| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Research Clinic for Functional Disorders and Psychosomatics (L.F., P.F., E.O., K.S.C., T.T.) and Research Unit of General Practice (F.O.), Aarhus University Hospital, Aarhus, Denmark; Unit of Psychology, Guy's, King's and St. Thomas', Guy's Campus, London, United Kingdom (J.W.).
Address correspondence and reprint requests to Lisbeth Frostholm, MSc, Research Clinic for Functional Disorders and Psychosomatics, Barthsgade 5.1, Aarhus University Hospital, DK-8200 Aarhus N, Denmark.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Methods: A randomized controlled follow-up study conducted in 28 general practices in Aarhus County, Denmark. Half of the physicians were randomized into an educational program on treatment of patients presenting with medically unexplained symptoms (somatization). One thousand seven hundred eighty-five general practice attenders presenting a new health problem completed questionnaires on illness perceptions, physical functioning, and mental distress before the consultation. After the consultation, a questionnaire including relational and communicative domains of patient satisfaction with the current consultation was completed. The physicians completed a questionnaire for each patient on diagnostics and prognostics. Predictors of patient satisfaction were determined by logistic regression.
Results: A large number of patient and practitioner variables predicted satisfaction in univariate logistic regression models. Results from a multivariate logistic model showed that the illness perceptions "uncertainty" (patient not knowing what is wrong) and "emotional representations" (the complaint making the patient feel worried, depressed, helpless, afraid, hopeless) predicted dissatisfaction at OR (CI) = 1.8 (1.3–2.4), p < .001 and OR (CI) = 1.5 (1–2.3), p = .03 respectively. Trained physicians were associated with dissatisfaction at OR (CI) 0.7 (0.5–1), p = .06 in the multivariate model. Furthermore, uncertain patients consulting a trained physician were less likely to be dissatisfied OR (CI) = 0.6 (0.3–1), p = .04.
Conclusions: A randomized controlled trial on the training of general practitioners' communication skills improved patient satisfaction. Illness perceptions predict satisfaction. In particular, patients feeling uncertain and negatively emotionally involved in their health problem were more inclined to being dissatisfied with the consultation.
Key Words: primary care • patient satisfaction • illness perceptions • communication skills • randomized controlled trial • medically unexplained symptoms • somatization
Abbreviations: IPQ = Illness Perception Questionnaire; PSCQ-7 = Patient Satisfaction Consultation Questionnaire; TERM = the extended reattribution model; SCL = Symptom Check List; SCL-SOM = SCL-90 somatization subscale.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Patients initiate most primary care consultations (5). This means that the content of the consultation will depend on the patient's view of his or her symptoms (6). Thus, one of the challenges for the consultation is the contrast between the physician's biomedical understanding and the patient's socioculturally embedded lay beliefs (7). Patients who present medically unexplained symptoms may feel that the physician does not take them seriously, as they often employ explanations of their illness that diverge substantially from that of the physician (6,8). However, interventions aimed at improving the communication skills of the physician toward patients with emotional distress and medically unexplained complaints improved function and reduced health care costs and psychopathology but did not improve overall satisfaction (9,10).
One of the more recent measures of the individual's understanding of his or her illness is the Illness Perception Questionnaire (IPQ) (11). The theory on which this questionnaire is based states that an individual experiencing symptoms will develop an understanding, based on preexisting knowledge and their current context, to make sense of the health threat. This understanding comprises the following distinct but interrelated themes: identity (the illness label and number of symptoms), cause, timeline, consequences, perceived control, and emotional representations. This understanding is crucial as it guides the individual in his or her attempts to cope with and solve his or her health problem. Previous research on individuals with a diagnosed disease or condition has provided evidence that a belief in serious consequences, a long timeline perspective and low perceived control, and a strong illness identity are associated with more negative outcomes (12–14). A recent randomized controlled trial with the purpose of changing illness perceptions after myocardial infarction found significant improvements on function and return to work (15). To our knowledge, no research has yet examined the association between patients' illness perception and their satisfaction with the consultation. This link could provide important information about preconsultation factors' impact on satisfaction with the consultation.
