ABSTRACT
Background
Frailty and depression are highly prevalent in haemodialysis recipients, exhibit a reciprocal relationship, and are associated with increased mortality and hospitalization, and lower quality of life. Despite this, there has been little exploration of the relationship between depression and frailty upon patient outcomes. We aimed to explore the relationship between depression and frailty, and their associations with mortality, hospitalization and quality of life.
Methods
We performed a prospective cohort study of prevalent haemodialysis recipients linked to national datasets for outcomes including mortality and hospitalization. Depression was assessed using the Patient Health Questionnaire-9 (PHQ-9), frailty using the Clinical Frailty Scale (CFS) and quality of life using the EuroQol 5-Dimension (EQ-5D) Summary Index.
Results
A total of 485 prevalent haemodialysis recipients were recruited, with 111 deaths and 1241 hospitalizations during follow-up. CFS was independently associated with mortality [hazard ratio (HR) 1.31; 95% confidence interval (CI) 1.08, 1.59; P = .006], hospitalization [incidence rate ratio (IRR) 1.13; 95% CI 1.03, 1.25; P = .010] and lower quality of life (Coef. −0.401; 95% CI −0.511, −0.292; P < .001). PHQ-9 score was independently associated with lower quality of life (Coef. −0.042; 95% CI −0.063, −0.021; P < .001), but not mortality (HR 1.00; 95% CI 0.96, 1.04; P = .901) or hospitalization (IRR 0.99; 95% CI 0.97, 1.01; P = .351). In an adjusted model including CFS, moderate depression was associated with reduced hospitalization (IRR 0.72; 95% CI 0.56, 0.93; P = .013).
Conclusions
With the addition of frailty, depression was associated with lower hospital admissions, but poorer quality of life. The relationship between frailty and depression, and their influence on outcomes is complex, requiring further study.
Keywords: depression, frailty, haemodialysis, health-related quality of life, hospitalization, mortality
Graphical Abstract
Graphical Abstract.
INTRODUCTION
Frailty is the syndrome of accelerated ageing, characterized by multi-system dysregulation, and vulnerability to stressors [1]. Agreement between frailty screening tools in haemodialysis recipients is poor, and we have previously argued that frailty screening must combine ease of use and predictive capability for adverse outcomes [2]. One commonly utilized definition is the Clinical Frailty Scale (CFS) [3], a simple global measure of frailty based upon activities of daily living after clinical interview. The CFS is being increasingly used in multiple settings [4, 5], and associates with mortality and hospitalization in haemodialysis [2, 6].
Frailty is also associated with poor quality of life for kidney patients, which begins early at a pre-dialysis stage [7, 8]. In haemodialysis patients, self-rated fair/poor quality of life was more likely in frail haemodialysis recipients, and frailty is associated with worsening quality of life over time [9, 10].
Depression also impacts upon quality of life [11, 12], but is under-recognized in haemodialysis populations [13]. The Patient Health Questionnaire-9 (PHQ-9) has 92% sensitivity and 92% specificity for detection of depression in haemodialysis recipients [14]. Depression of at least moderate severity was present in 28% of participants of a large prospective cohort, and both moderate depression and continuous PHQ-9 scores have been associated with mortality [15–18]. A depression diagnosis has also been associated with greater hospitalization, but lower mortality within hospital admission episodes in retrospective cohorts [19, 20].
Despite high prevalence of both frailty and depression, there has been little exploration of the relationship between the two in haemodialysis recipients. A systematic review of depression and frailty in non-dialysis older adults has suggested a reciprocal relationship between frailty and depression, with each being a risk factor for the other [21]. However, prevalent depression was associated with incident frailty, but not vice versa in a prevalent US dialysis cohort [22]. US and European dialysis demographics and outcomes differ significantly [23], therefore these relationships require further study in non-US haemodialysis populations. As a clear overlap exists between symptoms of frailty and depression, the relationship between them and their associations with patient-centred outcomes requires scrutiny. Therefore, the aims of this study were (i) to explore the relationship between frailty and depression, and (ii) to assess the association of depression with mortality, hospitalization and quality of life in haemodialysis recipients.
MATERIALS AND METHODS
Study design
The Frailty Intervention Trial iN End-Stage patientS on haemodialysis (FITNESS) study is a two-stage study that follows a cohort multiple randomized controlled trial (cmRCT) design [24]. The study protocol was subject to favourable opinion by the South Birmingham Research Ethics Committee (Ref: 17/WM/0381) and institutional review board assessment of University Hospitals Birmingham NHS Foundation Trust (RRK6082). The first stage is a cross-sectional assessment and long-term follow up of study participants on maintenance haemodialysis. The full protocol for the FITNESS study has been described in detail elsewhere [25]. The study is reported in accordance with STROBE guidelines [26].
Study setting
Patients were recruited from a single nephrology centre located in Birmingham, UK, with a diverse range of ethnic and socioeconomic groups. Eligible patients were identified by interrogation of hospital electronic patient records (EPR) and from discussion with clinicians at each dialysis unit. Eligible patients were approached, given written and verbal information about the study, before giving their consent to join the cohort study.
Eligibility criteria
Inclusion criteria included adults aged 18 years and over, anyone receiving regular haemodialysis for at least 3 months’ duration and the ability to give informed consent. The only exclusion criterion was inpatient care within 4 weeks of recruitment unless for vascular access purposes (clotted or non-functional access, but excluding access infection), to avoid confounding of baseline data with frailty secondary to recent hospitalization.
Baseline assessment
Baseline assessments of all study participants took place before and during one of their usual dialysis sessions. To negate the potential effect of the long break from dialysis upon frailty measurements, we avoided the first haemodialysis session after the weekend interval. Where participants dialysed twice weekly, the dialysis session after the shortest interval was chosen for baseline assessment.
