Pessimism and hostility scores as predictors of patient satisfaction ratings by medical out-patients
The Authors
Brian A. Costello, Division of General Internal Medicine, Mayo Clinic, Rochester, Minnesota, USA
Thomas G. McLeod, Division of General Internal Medicine, Mayo Clinic, Rochester, Minnesota, USA
G. Richard Locke III, Division of General Internal Medicine, Mayo Clinic, Rochester, Minnesota, USA
Ross A. Dierkhising, Division of Biostatistics, Mayo Clinic, Rochester, Minnesota, USA
Kenneth P. Offord, Division of Biostatistics, Mayo Clinic, Rochester, Minnesota, USA
Robert C. Colligan, Department of Psychiatry and Psychology, Mayo Clinic, Rochester, Minnesota, USA
Abstract
Purpose – The purpose of this research is to determine whether a pessimistic or hostile personality style adversely affects satisfaction with out-patient medical visits. Many patient and health care provider demographic characteristics have been related to patient satisfaction with a health care encounter, but little has been written about the association between patients' personality characteristics and their satisfaction ratings.
Design/methodology/approach – An eight-item patient satisfaction survey was completed by 11,636 randomly selected medical out-patients two to three months after their episode of care. Of these, 1,259 had previously completed a Minnesota Multiphasic Personality Inventory (MMPI). The association of pessimism and hostility scores with patient satisfaction ratings was assessed.
Findings – Among patients who scored high on the pessimism scale, 59 percent rated overall care by their physicians as excellent, while 72 percent with scores in the optimistic range rated it as excellent (p=0.003). Among the hostile patients, 57 percent rated their overall care by physicians as excellent, while 66 percent of the least hostile patients rated it as excellent (p=0.002).
Originality/value – Pessimistic or hostile patients were significantly less likely to rate their overall care as excellent than optimistic or non-hostile patients.
Article Type:
Research paper
Keyword(s):
Patients; Personality; United States of America; Health services.
Journal:
International Journal of Health Care Quality Assurance
Volume:
21
Number:
1
Year:
2008
pp:
39-49
Copyright ©
Emerald Group Publishing Limited
ISSN:
0952-6862
Introduction
Practicing clinicians know that certain patients are difficult to satisfy, inconsolable and personally challenging, no matter how comprehensive, efficient and expert their care, nor how polished the physician's interpersonal skills. Patient satisfaction has been widely studied. Satisfaction determinants range from structurally-based ones such as the type of health care delivery system, to physician characteristics, including interactional style and the physician's age (Kirsner and Federman, 1997). The Medical Outcomes Study (MOS) evaluated patient satisfaction based on practice type and payment method (Rubin et al., 1993). The MOS included more than 17,000 patients who had filled out a nine-item questionnaire after an out-patient visit. Patients seen in a solo or single-specialty practice, rather than in multi-specialty groups or health maintenance organizations, were most likely to rate the overall visit as excellent. Physician age and gender have also been correlated with patient satisfaction; specifically, lower satisfaction is reported by patients after seeing younger, female physicians (Hall et al., 1994). Effective communication skills and particular physician behaviors, such as performing a physical examination, have been associated with higher patient satisfaction ratings (DiMatteo et al., 1980; Bartlett et al., 1984; Robbins et al., 1993). Organizational factors, particularly patient waiting time, have been studied and, as expected, longer waiting times result in lower patient satisfaction (Probst et al., 1997).
Although much has been written about organizational, structural, and physician-specific factors related to patient satisfaction, surprisingly little has been written about patient characteristics and their relationship to medical care satisfaction ratings. For example, patients who believe their health status is good are more satisfied with their care (Probst et al., 1997). Patient satisfaction is associated with compliance and willingness to continue receiving care from a particular physician (Rubin et al., 1993). Conversely, patients with depression or anxiety disorders are more likely to express dissatisfaction from unmet expectations (Kroenke et al., 1997). Furthermore, in a study of patient characteristics among out-patients at our center, lower satisfaction ratings were reported by patients who were young, who were employees of the center, and who lived locally (Locke, G.R. personal communication, September 2002). Therefore, understanding factors associated with patient satisfaction, including the patients' personality characteristics, is important. This study investigated the personality characteristics of pessimism and hostility as they relate to patient satisfaction ratings. We hypothesized that pessimistic or hostile patients would report less satisfaction with their care.