In agreement with previous research, we hypothesized that in particular patients with a strong illness identity (many symptoms), patients perceiving their health problem to be long lasting, and patients perceiving their health problem to have serious consequences would report more dissatisfaction with the consultation. Regarding causal attributions, previous research has been less clear, but we expected patients with higher scores on psychosocial, lifestyle, and stress attributions to be more dissatisfied. Furthermore, we expected patients with high levels of emotional and physical distress to be more dissatisfied. Finally, in accordance with previous research (16), we hypothesized that physician ratings concerning diagnostics and prognostics would predict patient satisfaction in such a way that patients rated as sicker would report more dissatisfaction.
The aim of the present study was (1) to validate a questionnaire on relational and communicative domains of patient satisfaction; (2) to examine if relational and communicative domains of patient satisfaction was associated with (a) patient variables (in particular illness perceptions) obtained before the consultation, (b) practitioner ratings obtained after the consultation, (c) and a randomized controlled trial on the training of physicians in general communication skills and how to treat patients presenting with poorly defined illness; (3) to investigate if illness perceptions and the randomized controlled trial were associated with satisfaction when controlling for emotional distress, patient and physician sociodemographics, and physician ratings in a multivariate model.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Eighteen- to 65-year-old consecutive patients consulting their physician with a new health problem during a 3-week period with scheduled successive inclusion of practices between March 3 and May 1, 2000, were invited by the physicians' secretaries to participate in a study on improving the physicians' understanding and treatment of their patients to avoid unnecessary examination, treatment, and suffering. Denmark has a tax-financed public health care system. Within this system, 98% of the population is registered with a specific physician (family practice setting), who acts as a gatekeeper to the secondary health care system. Each person has a personal registration number used for all contacts with the health care system. The physician is reimbursed for the consultation based on a code, which specifies the content of the consultation. The patients were included in the study if they were given the code signifying a general consultation. The study excluded patients of non-Scandinavian descent, patients not speaking or reading Danish, and patients who the secretaries or physicians assessed were too ill or demented to complete questionnaires. Moreover, health checks and administrative consultations (e.g., driver's license and other certificates, pregnancy controls, and vaccinations) were excluded from the study.
The intervention consisted of a multifaceted educational program on assessment, treatment, and management of somatization (the extended reattribution model, TERM-model), which rests on present theoretical and scientific knowledge about somatization and adopts a cognitive oriented approach (16). The intervention took place in the month preceding the study and consisted of a 2-day residential course (16 hours) followed by 3–4 evening courses (2 hours' duration each). Moreover, a booster meeting (2 hours) after 3 months and a facilitator visit to the physician's practice (one-half hour) after 6 months was included. The residential course was mainly based on microskills training in groups using video supervision but also comprised theoretical presentations covering the somatization concept, patients' illness beliefs, iatrogenic factors, and the etiology and epidemiology of somatization. The objectives of the program were to (1) instill knowledge about functional disorders, (2) introduce physicians to general interview techniques and specific treatment techniques for functional disorders, and (3) instill a change of attitude toward functional disorders. To avoid patients, skills, or training material from being transferred between participating physicians, the included practices were stratified according to number of coworking doctors and randomized together for intervention or control group by drawing lots. Control physicians received a flowchart on psychiatric assessment (anxiety, depression, somatoform disorders, and alcohol abuse) in primary care based on ICD-10 classification (17).
