Quality of Life and Physical Function in Older Patients on Dialysis

A Comparison of Assisted Peritoneal Dialysis with Hemodialysis

Abstract

Background and objectives 

In-center hemodialysis (HD) is often the default dialysis modality for older patients. Few centers use assisted peritoneal dialysis (PD), which enables treatment at home. This observational study compared quality of life (QoL) and physical function between older patients on assisted PD and HD.

Design, setting, participants, & measurements 

Patients on assisted PD who were >60 years old and on dialysis for >3 months were recruited and matched to patients on HD (needing hospital transport) by age, sex, diabetes, dialysis vintage, ethnicity, and index of deprivation. Frailty was assessed using the Clinical Frailty Scale. QoL assessments included Hospital Anxiety and Depression Scale (HADS), Short Form-12, Palliative Outcomes Symptom Scale (renal), Illness Intrusiveness Rating Scale, and Renal Treatment Satisfaction Questionnaire (RTSQ). Physical function was evaluated by Barthel Score and timed up and go test.

Results 

In total, 251 patients (129 PD and 122 HD) were recruited. In unadjusted analysis, patients on assisted PD had a higher prevalence of possible depression (HADS>8; PD=38.8%; HD=23.8%; P=0.05) and higher HADS depression score (median: PD=6; HD=5; P=0.05) but higher RTSQ scores (median: PD=55; HD=51; P<0.01). In a generalized linear regression model adjusting for age, sex, comorbidity, dialysis vintage, and frailty, assisted PD continued to be associated with higher RTSQ scores (P=0.04) but not with other QoL measures.

Conclusions 

There are no differences in measures of QoL and physical function between older patients on assisted PD and comparable patients on HD, except for treatment satisfaction, which is higher in patients on PD. Assisted PD should be considered as an alternative to HD for older patients, allowing them to make their preferred choices.

Introduction

The management of older patients with advanced kidney disease remains a challenge. In 2012, the United Kingdom Renal Registry reported that the peak rates for starting RRT were in the 75- to 79-year-old age group for women and 80- to 84-year-old age group for men (¹). In the United States, the highest growth rate for starting RRT is for those ≥85 years old (²). Older patients tend to present later for dialysis (³), with multiple comorbidities, a higher risk of cognitive dysfunction (⁴), frailty (^5,6), and sensory impairments (⁷) as well as functional and psychologic dependence (⁸). These limitations have traditionally restricted the use of home-based therapies, including peritoneal dialysis (PD), in older people. However, with education and the availability of assistance, 60% of eligible older patients prefer PD, 40% of whom start dialysis on PD (⁹).

In the United Kingdom, older patients on dialysis typically receive in-center hemodialysis (HD), with only 10.5% of patients ≥75 years old on prevalent dialysis being on PD (¹⁰). This pattern is replicated in many European countries, with patients ≥70 years old being 56% less likely to receive PD than those between 20 and 44 years of age (¹¹). In contrast, dialysis delivery in France has been by assisted PD provided by private community nurses (^12,13). This has enabled older patients to dialyze at home, with PD being chosen more often than in-center HD by patients >85 years old (¹⁴). Assisted PD is becoming increasingly available as a model of care for older patients in other European countries (¹⁵).

In the United Kingdom, the use of assisted PD provided by the National Health Service has increased slowly since 2006. This involves the use of overnight cyclers (automated PD), which are set up by family members or health care assistants who visit once daily. The patients and their families remain responsible for connection to and disconnection from the cycling machine (¹⁶). Assisted PD is used to allow existing patients to continue PD as they become frailer or for patients who are considered too frail for HD. The restricted use of assisted PD is partly explained by limitations in current evidence.