Methods
Two archival data sets were abstracted. The first data set consisted of information from 11,636 medical out-patients who completed a patient satisfaction survey for the Department of Internal Medicine from March 1998 through March 1999. An eight-item questionnaire was mailed to randomly selected medical out-patients two to three months after their care episode. A five-point scale was used to rate seven items pertaining to satisfaction with access, interactions with physicians and allied health staff, general satisfaction with the visit and willingness to recommend our center to others. The topic of the eighth item was the completion of all scheduled tests and consultations (yes/no), not satisfaction. The survey response rate was 60 percent. The second data set comprised scores from approximately 335,000 Minnesota Multiphasic Personality Inventories (MMPIs) archived at our institution since 1959. When the two data sets were merged we found that 1,259 medical out-patients had completed both the patient satisfaction survey and, before that survey was completed, an MMPI. All MMPIs had been obtained before the satisfaction survey. Of the MMPI and the patient satisfaction responses, 1,259 out-patients formed the basis for our study. We then assessed the degree of association between:
- personality traits of pessimism and hostility, as assessed by the MMPI; and
- patients' reports of satisfaction with care.
The study was approved by our institutional review board, and all study subjects gave research authorization. Pessimism and hostility were chosen for study because these personality traits, as assessed by the MMPI, are considered relatively stable (Maruta et al., 1993; Maruta et al., 2000). Furthermore, in our experience, these qualities come closest to defining the intangible characteristics that physicians sense in patients who are difficult to satisfy. The MMPI consists of 550 unique true/false items about thoughts, feelings, attitudes, physical symptoms, emotional symptoms, and previous life experiences (Swenson et al., 1990). It was initially developed when McKinley and Hathaway (1943) noted that:
Competent internists have estimated variously that from 30 to 70 percent of the ambulatory patients who appear for medical attention come primarily because of one or more complaints that turn out to be psychoneurotic in nature.
The MMPI scales are reported as T-scores, which are standardized to a mean of 50 and an SD of 10. High scores on the pessimism (PSM) scale reflect a pessimistic explanatory style; low scores, optimistic. The PSM scale for the MMPI is based on Seligman's explanatory style theory (Colligan et al., 1994), which suggests that people who believe that the cause of an adverse event is internal and personal (rather than external), stable (rather than transient) and global (rather than specific) possess a pessimistic personality trait. Research shows that a pessimistic explanatory style is predictive of an increased likelihood of depression, poorer physical health, lower levels of achievement and increased use of medical and mental health services (Seligman, 1989, pp. 5-32). Hostility was measured by the MMPI hostility (Ho) scale (Cook and Medley, 1954). The scale's developers describe a “hostile person” (i.e. scoring high on the Ho scale) as:
… lone who has little confidence in his fellowman… [and] sees people as dishonest, unsocial, immoral, ugly and mean…Hostility amounts to chronic hate and anger.
Statistical methods
Logistic regression models were used to assess the association between patient satisfaction ratings and the PSM and Ho scores. Our large sample size allowed us to detect small, but significant, relationships that might not otherwise be uncovered in smaller or more highly selected samples. The dependent variables were seven patient satisfaction responses from the eight-item survey. For six questions, responses were coded “1” for “excellent” or “0” for all other response categories to that item. For the question about recommending the medical center to others, responses were coded “1” for “definitely would recommend” or “0” for any other response. This procedure for collapsing and dichotomizing the five-point scale is in keeping with the convention from marketing literature (Jones and Sasser, 1995). Additionally, from experience with previous internal analyses, we also adjusted for previously identified predictive covariates. These included patient's residence distance from our medical center, age, being a medical center employee or a dependent of an employee (“employee/dependent”; 1=yes; 0=no) and receiving primary care. Two interaction variables were included in the models as well: 1 – distance of residence from our center by employee/dependent status and 2 – age by employee/dependent status. These are known and important explanatory variables when modeling patient satisfaction at our center. Finally, we adjusted for the patient's rating of whether all the tests and consultations ordered had been completed during the patient's episode of care (yes/no). We considered this variable because many of our patients travel long distances to our center for intensive out-patient evaluations during a single episode of care. Therefore, we believed that scheduling a timely, efficient, and complete appointment itinerary could affect ratings of patient satisfaction. Three sets of explanatory variables were used in the modeling. The first model was done univariately, including only the MMPI scale of interest (i.e. either PSM or Ho). The second model included only one of the MMPI scales of interest and the adjusting variables described above. The third included both MMPI scales of interest and the adjusting variables already mentioned. Wald χ2 statistic p values were calculated from these logistic regression models. Since the MMPI and the patient satisfaction survey were not completed concurrently, we assessed whether the time between completing the MMPI and the patient satisfaction survey affected the associations we intended to study. Therefore, we modeled the interaction of time and the MMPI scale score within the context of two models. One included the MMPI scale of interest (i.e. either PSM or Ho) and the interaction between time and the MMPI scale of interest as explanatory variables. The other model consisted of the MMPI scale of interest, the adjusting variables described above and the interaction between time and the MMPI scale as explanatory variables.
