The study employed a cross-sectional design. Patients were recruited through the official German national patient association for kidney disease (Bundesverband Niere e.V.), in cooperation with Fresenius Medical Care Germany. An information letter, informed consent form, and the study questionnaire were published in the association’s quarterly journal Der Nierenpatient. Patients could detach the questionnaire from the journal, sign the informed consent form prior to completing it, and return both anonymously in a prepaid envelope to the patient association. Alternatively, after providing informed consent, participants had the option to complete the survey online via the survey tool LimeSurvey. Thus, participants could choose their preferred mode of participation and were given ten weeks to respond following publication in the journal. Inclusion criteria were age ≥18 years, a diagnosis of CKD, and residence in Germany. CKD status and treatment modality were self-reported.

A power analysis was conducted using GPower (32). Under a significant level of.05, a power of 95%, in total four covariates and medium effect size, a minimum of total 357 participants was required.

Ethical approval for the study was obtained from the Psychology Research Ethics Committee of Leiden University (proposal number CEP 18-0516/257). Data will be shared upon reasonable request to the first author.

The Short Form (SF)-12 (33) was used to assess physical and mental HRQoL. Twelve statements with varying response categories were converted into physical and mental standardized values. From the standardized values, a norm-based T-score is calculated for the physical and mental component, with higher scores indicating a better HRQoL. The SF-12 demonstrated good reliability, construct validity, and responsiveness (34) and test-retest reliability was supported in clinical patients (35).

Physical activity was assessed with the International Physical Activity Questionnaire-Short Form (IPAQ-SF), a valid and reliable tool (36) providing information on the time spent on moderate and vigorous-intensive physical activity, walking and sitting.

Fatigue was measured using the 20-item Multidimensional Fatigue Inventory (MFI-20) (37), consisting of five scales: general fatigue, physical fatigue, mental fatigue, reduced activity, and reduced motivation. A total MFI score can also be computed. Items were scored on a five-point scale (1 = not true, …, 5 = true). Reversely stated items were recoded and a higher score indicates more fatigue. The instrument has adequate reliability and validity (38). In the current study, reliability is good for all subscales (α = 0.77–0.84) except for reduced motivation (α = 0.61), which has acceptable internal consistency reliability.

Sleep quality was assessed with the 4-item sleep scale of the Kidney Disease Quality of Life - Short Form (KDQOL-SF) (39). Participants rated their sleep quality on a 0 to 10 scale (0 = bad, …, 10 = excellent) and three statements on other sleep problems in the last four weeks using a 6-point scale (1 = never, …, 6 = always). Scores were transformed and summed on a 0 to 100 scale. In the current study, internal consistency reliability is good (α = 0.75). Quantity of sleep was measured by self-reported effective sleeping time in hours.

Normative data of the healthy German population on physical and mental HRQoL (40), physical activity (41), fatigue (42) and sleep quantity (43), was compared with the sample data of the study (RQ-A). For the total sample and across gender groups, the average HRQoL score was tested against the population mean using one-sample T-tests. A one-sample Wilcoxon signed-rank test was used to evaluate physical activity against the reported median of the German population. Norm scores for fatigue were reported across age and gender, and were compared within each subgroup using two-samples T-tests. No norm-reference data was available on sleep quality. For the purpose of comparison, sleep quantity (effective sleeping time) of CKD patients was transformed into the categories “short”, “optimal” and “long” as Schlack and colleagues (43) only provided categorical reference data. This study’s sample distribution was tested against their reported percentages using a multinomial goodness-of-fit Chi-squared test. As measure of effect size, the Cohen’s d was used when applying T-tests (≥ 0.20 as “small”; ≥ 0.50 as “medium”; ≥ 0.8 as “large”); phi was used when a Chi-squared test was conducted (0.10 ~ small; 0.30 ~ medium; 0.50 ~ large). To control for multiple testing, p-values were adjusted using Bonferroni corrections.

