- Email: [email protected]
Mastery and Neuroticism Predict Recovery of Depression in Later Life
Mastery and Neuroticism Predict Recovery of Depression in Later Life Bas Steunenberg, M.Sc., Aartjan T. F. Beekman, M.D., Ph.D., Dorly J. H. Deeg, Ph.D., Marijke A. Bremmer, M.D., Ad J. F. M. Kerkhof, Ph.D.
Objective: The authors examined whether personality characteristics such as mastery, self-efficacy, and neuroticism predict the likelihood of recovery of depression among elderly in the community. It was hypothesized that these personality characteristics do predict recovery but that their effect is overwhelmed by the effect of deteriorations in physical health, cognitive decline, and loss of social resources. The second research question investigated whether these personality characteristics moderate the negative impact of the other prognostic factors on the chance of recovery. Methods: A prospective (nine-year) follow-up study of 206 depressed elderly (55– 85 years at baseline) participants of the Longitudinal Aging Study Amsterdam. Data on chance of recovery were analyzed using Cox proportional regression analyses. Results: Both in the univariate and in the multivariate model, the personality characteristics, especially neuroticism, predicted recovery of depression. The effect of neuroticism was similar to that of physical health and stronger than the impact of cognitive decline or social resources. No support was found for personality as a moderator of the negative impact of age-related stressors. Conclusions: Personality characteristics, i.e., neuroticism and physical health-related variables are separate but equally important domains for the chance of recovery of depression in later life. (Am J Geriatr Psychiatry 2007; 15:234–242) Key Words: Mastery, neuroticism, recovery, depression, late-life
A
growing body of literature has shown that the prognosis of late-life depression in communitydwelling elderly deteriorates with age.1,2 Both for clinical and theoretical purposes, understanding the pathways to a poor prognosis are of major importance. Identifying prognostic factors associated with
an unfavorable course may result in both a better understanding of the natural history and in more effective preventive interventions. In search of prognostic factors, much emphasis has been laid on the association of physical health and cognitive functioning with the course of late-life depression. Physical
Received October 31, 2005; revised April 7, 2006; accepted April 12, 2006. From the Departments of Clinical Psychology (BS), Psychiatry (ATFB), and Clinical Psychology (AJFMK), Vrije Universiteit, Amsterdam; the Research Institute Psychology and Health (BS, AJFMK), Utrecht; and the Institute for Research in Extramural Medicine (EMGO) (BS, ATFB, DJHD, MAB, AJFMK) and the Department of Psychiatry (DJHD, MAB), Vrije Universiteit Medical Centre. Send correspondence and reprint requests to Dr. Bas Steunenberg, Department of Clinical Psychology, Vrije Universiteit, Amsterdam, Van der Boechorststraat 1, 1081 BT Amsterdam. e-mail: [email protected] © 2007 American Association for Geriatric Psychiatry
234
Am J Geriatr Psychiatry 15:3, March 2007
Steunenberg et al. health problems such as chronic limiting diseases, functional limitations, and pain or cognitive impairment, which are frequently encountered in later life, are often shown to be closely related to the incidence and to a poor prognosis of depression.3–5 Physical health problems have even been said to overwhelm the impact of other prognostic factors.4,6 Other factors associated with a poor prognosis include older age, female gender, and lack of social support. Social demographic factors such as marital status and level of education have not been found to be related with outcome.1,6 – 8 There is some debate about the association of personality with the prognosis of depression in late life. Studies have documented that personality characteristics such as poor mastery, low self-efficacy, and neuroticism are risk factors for the onset and severity of depression in later life,9 –11 but these studies did not examine the association between personality and the prognosis of depression in late life. Mastery is defined as the extent to which a person feels that he or she has control over his or her life and environment.12 Bandura13 refers to self-efficacy as the conviction that one can successfully effectuate the behavior required to produce a given outcome in a specific situation. People with a strong sense of mastery (internal locus of control) and those with positive expectations of self-efficacy have been found to be less depressed than people with a low sense of mastery (i.e., external locus of control) or a low level of self-efficacy.14 Neuroticism refers to a tendency to experience negative emotions as well as a tendency toward emotional instability. Individuals who score high on neuroticism are more likely to experience a variety of problems, including negative moods (such as anxiety, fear, depression, and irritability) and physical symptoms. Personality may not only directly influence the level of depressive symptoms, but it may also modify the adverse effects of age-related stressful life situations. A higher level of mastery seems to facilitate adaptation under stressful events such as medical events and functional decline.9 –11,15,16 A study by Mendes de Leon et al.17 revealed that a high level of self-efficacy buffers the negative impact of functional decline on depressive symptoms. Contrary to onset and severity, little systematic research has been done into the association of personality characteristics with the prognosis of depression in late life.1,18 The aim of the present study was
Am J Geriatr Psychiatry 15:3, March 2007
to examine the association of mastery, self-efficacy, and neuroticism with the prognosis of depression in late life in the community. To this end, using a nineyear prospective longitudinal design, those with a recovery of symptoms were compared with those who did not recover, i.e., had a poor prognosis. We investigated the relative prognostic strength of personality characteristics compared with physical health-related prognostic and social resource factors. Next, we investigated whether personality characteristics modify the negative impact of other prognostic factors on depression. To our knowledge, we are the first to examine these relationships in a community sample of depressed elderly in a longitudinal study spanning a period of nine years.
METHODS Sampling and Procedures Data for this study were collected in the Longitudinal Aging Study Amsterdam (LASA), an ongoing interdisciplinary longitudinal study on predictors and consequences of changes in autonomy and wellbeing in the aging population.19 Sampling procedures and response were described in detail elsewhere.20,21 In short, a random sample of older (55– 85 years) men and women was drawn from the population registers in 11 municipalities in three geographic regions of The Netherlands. The sample was stratified by age and sex according to expected fiveyear mortality to ensure sufficient sample size for longitudinal analyses within age and sex strata. The LASA study started in 1992/1993 and includes follow-up measurements after three, six, and nine years. In the first cycle of LASA, 3,107 respondents were included. Respondents were visited at home by trained interviewers. Interviews and tests took approximately 1 hour 30 minutes. To spread the burden for the respondent, the baseline cycle consisted of two interviews and a self-administered questionnaire, including the personality variables. Loss of respondents by attrition or nonresponse was described in detail elsewhere,2,22 but is summarized subsequently for our study sample. For this study, data from all four available measurements were used spanning a period of nine years.
