Family Practice Vol. 19, No. 5, 476-483
© Oxford University Press 2002
Health Services Research |
Patient trust in the physician: relationship to patient requests
University of California at San Francisco School of Medicine, San Francisco, CA,
a University of California at Davis Center for Health Services Research in Primary Care,
b Department of Communication and
d Department of Statistics, University of California at Davis and
c Massachusetts College of Pharmacy and Health Sciences, USA.
David Thom, MD, PhD, Associate Professor of Family and Community Medicine, San Francisco General Hospital, 1001 Potrero Ave., Bldg 80/83, San Francisco, CA 94110, USA; E-mail: dthom{at}itsa.ucsf.edu
Thom DH, Kravitz RL, Bell RA, Krupat E and Azari R. Patient trust in the physician: relationship to patient requests. Family Practice 2002; 19: 476–483.
Received 8 November 2002; Accepted 13 May 2002.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
Background. Patient trust is a key component of the patient–physician relationship. A previous qualitative study has suggested that a low level of trust is associated with unfulfilled requests.
Objective. Our aim was to test the hypothesis that patients with a low level of trust will be more likely to report that requested or needed services were not provided during an office visit.
Methods. An observational study was carried out of office visits by 732 patients of 45 physicians (16 family physicians, 18 general internists and 11 cardiologists), within two managed care settings. Participants were consecutive, English-speaking patients, age 18 and older who had a significant health concern. Visit questionnaires were completed by 68% of patients known to be eligible. Post-visit measures included services requested (information, examination, prescription, test or referral); services provided; and requested or needed services not provided during the visit. Measures at 2-week follow-up included patient satisfaction, intended adherence to advice, interval contacts with the health system and symptom improvement.
Results. After adjustment for patient and physician characteristics, patient trust in the physician was not associated with the likelihood that a service was requested or provided during the visit, with the exception that prescription of a new medication was more frequent among patients with higher trust. In contrast, patients with low trust prior to the visit consistently were more likely to report that a needed or requested service was not provided (P < 0.001 for all services). Patients with a low level of trust were less satisfied with their care (P < 0.001), were less likely to intend to follow the doctor’s advice (P < 0.001) and were less likely to report symptom improvement at 2 weeks (P = 0.03).
Conclusions. Patients with a lower level of trust in their physician are more likely to report that requested or needed services are not provided. Understanding this relationship may lead to better ways of responding to patient requests that preserve or enhance patient trust, leading to better outcomes.
Keywords. Doctor-patient communication, doctor-patient relationship, trust.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Requests for services (and their fulfilment or lack of fulfilment) occur in the context of the doctor–patient relationship and, in turn, play an important role in defining the quality of the relationship. From the patient’s view, whether or not they receive the services they request or feel they need often defines the success of the visit.1
The association between patient trust and requests for services has not been studied previously. In business organizational theory, a low level of trust is believed to increase ‘transaction costs’ such as monitoring and verification. Applying this reasoning to the doctor–patient visit, one would expect that relationships with a low level of patient trust would be characterized by more patient requests for diagnostic tests, referrals or additional medical information, which may be seen as a way of verifying their doctor’s competency, and commitment. Other services, such as medication or an examination, may be requested more often by patients with a low level of trust if the patient has doubts about whether the doctor will write the prescription or perform the examination if not prompted by a request. In turn, requests prompted by lower levels of trust may be more likely to be denied by the physician if they are seen as motivated by lack of confidence rather then a ‘legitmate’ medical concern.
In the current study, we sought to test the hypotheses that patients with lower levels of pre-visit trust would be more likely to report requesting or receiving a service (medical information, examination, new prescription, testing or referral), and more likely to report that a needed service was not provided. We also tested the hypothesis that patients with lower levels of pre-visit level trust would be less satisfied with the office visit, less likely to report intending to follow their physicians’ advice, have more contacts with the health system after the visit and be less likely to report symptom improvement at 2 weeks after the visit.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
The Physician–Patient Communication Project is an observational study conducted within two large managed care health systems in the metropolitan area of Sacramento, California. The study gathered survey data from patients and physicians before, immediately after and 2 weeks after an out-patient visit. The study design has been described previously2 and will be summarized here.
