Introduction

The global prevalence of chronic kidney disease (CKD) is estimated to be 10–13% and increasing; consequently, demand for dialysis services is also likely to increase worldwide [1]. Few large randomized data sets in the last few years have characterized health-related quality of life (QoL) in patients on haemodialysis (HD) using validated metrics. However, from the available data, haemodialysis is associated with impaired QoL and high morbidity and mortality [2].

Quality of life is defined by the World Health Organization (WHO) as “an individual’s perception of their position in life in the context of their values and culture relative to both where they live and their aspirations, ideas, concerns and expectations” [3]. The subjective awareness of each individual of their physical, mental, social and functional well-being is critical to understanding the possible associations of treatments on QoL [4]. Both physical inactivity and impaired physical function as seen during the COVID-19 pandemic are also strongly associated with high morbidity, mortality and reduced QoL in patients on HD [5]. In prevalent haemodialysis patients a low physical component score (PCS) has been associated with increased adjusted rates of death [hazard ratio (HR), 1.55, 95% confidence interval (CI) 1.19–2.03] and hospitalisation (HR, 1.29, 95% CI 1.09–1.54), and a low mental component score (MCS), with increased rates of hospitalisation (HR, 1.39, 95% CI 1.17–1.65) but no association between the symptoms, effects, and burden of kidney disease subscales [6]. There are no data in incident dialysis patients.

The PIVOTAL (Proactive IV irOn Therapy in hemodiALysis) trial investigated the effects of proactive high-dose versus reactive low-dose intravenous (IV) iron in patients in the first year of haemodialysis who were on erythropoiesis-stimulating agent (ESA) therapy. A pre-specified secondary outcome of PIVOTAL was change in QoL measures using the validated European Quality of life 5-dimension score (EQ5D) and kidney disease quality of life (KD-QoL) questionnaires [7, 8]. This is a blunt health status instrument of the perception of health rather than quality of life per se. These scores were measured at baseline and every 3 months for the first year, then every 6 months until the end of the trial. There were no significant between-treatment (proactive high-dose versus reactive low-dose intravenous iron) differences in QoL during the trial but the results were limited by the number of completed questionnaires available during follow-up for analysis. We now report the detailed baseline data, which are not controlled, related to QoL measures in trial participants enrolled in the study [both the higher-dose approach (proactive regimen) and lower-dose approach (reactive regimen)]. We examine the potential relationship of QoL scores with the primary outcome (all-cause mortality, myocardial infarction (MI), stroke, and hospitalisation for heart failure), as well as associations of QoL scores with other key baseline characteristics including clinical and laboratory factors, presence of anaemia, or iron deficiency Specifically, we examined baseline predictors of QoL, as well as the association of haemoglobin and other variables at baseline on QoL scores. We also investigated whether the presence of iron deficiency, irrespective of haemoglobin, influences QoL, and the associations of baseline QoL measures on the primary outcome measure of the trial.

Methods

The design, details of baseline characteristics, and the main outcomes of this prospective randomized, controlled study (open-label with blinded end-point evaluation) have been published previously [7, 8]. In brief, a total of 2141 patients with end-stage kidney disease (ESKD) on maintenance haemodialysis initiated ≤ 12 months before randomization were included. At baseline, inclusion criteria included a transferrin saturation (TSAT) < 30% and a ferritin level < 400 ng/mL. Patients were randomized to a proactive regimen of IV (400 mg iron sucrose monthly, with cut-offs to discontinue IV iron therapy if ferritin rose above 700 ng/mL and/or the TSAT increased to over 40%) or to a reactive iron regimen (0–400 mg iron sucrose monthly to maintain a ferritin level of at least 200 ng/mL and a TSAT ≥ 20%). The median cumulative iron dose at one year was 3.8 g in the proactive arm and 1.8 g in the reactive arm. Patients received ESAs at a high enough dose to keep haemoglobin levels between 100 and 120 g/L.

