Depression is the most common psychiatric problem in patients with end stage renal disease [22]. One of the most common diagnostic tools for depression is the Patient Health Questionnaire (PHQ–9) questionnaire, which was designed by Spitzer, Williams and Kronenke in 2001 [23]. PHQ–9 is based on the diagnostic criteria for Major Depressive Disorder under the code of Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (DSM-IV). It is chosen due to many of its advantages. The use of this questionnaire is free of charge, and it is shorter than many other depression rating scales. PHQ–9 can be administered in person, over the phone or as self-report. It also facilitates the diagnosis of major and minor depression, and assesses the severity of depression symptom. It is available in multiple languages including traditional Chinese, which was used in our study. PHQ–9 questionnaire has also been validated in dialysis population [24]. PHQ–9 is sensitive and specific at identifying depression in general population [17] and in dialysis population [23]. However, it should be noted that PHQ–9 is best used as a monitoring tool for the assessment of depression severity rather than as a screening tool, and a slightly elevated PHQ–9 score (i.e. the “mild depression” group) may also always imply an actual diagnosis of depression. The use of PHQ–9 in this study may explain the high rate of depression in our study.
As for frailty, there are a few validated tools in measuring frailty. The Frailty Index (FI) is a tool to quantify and summarize patient’s vulnerability. It includes assessment of factors including of symptoms, underlying concomitant diseases, conditions and disability [18]. There are also other tools including Cardiovascular Health Study (CHS) scale [25], Study of Osteoporotic Fracture (SOF) scale [19], Canadian Study of Health and Aging (CSHA) Clinical Frailty Scale [26], to assess frailty. However, all of these scales require direct patient examination, which is time consuming and impractical as the screening test in large number population. Apart from technical difficulties, not all of the above-mentioned screening tools have been validated in CKD cohorts. In our study, we used a questionnaire developed by our university and has been validated in the Chinese population [6]. The questions are straightforward and easy to answer, which simplify the process of screening.
The prevalence of depression in our study was 58.8%. The result is comparable to another local study, which reported 55% prevalence of depression among Chinese PD patients [4]. On the other hand, the worldwide prevalence of depression among dialysis population is reported to be 20–30% [27]. However, most of the western studies focus on hemodialysis patients. A study in Korea reported 75% prevalence of depression among the population undergoing CAPD [28].
At the first glance, it seems an unusually high figure to have over 70% of patients who required hospitalization within one year. It should be noted, however, that our local policy mandates hospital admission for many intervention procedures, such as cardiac catheterization and colonoscopy. Unfortunately, we do not have the detailed breakdown on the cause of hospitalization for our patients, but the figure in this study is similar to our previous reports [29,30].
We did not find a significant association between depression and the number or duration of hospitalization after adjusting for clinical confounding factors. Although previous studies reported that depression in patients with end stage renal disease is significantly associated with adverse medical outcomes, including the number of hospitalization [31] and cumulative hospital days [32], these studies were based on hemodialysis patients and may not be applicable to our patients. A local study showed that depression is significantly associated with mortality rate in Chinese peritoneal dialysis patients [4]. However, this study only recruited patients who were newly started on PD.
In our study, depression was not an independent predictor of mortality after adjusting for frailty and other clinical confounding factors. In contrast, most of the previous studies reported an association between depression and mortality in dialysis patients. For example, Lopes et al [33] noted an approximately 40% increase in mortality in patients who indicated the presence of depressive symptoms on self-reported questionnaires, although the study did not deploy any depression scale and the patients were not grouped according to their depression severity. A meta-analysis reported the mortality risk in patients on dialysis was 1.5 times higher in the presence of depressive symptoms after adjusting for other confounding factors [5]. However, none of the previous study on depression adjusts for frailty, even though the two parameters have a close internal correlation.
The prevalence of frailty in our study population was 72.7%, which is similar the prevalence of 73% reported by Bao et al [34], and 67.7% by Johansen et al [10]. In analyzing the baseline demographics and clinical parameters in our study population, we noted that frailty score had significant correlations with Charlson Comorbidity Index and dialysis adequacy. Correlation of frailty and comorbidities in dialysis patients was also reported in a prior study [35].
In our study, frailty score is better correlated with the number of hospitalization and duration of hospitalization than depression score. However, after adjusting for the confounding factors by multivariate analysis, frailty score was not a significant predictive factor of the number or duration of hospitalization. Our result is in contrary to prior studies. For example, Johansen et al [10] showed that frail patients were more likely to be hospitalized for any reason even after adjusting for other potential risk factors for hospitalization [10]. Mara et al [36] reported that frailty was a predictor of hospitalization in hemodialysis. Inclusion of hemodialysis patients in these studies may account for the discrepancy of their results with our study. Other possibilities include difference in patients’ ethnicity and age distribution.
With multivariate Cox regression analysis, our study showed that frailty was a significant predictor of all-cause mortality. Our result is in line with previous studies. For example, Alfaadhel et al [37] demonstrated that frailty at the initiation of dialysis was associated with mortality. Johansen et al [10] team showed that frailty was independently associated with the risk of death. We believe that frailty screening can help to identify a high risk group.
There are several limitations of our study. First, it is a single-center study of limited sample size, which affects the external validity and statistical power to detect subtle subgroup effects. Because of the limitation in our original study design, we did not measure cognitive impairment, which is an integral part of frailty and is directly related to depression. The follow up duration of one year is also not sufficient to detect any effect on patient mortality. Second, our study population was entirely Chinese. Because of the potential contribution of genetic and environmental factors to the mortality risk, our result may not be applicable to patients from other parts of world or even other parts of China. Our study only examined the rate of adverse outcome after one year, which may not be sufficient for mortality. Since our study recruited prevalent PD patients, the onset of depression and frailty (i.e. before or after dialysis was initiated) was not known. It was also possible that survivor bias was present, which may affect the result of our study.
In conclusion, our study shows that depression and frailty are common in prevalent PD patients. There is also a significant relationship between frailty and mortality in PD patients. Further studies are needed to determine the benefit of treatment for frailty in PD patients.