Factors associated with survival among patients with end-stage kidney disease undergoing hemodialysis in Tanzania: a cohort study

Abstract

Background

Despite increased technological advances and trained health workers, morbidities and mortalities of patients on hemodialysis (HD) treatment remain unacceptably high, and factors affecting survival are largely unknown in Tanzania. This study aims to assess factors associated with survival among patients with end-stage kidney disease (ESKD) undergoing hemodialysis in Tanzania.

Methods

This retrospective study involved 386 patients with ESKD aged 18 years and older who underwent HD at Muhimbili National Hospital (MNH) hemodialysis unit from January 2016 to December 2021. The Kaplan-Meier method was used to determine the survival rate of patients, and the Cox proportional hazard model was used to evaluate the factors associated with survival.

Results

Overall, 386 participants were enrolled in the study, with a mean age of 50.7 ± 15.87 years, and the median survival time was 2.5 years. The estimated survival rates for 1-, 2-, and 3-year were 66%, 53%, and 47%, respectively. In the multivariable model, patients with higher risks of mortality were those in the age group ≥ 56 years (Hazard ratio, HR = 2.19, 95% confidence interval, CI: 1.35, 3.53; P = 0.001), with primary education (HR = 2.02, 95%CI: 1.15, 3.54; P = 0.014) and secondary education (HR = 1.81, 95%CI: 1.03, 3.18, P = 0.039), unemployed (HR = 1.83, 95%CI: 1.10, 3.03, P = 0.019), catheter vascular access (HR = 1.44, 95%CI: 1.05, 0.97; P = 0.023); WBC count (HR = 1.38, 95%CI: 1.02, 1.86; P = 0.038) and platelets count (HR = 1.79, 95%CI: 1.29, 2.50; P = 0.001) not in a normal ranges. Patients with a lower risk of mortality were those whose ESKD was due to renal causes (HR = 0.42, 95% CI: 0.18, 0.97; P = 0.041) and those whose model of payment for treatment was Health insurance (HR = 0.58, 95% CI: 0.40, 0.83; P = 0.003).

Conclusion

The results showed that the survival rates for 1-, 2-, and 3-year were 66%, 53%, and 47%, respectively. Younger age, college or university education, employment, health insurance, fistula vascular access, and normal range WBC count and platelet count were factors associated with higher survival rates.

Clinical trial number

Not applicable.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12882-026-04800-9.

Background

Chronic Kidney Disease (CKD) and its progression to End-Stage Kidney Disease (ESKD) are increasingly acknowledged as major risk factors for mortality and have become a critical global public health priority. CKD will be the 5th leading cause of death worldwide in 2040 [1].

Most African countries lack national kidney registries and community–based studies, making it challenging to understand the magnitude and impact of kidney disease in such communities. CKD prevalence in Sub-Saharan African countries is projected to be 13.9% [2]. In Tanzania, the prevalence of CKD was reported to be 7% [3] while in rural areas it was 3.0% [4]. However, there is limited data on the survival of hemodialysis (HD) patients.

Kidney replacement therapy (KRT) is a standard therapy for patients suffering from ESKD that prolongs survival, reduces morbidities, and improves the quality of life [5, 6]. About 1.75 million patients worldwide are on regular dialysis, of which 1.55 million (89%) are on HD and 197,000 (11%) on peritoneal dialysis (PD) [7].

In Tanzania, the number of HD centers has been increasing in both private and government health facilities [8], and it is estimated that more than 3231 patients are on hemodialysis [9]. In most sub-Saharan countries, the services remain unaffordable to most people owing to their high cost [10, 11]. Among approaches to improve the provision of sustained health care services in some countries, HD services have been provided either in full or partial coverage with Health Insurance [7]. Despite several technical advances, morbidities and mortalities of patients on HD remain excessively high, and their quality of life is habitually poor [12].

Although the survival rate differs in different countries, the quality of care received by the patients determines the patient’s survival, and good quality of care is attained through ensuring an adequate dose of dialysis, improving nutrition, timely identification and treatment, and preventing cardiovascular diseases [13–15] or hypertension [10, 15, 16]. Other factors that have been linked with survival among patients with ESKD on HD include gender [17, 18], age [19–21], low socioeconomic status, low social support, and being unemployed [21].

Despite advances in technology and trained health workers, high mortality rates are reported among patients receiving haemodialysis therapy in lower-income countries. The factors influencing survival are not well understood due to data scarcity in Tanzania. This study was conducted to determine factors associated with survival among patients with ESKD undergoing HD in Tanzania.

