Humanistic burden and economic impact of chronic kidney disease: a systematic literature review

Systematic literature review

Two systematic searches, one on humanistic burden/QoL and one on economic burden, were performed in MEDLINE and MEDLINE In-Process, Embase, and the Cochrane Library via Ovid. Cut-off dates were January 2002 and May 2017 for the humanistic burden SR and January 2007 and May 2017 for the economic burden SR; the shorter review window for the economic burden SR was specified because economic data are considered to decrease in relevance more quickly than QoL data. As supplementary searches, congress abstracts published between 2015 and 2017 (or the most recent 2 years available) from relevant health economics and outcomes research and nephrology meetings were reviewed. The search strings used to identify evidence are listed in Extended data: Supplementary content 1⁶.

Citation screening and full text review

Abstracts and titles identified were screened by an independent reviewer to determine whether they met the PICOS (patient, interventions, comparisons, outcomes, and study design) eligibility criteria (Table 1), in accordance with 2009 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines⁷. All publications that met entry criteria for review were obtained as full articles and reassessed against the predefined inclusion criteria. Owing to the large number of citations meeting these criteria, a decision was taken to restrict the extraction of data from eligible studies. For both SRs, data were extracted only from full publications. This meant that, although congress abstract screening was performed, no data from congress abstracts were extracted. For the humanistic burden SR, study publication dates for data extraction were restricted to 2007–2017; for the economic burden SR, the relevant time period was restricted to 2012–2017. Furthermore, for the humanistic burden SR, data were not extracted if the study population included fewer than 100 patients with CKD or if the study did not use any of the following instruments:

Data extraction

For each publication, information was extracted into a data extraction table. Studies were listed according to inclusion of QoL, symptom burden or cost, and resource use data. For each study, patients’ CKD stage, RRT status [pre-dialysis, dialysis (including dialysis modality) or renal transplant], age, and comorbidities including CVD and the presence of DM were recorded. No risk of bias assessment between or within studies was performed.

Focus of this review

These SRs were conducted to collect information on the overall impact of CKD development and progression on patients’ QoL or their healthcare costs. Therefore, we chose to focus this review on studies that compared QoL or costs for patients with CKD and individuals without CKD, or that compared by CKD severity, defined by either disease stage or estimated glomerular filtration rate (eGFR) category. Other studies identified in the SRs are grouped into key themes and listed separately to indicate the scope of the evidence identified in our searches.

Search results

In total, 5219 papers were identified in the initial searches, of which 1114 papers were removed as duplicates, and 4105 were included for screening by abstract and title. This screening identified 3539 papers that did not meet the inclusion criteria. In total, 444 papers were included for full paper review. Following full paper review, 284 references were identified for inclusion, plus 79 abstracts identified in supplementary searches. In total, data were extracted from 95 references reporting QoL data, 47 references reporting cost and resource use data, and eight references reporting descriptions of symptoms. A PRISMA flow diagram is shown in Figure 1. Details of the patient populations in the studies discussed in this review are given in Table 2. Figure 2 shows all studies identified in these SRs, grouped into key themes according to the data reported.

Figure 1. PRISMA flow diagram, showing the flow of studies identified through the systematic review process.

^aThe congresses included in this search were the International Society for Pharmacoeconomics and Outcomes Research US and European congresses, the European Renal Association–European Dialysis and Transplant Association congress, the World Congress of Nephrology, the American Society of Nephrology Kidney Week, and the National Kidney Foundation Spring Clinical Meeting. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses. QoL, quality of life. SR, systematic review.

Figure 2. Summary of evidence identified in these systematic reviews, showing the number of studies identified, grouped according to the key data presented.

^aSome studies fell into more than one category. CKD, chronic kidney disease. HRQoL, health-related quality of life.

QoL in patients with CKD compared with the general population

Several studies compared the QoL of patients with CKD with that of individuals in the general population. These studies reported overall measures associated with QoL, as well as physical component summary (PCS), mental component summary (MCS), and individual domain scores from the SF-36.