Questionnaires
The main outcome measure was a 7-item questionnaire (the Patient Satisfaction Consultation Questionnaire, PSCQ-7) designed to measure satisfaction with the relational and communicative aspects of the current consultation. The questionnaire was completed immediately after the consultation. The items concern empathic understanding, the "fit" of the diagnosis given, confidence in doctor, and overall satisfaction (Table 1). A sum score was calculated, and missing items (up to three) were averaged across completed items. The 20% most dissatisfied patients (n = 325) were identified as the focus for the logistic regression analysis.
|
The measure of illness perception was based on the IPQ (11). The IPQ was adapted for use in general practice. The items designed to measure emotional representations were altered to incorporate emotions such as helplessness and hopelessness. Moreover, two questions concerning patients' certainty of their health problem's nature were added. Principal components factor analysis showed a dimensional structure similar to the original IPQ, including the following dimensions: consequences, emotional representations, chronic timeline, personal control, timeline cyclical, and certainty (data available from authors). The various dimensions displayed from 10% to 15% missing values. Sum scores of the different dimensions were calculated. They were skewed with a low frequency of pessimistic illness perceptions and dichotomized to identify the 20% with the most pessimistic illness perceptions as cases. In addition, a number of causal illness attributions were included in the analysis (information available from authors).
Regarding screening for emotional distress, the following scales were used: an 8-item version of the Symptom Check List (SCL-8) (18) aimed at detecting depressive and anxiety symptoms, the 7-item Whiteley index screening for illness worry (19), the 12-item SCL-90 somatization subscale (SCL-SOM) (20) screening for common physical symptoms, and the four-item CAGE screening for alcohol abuse were completed (21; Table 2). All the scales were dichotomized in the present analysis identifying the highest quintile as cases. The IPQ and the screening measures were completed before the consultation.
|
The physicians completed a questionnaire for each patient immediately after the consultation (Table 3). Those questions were tailored and initially tested in a pilot study by a number of experienced Danish physicians to acquire information on diagnostics, prognostics, estimate of the doctor-patient relationship, and, in particular, functional somatic symptoms. The following physician ratings were included: physician's familiarity with patient (question 1), chronicity (question 2 and 3), nature of presenting health problem (question 4), physician's illness evaluations (question 5 to 7), physician's evaluation of the consultation (question 8), functional symptoms and illness worry (question 9 to 11), physician's evaluation of the patient's treatment within the last year (question 12).
|
Data Analysis
Regarding the satisfaction items, varimax-rotated principal components factor analysis was performed to assess latent domains, and the selection criterion was eigenvalues exceeding 1.0. Cronbach's 
was obtained to evaluate the internal reliability of the scale. Univariate predictors of patient satisfaction were determined in logistic regressions controlled for age and gender of the patient.
Multivariate Model
To examine if illness perceptions and the randomized controlled trial predicted satisfaction in a more comprehensive model, we performed a multivariate analysis. According to recommendations by Harrell (22) and Steyerberg et al. (23), the variable selection criterion should be based on a p level between .25 and .5 to include variables that may prove important in a multivariate model. Furthermore, to avoid overfitting, there should be a minimum of 15 cases (dissatisfied patients) for each independent variable (22,24).
Previous studies have suggested that illness perceptions are stronger predictors of outcome than causal attributions, wherefore we decided to omit the causal attributions with the exception of 1 variable that indicated the number of causal attributions stated by each patient (0–23). Emotional distress variables and patient and physician sociodemographics should also, as stated by earlier research, be included in the model. Furthermore, we hypothesized that physician ratings would have an effect on patient satisfaction. However, due to our sample size and missing values, we were not able to include all those variables and at the same time fulfill the abovementioned recommendations. Therefore, we did as follows: (1) To increase the power of the final model, we imputed the continuous illness perception sum scores based on best subsets regression with the other illness perceptions, emotional distress scores, and sociodemographics of the patients as covariates. We dichotomized the imputed sum scores to identify the 20% with the most pessimistic illness perceptions as cases and found almost no change in cut point between nonimputed and imputed data. (2) We performed a principal components analysis of the physician ratings, including the three first principal components explaining 67% of the variance in the physician ratings (22). We imputed those three principal components based on the scores on completed physician ratings, again using best subsets regression.