235
Mastery and Neuroticism Predict Recovery of Depression All 3,107 respondents to the baseline interview were screened for depression using the Center for Epidemiologic Studies Depression Scale (CES-D).23 To investigate the influence of personality on the chance of recovery of depressive symptoms, only respondents with depressive symptoms at baseline (N⫽448, CES-D ⱖ16; 14.9% relative to LASA baseline) and complete data at baseline and first follow up were included in this study (67 respondents were lost). Missing data on the personality questionnaires were imputed using the mean score only if respondents had answered at least 80% of the items on these questionnaires (175 respondents were lost). These restrictions resulted in a study sample of 206 depressed respondents. In multivariate analyses (logistic regression), the baseline characteristics of our study sample were compared with the subjects who were lost as a result of either loss to follow up between baseline and second measurement or item nonresponse. A higher age ( ⫽0.048; standard error [SE]: 0.02, p ⫽ 0.05), female gender ( ⫽⫺0.09; SE: 0.03, p⫽0.03), one or more chronic diseases ( ⫽0.60; SE: 0.28, p⫽0.03), or functional limitations (: 0.43; SE: 0.22, p⫽0.05) were found to be significantly associated with attrition. No difference between our study sample and those lost was found for level of depressive symptoms, urbanization or education level, pain, cognitive functioning, network size, level of emotional and instrumental support received, and loneliness. From the study sample, 51 respondents were lost between the second and third measurement. In the last three years between third and fourth measurements, another 32 respondents were lost. Reasons were: dead before approach (46 [55%]), no longer interested/motivated (11 [13%]), physically (2 [2%]) or cognitively (3 [4%]) not able to participate or reason not specified (3 [4%]), and 18 (22%) respondents had too many missing values on the key variables under study. Loss to follow up was found to be related to higher age ( ⫽0.07; SE: 0.02, p⫽0.001). Respondents who recovered, but were lost later after the second measurement, were included in the analyses. Measurements Course of Depression. Depressive symptoms were measured using the Dutch translation24 of the CES-D.25 This is a 20-item self-report scale developed to measure depressive symptoms in the community. Re-
236
spondents were asked how often they had experienced each symptom during the previous week. Items were scored on a four-point scale ranging from 0 (rarely or none of the time) to 3 (most of or all of the time). The total CES-D score ranges from 0 (no symptoms) to 60 (maximum number of symptoms). A score of ⱖ16 has generally been used as indicative for clinically relevant depressive syndromes.25 The CES-D has been widely used in older community samples, and the psychometric properties were found to be satisfactory.26,27 Because of the emphasis on affective items in the scale, the overlap with symptoms of physical illness is minimal.25,28 To study predictors related to the prognosis of depression, two course types were defined applying a definition of a clinically relevant change in depressive symptoms. A relevant change was defined as a decrease or an increase of at least five points on the CES-D between two measurements, thereby crossing the cutoff score of 16. The difference of five points is in line with the definition of a reliable change,29 which takes into account the reliability of the CES-D and with the principle of a medium effect size.30 A similar definition of a relevant change was used in earlier studies.6,31,32 The following two course types were distinguished: recovery (a relevant decline of symptoms between two adjacent measurements and the subject remaining nondepressed (CES-D score of ⬍16) throughout the rest of the study) and no recovery. The course type no recovery includes intermittent (after recovery, the subject had a relevant increase of symptoms later on in the study) and persistent (CES-D score ⱖ16 at all four measurements32) patterns of change of symptoms. The following variables measured at baseline were examined as potential predictors of the prognosis of depression during the nine-year study period. Personality Characteristics. Mastery was measured by means of a translated and abbreviated Dutch version of the Pearlin Mastery Scale.12 The questionnaire consists of five statements such as “I cannot seem to be able to solve some of my problems at all.” Scale scores range between 5 and 25. Higher scores on the scale indicate a high mastery. General self-efficacy was measured by means of the Perceived Self-Efficacy Scale, which is based on 17 items of the General Self-Efficacy Scale (GSES33). Based on the results of the pilot study of LASA, the
Am J Geriatr Psychiatry 15:3, March 2007
Steunenberg et al. GSES was shortened to a version with 12 items. An example-item is: “If I make plans, I am convinced I will succeed in executing them.” A minimum score of 12 indicates the most negative perceived self-efficacy score and 60 the most positive. Neuroticism was operationally defined by using a subset of 25 items, out of a list of 36 neuroticism items, from the Dutch Personality Questionnaire (DPQ34). The pilot studies of the LASA demonstrated that the original 36-item scale measuring neuroticism could be abbreviated.35,36 Respondents were asked to indicate whether various similar statements applied to them (yes, do not know, no). The scores can range between 0 and 50. An example of a neuroticism item is: “I often hate myself.” These DPQ items have strong negative relations with the Emotional Stability Scale of the NEO-PI-R.37 For reasons of clarity of presentations, predictors can now be directly compared; we have chosen to dichotomize the scales at the 50th percentile of the mean score. In analyses with the complete scales, we found similar results like with the dichotomized scales. Sociodemographic Variables. Sociodemographic variables included were age, sex, level of education, marital status, and urbanization level of the municipality. Age ranged from 55– 85 years at baseline. Education was dichotomized in: low education (uncompleted or completed primary school) and middle to high education (secondary education up to and including university). Marital status was dichotomized in married and not or no longer being married. Urbanization level of municipality was dichotomized in less urban and highly urban. Health Variables. Functional limitations in daily life was assessed by a three-item questionnaire,38 and for the present study, the number of functional limitations was dichotomized into 0⫽no limitations and 1⫽one or more limitations. The number of chronic diseases was calculated by summing all diseases reported by the respondent to be present.39 In a validation study, respondents’ selfreports were compared with information obtained from their general practitioners’ and proved to be sufficiently reliable.40 For the present study, the number of chronic diseases was dichotomized into 0⫽no disease and 1⫽one or more diseases. Pain was measured with a subscale of the Nottingham Health Profile.41 The following items were included: “I am in pain when I am standing,” “I find it
Am J Geriatr Psychiatry 15:3, March 2007
painful to change position,” “I am in pain when I am sitting,” “I am in pain when I walk,” and “I am in constant pain.” Response categories were “no” (0) and “yes.”1 Total scores range from 0 (low) to 5 (high). This scale was dichotomized at 0 ⫽ no pain and 1⫽ pain present. Cognitive functioning was measured by means of the Mini-Mental State Examination.42 On 23 questions and tasks, respondents received one or more points when they gave the correct answer or performed the task correctly. Scores ran from 0 to 30 using the commonly used cutoff score of ⱕ23 as an indicator for cognitive impairment. This score has been shown to be 80%–90% sensitive and 80%–100% specific for a diagnosis of dementia.43– 45 Social Resources. Measures of social support were constructed from a questionnaire, which included detailed questions on both the number and the quality of contacts with members of the social network.46 In this article, we report on the size of the contact network and the exchange of received emotional and instrumental support with network members. Loneliness was assessed with a scale developed by Jong-Gierveld and de Kamphuis.47 The scale has been used in several surveys and has been proven to be a robust and valid instrument.48 The scale consists of 11 items and the total scale score ranges between 1 and 11. All four predictors were dichotomized at the 50th percentile of the mean score. Statistical Analyses The direct effect of personality characteristics and the interaction with other predictors on the probability of recovery within the nine-year follow-up period was investigated in a Cox proportional hazards design.49,50 Cox regression model estimates the effect of predictor variables on the hazard rate, in this case being the time to recovery during follow up. P values lower than 0.05 were regarded as statistically significant. Hazard ratios are a measure of the strength of the association of the predictor with the time to recovery. A hazard ratio of 1 indicates no effect of the predictor on the time to recovery. The time from baseline measurement to recovery (event) or to the end of the study without recovery (censored data) was calculated per study cycle with intervals of three years. Respondents who recovered but were lost
237
Mastery and Neuroticism Predict Recovery of Depression later (i.e., deceased during the study) were included in the analyses as an event. First, we examined the chance of recovery for each predictor separately. Then, in a multivariate model, we included all predictors that were associated with recovery in the univariate analyses with a p value ⬍0.05. Finally, to find out whether personality acts as a buffer, mitigating the negative impact of potential predictors on the outcome of depression, interaction effects between the personality characteristics and potential predictors at baseline were investigated. We only investigated interaction effects between personality characteristics and potential predictors that were significantly associated with recovery in the univariate analyses with a p value ⬍0.05. When an interaction effect showed a p value ⬍0.10, stratified analyses were carried out. Stratified analyses explore the nature of the interaction effect, i.e., they estimate the strength of the association between predictor and depression for the different subgroups on the personality characteristics. Statistical analyses were performed using SPSS 12.0.
RESULTS Baseline demographic and clinical characteristics of the study sample are presented in Table 1. The mean age of the 206 respondents was 69.6 years at baseline. Sixty-eight percent were female. During the nineyear follow up, 91 (44%) respondents remitted. Most respondents (69 [33%]) remitted between baseline and first follow up three years later. Fourteen (7%) respondents remitted between first and second follow up and eight (4%) respondents remitted over the last three years. Prognostic Factors for Recovery of Late-life Depression Table 2 presents the results of the univariate Cox proportional hazard regression analyses. Recovery was predicted by high mastery (odds ratio [OR]: 1.7, p⫽0.01) and a low level of neuroticism (OR: 1.6, p⫽ 0.03) at baseline. Male respondents depressive at baseline had a higher chance on recovery during the
238
TABLE 1.
Demographic and Clinical Characteristics of the Baseline Depressed Study Sample (N ⴝ 206)
Characteristic Age ⬍70 years ⱖ70 years Sex Male Female Education level Low Middle/high Marital status Married Not or no longer married Municipality Less urban Highly urban Functional limitations None One or more Number of chronic diseases None One or more Pain No pain Low to severe pain Cognitive Functioning Normal Impaired
N
Percent
111 95
54 46
66 140
32 68
136 70
66 34
104 102
50 50
135 71
66 34
92 114
45 55
61 145
30 70
89 117
43 57
186 20
90 10
follow-up period (OR: 1.7, p⫽ 0.01). Absence of functional limitations (OR: 2.1, p ⫽0.001) and pain (OR: 1.8, p⫽0.01) at baseline also were positively associated with recovery. For no pain, a hazard ratio of 1.8 was found, meaning that the probability of recovery was 80% greater for respondents with lack of pain at baseline compared with respondents with pain. Controlling for confounding among correlated predictors, we included all predictors that were significantly associated with recovery in the univariate analyses in a multivariate model. Two of the predictors with a unique effect remained significant in the multivariate model (Table 2). Both low level of neuroticism (OR: 1.5, p⫽0.05) as well as no functional limitations (OR: 1.7, p⫽0.03) were found to be significantly related to recovery. Mastery no longer had a unique predictive effect, which holds for gender and pain as well. Third, we investigated whether the personality characteristics moderate the association between the other prognostic factors and the chance of recovery by entering interaction terms to the regression model. We investigated whether the personality characteristics mastery and neuroticism moderate
Am J Geriatr Psychiatry 15:3, March 2007
Steunenberg et al.
TABLE 2.