Across the studies identified, patients with CKD experienced lower physical and mental QoL compared with the general population. A study by Davison et al. that measured QoL using the SF-6D found that a mixed population of patients with CKD who were either pre-dialysis (CKD stage 3–4, anticipated to require dialysis within 12 months; 34%) or dialysis-dependent (66%) reported lower QoL than an age-matched subset of the general population (SF-6D score: 0.67 vs 0.79)⁸. A study by Nguyen et al. found that patients with pre-dialysis CKD (stages 3–5), the majority (69.1%) of whom had CKD stage 3a, spent, on average, a greater number of days over a 30-day period inactive owing to poor physical and mental health (3.1 vs 1.5 days; P < 0.001) and a greater number of days with poor physical health (5.1 vs 3.0 days; P < 0.01) compared with a sample of the US general population⁹. However, a study conducted by Agarwal et al. in the USA, which measured only sleep quality, found no difference between patients with pre-dialysis CKD or dialysis-dependent ESRD and a matched population without CKD¹⁰.

Compared with the general population, patients receiving dialysis experienced lower QoL. In total, four studies reported SF-12/SF-36 scores in patients receiving dialysis compared with the general population^11–14. Studies by Wan et al. and Wang et al. reported a statistically significant reduction in SF-12 and SF-36 PCS scores (9.9% and 37.6%, reductions, respectively; both P < 0.001; Figure 3)^12,13. Boini et al. also reported a reduction in SF-12 PCS scores in patients undergoing dialysis compared with the general population of the USA and France (21.0% and 21.5%, respectively; Figure 3); however, no statistical analysis was reported¹¹. Boini et al.¹¹ and Wang et al.¹³ reported a decrease in SF-12 (USA: 14.2%; France: 9.1%; P = NR) and SF-36 (23.0%; P < 0.001) MCS scores, respectively, in patients who were dialysis-dependent compared with the general population, whereas Wan et al.¹² reported a statistically significant increase in SF-12 MCS scores (9.8%; P < 0.001) in patients receiving dialysis compared with the general population of Hong Kong (Figure 3).

Figure 3. Scores across all domains of the Short-Form Health Survey in patients with end-stage renal disease or those undergoing dialysis compared with the general population, which provides a graphical representation of the distribution of scores across individual domains of the Short-Form Health Survey for different patient populations^11–14.

^aMH, P < 0.01 versus GP; RL-P and BP, P < 0.005 versus GP; PCS, MCS, PF, and GHP, P < 0.001 versus GP. ^bFor all domains, P < 0.001 versus average. ^cPatients with ESRD versus GP (age- and sex-adjusted): BP, P < 0.05 versus GP; MH, P < 0.005 versus GP; PF, RL-P, GHP, V, SF, RL-M, P < 0.001 versus GP. Patients on palliative care versus GP (age- and sex-adjusted): RL-P and V, P < 0.05 versus GP; PF and SF, P < 0.001 versus GP. ^dNo statistical analysis reported. ^eFor the general population of the USA, MCS and PCS scores were both reported as 50. BP, bodily pain, CKD, chronic kidney disease. ESRD, end-stage renal disease. GP, general population. HD, haemodialysis. MH, mental health. PF, physical functioning. RL-M, role-limiting (mental). RL-P, role-limiting (physical). SF, social functioning. SF-12/36, 12-/36-item Short-Form Health Survey. V, vitality.

The studies by Wan et al., Wang et al., and Yong et al. reported individual SF-12 and SF-36 domain scores for patients undergoing dialysis compared with the general population^12–14. Wan et al. found that patients receiving dialysis reported significantly higher scores for the mental health domain (7.0%; P = 0.006) but lower scores in the physical functioning (20.3%; P < 0.001), role-limiting (physical; 9.8%; P = 0.004), bodily pain (10.0%; P = 0.004), and general health (20.0%; P < 0.001) domains than those without CKD (Figure 3)¹². In studies by Wang et al. and Yong et al., a statistically significant reduction in scores across all domains of the SF-36 (P < 0.001 and P < 0.05, respectively) was reported in patients receiving dialysis compared with the general population (Figure 3)^13,14. Yong et al. also reported lower SF-36 domain scores in patients with ESRD undergoing palliative care compared with the general population. A statistically significant difference was found only in the physical functioning (P < 0.001), role-limiting (physical; P = 0.027), social functioning (P < 0.001), and vitality (P = 0.036) domains (Figure 3)¹⁴.

