Negative Attitudes Among Short-Term Stroke Survivors Predict Worse Long-Term Survival
Background and Purpose—Patients respond to serious illness in different ways. We wished to determine whether different attitudes toward illness are associated with survival after stroke.
Methods—Three hundred seventy-two stroke patients were identified and medically assessed as part of a randomized trial to evaluate a stroke family care worker. They had all survived 6 months from randomization. A research psychologist visited each patient and administered the Mental Adjustment to Stroke Scale (a self-rated attitude scale based on the Mental Adjustment to Cancer Scale). Disability and dependence (Barthel Index, modified Rankin Scale) and mood (Hospital Anxiety and Depression Scale, General Health Questionnaire 30) were also assessed. Patients were followed up in 1998 (3 to 5 years after the initial stroke) to establish their survival. We modeled the relationship between Mental Adjustment to Stroke scores and survival, adjusting for other factors associated with stroke survival.
Results—Eighty-two patients (22%) died within 3 years. After adjustment for other significant factors, fatalism and helplessness/hopelessness were both associated with decreased survival (P=0.03 and 0.04, respectively), but fighting spirit, anxious preoccupation, and denial/avoidance were not. Mood was not associated with survival.
Conclusions—Patients’ attitudes toward their illness seem to be associated with survival after stroke. Patients who feel that there is nothing they can do to help themselves 6 months after a stroke have a shorter survival. These findings need to be confirmed and any causal relationship between attitude and survival further explored in a randomized controlled trial to “improve” the attitude of stroke patients toward their illness.
Patients’ attitudes toward illness vary, and questionnaires have been developed to identify them. The Mental Adjustment to Cancer (MAC) Scale categorizes the attitudes of cancer patients toward their illness into fighting spirit, helplessness/hopelessness, anxious preoccupation, fatalism, and denial/avoidance.1 An attitude of fighting spirit or denial/avoidance, as opposed to helplessness/hopelessness or fatalism, predicts a greater likelihood of remaining alive and free of recurrence in patients with breast cancer for up to 15 years.2 3 A more recent study showed that helplessness/hopelessness was associated with decreased survival in breast cancer patients followed up for at least 5 years.4 We adapted the MAC Scale for use in stroke patients. We sought to test the hypothesis that helplessness/hopelessness and fatalism are also associated with reduced survival after stroke and to examine whether these effects are independent of demographic, physical, and mood factors that have been associated with poor outcomes after stroke.5 6 7
Subjects and Methods
We identified patients as part of a randomized trial to evaluate the effect of a stroke family care worker8 (the trial showed no statistically significant effect of the family care worker on patients’ outcome). They were eligible for inclusion into the trial if they were referred to our hospital as an inpatient or outpatient between October 1993 and September 1995, within 30 days of an acute stroke. They were excluded if they had a subarachnoid hemorrhage, lived >25 miles away, were unlikely to survive the acute stroke, or had another disease, such as cancer, that was likely to dominate their care. Over the study period, 65% (417 of 643) of all patients referred to our hospital with an acute stroke were included in the trial. Patients were included in the present observational study if they survived 6 months from entry into the trial.
In part 1, a study neurologist performed a clinical assessment of each patient on arrival at the hospital, including routine blood tests, an ECG, and a CT brain scan.
In part 2, a research psychologist visited each surviving patient (usually at the patient’s normal residence) 6 months after the part 1 assessment and administered the following measures: (1) modified Rankin Scale; (2) Barthel Index (modified version); (3) Hospital Anxiety and Depression Scale (HAD); (4) General Health Questionnaire 30 (GHQ30); and (5) Mental Adjustment to Stroke Scale (MASS). The modified Rankin Scale9 is an observer-rated ordinal scale indicating patients’ overall level of symptoms and dependence on others. A high score indicates greater dependence. The Barthel Index (modified version)10 is an observer-rated 10-item ordinal scale indicating patients’ degree of independence in activities of daily living. A high score indicates greater independence. The HAD11 is a widely used self-rated scale for symptoms of depression and anxiety developed for use in the medically ill. A high score indicates greater anxiety and depression. The GHQ3012 is a self-rated measure of psychiatric morbidity. A high score indicates greater psychiatric morbidity. The MASS is a self-rated scale developed for our study by rewording the MAC scale,1 replacing the word cancer with stroke. It includes 5 subscales: fighting spirit, helplessness/hopelessness, anxious preoccupation, fatalism, and denial/avoidance (Table 1⇓). Patients with fighting spirit are determined to get well and are optimistic. Patients with feelings of helplessness/hopelessness are overwhelmed by knowing they have had a stroke and are afraid they are dying. Patients with anxious preoccupation seek information and worry about their symptoms. Patients with a fatalistic attitude acknowledge their stroke but seek no further information and carry on with their lives. Patients with denial/avoidance either deny they have had a stroke or minimize its seriousness. For fighting spirit, a high score indicates a more positive attitude, but for the other subscales, a high score indicates a more negative attitude.
