Abstract
Background and objectives: There is growing identification of the need to seriously study the psychiatric presentations of end-stage renal disease patients treated with hemodialysis. This study reports on the course of depression and anxiety diagnoses and their impact on quality of life and health status.
Design, setting, participants, & measurements: The 16-mo course of psychiatric diagnoses in 50 end-stage renal disease patients treated with hemodialysis was measured by structured clinical interview.
Results: Three different pathways were identified: one subset of patients not having a psychiatric diagnosis at either baseline or 16-mo follow-up (68% for depression, 51% for anxiety), one group having an intermittent course (21% for depression, 34% for anxiety), and one group having a persistent course (11% for depression, 15% for anxiety), with diagnoses at both time 1 and time 2. For depression, the people with the persistent course showed marked decreases in quality of life and self-reported health status compared with the nondepressed and intermittently depressed cohorts. The most powerful predictor of depression at time 2 is degree of depressive affect at time 1(P < 0.05).
Conclusions: Patients at risk for short- and long-term complications of depression can be potentially identified by high levels of depressive affect even at a single time point. As nearly 20% of the sample had chronic depression or anxiety, identifying a psychiatric diagnosis in hemodialysis patients and then offering treatment are important because, in the absence of intervention, psychiatric disorders are likely to persist in a substantial proportion of patients.
Depression and anxiety are the primary psychiatric problems of end-stage renal disease (ESRD) patients (1–5). Psychiatric diagnosis, as opposed to levels of depressive affect or anxiety symptoms, has been gaining increasing attention as an authoritative measure of psychopathology in ESRD populations (6–8). In a previous report (9), we identified 29% of our urban hemodialysis sample as having either a current major depressive disorder or a milder more chronic depression, dysthymia, using the ‘gold standard’ of psychiatric diagnosis, the Structured Clinical Interview for the DSM-IV (Structured Clinical Interview for DSM-IV, SCID-I). We have also found that 46% of the sample had an anxiety diagnosis (10). This is the first known study, to our knowledge, to report on the course of depression and anxiety diagnoses in an ESRD population. Kimmel et al. (11) demonstrated the need to track the course of depression, as there were significant differences in morbidity and mortality for subjects with persistently high levels of depressive affect when compared with those with intermittently high levels or low levels of depressive affect. We present data for this ESRD sample at baseline and at 16-mo follow-up. We predicted that subjects with a chronic course of psychopathology (diagnosed at both time points) would report lower quality of life than subjects with either no psychiatric diagnosis or subjects with a psychiatric diagnosis at one time point, but not the other. We were also interested in identifying any baseline characteristics that would be associated with a persistent course of depression or anxiety.
Materials and Methods
This study was approved by the SUNY Downstate Institutional Review Board. Seventy participants were originally randomly selected from patients with ESRD treated at the adult hemodialysis center in central Brooklyn. A full description of the methodology has already been published (9). All subjects in the initial study were eligible for participation in this follow-up study. Subjects were compensated $20 for their time upon the successful completion of all measures, which took an estimated 1 to 1.5 h. All measures were administered at both time 1 and time 2.
Measures
The SCID-I Depression and Anxiety Modules (12).
The SCID is a semistructured interview that provides the major Axis I DSM-IV diagnoses. It has variable but acceptable reliability and validity and is accepted as the ‘gold standard’ for deriving psychiatric diagnoses in research studies. It has been previously used in ESRD populations (7,8,13). Only the depression and anxiety modules were readministered to the subjects to reduce participant burden.
Beck Depression Inventory-II (BDI).
The BDI (14) is a 21-item self-report instrument, with high scores indicating higher levels of depressive affect. It has been used extensively in ESRD populations (11,15–18).
Kidney Disease Quality of Life Short Form (KDQOL-SF).
The KDQOL-SF (19) assesses the quality of life of patients with kidney disease. This is accomplished with 43 disease-specific items and 36 generic (SF-36) items. Kidney disease-specific items include Symptom/problem list, Effects of kidney disease, Burden of kidney disease, Work status, Cognitive function, Quality of social interaction, Sexual function, Sleep, Social support, Dialysis staff encouragement, Patient satisfaction, and Overall health rating. The KDQOL has been used widely in ESRD populations (20–22).
Life Events Stress Scale.
The Life Events Stress Scale (Holmes and Rahe [23]) was used to measure the quantity of stressful life events experienced by the subjects in the time between the two assessment points. This scale lists 43 positive and negative life events, and each has a value between 1 and 100. A total corrected score (range, 1 to 2) indicates the quality and quantity of stressful life events, with higher scores reflecting more stressful events endured by the subject. It has been used widely in medical populations (24).
Clinical Variables
The patients’ charts were extracted for standard laboratory values. Monthly measures of serum albumin concentration, urea reduction rates (URR), and the calcium phosphate product were collected at baseline and at the 16-mo follow-up.