Results
Patient characteristics
At the time the satisfaction survey was completed by the 1,259 out-patient participants:
- (10 percent) were 18 to 40 years old;
- (49 percent) were 41 to 65 years old; and
- (41 percent) were 66 years or older.
Travel distance varied considerably:
- patients (31 percent) lived within 20 miles;
- (31 percent) lived 21 to 120 miles away;
- (16 percent) lived 121 to 250 miles away; and
- (23 percent) lived more than 250 miles away.
The survey was completed as follows:
- (15 percent) within 1 year after taking the MMPI;
- (18 percent) 1.1 to 5 years after the MMPI;
- (17 percent), 5.1 to 10 years after the MMPI;
- (50 percent), more than 10 years after the MMPI; and
- had missing time data.
Among these patients:
- (57 percent) were female;
- (16 percent) were medical center employees or dependents; and
- (22 percent) were seeking primary care.
During the visit studied, 1,039 patients (83 percent) had their tests and consultations completed (85 [7 percent] had missing data).
Findings
Table I shows the percentage of “excellent” ratings by PSM T-score categories and the p values from the logistic regression models with three sets of explanatory variables. In general, with higher PSM T-scores, reflecting an increasing pessimistic explanatory style, the percentage of patients giving “excellent” ratings was significantly lower. This pattern was fairly consistent for all survey questions. With higher PSM T-scores, the odds of a patient giving the center an “excellent” rating were significantly lower, even after adjusting for completion of tests/consultations and for all the previously identified predictors of satisfaction. However, the association weakened when adjusting for the Ho T-score. This can be explained by the relatively large positive correlation between the PSM and Ho T-scores (r=0.61; p<0.001). The exception is the question pertaining to the patient's willingness to recommend the center, for which the ratings were not associated with these MMPI scale scores.
Pessimistic patients (PSM T-score ≥60) were significantly less likely to give “excellent” ratings for various aspects of their care than those classified as optimistic (T-score ≤39). Specifically, 72 percent of the optimistic patients rated it excellent (p=0.003), while 59 percent of the pessimistic patients rated the overall care provided by their physicians as excellent. Furthermore, a significantly smaller proportion of pessimists than optimists rated other aspects of their care as excellent:
- respect shown by physicians (64 percent v. 75 percent; p=0.008);
- willingness of physicians to listen to the patient and family (60 percent v. 67 percent; p=0.047);
- physicians responding to questions about the patient's medical condition and treatment (57 percent v. 71 percent; p=0.002);
- helpfulness of allied health staff (53 percent v. 65 percent; p=0.002); and
- overall care received (55 percent v. 63 percent; p=0.003).
However, the PSM scale scores were not associated with willingness to recommend the center (86 percent of pessimists v. 89 percent of optimists; p=0.282). Table II displays the percentage of “excellent” ratings by Ho T-score groupings and the p values from the logistic regression models. The same patterns were present among the “excellent” ratings in relation to patient hostility (i.e. higher Ho T-scores) as for the PSM scale. The results from the models were also similar. A smaller proportion of patients who scored high on hostility (Ho scale T-score ≥60) gave excellent ratings on various aspects of their care than did patients who scored low on the Ho scale (T-score ≤39). This is evident in their responses on individual items:
- overall care provided by their physicians (57 percent v. 66 percent; p=0.002);
- respect shown by physicians (57 percent v. 70 percent; p=0.008);
- willingness of physicians to listen to patient and family (54 percent v. 66 percent; p=0.001);
- physicians responding to questions about the patient's medical condition and treatment (53 percent v. 67 percent; p<0.001);
- helpfulness of allied health staff (47 percent v. 64 percent; p<0.001); and
- overall care received (52 percent v. 63 percent; p=0.001).