Materials and methods

Population, setting, and study design

A retrospective cohort study was conducted to examine factors associated with three-year survival among patients with ESKD undergoing HD from January 2016 to December 2021 at Muhimbili National Hospital (MNH) in the hemodialysis unit in Tanzania. MNH is a national referral hospital and the teaching hospital for Muhimbili University of Health and Allied Sciences, located in Ilala Municipal, Dar es Salaam, Tanzania. It was the first government hospital to initiate HD treatment in Tanzania, with its effectiveness in May 2011 when the first patient was dialyzed. The unit has hemodialysis machines situated at the main HD unit, the medical ICU, and the obstetric ICU. It is staffed with nephrologists, nephrology nurses, medical specialists, dialysis nurses, medical officers, nursing officers, biomedical engineers, pharmacists, social workers, nutritionists, cashiers, medical recorders, and health attendants. On average, 120 patients with ESKD on maintenance hemodialysis are dialyzed daily in the unit, and dialyzers are single-use (not-reused) as per unit policy.

Sample size calculation

Sample size estimation was calculated according to [22] as follows;

Whereby,

n = sample size.

C = 7.85 (at 95% CI and power of 80%, Z_α/2 = Z_0.05/2 = 1.96 at type 1 error 5%).

P1-P2 = Difference in proportion of events between two groups.

Prevalence or probability in the case group = P1.

Prevalence or probability in the control group = P2.

P1 = 89% (0.89) survival prevalence of ESKD patients in HD on AVF access line.

P2 = 78% (0.78) survival prevalence of ESKD patients in HD on CVC access line (29).

Considering a 10% dropout of study participants.

Therefore, the estimated sample size for this retrospective cohort study is 386 participants.

Eligibility criteria

We included patients aged 18 years and older with ESKD who underwent hemodialysis at the MNH dialysis unit from January 2016 to December 2021. Patients were excluded if they were pregnant, aged younger than 18 years, or underwent hemodialysis for acute indications such as AKI, poisoning, overdose, or other reversible causes rather than maintenance hemodialysis for ESKD.

Sampling technique

Systematic sampling was used to recruit 386 participants who met the eligibility criteria of the study from January 2016 to December 2021. All participants were selected based on the Kth interval (3) from the larger population (N = 1326), with a random starting point (2). The Kth interval was calculated by dividing the population size (N = 1326) by the targeted sample size (n = 386). Ethical clearance to conduct this research, including the waiver of informed consent, was obtained from the MUHAS Institutional Review Board (Ref. No. DA 282/298/01.C/1130). The permit to conduct this study was obtained from the MNH administration (Ref. No.: MHN/TRCU/Perm/2022/157). Codes were used instead of names to ensure the confidentiality and privacy of participants.

Data collection

Data were retrospectively collected from the medical records for patients who had started maintenance hemodialysis at least one month before the study period. A pretested checklist was used to gather information on demographic characteristics such as age (grouped into 18–35, 36–55, and ≥ 56 years), age at initial HD, gender, marital status, residential area, education level, and occupation, as well as sociological factors like health insurance coverage and treatment payment mode (cost-sharing and Intramural Private Practice at Muhimbili (IPPM) lifestyle.

We also extracted clinical factors such as the primary cause of ESKD, presence of co-morbidity diseases, type of vascular access line (temporary dialysis catheters or tunnelled cuffed catheters), number of dialysis sessions per week, hours on single HD session, and laboratory findings, Blood pressure, ascites, erythropoietin (EPO) treatment, iron sucrose, pedal edema, and number of blood transfusions (BTs). The laboratory information such as potassium (normal range, 3.5-5mmol/L), sodium (normal range, 135–147 mmol/L), white blood cell (WBC) count [normal range, 4–10 K/µL]), platelets (normal range, 150–450 K/µL) and red blood cell (RBC) [3.8–4.8 K/µL] were categorized into the normal and abnormal range. The term abnormal range refers to biochemical parameters not within the respective normal ranges. These references were obtained from the Muhimbili National Hospital, laboratory services department. Blood pressure, systolic blood pressure (SBP) was categorized into < 120, 120–139, 140–159, and ≥ 160 mmHg; and diastolic blood pressure (DBP) was categorized into < 60, 60–69, 70–79, and ≥ 80 mmHg. BP was recorded before the beginning of HD.

Relevant information obtained from demographic, sociological, and clinical factors was assessed to determine whether they were associated with survival among patients with ESKD on HD. Data collected were cross-checked carefully for error reduction to maintain completeness and internal consistency. Comprehensive training of assistant researchers was conducted to enable reliability and proper data collection. Confidentiality and privacy were maintained at the maximum level.

Data analysis

Data entry and analysis were conducted using Stata version 15 with specifications (Stata Corp. 2017. Stata Statistical Software: Release 15. College Station, TX: Stata Corp LLC). In the descriptive analysis, categorical variables were summarized using frequency and percentage, while the numeric variables were summarized using mean and standard deviation (SD). The histogram was used to check the normality of the variables.