Physician recruitment and enrolment
The study was limited to physicians who practised at least 20 h per week in internal medicine, family practice or cardiology. Physicians were recruited by the principal investigator for the study (RLK) using direct contact, or using participating physicians to recruit additional physicians. A total of 45 physicians were recruited to participate: 22 from UCDMG and 23 from Kaiser. Sixteen were family physicians (FPs), 18 were general internists and 11 were cardiologists. The mean age of the physicians was 44 years (SD 8.3) and they reported spending an average of 39 h per week in direct patient care.
Patient recruitment and enrolment
Between January and November 1999, the study recruited and enrolled English-speaking patients, aged 18 and older who had a significant health concern, defined as a new or worsening problem or being at least somewhat concerned about a serious undiagnosed condition. Phone calls were attempted for all potentially eligible patients identified from practice appointment lists 1 to 2 days prior to a visit. A total of 2606 were contacted, of whom 537 declined to participate before eligibility could be determined and 737 were identified as ineligible (69% due to lack of a significant health concern). Of the remaining 1352 patients known to be eligible, 1071 (80.4%) completed the telephone consent and screening and 909 (68.2%) completed the questionnaires at the index visit. There were 177 patients (19.5%) who reported no previous visits to the doctor they were seeing. These patients did not complete the pre-visit trust questions. Thus, the current paper reports results on 732 patients with at least one previous visit.
Measures.
Patient and physician characteristics.
Patients provided the usual demographic data and information about the length of their relationship with their physician and number of previous visits. At the time of physician enrolment, physicians completed an information form that included demographics and practice characteristics.
Patient trust.
Patient trust in the physician was assessed immediately prior to the visit using a nine-item, Likert-type response scale, the Patient Trust in the Physician Scale, developed for this study. Items were selected based on the results from previous patient focus groups3 and previously validated trust scales,4,5 to assess three previously identified key domains: honesty, competency and agency/fidelity.6 The new scale was piloted for clarity and acceptability prior to being used in the current study. Based on experience with similar scales, the Patient Trust in the Physician Scale was expected to be moderately correlated with number of previous visits, patient satisfaction and intended adherence to physician advice.
Services requested, needed and provided.
Immediately after the visit, patients reported if they had asked the physician for medical information, physical examination, a diagnostic test or procedure, a new prescription medication or a referral to a specialist. When the patient reported requesting one of these services, they subsequently ticked one of four boxes: "I was given everything I asked for", "I was given some of what I asked for", "My request was ignored" or "My request was denied". Patients were also asked if a service that should have been provided was not, with response options being "yes", "uncertain" and "no". Physicians were asked what services were requested and provided during the visit.
Outcomes.
After their visit, patients were asked about their satisfaction with the care received using a fouritem scale (alpha = 0.88) based on a measure of patient satisfaction with care developed by Ware and Hays.7 Physicians were asked to rate (i) how demanding the visit was and (ii) their satisfaction with the visit on two single item scales (1 = far less demanding/satisfying than average to 5 = far more demanding/satisfying than average). A phone call follow-up survey at 2 weeks asked patients to report interval contacts with the health system (phone calls, office and emergency room visits, hospitalizations), their satisfaction with care received (same scale as after visit), if they intended to follow their doctor’s advice (1 = strongly agree to 5 = strongly disagree), and to rate their symptoms with respect to the time of the index visit on a five-point scale (1 = much worse to 5 = much better).