Study participants completed a number of investigations, which are detailed in our methodology paper [25]. Briefly, prior to connection to dialysis, participants underwent a timed 4-m walk from standing and bilateral hand-grip strength via dynamometer (Takei Grip D, Takei Scientific Instruments Co. Ltd, Japan), alongside a Montreal Cognitive Assessment (MoCA [27]). Once dialysis started, patients were clinically interviewed, including a series of questionnaires including assessments of activities of daily living (ADL) disability, demography, social history and frailty-specific questionnaires. PHQ-9 scores comprised nine questions each scored between 0 and 3 for frequency of depression symptoms, giving a total score of between 0 and 27. Moderate depression was defined as a PHQ-9 score of ≥10 [14]. The Physical Activity Index was derived from the GP Physical Activity Questionnaire via a validated formula [28, 29]. EPR were interrogated for comorbidities, drug history, dialysis vintage and adequacy, previous transplantation, and biochemical data. Determination of socioeconomic deprivation was based upon the Index of Multiple Deprivation 2015 (IMD), a multiple deprivation model calculated by local area, with 1 representing the most deprived and 5 the least deprived areas.
The CFS was obtained by interpretation of ADL questionnaire responses, with possible responses of 1–9. A CFS of 1 represented very fit, and 8 severely frail. A score of 9 was attributed to those who were terminally ill but not overtly frail. A CFS of 5–8 inclusive was considered frail.
Outcomes
Mortality data were obtained by electronic record linkage of all FITNESS study recruits to Office of National Statistics, a UK-wide repository of death certificate data, with robust coverage of mortality data. Hospital Episode Statistics, a database containing all secondary care episodes in any NHS hospital in England, provided robust national data for all hospitalizations. Hospital admissions were defined as any hospital episode lasting one or more night.
Health-related quality of life (HRQOL) was assessed using Euroqol 5-Dimension 3-Level (EQ-5D-3L). This was converted into the single-measure EQ-5D Summary Index via a standardized formula [30]. This results in a potential score of between 1 (if no HRQOL deficits identified) and −0.716 (if extreme deficit in every domain), and has been validated in UK populations.
Recruitment
Recruitment and power calculations are explored in detail in previous work from the FITNESS project [2, 25].
Statistics
Statistical analysis was performed using STATA 16 (StataCorp 2019, Stata Statistical Software: Release 16; StataCorp LLC, College Station, TX, USA) and R version 4.0.4 (R Foundation for Statistical Computing, Vienna, Austria). Categorical data were presented as numbers and percentages, with continuous variables reported as medians and interquartile ranges (IQRs).
Survival analyses were performed by Cox's Proportional Hazards Model, censored for end of follow-up. The proportional hazard assumption was satisfied by examination of plots of the log-negative-log of the within-group survivorship functions versus log time as well as comparing Kaplan–Meier (observed) with Cox (expected) survival curves with our study variables, alongside selected covariables for adjusted analyses.
Incidence rate ratios (IRR) were obtained for admissions data by negative binomial regression, death-censored and offset by length of follow-up. Negative binomial distribution was confirmed by interrogation of means and variances, and visual interpretation of expected versus observed distribution plots.
All Cox and negative binomial regressions were performed unadjusted and adjusted for an a priori list of covariables, based on a known or suspected relationship with dialysis-related mortality/admission [age, sex, ethnicity (grouped into white, south Asian, Black and other ethnicities), body mass index, index of multiple deprivation, Charlson comorbidity index (CKD omitted), number of hospitalization episodes, number of medications, smoking status, serum albumin, use of walking aids, dialysis vintage, self-reported change in health and kidney transplant wait-listing]. Further adjusted models were performed using these covariables plus CFS score.
Odds ratios (OR) of both frailty and moderate depression were obtained by logistic regression. Linear regressions were performed to explore the relationship between continuous frailty and PHQ-9 scores. Linearity was checked by comparison of observed values by linear and Lowess fit plots, and by plotting observed versus predicted residual values.
The EQ-5D summary index is right-censored, with a maximum value of 1. Therefore, it was transformed using the following equation 
to give a fractional score between 0 and 1, and fractional regressions were performed on these.
As each of frailty, depression and HRQOL have different potential associates, regression models were subject to adjustment for separate a priori covariables selected for known or suspected relationship with the outcome of interest. These models are detailed in Supplementary data, Table S1.
Variables that demonstrated significance in the final multivariable models were tested for interaction with both CFS and PHQ-9 scores and an interaction term for the outcome of interest. No such interactions were observed. Given recent evidence linking frailty to vascular access difficulty [31–33], sensitivity analyses were run on all final regression models with addition of vascular access type.
Missing values for IMD quintiles were handled with a separate category. Otherwise, missing data were assumed missing at random and handled via listwise deletion as all other covariables had <1% data missing. A P-value <.05 was considered significant in the statistical analysis.
RESULTS
Study cohort demographics
Figure 1 shows the PRISMA study flow of participant recruitment to the FITNESS study, with 485 prevalent haemodialysis patients with baseline frailty assessments and data linkage. Follow-up was 678 days (IQR 531–812 days), with all participants receiving a minimum potential follow-up of 365 days from recruitment. Baseline demographics are described in detail elsewhere [2]. Table 1 highlights key demographics stratified by frailty status at study recruitment.
Figure 1:
PRISMA flowchart of study participation.