The Kaplan-Meier method was used to determine the survival rate of the patients, and the Log-rank test was used to compare the survival probability between categorical variables. The variables that showed statistically significant differences at p-value < 0.05 in the log-rank test were entered into the univariable Cox proportional hazard model.

Finally, the Cox proportional hazards model was used to estimate the demographic, sociological, and clinical factors associated with survival among ESKD patients undergoing HD. A Multivariate Cox proportional hazards model was used to control all possible confounders by taking into the model all factors, which revealed statistically significant associations with survival in the univariate model. All analyses were two-tailed tests, and the significance level was set at 5%.

Results

Demographic and sociological characteristics

A total of 386 participants were recruited in this study, out of which 188 (48.7%) had died, and 198 (51.3%) were alive or censored. The mean age of participants was 50.7 ± 15.9 years, with 247 (64%) being males. The mean age at the start of HD was 49.5 ± 16.1 years. Most participants had health insurance (63.7%), reported a healthy lifestyle, such as no history of alcohol use, smoking, or traditional herbal remedies (81%), and 48.7% had primary education, with 43.8% unemployed. Detailed demographic and sociological characteristics are presented in Table 1.

Table 1.

Demographic and sociological characteristics of the study population (N = 386)

	Frequency (n)	Percentage/Mean ± SD
Age (Year)	386	50.73 ± 15.87
Range (years)		18–85
Age categories
18–35	74	19.17
36–55	143	37.05
≥ 56	169	43.78
Gender
Female	139	36.01
Male	247	63.99
Age at initiation of hemodialysis	386	49.52 ± 16.13
Range		18–75
Marital status
Single	81	20.98
Married/cohabiting	284	73.58
Widow/ Divorced/Separated	21	5.44
Resident
Dar es Salaam	220	56.99
Others	166	43.01
Education level
College/University	78	20.21
No formal education	27	6.99
Primary	188	48.70
Secondary	93	24.09
Occupation
Employed	82	21.24
Retired	55	14.25
Self employed	80	20.73
Unemployed	169	43.78
Outcomes
Died	188	48.70
Alive/ loss to follow-up*	198	51.30
Model of payment for treatment
Cost sharing	81	20.98
Health insurance	246	63.73
IPPM	24	6.22
Public	35	9.07
Lifestyle practice
History of alcohol	50	12.95
History of smoking	9	2.33
History of using traditional herbs	11	2.85
None	316	81.87

Note: * indicates patients who were censored; SD, Standard deviation; IPPM, Intramural Private Practice at Muhimbili

Clinical characteristics

Table 2 shows the clinical characteristics of the study participants (N = 386). The most common primary causes of ESKD were comorbid illness, accounting for 93.3%, with the remaining 6.74% attributed to primary kidney pathologies. Among the comorbid conditions contributing to ESKD, HTN predominantly accounts for 74.4%, followed by DM (14.3%). 227 (58.8%) participants used vascular catheters for dialysis, 236 participants (61.1%) had ≥ 3 dialysis sessions per week, and 354 participants (91.7%) utilized 4 h on a dialysis machine for a single HD session. Participants with SBP ≥ 160 mmHg were 117 (30.3%), and those with DBP ≥ 80 mmHg were 244 (63.2%). We observed that participants who received EPO and iron sucrose treatment were 305 (79.0%) and 134 (34.7%), respectively. Meanwhile, those who received one or more units of blood transfusion were 235 (60.9%). Ascites and pedal edema were observed in 64 (16.6%) and 108 (72.0%) patients, respectively.

Table 2.

Clinical characteristics of the study participants (N = 386)

Primary causes of ESKD	Frequency (n)	Percentage/mean ± SD
Primary kidney pathology	26	6.74
Comorbid illness	360	93.26
Comorbid contributing illness
Hypertension	287	74.35
Diabetic mellitus	55	14.25
Cancer	9	2.33
HIV/AIDS	16	4.15
Hepatitis	3	0.78
Unknown	16	4.15
Type of vascular access
Fistula	159	41.19
Catheter	227	58.81
Number of dialysis sessions per week
< 3	150	38.86
≥ 3	236	61.14
Hours on a single HD
3	32	8.29
4	354	91.71
Systolic blood pressure (SBP) mmHg	386	148.30 ± 30.50
Range		79–241
SBP
< 120	63	16.32
120–139	101	26.17
140–159	105	27.20
≥ 160	117	30.31
Diastolic blood pressure (DBP) mmHg	386	87.20 ± 21.63
Range		25–187
DBP
< 60	26	6.74
60–69	40	10.36
70–79	76	19.69
≥ 80	244	63.21
Ascites
Yes	64	16.58
No	322	83.42
EPO treatment given
Yes	305	79.02
No	81	20.98
Iron sucrose treatment
Yes	134	34.72
No	252	65.28
Pedal edema
Yes	108	72.02
No	278	27.98
Number of blood transfusions (BTs)
0	151	39.12
1	93	24.09
2	57	14.77
3	53	13.73
≥ 4	32	8.29