Participants were then requested to complete validated, structured health surveys (EQ5D and KD-QoL). The EQ5D includes overall index and a VAS with a range from 0 to 100, with a higher score indicating better health and 5 separate questions covering mobility, self care, usual activities, pain or discomfort and anxiety/depression. EQ5D has been used in many patient populations including CKD [9, 10]. The KD-QoL is a valid measure of generic but disease-specific health-related QoL scoring system which is separated into 5 subscales consisting of 2 general component summary measures (physical health composite scores (physical functioning (PF) + role physical (RP) + bodily pain (BP) + general health (GH)) and mental health composite scores (vitality (VT) + social functioning (SF) + role emotional (RE) + mental health (MH)). These include eight concept scales which can each be defined as follows; physical functioning—the level of limitation of physical activity caused by health limitations, role-physical—a measure of the limitations of patient-specific physical activity caused by health problems, bodily pain, general health, vitality—measurement of energy and fatigue, social functioning- the level of social life limitations caused by physical and emotional discomfort, role emotional and mental health– the level of psychological stress and well-being. The other 3 subscales consist of kidney disease-targeted specific scores on burden of kidney disease, symptoms of kidney disease and the effects of kidney disease. This score is calculated with norm-based scoring so that 50 is the average score, with higher scores indicating a better QoL [11, 12].

Ethics and regulatory approvals

The trial was conducted in compliance with the principles of the Declaration of Helsinki (1996), the principles of good clinical practice (GCP), and in accordance with all applicable regulatory requirements including, but not limited to the Research Governance Framework and the Medicines for Human Use (Clinical Trial) Regulations 2004, as amended in 2006 and any subsequent amendments. The trial and all elements of the protocol and revised protocols were approved by the South East Coast—Brighton and Sussex Research Ethics Committee (REC number 13/LO/1115), and by the Medicines and Healthcare products Regulatory Agency (MHRA) for Clinical Trial Authorisation.

Written Informed Consent was obtained from eligible patients who had been on haemodialysis for 0–12 months and based on the inclusion, exclusion criteria. Only patients who provided consent were subsequently randomised into the study.

Statistical analysis

Baseline descriptive characteristics and analysis were summarized as means ± standard deviations (SD) for normally distributed data, and medians and inter-quartile ranges for not normally distributed data. Percentages and frequencies were used where appropriate. p-values for between-variable differences based on two sample t-tests, analysis of variance or chi-squared tests/Fisher’s exact tests, as appropriate, are provided. Analyses were performed using SAS software, version 9.4 (SAS Institute), Minitab version 20.3 and R version 3.6.0.

A linear regression model adjusting for significant univariate predictors was used to identify independent predictors of health-related QoL using baseline characteristics and the randomized treatment allocation (proactive high-dose vs reactive low-dose intravenous iron). Time to first event outcomes were analysed using Cox proportional hazards models with estimation of hazard ratios, 95% confidence intervals and p-values from the Wald statistic.

The detailed analysis of QoL data included stratifications based on the following parameters:

  1. 1.

    baseline haemoglobin > 100 g/L vs ≤ 100 g/L

  2. 2.

    serum ferritin > 200 vs ≤ 200 ng/ml

  3. 3.

    TSAT > 20% vs ≤ 20%

  4. 4.

    erythropoiesis-stimulating agent dose (units/week)

  5. 5.

    serum albumin (g/L)

  6. 6.

    sex—males versus females

  7. 7.

    association of social demographic data including age

  8. 8.

    association of aetiology of CKD

  9. 9.

    diabetes vs no diabetes

  10. 10.

    ethnicity

  11. 11.

    vascular access

  12. 12.

    number of comorbidities

Results

Demographic data and disease-related characteristics

The participants’ demographics have been described previously, but in brief a total of 2141 haemodialysis patients (1398 male; 743 female) were randomized, of whom 79% were white, and whose mean age was 62.8 years. At baseline, 41% of patients were dialysed using a central venous catheter while 59% had an arteriovenous graft or fistula. Diabetes was reported to be the cause of ESKD in 587 patients (27.4%). Additional causes of renal disease and co-morbidities are detailed in Table 1, as well as laboratory measurements and ESA doses.