The model was adjusted for the effect of clustering at physician level to account for interdependence between the satisfaction scores of patients consulting the same physician. To validate the model, we used the Hosmer-Lemeshow fit statistic (25) for discrimination of the area under the ROC curve (22,23) and a heuristic shrinkage estimate (26). Statistical analysis was performed using SPSS and STATA.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
|
Among 2424 patients assessed for eligibility, 227 met exclusion criteria, 274 patients declined, and 138 could not participate for other reasons; for example, time pressure in clinic and patients not carrying reading glasses (Figure 1). Each patient was included only once; 1785 patients joined the study; 1512 (84.7%) of the included patients completed the satisfaction questionnaire (PSCQ-7) after the consultation. Noncompleters (n = 273) were significantly younger (mean age 35.8 versus 39.3), Mann Whitney's U Z = 4.1, p < .001, but did not differ in gender, 
2 (1) = 2.8, p = .09, from the 1512 patients completing the patient satisfaction questionnaire. Satisfied and dissatisfied patients varied on all explored sociodemographic variables except gender (Table 2).
Main Results
Principal components factor analysis (Table 1) of the satisfaction items demonstrated factor loadings from 0.68 to 0.88, with 67% of the variance explained in a 1-factor model with only 1 eigenvalue exceeding 1. The sum score displayed a ceiling effect, with most patients reporting high levels of satisfaction. Also, the questionnaire demonstrated good internal reliability, measured with a Cronbach's of 0.92.
Results from the individual logistic regressions with satisfaction (20% most dissatisfied/80% most satisfied) as the dependent variable showed that older patients were more likely to be satisfied, OR (CI) for a patient 50 years old compared with a patient 20 years old = 0.4 (0.3–0.7) (Table 3). Patients consulting a female physician or a more experienced physician were also more likely to be satisfied OR (CI) = 0.7 (0.6–0.9) for both variables. Patients consulting a trained physician reported higher levels of satisfaction, OR (CI) = 0.7 (0.6–0.9).
Regarding the patient predictors, patients with high illness worry and high symptomatology were more likely to be among the dissatisfied patients OR (CI) = 2.4 (1.8–3.2) and 1.7 (1.3–2.3), just as patients reporting higher levels of emotional distress OR (CI) = 2.2 (1.6–2.9). The illness perceptions associated with patient dissatisfaction were negative emotional representations OR (CI) = 2.4 (1.8–3.2), high levels of uncertainty OR (CI) = 2.1 (1.6–2.7), perceived negative consequences OR (CI) = 2 (1.5–2.7), and long timeline perspective OR (CI) = 1.6 (1.2–2.2). Psychosocial attributions and number of causal attributions were associated to patient dissatisfaction at OR (CI) = 1.6 (1.2–2.2), and OR (CI) = 1.07 (1.03–1.11), respectively, whereas lifestyle attributions had an OR (CI) of 1.4 (1–1.9).
Most of the physician ratings were associated with satisfaction (Table 3). Even when we adjusted the p value of the univariate analyses for multiple analyses using Bonferroni's rather conservative estimate (0.05/36 = 0.0014), most illness perceptions remained significant. However, the effect of the randomized controlled trial turned nonsignificant.
Multivariate Model
The multivariate model included 7 illness perceptions, the intervention, patient's age, gender, schooling, emotional distress variables (SCL-SOM, SCL-8, Whiteley-7), and physician's age, gender, seniority, and the three principal components of the physician ratings, all in all 20 variables (Table 4).
|
Two illness perceptions predicted dissatisfaction at p < .05: high scores on "uncertainty," OR (CI) = 1.8 (1.3–2.4), p < .001, and "emotional representations," OR (CI) = 1.5 (1–2.3), p = .03. No emotional distress variables were significant in the model. The trained physicians were associated with more satisfied patients at p = .06, OR (CI) 0.7 (0.5–1). Two of the principal components of the physician variables were associated with satisfaction in the analyses. Both patient and physician sociodemographics had significant or near-significant impact on satisfaction.
