Introduction

The increasing number of patients requiring treatment for end-stage renal disease (ESRD) have led to widespread efforts to promote home-based dialysis. The utilization of home-based dialysis such as peritoneal dialysis (PD), however, remains low relative to hemodialysis [1]. PD offers many potential benefits including increased autonomy and flexibility for patients, better preservation of kidney function and cost savings [2, 3], but it has also been shown to be associated with high emotional distress [4]. Distress has been shown to be associated with poor quality of life, but its role on clinical outcomes for patients on peritoneal dialysis is uncertain [5, 6].

Despite improvements in peritoneal dialysis techniques, there has been little change in patient and technique survival rates. Most studies on risk factors for poor clinical outcomes in PD patients focus on comorbidities (mainly diabetes and cardiovascular disease), residual renal function, inflammation and nutrition [7–9]. Consideration of the role of psychosocial factors such as symptoms of distress and perceived social support on PD outcomes is limited despite evidence for their prognostic value for mortality in hemodialysis [10, 11]. The purpose of this prospective single-cohort study was to evaluate the effect of depression, anxiety and social support on 1-year prognosis (patient and technique survival) in PD patients.

Methods

Participants

Participants were consecutively recruited from 2010 to 2011 from Singapore General Hospital, Singapore, if (1) they were aged ≥21, (2) on PD for at least 3 months, (3) not hospitalized at the time of assessment or the preceding 3 weeks, (4) no dementia or psychiatric diagnosis and (5) able to communicate in English, Mandarin or Malay. Patients were free to choose their preferred language of questionnaires (English, Malay or Mandarin). The study was approved by Centralized Institutional Review Board, SingHealth Research Facilities (reference number: 2010/588/E), and informed consent was obtained.

Measures

Study participants completed the Hospital Anxiety Depression Scale [12] and measures of social support (i.e., Kidney Disease Quality Of Life Short Form; KDQOL-SF subscales [13]: social support, quality of social interaction, staff encouragement) at baseline and were followed up for 1 year for technique and actuarial patient survival.

Both the KDQOL-SF [14] and the HADS have been linguistically validated in both Mandarin and Malay [15, 16], which is a paramount consideration for use in the local context to ensure fair representation of Singaporean population. Given the linguistic diversity in our sample [N = 104 (51.7 %) patients completed the English HADS, N = 29 (14.5 %) the Malay version of HADS, and N = 68 (33.8 %) the Mandarin HADS], we opted for the general cutoffs for classification of anxiety and depression as these have been well validated across a range of patient populations, cultures and languages [17]. Dialysis-specific cutoff thresholds have yet to be established conclusively in other languages with recent validation studies proposing different cutoff scores ranging from 6 to 8 [18, 19].

Demographic information on age, gender, relationship status, race/ethnicity, employment status, educational level and household monthly income was captured with self-report. Laboratory parameters (e.g., albumin) and clinical data (e.g., dialysis adequacy, dialysis vintage, PD dose/regimes, comorbidities) and characteristics (e.g., PD modality; assisted PD status) were extracted from medical records at study enrollment. Comorbidities were also abstracted using the Charlson comorbidity index [20].

Clinical endpoint

The primary endpoint was the combined event of all-cause mortality and PD technique failure. Survival time was calculated as number of months from baseline assessment (time 0) until one of the following: death, switch to hemodialysis, kidney transplant, transfer out of the facility or end of the study observation period (February 2012, at a maximum of 29 months; M = 13.58 ± 5.27). Status was monitored and verified through medical records and primary renal physician. For sensitivity analyses (patient and technique), survival times were also calculated based on the date of initiation onto PD (time 0) and analyses were repeated using the continuous anxiety and depression scores.

Statistical analyses

Mean ± standard deviation or percentage was determined for patient demographics, baseline clinical/laboratory results and outcomes, as appropriate. The differences between distress subgroups (as per clinical cutoffs) were analyzed by Chi-square test for categorical variables and by one-way analysis of variance test for continuous variables. Statistical procedures included Kaplan–Meier survival curves and Cox regressions. Variables associated with the combined clinical endpoint in the univariate analyses were considered for multivariate modeling. Multivariate Cox proportional-hazards model was performed to assess the relations between the combined adverse outcome and depression, anxiety controlling for significant sociodemographic, clinical covariates. Effects were considered significant when the p value was less than .05. All analyses were performed by using the SPSS version 22.