Several assisted PD programs have reported on patient survival (^17,18) and technique survival (¹⁹), with outcomes comparable with those of self-care PD. Quality of life (QoL), an important outcome in older patients, has been shown to influence decisions regarding modality choice. Two studies have compared QoL on HD and self-care PD in older patients. The North Thames Dialysis Study found no difference in QoL between patients >70 years old on PD or HD (²⁰). In the more recent Broadening Options for Long-Term Dialysis in the Elderly (BOLDE) Study, older patients on PD experienced less illness intrusion compared with those on HD after adjustment for comorbidities and other confounders (²¹). There are, however, no studies comparing QoL between older patients on assisted PD and the more prevalent HD.

We hypothesize that QoL is better for older patients on assisted PD compared with similar patients receiving in-center HD. The Frail and Elderly Patient Outcomes on Dialysis (FEPOD) Study is an observational study that aims to compare QoL and physical function between older patients on assisted PD and HD.

Materials and Methods

Patients were recruited from 22 renal centers in England and Northern Ireland. The study sample consists of two subsets. The FEPOD 1 subset (funded by The Dunhill Medical Trust) included patients from centers in London and Northern Ireland. The study was expanded with additional funding (a Baxter Healthcare Clinical Evidence Grant) to include a second subset of patients from other centers in England (FEPOD 2). Ethical approval was obtained from the National Research Ethics Committee (London–Fulham; reference nos. 11/LO/1428 and 11/LO/1886).

Inclusion and Exclusion Criteria

Patients on assisted PD were defined as being unable to perform PD at home without assistance from paid health care workers or family members. Patients on HD were eligible if they required hospital transport to attend dialysis sessions. All were ≥60 years old, on dialysis for ≥3 months, and free from hospitalization for ≥30 days. Patients with known cognitive impairment, unable to understand English, or with a life expectancy of <6 months were excluded from the study.

Subjects and Recruitment

Because of fewer patients eligible for assisted PD, these were recruited first. Each was matched by the study team to an eligible HD patient from the same center by age (±2 years), sex, diabetes status, time on dialysis (±1 year), ethnicity (where possible), and socioeconomic status as determined by the Index of Deprivation 2007 (²²). Matching was performed to ensure that both cohorts had similar baseline characteristics and reduce the influence of known confounders.

Study Visit

The study visit consisted of sequential QoL and physical function assessments performed in a standard order. Demographic and clinical characteristics were collected from medical records and during the assessment. Comorbidities were evaluated using the Stoke–Davies comorbidity score (²³). Frailty was also assessed by the researcher during the study visit. For patients on HD, assessments were conducted on a nondialysis day or before dialysis. Cognitive function and timed up and go tests were conducted in the FEPOD 1 subset of 106 patients.

Outcome Measures

QoL Measures.

Short Form-12, Version 2.

The Short Form-12 (SF-12) is a self-assessment of physical and mental health and has two scores: the Physical Component Summary (PCS) scale and the Mental Component Summary (MCS) scale. As an abbreviated version of the SF-36, it minimizes the burden of completion in older people. With >90% agreement between the SF-12 and SF-36, population norms for the SF-36 can be used to interpret SF-12 results (^24,25).

Hospital Anxiety and Depression Scale.

The Hospital Anxiety and Depression Scale (HADS) is a screening tool with scores ranging from zero to 21 for either depression or anxiety. A depression score of eight and above indicates possible depression (²⁶). Screening for depression in patients on dialysis using HADS correlates well with the gold standard diagnosis for mood disorders (depression) by the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders (²⁷).

Illness Intrusiveness Ratings Scale.

This self-reporting tool assesses the extent to which the illness and/or treatment interferes with 13 life domains. It has been validated in patients with ESRD (²⁸). Scores range from 13 to 91, with a higher score indicating more illness intrusion.

Symptoms.

Symptoms were measured using the Palliative Care Outcome Scale-S renal scale. This is an adaptation of the Palliative Care Outcome Scale, which was originally developed for oncology patients, allowing for renal-specific symptoms (²⁹).

Satisfaction with Treatment.

Satisfaction with treatment was assessed using the Renal Treatment Satisfaction Questionnaire (RTSQ) (³⁰). This 11-item tool has been validated in patients on HD, patients on PD, and patients with transplants.

