Family Practice Advance Access originally published online on July 26, 2007
Family Practice 2007 24(5):498-503; doi:10.1093/fampra/cmm043
Physician–patient questionnaire to assess physician–patient agreement at the consultation
a Home Care Unit, Primary Health Care, Göteborg
b University of Göteborg
c Research and Development Unit in Primary Health Care, Southern Elfsborg County, Sweden
Correspondence to G Carlsson Ahlén, Hemsjukvård NO, Box 47107, 402 58 Göteborg, Sweden; Email: gerd.carlsson.ahlen{at}vgregion.se
Received 2 January 2007; Revised 19 June 2007; Accepted 21 June 2007.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodResultsDiscussionDeclarationReferences |
|---|
Objective. The primary aim of this study was to validate an instrument of physician–patient agreement in the consultation. A secondary aim was to assess this agreement.
Method. The setting was a county in the southwest of Sweden with a cross-sectional survey of primary care patients and physicians using separate coded questionnaires. Forty-six physicians and 316 patients aged 16 or more with a new complaint lasting 1 week or more. Thirteen items were evaluated and index of proportional agreement for the dichotomized answers agree (Ppos) and disagree (Pneg) was calculated.
Results. In 10 of the 13 items, a high level of agreement between physician and patient was seen.
Discussion. Index of proportional agreement was useful in finding statements in a questionnaire on agreement for both physicians and patients that could be used for educational purposes and as a check-up for the GP in daily practice.
Keywords. Attitude of health personnel, communication, family practice, humans, office visits, patient participation, patient satisfaction, physician–patient relations, process assessment (health care), primary health care, quality of health care, questionnaires, self-evaluation programs.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionDeclarationReferences |
|---|
The consultation and how it is carried out is a key item on the GP's agenda. Aspects of the consultation subjected to study have usually been patient satisfaction, patient enablement, physician–patient interaction and physician–patient agreement.
By investigating patient satisfaction and patient enablement, important information on the consultation's quality has been highlighted.1 Research in patient satisfaction may be useful in assessing consultation quality.2,3 However, this research is difficult to evaluate due to differing study methodology and stringency.4
Physician–patient interaction has been measured by estimating differences of opinion between physician and patient. Martin5 pointed out that after consultations, the physician's perceptions differed from the patient's concerning illness level, cause and nature of the problem and the content of the consultation. The study was based on differences at a group level, whereas agreement at an individual level was not studied.
Greater physician–patient agreement on consultations concerning sufficiency of allotted time, urgency, number of problems presented and quality of communication was associated with higher patient satisfaction6. Physician patient agreement concerning problem identification was lower for psychosocial problems7. Starfield8 found that agreement on problems requiring follow-up was associated with a better outcome. Lack of agreement between the patient's and the physician's diagnosis was associated with a ‘negative medical consultation’.9 Thus, further knowledge of agreement between physician and patient might clarify the content of the consultation and enhance outcome. Validated questionnaires such as Medical Interview Satisfaction Scale (MISS), Consultation Satisfaction Questionnaire (CSQ), Patient Experience Questionnaire and Patient Enablement Instrument exclusively address the patient's perspective. No questionnaire addressing physician–patient agreement has, thus far, been validated.
The primary aim of this study was to validate an instrument of physician–patient agreement in the consultation. A secondary aim was to assess this agreement.
|
Method |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionDeclarationReferences |
|---|
The study was carried out from 2001 to 2003. Primary health care physicians in the municipalities of Alingsås and Lerum were asked to participate.
Selection of consultations
Each physician during a 6-week period asked eight consecutive patients fitting the inclusion criteria if they wanted to participate in the study. Only patients with new complaints of at least 1 week's duration, aged 
16 years and without need of an interpreter were asked to participate. Patients with dementia, psychosis or drug abuse were not asked to participate. When a physician failed to include eight consecutive patients, another physician was asked to assist with including the remaining patients.
Procedure
When the physician considered a patient fulfilling the inclusion criteria, the patient, directly after the consultation, was asked to participate in the study. Patients were informed of confidentiality and that participation was voluntary. The patient was asked to complete the questionnaire in the waiting room directly after the consultation and leave it in an envelope at the reception desk. If they did not wish to participate, they could leave in the incomplete questionnaire. Written information on the confidentiality and voluntary nature of participation was given to the patient together with the questionnaire.
In the patient's absence, the physician completed a similar questionnaire. Questionnaires were coded whereby the physician's questionnaire could be matched with the corresponding patient questionnaire.