QoL by severity of CKD

Several studies reporting QoL data stratified by CKD stage were identified, in addition to studies that examined the impact of demographic factors and comorbidities on the relationship between disease severity and QoL.

Overall, later stages of CKD were associated with worse QoL, but the trends for physical QoL were slightly different from those for mental QoL. In a study by Roumelioti et al., patients receiving maintenance dialysis had a near-significant decrease (17.7%; P = 0.05) in the SF-36 physical functioning domain score compared with patients with advanced CKD (eGFR < 30 mL/min/1.73 m²)¹⁵. Similarly, in two US studies by McClellan et al. and Porter et al., decreasing eGFR was associated with lower SF-12 (P = 0.001) and SF-36 (P = 0.004) PCS scores, but there was no apparent relationship between eGFR and MCS scores (data shown in Extended data: Supplementary content 2^16–18. McClellan et al. reported results from an unadjusted analysis as well as from an analysis adjusted for sociodemographic status and comorbidities; a correlation between eGFR and SF-12 PCS score was identified in both analyses (P < 0.001)¹⁶. Porter et al. reported results from an unadjusted analysis only¹⁷.

In several studies, the presence of comorbidities affected whether worsening CKD severity was linked to reductions in QoL. Porter et al. found that patients with a greater number of comorbidities experienced overall worse physical and mental QoL, assessed by SF-36 scores, than those with fewer comorbidities (mean PCS/MCS score: no comorbidities, 43.0/47.9; one comorbidity, 39.3/45.8; two or three comorbidities, 36.8/44.4; four or more comorbidities, 31.2/36.0; P < 0.001 for PCS and MCS)¹⁷. Furthermore, studies conducted by Campbell et al. and Wolfgram et al. in the USA found that adjustment for comorbidities and sociodemographic factors affected the relationship between QoL scores and eGFR^19,20. Campbell et al. found that in an unadjusted analysis, patients with an eGFR of less than 60 mL/min/1.73 m² reported a statistically significant reduction in SF-8 PCS scores (P < 0.05) compared with patients with an eGFR greater than 90 mL/min/1.72 m² (reference group), whereas those with an eGFR of 60–89 mL/min/1.72 m² reported higher SF-8 PCS scores (P < 0.05) than the reference group [mean PCS score by eGFR (mL/min/1.72 m²): ≥ 90, 44.04; 75–89, 44.97; 60–74, 45.00; 45–59, 42.32; 30–44, 40.41; ≤ 29, 40.02]¹⁹. However, following full adjustment for demographic factors and comorbidities, there were no significant differences between patients in any eGFR category and the reference group (Extended data: Supplementary content 2)¹⁸. Interestingly, there was a statistically significant difference in SF-8 MCS scores between patients who had the most severe disease (eGFR less than 29 mL/min/1.73 m²) and patients with the least severe disease in both unadjusted and adjusted analyses (fully adjusted analysis data shown in (Extended data: Supplementary content 2)^18,19. Similarly, a study by Wolfgram et al., also in the USA, found that decreasing eGFR was associated with significantly lower EQ-5D scores in both the unadjusted analysis [mean EQ-5D score by eGFR (mL/min/1.72 m²): ≥ 60, 0.85; > 44 to < 60, 0.85; ≥ 44, 0.82] and an analysis adjusted for sociodemographic factors [mean EQ-5D score by eGFR (mL/min/1.72 m²): ≥ 60, 0.84; > 44 to < 60, 0.84; ≥ 44, 0.82]²⁰. However, in an analysis adjusted for sociodemographics and comorbidities, including CVD, no relationship between eGFR and QoL was found [mean EQ-5D score by eGFR (mL/min/1.72 m²): ≥ 60, 0.73; > 44 to < 60, 0.75; ≥ 44, 0.73]²⁰.