Table 1:
Baseline demographics of the FITNESS cohort stratified by frailty status.
| Total cohort | Not frail | Frail | P | ||
|---|---|---|---|---|---|
| Frail | 261 (53.8) | ||||
| Median age (years) | 63 (53–74) | 60 (50–72) | 65 (55–76) | .016 | |
| Median PHQ-9 | 5 (2–10) | 3 (1–7) | 7 (3–6) | <.001 | |
| Median MoCA | 22 (17–25) | 23 (20–26) | 20 (16–23) | <.001 | |
| Median albumin (g/L) | 39 (35–41) | 39 (36–42) | 38 (34–41) | .024 | |
| Median BMI (kg/m2) | 26.8 (23.2–32.3) | 26 (23.0–30.7) | 27.9 (23.2–33.7) | 0.027 | |
| Median Charlson Scorea | 4 (3–6) | 4 (2–5) | 5 (4–7) | <.001 | |
| Median HD vintage (months) | 37 (17–76) | 33 (13–66) | 41 (19.9–81.5) | .015 | |
| Median Kt/V | 1.59 (1.39–1.85) | 1.58 (1.38–1.80) | 1.61 (1.41–1.88) | 0.474 | |
| Median EQ-5D Summary Index | 0.779 (0.516–1.00) | 1 (0.779–1.00) | 0.62 (0.189–0.796) | <.001 | |
| Health change | Better | 94 (19.4) | 52 (23.2) | 42 (16.1) | .003 |
| The same | 174 (35.9) | 90 (40.2) | 84 (32.2) | ||
| Worse | 217 (44.7) | 82 (36.6) | 135 (51.7) | ||
| IMD quintile | 1 | 212 (43.7) | 96 (42.9) | 116 (44.4) | |
| 2 | 87 (17.9) | 43 (19.2) | 44 (16.9) | ||
| 3 | 85 (17.5) | 38 (17.0) | 47 (18.0) | ||
| 4 | 38 (7.8) | 20 (8.9) | 18 (6.9) | .921 | |
| 5 | 33 (6.8) | 14 (6.3) | 19 (7.3) | ||
| Unknown | 30 (6.2) | 13 (5.8) | 17 (6.5) | ||
| Ethnicity | White | 281 (57.9) | 13 (61.2) | 144 (55.2) | .180 |
| South Asian | 115 (23.7) | 44 (19.6) | 71 (27.2) | ||
| Black | 76 (15.7) | 35 (15.6) | 41 (15.7) | ||
| Other | 13 (2.7) | 8 (3.6) | 5 (1.9) | ||
| Gender | Male | 284 (58.6) | 148 (66.1) | 136 (52.1) | .002 |
| Comorbidities | Diabetes | 138 (28.5) | 43 (19.2) | 95 (36.4) | <.001 |
| MI | 98 (20.2) | 34 (15.2) | 64 (24.5) | .011 | |
| CVA/TIA | 57 (11.8) | 17 (7.6) | 40 (15.3) | .008 | |
| Cancer | 56 (11.6) | 30 (13.4) | 26 (10.0) | .238 | |
| Heart failure | 52 (10.7) | 19 (8.5) | 33 (12.6) | .140 | |
| PVD | 47 (9.7) | 15 (6.7) | 32 (12.3) | .039 | |
| Primary renal disease |
Diabetic | 114 (23.5) | 33 (14.7) | 81 (31.3) | <.001 |
| Hypertensive | 39 (8.0) | 22 (9.8) | 17 (6.5) | .182 | |
| Ischaemic | 38 (7.8) | 14 (6.3) | 24 (9.2) | .229 | |
| IgA | 37 (7.6) | 20 (8.9) | 17 (6.5) | .318 | |
| PKD | 28 (5.8) | 17 (7.6) | 11 (4.2) | .122 | |
| FSGS | 24 (5.0) | 14 (6.3) | 10 (3.8) | .221 | |
| Reflux | 17 (3.5) | 7 (3.1) | 10 (3.8) | .673 | |
| Obstructive | 16 (3.3) | 10 (4.5) | 6 (2.3) | .183 | |
| AAV | 15 (3.1) | 11 (4.9) | 4 (1.5) | .032 | |
| Interstitial nephritis | 10 (2.1) | 6 (2.7) | 4 (1.5) | .376 | |
| Myeloma | 10 (2.1) | 8 (3.6) | 2 (0.8) | .030 | |
| Unknown | 68 (14.0) | 31 (13.8) | 37 (14.2) | .915 | |
| Smoking status | Current | 6814.1 | 38 (17.0) | 30 (11.5) | |
| Ex | 132 (27.3) | 64 (28.6) | 68 (26.2) | .133 | |
| Never | 284 (58.7) | 122 (54.5) | 162 (62.3) | ||
| Dialysis access | Line | 113 (23.3) | 47 (21.0) | 66 (25.3) | .264 |
| Tx list status | Active | 58 (12.0) | 36 (16.1) | 22 (8.4) | .036 |
| Suspended | 15 (3.1) | 9 (4.0) | 6 (2.3) | ||
| Not listed | 412 (85.0) | 179 (79.9) | 233 (89.3) | ||
| Employment status |
Employed | 69 (14.3) | 61 (27.2) | 8 (3.1) | <.001 |
| Unemployed | 148 (30.6) | 58 (25.9) | 90 (34.6) | ||
| Retired | 267 (55.2) | 105 (46.9) | 162 (62.3) | ||
| Job roleb | Unskilled manual | 181 (39.3) | 70 (32.1) | 111 (45.7) | .046 |
| Skilled manual | 101 (21.9) | 50 (22.9) | 51 (21.0) | ||
| Clerical | 52 (11.3) | 28 (12.8) | 24 (9.9) | ||
| Managerial | 46 (10.0) | 26 (11.9) | 20 (8.2) | ||
| Professional | 81 (17.6) | 44 (20.2) | 37 (15.2) | ||
| Education level | High school | 342 (70.7) | 146 (65.2) | 196 (75.4) | .042 |
| College/6th form | 92 (19.0) | 49 (21.9) | 43 (16.5) | ||
| University | 50 (10.3) | 29 (13.0) | 21 (8.1) | ||
| Residence | Own home | 462 (95.9) | 218 (97.8) | 244 (94.2) | .172 |
| Warden-controlled | 12 (2.5) | 3 (1.4) | 9 (3.5) | ||
| Residential home | 5 (1.0) | 2 (0.9) | 3 (1.2) | ||
| Nursing home | 3 (0.6) | 0 (0.0) | 3 (1.2) | ||
Median (IQR) or n (%) shown.
aCKD omitted from Charlson Score.
bOr prior job role if unemployed/retired.