Analysis. Values for missing items on the Patient Trust in the Physician Scale (a total of 71 items among 59 subjects) were input using the means of the reported items. For purposes of presentation, the average trust score was transformed to a 0–100 scale by subtracting 1 from the mean score and multiplying the result by 25. Patients were grouped into low, medium and high trust levels by tertiles and visit service request, and fulfilment during the visit was compared among these three groups. Categories of fulfilment of requests were collapsed into "given everything asked for" and "not given everything asked for". Multiple regression was used to test the statistical significance of associations, controlling for patient and physician characteristics. Similar analyses were performed for the other outcome variables. Outcome variables measured on a five-point Likert-type scale were dichotomized for ease of presentation, as indicated in the tables. Satisfaction immediately post-visit and at 2 weeks was dichotomized at the value of 4.50, approximately at the median. All P-values were calculated as two-tailed. Based on the principle that adjustment of P-values for multiple comparisons is needed only when multiple sampling is used (e.g. repeating the same test multiple times), but is inappropriate when testing for separate associations as done in the current study, we report the unadjusted P-values for all tests proposed a priori.8
Primary analyses were corrected for the clustering of patients within physicians using the Stata 6.0 svytab, svylogit, svymean and svyreg procedures for complex surveys. In these analyses, the physician was identified as the cluster (primary sampling unit), and a six-level stratification variable was created by crossing specialization (internal medicine, family practice and cardiology) with site (UCDMG and Kaiser). Probability weights were assigned to the patients in a cluster to account for differences among physicians in the number of patients enrolled in the study from their practice and the number of patients they see on a regular basis. Specifically, a weight was assigned to each patient within a cluster (i.e. physician practice) by (i) multiplying for each physician the number of patients seen weekly on an out-patient basis by sample size and (ii) dividing this value by the product of the number of patients seen weekly by all physicians in the study and the number of patients enrolled in the study by the index physician. Weights produced by this method ranged from 0.18 to 1.90.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Measurement of trust
On principal components analysis, the Patient Trust in the Physician Scale had a single factor which explained 63% of the variance, and high internal reliability (Cronbach’s alpha = 0.92) with item scale correlations ranging from 0.64 to 0.79 (Table 1
). When transformed to a 0–100 scale, the scale had a mean of 87, standard deviation of 15.6 and a range of 14–100. There was a substantial ceiling effect, with 29% of patients awarding their physician a perfect score. As expected, trust was moderately correlated with number of previous visits (r = 0.18, P < 0.001), patient satisfaction with the care received (r = 0.47 post-visit and r = 0.25 at 2 weeks, P < 0.001 for both), and intended adherence to physician advice (r = 0.41 post-visit and r = 0.31 at 2 weeks, P < 0.001 for both)
|
Correlates of pre-visit trust
Table 2
shows the average level of trust by patient and physician characteristics after adjustment for the other variables in the table. Patient trust showed a ‘U-shaped’ relationship with patient age, being highest in the youngest (age 19–29) and oldest (age 70+) age groups, and lowest in the age 40–49 age group (P = 0.03). The level of trust was significantly greater in patients who had seen the physician for >6 months compared with patients seeing the physician for <6 months, but showed no additional increase with length of time treated beyond 6 months. In contrast, trust increased linearly by grouped number of previous visits. Patients of female physicians reported a slightly higher level of trust compared with male physicians (P = 0.03). The level of trust was substantially higher among patients of cardiologists compared with general internists (P = 0.002) or FPs (P = 0.05). This difference in specialty was examined further at the level of individual trust scale items. For all items, cardiologists scored higher on average than did FPs or general internists. The difference in scores was greatest for item 3 ‘make excellent judgements’ (4.65 for cardiologists versus 4.51 for FPs and 4.38 for internists), item 4 ‘do everything that should be done’ (4.66 for cardiologists versus 4.44 for FPs and 4.42 for internists) and item 7 ‘put your medical needs above all other considerations, including costs’ (4.50 for cardiologists versus 4.34 for FPs and 4.17 for internists). Trust was also higher in the large multispecialty group compared with the group model HMO (P = 0.02). There were no significant differences in level of trust by patient gender, marital status, education or income.
|
Services requested and provided
Patients were grouped into groups of low, medium and high levels of trust by tertile for purposes of presentation of the remaining analyses. The associations between level of trust and requested or provided services were examined, adjusting for the variables in Table 2
. Patients with lower levels of trust were more likely to report that they requested medical information (P = 0.05) or a referral (P = 0.02). However, there was no significant association between trust and patient request for these services as reported by the physician. Prescription of a new medication was the only service requested or provided that was associated consistently with patient trust. By both physician and patient report, patients with higher levels of trust were more likely to request a new medication (P = 0.05), and, by physician report, were more likely to be prescribed a new medication. There were no other significant associations between trust and the request for, or provision of, other services.