Predictors of Remission of Depression in Older Adults: Results of Cox Proportional Hazard Regression Analysis (univariate and multivariate model, N ⴝ 206)
Predictors Multivariate Modela HR (95 % CI)
Univariate Model HR (95 % CI) Low level of depressive symptoms at baseline Personality characteristics High level of mastery High level of self-efficacy Low neuroticism Social demographics Age ⱕ70 years Male gender Middle/high level of education Married Less urban Physical health-related variables No functional limitations No chronic diseases No pain Normal cognitive functioning Social resources Large network size High level of instrumental support received High level of emotional support received No loneliness
1.3 (1.0–1.6) 1.7b (1.1–2.7) 1.5 (1.0–2.2) 1.6c (1.0–2.5)
1.4 (0.9–2.3) 1.5b (1.0–2.4)
1.5 (1.0–2.3) 1.7b (1.1–2.6) 1.5 (1.0–2.3) 1.3 (1.3–1.9) 1.2 (0.8–1.9)
1.4 (0.9–2.1)
2.1c (1.4–3.3) 1.2 (0.8–1.9) 1.8c (1.2–2.7) 1.3 (1.0–1.7)
1.7b (1.1–2.8) 1.3 (0.8–2.1) 1.1 (0.8–1.4)
1.3 (0.9–2.0) 1.3 (0.8–1.9) 1.1 (0.7–1.6) 1.4 (0.9–2.1)
In the multivariate model, only predictors that were associated with remission in the univariate analyses with a p value ⬍0.05 were included. p ⬍ 0.05. c p ⬍ 0.01. HR: hazard ratio; CI: confidence interval. a
b
the effects of deteriorations in the physical healthrelated predictors functional limitations and pain, which were found to be significantly related to the chance on recovery at univariate level. Results showed that none of the interaction terms were significant (Table 3), so the personality characteristics in this study do not moderate the influence of the physical health-related predictors on the chance on recovery. Finally, we investigated whether the association between the personality characteristics mastery and
TABLE 3.
neuroticism differed for male and female respondents. Results revealed there was no gender effect on the association between the personality characteristics and the course of depression (Table 3).
DISCUSSION We aimed to study whether the personality characteristics mastery, self-efficacy, and neuroticism are
Results of Two-Way Interaction Analyses Between Personality Characteristics and Predictors of Remission of LateLife Depression (N ⴝ 206) Personality Characteristics Mastery
Predictors
Estimated Coefficient (SE) ⫺0.05 (0.45) 0.56 (0.46) ⫺0.07 (0.46)
Physical Health-Related Variables Number of functional limitations Pain Gender
Neuroticism Wald 2
Estimated Coefficient (SE)
Wald 2
0.01 (p ⫽ 0.9) 1.5 (p ⫽ 0.2) 0.02 (p ⫽ 0.9)
⫺0.02 (0.45) 0.13 (0.44) 0.24 (0.44)
0.001 (p ⫽ 1.0) 0.09 (p ⫽ 0.9) 0.28 (p ⫽ 0.8)
SE: standard error.
Am J Geriatr Psychiatry 15:3, March 2007
239
Mastery and Neuroticism Predict Recovery of Depression associated with the nine-year prognosis of depression in later life and whether they modify the negative impact of other prognostic factors on the prognosis of depression. In the regression model for univariate effects, high mastery and low neuroticism were identified as significant predictors of recovery. We expected to find the negative impact of physical ill health on the chance of recovery to overwhelm the effect of other factors, especially personality. Our results did not lend support to this hypothesis. The prognostic strength of the physical health-related variables was similar to the strength of the personality characteristics. In a multivariate model correcting for the confounding effects of other predictors, low neuroticism and no functional limitations were found to be a unique significant prognostic factors for recovery. The relative prognostic strength of low neuroticism and no functional limitations was hardly affected by adjusting for the effects of the other significant predictors. Demonstrating that personality characteristics, like neuroticism, are stable in the context of acute change is crucial to models proposing that personality can influence the course of depression. However, although findings show that absolute changes in personality may be observed, effect sizes have been extremely small and have been taken as evidence for stability rather than as evidence for change.51,52 Both the relative stability of personality and its absolute change could not be accounted for by individual differences or absolute changes in depressive symptoms.53 In conclusion, we assumed the personality characteristics measured at baseline to be stable over the nine-year follow-up period. It is not precisely clear how a global personality dimension as neuroticism influences the chance of recovery. Scott et al.54 suggested that neuroticism primarily acts as an “amplifier” moderating the impact of dysfunctional cognitions on mood (and thereby, amplifying the entire vicious spiral of cognition and affect). According to Scott et al.,54 the ruminative response style as described by NolenHoeksema55 is the most evident way in which the amplification takes place. Nolen-Hoeksema55 hypothesized that rumination influences depressive mood by interfering with attention and concentration, by enhancing the recall of negative events, and by increasing the likelihood of using depressogenic
240
explanations for negative life events. Future research may subscribe this notion and further determine the complex relationship among neuroticism, cognitions, rumination, and the increased chance of recovery. Self-efficacy did not predict recovery of late-life depression, but higher levels of mastery did predict recovery. Self-efficacy and mastery are somewhat related concepts, because they both refer to a feeling of competence. Mastery, however, pertains to a general feeling of being in control and having power to use personal resources to influence the outcomes of life, whereas self-efficacy concerns the perceived ability of doing what needs to be done to achieve a desired goal in a specific situation. It seems plausible that in late-life self-efficacy, compared with mastery, is more strongly related to physical health. The strength of the association between physical health and depression increases with age, which possibly can explain why self-efficacy was not found to be significantly related to depression. In contrast to these direct effects, our results did not lend support for the hypothesis that personality characteristics moderate the association between the other prognostic factors and the chance on recovery. The present results favor a model of main effects adding onto each other rather than an interactive model with multiplicative interaction of variables, in which vulnerability factors such as personality characteristics and late-life stress such as deteriorations in physical health amplify each other. Our results constitute evidence for the validity of the construct of neuroticism, specifically the component of distress proneness; those high on neuroticism tend to experience more distress across time and regardless of the physical health, cognitive functioning, or social situational resources. Personality disorders are found in substantial numbers in the elderly population and have clinical implications for geriatric depression.56 Personality disorders represent extreme forms of personality characteristics. The highest rates of personality disorder were found among depressed elderly,56 and significant associations between personality disorders and chronicity of depression have been reported.57,58 Lifetime personality dysfunction may affect the course of depression by means of the persistence of somatic worry, distortion in perceived health status, prolonged bereavement, and noncompliance with medical and psychiatric treatment.56 These re-
Am J Geriatr Psychiatry 15:3, March 2007
Steunenberg et al. sults on personality disorders as well as personality characteristics underscore the need to pay attention to the impact of personality psychopathology on the prognosis of late-life depression. The strong points of this study are its long-term prospective design using a community sample of depressed older adults and including a wide range of potential prognostic factors. This study also had some methodological limitations. First, there was considerable attrition at all stages of the study. Analyses of loss of subjects revealed that the frailest elderly were at higher risk for attrition. The results of our study suggest the most frail elderly persons (those with a high level of pain and one or more functional limitations) also had the greatest chance of a chronic course in our nine-year follow-up period. Therefore, although loss of data limits the generalizability of the findings, the results of the analysis pertaining to attrition suggest that, if anything, the true prognosis of late-life depression is underestimated. A second limitation is that there was a measurement only once every three years. Therefore, it is possible that subjects have been depressed between measurements and recovered before the next measurement. The effect of this is that
change is underestimated. However, the purpose of this article was not to distinguish between persistent and chronic intermittent patterns of change of symptoms, but to identify the association of personality characteristics and other potential predictors with the long-term prognosis of depression to distinguish between elderly who recovered and remain nondepressed throughout the rest of the study and elderly with a recurrence later on or a chronic course. A third concern is that no adjustments were made for treatment of depression during our study period. However, in the LASA sample, the level of treatment was very low which is in accordance with earlier studies6,59,60 As a consequence, the results of this study most likely reflect the natural course of depression. In conclusion, the personality characteristics neuroticism and mastery have a strong and independent effect on the prognosis of depression in later life. In our view, along with the widely acknowledged importance of physical health-related variables, attention should be paid to the impact of personality characteristics on the prognosis of depression in later life. However, in the LASA sample, the level of treatment was very low, which is in accordance with earlier studies.