BMI: body mass index; HD: haemodialysis; CVA/TIA: cerebrovascular accident/transient ischaemic attack; PVD: peripheral vascular disease; PKD: polycystic kidney disease; FSGS: focal segmental glomerulosclerosis; AAV: ANCA-associated vasculitis.
Depression scores
The median PHQ-9 score was 5 (IQR 2–10), with 222 participants (46.3%) reporting minimal depression, 131 (27.3%) mild depression, 76 (15.8%) moderate depression, 38 (7.9%) moderately severe depression and 13 (2.7%) severe depression.
Correlation was weak between CFS and PHQ-9 scores (Rho = 0.361, P < 0.001). Figure 2 demonstrates that, on both univariable and each multivariable analysis, every increase in CFS score was associated with increased PHQ-9 scores. Table 2 shows full multivariable model results. Figure 3 demonstrates that PHQ-9 score was associated with CFS score on all adjusted models; Table 3 highlights the fully adjusted model.
Figure 2:
Change in PHQ-9 score with each 1-point increase in CFS.
Table 2:
Linear regression of change in PHQ-9 score associated with covariables in fully adjusted Model 4.
| Coefficient | Lower 95% CI | Upper 95% CI | P | |
|---|---|---|---|---|
| CFS score | 0.768 | 0.397 | 1.14 | <.001 |
| Age | −0.093 | −0.137 | −0.050 | <.001 |
| Ethnicity | ||||
| White | REFERENCE | |||
| South Asian | −1.54 | −2.74 | −0.344 | .012 |
| Black | −0.689 | −1.96 | 0.578 | .286 |
| Other | −0.007 | −2.83 | 2.81 | .996 |
| Gender | ||||
| Male | REFERENCE | |||
| Female | 0.527 | −0.366 | 1.42 | .246 |
| IMD quintile | ||||
| 1 | REFERENCE | |||
| 2 | −0.164 | −1.37 | 1.05 | .790 |
| 3 | −0.083 | −1.31 | 1.14 | .894 |
| 4 | −0.611 | −2.27 | 1.05 | .471 |
| 5 | −0.424 | −2.18 | 1.34 | .636 |
| Unknown | 0.062 | −1.88 | 2.00 | .950 |
| Education level | ||||
| High school | REFERENCE | |||
| College/6th form | −0.919 | −2.05 | 0.209 | .110 |
| University | −1.91 | −3.39 | −0.422 | .012 |
| Prior admissions | −0.008 | −0.188 | 0.172 | .931 |
| Charlson Indexa | −0.079 | −0.350 | 0.193 | .570 |
| Cognitive impairment | ||||
| No | REFERENCE | |||
| Yes | −0.366 | −1.49 | 0.753 | .521 |
| Antidepressant use | ||||
| No | REFERENCE | |||
| Yes | 3.69 | 2.29 | 5.09 | <.001 |
| Number medications | 0.075 | −0.056 | 0.206 | .260 |
| HD vintage | −0.001 | −0.010 | 0.007 | .778 |
| Self-rated pain | ||||
| None | REFERENCE | |||
| Moderate | 1.63 | 0.664 | 2.60 | .001 |
| Extreme | 2.61 | 0.867 | 4.36 | .003 |
| Self-rated health change | ||||
| Better/The same | REFERENCE | |||
| Worse | 1.73 | 0.808 | 2.65 | <.001 |
| Self-rated health/100 | −0.056 | −0.078 | −0.034 | <.001 |
| Constant | 10.6 | 7.19 | 14.1 | <.001 |
Coefficients derived from multiple linear regression. Bold text indicates significant at P < .05 level.
aCKD omitted from Charlson Index.
Figure 3:
Increase in CFS associated with each 1-point increase in PHQ-9.
Table 3:
Linear regression of change in CFS score associated with covariables in fully adjusted Model 4.
| Coefficient | Lower 95% CI | Upper 95% CI | P | |
|---|---|---|---|---|
| PHQ-9 score | 0.046 | 0.027 | 0.064 | <.001 |
| Age | −0.008 | −0.017 | 0.001 | .083 |
| Gender | ||||
| Male | REFERENCE | |||
| Female | 0.226 | 0.040 | 0.412 | .017 |
| Ethnicity | ||||
| White | REFERENCE | |||
| South Asian | 0.214 | −0.037 | 0.466 | .095 |
| Black | −0.012 | −0.278 | 0.254 | .929 |
| Other | 0.257 | −0.331 | 0.844 | .391 |
| Education level | ||||
| High school | REFERENCE | |||
| College/6th form | 0.008 | −0.224 | 0.240 | .945 |
| University | −0.152 | −0.468 | 0.163 | .344 |
| IMD quintile | ||||
| 1 | REFERENCE | |||
| 2 | 0.104 | −0.147 | 0.354 | .416 |
| 3 | 0.015 | −0.238 | 0.269 | .905 |
| 4 | −0.301 | −0.647 | 0.045 | .088 |
| 5 | 0.007 | −0.362 | 0.376 | .969 |
| Unknown | 0.156 | −0.240 | 0.551 | .439 |
| Social support | ||||
| No | REFERENCE | |||
| Yes | −0.698 | −1.07 | −0.332 | <.001 |
| Charlson Indexa | 0.094 | 0.040 | 0.147 | .001 |
| Cognitive Impairment | ||||
| No | REFERENCE | |||
| Yes | 0.030 | −0.201 | 0.261 | .798 |
| Smoking status | ||||
| Current | REFERENCE | |||
| Ex | 0.277 | −0.013 | 0.566 | .061 |
| Never | 0.172 | −0.100 | 0.444 | .214 |
| Self-rated health change | ||||
| Better/The same | REFERENCE | |||
| Worse | 0.075 | −0.117 | 0.267 | .441 |
| Self-rated health/100 | −0.006 | −0.011 | −0.002 | .006 |
| Walking aid use | ||||
| No | REFERENCE | |||
| Yes | 0.532 | 0.307 | 0.758 | <.001 |
| Physical Activity Index | ||||
| Inactive | REFERENCE | |||
| Moderately inactive | −0.384 | −0.732 | −0.035 | .031 |
| Moderately active | −1.07 | −1.49 | −0.641 | <.001 |
| Active | −1.86 | −2.30 | −1.42 | <.001 |
| Slow walking speed | ||||
| No | REFERENCE | |||
| Yes | 0.712 | 0.482 | 0.941 | <.001 |
| Low grip strength | ||||
| No | REFERENCE | |||
| Yes | 0.082 | −0.119 | 0.283 | .424 |
| Constant | 3.88 | 3.20 | 4.55 | <.001 |
Coefficients derived from multiple linear regression. Bold text indicates significant at P < 0.05 level.
aCKD omitted from Charlson Index.
