Correlates of symptom burden in renal dialysis patients: a systematic review and meta-analysis

Abstract

Background

Patients receiving renal dialysis often experience a wide range of symptoms. These symptoms contribute to a significant symptom burden that significantly affects patients’ quality of life and serves as a significant predictor of healthcare resource utilization and patient prognosis. It is necessary to synthesize existing evidence to draw reliable conclusions to deepen the understanding of symptom burden.

Objective

A systematic review and meta-analysis were conducted to identify the relevant factors of symptom burden in patients receiving renal dialysis.

Methods

The systematic review and meta-analysis was conducted by searching nine databases for studies reporting the correlates between symptom burden and demographic variables, disease factors, and psychosocial factors from inception to 24 June 2024. After two researchers independently conducted literature search, data extraction, and quality evaluation, meta-analysis was conducted using R Language and Stata 15.1 Software. This study has been registered in the PROSPERO.

Results

Sixty-two studies were included in this review. Results showed that the symptom burden of renal dialysis patients was positively correlated with age, gender, working status, medical cost, dialysis age, quality of sleep, nutritional status, comorbidities, depression, anxiety, disease uncertain, avoidance coping and resignation coping, and negatively correlated with marital status, income, serum sodium, quality of life, social support, subjective well-being, and self-management ability.

Conclusions

Our findings reveal that many factors, including demographic, disease-related, and psychosocial variables, affect symptom burden. The results can supply information for health promotion and relief symptom burden for patients receiving renal dialysis.

Registered number: CRD42024507577

1. Introduction

Chronic kidney disease (CKD) is a syndrome characterized by chronic kidney damage resulting from various factors. This condition ultimately leads to a range of clinical symptoms, including imbalances in water, electrolytes, acid-base balance, and metabolic disorders [1]. It poses a significant global health challenge due to its high prevalence and associated risks of end-stage renal disease, cardiovascular complications, and premature death. Projections suggest CKD will rank as the fifth leading cause of death worldwide by 2040 [2]. Currently, about 10% of the global population has CKD, with 6 million individuals reaching end-stage renal disease and 1.5 million starting renal replacement therapy annually [3], expected to rise to 5,439,000 by 2030 [4]. Treatment options include hemodialysis (HD), peritoneal dialysis (PD), and renal transplantation [5], with renal dialysis being the primary treatment due to limited kidney donors [5–8].

Studies have reported that the 5-year survival rate for patients undergoing maintenance hemodialysis (MHD) is 46.8% in the United States [9], 38.3% in Europe [10] and 60.6%∼85.3% in China [11]. Despite the life-extending benefits of dialysis treatment, patients often endure various symptoms during dialysis, such as itching, nausea, vomiting, sleep disturbances, and depression. These symptoms create a significant symptom burden that can affect disease progression and eventual outcomes [12].

Researchers initially focused on studying individual symptoms’ effects on patients’ well-being. However, clinicians noted that the accumulation of multiple symptoms, such as fatigue, sleep disorders, dry skin, and depression, significantly burdens patients. This observation led to the conceptualization of ‘symptom burden.’ In 1999, Desbiens [13] introduced the concept at the Marshfield Clinical Medical Research Foundation in the United States. Cleeland in 2007 described it as the combined severity and impact of patient-reported symptoms [14]. In 2008, Davis [15] from the Cleveland Center for Palliative Medicine in the USA refined the concept, defining symptom burden as the discomfort patients experience due to changing disease processes and treatment dynamics, encompassing both psychological and physiological burdens.

The results of studies on factors related to dialysis symptom burden have not been systematically reviewed. Considering the heterogeneity of the studies, it is necessary to synthesize existing evidence to draw reliable conclusions to deepen the understanding of symptom burden in renal dialysis patients. Therefore, this study conducted a meta-analysis of the factors associated with symptom burden in patients receiving renal dialysis.

2. Methods

2.1. Study design

This meta-analysis was conducted following the Preferred Reporting Items for Systematic Evaluation and Meta-Analyses (PRISMA) guidelines.

2.2. Search strategy

Nine databases were searched for all research investigating the correlates of symptom burden in renal dialysis patients published before 24 June 2024, including Web of Science, PubMed, Ovid, Embase, CINAHL Plus, CNKI, WANFANG DATA, VIP, and Scopus. To ensure a comprehensive search, we combined the Medical Subject Headings (MeSH) term with free terms. The searching strategy is described in Supplementary file.

2.3. Inclusion and exclusion criteria

Studies meeting the following criteria were included (1): Patients receiving renal dialysis (2); conducted on adult patients (aged ≥ 18 years) (3); reported values for correlations or effects of sociodemographic characteristics, disease factors and psychosocial variables on symptom burden (4); had an observational design, including cross sectional and longitudinal research (5); had clear tools for assessing symptom burden.

The following studies were excluded (1): conference reports, case studies, and editorials were excluded (2); studies that cannot be obtained in full text; and (3) studies that with inaccurate or incomplete data.

2.4. Data extraction and quality assessment

Data extraction and quality assessment were conducted independently by two researchers and any problems or discrepancies in the process were analyzed and resolved by a third researcher. We used the assessment criteria of the Agency for Healthcare Research and Quality (AHRQ) [16] to assess the quality of the included studies. The scoring levels for the scale are as follows: scores of 0–3 indicate low quality, 4–7 represent moderate quality, and 8–11 denote high quality [17].