Data Analysis
All data were analyzed using the computer-based statistical software package SPSS, version 16.0. Baseline characteristics of those subjects with follow-up information were compared with subjects without follow-up information using t test. For binary variables, a cross-tabs with χ2 was used. Subjects were divided by the course of their depression diagnosis (presence or absence at time 1 and/or time 2), and their quality of life, depressive affect, and self-reported health status were compared, as well as their average serum albumin concentration, URR, and calcium phosphate product. Group differences were compared with analysis of variance, using the Bonferroni method for post hoc comparisons. Then a similar analysis was undertaken for course of anxiety diagnoses, in which subjects were divided by course of SCID anxiety diagnosis and compared on quality of life and self-reported health status, as well as clinical variables. Finally, as an exploratory analysis in this pilot study, time 1 predictors of SCID diagnosis at time 2, hierarchical logistic regression analyses were used.
Results
Of the 70 randomly selected patients that were originally interviewed, at our follow-up interview about 16 mo later, 15 were ineligible for follow-up (7 died, 7 transplanted, and 1 discharged from hemodialysis treatment), 4 were lost to follow-up (moved out of state), and 4 refused, for a total of 47 responders. Of the eligible 55 subjects from baseline, 85% participated in the follow-up interview.
Baseline Differences
There were no statistically significant differences when baseline characteristics of those with and without follow-up data were compared. Age (t = 0.64, P > 0.05), gender (χ2 = 0.94, P > 0.05), length of dialysis treatment (t = 0.92, P > 0.05), proportion with a diagnosis of depression (χ2 = 0.65, P > 0.05), BDI (t = 0.058, P > 0.05), and QOL (t = −0.65, P > 0.05) were all not significantly different between groups. There were no baseline differences in demographic or psychologic variables for those who died by time 2 and those who were still alive (P > 0.05).
All subjects that were diagnosed with a mental health diagnosis at either time 1 or time 2 were informed of their diagnosis. Additionally, the participant was provided with a referral to the hospital's mental health clinic and the attending nephrologists were notified. Based on clinical records at the outpatient clinic, none of the participants had followed up with the referral at our outpatient clinic, and all of those that have follow-up data available denied pursuing treatment elsewhere. Although this statistic is overwhelming, it is unfortunately not surprising. Very few medical patients receive appropriate care for their mental health needs. The primary reasons identified for this sample's lack of follow-up at the mental health clinic are 1) the stigma of mental health treatment in our community, both from the subjects themselves and their social environment; and 2) the burden of the additional prescription/appointment. It is from these barriers to care that we have now implemented a psychosocial chairside intervention, as we think that this model of service delivery minimizes stigma and avoids an additional appointment as the intervention can be done while the person is being dialyzed.
Course of Depression
As reported previously, 29% (n = 20) of the original sample was diagnosed with a depressive disorder using the SCID-I. Of those with a depressive disorder at time 1, follow-up data were available on 12 subjects (60%). Of these subjects, 7 (58%) no longer qualified for a SCID diagnosed depressive disorder at time 2, and the remaining 42% still had a depression diagnosis. All of those that remitted had been diagnosed with a major depressive disorder at time 1. The two subjects from time 1 that had been diagnosed with dysthymia at time 1 were diagnosed with a major depressive disorder at time 2. An additional 3 subjects were newly diagnosed with depression (all dysthymia) at time 2, for a total of 17% (8 of 47) of the time 2 sample with an SCID-diagnosed depressive disorder.
Subjects were divided into three groups: nondepressed (no depression at either time 1 or time 2, n = 32), intermittently depressed (depression diagnosis at either one of the time points, n = 10), and persistently depressed (depression diagnosis at both time points, n = 5). Group differences are listed in Table 1. Baseline comparisons are displayed in Table 1 and follow-up data in Table 2. Subjects who were not depressed had statistically and clinically lower baseline BDI scores (6.3 ± 4.1) compared with both the intermittently depressed (16.9 ± 6.5) and the persistently depressed (24.6 ± 12.7) subjects (F(2) = 39.2, P < 0.001). There were no significant between group differences for any of the clinical markers (serum albumin (F(2) = 0.84, P > 0.05), URR (F(2) = 0.26, P > 0.05), or calcium phosphate product (F(2) = 1.5, P > 0.05).