Mortality and hospitalization
Over a follow-up period comprising a total of 862.1 patient-years, we observed 111 (22.9%) deaths, 1241 hospital admissions of at least one night's duration (1.44 admissions per patient-year) and 10 175 total nights spent in hospital. Overall, 382 (78.8%) participants had at least one hospital admission during follow-up, and the median number of admissions was 2 (IQR 0–4).
Depression and mortality
PHQ-9 score was not associated with mortality, either on univariable [hazard ratio (HR) 1.02; 95% confidence interval (CI) 0.988, 1.05; P = .220] or multivariable analysis (HR 1.00; 95% CI 0.96–1.04; P = .901). Moderate depression was not associated with mortality on univariable (HR 1.16; 95% CI 0.771, 1.75; P = .472), or multivariable analyses (HR 0.91; 95% CI 0.57, 1.47; P = .713). Similar results were obtained upon Cox regression with omission of frailty from the adjusted analyses, as shown in Supplementary data, Tables S1–S4.
Depression and hospital admissions
PHQ-9 score was associated with increased rates of hospitalization on univariable analysis (IRR 1.02; 95% CI 1.00, 1.04; P = .028), but lost significance on multivariable analyses both with CFS (IRR 0.99; 95% CI 0.97, 1.01; P = .351) and without (IRR 1.00; 95% CI 0.98, 1.02; P = .973) in the model. Full models are shown in Supplementary data, Tables S5 and S6.
Figure 4 shows moderate depression was not associated with rates of hospital admissions in univariable or multivariable analyses where CFS was omitted. However, admissions were significantly lower upon multivariable analysis with the addition of CFS into the model. Full model results are shown in Supplementary data, Tables S7 and S8.
Figure 4:
IRRs of hospital admissions associated with moderate depression.
Depression, frailty and quality of life
Correlation was moderate between EQ Summary Index and both PHQ-9 (Rho = 0.448; P < .001), and CFS (Rho = 0.600; P < .001). Figure 5 highlights both CFS and PHQ-9 scores were associated with lower fractional EQ-5D summary index scores on all models—univariable and multivariable. The fully adjusted model is shown in Table 4.
Figure 5:
Change in fractional EQ Summary Index with increase in PHQ-9 and CFS scores.
Table 4:
Associates of fractional EQ-5D Summary Index—fully adjusted Model 4.
| Coefficient | Lower 95% CI | Upper 95% CI | P | |
|---|---|---|---|---|
| PHQ-9 score | −0.042 | −0.063 | −0.021 | <.001 |
| CFS | −0.401 | −0.511 | −0.292 | <.001 |
| Age | 0.002 | −0.013 | 0.016 | .829 |
| Gender | ||||
| Male | REFERENCE | |||
| Female | −0.233 | −0.442 | −0.023 | .030 |
| Ethnicity | ||||
| White | REFERENCE | |||
| South Asian | 0.271 | −0.031 | 0.573 | .078 |
| Black | 0.135 | −0.191 | 0.462 | .416 |
| Other | −0.487 | −1.18 | 0.206 | .169 |
| Education level | ||||
| High School | REFERENCE | |||
| College/6th form | −0.262 | −0.553 | 0.029 | .078 |
| University | −0.379 | −0.690 | −0.067 | .017 |
| Social support | ||||
| Yes | REFERENCE | |||
| No | −0.486 | −0.807 | −0.165 | .003 |
| IMD quintile | ||||
| 1 | REFERENCE | |||
| 2 | −0.219 | −0.511 | 0.072 | .141 |
| 3 | 0.024 | −0.270 | 0.318 | .874 |
| 4 | 0.135 | −0.264 | 0.534 | .508 |
| 5 | 0.243 | −0.227 | 0.713 | .311 |
| Unknown | 0.127 | −0.398 | 0.652 | .636 |
| Employment status | ||||
| Employed | REFERENCE | |||
| Unemployed | −0.021 | −0.716 | 0.675 | .954 |
| Retired | 0.101 | −0.512 | 0.714 | .747 |
| HD vintage (months) | −0.003 | −0.005 | −0.001 | .009 |
| Charlson Indexa | −0.011 | −0.070 | 0.047 | .706 |
| Hb (g/L) | 0.004 | −0.005 | 0.013 | .361 |
| Kt/V | 0.242 | −0.060 | 0.544 | .117 |
| Antidepressant use | ||||
| No | REFERENCE | |||
| Yes | −0.100 | −0.435 | 0.235 | .558 |
| Use of walking aids | ||||
| No | REFERENCE | |||
| Yes | −0.233 | −0.484 | 0.017 | .068 |
| Walking speed | ||||
| Not slow | REFERENCE | |||
| Slow | −0.435 | −0.695 | −0.175 | .001 |
| Physical Activity Index | ||||
| Inactive | REFERENCE | |||
| Moderately inactive | −0.070 | −0.775 | 0.634 | .845 |
| Moderately active | −0.861 | −1.48 | −0.246 | .006 |
| Active | −0.091 | −0.964 | 0.783 | .838 |
| Self-rated health/100 | 0.012 | 0.007 | 0.017 | <.001 |
| Health change in the past year | ||||
| Better/The same | REFERENCE | |||
| Worse | −0.180 | −0.396 | 0.036 | .102 |
| Cognitive impairment | ||||
| No | REFERENCE | |||
| Yes | −0.308 | −0.571 | −0.045 | .022 |
| Constant | 3.110 | 1.296 | 4.924 | .001 |
Sensitivity analyses
Addition of vascular access type to each regression model did not affect interpretation of the data, as shown in Supplementary data, Tables S10–S20.
