| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Department of Psychiatry (N.F.-S., F.L.) and School of Nursing (N.F.-S.), McGill University; the Research Center, Montreal Heart Institute (N.F.-S., F.L., G.G., A.M., M. J., M.G.B.); the Departments of Psychiatry (N.F.-S., F.L.) and Medicine (M. J., M.G.B.), University of Montreal; and the Research Center, Centre Hospitalier de l’Université de Montréal (N.F.-S., F.L.), Montreal, Canada.
Address reprint requests to: Nancy Frasure-Smith, PhD, Montreal Heart Institute Research Center, 5000 Bélanger St. E., Montreal, QC, Canada, H1T 1C8. Email: frsm{at}icm.umontreal.ca
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
METHODS: Three subgroups were defined using median splits on the State Anxiety Inventory and Marlowe-Crowne Scale administered at baseline: truly low anxious, repressors, and high anxious. Quebec medicare data were used to track survival through 5 years.
RESULTS: The trend toward worse prognosis in women in the treatment group and no evidence of treatment impact in men that were seen during the program year were maintained during the follow-up. Analysis of results in terms of coping styles showed a significant long-term survival benefit of treatment in highly anxious men, for whom reductions in somatic symptoms of depression mediated program impact. However, the program was also associated with significantly worse survival in repressors of both sexes. By the end of the program, repressors in the treatment group were more likely to be prescribed benzodiazepines and to have visited emergency rooms without being readmitted than those in the control group, suggesting that the program may have increased distress in repressors.
CONCLUSIONS: Patients’ coping style is important in determining outcomes of psychosocial treatments and should be taken into account when tailoring interventions.
Key Words: myocardial infarction • psychosocial intervention • mortality • repressor • anxiety
Abbreviations: ACE = angiotensin converting enzyme; BDI = Beck Depression Inventory; GHQ = General Health Questionnaire; MC = Marlowe-Crowne Social Desirability Scale; M-HART = Montreal Heart Attack readjustment Trial; MI = myocardial infarction; PSSS = Perceived Social Support Scale; S-STAI = Spielberger State Anxiety Inventory
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
5) or who were rehospitalized received home-based nursing interventions. The nurses used a multifactorial, case management-type approach to help patients deal with their current sources of distress. Although a preliminary study showed evidence of positive program outcomes in men, no women were included, and methodological difficulties prevented drawing firm conclusions (4, 5). We conducted M-HART to test the hypothesis that program participation would reduce cardiac mortality for both women and men during the first post-MI year. One-year results were disappointing (1). Although more than three-quarters of patients in the intervention group received interventions consisting of an average of five to six 1-hour-long nursing visits, there was no overall impact on either survival or psychological outcomes. Furthermore, preplanned analyses by sex showed evidence of treatment-related increases in cardiac (p = .064) and all-cause mortality (p = .051) among women, with no evidence of any positive or negative impact for men. Considering the reasons for these outcomes, we speculated that the monthly telephone monitoring calls coupled with systematic intervention for even minor levels of psychological distress might have interfered with patients’ normal coping processes and, paradoxically, interfered with the trajectory of adjustment in some patients.
We have now used Quebec medicare data to follow M-HART participants’ prognosis through 5 years after hospital discharge. This allowed us to explore two issues: 1) whether after the end of the program, the sex-related patterns of program impact on prognosis continued in the same directions and 2) whether long-term outcomes varied in relation to patients’ coping style at the time of the baseline interview (low anxiety, repression, or high anxiety) (6–8).
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
The primary outcome for the long-term follow-up was cardiac deaths that occurred between randomization (hospital discharge) and 5 years after discharge. Patients with any physician contacts recorded in Quebec medicare data after 1825 days postdischarge were categorized as living at the end of follow-up. For other patients, additional information was obtained from hospital charts and death certificates. Survival status at 5 years was ascertained for all but 3.1% of patients (N = 42). A cardiologist and a trained research assistant, blind to study group and interview data, each independently coded the causes of death as cardiac and noncardiac. Disagreements were resolved by discussion with a second physician who also classified the data.
Statistical Analyses
Data analysis was carried out using SPSS for Windows (version 10.07) (15). All tests were two-tailed, with p values 
.05 considered statistically significant and p values >.05 but <.10 representing potentially meaningful trends. Visual inspection of log cumulative hazards plots for survival times over 5 years supported the proportional hazards assumption, and we used Cox proportional hazards regression analysis to assess the long-term prognostic impact of treatment group and other baseline measures both before and after adjustment for covariates.