Physical Function.

Barthel Index.

This index evaluates performance in activities of daily living (³¹). Scores range from 0 to 100, with lower scores suggesting increased disability.

Timed Up and Go Test.

This test times the patient standing up from a chair, walking a fixed distance, turning, walking back, and sitting down (³²). Completion time ≥14 seconds is associated with an increased risk of falls.

Other Clinical Assessments

Frailty.

The Canadian Study of Health and Aging Clinical Frailty Scale was used to measure frailty. Scores range from one to seven, with higher scores corresponding to increasing levels of frailty. It evaluates dependence for activities of daily living and has been shown to be predictive of death or the need to be institutionalized, similar to other established measures of frailty (³³).

Cognitive Function.

Miniature Mental State Examination.

This test is routinely used in clinical practice (³⁴), with available normative data stratified by age groups and education levels (³⁵). Scores range from one to 30, with a score <24 suggesting cognitive impairment.

Trail-Making Test B.

This test measures the time taken to follow a trail of alternating numbers and alphabets (³⁶). Completion time ≥300 seconds suggests executive dysfunction.

Others.

Questionnaires were designed to assess social networks, falls, and health care use.

Statistical Analyses

All statistical analyses were performed using SAS software (version 9.3; SAS Institute Inc., Cary, NC). Categorical variables are presented as percentages, and continuous variables are presented as median values with interquartile ranges. Mann–Whitney, Pearson chi-squared, or Fisher exact tests were used to compare baseline clinical characteristics between PD and HD. Statistical significance tests were not used to compare baseline demographic characteristics, because both samples were matched on demographics and therefore, do not constitute random samples from the respective populations. Relationships between QoL measures and participant characteristics were analyzed using generalized linear models with a γ-error structure, because the assumption of normality was not justified for most of the outcome measures. This modeling methodology followed standard practice for multiple regression analysis. Both univariate analyses and multivariate analyses were performed, the latter indicating the effect of the covariate on the outcome measure after adjustment for all other covariates in the model. All models adjusted for age, sex, dialysis vintage, Stoke–Davies comorbidity score, and frailty score in addition to dialysis modality. These covariates were selected a priori as potential confounders. Effect estimates were derived from the γ-regression models. A post hoc analysis of frail patients (frailty score ≥5) was also performed using the same method.

The relationship between dialysis modality and possible depression (HADS depression score ≥8) was evaluated using logistic regression analysis adjusting for baseline participant characteristics as stated above.

Because the models studied associations between participant characteristics and several outcomes, all P values were adjusted for multiple significance testing using the false discovery rate adjustment by Benjamini and Hochberg (³⁷). This adjustment provides strong protection against false-positive associations and has been recommended for use in health studies in a recent review (³⁸).

Power Calculation

A retrospective power calculation suggests that the study sample size would have 80% power to detect associations, which gives a small increase in multiple correlation (0.028) in a model that has low overall multiple correlation (0.100). Multiple regression sample size calculations are on the basis of many assumptions, and this retrospective power calculation should be regarded as approximate. However, it suggests that this sample size is capable of detecting relatively small associations.

Results

Patient Characteristics

In total, 251 patients (129 PD and 122 HD) were recruited. Seven patients on assisted PD included in the analysis remained unmatched. The assisted PD and HD groups were well matched by age, ethnicity, sex, dialysis vintage, diabetes status, and index of deprivation (Table 1); 53% of the patients on assisted PD connected and disconnected themselves to/from the cycling machine, whereas 40% and 7% were connected and disconnected by family and a paid caregiver, respectively. However, they all required assistance with the PD fluid bags and setting up the cycler.

