| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Department of Psychiatry, Massachusetts General Hospital/Harvard Medical School, Boston, Massachusetts (J.G.); Department of Psychology, University of Massachusetts, Amherst, Massachusetts (R.H.).
Address correspondence and reprint requests to Joseph Greer, PhD, Massachusetts General Hospital, WACC 812, 15 Parkman Street, Boston, MA 02114. E-mail: jgreer2{at}partners.org
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Methods: Primary care patients (n = 175) and their physicians rated the extent to which patients' presenting symptoms represented a medical versus a psychological problem. Patients also completed surveys regarding their demographic characteristics, level of psychological distress, health status, recent stress, history of mental health treatment, and reason for office visit.
Results: Physicians and patients agreed on the etiology of symptoms in approximately 59% of the cases. Patient sex, history of mental health treatment, and reason for office visit significantly predicted agreement on symptom etiology. Physicians perceived patients with whom they disagreed about symptom etiology as less cooperative.
Conclusion: Physician-patient agreement on symptom etiology is low and relates to several patient psychosocial and demographic factors. Identifying the variables associated with disagreement may help to improve communication and patient outcomes in primary care.
Key Words: primary care • physician-patient agreement or concordance • psychological distress
Abbreviations: SCL-90-R = Symptom Checklist 90-Revised; GSI = Global Severity Index; CEQ = Clinical Encounter Questionnaire.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Concordance between physicians and patients on symptoms is an important marker of effective communication (10), retention in outpatient care (11), and resolution of patient problems (12). For example, Starfield and colleagues (13) found that practitioner-patient agreement on the types of medical problems requiring follow-up care was associated with higher patient expectations for improvement and better outcomes for reported health concerns. By contrast, Gabbay and colleagues (14) found that mutual understanding of presenting complaints between general practitioners and patients with depression was not predictive of clinical outcomes. The authors argued that examining provider-patient concordance on the question of whether the primary problem has a psychological component might prove more beneficial.
Researchers have identified several correlates of clinical agreement between physicians and patients. For example, in attempting to determine patients' principal problems, physician-patient concordance tends to be lower when the main problem is psychosocial in nature (15) or chronic and less intense (16). Additionally, providers and patients more often disagree about the main reason for the medical visit when patients are female or report multiple complaints (17). Finally, Helman (18) examined physician-patient concordance on the etiology of symptoms in primary care, noting significant disagreement about perceptions of the diagnostic label given to the patient's condition, psychological factors in etiology, and ways that social relationships affect and are affected by the patient's condition.
Understanding the factors that relate to physicians' and patients' beliefs about presenting symptoms may improve communication and management of psychological distress in primary care. The goals of the present study were as follows: (1) to explore the level of agreement between physicians and patients regarding their perceptions of psychological factors in symptom etiology, (2) to identify predictors of physician-patient agreement, and (3) to determine whether agreement on symptom etiology is associated with increased cooperation in the physician-patient relationship.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
2 = 3.25, p = .52) but more likely to be female (2 = 9.37, p < .01).
|
Additionally, the five primary care physicians (four male, one female) of the medical practice were approached to take part in the study; all providers agreed to participate and complete brief assessments of their respective patients following the clinical encounter. The physicians, who ranged in age from 33 to 46 years, were all board-certified in internal medicine and provided services to approximately 8,000 patients.
Measures
Demographic and Psychosocial History Questionnaire
This initial questionnaire included questions about sex, age, ethnic background, and education. In addition, respondents provided global ratings of their general health (from 1 = "poor" to 7 = "excellent") and the extent to which they were under stress during the past week (from 1 = "no stress" to 7 = "severe stress"). Finally, respondents indicated whether they had ever taken medication for psychiatric or emotional reasons (as "Yes" or "No") and whether they had ever seen a therapist, counselor, minister, or other professional for emotional problems (as "Yes" or "No").
Reason for Office Visit
Patients were asked to specify the symptoms for which they were visiting their doctor that day. The format of this question was open ended, with patients writing their responses in provided blank spaces.