Note: SD, Standard deviation; ESKD, End-Stage Kidney Disease; HD, Hemodialysis; EPO, Erythropoietin. Abnormal*, indicates low or high parameters. Primary kidney pathology refers to conditions that directly originate in the kidneys, causing ESKD. Comorbid contributing illness refers to a systemic condition not directly originating from the kidneys that may have contributed to ESKD

Most of the participants’ biochemical parameters were outside the normal ranges. Particularly, BUN, creatinine, and RBC were outside the normal range. Participants with HIV, Hepatitis B, and Hepatitis C were 9.1%, 8.3%, and 4.2%, respectively (Table 3).

Table 3.

Laboratory characteristics of the study participants (N = 386)

HIV status	Frequency (n)	Percentage/ IQR
Negative	351	90.93
Positive	35	9.07
Hepatitis B (HBsAg)
Negative	354	8.29
Positive	32	91.71
Hepatitis C (HCV)
Negative	370	95.85
Positive	16	4.15
Potassium (mmol/L)
Normal (3.5-5mmol/L)	200	51.81
Abnormal*	186	48.19
Sodium (mmol/L)
Normal (135–147)	187	48.45
Abnormal*	199	51.55
WBC Median (IQR) (K/µL)	386	6.87 (6.7,7.25)
Range		0.13–147
WBC count category
Normal (4–10)	251	65.03
Abnormal*	135	34.97
HB level (g/dL)
Anemic (Hb < 12 for females and < 13 for males)	360	93.26
Non anemic	26	6.74
Platelet median (K/µL)	386	224.5 (215, 234.55)
Range	386	18.5–2869
Platelets category
Normal (150–450)	293	75.91
Abnormal*	93	24.09
RBC median (K/µL)	386	3.01(2.9, 3.12)
Range	386	1.07–16.3
RBC category
Normal (3.8–4.8)	60	15.54
Abnormal*	326	84.46

Note: IQR, interquartile range; HB, hemoglobin; WBC, white blood cell; Abnormal*; indicates low or high parameters

Kaplan-Meier findings

Kaplan-Meier survival curves were generated to provide pivotal information on the shape of the survival of ESKD patients who underwent HD in the cohort study. The results showed that there was a steady decrease in survival function during the study period. The survival time for 25% of the population was less than a year (0.6 year), and the median survival (50%) of the entire population (N = 386) was less than 2.5 years (Fig. 1, Table S1). The survival estimates for 1-year, 2-year, and 3-year were 66%, 53%, and 47%, respectively (Table S1). These results indicated that the survival rates of these ESKD patients who underwent HD decreased with time.

Fig. 1.

Survival estimate of the ESKD patients on HD during the study period (Median survival time was 2.49 years)

Figure 2 shows the Kaplan-Meier curve for survival estimates of demographic characteristics. Comparing survival by age groups, the estimated survival probability showed that younger adults (18–35 years) had the highest survival compared to other age groups, and older adults had the lowest survival (Fig. 2A). Comparing survival by gender, the curve showed that male patients had a higher estimated survival probability than females (Fig. 2B). Comparing survival by education level, the curve showed that patients with college /university education levels had the highest survival, while those with primary education had the lowest survival (Fig. 2C). The Kaplan Meier curve for survival estimated by occupation showed that employed patients had the highest survival compared to unemployed patients (Fig. 2D).

Fig. 2.

Kaplan-Meier curve for survival estimates demographic characteristics. (A) Age group. (B) Gender. (C) Education level. (D) Occupation

The Kaplan Meier curve for survival estimated by a model of payment for treatment showed that patients with health insurance had higher estimated survival, and those with a public mode of payment had the lowest survival (Fig. 3).

Fig. 3.

Survival estimates by a model of payment for treatment

Figure 4 shows the Kaplan-Meier curve for survival estimates by baseline clinical characteristics. Comparing the survival by the primary causes of ESKD, the curve showed that patients with ESKD due to renal causes had higher estimated survival compared to those whose ESKD was due to other causes (comorbidities) (Fig. 4A). Patients with fistula vascular access had higher estimated survival probabilities compared to those with catheters (Fig. 4B). Patients who had dialysis sessions ≥ 3 per week had higher estimated survival compared to those with < 3 Dialysis sessions per week (Fig. 4C), and patients who utilized 4 h in a single HD had higher estimated survival probabilities compared to those who utilized only 3 h (Fig. 4D).

Fig. 4.