Request fulfilment
In contrast, there was a striking and consistent association between patient trust and patient report of whether a requested service was provided (Table 3). Patients in the lowest tertile of pre-visit trust were far more likely to report that a requested or needed service was not fulfilled, with the differences between patients in the middle and highest trust tertiles being consistent, but less dramatic. Similarly, a lower level of trust was associated strongly and consistently with patients reporting that a needed service was not provided for all five medical services (Table 4).
|
|
Outcomes post-visit
Table 5
presents the adjusted associations between patient pre-visit trust, divided by tertiles, and visit outcomes. Patients with a lower level of trust were perceived by their physician to be more demanding than average (P < 0.01), but physicians reported being no less satisfied with these visits. Patient trust was postively associated with intent to follow advice (P < 0.001), with the difference being the most striking for patients in the lowest tertile of trust. The level of trust was strongly associated with patient satisfaction immediately following the visit and 2 weeks later (P < 0.001 for both). There was a positive association between patient trust and symptom improvement 2 weeks following the visit (P = 0.03), but no difference in interval contact with the health system.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
The Patient Trust in the Physician Scale employed in the current study performed similarly to previously published patient trust scales in that it showed a single factor structure, high internal reliability and a mean score and standard deviation in the range of previously reported trust measures.4,5,9,10 The scale had considerable face validity, as it was designed to be similar to previously published measures of trust and to include an equal number of items from the three domains believed to be essential to trust: integrity/honesty, competence and agency/fidelity.3 Construct validity was supported by the positive correlation between trust and number of previous visits, patient satisfaction with the care received and intended adherence to physician advice. Predictive validity was supported by the significant association between trust prior to the visit and symptom improvement 2 weeks after the visit.
Trust was substantially higher in patients seeing a cardiologist compared with those seeing an FP or general internist, even after adjusting for multiple patient and physician characteristics. No previous study has reported trust by physician specialty. The additional finding that this difference was greatest for items referring to ‘making excellent judgments’, ‘doing everything that should be done’ and ‘putting the patient’s medical needs above all other considerations, including cost’ is consistent with a perception of the specialist as being more of a clinical expert in their area, more of an interventionist and less of a gatekeeper, compared with the primary care physician.
Patients who reported a low level of trust in their physician did not, in general, appear to make more (or fewer) requests, nor did their physicians report providing more (or fewer) services, with the exception of new medication, which was actually more likely to be requested by, or provided to, patients with greater trust. These results do not support the original hypothesis that less trust would be associated with more requests and more services. This lack of association may be the result of competing forces. For example, low trust might prompt patients to make requests because they do not trust the doctor to provide what they need, while patients with high levels of trust might feel more comfortable asking for what they want.
In contrast, patients with lower levels of trust were much more likely to report that they did not receive services they requested, or services they felt they needed (whether or not requested). While this result could be interpreted as indicating that a low level of pre-visit trust leads to a denial of patient requests, it seems more reasonable to interpret it as indicating that relationships involving less trust are also characterized by patients feeling they do not receive the services they request and/or need. In a previous study of patient trust using patient focus groups,3 instances of denial of requested tests or treatments were cited as reducing or obliterating patient trust in the physician. In practice, the relationship between patient trust, service request and the experience of a requested or needed service being denied may be mutually reinforcing. It is likely that patients with low levels of trust have a different experience with the physician, resulting in feeling they have not received the services they desire or feel they need. Possible reasons for this difference could include patients not making requests in a way that is heard by the physician, physicians hearing requests but not responding and physicians responding to requests but patients not perceiving the response as sufficient.