Psychological and clinical values by depression diagnosis course
Psychological and clinical values by anxiety diagnosis course
Course of Anxiety
A total of 45% (n = 33) of the original sample was diagnosed with an anxiety disorder (Specific Phobias, Panic Disorder (± agoraphobia), Obsessive Compulsive Disorder, Posttraumatic Stress Disorder, Generalized Anxiety Disorder, Social Phobia, Anxiety Disorder NOS) at baseline (10). Of those with an anxiety disorder at time 1, follow-up data are available on 21 subjects (64%). Of these subjects, 14 (66%) no longer qualified for an SCID-diagnosed anxiety disorder at time 2, but the remaining 33% (n = 7) still had an anxiety diagnosis. An additional 2 subjects were newly diagnosed with anxiety at time 2, for a total of 19% (9 of 47) of the time 2 sample with an SCID-diagnosed anxiety disorder.
Similar to the depression analysis, subjects were divided into three groups: nonanxious (no anxiety diagnosis at either time 1 or time 2, n = 24), intermittently anxious (anxiety diagnosis at either one of the time points, n = 16), and persistently anxious (anxiety diagnosis at both time points, n = 7) by SCID diagnosis. Group differences for baseline and follow-up data are given in Table 2. There were no significant between-group differences on any of the baseline psychologic (BDI (F(2) = 3.3, P > 0.05), KDQOL (F(2) = 2.6, P > 0.05), SF-36 (F(2) = 0.85, P > 0.05)) or clinical markers (serum albumin (F(2) = 0.33, P > 0.05), URR (F(2) = 1.8, P > 0.05), or calcium phosphate product (F(2) = 0.50, P > 0.05)). There were also no statistical differences between anxiety groups on follow-up regarding psychologic (BDI (F(2) = 4.3, P > 0.01), KDQOL (F(2) = 1.7, P > 0.05), SF-36 (F(2) = 2.4, P > 0.05)) or clinical markers (serum albumin (F(2) = 0.52, P > 0.05), URR (F(2) = 2.5, P > 0.05)), or calcium phosphate product (F(2) = 0.60, P > 0.05).
Comorbid Depression and Anxiety
Of those with follow-up data available, 9% of the original sample had both anxiety and depression. At time 2, 13% of the sample had comorbid depression and anxiety. Two thirds of the individuals with comorbid depression and anxiety at baseline still carried both diagnoses at 16 mo.
Predictors of Depression
Hierarchical logistic regression analyses were used to determine the relationships among time 1 variables (QOL, SF-36, BDI, and baseline depression diagnosis) and SCID diagnosis of depression at time 2 while accounting for variance in gender, age, and length of time on dialysis. Results of these analyses are presented in Table 3. Presence/absence of SCID diagnosed depression at time 2 was the dependent variable and sociodemographic variables were entered first. This step of the model was not significant (χ2 = 2.58, P = 0.46) and only accounted for 9% of the variance in SCID diagnosis. In the second step, time 1 psychologic variables (QOL, SF-36, BDI, baseline SCID) accounted for an additional 72% of the variance (χ2 = 24.02, P < 0.001) in SCID depression diagnosis. Only BDI at time 1 (P < 0.05) was a statistically significant predictor of time 2 depression diagnosis. Depression diagnoses at time 1 (P > 0.05), QOL (P > 0.05) or SF-36 (P > 0.05) were not significantly associated with time 2 depression diagnosis.
Logistic regression of sociodemographic and time 1 psychological variables on depression diagnosis at time 2
Predictors of Anxiety
In a similar analysis, hierarchical logistic regression was used to determine the relationships among time 1 variables (QOL, SF-36, BDI, and baseline anxiety diagnosis) and SCID diagnosis of anxiety at time 2 while accounting for variance in gender, age, and length of time on dialysis. Results of these analyses are presented in Table 4. Presence/absence of SCID diagnosed anxiety at time 2 was the dependent variable and sociodemographic variables were entered first. This step of the model was not significant (χ2 = 3.43, P = 0.33) and only accounted for 11% of the variance in SCID diagnosis. In the second step, time 1 psychologic variables (QOL, SF-36, BDI, baseline SCID) accounted for an additional 42% of the variance (χ2 = 11.48, P < 0.05) in SCID anxiety diagnosis. Only anxiety diagnosis at time 1 (P < 0.05) was a statistically significant predictor of time 2 anxiety diagnosis. Depressive affect at time 1 (P > 0.05), QOL (P > 0.05), or SF-36 (P > 0.05) was not significantly associated with time 2 anxiety diagnosis.
Logistic regression of sociodemographic and time 1 psychological variables on anxiety diagnosis at time 2
Discussion
This study reports the results of a 16-mo follow-up of the baseline rates of psychopathology in an ESRD population treated with hemodialysis. Of those with follow-up data available, 42% of those with diagnosed depression at baseline still warranted a depression diagnosis 16 mo later, and 33% of those with an anxiety diagnosis still had an anxiety diagnosis at follow-up. Those who had a depression diagnosis at both time periods had significantly lower quality of life and more severe depressive affect, both at baseline and follow-up. It would appear that there are three different courses for depression in this population. The majority of people do not have a depression diagnosis and have low levels of depressive affect, similar to nonmedical samples. There is another subgroup that has intermittent problems with depression. This group has moderately high levels of depressive affect and a reduced quality of life compared with people with no depression. The most severe course is for those people who have persistent depression. This course is associated with markedly higher levels of depressive affect and significantly reduced quality of life. These data are consistent with recent findings on the course of chronic depression in psychiatric populations (25) and earlier findings that consistently high levels of depressive affect are a significant predictor of mortality in ESRD patients treated with hemodialysis (11).