Table 1 Overall statistical analysis of the total scores of EQ5D Index (mean: 0.68: SD 0.26; n = 1777) and EQ5D VAS (mean: 60.75: SD 20.87; n = 1787) summaries of patients with end stage kidney disease on haemodialysis, according to the categorical socio-demographic variables and categorical clinical and laboratory parameters
Full size table

Predictors of the primary endpoint and all-cause mortality

After adjustment for randomized treatment (model 1), then adjusting for other baseline predictive variables (model 2), EQ5D Index (HR: 0.93; 95% CI 0.9–0.97; p < 0.001), EQ5D visual analogue scale (VAS) (0.94: 0.89–0.98; p = 0.006), KD-QoL PCS (0.84: 0.76–0.94) and MCS (0.90: 0.83–0.93) were found to be independently associated with all-cause mortality (Fig. 1). Similar associations were observed for the primary outcome (all-cause death, myocardial infarction, stroke or hospitalisation for heart failure): EQ5D Index (HR 0.93; 95% CI 0.9–0.96; p < 0.001), EQ5D VAS (0.93: 0.87–0.97; p < 0.001), as well as KD-QoL PCS (0.84: 0.77–0.93; p < 0.001) and MCS (0.88: 0.82–0.95; p = 0.0017). There was also an association with symptoms of kidney disease (0.95: 0.90–0.99; p = 0.019) and effects of kidney disease (096: 0.92–1.99; p = 0.046) (Fig. 2).

Fig. 1
figure 1

All-cause mortality forest plot. Model 1 adjusts for randomized treatment group. Model 2 additionally adjusts for log (CRP), albumin, ESA dose, vascular access status, smoking (current former, never), cause of end stage kidney disease (ESKD), age, duration of dialysis and histories (all yes/no) of myocardial infarction (MI), heart failure, atrial fibrillation, peripheral arterial disease and diabetes. PCS physical component score, MCS mental component score. Results are Hazard ratio (HR) and 95% confidence interval (CI). Values for EQ5D index are per 0.1, while all others are per 10

Full size image
Fig. 2
figure 2

Primary endpoint forest plot (fatal and non fatal myocardial infarction, stroke, hospitalisations for heart failure and all-cause mortality). Model 1 adjusts for randomized treatment group. Model 2 additionally adjusts for log (CRP), albumin, ESA dose, vascular access status, smoking (current former, never), cause of renal disease, age, duration of dialysis and histories (all yes/no) of MI, heart failure, atrial fibrillation, peripheral vascular disease and diabetes. Results are Hazard ratio (HR) and 95% confidence interval (CI). Values for EQ5D index are per 0.1, while all others are per 10

Full size image

Key laboratory measurements associated with poor QoL were a high CRP level and low TSAT, whilst haemoglobin and ferritin concentrations were not independent predictors of QoL. A low TSAT was an independent predictor of KD-QoL PCS (Suppl Table 2). Variables associated with reduced QoL scores at baseline included age (≤ 65 years) sex (female), BMI (> 27.5), and a history of stroke.

Baseline predictors of quality of life

Variables associated with EQ5D index and EQ5D VAS

At baseline, the mean (SD) EQ5D index overall was 0.68 (0.26) and the mean (SD) overall EQ5D VAS was 60.7 (20.8). In the univariate analysis, being female (p = 0.006 and 0.015), having a higher BMI (p < 0.001: BMI > 27.5) and the presence of more co-morbidities, including diabetes mellitus, a history of MI, stroke or heart failure were associated with significantly worse EQ5D index and EQ5D VAS scores (Table 1 and Suppl Table 1). The lowest scores and hence worse measures were observed in patients with a history of heart failure or stroke. A high CRP level was a consistent association with lower EQ5D index and EQ5D VAS scores.