References 1. Cole MG, Bellevance F, Mansour A: Prognosis of depression in elderly community and primary care populations: a systematic review and meta-analysis. Am J Psychiatry 1999; 156:1182–1189 2. Beekman ATF, Geerlings SW, Deeg DJH, et al: The natural history of late-life depression: a 6-year prospective study in the community. Arch Gen Psychiatry 2002; 59:605–611 3. Henderson AS, Korten AE, Jacomb PA, et al: The course of depression in the elderly: a longitudinal community-based study in Australia. Psychol Med 1997; 27:119 –129 4. Prince MJ, Harwood RH, Thomas A, et al: A prospective population-based cohort study of the effects of disablement and social milieu on the onset and the maintenance of late life depression. The Gospel Oak Project VII. Psychol Med 1998; 28:337–350 5. Snowdon J, Lane F: The botany survey: a longitudinal study of depression and cognitive impairment in an elderly population. Int J Geriatr Psychiatry 1995; 10:349 –358 6. Kennedy GJ, Kelman HR, Thomas C: Persistence and remission of depressive symptoms in late life. Am J Psychiatry 1991; 148:174 – 178 7. Kivela S-L, Pahkala K, Laippala P: A one-year prognosis of dysthymic disorder and major depression in old age. Int J Geriatr Psychiatry 1991; 6:81–87 8. Kivela S-L: Long-term prognosis of major depression in old age: a comparison with the prognosis of dysthymic disorder. Int Psychiatry 1995; 7(suppl): 69 –82 9. Jang Y, Haley WE, Small BJ, et al: The role of mastery and social
Am J Geriatr Psychiatry 15:3, March 2007
resources in the associations between disability and depression in later life. Gerontologist 2002; 6:807–813 10. Roberts BL, Dunkle R, Haug M: Physical, psychological, and social resources as moderators of the relationship of stress to mental health in the very old. J Gerontol Soc 1994; 49:535–543 11. Schieman S, Turner HA: Age, disability and the sense of mastery. J Health Soc Behav 1998; 39:169 –187 12. Pearlin LI, Schooler C: The structure of coping. J Health Soc Behav 1978; 19:2–21 13. Bandura A: Self-efficacy: toward a unifying theory of behavioural change. Psychol Rev 1977; 84:191–215 14. Flammer A: Developmental analysis of control beliefs, in SelfEfficacy in Changing Societies. Edited by Bandura A. Cambridge, UK, Cambridge University Press, 1995, pp 69 –113 15. Kempen GIJM, Jelicic M, Ormel J: Personality, chronic medical morbidity, and health elated quality of life among older persons. Health Psychol 1997; 16:539 –549 16. Kempen GIJM, van Heuvelen MJG, van Sonderen E, et al: The relationship of functional limitations to disability and the moderating effects of psychological attributes in community-dwelling older adults. Soc Sci 1999; 56B:S229 –236 17. Mendes de Leon CF, Seeman TE, Baker DI, et al: Self-efficacy, physical decline, and change in functioning in community-living elderly: a prospective study. J Gerontol B Psychol Sci Soc Sci 1996; 51B:S183–190 18. Blazer D: Depression in the elderly. N Engl J Med 1989; 320:164 – 166
241
Mastery and Neuroticism Predict Recovery of Depression 19. Deeg DJH, Knipscheer CPM, van Tilburg W: Autonomy and WellBeing in the Aging Population: Concepts and Design of the Longitudinal Aging Study Amsterdam. NIG-Trend-Studies Number 7. Bunnik, The Netherlands, Netherlands Institute Gerontology, 1993 20. Penninx BWJH, van Tilburg TG, Kriegsman DMW, et al: Effects of social support and personal coping resources on mortality in older age: the Longitudinal Aging Study Amsterdam. Am J Epidemiol 1997; 146:510 –519 21. Beekman ATF, Deeg DJH, van Tilburg T, et al: Major and minor depression in later life: a study of prevalence and risk factors. J Affect Disord 1995; 36:65–75 22. Deeg DJH, van Tilburg TG, Smit JH, et al: Attrition in the Longitudinal Aging Study Amsterdam: the effect of differential inclusion in side studies. J Clin Epidemiol 2002; 55:319 –328 23. Radloff LS: The CES-D scale: a self-report depression scale for research in the general population. Appl Psychol Meas 1977; 1:385–401 24. Beekman ATF, van Limbeek J, Deeg DJH, et al: Screening for depression in the elderly in the community: using the Center for Epidemiologic Studies Depression Scale (CES-D) in The Netherlands. Tijdschr Gerontol Geriatr 1994; 25:95–103 25. Berkman LF, Berkman CS, Kasl SV, et al: Depressive symptoms in relation to physical health and functioning in the elderly. Am J Epidemiol 1986; 124:372–388 26. Himmelfarb S, Murrel SA: Reliability and validity of five mental health scales in older persons. J Gerontol 1983; 38:333–339 27. Radloff LS, Teri L: Use of the CES-D with older adults. Clin Gerontol 1986; 5:119 –136 28. Foelker GA, Shewchuk RM: Somatic complaints and the CES-D. J Am Geriatr Soc 1992; 40:259 –262 29. Jacobsen NS, Truax P: Clinical significance: a statistical approach to defining meaningful change in psychotherapy research. J Consult Clin Psychol 1991; 59:12–19 30. Cohen J: Statistical Power Analysis for the Behavioral Sciences. Hillsdale, NJ, Lawrence A. Erlbaum Associates, 1985 31. Beekman ATF, Deeg DJH, Smit JH, et al: Predicting the course of depression in the elderly: results from a community-based study in The Netherlands. J Affect Disord 1995; 34:41–49 32. Geerlings SW, Beekman ATF, Deeg DJH, et al: Physical health and the onset and persistence of depression in older adults: an eightwave prospective community based study. Psychol Med 2000; 30:369 –380 33. Sherer M, Maddux JE, Mercandante B, et al: The self-efficacy scale: construction and validation. Psychol Rep 1982; 51:663–671 34. Luteijn F, Starren J, van Dijk H: Handleiding bij de NPV. Lisse, Swets & Zeitlinger, 1975 35. Smits CHM, Deeg DJH, Bosscher R: Well-being and control in older persons: the prediction of well-being from control measures. Int J Aging Hum Dev 1995; 40:237–250 36. Steunenberg B, Beekman ATF, Deeg DJH, et al: Neuroticism in the elderly: the utility of the shortened DPQ-scales [in Dutch]. Tijdschrift voor Gerontologie en Geriatrie 2003; 34:118 –124 37. Luteijn F, Starren J, van Dijk H: Manual of the Dutch Personality Inventory, rev ed. Lisse, Swets en Zeitlinger, 2000 38. Kriegsman DMW, van Eijk JT, Penninx BW: Does family support buffer the impact of specific chronic diseases on mobility in community-dwelling elderly? Disabil Rehabil 1997; 19:71–83 39. Health Interview Questionnaire. Heerlen, Central Bureau of Statistics, 1989
242
40. Kriegsman DMW, Penninx BW, Eijk JT, et al: Self-reports and general practitioner information on the presence of chronic diseases in community dwelling elderly. A study on the accuracy of patients’ self-reports and on determinants of inaccuracy. J Clin Epidemiol 1996; 49:1407–1417 41. Hunt SM, McEwen J, McKenna SP: Measuring health status: a new tool for clinicians and epidemiologist. J R Coll Gen Pract 1985; 35:185–188 42. Folstein MF, Folstein SE, McHugh PR: Mini-Mental State: a practical method for the clinician. J Psychiatr Res 1975; 12:189 –198 43. Anthony JC, LeResche L, Niaz U: Limits of the ‘Mini-Mental State’ as a screening test for dementia and delirium among hospital patient. Psychol Med 1982; 12:397–408 44. Tomhaugh TN, McIntyre NJ: The Mini-Mental State Exam: a comprehensive review. J Am Geriatr Soc 1992; 40:922–935 45. Langley LK: Cognitive assessment of older adults, in Assessing Older Persons. Edited by Kane RI. Oxford, UK, Oxford University Press, 2000, pp 65–128 46. van Tilburg TG: Losing and gaining in old age: changes in personal network size and social support in a four-year longitudinal study. J Gerontol 1998; 53B:S313–S323 47. Jong-Gierveld J, de Kamphuis FH: The development of a Raschtype loneliness scale. Appl Psychol Meas 1985; 9:289 –299 48. van Tilburg TG, de Leeuw ED: Stability of scale quality under various data collection procedures: a mode comparison of the de Jong-Gierveld Loneliness Scale. International Journal of Public Opinion Research 1991; 3:69 –85 49. Cox DR: Regression models and life-tables. J R Stat Soc 1972; 34:187–220 50. Cox DR, Oakes D: Analysis of Survival Data. London, Chapman & Hall, 1984 51. Costa Jr, PT McCrae RR, Zonderman AB, et al: Cross-sectional studies of personality in a national sample: 2. Stability in neuroticism, extraversion and openness. Psychol Bull 1986; 94:472–505 52. Steunenberg B, Twisk JWR, Beekman ATF, et al: Stability and change of neuroticism in aging. J Gerontol B Psychol Sci Soc Sci 2005; 60:P27–33 53. Santor DA, Bagby RM, Joffe RT: Evaluating stability and change in personality and depression. J Pers Soc Psychol 1997; 73:1354 – 1362 54. Scott J, Williams MJ, Brittlebank A, et al: The relationship between premorbid neuroticism, cognitive dysfunction and persistence of depression: a one-year follow-up. J Affect Disord 1995; 33:167–172 55. Nolen-Hoeksema S: Responses to depression and their effect on the duration of depressive episodes. J Abnorm Psychol 1991; 100:569 –582 56. Abrams RC: Personality disorders in the elderly. Int J Geriatr Psychiatry 1996; 11:759 –763 57. Devanand DP, Nobler MS, Singer T, et al: Is dysthymia a different disorder in the elderly? Am J Geriatr Psychiatry 1994; 151:1592– 1599 58. Thompson LW, Gallagher D, Czirr R: Personality disorder and outcome in the treatment of late-life depression. J Geriatr Psychiatry 1988; 21:133–146 59. Eisenberg L: Treating depression and anxiety in primary care. Closing the gap between knowledge and practice. N Engl J Med 1992; 326:1080 –1083 60. Freeling P: Diagnosis and treatment of depression in general practice. Br J Psychiatry 1993; 163(suppl): 14 –19
Am J Geriatr Psychiatry 15:3, March 2007