Kaplan-Meier curve for survival estimates by baseline clinical characteristics. (A) Primary causes of ESKD. (B) Type of vascular access. (C) Number of dialysis sessions per week. (D) Hours utilized in a single HD session

Figure 5 shows the Kaplan-Meier curve for survival estimates by laboratory parameters. Comparing survival by Hepatitis C status, the Kaplan Meier survival estimate curve showed that patients who had negative Hepatitis C had higher estimated survival probabilities compared to those with positive status (Fig. 5A) and those with normal ranges of WBC counts, platelets, and HB had higher estimated survival probabilities compared to those with abnormal ranges (Fig. 5B, C, and D).

Fig. 5.

Kaplan-Meier curve for survival estimates by laboratory parameters. (A) Hepatitis C status, (B) WBC count categories. (C) Platelet categories. (D) HB status

Log-rank test

We performed a Log-rank test to compare the survival probability between the two groups by considering the demographic, sociological, and clinical characteristics. Table S2 shows the log-rank test for equality of survivor functions for demographic and sociological factors (N = 386). We found that there were significant differences in survival probabilities among the different age groups (ꭓ² =6.46, P = 0.040), residential areas (ꭓ² =4.43, P = 0.035), level of education (ꭓ² = 27.57, P = 0.000), occupations (ꭓ² =20.99, P = 0.000), and model of payment for treatment (ꭓ² =18.80, P = 0.000). The demographic and sociological factors that revealed statistically significant differences were considered predictors and entered into the univariate Cox proportional hazard model, except gender and marital status, which were not statistically significant but were included in the model due to prior knowledge that these variables may be predictors. The residential area was not included in the model because of various reasons, such as it is well known that most people claim their residential area is in Dar es Salaam, while not.

Table S3 shows the log-rank test for the equality of survival functions for clinical factors. The log-rank test showed that the survival probabilities were statistically significantly different on the primary causes of ESKD (ꭓ² =6.27, P = 0.012), type of vascular access line (ꭓ² =5.96, P = 0.015), and number of dialysis sessions per week (ꭓ² =4.09, P = 0.043). Laboratory characteristics that revealed statistical differences were Hepatitis C status (ꭓ² =7.37, P = 0.007), WBC count category (ꭓ² =8.41, P = 0.004), platelets category (ꭓ² =8.08, P = 0,006), SBP categories (ꭓ² =10.13, P = 0.018), and DBP category (ꭓ² =9.52, P = 0.023) (Table S4).

The clinical and laboratory factors that revealed statistically significant differences were considered as predictors and entered into the univariable Cox proportional hazard model, except HB levels, which were not statistically significant but were included in the model due to prior knowledge that this variable may be a predictor.

Cox proportional hazard ratios

We performed univariate and multivariate Cox proportional hazard ratios to identify whether demographic, social, and clinical factors were associated with survival outcomes. The variables that were found to be statistically significant in the univariate model were entered into the multivariate model to control for all possible confounders. Table 4 shows factors associated with survival among ESKD patients with HD (N = 386). In the univariate model, patients in age groups 36–55 and > 56 years had a higher risk of mortality compared to those in age group 18–35 years (HR = 1.63, 95% CI: 1.03, 2.57; P = 0.036) and (HR = 1.75, 95% CI: 1.13, 2.73; P = 0.013), respectively. Nevertheless, in the multivariate model, only patients in the age group ≥ 56 years were found to have a higher risk of mortality compared with the reference age group 18–35 years (HR = 2.19, 95% CI: 1.35, 3.53; P = 0.001).

Table 4.

Factors associated with survival among ESKD patients on HD (N = 386)