Because we had altered the scale and because it had not been previously used for stroke, we assessed its test-retest reliability over 1 to 2 weeks in the first 97 patients entered into the study. The κ values for individual questions ranged from 0.18 to 0.89 (Table 1⇑) but were generally satisfactory. The internal consistency of the scales was assessed with the use of Cronbach’s α (Table 1⇑), which suggested reasonable internal consistency.
The Barthel Index, modified Rankin Scale, and GHQ30 were completed at the part 2 (6-month) interview, while the HAD and MASS were left with the patient to complete independently.
Patients were followed up in 1998 (3 to 5 years after the initial stroke) to establish their survival.
The statistical package SAS was used.13 Pearson correlation coefficients were calculated for the relationships between the MASS subscales and the mood measures (HAD and GHQ30). Survival was modeled with Cox proportional hazards regression.14 Variables included from the part 1 assessment were as follows: age, sex, diabetes mellitus, ischemic heart disease, peripheral vascular disease, living alone before the stroke, prestroke dependence (modified Rankin Scale score of >2), urinary incontinence, inability to lift both arms, inability to walk, and verbal deficit according to the Glasgow Coma Scale.15 Variables included from the part 2 assessment were as follows: modified Rankin Scale, Barthel Index, HAD subscales, GHQ30, and MASS subscales.
We used a forward selection procedure with the level of significance for entry into the model set at 0.2. Entry into the model was determined by the change in magnitude of the log likelihood. Variables describing age, stroke severity, and comorbidity were entered first, and when no more were statistically significant, the psychological factors were entered into the model. In the regression procedure, missing values for each variable were set equal to the mean value of the nonmissing data for that variable (ie, the mean value of the other patients’ values). The quantity of missing data depended on the variable, but the maximum was 93 patients (25%) for MASS denial/avoidance. We also analyzed the data excluding patients with any missing data, and it made little difference to the overall results.
The MASS subscales were analyzed as continuous variables of raw scores. We also ran the analysis breaking the subscales into categories, and the results were very similar.
We included 372 of the 417 randomized trial patients in this observational study. Reasons for exclusion were dying before the 6-month assessment (n=41), brain tumor (n=1), refusing follow-up (n=2), and emigration (n=1). Forty-seven of the 372 patients (13%) had a total anterior circulation syndrome, 147 (40%) a partial anterior circulation syndrome, 103 (28%) a lacunar syndrome, 60 (16%) a posterior circulation syndrome, and 15 (4%) an unknown syndrome.16 CT or MR scans were performed on 342 patients (92%); 312 (84%) had ischemic strokes, and 30 (8%) had primary intracerebral hemorrhage. Table 2⇓ shows the characteristics of the patients at the part 1 (stroke onset) assessment. Table 3⇓ shows the variables measured at the part 2 (6-month) assessment, including the proportion of patients from whom we obtained data. The main reasons for missing data were patients’ cognitive and communication problems and turning over 2 pages at once when completing self-report questionnaires.
Relationships Between the Psychological Measures at Part 2 (6-Month) Assessment
Correlation coefficients between the HAD subscales, the GHQ30, and the MASS subscales are shown in Table 4⇓.
Eighty-two of 372 patients (22%) died within 3 years of the initial stroke assessment (ie, within 2.5 years of the assessment of attitude and mood). A univariate Cox proportional hazards regression demonstrated strong associations between several demographic and clinical features and survival (Table 5⇓). It also demonstrated that fighting spirit was associated with increased survival, while helplessness/hopelessness and fatalism were associated with decreased survival. Anxious preoccupation, denial/avoidance, and mood were not significantly associated with survival.