2.5. Statistical analysis

Meta-analysis was conducted by R Language and Stata 15.1 Software and p < 0.05 indicates statistical significance. In this study, the Pearson’s correlation coefficient (r) was used as an effect value. If the values are not reported, the Spearman’s correlation coefficient (rs), odds ratios (OR), or standardized regression coefficient (β) need to be transformed to Pearson correlation coefficient (r). The Spearman’s correlation coefficient (rs) values were converted to Pearson’s correlation coefficient (r) using the formula: r = 2 × sin(rs × π/6), and odds ratios (OR) and standardized regression coefficients (β) can also transformed into Pearson correlation coefficients (r) based on the methodology described in a previous study [18–21]. We calculate the combined z-values using Pearson’s correlation coefficients (r) transformed by Fisher z Transform. When more than two studies were included in the meta-analysis, heterogeneity was assessed using the I²test. I² values of <25%, ∼50%, ∼75%, and ∼100% are considered as indicating mild, moderate, high, and very high heterogeneity [22]. A random-effects model was employed if between-study heterogeneity was statistically significant (I² > 50% or p < 0.05), while a fixed-effects model was used otherwise. Subgroup analysis was conducted to find the source of heterogeneity. Funnel plots and Egger tests were utilized to investigate potential publication bias if more than 10 studies were included. Sensitivity analyses were conducted to exclude individual studies one individually to assess the stability of the meta-analytic results (Supplementary file) [23].

3. Results

3.1. Study flow and characteristics

A total of 4185 articles were obtained after initial searching, and 62 studies were finally included (Figure 1), which included 12,967 patients, with samples sizes ranging from 59 to 700. Types of dialysis for the subjects included hemodialysis, peritoneal dialysis or mixed dialysis modalities. Regarding the geographic location of the studies, 45 studies were conducted in China; 3 in the United States; 3 in Turkey, 2 in Spain; 1 in Egypt; 2 in Korea; 1 in Indonesia; 1 in the United Kingdom; 1 in Denmark; 1 in Malaysia, 1 in Cameroon, and 1 in Australia. Symptom burden was measured using different assessment tools and among the included studies, the scores ranged from 5 to 10, all of which were moderate and high-quality literature, and these articles were of sufficient quality to participate in systematic reviews and meta-analyses (Table 1).

Figure 1.

Literature selection process and results.

Table 1.

Characteristics of the included studies.