Table 1: Demographic characteristics of the studied cohort

Clinical Characteristics

Table 2 shows the clinical characteristics for the cohort; 48% met the criteria for frailty (frailty scores ≥5; 51.9% PD; 42.6% HD; P=0.32). The comorbidity burden was not different between the groups; 46% of the study group had been admitted in the previous 3 months, and 33% sustained one or more falls in the preceding 6 months. Cognitive function was assessed in the FEPOD 1 subset (n=106 patients). Only four patients (4.8%) had abnormal Mini Mental State Examination scores. In contrast, 41% had executive dysfunction (Trail-Making Test B time >300 seconds), with no significant difference in prevalence between patients on assisted PD and patients on HD; 38% of patients with executive dysfunction met the criteria for possible depression (HADS depression score ≥8).

Table 2: Clinical characteristics by dialysis modality

Study Outcomes

Table 3 presents the unadjusted QoL measures for patients on assisted PD and patients on HD. The assisted PD group had higher HADS depression scores (median: PD=6; interquartile range [IQR], 3–7.75; HD=5; IQR, 3–8.75; P=0.05) and a higher prevalence of possible depression (38.8% versus 23.8%; P=0.05). Conversely, the RTSQ score was higher in patients on assisted PD (median: 55; IQR, 48–59.75 versus 51; IQR, 44–57; P<0.01). There were no significant differences in the other QoL measures.

Table 3: Univariate comparison: Measures of quality of life and physical function

In multivariate analysis using generalized linear models (each adjusted for age, sex, dialysis vintage, comorbidity score, and frailty), HD was associated with lower RTSQ scores (P=0.04) compared with assisted PD. There were no significant differences in other QoL measures between HD and assisted PD. Frailty was associated with worse SF-12 MCS, SF-12 PCS, Barthel Index, symptoms, illness intrusion, and HADS scores (P<0.01), whereas age was associated with lower illness intrusion (P<0.01). Table 4 shows the effect estimates of each covariate on each QoL measure.

Table 4: Effect estimates from multivariate analysis using generalized linear models

In a post hoc analysis of 119 frail patients (frailty score ≥5), there was no significant difference in any QoL measure between assisted PD and HD after adjusting for the other covariates (Table 5).

Table 5: Multivariate analysis using generalized linear models in patients with frailty score ≥5 (n=119)

In logistic regression analysis with possible depression (HADS depression score ≥8) as the outcome variable, there was no difference in the odds for depression between assisted PD and HD. However, the odds for depression were 53% higher for every unit increase in the frailty score (odds ratio, 1.53; 95% confidence interval, 1.12 to 2.07).

Discussion

The principal determinants of QoL as rated by the older person are being independent and being in control of one’s own life (³⁹). We, therefore, postulated a priori that QoL is better for patients on assisted PD compared with patients on in-center HD. However, the study showed no differences in measures of QoL apart from treatment satisfaction, which was higher in older patients on assisted PD compared with patients on HD.

Reports on the value of assisted PD in older patients have been published recently from developed and emerging economies (^9,40,41). A registry study from France, which has the largest experience with assisted PD (^12,13), reported a median survival of 27.1 months in 1613 patients >75 years old on PD (89% on assisted PD) (¹⁷). This compares with a median survival of 2.4 years for patients of the same age but mostly on HD in the United Kingdom (⁴²).

Two prior studies compared QoL in older patients on self-care PD and HD. They also found no significant difference in outcomes, similar to the FEPOD Study. The BOLDE Study did, however, report less illness intrusion in the PD group (^20,21). The SF-12 scores were lower in the FEPOD Study cohort compared with those in the BOLDE Study, with higher illness intrusion scores. Possible depression was also more prevalent in the FEPOD Study compared with the BOLDE Study (32% versus 18%). These differences may reflect the influence of frailty in our cohort, a common indication for assisted PD.

Although dialysis modality was not statistically significantly associated with measures of QoL in either unadjusted or adjusted analysis, higher frailty scores were associated with lower QoL, even after adjustment for dialysis modality and other patient characteristics. Frailty is characterized by a lack of physiologic reserve and an impaired response to acute stressors. It is predictive of increasing disability, hospitalization, and mortality (⁴³); 48% of our study participants were characterized as frail (frailty score ≤5). Collectively, these findings may suggest that, among patients requiring assistance with dialysis treatments, choice of modality is a less important determinant of QoL than characteristics, such as frailty, that drive the need for assistance.