DISCUSSION
Frailty and depression are both associated with adverse outcomes in haemodialysis recipients, but there has been little investigation of the relationship between these important conditions and patient outcomes. To our knowledge, this study is the first to demonstrate that frailty and poor self-reported health over the past year are associated with increased PHQ-9 scores for haemodialysis patients, whereas self-reported physical activity is inversely associated with PHQ-9 score. We did not replicate findings of an association between depression and mortality, or with hospital admissions on multivariable analyses. Our data highlight the complex association between frailty and depression in haemodialysis recipients, but do not support depression as an associate of mortality or hospitalization. Depression remains an important outcome as evidenced by its association with lower quality of life.
Our data corroborate work suggesting a reciprocal relationship between depression and frailty. In a systematic review in non-dialysis elderly people, a bi-directional relationship was observed between depression and frailty [21]. By contrast, Sy and colleagues observed in a large US multicentre longitudinal haemodialysis cohort that depression was associated with incident frailty but not vice versa [22]. However, the same cohort demonstrated an association baseline frailty and baseline depression. Our cohort confirms this association even after adjustment for numerous a priori covariables, including many that may be considered surrogates for frailty and/or depression, suggesting a robust relationship between the two.
The lack of association between depression and mortality runs counter to work from other groups and warrants discussion. In a large prospective cohort study, Saglimbene and colleagues observed moderate depression by the Beck Depression Inventory (BDI) to be associated with all-cause mortality in haemodialysis [15]. Chilcot and colleagues also observed an association between moderate depression and mortality using both the BDI and PHQ-9 [16]. In a meta-analysis, depression scores were associated with a modest mortality HR of 1.04 [18], though heterogeneity in study populations and depression measures limit interpretation of these results. The effect size seen in this meta-analysis was small, from a sample population of 8267 participants, suggesting our lack of significant association between depression and mortality may represent an underpowered sample. A lack of appreciable effect size on mortality hazard in our cohort, however, undermines this hypothesis. Machado and colleagues, in an umbrella review of published systematic reviews and meta-analyses, observed nominally significant results to suggest an association between depression and all-cause mortality in the general population [34]. However, the evidence becomes weaker when focusing on studies using structured interviews and those adjusting for potential confounders. Therefore, a causal effect of depression on all-cause and cause-specific mortality remains unproven and is consistent with our data.
We can speculate depressed and frail haemodialysis recipients may be less inclined to seek medical attention for the sequelae of their frailty, or that healthcare providers may be more likely to attribute their presenting complaints to comorbid depression rather than organic disease. There was, however, no appreciable interaction effect between these two variables. Alternatively, serial measures of depression may be more sensitive at identifying patients at risk of adverse outcomes. Weisbord and colleagues have reported an associated increase in hospitalization and mortality for haemodialysis recipients with moderate to severe depression upon serial prospective measures of depression [17]. To further illustrate the heterogenous nature of reported outcomes, a coded diagnosis of depression in retrospective registry studies amongst haemodialysis recipients was associated with greater admission rates and length of stay in hospital [20], but lower mortality risk during inpatient stay [19]. Caution is required with interpretation, as coding data have been shown to have limited accuracy when applied to comorbidities—rather than primary admission diagnoses—and thus limit the potential utility in research [35]. Further work is warranted to better understand these important outcomes for haemodialysis patients.
The association between frailty, depression and HRQOL is less well described in haemodialysis recipients. Jafari and colleagues demonstrated frailty to be associated with lower EQ-5D scores in a small Canadian haemodialysis cohort [10]. McAdams-DeMarco and colleagues found an association between frailty and self-rated fair or poor health, though this might not be considered as an HRQOL measure per se [9]. Depression has also been associated with lower quality of life in haemodialysis recipients [11, 12]. Our data demonstrate that frailty and depression independently associate with poorer HRQOL, which has not been reported before. Furthermore, we have demonstrated sex, education, social support, haemodialysis vintage, walking speed, cognitive impairment and self-rated health all have independent associations with poorer HRQOL. Whilst these data only describe associations, rather than predictors of lower HRQOL, they suggest that the determinants of HRQOL may be complex and require detailed study.
FITNESS is amongst the first studies to examine the relationship between frailty and depression in a large prospective haemodialysis cohort. Strengths of this study include the large prospective cohort with robust data linkage for outcomes. The cohort is well described with multiple areas of frailty, quality of life, comorbidity, cognition, depression and musculoskeletal function accounted for in the analysis. Further strengths include the diversity of demographics, comorbidities and socioeconomic backgrounds representative of the local population [36]. However, our data should be interpreted with caution in non-English populations. Furthermore, as a single-centre analysis, validation of our findings elsewhere is required. Limitations also include the single baseline data collection for frailty and depression phenotyping; both are dynamic syndromes and repeated measures would offer greater detail and arguably accuracy [17, 37]. Our dataset also lacked adjustment for a validated malnutrition score. Our CFS definition deviates slightly from the original validation sample [3], in that no multidisciplinary team discussion was held to determine frailty status after clinical interview. However, we would argue that our approach is analogous to CFS use in clinical practice [5]. Use of the EQ-5D summary index for HRQOL is validated in UK populations, but its relationship with other HRQOL measures is uncertain. Finally, we report associations in these observational data rather than causation, and due caution should be maintained when applying these findings to the individual haemodialysis recipient in clinical practice.