Author/ year	Nation	Sample size	Study design	Dialysis type	Symptom burden scale	Correlates	Quality score
Weisbord, 2005 [24]	USA	162	Cross-sectional	HD	DSI	(51)	8
Weisbord, 2007 [25]	USA	160	Cross-sectional	HD	DSI	(12)(50)	8
Ibrahim, 2008 [26]	Egypt	60	Cross-sectional	HD	DSI	(26)(51)	6
Caplin, 2011 [27]	Britain	550	Cross-sectional	HD	other	(1)(2)(11) (47)	7
Shan, 2012 [28]	China	59	Cross-sectional	HD	other	(53)	6
Shi, 2012 [29]	China	133	Cross-sectional	HD	other	(53)	8
Gao, 2012 [30]	China	103	Cross-sectional	HD	DSI	(62)	7
Zhou, 2013 [31]	China	136	Cross-sectional	HD	DFSSBI	(1)(2)(4)(6)(7)(26)	6
Montilla, 2013 [32]	Spain	46	Cross-sectional	HD	ESSR	(52)	8
Wang, 2014 [33]	China	100	Cross-sectional	PD	PSDS	(1)(17)	8
Li, 2015 [34]	China	403	Cross-sectional	PD	DSI	(1)(17)(22)(24)(46)(50)	6
Seo, 2015 [35]	Korea	243	Cross-sectional	HD	DSI	(1)(10)(17)(55)	7
Wang, 2016 [36]	China	301	Cross-sectional	HD	DSI	(26)	6
Zhang, 2016 [38]	China	194	Cross-sectional	HD	DFSSBI	(1)(2)(4)(5)(6) (21)(25) (27)	7
Li, 2016 [39]	China	143	Cross-sectional	HD	DSI	(18)	7
Hao, 2016 [40]	China	186	Cross-sectional	HD	DSI	(1)(17)	9
Liu, 2016 [41]	China	118	Cross-sectional	HD	DFSSBI	(26)	7
Wang, 2016 [37]	China	301	Cross-sectional	HD	DSI	(18)(19)(50)(51)	6
He, 2017 [42]	China	72	Cross-sectional	PD	DSI	(18)(56)	6
Ma, 2017 [43]	China	149	Cross-sectional	HD	DSI	(54)	8
Zou, 2017 [44]	China	96	Cross-sectional	HD	DSI	(54)	8
Tang, 2017 [45]	China	310	Cross-sectional	HD	DSI	(19)	6
Liu, 2017 [46]	China	230	Cross-sectional	HD	DFSSBI	(21)(22)(23)(24)(25)(28)(29)(37)(40)(44)	7
Zou, 2017 [47]	China	76	Cross-sectional	HD	DSI	(50)(51)	9
Cao, 2017 [48]	China	273	Cross-sectional	HD	MST	(1)(3)	6
Luo, 2018 [49]	China	384	Cross-sectional	HD	DFSSBI	(54)	6
Chen, 2018 [50]	China	68	Cross-sectional	HD	DSI	(22)	9
Chen, 2018 [51]	China	295	Cross-sectional	HD	DFSSBI	(26)(56)	6
Kou, 2018 [52]	China	136	Cross-sectional	HD	other	(1)(2)(4)(5)(6)(7)(48)	6
Dong, 2018 [53]	China	143	Cross-sectional	HD	DSI	(50)	7
Li, 2019 [54]	China	120	Cross-sectional	HD	DSI	(18)	7
Ducharlet, 2019 [55]	Australia	102	Longitudinal-study	MIX	POS-S	(26)(69)	7
Sánchez, 2019 [56]	Spain	60	Cross-sectional	HD	POS-S	(49)	6
Li, 2019 [57]	China	398	Cross-sectional	HD	DFSSBI	(65)	7
Lu, 2020 [58]	China	136	Cross-sectional	HD	DSI	(1)(5)(6)(17)	7
Kim, 2020 [59]	Korea	120	Cross-sectional	HD	other	(50)(53)(68)	9
Zou, 2020 [60]	China	184	Cross-sectional	HD	DFSSBI	(30)(71)	7
Limbong, 2020 [61]	Indonesia	118	Cross-sectional	HD	CKD-SBI	(1)(2)(19)(23)(32)(47)(50)(57)(70)	8
Geng, 2021 [62]	China	96	Cross-sectional	HD	DFSSBI	(31)(38)(39)(70)	5
Kang, 2021 [63]	China	327	Cross-sectional	HD	DSI	(52)	6
Kang, 2021 [64]	China	92	Cross-sectional	HD	DSI	(52)	8
Gao, 2021 [65]	China	253	Cross-sectional	HD	DSI	(52)(58)	8
Liu, 2021 [66]	China	253	Cross-sectional	HD	DFSSBI	(1)(3)(4)(6)(20)(63)(64)	8
Geng, 2021 [67]	China	96	Cross-sectional	HD	DFSSBI	(55)(57)	7
Xiao, 2021 [68]	China	156	Cross-sectional	HD	DSI	(28)	6
Wu, 2021 [69]	China	271	Longitudinal-study	MIX	DSI	(1)(3)(4)(5)(6)(23)(24)(25)(38)	7
Ozen, 2021 [70]	Turkey	128	Cross-sectional	HD	DSI	(35)(36)	6
Song, 2021 [71]	China	382	Cross-sectional	HD	DSI	(17)(20)(56)(57) (58)	8
Zhang, 2021 [72]	China	309	Cross-sectional	HD	DSI	(26)(50)(66)	9
Zhao, 2022 [73]	China	160	Cross-sectional	HD	DSI	(28)	5
Zhang, 2022 [74]	China	160	Cross-sectional	HD	DSI	(1)(4)(6)(8)(20)(17)	8
Cukor, 2022 [75]	USA	100	Cross-sectional	HD	DSI	(50)(61)	8
Georges, 2022 [76]	Cameroon	181	Cross-sectional	HD	DSI	(1)(2)(4)(17)(19)(20)(29)(33)(72)	8
Gunarathne, 2022 [77]	Malaysia	118	Cross-sectional	HD	CKDSI-SL	(14)(15)(23)(24)(28)(37)(40)(57)(63)(74)	10
Kaplan, 2022 [78]	Turkey	596	Cross-sectional	HD	DSI	(14)(73)	8
Ding, 2023 [79]	China	98	Cross-sectional	HD	DSI	(52)	8
Hao, 2023 [80]	China	700	Cross-sectional	HD	DFSSBI	(59)	9
Wang, 2023 [81]	China	286	Cross-sectional	HD	DSI	(17)(19)(22) (34)	8
Hao, 2023 [82]	China	700	Cross-sectional	HD	DFSSBI	(1)(2)(4)(5)(6)(8)(9)(16)(37)	8
Zhang, 2023 [83]	China	160	Cross-sectional	HD	DSI	(60)	8
Karaaslan, 2023 [84]	Turkey	92	Cross-sectional	HD	DSI	(2)(3)(4)(5)(6)(21)(22)(23)(25)(27)(33)(36)(37)(39)(41)(42)(43)(45)(48)	9
Nielsen, 2023 [85]	Denmark	385	Cross-sectional	HD	DSI	(67)	8

Note: HD, Hemodialysis; PD, Peritoneal Dialysis; DSI, Dialysis Symptom Index; DFSSBI; CKD Symptom Burden Index (CKD-SBI); other, Self-made scale; ESS-R: Revised Somatic Symptom Scale; Physical Symptom Distress Scale (PSDS); Multidimensional Symptoms Tool; MST; (1) age; (2) gender; (3) marital status; (4) education; (5) employment status; (6) income; (7) cost; (8) residence; (9) living status; (10) family support; (11) race; (12) religion; (13) height; (14) weight; (15)BMI; (16) medical insurance; (17) Dialysis age; (18) quality of sleep; (19) nutritional status; (20) comorbidities; (21) IPTH; (22) serum phosphorus; (23) hemoglobin; (24) albumin; (25) Serum Calcium (26) quality of life; (27) Calcium-Phosphorus Product (28) Blood Pressure Variation rate (BPV); (29) Dialysis frequency; (30) Glomerular Filtration Rate(GFR); (31) Underlying disease; (32) Dialysis recovery time; (33) vascular access; (34) weak; (35) Urea reduction rate (URR); (36) urea excretion ratio; (37) serum creatinine; (38) serum phosphate; (39) serum sodium; (40) urea nitrogen; (41) C-reactive protein; (42) ferroprotein; (43) glucose; (44) Uric Acid; (45) serum potassium; (46) prealbumin; (47) dialysis duration; (48) concomitant disease; (49) comorbidity index (CDI); (50) depression; (51) anxiety; (52) subjective well-being; (53) disease uncertainty; (54) coping style; (55) self-efficacy; (56) self-management skills; (57) social support; (58) sense of coherence; (59) Benefit finding; (60) self-perceived burden; (61) somatic anxiety; (62) hope; (63) disease perception (64) mental elasticity; (65) health literacy; (66) fear of progression; (67) drug compliance (68) family function; (69) performance status; (70) physical activity; (71) three dimensional positive psychological qualities; (72) physical disability; (73) fluid control compliance; (74) stress.