Studies by Rajan et al. and Eriksson et al. assessed the impact of increasing disease severity in patient populations with specific comorbidities^21,22. Rajan et al. reported QoL scores for patients with DM that was either prevalent (longer than 3 years) or recent-onset (less than 3 years)²². In both patient populations, scores for the physical functioning and role-limiting (physical) domains of the SF-36 were lower with increasing CKD stage; however, scores for the mental health domain increased up to stage 3, but were lower for stage 4 and lower still for stage 5 (Extended data: Supplementary content 2)^18,22. The authors acknowledged that many of the patients with CKD stage 0/1 had claims for International Classification of Diseases (Ninth Revision) codes relating to mental illness, which may account for the lower scores in the mental health domain in this subgroup²². Eriksson et al. reported QoL in patients with pre-dialysis CKD (stage 3–4) or receiving dialysis, with or without anaemia²¹. Lower SF-12 and EQ-5D scores were reported by patients with CKD stage 4 than those with CKD stage 3, and the lowest scores were reported by patients receiving dialysis; however, no statistical analysis of the difference between stages was reported. This trend was present both in patients with anaemia and in those without anaemia (Figure 4)²¹.

Figure 4. Quality of life with increasing disease severity in patients with and without anaemia: (a) SF-12 PCS scores; (b) SF-12 MCS scores; (c) EQ-5D scores, showing the change in quality of life as CKD stage increases²¹.

No statistical analysis of the between-stage difference was reported. EQ-5D, 5-dimension EuroQol questionnaire. MCS, mental component summary. PCS, physical component summary. SF-12, 12-item Short-Form Health Survey.

Other QoL studies and symptom burden studies

Studies were identified that reported QoL data associated with treatment modality and setting^8,12,23–41; comorbidities in CKD^{17,20–22,41–60}; mental health^61,62; sleep quality^{10,59,63–65}; patient support, care, and disease management^11,66–83; sociodemographic factors^84,85; and other factors predictive of CKD development^15,86–103. These studies are listed in Extended data: Supplementary content 3¹⁰⁴. A total of eight studies were identified on symptom burden in patients with CKD, of which two studies^105,106 discussed the symptoms experienced by patients with CKD stage 5 who were not receiving dialysis, and six studies^{24,34,67,107–109} discussed symptom burden in patients receiving dialysis (Extended data: Supplementary content 3¹⁰⁴).

Costs for CKD or ESRD, compared with normal kidney function

The SR identified several studies that compared costs and resource use between individuals with CKD and those with normal kidney function. Some of the studies differentiated between patients with pre-dialysis CKD and those who required RRT, and the patients in several of the studies had additional comorbidities.

Across all studies, pre-dialysis CKD and ESRD requiring RRT were associated with significant increases in cost and resource use. Kim et al. found that patients with pre-dialysis CKD were significantly more likely to experience hip fracture-related hospitalization than patients with normal kidney function, and those receiving RRT were at higher risk than patients with pre-dialysis CKD. Median hospital costs were significantly greater in patients with CKD than for the general population and were highest for patients requiring RRT. This trend was also observed in the length of hospital stay, although the absolute differences between groups were small (Figure 5)¹¹⁰. Kumar et al. found very similar results in a study examining hospitalization rates and length of stay for pulmonary embolism in patient populations with normal kidney function, with pre-dialysis CKD or receiving dialysis (Figure 5)¹¹². In a study of patients with hepatitis C, total healthcare costs were 2.9 times higher for those who had CKD, compared with those who had hepatitis C without CKD (USD 548 vs USD 1922; P < 0.001)¹¹². Some patients with CKD in this study were defined as having ESRD; however, patients’ dialysis or transplant status was not specified. In a fourth study, patients with bone metastases and impaired kidney function incurred 60% higher total healthcare costs than a control group of patients with bone metastases and normal kidney function (USD 142,267 vs USD 88,839; P < 0.001). These increased costs were driven by hospital admission, emergency department and outpatient visits, longer length of stay in hospital, and higher pharmacy costs¹¹³. An Australian study by Wyld et al. also showed that patients with any stage of pre-dialysis CKD incurred significantly higher costs than individuals with no CKD (Figure 6a)¹¹⁴.

