = 341.

Sampling procedures

From a total of 20 dialysis units, five hospitals that provide dialysis were selected through simple random sampling methods. Within five years, a total of 545 hemodialysis patients with CKD were registered. In Zewditu Memorial Hospital, a total of 45 hemodialysis patients were registered, but finally 20 hemodialysis patients were used according to that fulfilled inclusion criteria; in St. Paul’s Hospital, Millennium Medical College, 146 hemodialysis patients were registered, but finally 60 hemodialysis patients were used; in Addis Hiwot General Hospital, 132 hemodialysis patients were registered, but 120 hemodialysis patients were used according to inclusion criteria; and in Bethzatha General Hospital, 112 hemodialysis patients were registered from the total of registered according to inclusion criteria, 90 hemodialysis patients were included, EthioTebib General Hospital 110 hemodialysis patients of CKD were registered but finally 80 hemodialysis patients were included in the study. Finally, a total of 370 hemodialysis patients with CKD met the inclusion criteria; the calculated sample size was 341. However, since it was manageable, all 370 eligible chronic kidney patients who started hemodialysis in a selected hospital were included.

Study variables

The dependent variable was Time to death whereas the independent variable were Socio-demographic factors(Age, Sex, Marital status &Residence) Clinical characteristics of the patient (Comorbidity, Charlson comorbidity index, Cause of chronic kidney disease, Family history & blood group) Biochemical parameters (Albumin, Hemoglobin, Potassium, Creatinine &Calcium level) Treatment-related factors (Types of vascular access, Dialysis duration per session, Dialysis frequency per week, adequacy of dialysis, Erythropoietin Treatment, Health facilities, Existence of infection).

Operational definitions

CKD

Gradual reduced kidney function > 3 months and a glomerular filtration rate (GFR) of less than 60 mL/min [1].

Acute kidney disease

Sudden reduction of kidney function with a duration of ≤3 months [15].

Time to death

Length of time it takes for hemodialysis patients to develop an event of interest measured in days from start of hemodialysis to the date of death.

Event

Death of patients due to chronic kidney disease on hemodialysis.

Censored

Those patients who didn’t develop the outcome of interest at the end of the follow-up period or were alive; loss of follow-up; transfer to another dialysis center; or kidney transplant.

Charlson comorbidity index

An index consist 17 medical condition weighted 0–6 and weighted as mild, moderate and severe based on number of comorbidity where 1–2, 3–4 & ≥ 5 respectively [16].

Data collection tool and techniques

The data were collected by using a data extraction checklist that was adapted from different studies [17,18,19,20,21].Which consist of socio-demographic-related factors, clinical characteristics of the patient, biochemical parameters, and treatment-related factors. Data were gathered by reviewing patients’ medical records, hemodialysis charts, and registration book. A checklist with an English version was used for data collection to facilitate the extraction of data. Ten BSc nurses (two in every hospital) were involved in the data collection.

Data quality control and management

Data collectors and supervisors received one day of training on the data gathering procedure regarding to objective of the study, and the study population’s eligibility criteria. The investigator and supervisors daily checked the obtained data for consistency, and completeness to minimize the chance of loss of data and to ensure data quality.

Data processing and analysis

After the data was collected, each variable was coded and entered into Epi Data version 3.1 and then exported into STATA version 15 software for analysis. In order to handle missing values for the variables, the imputation method was used. Kaplan-Meier and the log-rank test were done. Bivariable Cox-proportional regression was done, and a variable whose p-value < 0.25 entered into multivariable Cox-proportional regression analysis. In order to check the proportional hazard assumption, the log-log plot and Shenfield residual test was used, which had a P-value >0.05 was considered as fulfill the assumptions. And multicollinearity was done by using variance inflation factor a mean VIF was 1.18. Finally, a variable whose p-value < 0.05 in multivariable Cox-proportional regression analysis was considered a statistically significant predictors for time to death among patients on hemodialysis. Overall model fitness was also checked by the Cox-Snell residual. Finally, the result was presented through tables, text, and graphs.

Result

Socio demographical characteristics of study participant

A total of 370 hemodialysis patient records were included in the study, with the mean age of the participants 46.38 ± 15.72 years. Two hundred thirty three (62.97%) participants were males, and 319 (86.45%) of the patients were residents of Addis Ababa (Table 2).