	Univariable		Multivariable
	HR (95%CI)	P-value	HR (95%CI)	P-value
Age categories
18–35	1		1
36–55	1.63(1.03, 2.57)	0.036	1.75(1.09, 2.82)	0.021
> 56	1.75(1.13, 2.73)	0.013	2.19(1.35, 3.53)	0.001
Gender			-
Female	1
Male	0.79(0.59, 1.05)	0.106
Marital status			-
Married/cohabiting	1
Widow/ Divorced/Separated	1.05(0.57, 1.95)	0.866
Single	0.69(0.46, 1.02)	0.060
Education level
College/University	1		1
No formal education	1.38(0.63, 3.01)	0.421	1.00(0.42, 2.36)	0.999
Primary	3.04(1.91, 4.85)	0.000	2.02(1.15, 3.54)	0.014
Secondary	2.22(1.32, 3.73)	0.003	1.81(1.03, 3.18)	0.039
Occupation
Employed	1		1
Retired	1.30(0.73, 2.31)	0.374	0.86(0.46, 1.61)	0.634
Self employed	2.09(1.28, 3.40)	0.003	1.35(0.79, 2.29)	0.268
Unemployed	2.43 (1.57, 3.75)	0.000	1.83(1.10, 3.03)	0.019
Primary causes of ESKD
Comorbid illness	1		1
Primary kidney pathology	0.37(0.16, 0.83)	0.016	0.42(0.18, 0.97)	0.041
Type of vascular access
Fistula	1		1
Catheters	1.45(1.07, 1.95)	0.015	1.44(1.05, 1.96)	0.023
Number of dialysis sessions per week
< 3	1		1
≥ 3	0.74(0.56, 0.99)	0.044	1.03(0.75, 1.42)	0.847
HCV
Negative	1		1
Positive	2.27(1.23, 4.19)	0.008	1.75(0.87, 3.49)	0.114
WBC category
Normal	1		1
Abnormal	1.53 (1.14, 2.04)	0.004	1.38(1.02, 1.86)	0.038
HB level
Anemic	1		-
Non anemic	1.54(0.97, 2.45)	0.067
Platelets category
Normal	1		1
Abnormal	1.57(1.15, 2.15)	0.005	1.79(1.29, 2.50)	0.001
Systolic blood pressure (SBP)
< 120	1		1
120–139	1.084(0.71, 1.66)	0.711	1.48(0.91, 2.42)	0.114
140–159	0.59(0.37, 0.94)	0.026	0.74(0.44, 1.24)	0.252
≥ 159	0.79(0.51, 1.21)	0.280	1.15(0.68, 1.94)	0.611
Diastolic blood pressure (DBP)
< 60	1		1
60–69	1.35(0.69, 2.64)	0.378	1.01(0.49, 2.09)	0.976
70–79	1.13(0.61, 2.11)	0.695	1.07(0.53, 2.14)	0.857
≥ 80	0.76(0.43, 1.36)	0.360	0.80(0.41, 1.60)	0.534
Model of payment for treatment
Cost sharing	1		1
Health insurance	0.49(0.35, 0.68)	0.000	0.58(0.40, 0.83)	0.003
IPPM	0.611(0.33, 1.14)	0.123	0.79(0.41, 1.51)	0.469
Public	0.70(0.41, 1.19)	0.179	0.80(0.46, 1.41)	0.443

Note: -, Indicate not entered in the multivariable model; HR, Hazard ratio; CI, Confidence interval; ESKD, End-Stage Kidney Disease; HD, Hemodialysis; IPPM, Intramural Private Practice at Muhimbili. Univariable and multivariable Cox proportional hazard regression was computed. Hazard ratios (HRs) were calculated and compared with each factor. P-value < 0.05

Compared with college/ university level of education, patients with primary and secondary education had a higher risk of mortality in both the univariate and multivariable models (adjusted HR = 2.02, 95% CI: 1.15, 3.54; P = 0.014) and (adjusted HR = 1.81, 95% CI: 1.03, 3.18, P = 0.039), respectively.

Patients who were self-employed and unemployed had a higher risk of mortality compared with those who were employed (HR = 2.09, 95% CI: 1.28, 3.40; P = 0.003) and (HR = 2.43, 95% CI: 1.57, 3.75), respectively. In the multivariable model, only unemployed patients remained with a higher risk of mortality (HR = 1.83, 95% CI: 1.10, 3.03, P = 0.019).

Compared to comorbid illness, patients whose ESKD was due to primary renal pathology had a lower risk of mortality in both the univariate and multivariable models (adjusted HR = 0.42, 95%CI: 0.18, 0.97; P = 0.041).

Patients who had catheter vascular access had a higher risk of mortality compared to those with fistula vascular access in both univariate and multivariable models (adjusted HR = 1.44, 95% CI: 1.05, 0.97; P = 0.023). Patients whose WBC count and platelets were not in normal ranges had a higher risk of mortality compared to those in normal ranges in both the univariate and multivariable models (adjusted HR = 1.38, 95%CI: 1.02, 1.86; P = 0.038) and (adjusted HR = 1.79, 95% CI: 1.29, 2.50; P = 0.001), respectively.

Compared with the cost-sharing model of payment for treatment, patients with Health insurance had a lower risk of death in both the univariate and multivariable models (adjusted HR = 0.58, 95% CI: 0.40,0.83; P = 0.003).

Patients who performed three or above dialysis sessions per week had a lower risk of mortality compared to those with fewer than three dialysis sessions per week (HR = 0.74, 95% CI: 0.56, 0.99; P = 0.847). However, these associations were not observed in the multivariable model (HR = 1.03, 95% CI: 0.75, 1.42; P = 0.847).

Patients who were Hepatitis C positive had a higher risk of mortality compared to those with Hepatitis C negative (HR = 2.27, 95%CI: 1.23, 4.19; P = 0,008). However, this association was not present in the multivariable model (HR = 1.75, 95%CI: 0.87, 3.49; P = 0.114).

Compared with patients with SBP < 120 mmHg, those with SBP 140–159 mmHg had a lower risk of mortality (HR = 0.59, 95% CI: 0.37, 0.94; p = 0.026). However, this association was not present in the multivariable model (HR = 0.74, 95%CI: 0.44, 1.24; P = 0.252).