Figure 5. Hospitalizations in patients with normal kidney function, chronic kidney disease, or end-stage renal disease, showing the number of hospitalizations per 1000 individuals in patient with different severity of disease^110,111.

^aP < 0.001 for each pairwise comparison. ^bIncluded patients receiving dialysis or renal transplant. ^cIncluded patients receiving dialysis. ^dP < 0.05 versus normal kidney function. CKD, chronic kidney disease. ESRD, end-stage renal disease. IQR, interquartile range. LoS, length of stay. USD, US dollars.

Figure 6. Total costs for patients with chronic kidney disease (pre-dialysis or dialysis status not specified), by disease stage, showing the total cost of healthcare for patients with different severity of disease from three studies in Australia, Italy, and the USA^114–116.

^aOne-way ANOVA. ^bDialysis status not specified. ^cHypertension, DM, history of CHF, hyperkalaemia, or proteinuria. ANOVA, analysis of variance. AUD, Australian dollars. CHF, congestive heart failure. CKD, chronic kidney disease. DM, diabetes mellitus. EUR, euros. USD, US dollars.

Costs by severity of CKD

In total, six studies reported costs for patient populations stratified by CKD stage. Across the evidence base, later-stage disease was associated with comparatively higher costs, both in studies including only patients with pre-dialysis CKD and in those in which some patients were receiving RRT.

Wyld et al. showed that costs increased significantly across pre-dialysis CKD stages (Figure 6a)¹¹⁴. Patients with CKD stage 3 incurred approximately 28% higher direct annual healthcare costs than those with CKD stage 1–2; however, there was a much larger cost increase for CKD stage 4–5, with patients in this group incurring more than fourfold higher costs than patients with CKD stage 3. This cost increase was apparent in subgroups of patients both with and without DM, suggesting that the high costs associated with CKD stage 4 and 5 were not attributable solely to the presence of DM. However, only 18 patients in this study had CKD stage 4–5, meaning that the results may have been skewed by a small number of individuals who incurred exceptionally high healthcare expenditure.

Two studies examined the costs of CKD across stages 4–5 or 3–5. An Italian study by Turchetti et al. of patients with pre-dialysis CKD showed a 31% increase in the direct annual medical costs associated with CKD stage 5, compared with CKD stage 4 (P < 0.01; Figure 6b)¹¹⁵. DM and CVD were shown to have an impact on costs incurred by patients at either CKD stage. Smith et al., who conducted a study in the USA, found that patients with CKD stage 3a who had comorbidities and patients with CKD stage 3b incurred similar monthly costs associated with health insurance claims (Figure 6c)¹¹⁶. Those with CKD stage 4–5, however, incurred more than double these costs (P < 0.05). Patients’ dialysis status was not specified in the study publication, but a breakdown of costs was reported. Later-stage CKD was associated with higher inpatient, outpatient, and professional services costs than CKD stage 3a or 3b, whereas the highest prescription costs were incurred by patients with CKD stage 3b.

Several studies examining costs by CKD stage included patients who had progressed to RRT and therefore incurred additional costs^117,118. Two studies were conducted in patient populations with CKD and DM in the USA. McQueen et al. found an increase in annual costs for each successive CKD stage, with the exception of CKD stage 2, which was associated with slightly lower costs than CKD stage 1. As in other studies, the largest increase was between the later disease stages, with a 71% increase in costs for stage 5 compared with stage 4 (Figure 7a)¹¹⁷. A similar cost increase was reported by Vupputuri et al. who compared costs between patients whose disease progressed and those who remained stable. Individuals who progressed to RRT incurred 77% higher annual medical costs than patients who remained at CKD stage 4 (Figure 7b)¹¹⁸. In a third study, Ariyaratne et al. examined the impact of CKD stage on patients undergoing percutaneous coronary intervention (PCI) in Australia. Direct cardiovascular costs, assessed 1 year after PCI, did not increase significantly for CKD stage 3 from stage 1–2 (AUD 4851 vs AUD 4442; 9% increase; P = 0.052), whereas patients with CKD stage 4–5, some of whom were receiving dialysis, incurred significantly higher costs than those at stage 1–2 (AUD 6958; 57% increase; P < 0.001)¹¹⁹.