For the multivariate analyses, a logarithmic transformation was applied to the physical activity variable in accordance with the recommendations of Tabachnick and Fidell (44). To compare fatigue, sleep, physical activity, and physical and mental HRQoL between treatment modalities, ANCOVAs were conducted. The PROCESS macro developed by Hayes (45) was used to test mediation (RQ-C), moderation, and moderated mediation (RQ-D and RQ-E). These analyses, were performed for each aspect of fatigue separately to avoid issues of collinearity among the different MFI (sub)scales. All multivariate analyses were controlled for gender, age, comorbidity, and disease duration. For each model, the assumptions of linearity, homoscedasticity, and normality of residuals were inspected using residual-versus-fitted plots and partial regression plots. Variance Inflation Factor values indicated no problems with multicollinearity (all VIFs < 2). Missing data were handled using pairwise deletion, so that all available data for each analysis could be included.

Statistical analyses were conducted using IBM SPSS Statistics version 22 with the PROCESS macro version 4.2 (45).

A total of 465 patients participated in the study. The average age was 53.78 years (SD = 15.00); 50.3 percent were women. Demographic and medical characteristics of the patient sample can be found in Table 1.

In the CKD patient group, the mean HRQoL scores were significantly lower than the normative means for the physical (M = 39.32 vs. 51.40) and mental component (M = 45.55 vs. 49.30). The associated effect size was medium to large for the physical (d = -1.13), and small for the mental component (d = -0.33). The same pattern of differences was found within the female and male subsamples, respectively (all ps <.001).

Physical activity was significantly lower in the patient group than in the general population. The median score in the patient sample was 2091 (IQR = 3412), which is significantly lower than the median of 5070 reported by Rütten et al. (41) (p <.001). The CKD sample mean (M = 3249.7, SD = 3925.1) was also substantially lower than the reported population mean (M = 8534.2, SD = 9024.5).

An overview of norm comparisons of the MFI subscales, stratified by gender and age, is presented in Figure 1. Higher fatigue was reported in the sample of CKD patients, except for reduced motivation and mental fatigue for the oldest group of women and mental fatigue for the oldest group of men. With respect to the age-groups, the largest differences were found in younger men and women (≤ 40 years; Cohen’s d = 1.02–1.97). With respect to gender, a similar pattern was found for females and males. With respect to the MFI-subscales, overall, the largest differences were found for general fatigue (d = 0.71–1.97), followed by reduced activity (d = 0.62–1.97) and physical fatigue (d = 0.55–1.68); the differences for reduced motivation (d = 0.24–1.44) and mental fatigue (d = 0.13–1.21) were smaller.

A significant difference in sleeping time was found between the CKD sample and the healthy German population, χ²(2) = 32.52, p <.001; φ = 0.27. Short sleep (≤ 5 hours) was more common in the CKD group (21.1% vs. 12.3%), while optimal sleep was less frequent (72.2% vs. 81.6%). Long sleep (≥ 9 hours) occurred at similar rates (6.2% vs. 6.1%). This pattern was significant for both females, χ²(2) = 36.36, p <.001, φ = 0.40, and males, χ²(2) = 6.05, p = .049, φ = 0.17, with a larger effect for females. Female patients reported optimal sleep less often (68.4%) than male patients (76.6%) compared to the norm (80.5%).

Table 2 shows results of mean comparison tests across treatment modalities. Overall, transplant patients had the highest sleep quality, the lowest levels of (general) fatigue and reduced motivation, and the highest mental HRQoL, followed by hemodialysis patients. In contrast, the lowest sleep quality and mental HRQoL, and the highest (general) fatigue was found in the group of patients without RRT. Level of physical activity, physical fatigue, reduced activity, mental fatigue and the physical component of HRQoL was not significantly different across treatment modalities.

In Table 3 (Panel A), the results of the mediation analysis are presented. The indirect effects were all significantly positive. Patients who were physically more active, had a lower level of fatigue, which in turn was related to better physical and mental HRQoL. For physical quality of life, physical fatigue showed the strongest indirect association (β = 0.195); for the mental component of HRQoL, total fatigue showed the largest indirect association (β = 0.243). The smallest indirect effects were found for reduced motivation and mental fatigue. To illustrate the mediating effect of fatigue, results for general fatigue (MFI-G) are presented in Figure 2.

Panel B (Table 3) gives an overview of the moderating effects of sleep quality and sleep quantity. For patients reporting better sleep quality, the association between physical activity and physical fatigue was stronger, indicating a moderating effect of sleep quality (see Figure 3, left). A similar moderating effect was observed for the MFI subscale measuring reduced activity: patients with better sleep quality and higher physical activity levels showed more favorable scores on this subscale (see Figure 3, right). By contrast, sleep quantity did not significantly moderate the association between physical activity and fatigue.