One thousand four hundred eighty-six patients and 322 cases (dissatisfied patients) were included in the model, allowing 322/15 = 21 variables in the model. The model without imputed data contained only 857 patients and 186 cases. When the model was estimated on the nonimputed data, we, to a large extent, found similar odds ratios. However, there was a tendency toward smaller odds ratios and narrower confidence intervals for imputed data. All variables with a p value less than .5, except causal attributions, from the univariate analyses were included in the model. However, the physician ratings were included in a condensed form. The goodness-of-fit of the final model had a Hosmer-Lemeshow fit statistic 2 (8) = 11.16, p = .19. The model discrimination given as the area under the ROC curve was 0.72, and the model had a heuristic shrinkage estimate = 0.87.
With regard to participation in the educational program, we tested for interactions with the two strongest patient predictors (uncertainty and emotional representations) and found a statistically significant interaction between the trained physicians and uncertainty; the intervention was insignificant, OR (CI) = 0.9 (0.5–1.4), p = .50; uncertainty rose to OR (CI) = 2.3 (1.7–3.2), p < .001; and the odds ratio for the interaction was OR (CI) = 0.6 (0.3–1), p = .04. Likelihood ratio test of the interaction was 2 (1) = 4.23, p = .04. This means that the patients reporting most uncertainty regarding the nature of their health problem were more likely to be satisfied when consulting a trained as opposed to an untrained physician.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
The satisfaction questionnaire (PSCQ-7) deals with relational and communicative aspects of the consultation only and cannot be generalized to other aspects of care. It appears to have an acceptable internal validity based on factor analysis and Cronbach's , but a new study is needed to evaluate its external validity more thoroughly. We dichotomized the sum score of the PSCQ-7 primarily because the variable is very skewed. We chose as cut point the lowest quintile. Again, a new study is needed to gain information of an appropriate cut point. At this point, we cannot make any conclusions regarding the relative contribution to dissatisfaction among the 20%. Thus, we are aware that the dichotomization entails loss of information
A major limitation of this study is the relatively large proportion of missing values concerning satisfaction, as well as illness perceptions. Patient difficulties in translating still vague perceptions of a new complaint into questionnaire statements may have increased the number of incomplete data. A qualitative approach may in this aspect have elicited a higher and more valid response. We made imputations of the illness perceptions to increase the power of the multivariate model. This, of course, entails the risk of introducing bias. However, the mean sum scores of the illness perceptions only changed slightly after imputing data, and when performing the multivariate model without computed data, we found similar odds ratios. Furthermore, illness perceptions were correlated, and they were all completed in the waiting room. We decided not to impute satisfaction scores, because the data on which to base such imputations were unclear.
One should exercise caution when drawing conclusions from the multivariate model due to colinearity between independent variables. Furthermore, the heuristic shrinkage estimate indicates that overfitting, i.e., that the model has too many parameters for the amount of information in the data leading to an overestimation of the worth of the model, is present (22). An evaluation study would provide a more pragmatic estimate of the degree of overfitting. The area under the ROC curve of the model indicates an acceptable but not impressive prognostic value of the multivariate model (25).
With respect to the intervention, we randomized the participating physicians at clinic level to avoid crosstalk between intervention and control physicians. However, statistically we adjusted for clustering at physician level as our primary intention was to implement change in the individual physician, and differences in patient satisfaction scores were larger between physicians than between clinics. The p value of the intervention was very influenced by the adjustment for clustering. The p value was .01 without cluster and .06 when cluster was added to the analysis.
In line with other studies (27,28), we found, in the univariate analyses, that illness worry predicted dissatisfaction. Reporting a high number of causal attributions may be associated with illness worry too; those patients may ruminate about what causes their health problem and thus be less inclined to accept the physician's explanation and feel less reassured after the consultation.
A recurrent finding is that poor mental health is associated with dissatisfaction (29). However, in this study the emotional representations were stronger predictors of dissatisfaction than the well-established measures of mental distress. Thus, it was the patients' specific emotional reactions to their illness, rather than their more general level of psychological distress, which were more strongly associated with their dissatisfaction with the consultation. Likewise, the association between psychosocial attributions and dissatisfaction probably rests on the link between poorer mental health and dissatisfaction. However, emotional representations and measures of emotional distress were correlated.