A Hb level > 100 g/L and TSAT > 20% without raised inflammatory markers (CRP) was associated with better EQ5D index and EQ5D VAS scores in univariable analysis (Table 1). However, after multivariable analysis, haemoglobin was not an independent predictor of any aspect of quality-of-life score at baseline (Suppl Table 4). The variables independently associated with EQ5D index were age, CRP, BMI, sex, heart failure and stroke, and those independently predictive of worse EQ5D VAS score were younger age, higher CRP, female sex and stroke.

Factors associated with KD-QoL PCS and MCS score

At baseline, the mean (SD) PCS overall was 33.7 (10.2) and the mean (SD) overall MCS was 46.0 (11.3). The differences observed across the subgroups examined were qualitatively consistent with those observed for the EQ5D index and EQ5D VAS, although atrial fibrillation was associated with a lower PCS than no atrial fibrillation (Table 2). PCS was substantially lower than MCS in all subgroups examined.

Table 2 Comparative statistical analysis of the mean scores of physical component score (PCS) and mental component score (MCS) summaries of patients with end stage kidney disease on HD, according to the categorical socio-demographic variables and categorical clinical and laboratory parameters
Full size table

Older age was associated with a better MCS score. After adjustment, a history of stroke (p = 0.048) and myocardial infarction (p = 0.01) remained independent predictors of a worse MCS, while transferrin saturation (of > 20%) was a strong independent predictor of a better PCS score (p = 0.005) (Suppl Table 2). Serum ferritin had no independent effect on QoL scores in any domain analysed (Suppl Table 3).

Factors associated with KD-QoL eight constitute domains analysis of PCS and MCS

Overall scores for RP and RE were much lower than other domains, while PF was lower than the others. Further comparative statistical analysis of the 8 domains of the KD-QoL scores demonstrated higher scores in PF, BP, GH, and VT in males compared to females (p < 0.01; < 0.01; < 0.012 and < 0.003, respectively) (Suppl Table 4). Higher CRP was associated with lower scores on all 8 subdomains including age, sex, BMI, comorbidities of diabetes, stroke and heart failure. In univariate analysis a low haemoglobin level and low TSAT were associated with the domains BP, GH, and social functioning (SF), (Suppl Table 4).

Factors associated with kidney-specific scales: KD-Qol Burden, Symptoms and Effects of kidney disease subscales

The comparative statistical analysis of the scores of KD-QoL burden, symptoms and effects of kidney disease according to the categorical socio-demographic variables and categorical clinical and laboratory parameters are shown in Suppl Table 5. The findings related to burden, symptoms and effects again indicated that males (p = 0.004, 0.046, 0.005, respectively, for burden, symptoms and effects) were independently less affected; the presence of diabetes led to significantly worse scores in all 3 domains (p < 0.001) and the presence of co-morbidities was less important. A higher CRP was again an important significant independent predictive factor of burden (p = 0.001), symptoms (p = 0.004) and effects (p = 0.04) (Suppl Table 5). A low haemoglobin level was associated with increased burden but not symptoms or effects of renal disease.

Discussion

In this post hoc analysis of a large cohort of patients within the first year of commencing haemodialysis in the UK, overall QoL scores were 30% lower than the general population and similar to those found in prevalent dialysis patients, approximately 10% lower than patients on peritoneal dialysis and approximately 12% lower in comparison to kidney transplant patients with regard to the PCS [6, 13, 14]. Specific reductions were seen in our cohort in the subdomains of physical function, role physical and role emotional but not renal-specific subdomains. A worse QoL at baseline was influenced by several parameters, the most consistent independent variable being a high CRP level. Our study showed that at baseline these factors were consistent across several QoL scores, and baseline QoL was predictive of all-cause mortality and the primary outcome measure. This has not been previously studied in patients within the first year of haemodialysis.