Table 2 Socio demographical characteristics of study participant of chronic kidney diseases patients on hemodialysis at selected hospitals giving dialysis Service, Addis Ababa, Ethiopia, 2023
Full size table

Clinical characteristics of the study participant

From a total of 370 patients, 313 (84.6%) had comorbidities, and from the comorbidities, the leading one was hypertension 208 (56.2%). The most common potential causes of chronic kidney disease were hypertension 113 (30.54%), diabetes mellitus 108 (29.19%), and both hypertension and diabetes mellitus 64 (17.3%), whereas 48 (12.97%) of patients had chronic kidney disease for which the etiology was unknown (Table 3).

Table 3 Clinical characteristics of the study participant of chronic kidney diseases patients on hemodialysis at selected hospitals giving dialysis Service, Addis Ababa, Ethiopia, 2023
Full size table

Treatment related and biochemical parameters of study participant

The majority of the patients used Arteriovenous fistula for vascular access, which is 224 (60.54%), and most of the patients there hemoglobin levels were < 11mg/dl 215 (58.1%). About 123 (33.24%) patients undergo hemodialysis three times a weeks, whereas 137 (37.03%) and 110 (29.73%) used twice and once per week, respectively. Regarding blood groups, 120 (32.4%) were blood groups O, whereas, 82 (22.2%), 114 (30.8%), and 54 (14.6%) were blood groups A, B and AB, respectively (Table 4).

Table 4 Treatment related and biochemical parameters of chronic kidney diseases patients on hemodialysis at selected hospitals giving dialysis Service, Addis Ababa, Ethiopia, 2023
Full size table

Time to death of chronic kidney disease patients on hemodialysis

Of the total of 370 patients who were on HD for at least one month, were followed for a total of 60 months, with a minimum and maximum follow-up time of 1 and 59 month, respectively. From the total of 370 eligible patients, 265 (71.6%) were censored and 105 (28.4%) (95% CI: 23.8-33.3) died. The study participants were followed for a total of 6245 person-month, with an incidence rate of death 16.8/1000 person-month of observation (95% CI: 13.8-20.3). The median time to survival was 47 months (95% CI: 36.7, 56). The probability of survival at 3 month, 1st, 3rd, and end of follow-up years was 92.76%, 80.1%, 58.14%, and 24.95 %, respectively (Figs. 1, 2, 3 and 4).

Fig. 1
figure 1

Kaplan–Meier survival estimate of chronic kidney disease patients on hemodialysis at selected hospitals giving dialysis Service, Addis Ababa, Ethiopia, 2023

Full size image
Fig. 2
figure 2

The outcome of chronic kidney disease patients on hemodialysis at selected hospitals giving dialysis Service, Addis Ababa, Ethiopia, 2023

Full size image
Fig. 3
figure 3

There is survival difference b/n a vascular access with log rank test at p value of 0.0000 and chi.2 of 41.89. Kaplan–Meier survival curve for vascular access of chronic kidney disease patients on hemodialysis at a selected dialysis center in Addis Ababa, Ethiopia, 2023

Full size image
Fig. 4
figure 4

It indicated that there is a difference of survival between those who had cardiovascular disease than of without cardiovascular disease at p-value of 0.0000. Kaplan–Meier survival curve for cardiovascular disease of chronic kidney diseases patients on hemodialysis at selected hospitals giving dialysis Service, Addis Ababa, Ethiopia, 2023

Full size image

Predicators of mortality among chronic kidney disease on hemodialysis

At a p-value of <0.25 in Bivariable Cox regression analysis eleven variables were selected, those were sex, age, infection, hypertension, vascular access, CVD, blood type, hemoglobin level, adequacy of dialysis, existence of comorbidity, and the Charlson comorbidity index. The global goodness-of-fit test of Shenfield residuals was used to verify the proportional hazard assumption. Which had a global test p-value of 0.8092 >0.05 which satisfy PH assumption.

Those eleven variables from Bivariable cox regression entered to multivariable cox regression. Among the eleven variables that were entered into cox multivariable cox regression analysis, four variables became statistically significant at a p value of <0.05, those are central venous catheter, older age, CVD, and blood type B.

Those patients who used central venous catheter for vascular access were 3.47 times risk of death than those with AVF. Those patients who were aged 64 and older were 2.8 times more likely to die than those in the age group of 18–47 years. The risk of death in those hemodialysis patients whose blood group B has 2.07 times than blood type O Those patients with cardiovascular disease were 1.88 times more likely to die than those patients without cardiovascular disease (Table 5) (Fig. 5).