3.2. Meta-analysis

According to the data, 28 variables were quantitatively analyzed, including demographic variable, disease-related variables, and psychosocial variables. After combining effect sizes, 20 factors were statistically significant. The results were as follows.

3.2.1. Sociodemographic variables

Seventeen studies [27,31,33–35,38,48,52,58,61,66,69,74,76,82,84,86] reported the association between age and symptom burden. Studies have shown that age is positively correlated with symptom burden, and the older the age, the heavier the symptom burden. (Z = 0.12, CI: 0.01, 0.23). There was significant heterogeneity among the seventeen studies (I² = 92%, p < 0.05). Subgroup analysis was conducted to find the source of heterogeneity according to sample size, assessment tools, geographic location and quality assessment levels, but no influencing factors of heterogeneity were found. In addition, the funnel plot and Egger test showed there is no publication bias. Sensitivity analysis shows that our results are stable.

Eight studies [27,31,38,52,61,76,82,84] reported the association between gender and symptom burden, and the studies showed that gender was positively correlated with symptom burden, and the symptom burden was heavier in women than in men. (Z = 0.19, CI: 0.11, 0.28). Significant heterogeneity was found among the included studies (I² = 75.4%, p < 0.05). In order to find the source of heterogeneity, subgroup analysis was conducted according to assessment tools and quality assessment levels, and we found that different literature quality rating levels may be the source of heterogeneity. Sensitivity analysis shows that our results are stable.

Four studies [48,66,69,84] reported the association between marital status and symptom burden, which showed that marital status was negatively correlated with symptom burden, and married patients had lower symptom burden (Z = -0.09, CI: −0.18, −0.01). There was no significant heterogeneity among the included studies (I² = 37.8% < 50%, p = 0.033 > 0.05). Sensitivity analysis shows that our results are stable.

Nine studies [31,38,52,66,69,74,76,82,84] reported the association between education and symptom burden. However, significant heterogeneity was detected in the eligible studies (I² = 89.9%, p < 0.01), and no significant differences were observed in the study effect size estimates, with results not statistically significant (Z = -0.05, CI: −0.19,0.09).

Six studies [38,52,58,69,82,84] reported the association between work status and symptom burden. The study showed that non-employed patients had higher symptom burden (Z = 0.24, CI: 0.12, 0.35). There was significant heterogeneity among the included studies (I² = 75.8%, p < 0.05). We omit one study in each round for sensitivity analysis. The results show that after the deletion of the study by Hao [82], the variation of the results decreases, which may be the source of heterogeneity.

Nine studies [31,38,52,58,66,69,74,82,84] reported the association between income and symptom burden. The study showed that patients with lower income had higher symptom burden (Z = -0.24, CI: −0.36, −0.13). There was significant heterogeneity among the included studies (I² = 83.4%, p < 0.01). In order to find the source of heterogeneity, subgroup analysis was conducted according to sample size, and it was found that sample size may be the influencing factor of heterogeneity. Sensitivity analysis shows that our results are stable.

Two studies [31,52] reported the association between medical cost and symptom burden. The study showed that patients with higher medical cost had more severe symptom burden (Z = 0.22, CI: 0.10, 0.34). No significant heterogeneity was found among the included studies (I² = 0%, p > 0.05).

3.2.2. Disease-related variables

Ten studies [33–35,46,58,71,74,76,81,86] reported the association between income and symptom burden. The study showed that patients with lower income had higher burden of symptoms (Z = 0.29, CI: 0.01, 0.57). There was significant heterogeneity among the included studies (I² = 97.8%, p < 0.01). Subgroup analysis was conducted to find the source of heterogeneity according to sample size, which may be the influencing factor of heterogeneity. In addition, the funnel plot and Egger test showed there is no publication bias. Sensitivity analysis shows that our results are stable.

Four studies [39,41,42,54] reported the association between sleep quality and symptom burden. The studies showed that patients with poorer sleep quality had higher symptom burden (Z = 0.71, CI: 0.55, 0.87). Heterogeneity among the included studies was large (I² = 65.5%, p < 0.05). No factors influencing heterogeneity were found by subgroup analysis. Sensitivity analysis shows that our results are stable.

Five studies [37,45,61,76,81] reported the association between nutritional status and symptom burden. The studies showed that patients with poorer nutritional status had heavier symptom burden (Z = 0.43, CI: 0.20, 0.57). Significant heterogeneity was found among the included studies (I² = 91.7%, p < 0.05). Sensitivity analysis shows that our results are stable.

Four studies [66,71,74,76] reported the association between comorbidities and symptom burden. The studies showed that patients with more comorbidities had higher symptom burden (Z = 0.24, CI: 0.18, 0.30). No significant heterogeneity was found among the included studies (I² = 0%, p > 0.05). Sensitivity analysis shows that our results are stable.