Again, the exception to this trend was the question pertaining to a patient's willingness to recommend the medical center to others. There was no significant difference in the percentages of hostile patients (Ho T score ≥60; 83 percent) and non-hostile patients (Ho T score ≤39; 88 percent) who reported “definitely would recommend the center” (p=0.434).
Table III displays the odds ratios for “excellent” ratings corresponding to 10-unit (1 SD) increases in MMPI scale scores for PSM and Ho. The odds ratios were estimated for each of the three models. When significant associations existed, there was an approximately 15 percent to 20 percent decrease in the odds of giving an excellent rating for every 10-point increase in the PSM or Ho T-scores. The time interval between completing the MMPI and the patient satisfaction survey did not affect the association between the MMPI T-scores and the satisfaction ratings, except for the question about willingness to recommend the center. Patients who had a long interval between completing the MMPI and the patient satisfaction survey were more likely to “definitely recommend” the center to others than were those with a shorter interval. Among patients who would definitely recommend the center, the estimated odds ratios that corresponded to a 10-unit (1 SD) increase in the MMPI T-scores increased with the interval. In general, these odds ratios increased by about 0.02 to 0.03 for every 5-year interval increase.
Discussion
Patient satisfaction has been studied from various vantage points. However, our approach to studying the contributions of patient personality has not been previously taken. Our analysis shows that pessimistic patients (i.e. having a pessimistic explanatory style as theorized by Seligman) or hostile patients (i.e. high scores on the Ho scale) are significantly less likely to rate satisfaction with their care as excellent. Our results are a first step towards understanding that certain aspects of the patient's personality affect ratings of their satisfaction with care. Previously, these aspects of patient personality were experienced subjectively and understood solely through physician's intuition and judgment. Adverse encounters were believed to result from the physician's characteristics and behaviors or from organizational and structural factors surrounding the care episode. Now it is evident that some aspects of the patient's personality affect ratings of satisfaction with care. These occur independently of other factors already known to contribute to variations in ratings of patient satisfaction. Clearly there are likely to be other patient-related characteristics that affect patient satisfaction ratings. These may include emotional states (e.g., depression) at the time of the survey or episode of care, or certain patient expectations, unknown to the physician, before the encounter. Additionally, patient satisfaction ratings for particular physicians and health care organizations are tacitly assumed to be a reflection of physicians or health care system. However, our findings indicate that important data about the factors contributing to patient satisfaction ratings are missing if patient personality characteristics are not considered.
It is notable that willingness to recommend our center to family and friends was not associated with hostility or pessimism. This may result from feelings that, regardless of their opinion about some aspects of their experience during a particular care episode, these patients would still recommend the center because of other factors such as the center's reputation. We included all patients who had completed both a patient satisfaction survey and an MMPI and did not exclude cases in which there was a long interval between the survey and the MMPI. Time limits were not imposed because we were studying two personality traits that are relatively stable. Furthermore, one might speculate that patients coming to a tertiary care medical center for evaluation, who had also been asked to complete an MMPI, had a combination of medical and psychological issues requiring multidisciplinary investigation. Such patients are diagnostically and personally challenging for physicians. Patient satisfaction ratings may be decreased if these patients are also characterized by traits of pessimism or hostility.
Patient satisfaction at our center is associated with several variables, including status as employee/dependent, age of patient, distance traveled for care, and type of care received. We included these independent variables in the models as adjusting variables. Nonetheless, significant differences were noted in satisfaction on the basis of levels of hostility or pessimism. Although these findings are intriguing, the possibility exists that physician care or manner may be modified by patient personality factors, since astute physicians are attuned to the personal qualities of their patients. Finally, while contributing little to the practical management of hostile or pessimistic patients, this research is a first step towards understanding how patient personality affects reported levels of satisfaction with medical care and why some patients are difficult to satisfy regardless of other factors related to their care.
Conclusions and recommendations
This research clearly demonstrates that patient personality characteristics play an important role in the results derived from patient satisfaction surveys. Physicians are much less likely to obtain excellent ratings from pessimistic or hostile patients, as defined in this article. This influence is independent of physician or practice characteristics. Patient mix may temper an individual doctor's satisfaction ratings. Therefore, institutional policy makers need to be aware of the potentially significant contribution of patient personality factors in adversely affecting the patient satisfaction ratings of individual physicians. This research also suggests that it may be informative to identify personality characteristics of those who were sent satisfaction questionnaires and did not return them.