To examine the potential role of patients’ baseline coping tendencies, we used median splits on the baseline S-STAI (35.0) and MC (>9.0) scales to define three groups: subjects with scores below the median on both measures (truly low anxious), subjects above the median on the MC and below the median on the S-STAI (repressors), and subjects above the median on the S-STAI regardless of their MC scores (highly anxious). We used the approach of Shaw et al.(8) and did not divide the high-anxious patients into defensive and nondefensive groups because of previous research showing little difference between them. The survival times for the three coping groups were compared using Cox proportional hazards regression including assessment of the two-way and three-way interaction effects of the repression/anxiety subgroups, sex, and treatment group.
To identify imbalances in background characteristics between treatment and control group patients in the sex by treatment and repression/anxiety subgroups involved in significant interactions, we used logistic regression to compare treatment and control patients for dichotomous factors and analysis of variance for continuous factors. Baseline factors differing at p .10 were adjusted in subsequent Cox regression analyses.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
|
|
35.0) and MC scores (>9.0) to define three groups suggested by Weinberger et al. (6): those with truly low anxiety (N = 274), the repressors (N = 408), and those with high anxiety levels, who have also been referred to as sensitizers (N = 676). The S-STAI and/or MC were missing for 18 patients. Although repressors were less likely than the truly low-anxious or high-anxious patients to have any high GHQ scores during the year and to receive nursing visits (p < .001), the great majority did receive visits (70% compared with 85% of the low anxious and 89% of the high anxious). Repressors also had the shortest duration of contact with nurses over the program year (on average 198 days vs. 214 days for the low anxious and 232 for the high-anxious group). There were no sex differences in these patterns in the three anxiety/repression subgroups.
There was a significant difference in long-term cardiac mortality among the anxiety/repression subgroups (unadjusted p = .026), with patients in the truly low-anxious group having better long-term prognosis than the repressors or the highly anxious. However, this overall difference was complicated by a significant interaction involving anxiety/repression and treatment group (p = .021), as well as a significant three-way interaction of the anxiety/repression subgroups, sex, and treatment group (p = .015). To explore these results, we assessed the two-way interaction of group by sex separately within each of the anxiety/repression subgroups. The interaction was not significant for the truly low anxious (p = .11) or the repressors (p = .95), but was highly significant for the high-anxiety group (p = .002). Table 2 shows the data for long-term cardiac mortality in relation to these subgroups.
|
|
|
Our analyses of mediators were based on the criteria suggested by Baron and Kenny (16) that to be a mediator, a mechanism must be linked to both treatment group and treatment outcome and must remain linked to outcome after control for treatment group. If these three criteria are fulfilled, control for the mediator will also eliminate a large portion of the apparent impact of the treatment. Table 3 summarizes the results of these analyses for those potential mechanisms significantly linked to treatment group in the repressors and in the men with high anxiety. Data for the mechanistic variables not significantly related to treatment in these groups are not shown.
|
Our examination of potential medical care differences between the highly anxious men in the treatment and usual care groups revealed no differences in overall contacts with physicians or in contacts with cardiologists. However, a greater percentage of them were seen by psychiatrists during the year than was the case for highly anxious men in the usual care group. Nonetheless, adjustment for psychiatric contact did not remove the impact of treatment group, suggesting that psychiatric care was not a mechanism explaining group impact. In addition, despite their higher frequency of psychiatrist contacts, there were no group differences in prescription of antidepressants or benzodiazepines. There were, however, significant unadjusted treatment/usual care group differences for somatic symptoms of depression and anxiety at 1 year, and evidence of these differences remained after control for diabetes, the only imbalanced background variable for the highly anxious men. Exploration of Baron and Kenny’s criteria for mediation showed that changes in anxiety were unrelated to outcome after control for treatment group, but that changes on the somatic subscore of the BDI continued to predict the long-term survival of anxious men after control for treatment group. In fact, adjustment for changes in somatic depression symptoms removed the positive impact of treatment in highly anxious men, suggesting that treatment-related beneficial changes in somatic symptoms may have been a mechanism associated with their improved long-term prognosis.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
We also observed a striking pattern of outcomes in relation to the anxiety/repression categorization of both men and women. There was evidence of a long-term negative impact of treatment in repressors regardless of their sex. Assessment of the potential program-related mechanisms underlying the worse prognosis in treated repressors showed that although there were no differences in discharge prescriptions for benzodiazepines, there were almost twice as many treatment as control group repressors with prescriptions for benzodiazepines at the time of the 1-year interview. Treated patients were also more likely to have gone to an emergency room for something minor enough that they were not admitted to hospital. Subsequent analyses supported the role of both benzodiazepines and emergency room utilization as mediators of the long-term impact of treatment in repressors. Although there has been some speculation in the literature that anxiolytic use may be linked to increased mortality risks in some patients (17), the existing data are mostly based on older medications (primarily barbiturates). Therefore, there is no evidence that prescription of benzodiazepines per se was responsible for the difference in cardiac mortality. Instead, we believe that these results suggest that the type of person most likely to be prescribed benzodiazepines and to go to an emergency room for relatively minor problems was at greater risk in the treatment group.