The inverse association between patient trust and how demanding the visit was as reported by the physician was in the direction expected, even though physicians did not report patients with a low level of trust to have requested more services. This suggests that the perception of how demanding a visit was is due to other factors, perhaps the demeanour or tone or the interaction, that were not measured in the current study. Interestingly, there was no association between trust and physician satisfaction with the visit. The lack of association with physician satisfaction has been found in at least one previous study10 and may reflect the fact that physician satisfaction is dependent on many factors other than patient trust.
Patients with low levels of trust were substantially less likely to report that they intended to adhere to their physician’s advice, even after adjustment for multiple patient and physician characteristics. This result is consistent with a previous study10 but is limited by the nature of patient self-report. A future study is needed to examine the relationship between trust and observed adherence.
The finding that patients with greater trust were more likely to report improvement in symptoms (2 weeks after the office visit), even after adjustment of multiple patient and physician characteristics, is intriguing and has not been reported previously. This finding may reflect a beneficial effect of trust on patient symptoms. Whether there is any beneficial effect of trust on objective functioning is not known.
This study has several limitations that should be borne in mind when interpreting the results. First, the study only enrolled patients scheduled for a visit for a new or worsening problem in two large, mature, managed care systems. The results reported may not generalize to other settings or populations of doctors and patients. Secondly, the results represent patients’ perceptions of care rather than care itself. It is possible that patients with lower levels of trust were predisposed to report that requested services were denied, even if there was no objective difference in the likelihood of denial by level of trust. However, in the context of patient-centred care, the patient’s perceptions may be what is most relevant.
In summary, our study found that patients with a lower level of pre-visit trust in their physician are much more likely to report that requested or needed services are not provided during their office visit. The causal direction of this relationship is not clear, and may, in fact, be bi-directional, with a low level of trust leading to the perception that requests are denied, or needed services are not provided, which further erodes trust. The strength of this association suggests that this would be a fruitful area for further research.
|
Acknowledgments |
|---|
The authors gratefully acknowledge the assistance of the 45 participating physicians and their patients. Thanks are also due to Sara Lu Vorhes, Steven Kelly-Reif and David Ormerod for assistance with physician recruitment and data collection; to Christine Harlan for budgetary management; and to the staff of the Patient–Provider Relationship Initiative (Bernard Lo, Director) for technical assistance. Funded by a grant from the Robert Wood Johnson Foundation (#034384).
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
1 Kravitz, RL, Cope DW, Bhrany V, Leake B. Internal medicine patients’ expectations for care during office visits. J Gen Int Med 1994; 9: 75–81.[ISI][Medline]
2 Kravitz RL, Bell RA, Thom DH, Krupat E, Azari R. Antecedents and consequences of request fulfillment in office practice: results from the physician–patient communication project. Med Care 2002; 40: 38–51.[ISI][Medline]
3 Thom DH, Campbell B. Patient–physician trust: an exploratory study. J Fam Pract 1997; 44: 169–176.[ISI][Medline]
4 Safran DG, Taira DA, Rogers WH, Kosinski M, Ware JE, Tarlov AR. Linking primary care performance to outcomes of care. Fam Pract 1998; 47: 213–220.
5 Anderson LA, Dedrick RF. Development of the trust in physician scale: a measure to assess interpersonal trust in patient– physician relationships. Psych Rep 1990; 67: 1091–1100.
6 Mechanic D, Schlesinger M. The impact of managed care on patents’ trust in medical care and their physicians. J Am Med Assoc 1996; 275: 1693–1697.[Abstract]
7 Ware JE, Hays RD. Methods for measuring patient satisfaction with specific medical encounters. Med Care 1988; 26: 393–402.[ISI][Medline]
8 Perneger TV. What’s wrong with Bonferroni adjustments? Br Med J 1998; 316: 1236–1238.
9 Kao AC, Green DC, Zaslavsky AM, Koplan JP, Cleary PD. The relationship between method of physician payment and patient trust. J Am Med Assoc 1998; 280: 1708–1714.
10 Thom DH, Ribisl KM, Stewart AL, Luke DA. Validation of a measure of patients’ trust in their physician: the Trust in Physician Scale. Med Care 1999; 37: 510–517.[ISI][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||