Objective: The authors examined whether personality characteristics such as mastery, self-efficacy, and neuroticism predict the likelihood of recovery of depression among elderly in the community. It was hypothesized that these personality characteristics do predict recovery but that their effect is overwhelmed by the effect of deteriorations in physical health, cognitive decline, and loss of social resources. The second research question investigated whether these personality characteristics moderate the negative impact of the other prognostic factors on the chance of recovery. Methods: A prospective (nine-year) follow-up study of 206 depressed elderly (55– 85 years at baseline) participants of the Longitudinal Aging Study Amsterdam. Data on chance of recovery were analyzed using Cox proportional regression analyses. Results: Both in the univariate and in the multivariate model, the personality characteristics, especially neuroticism, predicted recovery of depression. The effect of neuroticism was similar to that of physical health and stronger than the impact of cognitive decline or social resources. No support was found for personality as a moderator of the negative impact of age-related stressors. Conclusions: Personality characteristics, i.e., neuroticism and physical health-related variables are separate but equally important domains for the chance of recovery of depression in later life. (Am J Geriatr Psychiatry 2007; 15:234–242) Key Words: Mastery, neuroticism, recovery, depression, late-life
A
growing body of literature has shown that the prognosis of late-life depression in communitydwelling elderly deteriorates with age.1,2 Both for clinical and theoretical purposes, understanding the pathways to a poor prognosis are of major importance. Identifying prognostic factors associated with
an unfavorable course may result in both a better understanding of the natural history and in more effective preventive interventions. In search of prognostic factors, much emphasis has been laid on the association of physical health and cognitive functioning with the course of late-life depression. Physical
Received October 31, 2005; revised April 7, 2006; accepted April 12, 2006. From the Departments of Clinical Psychology (BS), Psychiatry (ATFB), and Clinical Psychology (AJFMK), Vrije Universiteit, Amsterdam; the Research Institute Psychology and Health (BS, AJFMK), Utrecht; and the Institute for Research in Extramural Medicine (EMGO) (BS, ATFB, DJHD, MAB, AJFMK) and the Department of Psychiatry (DJHD, MAB), Vrije Universiteit Medical Centre. Send correspondence and reprint requests to Dr. Bas Steunenberg, Department of Clinical Psychology, Vrije Universiteit, Amsterdam, Van der Boechorststraat 1, 1081 BT Amsterdam. e-mail: [email protected] © 2007 American Association for Geriatric Psychiatry
234
Am J Geriatr Psychiatry 15:3, March 2007
Steunenberg et al. health problems such as chronic limiting diseases, functional limitations, and pain or cognitive impairment, which are frequently encountered in later life, are often shown to be closely related to the incidence and to a poor prognosis of depression.3–5 Physical health problems have even been said to overwhelm the impact of other prognostic factors.4,6 Other factors associated with a poor prognosis include older age, female gender, and lack of social support. Social demographic factors such as marital status and level of education have not been found to be related with outcome.1,6 – 8 There is some debate about the association of personality with the prognosis of depression in late life. Studies have documented that personality characteristics such as poor mastery, low self-efficacy, and neuroticism are risk factors for the onset and severity of depression in later life,9 –11 but these studies did not examine the association between personality and the prognosis of depression in late life. Mastery is defined as the extent to which a person feels that he or she has control over his or her life and environment.12 Bandura13 refers to self-efficacy as the conviction that one can successfully effectuate the behavior required to produce a given outcome in a specific situation. People with a strong sense of mastery (internal locus of control) and those with positive expectations of self-efficacy have been found to be less depressed than people with a low sense of mastery (i.e., external locus of control) or a low level of self-efficacy.14 Neuroticism refers to a tendency to experience negative emotions as well as a tendency toward emotional instability. Individuals who score high on neuroticism are more likely to experience a variety of problems, including negative moods (such as anxiety, fear, depression, and irritability) and physical symptoms. Personality may not only directly influence the level of depressive symptoms, but it may also modify the adverse effects of age-related stressful life situations. A higher level of mastery seems to facilitate adaptation under stressful events such as medical events and functional decline.9 –11,15,16 A study by Mendes de Leon et al.17 revealed that a high level of self-efficacy buffers the negative impact of functional decline on depressive symptoms. Contrary to onset and severity, little systematic research has been done into the association of personality characteristics with the prognosis of depression in late life.1,18 The aim of the present study was
Am J Geriatr Psychiatry 15:3, March 2007
to examine the association of mastery, self-efficacy, and neuroticism with the prognosis of depression in late life in the community. To this end, using a nineyear prospective longitudinal design, those with a recovery of symptoms were compared with those who did not recover, i.e., had a poor prognosis. We investigated the relative prognostic strength of personality characteristics compared with physical health-related prognostic and social resource factors. Next, we investigated whether personality characteristics modify the negative impact of other prognostic factors on depression. To our knowledge, we are the first to examine these relationships in a community sample of depressed elderly in a longitudinal study spanning a period of nine years.
METHODS Sampling and Procedures Data for this study were collected in the Longitudinal Aging Study Amsterdam (LASA), an ongoing interdisciplinary longitudinal study on predictors and consequences of changes in autonomy and wellbeing in the aging population.19 Sampling procedures and response were described in detail elsewhere.20,21 In short, a random sample of older (55– 85 years) men and women was drawn from the population registers in 11 municipalities in three geographic regions of The Netherlands. The sample was stratified by age and sex according to expected fiveyear mortality to ensure sufficient sample size for longitudinal analyses within age and sex strata. The LASA study started in 1992/1993 and includes follow-up measurements after three, six, and nine years. In the first cycle of LASA, 3,107 respondents were included. Respondents were visited at home by trained interviewers. Interviews and tests took approximately 1 hour 30 minutes. To spread the burden for the respondent, the baseline cycle consisted of two interviews and a self-administered questionnaire, including the personality variables. Loss of respondents by attrition or nonresponse was described in detail elsewhere,2,22 but is summarized subsequently for our study sample. For this study, data from all four available measurements were used spanning a period of nine years.
235
Mastery and Neuroticism Predict Recovery of Depression All 3,107 respondents to the baseline interview were screened for depression using the Center for Epidemiologic Studies Depression Scale (CES-D).23 To investigate the influence of personality on the chance of recovery of depressive symptoms, only respondents with depressive symptoms at baseline (N⫽448, CES-D ⱖ16; 14.9% relative to LASA baseline) and complete data at baseline and first follow up were included in this study (67 respondents were lost). Missing data on the personality questionnaires were imputed using the mean score only if respondents had answered at least 80% of the items on these questionnaires (175 respondents were lost). These restrictions resulted in a study sample of 206 depressed respondents. In multivariate analyses (logistic regression), the baseline characteristics of our study sample were compared with the subjects who were lost as a result of either loss to follow up between baseline and second measurement or item nonresponse. A higher age ( ⫽0.048; standard error [SE]: 0.02, p ⫽ 0.05), female gender ( ⫽⫺0.09; SE: 0.03, p⫽0.03), one or more chronic diseases ( ⫽0.60; SE: 0.28, p⫽0.03), or functional limitations (: 0.43; SE: 0.22, p⫽0.05) were found to be significantly associated with attrition. No difference between our study sample and those lost was found for level of depressive symptoms, urbanization or education level, pain, cognitive functioning, network size, level of emotional and instrumental support received, and loneliness. From the study sample, 51 respondents were lost between the second and third measurement. In the last three years between third and fourth measurements, another 32 respondents were lost. Reasons were: dead before approach (46 [55%]), no longer interested/motivated (11 [13%]), physically (2 [2%]) or cognitively (3 [4%]) not able to participate or reason not specified (3 [4%]), and 18 (22%) respondents had too many missing values on the key variables under study. Loss to follow up was found to be related to higher age ( ⫽0.07; SE: 0.02, p⫽0.001). Respondents who recovered, but were lost later after the second measurement, were included in the analyses. Measurements Course of Depression. Depressive symptoms were measured using the Dutch translation24 of the CES-D.25 This is a 20-item self-report scale developed to measure depressive symptoms in the community. Re-
236