Symptom Checklist-90-R (SCL-90-R)
The SCL-90-R is a self-report instrument that assesses general psychological distress. Using a 5-point scale (from 0 = "not at all" to 4 = "extremely"), respondents indicate the extent to which they have been distressed by 90 distinct symptoms during the past week. The instrument consists of nine subscales that are averaged into three global indices, the most commonly used of which is the Global Severity Index (GSI). Scaled T-scores of the GSI were used for all statistical analyses; higher aggregate scores indicate greater distress. The psychometric properties and use of the SCL-90-R in general medical settings have been well established (19–22).
Clinical Encounter Questionnaire (CEQ)
Developed by the authors, the CEQ asks physicians and patients to rate the extent to which they believe the presenting symptom(s) represent a psychological problem (i.e., CEQ/Attribution scale from 1 = "completely medical" to 7 = "completely psychological"). This item possesses adequate face validity and has been shown in a previous study to relate to physician recognition and treatment of psychological distress in primary care (23). In addition, physicians and patients each provide a global rating of how they would characterize one another on their respective versions of the questionnaire (i.e., CEQ/Cooperative scale from 1 = "extremely cooperative" to 7 = "extremely difficult").
Procedure
The institutional review board of the Department of Psychology at the University of Massachusetts-Amherst approved the study before its initiation, and confidentiality procedures were strictly maintained. Data were collected from September 15, 1999, to January 15, 2000. Questionnaires were distributed to the physicians and their patients at the medical office during regular business hours. As patients arrived at the practice, a researcher approached them consecutively, explained the purpose of the study, assessed eligibility, and requested participation. Eligible patients willing to enroll were asked to sign a consent form and assured that participation was voluntary and confidential. The participants then received questionnaires about their demographic information, psychosocial history, and reason for office, as well as the SCL-90-R in the waiting room, before meeting with their physicians. After the medical visit, patients and physicians completed their respective versions of the CEQ.
Statistical Methods
Using standard statistical software for the social sciences (i.e., SYSTAT), the means, standard deviations, and frequencies for the study variables were calculated. All categoric variables were dummy coded, and patient ethnicity was categorized as a dichotomous variable (i.e., ethnic minority or European American) due to the relatively small number of participants in each ethnic group. Cohen's 
was calculated to test reliability for the coding of patients' reasons for office visits, as well as to assess agreement between physicians and patients regarding the etiology of presenting symptoms. Correlation and simultaneous linear regression analyses were conducted to identify predictors of physician-patient agreement on symptom etiology. An 
level of 0.05 was used to determine significance for all statistical tests.
Patients were initially grouped by attending physician to conduct analyses of variance and 2 tests on the predictor and outcome variables. Comparing patients of individual physicians, no significant differences were observed with respect to patients' level of psychological distress, health status, recent stress, history of psychotropic medication use, history of counseling or therapy, reason for office visit, or level of agreement with the medical providers regarding the etiology of the presenting symptoms. Consequently, a physician variable was not added in the subsequent statistical analyses as a control factor.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Reason for Office Visit
Patients sought care from physicians for a variety of reasons, including well care (e.g., checkup), respiratory problems (e.g., sore throat, sinus problems, asthma), follow-up for chronic medical conditions (e.g., hypertension, hyperlipidemia, diabetes), musculoskeletal pain (e.g., back, leg, shoulder pain), gastrointestinal/genitourinary difficulties (e.g., urinary tract infections, abdominal discomfort), neurologic problems (e.g., headache, dizziness), psychological distress (e.g., depression, anxiety), and other complaints. Table 2 provides a list of the types of clinical conditions for which patients visited the doctors' office.
|
Primary care clinicians often utilize a strategy of "ruling out" medical/organic causes when evaluating patients' presenting symptoms for possible diagnosis of depression or other psychological problem (9). For the present study, since subgroup analyses according to patients' reasons for medical visit were not possible due to the sample size, the presenting symptoms were categorized based on the "ruling out" strategy, such that symptoms with a likely organic cause (e.g., sore throat or blood pressure check) were coded as medical, whereas those presenting symptoms for which there was less clearly an organic cause (e.g., fatigue, headache, or dizziness) were coded as potentially psychological. Patients who offered multiple reasons for the office visit were coded as potentially psychological if any one of their presenting symptoms could be so categorized. Two doctoral-level clinicians independently rated the patients' presenting symptoms using this coding scheme, yielding a Cohen's of 0.93. Discrepancies between the coders were discussed and resolved for the final coding of the reasons for office visit used in subsequent analyses. Of the reasons for office visits reported by patients, 63.2% were coded as medical and 36.8% as potentially psychological in nature.