Table IRelationship between patient satisfaction rating of “Excellent” and the PSM T-score
Table IIRelationship between patient satisfaction rating of “Excellent” and the Ho T-score
Table IIIOdds ratios (95 percent CI) for percentage of “Excellent” ratings for a 10-unit (1 SD) increase in the MMPI T-scores*
References
Bartlett, E.E., Grayson, M., Barker, R., Levine, D.M., Golden, A., Libber, S. (1984), "The effects of physician communications skills on patient satisfaction, recall, and adherence", Journal of Chronic Diseases, Vol. 37 No.9-10, pp.755-64.
Colligan, R.C., Offord, K.P., Malinchoc, M., Schulman, P., Seligman, M.E. (1994), "CAVEing the MMPI for an optimism-pessimism scale: Seligman's attributional model and the assessment of explanatory style", Journal of Clinical Psychology, Vol. 50 No.1, pp.71-95.
Cook, W.W., Medley, D.M. (1954), "Proposed hostility and pharisaic-virtue scales for the MMPPI", Journal of Applied Psychology, Vol. 38 pp.414-8.
DiMatteo, M.R., Taranta, A., Friedman, H.S., Prince, L.M. (1980), "Predicting patient satisfaction from physicians' nonverbal communication skills", Medical Care, Vol. 18 No.4, pp.376-87.
Hall, J.A., Irish, J.T., Roter, D.L., Ehrlich, C.M., Miller, L.H. (1994), "Satisfaction, gender, and communication in medical visits", Medical Care, Vol. 32 No.12, pp.1216-31.
Jones, T.O., Sasser, W.E. Jr. (1995), "Why satisfied customers defect", Harvard Business Review, Vol. 736 pp.88-99.
Kirsner, R.S., Federman, D.G. (1997), "Patient satisfaction: quality of care from the patients' perspective", Archives of Dermatology, Vol. 133 No.11, pp.1427-31.
Kroenke, K., Jackson, J.L., Chamberlin, J. (1997), "Depressive and anxiety disorders in patients presenting with physical complaints: clinical predictors and outcome", American Journal of Medicine, Vol. 103 No.5, pp.339-47.
McKinley, J.C., Hathaway, S.R. (1943), "The identification and measurement of the psychoneuroses in medical practice", Journal of the American Medical Association, Vol. 122 pp.161-7.
Maruta, T., Colligan, R.C., Malinchoc, M., Offord, K.P. (2000), “Optimists vs pessimists: survival rate among medical patients over a 30-year period”, Mayo Clinic Proceedings, Vol. 75 No. 2, pp. 140-3 (“Erratum”, in Mayo Clinic Proceedings, (2000), Vol. 75 No. 3, p. 318), .
Maruta, T., Hamburgen, M.E., Jennings, C.A., Offord, K.P., Colligan, R.C., Frye, R.L., Malinchoc, M. (1993), "Keeping hostility in perspective: coronary heart disease and the Hostility Scale on the Minnesota Multiphasic Personality Inventory", Mayo Clinic Proceedings, Vol. 68 No.2, pp.109-14.
Probst, J.C., Greenhouse, D.L., Selassie, A.W. (1997), "Patient and physician satisfaction with an out-patient care visit", Journal of Family Practice, Vol. 45 No.5, pp.418-25.
Robbins, J.A., Bertakis, K.D., Helms, L.J., Azari, R., Callahan, E.J., Creten, D.A. (1993), "The influence of physician practice behaviors on patient satisfaction", Family Medicine, Vol. 25 No.1, pp.17-20.
Rubin, H.R., Gandek, B., Rogers, W.H., Kosinski, M., McHorney, C.A., Ware, J.E. Jr. (1993), "Patients' ratings of out-patient visits in different practice settings: results from the Medical Outcomes Study", Journal of the American Medical Association, Vol. 270 No.7, pp.835-40.
Seligman, M.E.P. (1989), "Explanatory style: predicting depression, achievement, and health", Brief Therapy Approaches to Treating Anxiety and Depression, Brunner/Mazel, New York, NY, .
Swenson, W.M., Osborne, D., Colligan, R.C. (1990), A User's Guide to the Mayo Clinic Computerized Scoring and Interpretative System for the Minnesota Multiphasic Personality Inventory (MMPI), 3rd ed., Mayo Foundation, Rochester, .
Corresponding author
Brian A. Costello can be contacted at: costello.brian@mayo.edu