Table 5 Cox proportional hazard model analysis for predictors of time to death among chronic kidney diseases patients on hemodialysis at selected hospitals giving dialysis service, Addis Ababa, Ethiopia, 2023
Full size table
Fig. 5
figure 5

Cox-Snell residual plot to test Cox regression overall model fitness test for chronic kidney disease patients on hemodialysis at selected hospitals giving dialysis Service, Addis Ababa, Ethiopia, 2023

Full size image

Discussion

The objective of this study was to assess time to death and its predictors among patients with chronic kidney disease on hemodialysis at selected hospitals giving dialysis service in Addis Ababa, Ethiopia, 2023. The incidence rate of death in this study was 16.8 per1000 person-month. which is consistent with study conducted in Tigray 1.4 per 1000 person year [18]. However lower than study conducted in Amhara 1.89 per 1000 person-days [17]. Which might be due to a variation in sample size and study setting.

The median survival time in this study was 47 months, which is higher than the study conducted in Ayder hospital in Tigray region which was 378 days [18].This might be a result of the fact that the majority of the patients in Tigray were older aged, and also it was carried out over a three-year period in a single public dialysis facility.

On the other hand the overall survival probabilities at 3 month, 1, 3, and end of follow up years were 92.76% 80.1%, 58.14%, and 24.95% respectively. The finding is consistent with study conducted in Vietnam 1st year was (85%) [22], but lower than a study conducted in china 1st year survival was 94% [20].This variation might be due to a difference in frequency of dialysis and socio demographical characteristics of participants. However, higher than study conducted in Nigeria, 90thday and 1st year survival probability were (24.86%) & (66.3%) respectively [23] and in Ethiopia, Amhara 1st year survival was (49.58%) [17] and Saint Gabriel General Hospital 90th days and 1st year survival was (62.1%) and (42.1%) respectively [10]. This might be due to variation on socio demographical characteristics of participant, study setting and study period.

Those patients whose (aged ≥64 years) had 2.8 times more likely to die than of young adults (aged 18–47 years). This finding is supported by other previous studies conducted in Vietnam [22], Iran [24] and Adama [14]. This might be due to the fact that older age is associated with lower levels of physical functioning [25] as a result it increase the risk of different comorbidities and makes them more susceptible to hemodialysis instability. However there was no significant association b/n age and mortality of hemodialysis patients in study conducted in Tigray [18]. Thus, the difference may be due to disparities among the study sample size and study period.

Patients using central venous catheters have 3.47 times higher risk of death than patients using an arteriovenous fistula as their vascular access. The result was supported by studies done in Taiwan [21], Palestine [26], Japan [27] and Amhara [17]. This could be because of central venous catheter had a high risk of infection, thrombosis and mechanical stress, which caused vein stenosis, which lowers blood flow and lowers the required dose of dialysis, as a result decreased patient survival [28]. However this finding not supported by a study conducted in china [20]. This difference could be due to low usage of central veins catheter or due to sample difference.

Hemodialysis patients with blood group B had twofold risk of death than blood group O the result of this study supported by a study conducted in china [29].This might be due to the reason that non-O blood groups had a great probability of thrombosis, due to increased amounts of von Willebrand factor (VWF) and factor VIII (FVIII), as a result it makes more susceptible to thromboembolic disorders and it affect patency and longevity of the vascular system [30]. However there was no significant association in study conducted in Iran [24].

Patients with cardiovascular disease were increase risk die by 88% than those without cardiovascular. This finding is in line with the studies carried out in Casablanca [31],China [32].This may be due to there is high risk of fluid overload due to loss of renal function as well as due to dialysis treatment, which increases the risk of the pathophysiological processes of hypertension and left ventricular hypertrophy it worsens the progression of CVD [33].

In this study there is no survival difference b/n public and private dialysis unit thus this study inconsistent with study conducted in Romanian in which those private dialysis unit have better outcome than of governmental (non-profit) hemodialysis [34]. This difference might be due to dialysis treatment protocols and practices may be standardized across public and private units, ensuring consistent care delivery regardless of the type of facility.

Conclusion

Overall, in this study show that the median survival time was 47 month and those predictors or factors cardiovascular disease, older adults (aged ≥64), catheter vascular access, and blood type B were significant predictors of mortality that affect the survival of hemodialysis patients. Therefore, every concerned bodies who works in health sectors particularly in dialysis unit must be to intervene for those predictors of mortality and create interventions accordingly in order to increase survival time hemodialysis patients. In general this study will be contribute to improving the care of dialysis patients in the country that’s why nephrology policy maker also must give concern and prepared guideline for dialysis patients.

Strengths of the study

This study collect a sample from multiple dialysis unit and in cooperate both private and public dialysis unit and its make the study more generalizable, additional to this it try to incorporate some predicators which has not address from previous study.

Limitations of the study

Due to the secondary sources used to gather the data, some crucial variables like albumin level, creatinine level, calcium and potassium level were overlooked.