Physician-Patient Agreement on Symptom Etiology
Physicians and patients both reported the extent to which they believed the presenting symptoms represented a psychological problem on their respective versions of the CEQ following the medical visit. These CEQ/Attribution ratings were first analyzed using the Pearson correlation coefficient to ascertain a global measure of agreement. The results of this analysis suggest low correspondence between physicians and patients (r = 0.31, n = 175, p < .001).
The 7-point CEQ/Attribution ratings were then collapsed into three categories (i.e., ratings of 1–2, 3–5, and 6–7), permitting the calculation of Cohen's coefficient to examine agreement between physician and patient perceptions of the presenting symptoms. Using a 2 x 3 matrix, physicians and patients agreed on the etiology of the presenting symptoms in 58.3% (n = 102) of the cases. Physicians attributed the symptoms more psychologically in 24.6% (n = 43) of the cases, whereas patients believed the presenting symptoms were more psychological in nature than physicians in 17.1% (n = 30) of the cases, Cohen's = 0.16.
To assess level of agreement on symptom etiology most directly, the patients' CEQ/Attribution ratings were subtracted from the physicians'. The average agreement score was slightly greater than zero (M = 0.31, SD = 1.98, range = –6 to +6). Based on these scale-point deviations, 32.6% of physicians and patients had perfect agreement on the etiology of the presenting symptoms (i.e., difference score of 0), with an additional 26.3% deviating by only 1 point (i.e., difference score of –1 or +1), yielding an estimate of overall agreement of 58.9% (n = 103).
Finally, in addition to the breakdown of reasons for office visit, Table 2 provides a rough estimate of agreement on symptom etiology between physicians and patients for each symptom category. These mean scores reveal that the top two clinical conditions for which physicians were more likely than patients to view the presenting problems as psychological in nature were neurologic and GI/GU symptoms. In contrast, among patients seeking well care, the average difference score for the ratings of symptom etiology was 0.00 (SD = 2.35).
Predictors of Agreement on Symptom Etiology
For the following correlation and linear regression analyses, the difference scores between physicians' and patients' CEQ/Attribution ratings were used as the primary measure of agreement on symptom etiology. Pearson and point-biserial correlations were first calculated to examine associations among all of the study variables (see Table 3). According to these analyses, physician-patient agreement on symptom etiology was significantly associated with patient sex, health status, and reasons for the office visit.
|
To identify independent predictors of agreement between physicians and patients, a linear regression analysis was performed. For this analysis, the CEQ/Attribution difference scores were regressed on the following variables: patient demographic characteristics, level of psychological distress (i.e., SCL-90-R ratings), general health, recent stress, history of taking psychotropic medication or seeing a professional for counseling, and reason for office visit. These specific variables, which were simultaneously entered into the regression equation, have been shown empirically to relate to the interpretation of symptoms and recognition of psychological distress by providers in primary care settings (23–25). The results of the regression analysis (see Table 4) demonstrated that physicians were more likely to disagree with patients and believe that the presenting symptoms were more psychological in nature when the patient was female, indicated having taken medication for emotional reasons in the past, or presented with symptoms with no clear medical cause. In addition, patients who had seen a therapist, counselor, or some other professional for emotional problems in the past were more likely than physicians to rate the presenting symptoms as psychological in nature.
|
Presenting Symptoms and the Physician-Patient Relationship
Physicians and patients each rated the extent to which they perceived one another as cooperative versus difficult (i.e., CEQ/Cooperative). Using the reason for the office visit as a grouping variable (i.e., medical versus potentially psychological), an independent-samples t test was conducted on the CEQ/Cooperative scores to explore any associations between the type of presenting concern and perceptions of the physician-patient relationship. Although no difference was found in patients' ratings of their physicians according to the reason for office visit, a marginally significant trend emerged from the physicians' perspective. Specifically, the doctors rated patients who presented with potentially psychological symptoms as more difficult than patients with medically explained symptoms (medical M = 2.30, SD = 1.40; potentially psychological M = 2.69, SD = 1.45; t = –1.75, p = .08). Moreover, although statistical testing by each type of clinical condition was not possible due to the small sample size, the results in Table 2 further demonstrate that patients who presented primarily with psychological concerns were perceived as most difficult by their providers (CEQ/Cooperative M = 3.67, SD = 1.41), compared with patients with other clinical complaints.