Although the literature on defensiveness and related constructs like denial and repression varies in the way they are operationalized (18–20), repression is usually defined as a form of defensiveness in which individuals are not consciously aware of their heightened arousal to various anxiety-provoking stimuli, but nonetheless use defensive strategies to avoid the stimuli (6). The term repressor is also used to describe individuals with a coping style in which information that is potentially threatening is avoided. This type of coping style has also sometimes been referred to as "blunting" or "information/avoidance" (21, 22). Despite some conflicting data, several studies with undergraduate students and young adults have shown that although repressors do not directly report elevated levels of negative emotions or physical symptoms in response to stressors, indirect and physiological measures, including skin resistance, heart rate, blood pressure, cortisol, and measures of immune function, indicate high levels of arousal (20, 23). The present results fit in well with this literature. We speculate that the greater physiological emotional arousal in the repressors in the treatment group may have been sufficient to provoke enough concern in patients and family members to lead to increased emergency room visits and enough concern in physicians to have increased prescriptions of benzodiazepines, but may not have been perceived by patients as changes in their emotional state and so was not reported as such. However, it is also possible that by the time of the 1-year measurement, repressors’ anxiety levels had been reduced by benzodiazepine use.
Clearly, for there to have been no overall program impact as well as a negative impact in repressors, there must have been a positive impact of the program in some subgroup. In fact, we observed an improvement in long-term prognosis in men who had high levels of anxiety during their index hospitalization. In contrast, there was evidence of a negative prognostic impact in women with high anxiety. However, as for the women in general, adjustment for baseline imbalances between treatment and usual care women reduced the apparent negative impact of the program. For highly anxious men, adjustment for baseline differences did not alter the long-term positive impact of treatment. Furthermore, analysis of the potential mediators of this outcome suggest that improvements in depression symptoms, in particular in the somatic symptoms of depression, explained the long-term beneficial impact of treatment for highly anxious men. Intriguingly, this is the group for whom the program was originally designed and initially positively evaluated in the Ischemic Heart Disease Life Stress Monitoring Program (4, 5). However, we can only speculate as to why the treatment-related difference in prognosis in high-anxious men did not begin to become apparent until more than 1 year after the end of the program. One factor, may have been that consequent to the treatment-related reduction in the somatic symptoms of depression, a significantly greater proportion of high-anxious men in the control group (37.5%) had total 1-year BDI scores 10 than their counterparts in the treatment group (27.6%, p = .042). Scores at this level on the BDI are considered indicative of at least mild to moderate levels of depressive symptoms, and when measured during hospitalization for an MI, such scores have been linked to cardiac prognosis (24). However, our recent long-term analyses of BDI scores in hospital and at 1 year in patients receiving usual care (25) suggest that although both scores have prognostic importance, the impact is less marked at 1 year. Outside the pathophysiological instability during the first few months after an acute MI, it may take longer for chronic depression to influence prognosis.
An early study, carried out by Shaw et al. (7) in a sample of 30 post-MI patients, suggested that the degree of fit between repressive style and cardiac rehabilitation approaches might influence outcomes. Self-reported outcomes at 6 months suggested that repressors (high MC scores but low in anxiety) who had retained high levels of information at discharge had more subsequent complications and worse psycho-medical functioning (sleep, depression, and tension) than those with low levels of knowledge. In contrast, highly anxious patients with high levels of knowledge had better social functioning than those with low knowledge. Shaw’s group speculated that repressors may be more comfortable with low levels of information about cardiac disease, and that higher levels result in information overload and increases in arousal. However, the highly anxious or sensitizers may do better when provided with additional information that may act to lower anxiety.