In patients with chronic kidney disease not on dialysis, scores tend to be higher and associations exist with hospitalisations. There were strikingly low scores for heart failure and stroke which is not surprising given their association on health. In the Therapeutic Response Evaluation and Adherence Trial (TREAT) trial in patients with CKD and type 2 diabetes, treatment with darbepoetin led to a small improvement in fatigue and overall QoL over placebo, although there was no benefit on other domains [15]. Despite finding in this study that low transferrin saturation or haemoglobin was predictive of poor QoL, the subsequent follow-up data on proactive versus reactive iron did not show any significant benefit which might relate in part to the quantity of missing follow-up data, hence this was not addressed in this current study. Also the lack of association, after adjustment, between a higher haemoglobin level and better measures of health status or quality of life may be due to the relatively small range of haemoglobin seen in the study leading to relatively small corrections in haemoglobin.

Use of the disease-specific elements of the KD-QoL did not add any further information to the generic scores. These again were approximately 15% lower than those in transplant patients but higher than those in prevalent dialysis patients [13, 14].

Our data confirmed that both the PCS component and the MCS component of QoL is lower in the elderly and is associated with death and hospitalisation regardless of demographic data [16]. They were also associated with the composite outcome of all-cause death, myocardial infarction, stroke or hospitalisation for heart failure. Quality of life was an independent prognostic predictor and was associated with mortality, consistent with other studies [15], and increased hospitalisations [17, 18]. Indeed, impaired functional status is associated with early death after commencement of haemodialysis. [19, 20]. Our patient group included incident patients, some of whom would have been in the first three months of dialysis and therefore this may have influenced the association. However, with such a large data set of over 2000 dialysis patients the results remain important and warrant consideration of methods of intervention to reduce the future impact. Both physical inactivity and impaired physical function are strongly associated with increased morbidity, mortality and reduced QoL. Reduced QoL is also independently associated with mortality in patients on haemodialysis [5].

Male patients had better QoL scores, especially for KD-QoL burden, symptoms and effects, but age had less of an association. This was seen in a previous systematic review and meta-analysis and analysis of the Dialysis Outcomes and Practice Patterns Study (DoPPS) data [21, 22]. Chesnaye and colleagues in a recent analysis in patients with advanced chronic kidney disease also noted this association and provided several possible explanations including reduced reporting of physical weakness by men, the applications of different coping strategies, the different perceptions of symptom severity and the higher rates of depression and anxiety in women [23]. A recent analysis of the DoPPS data concerning dialysis patients in Japan found that in 892 maintenance HD patients, those > 70–79 years or > 80 years had lower PCS scores compared to those aged > 60 years (43.1 vs 35.2) [24]. Again in a population of 980 dialysis patients in Singapore, age and male sex had higher scores [13]. Interestingly, in our population, the scores were much lower (34.4 vs 33.6) and were not significantly different. In addition, whereas there was a significant difference in MCS scores in our study which was better in the older age group, Ishiwatari et al. [24] did not find a significant difference in their population of Japanese patients. The reasons for this are unclear but may relate to the age differences examined or cultural differences between the two populations.

Our patient population was identified on the basis of laboratory biomarkers of anaemia and iron deficiency. Both Hb and TSAT led to a statistically significant difference in QoL, with a higher Hb and higher TSAT reflecting better baseline QoL scores for EQ5D index and EQ5D VAS, and amongst the KD-QoL scores, BP,GH, and SF but not VT or PF as seen in other studies of patients with CKD or on dialysis therapy [25,26,27,28,29,30]. However, after analysis for independent variables, only a lower TSAT was an independent predictor of worse QoL for PCS. It may be that TSAT is a better predictor of functional iron deficiency which is associated with functional capacity and possible fatigue scores [31,32,33,34]. The TREAT study led to an improvement in The Functional Assessment of Cancer Therapy-Fatigue (FACT-Fatigue) score in these studies with ESA therapy [15, 35], while the Correction of Hemoglobin Outcomes in Renal Insufficiency Study (CHOIR) demonstrated an improvement in the linear analogue scale assessment in both groups (higher and lower Hb groups) and a trend for increased energy scores in the lower Hb group [36, 37]. However, there was no difference in the KD-QoL scores of energy or physical functioning domains in these studies [15, 35,36,37]. TREAT demonstrated that a 5 point or greater increase in score was clinically meaningful, leading to better outcomes (54% in the ESA group vs 49% in the placebo group; p = 0.027). Our study showed that QoL scores, independent of Hb or iron status at baseline, were predictive of mortality and the primary outcome measure. This finding is perhaps important when considering those patients commencing dialysis therapy, at least in the UK. It would suggest that lower QoL scores are perhaps simply a marker of bad outcomes. Whether trying to improve these factors which affect quality of life, such as anaemia and iron deficiency, prior to commencing dialysis with maximization of preventative therapies to minimize cardiovascular risk (reducing lipids, glycemic control) may impact Quality of life or mortality on dialysis is unknown. In addition, consideration of dialysis therapy itself and strategies such as incremental dialysis which impact patient wellbeing also remains unknown [38, 39].