Questionnaire
Based on studies on aspects important to the patient in the consultation,10–13 the new questionnaire ‘Physician–Patient Questionnaire’ (PPQ) with 13 identical items was constructed; one for the physician and one for the patient. Items were constructed as statements with degree of agreement recorded on a five-point Likert scale. Global aspects of the consultation were represented by Items 2, 4, 11, 12 and 13. The remaining items represented more specific aspects. Many of the items in our questionnaires are similar to the ones in MISS or CSQ. The content validity was ensured through discussion with experienced GPs and researchers in rephrasing the items several times before the final version. In addition to the 13 items, physicians were also questioned on age, gender, years in practice and diagnoses. Patients were asked about age, gender and duration of the problem.
Statistical methods
The five-grade scale was dichotomized as ‘agree’ (response 4 to 5) or ‘disagree’ (response 1 to 3) with the statement. Descriptive data on the proportion of positive responses (response 4 to 5) were presented for physicians and patients separately. Using this dichotomized scale, each item was then analysed according to the level of agreement between physician and patient with the omnibus indexes index of validity (Iv) and kappa coefficient (
). Furthermore, the more specific indexes of proportional agreement for the two responses ‘agree’ (Ppos) and ‘disagree’ (Pneg) were calculated.14,15Ppos is not an omnibus index since it only includes the positive responses. Ppos is the number of consultations where physicians and patients both state that they agree (response 4 to 5) compared to the number of consultations where physicians or patients state that they agree with the statement. An item was considered useful if Ppos or Pneg was at least 0.85.
Gender influence on agreement was tested in a logistic regression using agreement between physician and patient as the dependent variable, and age, patient and physician gender as independent variables. Agreement was constructed either based on agreement between physician and patient using a five-grade scale or on a dichotomized scale. Thus, two logistic regressions were constructed for each item.
|
Results |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionDeclarationReferences |
|---|
Eighty-six physicians in 23 primary health care centres were asked to participate. Sixty physicians (75%) accepted to participate but 14 never participated in the study. Thus, 46 of 60 (77%) actually participated in the study. Seventeen physicians failed to register eight suitable patients during their 6-week period. Seven physicians were asked to assist in including patients. Questionnaires were obtained from 316 consultations. The response rate of the questionnaire among physicians was 99% (314/316) and among patients 91% (289/316). The 314 physician questionnaires had 289 matched questionnaires from the patients. We also received two patient questionnaires with no corresponding questionnaire completed by the physician.
The physicians responded to nearly all items while patients failed to respond to some items (Table 1). Among questionnaires filled in by both physicians and patients, responses in single items were missing, mainly from patients, in a frequency of 8.5–21.5%.
|
Mean patient age was 50.0 years ranging from 17 to 84 years, with 68% females. In all, 41% of patients encountered a female physician. Eighty-three per cent of physicians were GPs for 0–24 years; 17% of physicians were interns or physician substitutes. There were 383 diagnoses which averaged out to 1.3 diagnoses per consultation. The most common causes for consultations were ICD10-13 diseases of the musculoskeletal system and connective tissue (N = 101), ICD10-10 diseases of the respiratory system (N = 61), ICD10-18 symptoms, signs and abnormal clinical and laboratory findings (N = 46) and ICD10-5 mental and behavioural disorders (N = 26).
Both patients and physicians perceived that the statements in the questionnaire correctly described what happened during the consultation but for Items 8–10 (Table 2).
|
Patients and physicians had a high level of agreement in responses. Thus, index of validity was in general high for each item (Table 2). Due to an imbalance in responses to statements according to ‘agree’ or ‘disagree’, kappa was low. The more useful proportional indexes show that Items 8–10 are of limited value while the other items are more useful (Table 2).
When evaluating gender influence on agreement based on the five-grade scale, gender differences were found in Items 5, 6, 10 and 12 (Table 3). In Item 5, male patients agreed less often with the physician than female patients. In Item 6, male physicians agreed less often with the patients than female physicians. In Item 10, male patients agreed more often with their physicians than female patients. In Item 12, male physicians agreed less often with their patients than female physicians. If agreement was based on dichotomization of the five-grade scale, a gender difference was only found for Item 6 where male patients agreed with their physicians less often than female patients. Confidence intervals for odds ratios in these items were close to one indicating that gender is of minor clinical importance.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionDeclarationReferences |
|---|
The main finding in this study was the identification of 10 items useful in PPQ to assess agreement in the consultation.