Moreover, we did not find enough evidence to pronounce the associations between gender, marital status, HB level, and DBP with survival in ESKD patients on HD.

Discussion

This study was conducted at MNH to determine three-year survival among patients receiving maintenance haemodialysis. The survival rates decline with time, and the factors that influenced higher survival rates were younger age, college or university education, employment, health insurance, fistula vascular access, and normal range WBC count and platelet count. We found that, among 386 patients with ESKD, 48.7% died during the follow-up period, and 51.3% of patients were alive or censored. The median survival time was less than 2.5 years.

In our study, the survival rates for 1, 2, and 3 years were 66%, 53%, and 47%, respectively. Previous studies have reported varying survival rates within a specific time frame. A study conducted in Canada showed that the survival rates for 1, 2, and 3 years were 80%, 65%, and 54%, respectively [23]. Teixeira et al. [24] reported survival rates of 84.7%, 81.2%, and 78.7% at 1, 2, and 3 years, respectively. Mousavi et al. [25] showed that the survival rates for 1, 2, and 3 years were 63%, 53%, and 38%, respectively. The differences in our findings and previous studies could be due to variations in study designs, populations, clinical practices, and patient characteristics across different centers and countries.

We found that older patients had a lower survival rate compared to younger patients. In agreement with our findings, previous studies have reported that an increase in age is associated with a decline in survival [19–21]. This can be explained by the fact that elderly patients on hemodialysis are subjected to functional and cognitive impairments [26]. Functional impairment contributes to a disability, multiple readmissions, and a decreased survival rate [27, 28], whereas cognitive impairment is positively linked to mortality [26].

Our study demonstrated that patients with college or university education, employed, and those with health insurance had higher survival rates compared to uneducated, unemployed, and those without health insurance. Generally, education substantially affects survival rates and health insurance coverage. Higher education levels are related to improved health outcomes, including lower mortality rates attributable to better access to healthcare. In addition, education is connected to better employment opportunities and higher income, which can further influence health insurance coverage and access to necessary medical services [29–31]. In agreement with our findings, previous research reported that good literacy is related to having a healthy lifestyle, early diagnosis and good management of acute and chronic diseases, good long-term treatment and compliance, as well as lower prevalence of risk factors [32, 33]. Marinovich et al. [19] found that low or no income was an independent risk factor for immediate lower survival among hemodialysis patients. Furthermore, a study conducted in Dodoma, Tanzania, from January 2013 to June 2015, reported that out of 116 patients on HD, 77 patients (66.4%) were not on health insurance and had poor outcomes [10].

Studies have shown that diabetes mellitus and hypertension (HTN) are the primary causes of ESKD [10, 21]. In our study, HTN was the most prevalent, followed by diabetes mellitus. Additionally, patients whose ESKD was due to primary kidney pathology causes had higher survival compared to those with comorbid illness (HTN, diabetes mellitus). Likewise, several studies have reported lower survival among diabetic and hypertensive patients on dialysis [16, 18, 19, 21].

We found that patients with catheter vascular access had a higher risk of mortality compared to those with fistula vascular access, similar to the studies conducted in Taiwan and Ethiopia [34, 35]. This can be explained by the fact that vascular access type has been reported as among the factors for the short-term and long-term survival of HD patients [34], because a faster and more efficient AVF access provides a better clearance compared to the unstable one, hence better therapeutic outcomes [35]. In addition, studies have reported that native AVF has fewer complications of infection and longer durability than other types of vascular access [36, 37].

Patients with WBC counts not in the normal range had a lower survival rate compared to those with normal WBC counts. Likewise, a study involving 959 patients on maintenance HD has observed that WBC count was a robust predictor of 1-year mortality from all causes, including infection and cardiovascular diseases [38]. In addition, a retrospective single-center analysis study found that WBC count had a long-term effect on the survival of patients with peritoneal dialysis [39], which could probably be similar to HD patients.

Patients with platelets not in the normal range had a higher risk of mortality compared to those with normal range. Indeed, CKD patients have a substantially higher risk of cardiovascular, thromboembolic, and hemorrhage problems, and half of the patients with advanced CKD eventually die due to cardiovascular disease. Since platelets are central to hemostasis, changes in platelet function could lead to platelet hyper- or hypo-reactivity, which may, in turn into thrombotic or hemorrhagic problems [40].