To conclude, depression and frailty are intimately linked in haemodialysis recipients, and independently associate with poorer quality of life, but only frailty is associated with mortality and hospitalization in this cohort. Our data highlights some of the complexity in assessing frailty and its effects in individuals and populations, including associations with other competing risks like depression. Further research is warranted in this area to understand the cause and effect of frailty, and its associated outcomes, among haemodialysis patients.
Supplementary Material
Contributor Information
Benjamin M Anderson, Department of Nephrology and Transplantation, Queen Elizabeth Hospital, Birmingham, UK; Institute of Inflammation and Ageing, University of Birmingham, Birmingham, UK.
Muhammad Qasim, Department of Nephrology and Transplantation, Queen Elizabeth Hospital, Birmingham, UK; Institute of Immunology and Immunotherapy, University of Birmingham, Birmingham, UK.
Gonzalo Correa, Department of Nephrology, Hospital del Salvador, Santiago, Chile.
Felicity Evison, Department of Health Informatics, Queen Elizabeth Hospital, Birmingham, UK.
Suzy Gallier, Department of Health Informatics, Queen Elizabeth Hospital, Birmingham, UK; PIONEER: HDR-UK hub in Acute Care, Edgbaston, Birmingham, UK.
Charles J Ferro, Department of Nephrology and Transplantation, Queen Elizabeth Hospital, Birmingham, UK; Institute of Cardiovascular Sciences, University of Birmingham, Birmingham, UK.
Thomas A Jackson, Institute of Inflammation and Ageing, University of Birmingham, Birmingham, UK; Department of Healthcare for Older People, Queen Elizabeth Hospital, Birmingham, UK.
Adnan Sharif, Department of Nephrology and Transplantation, Queen Elizabeth Hospital, Birmingham, UK; Institute of Immunology and Immunotherapy, University of Birmingham, Birmingham, UK.
FUNDING
Queen Elizabeth Hospital Charity, fund number 17-3-886
DATA AVAILABILITY STATEMENT
The data underlying this article will be shared on reasonable request to the corresponding author.
CONFLICT OF INTEREST STATEMENT
The authors have no relevant conflicts of interest to declare. The results presented in this paper have not been published previously in whole or part, except in abstract format.
ETHICS COMMITTEE APPROVAL
The study protocol was subject to favourable opinion by the South Birmingham Research Ethics Committee (Ref: 17/WM/0 381) and institutional review board assessment of University Hospitals Birmingham NHS Foundation Trust (RRK6082).
REFERENCES
- 1. Morley JE, Vellas B, van Kan GAet al. Frailty consensus: a call to action. J Am Med Dir Assoc 2013;14:392–7. 10.1016/j.jamda.2013.03.022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Anderson BM, Qasim M, Correa Get al. Correlations, agreement and utility of frailty instruments in prevalent haemodialysis patients: baseline cohort data from the FITNESS study. Clin Kidney J 2021;15:145–52. 10.1093/ckj/sfab137. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Rockwood K, Song X, MacKnight Cet al. A global clinical measure of fitness and frailty in elderly people. Can Med Assoc J 2005;173:489–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Church S, Rogers E, Rockwood Ket al. A scoping review of the Clinical Frailty Scale. BMC Geriatr 2020;20:393. 10.1186/s12877-020-01801-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Rockwood K, Theou O.. Using the Clinical Frailty Scale in allocating scarce health care resources. Can Geriatr J 2020;23:210–5. 10.5770/cgj.23.463. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Alfaadhel TA, Soroka SD, Kiberd BAet al. Frailty and mortality in dialysis: evaluation of a clinical frailty scale. Clin J Am Soc Nephrol 2015;10:832–40. 10.2215/cjn.07760814. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Mansur HN, Colugnati FAB, Grincenkov FRdSet al. Frailty and quality of life: a cross-sectional study of Brazilian patients with pre-dialysis chronic kidney disease. Health Qual Life Outcomes 2014;12:27. 10.1186/1477-7525-12-27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Nixon AC, Bampouras TM, Pendleton Net al. Frailty is independently associated with worse health-related quality of life in chronic kidney disease: a secondary analysis of the Frailty Assessment in Chronic Kidney Disease study. Clin Kidney J 2020;13:85–94. 10.1093/ckj/sfz038. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. McAdams-DeMarco MA, Ying H, Olorundare Iet al. Frailty and health-related quality of life in end stage renal disease patients of all ages. J Frailty Aging 2016;5:174–9. 10.14283/jfa.2016.106. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Jafari M, Kour K, Giebel Set al. The burden of frailty on mood, cognition, quality of life, and level of independence in patients on hemodialysis: Regina Hemodialysis Frailty Study. Can J Kidney Health Dis 2020;7:2054358120917780. 10.1177/2054358120917780. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Drayer RA, Piraino B, Reynolds CF IIIet al. Characteristics of depression in hemodialysis patients: symptoms, quality of life and mortality risk. Gen Hosp Psychiatry 2006;28:306–12. 10.1016/j.genhosppsych.2006.03.008. [DOI] [PubMed] [Google Scholar]
- 12. Shimizu U, Aoki H, Sakagami Met al. Walking ability, anxiety and depression, significantly decrease EuroQol 5-Dimension 5-Level scores in older hemodialysis patients in Japan. Arch Gerontol Geriatr 2018;78:96–100. 10.1016/j.archger.2018.06.006. [DOI] [PubMed] [Google Scholar]
- 13. King-Wing Ma T, Kam-Tao Li P. Depression in dialysis patients. Nephrology (Carlton) 2016;21:639–46. 10.1111/nep.12742. [DOI] [PubMed] [Google Scholar]
- 14. Watnick S, Wang P-L, Demadura Tet al. Validation of 2 depression screening tools in dialysis patients. Am J Kidney Dis 2005;46:919–24. 10.1053/j.ajkd.2005.08.006. [DOI] [PubMed] [Google Scholar]