Results

Study sample

Of those eligible (N = 241), a total of 201 PD patients (age, 58.9 ± 12.59 years; 45 % men) were studied (response rate 83.3 %). A total 59.7 % of the participants were indicated with depression and 41.5 % with anxiety at baseline based on recommended clinical cutoff of ≥8. Patient characteristics of the total sample stratified by depression and anxiety are presented in Table 1.

Table 1 Demographics and clinical characteristics of total sample stratified based on symptoms of depression and anxiety
Full size table

Symptoms of anxiety and depression were highly interrelated (r = .80; p = 0.001). Patients with scores above the cutoffs had lower education, were more likely to be employed and with more comorbidities as per CCI relative to those with score below cutoffs. Significant difference in age was noted only between the depression subgroups with those with lower scores being significantly older than patients above depression cutoffs.

During the 12 ± 8.1 month follow-up, N = 20 patients died (N = 11 Cardiovascular deaths) and N = 12 switched to hemodialysis due to technique failure. The most frequent causes of death were cardiovascular events and/or sepsis, whereas most frequent causes of transfer to HD were peritonitis and/or sepsis. Mean survival time was 13.6 ± 5.27 months.

Factors associated with clinical outcomes

Univariate analyses indicated that assisted PD, anxiety, Charlson comorbidity index, primary diagnosis of diabetic nephropathy, and albumin were associated with actuarial/technique survival (Table 2). Survival rates were 90 % for non-anxious patients and 76.5 % for those with anxiety (log-rank χ2 = 7.12; p = 0.003). The effect of depression approached, but did not reach significance (p = 0.06). None of the remaining demographic, laboratory/clinical parameters or social support indicators were significant.

Table 2 Unadjusted and adjusted multivariate Cox regression analyses
Full size table

Multivariate proportional hazard model to adjust for sociodemographic and clinical characteristics indicated that anxiety remained independently associated with all-cause mortality/technique (HR 2.25 [95 % CI 1.09, 4.63, p = 0.02]). Albumin and CCI were also associated with increased risk for all-cause mortality/technique failure in the multivariate model (Table 2). Sensitivity analysis using survival times calculated based on PD initiation rather than study entry/baseline yielded similar results (HR for anxiety = 2.54 [95 % CI 1.19, 5.42, p = 0.02]). Effects also remained significant in analysis using the continuous anxiety scores rather than clinical cutoffs (HR = 1.09, [95 % CI 1.02, 1.12, p = 0.01]).

Discussion

This is the first prospective study to explore the role of psychological parameters as risk factors for death and technique failure in patients on peritoneal dialysis.

Several important findings are evident from this study. Firstly, anxiety and depressive symptoms are common in PD patients and require attention. The majority of our patients experienced symptoms of depression or anxiety or both that have been shown in previous work to be related to treatment disruptiveness [21].

Although anxiety and depression are highly comorbid, only anxiety emerged as a (stronger) independent predictor of actuarial and technique survival in PD patients. The effect remained significant even after adjustment for other risk factors, in line with some of the evidence on myocardial infarction patients [22].

Notably, the risk conferred was higher than those of traditional clinical risk factors. Anxiety may exert an effect on prognosis through compromised health behaviors or physiological pathways such as reduced heart rate variability [23] and/or stress-induced increases in sympathetic nervous system activities and catecholamine release, which are linked to cardiovascular strain. Furthermore, PD patients have significantly increased norepinephrine plasma levels [24], which ex vivo work has shown to reduce ionic permeability of the peritoneal membrane pertinent to deterioration of PD technique efficacy [25] and thereby technique failure.

More work is needed to elucidate mechanisms and replicate findings with larger samples. This preliminary work, however, supports the call for screening and interventions integrated in routine renal care to improve emotional outcomes in end-stage renal disease and in PD in particular. Symptoms of anxiety have received less attention in both research and clinical practice. Regular assessment of distress symptoms (not confined to depression) may help to identify at risk patients who will benefit from treatment and/or referral to psychological services. Effective treatment for symptoms of distress may represent an easily achievable means of improving both quality of life and the clinical outcome of PD patients. Several limitations need to be recognized including the observational design that precludes causal inferences, the small sample size and number of death/events that made the derivation of combined events endpoint necessary, the lack of serial assessments of emotional distress over study window and the use of self-reported measurement of symptoms of anxiety and depression rather than diagnostic interviews.