Three studies [38,46,84] reported the association between ipth and symptom burden. However, significant heterogeneity was detected among eligible studies (I² = 89.9%, p < 0.05) and no significant difference was observed. (Z = 0.23, CI: −0.06, 0.51). Sensitivity analysis shows that our results are stable.

Five studies [34,46,50,81,84] reported the association between serum phosphorus and symptom burden. The studies showed that patients with higher blood phosphorus had higher symptom burden (Z = 0.47, CI: 0.15, 1.10). Significant heterogeneity was detected among the included studies (I² = 99%, p < 0.05). Sensitivity analysis shows that our results are stable.

Five studies [46,61,69,77,84] reported the association between hemoglobin and symptom burden. However, significant heterogeneity was detected among eligible studies (I² = 85.7%, p < 0.05) and no significant difference was observed in the study effect size estimates (Z = -0.11, CI: −0.07, 0.30).

Four studies [46,69,77,84] reported the association between serum albumin and symptom burden. However, significant heterogeneity was detected among eligible studies (I² = 92.8%, p < 0.05) and no significant difference was observed in the study effect size estimates (Z = 0.03, CI: −0.26, 0.32).

Four studies [38,46,69,84] reported the association between serum calcium and symptom burden. However, significant heterogeneity was detected among eligible studies (I² = 97.4%, p < 0.05) and no significant difference was observed in the study effect size estimates (Z = 0.30, CI: 0.15, −0.75).

Four studies [46,77,82,84] reported the association between creatinine and symptom burden. However, significant heterogeneity was detected among eligible studies (I2 = 88.9%, p < 0.05) and no significant difference was observed in the study effect size estimates (Z = -0.15, CI: −0.36, 0.16).

Two studies [62,84] reported the association between medical cost and symptom burden. The study showed that patients with higher medical cost had more severe symptom burden (Z = -0.17, CI: −0.31, −0.02). No significant heterogeneity was found among the included studies (I² = 0%, p > 0.05).

3.2.3. Psychosocial variables

Seven studies [26,31,37,41,51,55,72] reported the association between quality of life and symptom burden. The studies showed that patients with poorer quality of life had a higher symptom burden (Z = -0.88, CI: −1.13, −0.63). There was significant heterogeneity among the included studies (I² = 94.8%, p < 0.05). No factors influencing heterogeneity were found by subgroup analysis. Sensitivity analysis shows that our results are stable.

Eleven studies [24–26,34,37,47,53,59,61,72,75] reported the association between depression and symptom burden. The studies showed that patients with poorer quality of life had higher symptom burden (Z = 0.62, CI: 0.50, 0.74). Significant heterogeneity was detected among the eleven studies (I² = 85.5%, p < 0.05). In addition, the funnel plot and the Egger test revealed no significant publication bias. In the sensitivity analysis, the results showed that heterogeneity was reduced by omitting the study by Zhang et al. [72], suggesting that the study may be the source of heterogeneity.

Three studies [32,37,47] reported the association between anxiety and symptom burden. The studies showed that patients with anxiety had higher symptom burden (Z = 0.78, CI: 0.68, 0.87). No heterogeneity was detected among the three studies (I² = 0%, p > 0.05). Sensitivity analysis shows that our results are stable.

Four studies [61,62,71,77] reported the association between social support and symptom burden. The studies showed that patients with poorer social support had higher symptom burden (Z = -0.23, CI: −0.36, −0.11). No significant heterogeneity was detected among the four studies (I² = 58.6%, p = 0.065 > 0.05). Sensitivity analysis shows that our results are stable.

Four studies [63–65,79] reported the association between subjective well-being and symptom burden. The studies showed that patients with lower subjective well-being had higher symptom burden (Z = -0.60, CI: −0.67, −0.53). Significant heterogeneity was detected among the four studies (I² = 87.4%, p < 0.05). No factors influencing heterogeneity were found by subgroup analysis. Sensitivity analysis shows that our results are stable.

Three studies [28,29,59] reported the association between disease uncertainty and symptom burden. The studies showed that patients with higher disease uncertainty had higher symptom burden (Z = 0.30, CI: 0.04, 0.57). Significant heterogeneity was detected among the three studies (I² = 81.1%, p < 0.05). Sensitivity analysis shows that our results are stable.

Three studies [42,51,71] reported the association between self-management skills and symptom burden. The studies showed that patients with poorer self-management skills had higher symptom burden (Z = -0.49, CI: −0.69, −0.30). Significant heterogeneity was detected among the three studies (I² = 82.5%, p < 0.05). In addition, the factor was highly heterogeneous, and sensitivity analyses indicated that the study by Song et al. [71] may have been the source of heterogeneity, and may have been the source of heterogeneity. (I² = 39.6%, p = 0.198 > 0.05).

Three studies [43,44,49] showed that symptom burden was related to coping styles, including facing coping, avoidance coping and yielding coping. There was significant heterogeneity in the coping studies (I² = 89.3%, p < 0.05) and no statistical significance (Z = -0.17, 95%CI: −0.43, 0.10). No heterogeneity was detected in the included avoidance coping studies (I² = 0%, p = 0.78 > 0.05), and avoidance coping was positively correlated with symptom burden (Z = 0.41, 95%Cl: 0.33, 0.48). There was significant heterogeneity in the yield coping study (I² = 85.3%, p < 0.05), and a positive correlation between yield coping and symptom burden (Z = 0.31, 95% CI: 0.08, 0.54). Sensitivity analysis shows that our results are stable (Table 2).