The results of the analysis are shown in Table 3 (Panel C). The indirect association between physical activity and physical and mental HRQoL via physical fatigue was moderated by sleep quality. Specifically, for patients with better sleep quality, the negative association between physical activity and physical fatigue was stronger, which in turn was related to higher HRQoL. A similar moderated mediation was observed for the MFI subscale measuring reduced activity: patients with higher sleep quality and greater physical activity reported less reduced activity, which was associated with higher physical and mental HRQoL.

A major strength of the study is that physical activity, fatigue, and sleep, factors that are all known to influence HRQoL as well as disease progression and mortality in CKD patients (18, 20, 27), were examined together. Another strength is the inclusion of a heterogeneous CKD population, consisting of patients on hemodialysis, peritoneal dialysis, and kidney transplantation, as well as patients not yet receiving RRT. This contrasts with most prior studies, which have typically focused only on hemodialysis populations.

At the same time, several limitations should be acknowledged. First, while mediation and moderation analyses provide valuable insights into potential pathways, the cross-sectional design of the study does not allow conclusions regarding temporal sequence or causality. Longitudinal studies are needed to establish causal inferences. Second, recruitment took place via a patient association newsletter and an anonymous online survey, which may have introduced self-selection bias, for example, by attracting more motivated or health-literate patients. This, in turn, may limit the generalizability of our findings. Third, physical activity was measured using the IPAQ-SF, a widely used tool, but one known to overestimate activity and subject to recall bias. Similarly, sleep was assessed with the abbreviated 4-item KDQOL sleep scale, which may not fully capture all dimensions of sleep quality. Fourth, all measures in the study - including CKD status and treatment modality - were self-reported; no objective measures such as actigraphy for activity (using an accelerometer) or polysomnography for sleep were employed. Additional medical information such as CKD stage and medication use was not available, nor were bloodwork and clinical data (e.g., hemoglobin, eGFR), as the study was conducted anonymously in collaboration with a patient organization. This limits the ability to account for important medical confounders; for instance, anemia or uremic toxins may contribute to fatigue independently of physical activity or sleep. Finally, some CKD subgroups were relatively small, which may have reduced the sensitivity to detect subtle differences across treatment modalities.

Beyond statistical significance, the present findings also have clinical relevance. Fatigue and sleep quality emerged as important factors associated with physical activity and HRQoL in patients with CKD. Although some of the observed effect sizes were modest, they highlight domains directly relevant to patients’ daily functioning and well-being. For example, even small improvements in sleep quality may strengthen the association between physical activity and reduced fatigue, which can translate into better energy levels, adherence to treatment, and overall quality of life. Building on this, the results suggest that interventions incorporating physical activity could be a promising strategy to address both fatigue and quality of life in CKD patients.

At the same time, prescribing exercise to fatigued patients with poor sleep quality may be challenging in practice. Tailored approaches that account for sleep quality, and that integrate sleep-focused strategies before or alongside physical activity, may therefore be more feasible. While previous studies have established that fatigue is related to lower QoL and that exercise is associated with reduced fatigue in CKD (24), the present study extends this work by examining mediated and moderated associations. These findings underscore the relevance of fatigue as a potential pathway and sleep quality as a potential modifier in the relationship between physical activity and HRQoL. To investigate these relationships more rigorously, it will be important to conduct longitudinal studies to further examine the observed associations and to establish temporal sequence and causality. In addition, the use of objective measures for physical activity and sleep would strengthen the validity of future findings. Subsequently, intervention studies are needed to test whether interventions targeting sleep and physical activity have the desired effect on reducing fatigue.

The results also show a substantial gap in fatigue between younger CKD patients and their peers in the general population, suggesting that younger patients may be a primary target group for intervention development. Moreover, the impact of untreated fatigue on later treatment outcomes has not yet been adequately studied from earlier stages of CKD through to renal replacement therapy. Addressing this gap would allow for the more targeted development of interventions across the disease trajectory.

For future research, it will be important to include biological indicators of renal function in addition to the psychosocial variables examined here. This would make it possible to investigate a broader range of factors within the biopsychosocial model of fatigue in CKD patients.