All M-HART patients who were visited by nurses received some degree of education about cardiac disease and risk factors. In addition, it is likely that the monthly telephone monitoring calls from a telephonist who identified herself as "from the Montreal Heart Attack Readjustment Trial" kept most treatment group patients’ awareness of their cardiac conditions at fairly high levels. Although this may have been comforting for the type of patient who seeks out information or health care system contact as a form of reassurance, it may well have been a source of distress for patients whose preferred mode of coping was repression, or information avoidance.
Our research has several limitations. First, although this is the first report of the long-term follow-up of the M-HART study, it nonetheless constitutes a secondary data analysis and shares the limitations of the original study. However, in an attempt to minimize the problems associated with multiple statistical testing of subgroups, we focused on two questions: the durability of sex differences in outcomes observed during the first program year and the possibility that the program had different outcomes depending on the coping style of the participants. Nonetheless, there were many tests involved in assessing the various potential mediators in each anxiety and repression subgroup, and this limitation should be kept in mind when considering results. We used median splits for the S-STAI and MC scales in the overall sample to define the repressor/anxiety groups. Medians for these scales differed for men and women. Nonetheless, when patients were classified using the sex-specific medians, the pattern of results was the same. The literature varies in exactly how the combination of the S-STAI and MC scores is used to define various repressor groups (20). Some authors have used sample-specific means, some have used medians, and others have used upper quartiles. Our choice of median splits was based on an attempt to assure adequate power in the resulting subgroups. However, the lack of consistent thresholds for defining repressor and anxiety subgroups makes the reliability of the classifications questionable across studies. Despite these potential limitations, the mechanisms our analyses identified as potentially responsible for the treatment-related success in highly anxious men and poor prognosis in treatment group repressors make sense in terms of the existing literature and are internally consistent, lending additional support to the repressor/sensitizer concept.
There is increasing evidence that repressors have worse health outcomes than their less defensive peers (20). Although the clinical importance of "information coupling" or tailoring has been discussed for years (22), we know very little about the best way to assure that cardiac patients’ with repressive coping styles receive enough information to assure compliance with needed lifestyle changes without increasing psychological distress. The development and evaluation of treatment packages optimizing the fit between intervention styles; disease-specific treatment needs; and patients’ personalities and social and family realities constitutes a major clinical and research challenge for the future.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
In addition to all the patients who took part in the M-HART project and all the people who helped with the study, whom we have acknowledged in previous publications, we are indebted to the Régie de l’assurance maladie du Québec for providing information on physician contacts. We also thank Sylvie Cossette, RN, PhD, for her helpful comments on an earlier draft.
Received for publication May 31, 2001.
Revision received September 12, 2001.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  W. Linden, M. J. Phillips, and J. Leclerc Psychological treatment of cardiac patients: a meta-analysis Eur. Heart J., December 2, 2007; 28(24): 2972 - 2984. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K.A. Ertel, M.M. Glymour, T.A. Glass, and L.F. Berkman Frailty modifies effectiveness of psychosocial intervention in recovery from stroke Clinical Rehabilitation, June 1, 2007; 21(6): 511 - 522. [Abstract] [PDF]
|
|
|
|
 |
|
 J. Sareen, F. Jacobi, B. J. Cox, S.-L. Belik, I. Clara, and M. B. Stein Disability and poor quality of life associated with comorbid anxiety disorders and physical conditions. Arch Intern Med, October 23, 2006; 166(19): 2109 - 2116. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. Jiang, M. Kuchibhatla, M. S. Cuffe, E. J. Christopher, J. D. Alexander, G. L. Clary, M. A. Blazing, L. H. Gaulden, R. M. Califf, R. R. Krishnan, et al. Prognostic Value of Anxiety and Depression in Patients With Chronic Heart Failure Circulation, November 30, 2004; 110(22): 3452 - 3456. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. J. Cunningham and K. Watson How Psychological Therapy may Prolong Survival in Cancer Patients: New Evidence and a Simple Theory Integr Cancer Ther, September 1, 2004; 3(3): 214 - 229. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