Table 2.

Meta‐analysis of the pooled correlates for symptom burden.

Correlates	N	Heterogeneity Test		Model	Pooled z value	95% CI
		I2 (%)	P
Demographic variables
Age	17	92.6	<0.05	Random	0.10	(0.01, 0.23)
Gender (female)	8	75.4	<0.05	Random	0.19	(0.11, 0.28)
Marital status (married)	4	37.8	0.19	Fixed	−0.09	(−0.18, −0.01)
Education	9	89.9	<0.05	Random	−0.05	(−0.19, 0.09)
Working status	6	75.8	<0.05	Random	0.24	(0.12, 0.35)
Income	9	83.4	<0.05	Random	−0.24	(−0.36, −0.13)
Medical cost	2	0	0.71	Fixed	0.22	(0.10, 0.34)
Disease-related variables
Dialysis age	10	97.8	<0.05	Random	0.29	(0.01, 0.57)
Quality of sleep	4	65.5	<0.05	Random	0.71	(0.55, 0.87)
Nutritional status	4	91.7	<0.05	Random	0.43	(0.20, 0.57)
Comorbidities	4	0	0.56	Fixed	0.24	(0.18, 0.30)
IPTH	3	89.9	<0.05	Random	0.23	(−0.06, 0.51)
Serum phosphorus	5	99	<0.05	Random	0.47	(−0.15, 1.10)
Hemoglobin	5	85.7	<0.05	Random	−0.11	(−0.07, 0.30)
Albumin	4	92.8	<0.05	Random	0.03	(−0.26, 0.32)
Serum Calcium	4	97.4	<0.05	Random	0.30	(0.15, −0.75)
Creatinine	4	88.9	<0.05	Random	−0.15	(−0.36, 0.16)
Serum sodium	2	0	0.64	Fixed	−0.17	(−0.31,-0.02)
Psychosocial variables
QOL	7	94.8	<0.05	Random	−0.88	(−1.13, −0.63)
Depression	11	85.5	<0.05	Random	0.62	(0.50, 0.74)
Anxiety	3	0	0.63	Fixed	0.78	(0.68, 0.87)
Social support	4	58.6	0.65	Fixed	−0.23	(−0.36, −0.11)
Subjective well-being	4	87.4	<0.05	Random	−0.60	(−0.67, −0.53)
Disease uncertainty	3	81.1	<0.05	Random	0.30	(0.04, 0.57)
Self-management ability	3	82.5	<0.05	Random	−0.49	(−0.69, −0.30)
Confrontation coping	3	89.3	<0.05	Random	−0.17	(−0.43, 0.10)
Avoidance coping	3	0	0.78	Fixed	0.41	(0.33, 0.48)
Resignation coping	3	85.3	<0.05	Random	0.31	(0.08, 0.54)

Note: N = number of studies.

4. Discussion

Through systematic review and meta-analysis of existing studies, symptom burden of dialysis patients was correlated with 6 demographic factors, 5 disease-related factors, and 9 socio-psychological factors.

4.1. Demographic variables

Our study found that age was positively correlated with symptom burden of renal dialysis patients, which was consistent with previous studies [27,31,33–35,38,48,52,58,66,74,76,86]. Younger patients generally exhibit better physiological function, enabling them to withstand the adverse effects of long-term dialysis more effectively [66]. Additionally, many young patients maintain a hopeful outlook, which may lessen the burden of symptoms and enhance their resilience to the challenges of long-term dialysis [66]. Conversely, as patients age, their physiological functions decline, chronic disease incidence rises, and they may develop more concurrent conditions, leading to a heavier perceived burden of symptoms [30,48]. Therefore, special attention should be paid to the symptom burden of the elderly in clinical work.

Our study found that gender was positively associated with symptom burden, indicating that women may have heavier symptom burden, consistent with previous research [27,31,37,52,84]. Female patients exhibited a higher symptom burden than male patients, possibly due to their tendency to express discomfort and seek help more openly, as opposed to the more reserved nature often observed in male patients [87,88]. However, other studies [61,64,82] reported no significant correlation between gender and symptom burden. Upon analysis, Hao et al. ‘s study [82] may be the bias caused by fewer female patients (37,6%) and more male patients in the included samples. The sample size of Limbong’s study [61] was small, and the results obtained may be accidental. The female patients in Georges et al. ‘s study [76] were significantly younger and therefore had a lower burden of symptoms.

Marital status was inversely associated with symptom burden, suggesting that married patients had lower symptom burden. This is consistent with the research results in the studies [48,66,69,84]. There was no significant heterogeneity among the included studies, suggesting that marital status was associated with symptom burden. Married patients can get social support from relatives, which enable patients to take the disease seriously, so the symptom burden is lower.

Work status was positively correlated with symptom burden, indicating that patients who were not employed had higher symptom burden, consistent with the previous studies [38,58,69,82,84], which may be because working patients can get the support of family and friends as well as the care and support of colleagues, and can better return to work [58,89]. At the same time, earning income through work can help reduce the economic burden caused by illness. It is suggested that patients should be encouraged to return to society and reflect their self-worth, which is conducive to reducing the symptom burden. However, significant heterogeneity was observed among the studies. Sensitivity analysis revealed that Hao et al.’s study [82] may be the source of this heterogeneity, potentially because their study included only retired unemployed patients, unlike the other studies [38,58,69,84] which encompassed unemployed patients for various reasons.