Figure 7. Total costs for patients with diabetes mellitus and chronic kidney disease by disease stage, including patients receiving dialysis, providing a graphical representation of annual costs across disease stage in two US studies^117,118.

^a14.4% of patients at CKD stage 4 had a procedure code for dialysis during the baseline period. ^b70.3% of patients at CKD stage 5 had a procedure code for dialysis during the baseline period. ^cESRD was defined as a requirement for dialysis or renal transplant, and was considered to be equivalent to CKD stage 5 in this study. CKD, chronic kidney disease. ESRD, end-stage renal disease. NS, nonsignificant. T2DM, type 2 diabetes mellitus. USD, US dollars.

Costs for pre-dialysis CKD compared with RRT

Two studies were identified that compared costs for patients with pre-dialysis CKD with those for patients receiving dialysis (Extended data: Supplementary content 4)^120–122. Eriksson et al. compared costs and resource use in Sweden between four treatment groups: patients with pre-dialysis CKD (stage 4–5), those receiving haemodialysis, those receiving peritoneal dialysis, and patients with renal transplant. Patients with pre-dialysis CKD incurred the lowest total costs and had the lowest rate of outpatient visits of any treatment group, but had a slightly higher hospitalization rate and a greater mean number of hospital days annually than those who had received a renal transplant¹²¹. Both types of dialysis incurred higher inpatient and outpatient costs than pre-dialysis CKD or renal transplant, with more than 70% of the costs associated with haemodialysis contributed by outpatient care. All types of RRT were associated with additional expenditure on drugs, compared with pre-dialysis CKD. In addition, a study by Escudero-Vilaplana et al. showed that among patients receiving erythropoietin-stimulating agents (ESAs), the monthly cost of ESA therapy was significantly higher for patients receiving peritoneal dialysis than for those with pre-dialysis CKD (stage 2–5)¹²².

Other cost and resource use studies

The other studies identified in the economic burden SR did not compare patients with CKD with the general population or report data by disease severity. Data from these studies are not reported in this manuscript; however, the studies are listed in Extended data: Supplementary content 3¹⁰⁴ to illustrate key trends and groupings in the economic evidence base for CKD. The studies that we identified reported data on the costs associated with different treatment modalities or settings^{26,121,123–127}; on the cost of comorbidities in CKD, including DM, cardiovascular complications, and anaemia^{112–115,117–119,122,128–141}; on hospitalization costs and rates for patients with CKD^{62,66,142–148}; and on other costs in CKD^149–155.

Underlying data

All data underlying the results are available as part of the article and no additional source data are required.

Extended data

Figshare: Freeman et al. Supplementary content 1 – Table S1. http://doi.org/10.6084/m9.figshare.10011476⁶.

This project contains Table S1: Electronic search strategy.

Figshare: Freeman et al. Supplementary content 2 – Figure S1. https://doi.org/10.6084/m9.figshare.10011683¹⁸.

This project contains Figure S1: Physical and mental component summary scores with increasing severity of chronic kidney disease.

Figshare: Freeman et al. Supplementary content 3 – Table S2. http://doi.org/10.6084/m9.figshare.10011692¹⁰⁴.

This project contains Table S2: Additional quality of life, symptom burden and cost and resource use studies identified in the systematic review.

Figshare: Freeman et al. Supplementary content 4 – Table S3. https://doi.org/10.6084/m9.figshare.10011710¹²⁰.

This project contains Table S3: Summary of studies that compared costs for patients with chronic kidney disease by renal replacement therapy status.

Figshare: Freeman et al. Supplementary content 5 – Protocol. https://doi.org/10.6084/m9.figshare.11112578¹⁵⁶.

This project contains the protocol for this systematic review.

Reporting guidelines

Figshare: PRISMA checklist for ‘Humanistic burden and economic impact of chronic kidney disease: a systematic literature review’. http://doi.org/10.6084/m9.figshare.10011893¹⁵⁷.

Data are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).