Treatment time is a vital prescription component of dialysis therapy and has an influence on the safe removal of solutes and accumulated excess fluid in the body; longer dialysis sessions provide benefits and reduce the risk of complications [41]. In our study, patients who performed three or more dialysis sessions per week had lower risks of mortality compared to those with fewer than three dialysis sessions per week. However, these associations were not observed in the multivariable model, indicating insufficient evidence for the association between the number of dialysis sessions and survival. In agreement with our findings, a study conducted in India reported no difference in survival and hospitalization rates among patients on twice or thrice weekly hemodialysis at one year [42]. In addition, a systematic review reported inconsistend survival data in different studies, with 53% studies documented no survival differences, 24% worse survival, and 23% reported improved survival between twice weekely and thric weekely groups [43]. Overall, our findings should be interpreted with caution because of the retrospective design of our study; we could not accurately identify the predominant vascular access type, particularly for patients who underwent dialysis twice or thrice-weekly dialysis. In addition, detailed measurements of residual renal function were not available for the majority of patients.

Studies have reported that normocytic normochromic anemia develops due to decreased renal production of erythropoietin (EPO) [44]. Despite the predominant use of recombinant human erythropoietin (rhEPO), anemia is a frequent finding in hemodialysis, which significantly impairs quality of life. As a result, anemia may be predictive of an increased risk of mortality in some hemodialysis patients [45]. In addition, another study found that anemia reduces survival in patients with kidney disease, heart failure, or both [14, 15]. However, in the present study, we did not find enough evidence to pronounce the associations between HB level (anemic or non-anemic) and survival in patients with ESKD on HD.

The strengths of this study include: To the best of our knowledge, this is the first study to assess survival rate and associated factors conducted at MHN-Upanga, Dar es Salaam, Tanzania. The findings obtained from the current study contribute to the body of literature on survival status and associated factors of patients with ESKD undergoing HD in Tanzania. We used robust statistical methods, including Kaplan-Meier and Cox proportional hazard ratios, to account for censoring and provide accurate estimates for survival probabilities. Several variables and the survival rate were compared in different categories to give an idea of whether survival functions are approximately parallel.

This study has some limitations as follows: Firstly, its retrospective design does not allow assessing causality; instead, we assessed the association between the predictors and survival outcomes. Secondly, our study is susceptible to residual confounding despite extensive adjustment for known confounders in the multivariate analysis. Important variables such as hospitalization rate, length of stay, ICU admission, residual kidney function, etc., were not measured due to either, nature of the study, a retrospective cohort study design. Thirdly, retrospective studies are subject to selection bias. We addressed this by generating computerized general numbers for study selection. Fourthly, single-centre extraction of data limits the generalizability of the results. Finally, regardless of such limitations mentioned, our study revealed the survival rate and associated factors of patients with ESKD who underwent HD in Tanzania.

Conclusions

The results showed that the survival rates for 1-, 2-, and 3-year were 66%, 53%, and 47%, respectively. Younger age, college or university education, employment, health insurance, fistula vascular access, and normal range WBC count and platelet count were factors associated with higher survival rates. Financing for haemodialysis was noted to be one of the contributing factors to poor survival; therefore, the government and other stakeholders need to consider mitigations to ensure equitable access to kidney replacement therapy including haemodialysis.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviation

AVF

Arteriovenous Fistula

BUN

Blood urea

CI

Confidence interval

CKD

Chronic Kidney Disease

CVC

Central venous catheter

CVD

Cardiovascular Disease

DBP

Diastolic blood pressure

DM

Diabetic Mellitus

EPO

Erythropoietin

ESKD

End-Stage Kidney Disease

GFR

Glomerular Filtration Rate

HB

Hemoglobin

HD

Hemodialysis

HIV

Human Immunodeficiency Virus

HR

Hazard ratio

HTN

Hypertension

IPPM

Intramural Private Practice at Muhimbili

IRB

Institute review board

KT

Kidney Transplantation

MNH

Muhimbili National Hospital

MUHAS

Muhimbili University of Health and Allied Sciences

NHIF

National Health Insurance Fund

NPC

New Pediatric Complex

PD

Peritoneal Dialysis

RBC

Red blood cell

KRT

Kidney Replacement Therapy

SBP

Systolic blood pressure

URR

Urea Reduction Ratio

WBC

White blood cell

Author contributions

SMM: Conception and study design, data collection and analysis, writing the manuscript first draft, and revising and editing. RH and JS: Conception and study design, revising and editing the manuscript, and supervision. FMB: Conception and study design, help with data analysis, and revise the manuscript. FFF. Conception and study design, revising the manuscript for intellectual content, project administration, and supervision. All authors have approved the final manuscript.

Funding

No funding was received for this study.

Data availability

The datasets used and/ or analyzed during this study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

MUHAS Institution Review Committee approved this study. MUHAS IRB also granted a waiver of consent because this study was based on reviewing medical records without the patient’s intervention (Ref. No. DA 282/298/01.C/1130). In addition, MNH granted permission to carry out this study at the site (Ref. No.: MHN/TRCU/Perm/2022/157). This study was conducted in accordance with the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.