Income was negatively correlated with the burden of symptoms, indicating that the lower the income, the heavier symptom burden, consistent with previous studies [31,38,52,58,69,74,84]. Medical cost was positively correlated with symptom burden [31,52]. Patients with higher monthly income can afford the long-term dialysis costs, leading to a more serious approach to managing the disease [66]. Conversely, patients with limited economic resources face a dual psychological burden from the disease, treatment, and high costs, resulting in heavier symptom burden. However, Liu and Hao [66,82] reported no significant relationship between income and symptom burden, possibly, due to their inclusion criteria, which focused on 18–55-year-old patients with end-stage chronic kidney disease, leading to potential biases compared to other studies. On the other hand, Hao primarily included subjects with low incomes, potentially biasing the results.

4.2. Disease-related factors

Studies have shown that dialysis age is positively correlated with symptom burden, which is consistent with studies [34,46,58,71,76]. It may be due to prolonged dialysis leading to increased complications such as progressive renal function decline, elevated volume load, vascular calcification, renal hypertension, and subsequent heart failure, resulting in reduced activity endurance and fatigue [81] Peripheral neuropathy, including restless leg syndrome, is also common with extended dialysis, further adding to patient symptom burden. However, Hao [86] did not find a correlation between dialysis age and symptom burden, which may be due to the fact that the subjects in the study were all from the blood purification center of Beijing tertiary General Hospital, with high medical and nursing quality and high survival rate. Moreover, due to the limitation of the scale of the blood purification center, these factors, including the saturation of hemodialysis patients in various hospitals, contributed to the small number of patients in the study who underwent short dialysis.

This study also found a significant positive correlation between sleep quality and symptom burden, the worse the patient’s sleep quality, the heavier symptom burden, which is consistent with previous research results [39,41,42,54]. NovakM et al. [90] reported that 85% of patients receiving renal dialysis experienced sleep issues, stemming from psychological factors and long-term dialysis complications. Chronic sleep problems not only worsen psychological distress like anxiety and depression but also lead to serious health issues such as unstable blood pressure, elevated blood sugar, increased vascular tension, higher cardiac workload, and a significantly raised risk of fatal complications, all contributing to heavier symptom burden [39]. Therefore, actively addressing discomfort symptoms, minimizing symptom interference, and improving sleep quality are essential strategies during dialysis to alleviate overall symptom burden.

There was a significant positive correlation between nutritional status and symptom burden, the worse the nutritional status of patients, heavier the symptom burden, which is consistent with previous research [37,45,61,76,81]. Malnutrition leads to organ function decline, electrolyte imbalances, and an increased likelihood of heightened symptom burden. Therefore, it is crucial to evaluate patients’ nutritional status during treatment, conduct timely assessments, and correct influencing factors to effectively reduce the burden of patients’ symptoms.

Comorbidities were positively correlated with symptom burden, patients with more comorbidities had higher symptom burden, which was consistent with the research results [66,71,74,76]. Therefore, clinical staff should also pay attention to the management of complications patients receiving renal dialysis.

Serum sodium level was negatively correlated with symptom burden, which was consistent with previous research results [62,84]. The pre-dialysis serum sodium level in patients was closely associated with overall mortality and increased cardiovascular mortality rates [91], serving as an independent factor influencing symptom burdens [67].

4.3. Psychosocial variables

The study found a significant negative correlation between quality of life and symptom burden in patients, consistent with prior research [26,31,37,41,51,55,72]. Itchy skin can disrupt sleep, and severe sleep problems can worsen patients’ quality of life. Cheikh [92] pruritus and restless leg syndrome were associated with reduced quality of life. Symptom burden can objectively reflect patients’ quality of life, showing a negative correlation between the two.

Depression showed a significant positive correlation with symptom burden, consistent with previous research [24,26,34,37,47,61,72]. The patient’s depressed mood can impact their appetite, leading to lower food intake, worsened nutritional status, and increased severity of symptoms [34]. Additionally, a heavier symptom burden can also contribute to depression, suggesting a potential bidirectional relationship [34]. This reminded clinicians to pay attention to screening for depression when facing patients with more somatic complaints. The heterogeneity was high, and sensitivity analysis indicated that the study of Zhang [72] may be the source of heterogeneity.

Anxiety was significantly positively correlated with symptom burden, consistent with previous findings [32,37,47]. It is likely due to the characteristics of kidney dialysis treatment, such as its long cycle, high frequency, and cost, leading to various symptoms during the process, which can induce anxiety and increase psychological distress, thereby amplifying the symptom burden.

Social support was negatively correlated with symptom burden, and patients with high social support had lower symptom burden, which was consistent with previous research [61,67,71,77] Patients who received substantial social support felt respected, supported, and understood by society. This support enabled them to better cope with uncomfortable symptoms and potentially overlooks their occurrence temporarily, thus alleviating the overall burden of symptoms [67].

Subjective well-being showed a negative correlation with symptom burden, consistent with earlier research findings [63–65,79]. Poor subjective well-being in patients can lead to depression and other negative emotions, increasing the psychological burden and potentially causing patients to lose confidence in their treatment when symptoms worsen or new symptoms arise [79,93]. This highlights the importance for medical staff to inform patients about potential symptoms in advance, explain the underlying causes, provide necessary disease knowledge guidance, and ensure patients have a good understanding of their symptoms to improve their subjective well-being.