Interestingly, this pattern did not emerge for the anxiety diagnoses. There were no differences between those with intermittent or chronic courses to their anxiety on measures of quality of life, health status or clinical variables. In comparison, two thirds of the people that had comorbid depression and anxiety still carried both diagnoses at time 2, possibly suggesting that the co-occurrence of these two disorders yields a course less likely to naturally remit.
It is interesting to note that subjects with intermittent anxiety appeared to fare worse than those with chronic anxiety. It is unclear why this might be true, but perhaps a new incidence of an anxiety disorder is in reaction to a change (emotional or medical) in the patient's life, and a persistent course of anxiety is less related to environmental triggers and more related to a lifelong biologic/cognitive vulnerability to anxiety.
The observed differences between the course of anxiety and depression could either be due to measurement error or they could reflect legitimate differences between depression and anxiety. As this study has a limited sample size and was based in a single hemodialysis site, it is possible that the effect of depression is more robust and easily detectable and that, to observe the effect for anxiety, a larger sample would be required. However, it is also possible that depression has a greater effect on quality of life and health status than anxiety in ESRD populations.
Depression has been suggested to affect medical outcomes in ESRD patients through modification of immunologic and stress responses, impact on nutritional status, and/or reduction of compliance with prescribed dialysis and medical regimens (3,5,26). Recent biologic studies suggest that proinflammatory cytokines may mediate the behavioral and neurochemical features of depression (27–30). Many of the same inflammatory biomarkers are known to be dysregulated in ESRD patients, so perhaps there is a direct biologic link between increased levels of depression and renal disease (31–33). There is some indication that more depressed people tend to be more malnourished, as some studies have shown an association between depression and markers of malnutrition (15,34,35). Finally, a relationship between depressive affect and both laboratory and behavioral markers of poor compliance in dialysis patients has been demonstrated (36–38) in which decreased compliance with the dialysis prescription was associated with increased depressive affect in hemodialysis patients (15,37,39). It is not known if people who are depressed lack the energy/motivation to comply with the dialysis prescription, or people who are noncompliant become more ill and more depressed, but in either scenario there exists a vicious cycle between noncompliance and depression.
Another major finding of this paper is that level of depressive affect is a more powerful predictor of depression 16 mo later, than other psychologic measures, including baseline depression diagnosis. This is important as the call to screen for depression in hemodialysis centers has been growing (40), and this study suggests that patients at risk for short- and long-term complications of depression potentially can be easily identified by high levels of depressive affect even at a single time point.
In contrast to this pattern, the best predictor of anxiety diagnosis at time 2 was anxiety diagnosis at time 1. We did not include any continuous measure of anxiety intensity, and perhaps the Anxiety Sensitivity Index would have been an appropriate measure. However, the strength of the association between anxiety diagnosis at time 1 and time 2 clearly highlights the need for psychiatric intervention.
Our results suggest that, for a significant minority of hemodialysis patients, psychiatric difficulties are more than transient reactive states but persistent potentially complicating chronic illnesses. Our findings suggest that those with the highest levels of depressive affect are at most risk for this chronic course. These results are preliminary, and these findings must be replicated in a larger sample where regression analyses with several variables would be more reliable, and in a more diverse population to establish generalizability.
Identifying psychiatric diagnosis in HD patients is important because the disorder is likely to be persistent in a substantial proportion of patients. In our sample, nearly 20% of patients had chronic depression or anxiety. Making a specific psychiatric diagnosis versus a more general assessment of ‘psychic distress' is important as depression and anxiety are not exclusively co-occurring disorders, and they have different courses that require different treatments (41).
The need for aggressive targeted treatment of depression and anxiety at the hemodialysis center is clear. As there is mounting evidence that both pharmacologic (16,42–46) and cognitive behavioral strategies (4,47,48) are effective choices for depression treatment for hemodialysis patients, every center should routinely screen their patients for depression and actively treat those with high levels of depressive affect.
Disclosures
None.
Acknowledgments
Dr. Cukor is currently supported by a K-23 (DK076980-01) award from the National Institute of Diabetes and Digestive and Kidney Diseases.
Footnotes
Published online ahead of print. Publication date available at www.cjasn.org.
- Received March 10, 2008.
- Accepted June 27, 2008.
- Copyright © 2008 by the American Society of Nephrology
References
In this issue