To examine the extent to which agreement on symptom etiology relates to perceptions of the physician-patient relationship, zero-order correlation coefficients were calculated between the CEQ/Attribution difference scores and physician/patient ratings of one another as cooperative versus difficult (i.e., CEQ/Cooperative). These analyses revealed no significant linear relationship between patients' ratings of their physicians as cooperative and the CEQ/Attribution difference scores (r = –0.06, n = 174, p = .41). However, physicians were more likely to characterize the patient as difficult when they disagreed with the patient and believed that the presenting symptoms were more psychological in nature (r = 0.30, n = 174, p < .001).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Several patient factors seem to relate to disagreement on symptom etiology between physicians and patients. Specifically, physicians were more likely to disagree with patients and rate the presenting symptoms as representing a psychological problem when the patients were female, indicated having a history of psychotropic medication use, or presented with symptoms with no clear medical cause. In addition, patients with a history of mental health counseling were more likely to disagree with physicians and rate the presenting symptoms as representing a psychological problem. The reverse direction of the latter finding may be due to the type of information providers elicit from patients. For example, it is conceivable that primary care physicians are likely more aware of their patients' psychotropic medication use but not their history of participation in mental health counseling. These findings highlight the demographic and psychological factors that may influence the ways in which physicians and patients understand the etiology of symptoms.
Conceptualizing presenting symptoms as medical versus psychological in nature may seem to perpetuate the oversimplified view of mind-body dualism. Nevertheless, physicians continue to use a strategy whereby they typically shift to consideration of a mental health problem only when no medical diagnosis can be made, especially if the physician is less experienced or if the patient is unfamiliar (9). This diagnostic approach may in part account for the low rates of detection of patient psychiatric morbidity by primary care providers (1–3). Additionally, patients who consult medical professionals regarding medically unexplained symptoms often resist the notion that their symptoms may be psychologically related, not only because the choice of provider demonstrates patients' likely expectation of a biologic explanation but also because of the stigma associated with mental illness.
Even in cases when physicians accurately detect patient psychological distress, patients may still be unwilling to accept such explanations, potentially leading to difficulties in the physician-patient relationship. Not surprisingly, patients considered difficult by their physicians are more likely to have a depressive or anxiety disorder (26). Findings of the present study indeed confirm that physicians view patients with potentially psychologically related symptoms as more difficult, and the failure of patients and physicians to agree on the etiology of presenting symptoms is associated with physician perceptions of patients as less cooperative.
Limitations of the Study
Although this study is one of the first of its kind to identify independent predictors of agreement between primary care physicians and patients regarding the etiology of presenting symptoms, the small number of physicians in the sample and low participation rate among patients (i.e., 68%) may limit its generalizability. Additionally, further characterization of the participants, such as length of care, illness severity, and existing psychiatric diagnoses, may have revealed other salient predictors of agreement.
A final concern regarding the study pertains to the manner of data collection. Since patients completed surveys regarding their psychological health before meeting with physicians, they may have considered a possible association between their presenting symptoms and psychosocial distress. As a result, patients may have raised such concerns with their providers during the clinical encounter, possibly affecting rates of agreement. However, this possibility should have led to greater concordance than typically occurs in usual practice, making our results a conservative estimate of the prevalence of patient-physician disagreement.
Future Directions and Conclusions
Recent studies have shown that physicians employ different patterns of communication during the clinical encounter (27,28). For example, patient-centered communication has been shown to be associated with patient perceptions of finding "common ground" with providers, improved health status, and decreased utilization of medical services (28). In the future, researchers may wish to explore which styles of communication and provider behaviors facilitate physician-patient concordance on symptom etiology and improve management of psychological distress.
Agreement between physicians and patients about the etiology of presenting symptoms does not ensure that medical providers will accurately recognize and treat patient psychological distress. Yet understanding the patient factors that predict disagreement, such as patient sex, history of mental health treatment, and reason for office visit, may sensitize providers to the ways that perceptions of symptom etiology are formed.
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
Received for publication January 2, 2005; revision received September 19, 2005.
DOI:10.1097/01.psy.0000203239.74461.db
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