We entered the explanatory variables collected at the part 1 and part 2 assessments into the regression analysis using a forward selection procedure. In the final model, older age, diabetes, ischemic heart disease, peripheral vascular disease, prestroke dependence, inability to walk at initial assessment, and living alone were all associated with decreased survival. After adjustment for these factors, the association of helplessness/hopelessness and fatalism with survival remained statistically significant (P=0.04 and 0.03, respectively), but the association of fighting spirit did not (P=0.2) (Table 6⇓). Anxiety and depression, as measured by the HAD subscales and the GHQ30, did not add significantly to this model irrespective of whether the MASS subscales were included. Table 6⇓ also shows the relative magnitude of the effects of the MASS subscales, after adjustment for other prognostic factors. Patients who had one of the highest 10% of fighting spirit scores were 39% more likely to die at any given time than patients with one of the lowest 10% of fighting sprit scores.
In our study reduced survival is predicted by negative attitudes toward illness but not by depression or anxiety, even after stroke severity and other prognostic variables have been accounted for.
The MASS has not been used elsewhere, and confirmatory observational studies are needed. However, Watson et al4 found that helplessness/hopelessness was a predictor of decreased survival in breast cancer. They did not find a significant association between fatalism and survival. In lung cancer, depressive coping (measured by the Freiburg Questionnaire of Coping With Illness) has been associated with decreased survival.17 Depressive coping was characterized by brooding, arguing with fate, pitying oneself, acting impatiently, and taking it out on others.
Depression has been previously linked with increased mortality after stroke. Morris et al18 followed up a cohort of 103 stroke patients and found that patients with depression were 3 times more likely than others to die within 10 years, even after adjustment for other prognostic variables. They identified depression using the Present State Examination and measured its severity with the Hamilton Rating Scale. Arfken et al19 showed a similar relationship among 455 medically ill older adults followed up for 1 year. They used the Geriatric Depression Scale and showed that moderate depression (odds ratio, 5.0) and male sex (odds ratio, 3.4) were independent risk factors for dying. Frasure-Smith et al20 followed up 222 patients after myocardial infarction and showed that the Beck Depression Inventory was significantly associated (odds ratio comparing scores of ≥10 to scores of <10=6.6; P=0.003) with 18-month mortality after adjustment for other predictors of mortality. Our failure to find an association between depression and survival could be explained if the HAD and GHQ30 were not good measures of depression in our population. However, we have previously shown that the HAD and GHQ are reasonable measures of depression in a subgroup of 105 (25%) of the patients included in the present study.21
We studied the HAD scale together with the MASS to explore the differences between anxiety and depression and “attitude.” There was some overlap between what the HAD scale and the MASS measured, as shown by the correlations in Table 4⇑. Although our data suggest a relationship between HAD depression and helplessness/hopelessness, the latter is not purely a measure of depression since it predicted survival, whereas HAD depression did not. On the other hand, anxious preoccupation showed associations similar to HAD anxiety, in terms of both correlations with other MASS subscales and a lack of association with survival, and perhaps they are measuring the same entity. The MASS clearly contains items that reflect mood. However, our data suggest that it seems to capture something distinct from mood, at least as measured by the HAD and GHQ30, given that only the MASS was associated with survival. There was a substantial amount of missing data for the MASS. Different methods of dealing with the missing data made no difference to the overall results.
There are several plausible explanations for our observations. Patients’ attitudes toward their stroke may change over time. For instance, if a patient with a severe stroke recovered a great deal of function in the first few months, he/she may then have a more positive attitude 6 months after the stroke. A patient with a stroke of similar severity who had not recovered much function may have felt very negative after 6 months. Thus, the MASS may be related to the amount of recovery a patient had made thus far, which in turn might be a good predictor of survival. We attempted to control for this by adjusting for other predictive factors, including the patient’s functional status at 6 months after the stroke (which was significantly related to survival in a univariate analysis but not in a multivariate analysis after adjustment for more strongly significant variables).