spondents were asked how often they had experienced each symptom during the previous week. Items were scored on a four-point scale ranging from 0 (rarely or none of the time) to 3 (most of or all of the time). The total CES-D score ranges from 0 (no symptoms) to 60 (maximum number of symptoms). A score of ⱖ16 has generally been used as indicative for clinically relevant depressive syndromes.25 The CES-D has been widely used in older community samples, and the psychometric properties were found to be satisfactory.26,27 Because of the emphasis on affective items in the scale, the overlap with symptoms of physical illness is minimal.25,28 To study predictors related to the prognosis of depression, two course types were defined applying a definition of a clinically relevant change in depressive symptoms. A relevant change was defined as a decrease or an increase of at least five points on the CES-D between two measurements, thereby crossing the cutoff score of 16. The difference of five points is in line with the definition of a reliable change,29 which takes into account the reliability of the CES-D and with the principle of a medium effect size.30 A similar definition of a relevant change was used in earlier studies.6,31,32 The following two course types were distinguished: recovery (a relevant decline of symptoms between two adjacent measurements and the subject remaining nondepressed (CES-D score of ⬍16) throughout the rest of the study) and no recovery. The course type no recovery includes intermittent (after recovery, the subject had a relevant increase of symptoms later on in the study) and persistent (CES-D score ⱖ16 at all four measurements32) patterns of change of symptoms. The following variables measured at baseline were examined as potential predictors of the prognosis of depression during the nine-year study period. Personality Characteristics. Mastery was measured by means of a translated and abbreviated Dutch version of the Pearlin Mastery Scale.12 The questionnaire consists of five statements such as “I cannot seem to be able to solve some of my problems at all.” Scale scores range between 5 and 25. Higher scores on the scale indicate a high mastery. General self-efficacy was measured by means of the Perceived Self-Efficacy Scale, which is based on 17 items of the General Self-Efficacy Scale (GSES33). Based on the results of the pilot study of LASA, the
Am J Geriatr Psychiatry 15:3, March 2007
Steunenberg et al. GSES was shortened to a version with 12 items. An example-item is: “If I make plans, I am convinced I will succeed in executing them.” A minimum score of 12 indicates the most negative perceived self-efficacy score and 60 the most positive. Neuroticism was operationally defined by using a subset of 25 items, out of a list of 36 neuroticism items, from the Dutch Personality Questionnaire (DPQ34). The pilot studies of the LASA demonstrated that the original 36-item scale measuring neuroticism could be abbreviated.35,36 Respondents were asked to indicate whether various similar statements applied to them (yes, do not know, no). The scores can range between 0 and 50. An example of a neuroticism item is: “I often hate myself.” These DPQ items have strong negative relations with the Emotional Stability Scale of the NEO-PI-R.37 For reasons of clarity of presentations, predictors can now be directly compared; we have chosen to dichotomize the scales at the 50th percentile of the mean score. In analyses with the complete scales, we found similar results like with the dichotomized scales. Sociodemographic Variables. Sociodemographic variables included were age, sex, level of education, marital status, and urbanization level of the municipality. Age ranged from 55– 85 years at baseline. Education was dichotomized in: low education (uncompleted or completed primary school) and middle to high education (secondary education up to and including university). Marital status was dichotomized in married and not or no longer being married. Urbanization level of municipality was dichotomized in less urban and highly urban. Health Variables. Functional limitations in daily life was assessed by a three-item questionnaire,38 and for the present study, the number of functional limitations was dichotomized into 0⫽no limitations and 1⫽one or more limitations. The number of chronic diseases was calculated by summing all diseases reported by the respondent to be present.39 In a validation study, respondents’ selfreports were compared with information obtained from their general practitioners’ and proved to be sufficiently reliable.40 For the present study, the number of chronic diseases was dichotomized into 0⫽no disease and 1⫽one or more diseases. Pain was measured with a subscale of the Nottingham Health Profile.41 The following items were included: “I am in pain when I am standing,” “I find it
Am J Geriatr Psychiatry 15:3, March 2007
painful to change position,” “I am in pain when I am sitting,” “I am in pain when I walk,” and “I am in constant pain.” Response categories were “no” (0) and “yes.”1 Total scores range from 0 (low) to 5 (high). This scale was dichotomized at 0 ⫽ no pain and 1⫽ pain present. Cognitive functioning was measured by means of the Mini-Mental State Examination.42 On 23 questions and tasks, respondents received one or more points when they gave the correct answer or performed the task correctly. Scores ran from 0 to 30 using the commonly used cutoff score of ⱕ23 as an indicator for cognitive impairment. This score has been shown to be 80%–90% sensitive and 80%–100% specific for a diagnosis of dementia.43– 45 Social Resources. Measures of social support were constructed from a questionnaire, which included detailed questions on both the number and the quality of contacts with members of the social network.46 In this article, we report on the size of the contact network and the exchange of received emotional and instrumental support with network members. Loneliness was assessed with a scale developed by Jong-Gierveld and de Kamphuis.47 The scale has been used in several surveys and has been proven to be a robust and valid instrument.48 The scale consists of 11 items and the total scale score ranges between 1 and 11. All four predictors were dichotomized at the 50th percentile of the mean score. Statistical Analyses The direct effect of personality characteristics and the interaction with other predictors on the probability of recovery within the nine-year follow-up period was investigated in a Cox proportional hazards design.49,50 Cox regression model estimates the effect of predictor variables on the hazard rate, in this case being the time to recovery during follow up. P values lower than 0.05 were regarded as statistically significant. Hazard ratios are a measure of the strength of the association of the predictor with the time to recovery. A hazard ratio of 1 indicates no effect of the predictor on the time to recovery. The time from baseline measurement to recovery (event) or to the end of the study without recovery (censored data) was calculated per study cycle with intervals of three years. Respondents who recovered but were lost
237
Mastery and Neuroticism Predict Recovery of Depression later (i.e., deceased during the study) were included in the analyses as an event. First, we examined the chance of recovery for each predictor separately. Then, in a multivariate model, we included all predictors that were associated with recovery in the univariate analyses with a p value ⬍0.05. Finally, to find out whether personality acts as a buffer, mitigating the negative impact of potential predictors on the outcome of depression, interaction effects between the personality characteristics and potential predictors at baseline were investigated. We only investigated interaction effects between personality characteristics and potential predictors that were significantly associated with recovery in the univariate analyses with a p value ⬍0.05. When an interaction effect showed a p value ⬍0.10, stratified analyses were carried out. Stratified analyses explore the nature of the interaction effect, i.e., they estimate the strength of the association between predictor and depression for the different subgroups on the personality characteristics. Statistical analyses were performed using SPSS 12.0.
RESULTS Baseline demographic and clinical characteristics of the study sample are presented in Table 1. The mean age of the 206 respondents was 69.6 years at baseline. Sixty-eight percent were female. During the nineyear follow up, 91 (44%) respondents remitted. Most respondents (69 [33%]) remitted between baseline and first follow up three years later. Fourteen (7%) respondents remitted between first and second follow up and eight (4%) respondents remitted over the last three years. Prognostic Factors for Recovery of Late-life Depression Table 2 presents the results of the univariate Cox proportional hazard regression analyses. Recovery was predicted by high mastery (odds ratio [OR]: 1.7, p⫽0.01) and a low level of neuroticism (OR: 1.6, p⫽ 0.03) at baseline. Male respondents depressive at baseline had a higher chance on recovery during the
238
TABLE 1.
Demographic and Clinical Characteristics of the Baseline Depressed Study Sample (N ⴝ 206)
Characteristic Age ⬍70 years ⱖ70 years Sex Male Female Education level Low Middle/high Marital status Married Not or no longer married Municipality Less urban Highly urban Functional limitations None One or more Number of chronic diseases None One or more Pain No pain Low to severe pain Cognitive Functioning Normal Impaired
N
Percent
111 95
54 46
66 140
32 68
136 70
66 34
104 102
50 50
135 71
66 34
92 114
45 55
61 145
30 70
89 117
43 57
186 20
90 10
follow-up period (OR: 1.7, p⫽ 0.01). Absence of functional limitations (OR: 2.1, p ⫽0.001) and pain (OR: 1.8, p⫽0.01) at baseline also were positively associated with recovery. For no pain, a hazard ratio of 1.8 was found, meaning that the probability of recovery was 80% greater for respondents with lack of pain at baseline compared with respondents with pain. Controlling for confounding among correlated predictors, we included all predictors that were significantly associated with recovery in the univariate analyses in a multivariate model. Two of the predictors with a unique effect remained significant in the multivariate model (Table 2). Both low level of neuroticism (OR: 1.5, p⫽0.05) as well as no functional limitations (OR: 1.7, p⫽0.03) were found to be significantly related to recovery. Mastery no longer had a unique predictive effect, which holds for gender and pain as well. Third, we investigated whether the personality characteristics moderate the association between the other prognostic factors and the chance of recovery by entering interaction terms to the regression model. We investigated whether the personality characteristics mastery and neuroticism moderate
Am J Geriatr Psychiatry 15:3, March 2007
Steunenberg et al.
TABLE 2.