High illness identity (number of reported symptoms) and perceiving the health problem to have serious consequences were associated with dissatisfaction. However, neither of these in most studies (11–15) very strong predictors of worse outcome was significant in the final model. This may be due to the patients' emotional state and particular feelings concerning the current health problem being more decisive in how they evaluate the here-and-now relation, as the PSCQ-7 was completed immediately after the consultation. Furthermore, the fact that the study population was limited to patients presenting a new complaint may have contributed to the importance of emotions and uncertainty.
A number of practitioner ratings with a seemingly heterogeneous content such as illness worry, many physical symptoms at first contact, insufficient treatment offer, and the physician's illness evaluations and the physician's satisfaction with the consultation were associated with patient dissatisfaction. They may be an indicator of the complexity or severity of the patients' illness history both retrospectively and prospectively and may help the physicians pinpoint cases in need of additional management and care (30,31).
According to previous reports, patients not only prefer a patient-centered approach with greater focus on communication, partnership, and health promotion compared with a narrow biomedical approach (32) but also report greater satisfaction after consulting physicians using psychosocial and affective communication (30–32). However, Morriss et al. (10) did not find differences in overall patient satisfaction between the intervention and control group of their treatment program for somatized mental disorders. This may be due to a less sensitive measure of satisfaction or the fact that they only included patients with somatized mental disorders. Furthermore, the educational program we used included most of the elements used by Morriss et al. (10), and several other elements focused on patients presenting with diffuse symptoms of uncertain genesis. The physicians were especially trained in addressing the patients' illness beliefs and in modifying these to empower the patients to a new understanding of their health problem thus reducing uncertainty and illness worry. And we did find that patients feeling uncertain about the nature of their health problem were more likely to be satisfied when they consulted a trained physician.
Balint (33) in 1965 pointed out the importance of uncertainty in the consultation process. He identified two stages in the communication between the physician and the patient: the "unorganized" stage, where the patient presents his or her vague and uncertain worries, and the "organized" stage, where the illness gets a name and the roles of the doctor and patient are settled. Our findings support the importance Balint attached to uncertainty as we demonstrate that patients being more uncertain and emotionally involved in their complaint are more likely to be dissatisfied. Thus, it is essential that the physicians be able to address their own, as well as the patients', feeling of uncertainty. More important, the physicians need tools to manage such consultations as they are not likely to acquire those skills through traditional biomedical training. Our findings suggest that supplementary education, in this context represented by the TERM-model, may be a usable tool. Future communication skills training for physicians should continue to focus on patients' illness perceptions as they influence the quality of the doctor-patient communication and furthermore offer an obvious starting point for a biopsychosocial approach. However, more research is required into if and how such training can lead to changes in patients' illness perceptions.
We thank the physicians, the secretaries, and the patients who participated in this study.
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
This study is part of the FIP-study (Functional Illness in Primary Care), which is a joint interdisciplinary project with the participation of the Research Clinic for Functional Disorders and Psychosomatics, Aarhus University Hospital, the Research Unit for General Practice, University of Aarhus, the Department of Ethnography and Social Anthropology, University of Aarhus, the Department of Psychology, University of Aarhus.
DOI:10.1097/01.psy.0000188403.94327.5b
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  P. R. Wysong and E. Driver Patients' Perceptions of Nurses' Skill Crit. Care Nurse, August 1, 2009; 29(4): 24 - 37. [Full Text] [PDF]
|
|
|
|
 |
|
 E. Guthrie Medically unexplained symptoms in primary care Adv. Psychiatr. Treat., November 1, 2008; 14(6): 432 - 440. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Palmer, M Calnan, D Wainwright, C O'Neill, A Winterbottom, C Watkins, J Poole, and D Coggon Upper limb pain in primary care: health beliefs, somatic distress, consulting and patient satisfaction Fam. Pract., December 1, 2006; 23(6): 609 - 617. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