The relationships between survival and aspects of the MASS held after adjustment for other prognostic factors, although the relationships were weakened. This could indicate that negative attitudes hasten death, or we may have simply failed to adjust fully for stroke severity. It is possible that patients intuitively know how severe their disease is, in a way that simple clinical variables cannot match. This may not detract from the usefulness of attitude in prognostic models, but it would undermine the notion of causality. It is also possible that a particular attitude could in some way cause a patient to have more severe stroke symptoms, which in turn might cause the patient to survive for a shorter time. If this is the case, then in our multivariate analysis we will have adjusted for something that is between attitude and survival on the causal pathway. In other words, we may have hidden important relationships between attitude and survival in our multivariate analyses.
A causal relationship is biologically plausible, although mechanisms remain speculative. In experimental animals, Spiegel22 has shown that acute stress affects hypothalamic function, which leads to glucocorticoid receptor hypersensitivity and causes immunosuppression. The functions of the hypothalamus and pituitary adrenal axis have been associated with both mood and survival after stroke.23 Rozanski et al24 outline possible mechanisms in cardiovascular disease.
There is no evidence that any intervention can alter a patient’s attitude to stroke. However, if such an intervention could be developed, its effect on survival could be tested in a randomized trial. A positive trial would establish a causal link between attitude and survival.
Patients’ attitudes to their stroke are associated with survival. Patients who are fatalistic and feel helpless or hopeless, ie, who feel that there is nothing they can do to help themselves, do not survive as long as other patients. This seems to remain true even when physical factors, such as stroke severity, are accounted for. We now need to confirm this observation in further observational studies and then examine whether any intervention can decrease feelings of fatalism and helplessness and in turn improve survival after stroke.
This work was supported by grants from the Chief Scientist’s Office (Scotland) and Chest, Heart, and Stroke Scotland.
Reprint requests to Dr S.C. Lewis, Bramwell Dott Building, Department of Clinical Neurosciences, Western General Hospital NHS Trust, Crewe Rd, Edinburgh EH4 2XU, UK.
- Received January 18, 2001.
- Revision received March 8, 2001.
- Accepted March 12, 2001.
- Copyright © 2001 by American Heart Association
Watson M, Greer S, Bliss JM. Mental Adjustment to Cancer (MAC) Scale Users’ Manual. 1989.
Lai SM, Alter M, Friday G, Sobel E. Prognosis for survival after an initial stroke. Stroke. 1995;26:2011–2015.
Counsell CE. The Prediction of Outcome in Patients With Acute Stroke [PhD thesis]. Cambridge, UK: Girton College; 1998:138–143.
Anderson CS, Jamrozik KD, Broadhurst RJ, Stewart-Wynne EG. Predicting survival for 1 year among different subtypes of stroke: results from the Perth Community Stroke Study. Stroke. 1994;25:1935–1944.
Dennis M, O’Rourke S, Slattery J, Staniforth T, Warlow C. Evaluation of a stroke family care worker: results of a randomised controlled trial. BMJ. 1997;314:1071–1077.
Wade DT. Activities of Daily Living (ADL) and Extended ADL Tests: Measurement in Neurological Rehabilitation. Oxford, UK: Oxford University Press; 1992:175–194.
Goldberg DP. The Detection of Psychiatric Illness by Questionnaire: A Technique for the Identification and Assessment of Non-psychotic Psychiatric Illness. London, UK: Oxford University Press; 1972. Maudsley Monograph No. 21.
SAS Version 6.12. Cary, NC: SAS Institute Inc; 1997.
Cox DR. Regression models and life-tables. J R Stat Soc B. 1972;34:187–220.
Arfken CL, Lichtenberg PA, Tancer ME. Cognitive impairment and depression predict mortality in medically ill older adults. J Gerontol. 1999;54A:M152–M156.
Frasure-Smith N, Lesperance F, Talajic M. Coronary heart disease/myocardial infarction: depression and 18-month prognosis after myocardial infarction. Circulation. 1995;91:999–1005.
O’Rourke S, MacHale S, Signorini D, Dennis M. Detecting psychiatric morbidity after stroke: comparison of the GHQ and the HAD scale. Stroke. 1998;29:980–985.
Rozanski A, Blumenthal JA, Kaplan J. Impact of psychological factors on the pathogenesis of cardiovascular disease and implications for therapy. Circulation. 1999;99:2192–2217.