Predictors of Remission of Depression in Older Adults: Results of Cox Proportional Hazard Regression Analysis (univariate and multivariate model, N ⴝ 206)
Predictors Multivariate Modela HR (95 % CI)
Univariate Model HR (95 % CI) Low level of depressive symptoms at baseline Personality characteristics High level of mastery High level of self-efficacy Low neuroticism Social demographics Age ⱕ70 years Male gender Middle/high level of education Married Less urban Physical health-related variables No functional limitations No chronic diseases No pain Normal cognitive functioning Social resources Large network size High level of instrumental support received High level of emotional support received No loneliness
1.3 (1.0–1.6) 1.7b (1.1–2.7) 1.5 (1.0–2.2) 1.6c (1.0–2.5)
1.4 (0.9–2.3) 1.5b (1.0–2.4)
1.5 (1.0–2.3) 1.7b (1.1–2.6) 1.5 (1.0–2.3) 1.3 (1.3–1.9) 1.2 (0.8–1.9)
1.4 (0.9–2.1)
2.1c (1.4–3.3) 1.2 (0.8–1.9) 1.8c (1.2–2.7) 1.3 (1.0–1.7)
1.7b (1.1–2.8) 1.3 (0.8–2.1) 1.1 (0.8–1.4)
1.3 (0.9–2.0) 1.3 (0.8–1.9) 1.1 (0.7–1.6) 1.4 (0.9–2.1)
In the multivariate model, only predictors that were associated with remission in the univariate analyses with a p value ⬍0.05 were included. p ⬍ 0.05. c p ⬍ 0.01. HR: hazard ratio; CI: confidence interval. a
b
the effects of deteriorations in the physical healthrelated predictors functional limitations and pain, which were found to be significantly related to the chance on recovery at univariate level. Results showed that none of the interaction terms were significant (Table 3), so the personality characteristics in this study do not moderate the influence of the physical health-related predictors on the chance on recovery. Finally, we investigated whether the association between the personality characteristics mastery and
TABLE 3.
neuroticism differed for male and female respondents. Results revealed there was no gender effect on the association between the personality characteristics and the course of depression (Table 3).
DISCUSSION We aimed to study whether the personality characteristics mastery, self-efficacy, and neuroticism are
Results of Two-Way Interaction Analyses Between Personality Characteristics and Predictors of Remission of LateLife Depression (N ⴝ 206) Personality Characteristics Mastery
Predictors
Estimated Coefficient (SE) ⫺0.05 (0.45) 0.56 (0.46) ⫺0.07 (0.46)
Physical Health-Related Variables Number of functional limitations Pain Gender
Neuroticism Wald 2
Estimated Coefficient (SE)
Wald 2
0.01 (p ⫽ 0.9) 1.5 (p ⫽ 0.2) 0.02 (p ⫽ 0.9)
⫺0.02 (0.45) 0.13 (0.44) 0.24 (0.44)
0.001 (p ⫽ 1.0) 0.09 (p ⫽ 0.9) 0.28 (p ⫽ 0.8)
SE: standard error.
Am J Geriatr Psychiatry 15:3, March 2007
239
Mastery and Neuroticism Predict Recovery of Depression associated with the nine-year prognosis of depression in later life and whether they modify the negative impact of other prognostic factors on the prognosis of depression. In the regression model for univariate effects, high mastery and low neuroticism were identified as significant predictors of recovery. We expected to find the negative impact of physical ill health on the chance of recovery to overwhelm the effect of other factors, especially personality. Our results did not lend support to this hypothesis. The prognostic strength of the physical health-related variables was similar to the strength of the personality characteristics. In a multivariate model correcting for the confounding effects of other predictors, low neuroticism and no functional limitations were found to be a unique significant prognostic factors for recovery. The relative prognostic strength of low neuroticism and no functional limitations was hardly affected by adjusting for the effects of the other significant predictors. Demonstrating that personality characteristics, like neuroticism, are stable in the context of acute change is crucial to models proposing that personality can influence the course of depression. However, although findings show that absolute changes in personality may be observed, effect sizes have been extremely small and have been taken as evidence for stability rather than as evidence for change.51,52 Both the relative stability of personality and its absolute change could not be accounted for by individual differences or absolute changes in depressive symptoms.53 In conclusion, we assumed the personality characteristics measured at baseline to be stable over the nine-year follow-up period. It is not precisely clear how a global personality dimension as neuroticism influences the chance of recovery. Scott et al.54 suggested that neuroticism primarily acts as an “amplifier” moderating the impact of dysfunctional cognitions on mood (and thereby, amplifying the entire vicious spiral of cognition and affect). According to Scott et al.,54 the ruminative response style as described by NolenHoeksema55 is the most evident way in which the amplification takes place. Nolen-Hoeksema55 hypothesized that rumination influences depressive mood by interfering with attention and concentration, by enhancing the recall of negative events, and by increasing the likelihood of using depressogenic
240
explanations for negative life events. Future research may subscribe this notion and further determine the complex relationship among neuroticism, cognitions, rumination, and the increased chance of recovery. Self-efficacy did not predict recovery of late-life depression, but higher levels of mastery did predict recovery. Self-efficacy and mastery are somewhat related concepts, because they both refer to a feeling of competence. Mastery, however, pertains to a general feeling of being in control and having power to use personal resources to influence the outcomes of life, whereas self-efficacy concerns the perceived ability of doing what needs to be done to achieve a desired goal in a specific situation. It seems plausible that in late-life self-efficacy, compared with mastery, is more strongly related to physical health. The strength of the association between physical health and depression increases with age, which possibly can explain why self-efficacy was not found to be significantly related to depression. In contrast to these direct effects, our results did not lend support for the hypothesis that personality characteristics moderate the association between the other prognostic factors and the chance on recovery. The present results favor a model of main effects adding onto each other rather than an interactive model with multiplicative interaction of variables, in which vulnerability factors such as personality characteristics and late-life stress such as deteriorations in physical health amplify each other. Our results constitute evidence for the validity of the construct of neuroticism, specifically the component of distress proneness; those high on neuroticism tend to experience more distress across time and regardless of the physical health, cognitive functioning, or social situational resources. Personality disorders are found in substantial numbers in the elderly population and have clinical implications for geriatric depression.56 Personality disorders represent extreme forms of personality characteristics. The highest rates of personality disorder were found among depressed elderly,56 and significant associations between personality disorders and chronicity of depression have been reported.57,58 Lifetime personality dysfunction may affect the course of depression by means of the persistence of somatic worry, distortion in perceived health status, prolonged bereavement, and noncompliance with medical and psychiatric treatment.56 These re-
Am J Geriatr Psychiatry 15:3, March 2007
Steunenberg et al. sults on personality disorders as well as personality characteristics underscore the need to pay attention to the impact of personality psychopathology on the prognosis of late-life depression. The strong points of this study are its long-term prospective design using a community sample of depressed older adults and including a wide range of potential prognostic factors. This study also had some methodological limitations. First, there was considerable attrition at all stages of the study. Analyses of loss of subjects revealed that the frailest elderly were at higher risk for attrition. The results of our study suggest the most frail elderly persons (those with a high level of pain and one or more functional limitations) also had the greatest chance of a chronic course in our nine-year follow-up period. Therefore, although loss of data limits the generalizability of the findings, the results of the analysis pertaining to attrition suggest that, if anything, the true prognosis of late-life depression is underestimated. A second limitation is that there was a measurement only once every three years. Therefore, it is possible that subjects have been depressed between measurements and recovered before the next measurement. The effect of this is that
change is underestimated. However, the purpose of this article was not to distinguish between persistent and chronic intermittent patterns of change of symptoms, but to identify the association of personality characteristics and other potential predictors with the long-term prognosis of depression to distinguish between elderly who recovered and remain nondepressed throughout the rest of the study and elderly with a recurrence later on or a chronic course. A third concern is that no adjustments were made for treatment of depression during our study period. However, in the LASA sample, the level of treatment was very low which is in accordance with earlier studies6,59,60 As a consequence, the results of this study most likely reflect the natural course of depression. In conclusion, the personality characteristics neuroticism and mastery have a strong and independent effect on the prognosis of depression in later life. In our view, along with the widely acknowledged importance of physical health-related variables, attention should be paid to the impact of personality characteristics on the prognosis of depression in later life. However, in the LASA sample, the level of treatment was very low, which is in accordance with earlier studies.