Disease uncertainty was positively correlated with symptom burden, consistent with previous studies [28,29,59]. Patients with chronic kidney disease experience a heavy symptom burden during dialysis, leading to uncertainty about the duration and impact of these symptoms on their treatment. This uncertainty often causes worry about symptom exacerbation or increased treatment needs. Additionally, research indicates that negative emotions stemming from disease uncertainty can affect patients’ treatment compliance and effectiveness [29]. Therefore, clinical nursing efforts should focus on assessing patients’ disease uncertainty, identifying its level and sources, and implementing targeted nursing interventions early on to alleviate uncertainty and improve treatment compliance.

Self-management ability is negatively correlated with symptom burden, and the better the patient’s self-management ability, the lighter the symptom burden, which is consistent with previous studies [42,51,71]. Patients with strong self-management skills lead a more structured lifestyle, prioritize their health, recognize the significance of disease management and adherence to medical recommendations, and exhibit good compliance behavior. Consequently, they can effectively manage their symptoms and reduce the overall symptom burden. However, there was high heterogeneity, and sensitivity analysis indicates that the study by Song [71] may be the source of this heterogeneity, likely due to its larger sample size compared to other studies.

Negative medical coping styles, such as avoidance coping and submission coping, are positively correlated with symptom burden, as indicated by previous studies [43,44,49]. These negative coping styles may exacerbate patients’ dialysis-related symptoms, as they are ineffective in problem-solving and can increase negative emotions, leading to excessive worry about the disease and further deterioration of physical and mental health [49]. Therefore, nursing staff should proactively provide patients with information, support, and encouragement to help them better understands dialysis symptoms and enhance their self-confidence in managing these symptoms.

4.4. Psychosocial variables

Due to the limited evidence available for certain relevant factors, we were unable to conduct a meta-analysis. Therefore, we performed a systematic review of these factors.

Zhang [74] found that urban patients have a lower symptom burden due to better access to medical resources, more health education, and stronger self-management skills [94]. In contrast, rural patients may face challenges adopting healthy lifestyles due to limited education and healthcare access [95]. But another study [82] suggested that urban patients with higher cultural awareness and disease knowledge might experience greater emotional fluctuations and a heavier symptom burden if their disease progression doesn’t meet their expectations [82]. And Zhou and Kou [31,52] highlighted a positive correlation between medical expenses and symptom burden. High medical costs can strain patients financially, leading to poor compliance and worsening symptoms.

Zhao [73] showed that the rate of blood pressure variation was related to symptom burden, explaining that excessive fluctuations can disrupt vasoconstriction and dilation functions, leading to abnormal brain perfusion and related symptoms like dizziness and waking difficulties [96]. Geng and Karaaslan [62,84] found a negative correlation between serum sodium levels and symptom burden. Karaaslan [84] also showed that urea clearance index, reflecting dialysis adequacy, was significantly associated with higher symptom burden, suggesting adequate dialysis doses can alleviate symptoms. However, Ozen [70] found no significant link between Kt/v (a dialysis adequacy measure) and symptom burden. Zhang et al. [38] showed that the calcium-phosphorus product was related to it, contrary to Karaaslan et al. [84]. These studies underscore that symptom burden is influenced not only by lab data but also individual, environmental, and disease factors.

Previous studies [66,77] have examined disease perception, which encompasses organized thoughts and beliefs triggered by a health threat [97]. As chronic kidney disease progresses and more dialysis cycles are received, middle-aged and elderly hemodialysis patients develop negative thoughts about chronic kidney disease. Gunarathne [77] found that disease perception influences symptom burden. The studies of Gao and Song et al. [65,71]indicated a negative relationship between psychological congruence and symptom burden. Patients with lower psychological congruence tend to perceive more symptoms due to reduced confidence in disease management, limited perception of available resources, and inadequate utilization of coping strategies and health maintenance resources.

5. Limitations

The study has several limitations. Firstly, some data cannot be used directly; the need for data conversion, data conversion may be biased. Secondly, the study included all relevant factors addressed in the published literature as far as possible in order to obtain comprehensive information for clinical reference, but this may have resulted in the inclusion of a large number of factors and an insufficiently focused study. Thirdly, certain correlations showed high heterogeneity, possibly due to variations in geographical areas, study populations, and literature quality. However, due to limited data on some factors, the sources of heterogeneity could not be explored through subgroup analysis. Moreover, the small number of studies on some of the relevant factors has the potential to introduce uncertainty. Therefore, more research is needed in the future to validate these results.

6. Conclusion

The systematic review and meta-analysis identified the factors associated with symptom burden. These results will provide information for early health education or targeted interventions for patients receiving renal dialysis to reduce the burden of symptoms and improve the quality of life of patients. However, more empirical researches are needed to confirm these results in the future.

Funding Statement

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Authors’ contributions

Yifan Lu: Conceptualization, Data Curation, Visualization, Software, Methodology, Investigation, Formal analysis, Writing - Original Draft, Writing - Review & Editing. Shuqi Zhai: Conceptualization, Data Curation, Visualization, Validation, Software, Investigation, Formal analysis, Writing - Review & Editing. Qinqin Liu: Investigation, Formal analysis. Congcong Dai: Visualization, Formal analysis. Shejuan Liu: Methodology, Writing - Review & Editing. Yanqing Shang: Investigation, Methodology, Writing-Review and Modifying. Chaoran Chen: Supervision, Writing - Review & Editing.

Disclosure statement

The authors declare no conflict of interest.