- 15. Saglimbene V, Palmer S, Scardapane Met al. Depression and all-cause and cardiovascular mortality in patients on haemodialysis: a multinational cohort study. Nephrol Dial Transplant 2017;32:377–84. 10.1093/ndt/gfw016. [DOI] [PubMed] [Google Scholar]
- 16. Chilcot J, Guirguis A, Friedli Ket al. Depression symptoms in haemodialysis patients predict all-cause mortality but not kidney transplantation: a cause-specific outcome analysis. Ann Behav Med 2018;52:1–8. 10.1007/s12160-017-9918-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Weisbord SD, Mor MK, Sevick MAet al. Associations of depressive symptoms and pain with dialysis adherence, health resource utilization, and mortality in patients receiving chronic hemodialysis. Clin J Am Soc Nephrol 2014;9:1594–602. 10.2215/CJN.00220114. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Farrokhi F, Abedi N, Beyene Jet al. Association between depression and mortality in patients receiving long-term dialysis: a systematic review and meta-analysis. Am J Kidney Dis 2014;63:623–35. 10.1053/j.ajkd.2013.08.024. [DOI] [PubMed] [Google Scholar]
- 19. Chan L, Tummalapalli SL, Ferrandino Ret al. The effect of depression in chronic hemodialysis patients on inpatient hospitalization outcomes. Blood Purif 2017;43:226–34. 10.1159/000452750. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Hedayati SS, Grambow SC, Szczech LAet al. Physician-diagnosed depression as a correlate of hospitalizations in patients receiving long-term hemodialysis. Am J Kidney Dis 2005;46:642–9. 10.1053/j.ajkd.2005.07.002. [DOI] [PubMed] [Google Scholar]
- 21. Soysal P, Veronese N, Thompson Tet al. Relationship between depression and frailty in older adults: a systematic review and meta-analysis. Ageing Res Rev 2017;36:78–87. 10.1016/j.arr.2017.03.005. [DOI] [PubMed] [Google Scholar]
- 22. Sy J, McCulloch CE, Johansen KL.. Depressive symptoms, frailty, and mortality among dialysis patients. Hemodial Int 2019;23:239–46. 10.1111/hdi.12747. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Foley RN, Hakim RM.. Why is the mortality of dialysis patients in the United States much higher than the rest of the world? J Am Soc Nephrol 2009;20:1432–5. 10.1681/asn.2009030282. [DOI] [PubMed] [Google Scholar]
- 24. Relton C, Torgerson D, O'Cathain Aet al. Rethinking pragmatic randomised controlled trials: introducing the “cohort multiple randomised controlled trial” design. BMJ 2010;340:c1066. [DOI] [PubMed] [Google Scholar]
- 25. Anderson BM, Dutton M, Day Eet al. Frailty Intervention Trial iN End-Stage patientS on haemodialysis (FITNESS): study protocol for a randomised controlled trial. Trials 2018;19:457. 10.1186/s13063-018-2842-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. von Elm E, Altman DG, Egger Met al. Strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. BMJ 2007;335:806–8. 10.1136/bmj.39335.541782.AD. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Nasreddine ZS, Phillips NA, Bédirian Vet al. The Montreal Cognitive Assessment, MoCA: a brief screening tool for mild cognitive impairment. J Am Geriatr Soc 2005;53:695–9. http://onlinelibrary.wiley.com/doi/10.1111/j.1532-5415.2005.53221.x/abstract. [DOI] [PubMed] [Google Scholar]
- 28. Khaw K-T, Jakes R, Bingham Set al. Work and leisure time physical activity assessed using a simple, pragmatic, validated questionnaire and incident cardiovascular disease and all-cause mortality in men and women: the European Prospective Investigation into Cancer in Norfolk prospective population study. Int J Epidemiol 2006;35:1034–43. 10.1093/ije/dyl079. [DOI] [PubMed] [Google Scholar]
- 29. Physical Activity Policy HID . General Practice Physical Activity Questionnaire. London: UK Department of Health, 2009. [Google Scholar]
- 30. Dolan P. Modeling valuations for EuroQol health states. Med Care 1997;35:1095–108. 10.1097/00005650-199711000-00002. [DOI] [PubMed] [Google Scholar]
- 31. Kuningas K, Inston N.. Tailoring vascular access for dialysis: can frailty assessment improve the fit? Am J Kidney Dis 2022;80:7–8. 10.1053/j.ajkd.2021.12.003. [DOI] [PubMed] [Google Scholar]
- 32. Luo C-M, Hsieh M-Y, Cheng C-Het al. Association of frailty with thrombosis of hemodialysis vascular access: a prospective Taiwanese cohort study. Am J Kidney Dis 2022;80:353–63.e1. 10.1053/j.ajkd.2021.12.017. [DOI] [PubMed] [Google Scholar]
- 33. Woo K, Gascue L, Norris Ket al. Patient frailty and functional use of hemodialysis vascular access: a retrospective study of the US Renal Data System. Am J Kidney Dis 2022;80:30–45. 10.1053/j.ajkd.2021.10.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Machado MO, Veronese N, Sanches Met al. The association of depression and all-cause and cause-specific mortality: an umbrella review of systematic reviews and meta-analyses. BMC Med 2018;16:112. 10.1186/s12916-018-1101-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Burns EM, Rigby E, Mamidanna Ret al. Systematic review of discharge coding accuracy. J Public Health (Oxf) 2012;34:138–48. 10.1093/pubmed/fdr054. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. UK Renal Registry (2021) UK Renal Registry 23rd Annual Report – data to 31/12/2019, Bristol, UK. Available from; https://ukkidney.org/sites/renal.org/files/publication/file-attachments/23rd_UKRR_ANNUAL_REPORT.pdf (September 14, 2022, date last accessed). [Google Scholar]
- 37. Johansen KL, Dalrymple LS, Delgado Cet al. Factors associated with frailty and its trajectory among patients on hemodialysis. Clin J Am Soc Nephrol 2017;12:1100–8. 10.2215/cjn.12131116. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data underlying this article will be shared on reasonable request to the corresponding author.