References 1. Cole MG, Bellevance F, Mansour A: Prognosis of depression in elderly community and primary care populations: a systematic review and meta-analysis. Am J Psychiatry 1999; 156:1182–1189 2. Beekman ATF, Geerlings SW, Deeg DJH, et al: The natural history of late-life depression: a 6-year prospective study in the community. Arch Gen Psychiatry 2002; 59:605–611 3. Henderson AS, Korten AE, Jacomb PA, et al: The course of depression in the elderly: a longitudinal community-based study in Australia. Psychol Med 1997; 27:119 –129 4. Prince MJ, Harwood RH, Thomas A, et al: A prospective population-based cohort study of the effects of disablement and social milieu on the onset and the maintenance of late life depression. The Gospel Oak Project VII. Psychol Med 1998; 28:337–350 5. Snowdon J, Lane F: The botany survey: a longitudinal study of depression and cognitive impairment in an elderly population. Int J Geriatr Psychiatry 1995; 10:349 –358 6. Kennedy GJ, Kelman HR, Thomas C: Persistence and remission of depressive symptoms in late life. Am J Psychiatry 1991; 148:174 – 178 7. Kivela S-L, Pahkala K, Laippala P: A one-year prognosis of dysthymic disorder and major depression in old age. Int J Geriatr Psychiatry 1991; 6:81–87 8. Kivela S-L: Long-term prognosis of major depression in old age: a comparison with the prognosis of dysthymic disorder. Int Psychiatry 1995; 7(suppl): 69 –82 9. Jang Y, Haley WE, Small BJ, et al: The role of mastery and social
Am J Geriatr Psychiatry 15:3, March 2007
resources in the associations between disability and depression in later life. Gerontologist 2002; 6:807–813 10. Roberts BL, Dunkle R, Haug M: Physical, psychological, and social resources as moderators of the relationship of stress to mental health in the very old. J Gerontol Soc 1994; 49:535–543 11. Schieman S, Turner HA: Age, disability and the sense of mastery. J Health Soc Behav 1998; 39:169 –187 12. Pearlin LI, Schooler C: The structure of coping. J Health Soc Behav 1978; 19:2–21 13. Bandura A: Self-efficacy: toward a unifying theory of behavioural change. Psychol Rev 1977; 84:191–215 14. Flammer A: Developmental analysis of control beliefs, in SelfEfficacy in Changing Societies. Edited by Bandura A. Cambridge, UK, Cambridge University Press, 1995, pp 69 –113 15. Kempen GIJM, Jelicic M, Ormel J: Personality, chronic medical morbidity, and health elated quality of life among older persons. Health Psychol 1997; 16:539 –549 16. Kempen GIJM, van Heuvelen MJG, van Sonderen E, et al: The relationship of functional limitations to disability and the moderating effects of psychological attributes in community-dwelling older adults. Soc Sci 1999; 56B:S229 –236 17. Mendes de Leon CF, Seeman TE, Baker DI, et al: Self-efficacy, physical decline, and change in functioning in community-living elderly: a prospective study. J Gerontol B Psychol Sci Soc Sci 1996; 51B:S183–190 18. Blazer D: Depression in the elderly. N Engl J Med 1989; 320:164 – 166
241
Mastery and Neuroticism Predict Recovery of Depression 19. Deeg DJH, Knipscheer CPM, van Tilburg W: Autonomy and WellBeing in the Aging Population: Concepts and Design of the Longitudinal Aging Study Amsterdam. NIG-Trend-Studies Number 7. Bunnik, The Netherlands, Netherlands Institute Gerontology, 1993 20. Penninx BWJH, van Tilburg TG, Kriegsman DMW, et al: Effects of social support and personal coping resources on mortality in older age: the Longitudinal Aging Study Amsterdam. Am J Epidemiol 1997; 146:510 –519 21. Beekman ATF, Deeg DJH, van Tilburg T, et al: Major and minor depression in later life: a study of prevalence and risk factors. J Affect Disord 1995; 36:65–75 22. Deeg DJH, van Tilburg TG, Smit JH, et al: Attrition in the Longitudinal Aging Study Amsterdam: the effect of differential inclusion in side studies. J Clin Epidemiol 2002; 55:319 –328 23. Radloff LS: The CES-D scale: a self-report depression scale for research in the general population. Appl Psychol Meas 1977; 1:385–401 24. Beekman ATF, van Limbeek J, Deeg DJH, et al: Screening for depression in the elderly in the community: using the Center for Epidemiologic Studies Depression Scale (CES-D) in The Netherlands. Tijdschr Gerontol Geriatr 1994; 25:95–103 25. Berkman LF, Berkman CS, Kasl SV, et al: Depressive symptoms in relation to physical health and functioning in the elderly. Am J Epidemiol 1986; 124:372–388 26. Himmelfarb S, Murrel SA: Reliability and validity of five mental health scales in older persons. J Gerontol 1983; 38:333–339 27. Radloff LS, Teri L: Use of the CES-D with older adults. Clin Gerontol 1986; 5:119 –136 28. Foelker GA, Shewchuk RM: Somatic complaints and the CES-D. J Am Geriatr Soc 1992; 40:259 –262 29. Jacobsen NS, Truax P: Clinical significance: a statistical approach to defining meaningful change in psychotherapy research. J Consult Clin Psychol 1991; 59:12–19 30. Cohen J: Statistical Power Analysis for the Behavioral Sciences. Hillsdale, NJ, Lawrence A. Erlbaum Associates, 1985 31. Beekman ATF, Deeg DJH, Smit JH, et al: Predicting the course of depression in the elderly: results from a community-based study in The Netherlands. J Affect Disord 1995; 34:41–49 32. Geerlings SW, Beekman ATF, Deeg DJH, et al: Physical health and the onset and persistence of depression in older adults: an eightwave prospective community based study. Psychol Med 2000; 30:369 –380 33. Sherer M, Maddux JE, Mercandante B, et al: The self-efficacy scale: construction and validation. Psychol Rep 1982; 51:663–671 34. Luteijn F, Starren J, van Dijk H: Handleiding bij de NPV. Lisse, Swets & Zeitlinger, 1975 35. Smits CHM, Deeg DJH, Bosscher R: Well-being and control in older persons: the prediction of well-being from control measures. Int J Aging Hum Dev 1995; 40:237–250 36. Steunenberg B, Beekman ATF, Deeg DJH, et al: Neuroticism in the elderly: the utility of the shortened DPQ-scales [in Dutch]. Tijdschrift voor Gerontologie en Geriatrie 2003; 34:118 –124 37. Luteijn F, Starren J, van Dijk H: Manual of the Dutch Personality Inventory, rev ed. Lisse, Swets en Zeitlinger, 2000 38. Kriegsman DMW, van Eijk JT, Penninx BW: Does family support buffer the impact of specific chronic diseases on mobility in community-dwelling elderly? Disabil Rehabil 1997; 19:71–83 39. Health Interview Questionnaire. Heerlen, Central Bureau of Statistics, 1989
242
40. Kriegsman DMW, Penninx BW, Eijk JT, et al: Self-reports and general practitioner information on the presence of chronic diseases in community dwelling elderly. A study on the accuracy of patients’ self-reports and on determinants of inaccuracy. J Clin Epidemiol 1996; 49:1407–1417 41. Hunt SM, McEwen J, McKenna SP: Measuring health status: a new tool for clinicians and epidemiologist. J R Coll Gen Pract 1985; 35:185–188 42. Folstein MF, Folstein SE, McHugh PR: Mini-Mental State: a practical method for the clinician. J Psychiatr Res 1975; 12:189 –198 43. Anthony JC, LeResche L, Niaz U: Limits of the ‘Mini-Mental State’ as a screening test for dementia and delirium among hospital patient. Psychol Med 1982; 12:397–408 44. Tomhaugh TN, McIntyre NJ: The Mini-Mental State Exam: a comprehensive review. J Am Geriatr Soc 1992; 40:922–935 45. Langley LK: Cognitive assessment of older adults, in Assessing Older Persons. Edited by Kane RI. Oxford, UK, Oxford University Press, 2000, pp 65–128 46. van Tilburg TG: Losing and gaining in old age: changes in personal network size and social support in a four-year longitudinal study. J Gerontol 1998; 53B:S313–S323 47. Jong-Gierveld J, de Kamphuis FH: The development of a Raschtype loneliness scale. Appl Psychol Meas 1985; 9:289 –299 48. van Tilburg TG, de Leeuw ED: Stability of scale quality under various data collection procedures: a mode comparison of the de Jong-Gierveld Loneliness Scale. International Journal of Public Opinion Research 1991; 3:69 –85 49. Cox DR: Regression models and life-tables. J R Stat Soc 1972; 34:187–220 50. Cox DR, Oakes D: Analysis of Survival Data. London, Chapman & Hall, 1984 51. Costa Jr, PT McCrae RR, Zonderman AB, et al: Cross-sectional studies of personality in a national sample: 2. Stability in neuroticism, extraversion and openness. Psychol Bull 1986; 94:472–505 52. Steunenberg B, Twisk JWR, Beekman ATF, et al: Stability and change of neuroticism in aging. J Gerontol B Psychol Sci Soc Sci 2005; 60:P27–33 53. Santor DA, Bagby RM, Joffe RT: Evaluating stability and change in personality and depression. J Pers Soc Psychol 1997; 73:1354 – 1362 54. Scott J, Williams MJ, Brittlebank A, et al: The relationship between premorbid neuroticism, cognitive dysfunction and persistence of depression: a one-year follow-up. J Affect Disord 1995; 33:167–172 55. Nolen-Hoeksema S: Responses to depression and their effect on the duration of depressive episodes. J Abnorm Psychol 1991; 100:569 –582 56. Abrams RC: Personality disorders in the elderly. Int J Geriatr Psychiatry 1996; 11:759 –763 57. Devanand DP, Nobler MS, Singer T, et al: Is dysthymia a different disorder in the elderly? Am J Geriatr Psychiatry 1994; 151:1592– 1599 58. Thompson LW, Gallagher D, Czirr R: Personality disorder and outcome in the treatment of late-life depression. J Geriatr Psychiatry 1988; 21:133–146 59. Eisenberg L: Treating depression and anxiety in primary care. Closing the gap between knowledge and practice. N Engl J Med 1992; 326:1080 –1083 60. Freeling P: Diagnosis and treatment of depression in general practice. Br J Psychiatry 1993; 163(suppl): 14 –19
Am J Geriatr Psychiatry 15:3, March 2007