There have been recent developments in assessing patient-related outcomes measures both generally in patients with CKD and specifically in those with associated anaemia, nonetheless, several of these developments require validation. Health-related QoL represents a patient’s overall perception of the effect of disease and treatment on their QoL, thus measures of health-related QoL are often measures of self-perceived health status. Such measures do not directly capture how patients feel and function. Assessment of patient well-being is valuable for optimal patient care. The SF-36 questionnaire has been widely used to evaluate the impact of anaemia in CKD trials and the vitality domain has been shown to correlate with Hb level and to be beneficially affected by treatment [40]. The Physical function scales and the FACIT-An Total, Fatigue and Anaemia scales, are also reliable and valid measures for assessing health-related quality of life in anaemia associated with CKD [41]. Patient reported outcome measures are being developed [42, 43]. However, the chronic kidney disease Anaemia Questionnaire (CKD-AQ) is the latest advance and appears to be a reliable and sound patient-related outcome measure for use in patients with anaemia of CKD [44]. The objectives of the CKD-AQ questionnaire are to assess the frequency, duration and severity of symptoms and impact associated with anaemia in CKD.

Our study has several strengths and limitations. This was a large study encompassing a diverse real world UK dialysis population with good baseline data collection. However, a prospective analysis of QoL over time was futile due to the volume of missing data. Another possible limitation is the lack of collection of socio- economic level data which has been previously shown to be associated with outcomes [45]. Ferritin was not adjusted for CRP and hence it cannot be said that this is not an explanation for the lack of association of ferritin with QoL. Other possible variables that could have influenced the outcomes and QoL, such as the impact of central venous catheter vs arteriovenous fistula or dialysis adequacy, were not studied in the current model. In addition, it is recognised that within the first three months of commencing dialysis, patients have worse quality of life measures and outcomes; this was not included in the modeling.

A final consideration is that health-related QoL is a term that is commonly and misleadingly used to refer to health status tools such as the KD-QoL and EQ-5D, which may assess health and functioning but do not assess the association of health and functioning on QoL [46, 47]. Health status includes aspects of a person’s life such as their physical ability, daily functioning, and experience of symptoms. Indeed, it is controversial whether these health-status measures should be considered measures of quality of life at all [46]. Therefore, assumptions about the overall quality of life of individual patients should not be based on measures of their health status alone.

Conclusions

Overall, we have shown that in the first year of commencing haemodialysis, baseline QoL is predictive of mortality and the primary outcome measure of the PIVOTAL trial. Several factors influence a poorer quality of life score including female gender, the presence of diabetes and other co-morbidities. Markers of iron deficiency (a low TSAT) and inflammation (a high CRP) were also associated with poorer outcomes in many domains of QoL. Possible earlier optimization of patients (such as reducing cardiovascular risk factors, diabetes control, correcting iron deficiency and lifestyle changes) prior to commencement of dialysis so as to tackle those modifiable factors associated with poor QoL scores needs attention to potentially reduce future mortality.