Methodological aspects
The primary aim of this study was to evaluate an instrument of physician–patient agreement for the consultation. Since the consultation is a unique experience that cannot be identically repeated, a test–retest procedure often used for measuring reliability of questionnaires, was not an option. Content validity was ensured as described in the ‘Method’ section. External criterion validation comparing with other questionnaires was not an option since the few previously published questionnaires evaluating agreement focus on other items than the questionnaires in this study. Validating each item by Ppos and Pneg estimating their ability to predict patient–physician agreement was the most logical choice to validate PPQ. This validation procedure could be considered to be internal criterion validity. Since validation in this study compares agreement between two experiences of the same consultation it could also be considered as construct validity.
The secondary aim of this study was to assess physician–patient agreement. Due to the inclusion criteria, the results presented might not be applicable to patients with chronic conditions or with self-limiting conditions of very short duration. The external validity of the PPQ is, therefore until further studies are performed, restricted to these categories of patients. Furthermore, in this type of study, participating physicians are selected by their agreement to participate. Thus, they are probably interested in the consultation process. At some primary care health centres, all physicians accepted participation. At others, some of the physicians accepted while at some centres all physicians declined. It is not possible to eliminate this bias but should be accounted for when interpreting results chiefly concerning the second aim.
Responses in single items were missing in a frequency of 8.5–21.5% which seemed acceptable compared to a similar study by Fagerberg et al.,6 where responses in a paired questionnaire were missing from the GP or patient in 22–26% of the matches.
The fact that 17 of 46 participating physicians did not register eight consecutive patients was partly due to the restrictions in the inclusion criteria and engagement in other activities such as child health care or medical care at nursing homes. Another explanation might have been physicians’ reality of being pressured for time. This incomplete compliance might have affected the result related to the secondary aim but probably not to the primary aim.
Summarizing item scores to dimension score or total score may not be appropriate considering the nature of ordinal scales. Factor rotation is a statistical analysis often used to determine how items should be merged into dimensions. However, one assumption for factor analysis and factor rotation is that item score consists of data having equidistant scale intervals. Thus, no factor rotation was performed in this study.
High index of validity but low kappa
We know that using a questionnaire with closed-ended questions and a graded scale in evaluation of patient satisfaction often lead to high scores; patients tend to be very positive. When patients completed the questionnaire at home, results were lower compared with when they filled it in directly after the consultation.16
Marcinowicz et al.17 compared closed-ended questions with open-ended questions and found that among 216 individuals grading family physician care as fairly good in an answer to a closed question, 42 patients (19.4%) used phrases expressing negative feelings in response to an open question.
Thus, questionnaires where patients are asked to grade physicians in closed-ended questions almost always yield positive responses. In this study, physician responses tended to resemble patient responses with very few negative responses. Prerequisites for high kappa are good agreement, often measured by index of validity and a fairly even distribution between positive versus negative responses. Thus, it can be predicted that kappa in studies comparing patient and physician attitude towards the consultation will be very low. The solution is to use proportional indexes rather than kappa.15 However, this requires that the ordinal Likert scale be rearranged to a dichotomized scale.
Choosing items
An item was considered useful if Ppos or Pneg was at least 0.85. Therefore, Items 8–10 were inadequate for assessing agreement. Thus, in the final version of the PPQ, only 10 items, Items 1–7 and 11–13, were considered useful.
Specific and global assessment of the consultation
Seeing the patient as an individual, listening intently to the patient, reaching an agreement with an understandable explanation to symptoms or illnesses and presenting treatment are specific ingredients essential to a successful consultation.10 If these conditions are met they are associated with better outcome.8,18
The aspect of agreeing on an understandable explanation is embraced by Items 1, 3, 5 and 6 and treatment by Item 7. If the physician responded affirmatively for these items it was likely that the patient had the same attitude towards the consultation, suggesting that the consultation may have contributed to an optimal outcome for the patient.
While Items 1, 3 and 5–7 mentioned in the paragraph above dealt with specific issues in the consultation, Items 2, 4 and 11–13 dealt with more general aspects. If the physician was positive in these global items it was likely that also the patient was satisfied with the consultation.
Gender aspects
Although statistical gender differences could be seen for some items, gender seemed to be of no major clinical importance.
Usability of PPQ
The physician version of PPQ with its 10 items could be used for educational purposes on all levels of training consultation skills.