Because one half of the patients on PD were not frail, there would have been other indications for assistance (e.g., visual impairment or stroke).There are no standardized criteria for assisted PD other than a requirement for assistance to enable PD at home. We did not assess the indications for assisted PD at individual centers.

Treatment satisfaction (RTSQ) scores were higher in patients on assisted PD compared with those receiving in-center HD, even when baseline characteristics were considered. This is consistent with the results from comparative studies involving patients on self-care PD (^44,45). One may speculate that assisted PD at home increases the likelihood of maintaining independence and control, leading to higher treatment satisfaction in older patients.

This study has some notable limitations. The cross-sectional design means that results are limited to associations and not causality. The timing of the two sources of funding also affected the study design. Cognitive function and timed up and go tests were, therefore, not assessed in all patients. To limit confounding, patients on assisted PD were recruited and matched to eligible patients on HD requiring hospital transport at individual centers. We have, therefore, been unable to collect information about the denominator PD and HD populations. Because dialysis transport in the United Kingdom is most often provided (and funded) for multimorbid, dependent patients, this criterion was included to help ensure that suitable HD matches were recruited. It is, however, recognized that this matching process may increase the risk of selection bias. In addition, the findings may not be generalizable, because the frailest patients with short life expectancy and significant cognitive impairment were excluded. Study outcomes may also differ in a population with a different model of assisted PD compared with that in the United Kingdom. The sample size was limited by the number of patients eligible for assisted PD, although a retrospective power calculation suggests that the study is suitably sized to detect associations. Despite these limitations, this is the largest study to compare QoL between patients on assisted PD and patients on HD.

In summary, our study has shown that QoL measures do not differ between older patients on assisted PD and comparable patients receiving in-center HD, with the exception of treatment satisfaction. This is higher for patients on assisted PD compared with patients on HD. QoL in these patients may be influenced by nondialysis factors, including frailty. The morbidity risk is also considerable, with a high prevalence of falls and hospitalization. These issues should be discussed with older patients with advanced kidney disease and their families to allow them choose the most appropriate therapy (^46,47).

Longitudinal studies are needed to determine comparative outcomes, use of health care resources, and overall costs. This study suggests that assisted PD should be considered as an alternative to in-center HD for older patients, at least from the viewpoint of QoL.

Disclosures

E.A.B. has received speaker honoraria and research funding from Baxter Healthcare, Berkshire, United Kingdom. All other authors have declared no conflict of interest.

Acknowledgments

This work was supported by The Dunhill Medical Trust Grant R202/0511, Baxter Clinical Evidence Council Grant 11CECPDEU1006, the Imperial National Institute for Health Research Biomedical Research Centre, and the National Institute for Health Research through the Comprehensive Clinical Research Network.

We thank the Frail and Elderly Patient Outcomes on Dialysis Study investigators and centers: Peter Maxwell, Belfast Health and Social Care Trust; John Stoves, Bradford Teaching Hospitals; Anand Vardhan, Central Manchester University Hospitals; Richard Fluck, Derby City General Hospital; E.A.B., Hammersmith Hospital; Helen Collinson, Hull Royal Infirmary; Sally Krause, Kent and Canterbury Hospital; Maxine Keddo, Kings College Hospital; Graham Warwick, Leicester General Hospital; K.F., Lister Hospital; Richard Fielding, Newcastle Hospitals; Sarah Jenkins, Northern General Hospital; Camille Harron, Northern HSC Trust; A.D., Royal Free Hospital; Stan Fan, Royal London Hospital; David Lewis, Salford Royal Hospital; Neal Morgan, Southern HSC Trust; Steve Nelson, St. George’s Hospital; Hugh Gallagher, St. Helier Hospital; Graham Woodrow, St. James University Hospital; Michael Quinn, Western HSC Trust; and Simon Davies, University Hospitals of North Staffordshire.