Another way of using the physician version of PPQ is as a mental checklist for the GP in daily practice. However, in this situation, it can be practical to select a few items that the GP at regular intervals could pose after some consecutive consultations. Any of the 10 items could be chosen. Items 5 and 6, dealing with the patient's view of the reason for her complaints and agreement on the reason to the complaints, could be recommended for this purpose as the importance of these items are supported by Starfield's observation8 that a better outcome was associated with high agreement between patients and physicians about problems needing follow-up.
The matched physician–patient version of PPQ could be used in future research to further clarify physician–patient agreement in other settings.
|
Declaration |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionDeclarationReferences |
|---|
Ethical approval: The study was approved by The Göteborg University ethical committee.
Conflicts of Interest: None.
|
Notes |
|---|
Ahlén GC, Mattsson B, Gunnarsson RK. Physician–patient questionnaire to assess physician–patient agreement at the consultation. Family Practice 2007; 24: 498–503.
|
References |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionDeclarationReferences |
|---|
1 Howie JG, Heaney DJ, Maxwell M, Walker JJ, Freeman GK. Developing a ‘consultation quality index’ (CQI) for use in general practice. Fam Pract (2000) 17:455–461.
2 Fitzpatrick R. Surveys of patients satisfaction: I—important general considerations. BMJ (1991) 302:887–889.
3 Lundkvist J, Akerlind I, Borgquist L, Mölstad S. The more time spent on listening, the less time spent on prescribing antibiotics in general practice. Fam Pract (2002) 19:638–640.
4 Lewis JR. Patient views on quality care in general practice: literature review. Soc Sci Med (1994) 39:655–670.[CrossRef][Web of Science][Medline]
5 Martin E, Russell D, Goodwin S, Chapman R, North M, Sheridan P. Why patients consult and what happens when they do. BMJ (1991) 303:289–292.
6 Fagerberg CR, Kragstrup J, Støvring H, Rasmussen NK. How well do patient and general practitioner agree about the content of consultations? Scand J Prim Health Care (1999) 17:149–152.[CrossRef][Web of Science][Medline]
7 Freidin RB, Goldman L, Cecil RR. Patient-physician concordance in problem identification in the primary care setting. Ann Intern Med (1980) 93:490–493.
8 Starfield B, Wray C, Hess K, Gross R, Birk PS, D'Lugoff BC. The influence of patient-practitioner agreement on outcome of care. Am J Public Health (1981) 71:127–131.
9 Punamäki R-L, Kokko S. Content and predictors of consultation experiences among Finnish primary care patients. Soc Sci Med (1995) 40:231–243.[CrossRef][Web of Science][Medline]
10 Arborelius E, Timpka T, Nyce JM. Patients comment on video-recorded consultations—the "good" GP and the "bad". Scand J Soc Med (1992) 20:213–216.[Web of Science][Medline]
11 Williams S, Weinman J, Dale J, Newman S. Patient expectations: what do primary care patients want from the GP and how far does meeting expectations affect patient satisfaction? Fam Pract (1995) 12:193–201.
12 Steine S, Finset A, Laerum E. What is the most important for the patient in the meeting with a general practitioner? Tidsskr Nor Laegeforen (2000) 120:349–353.[Medline]
13 Larsen JH, Risør O, Putnam S. P-R-A-C-T-I-C-A-L: a step-by-step model for conducting the consultation in general practice. Fam Pract (1997) 14:295–301.
14 Feinstein AR, Cicchetti DV. High agreement but low kappa: I. The problems of two paradoxes. J Clin Epidemiol (1990) 43:543–549.[CrossRef][Web of Science][Medline]
15 Cicchetti DV, Feinstein AR. High agreement but low kappa: II. Resolving the paradoxes. J Clin Epidemiol (1990) 43:551–558.[CrossRef][Web of Science][Medline]
16 Kinnersley P, Stott N, Peters T, Harvey I, Hackett P. A comparison of methods for measuring patient satisfaction with consultations in primary care. Fam Pract (1996) 13:41–51.
17 Marcinowicz L, Borzuchowska A, Grebowski R. Methodologic difficulties in measuring patient satisfaction—discrepancy coming from formulating questions. Wiad Lek (2002) 55(suppl 1):335–340.[Medline]
18 Bass MJ, Buck C, Turner L, Dickie G, Pratt G, Robinson HC. The physician's actions and the outcome of illness in family practice. J Fam Pract (1986) 23:43–47.[Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||