These results have been presented at the 66th Annual Scientific Meeting of the American Diabetes Association (Washington, DC, USA, 2006) and at the American Academy on Communication in Healthcare Conference (Atlanta, GA, USA, 2006).
A treatment decision aid may increase patient trust in the diabetes specialist. The Statin Choice randomized trial
Article first published online: 27 FEB 2009
© 2009 The Authors. Journal compilation © 2009 Blackwell Publishing Ltd
Volume 12, Issue 1, pages 38–44, March 2009
How to Cite
Nannenga, M. R., Montori, V. M., Weymiller, A. J., Smith, S. A., Christianson, T. J. H., Bryant, S. C., Gafni, A., Charles, C., Mullan, R. J., Jones, L. A., Bolona, E. R. and Guyatt, G. H. (2009), A treatment decision aid may increase patient trust in the diabetes specialist. The Statin Choice randomized trial. Health Expectations, 12: 38–44. doi: 10.1111/j.1369-7625.2008.00521.x
- Issue published online: 27 FEB 2009
- Article first published online: 27 FEB 2009
- Accepted for publication 1 August 2008
- decision aids;
- health-care delivery;
- patient experience;
Aims Decision aids in practice may affect patient trust in the clinician, a requirement for optimal diabetes care. We sought to determine the impact of a decision aid to help patients with diabetes decide about statins (Statin Choice) on patients’ trust in the clinician.
Methods We randomized 16 diabetologists and 98 patients with type 2 diabetes referred to a subspecialty diabetes clinic to use the Statin Choice decision aid or a patient pamphlet about dyslipidaemia, and then to receive these materials from either the clinician during the visit or a researcher prior to the visit. Providers and patients were blinded to the study hypothesis. Immediately after the clinical encounter, patients completed a survey including questions on trust (range 0 to total trust = 100), knowledge, and decisional conflict. Researchers reviewed videotaped encounters and assessed patient participation (using the OPTION scale) and visit length.
Results Overall mean trust score was 91 (median 97.2, IQR 86, 100). After adjustment for patient characteristics, results suggested greater total trust (trust = 100) with the decision aid [odds ratio (OR) 1.77, 95% CI 0.94, 3.35]. Total trust was associated with knowledge (for each additional knowledge point, OR 1.3, 95% CI 1.1, 1.6), patient participation (for each additional point in the OPTION scale, OR 1.1, 95% CI 1.1, 1.2), and decisional conflict (for every 5-point decrease in conflict, OR 1.5, 95% CI 1.2, 1.9). Total trust was not associated with visit length, which the decision aid did not significantly affect. There was no significant effect interaction across the trial factors.
Conclusions Preliminary evidence suggests that decision aids do not have a large negative impact on trust in the physician and may increase trust through improvements in the decision-making process.
Trust is a fundamental feature of the patient–physician relationship1–5 that correlates with patient satisfaction, continuity of care, adherence to medical therapy, and other desirable practice metrics.1,4,6–9 Despite the importance of trust, efforts to improve patient trust in their clinicians have been sparse and largely inconclusive.1,2
Certain patient and clinician characteristics may affect trust. Observational studies show that patients preferring passive or shared roles in decision making have higher levels of trust than their more autonomous counterparts.5,10,11 Clinician characteristics best correlated with patient trust include manifestations of caring and comfort, demonstration of competency, explaining actions, and encouraging and answering patient questions.4
The nature of the clinical visit could also affect trust. Decision aids are tools that, among other purposes, help clinicians share unbiased evidence-based information and uncertainty with patients to improve the quality of clinical decisions.12 The extent to which using decision aids in practice can favourably or unfavourably affect patient-perceived trust in the clinician who uses them remains unknown.
We have previously reported on the development and purpose of the decision aid Statin Choice13 which can be found on our website (http://kerunit.e-bm.org), and on the main results of the Statin Choice trial. This trial showed that a decision aid was acceptable to patients, positively influenced patient knowledge and decisional conflict about using statins, enhanced the risk-appropriate use of statins, and improved short-term adherence to statins.14 Here, we report on the extent to which use of this treatment decision aid in patients with type 2 diabetes improved their trust in a specialty provider, a pre-specified secondary endpoint. We also examined the association between trust and other variables to begin to elucidate determinants of trust.
Patients and methods
We conducted the Statin Choice randomized trial, a single-centred study of a decision aid addressing use of statin drugs in patients with type 2 diabetes, in a subspecialty clinic for diabetes at Mayo Clinic in Rochester, MN, USA15 between April and July of 2005. We videotaped patients during clinical encounters and surveyed patients immediately after the visit and at 3 months. Readers can find the design and the main results of the trial elsewhere.13,14 The Mayo Clinic Institutional Review Board approved the protocol.
Eligible patients had a clinical diagnosis of type 2 diabetes, were able (had no major hearing, visual, or cognitive impairment or did not require translation) and willing to provide informed consent, had no reported contraindications to statin use, and were available for follow-up at 3 months. Eligible providers were endocrinologists (consultants and fellows), stratified by sex and level of training (fellow vs. consultant) prior to randomization. All encounters were initial consultations; there were no pre-existing patient–provider relationships.
Blinding and randomization
Providers and patients were naïve to this study objective (the relationship of decision aids to patient trust) and randomized by concealed central allocation to a 2 × 2 clustered factorial design to intervention (Statin Choice decision aid or control) from their clinician during the visit or from the researcher prior to the visit, thus creating four groups (Fig. 1).
The experimental intervention was the use of the Statin Choice decision aid, a one-page document tailored to the individual patient including the patient’s name, cardiovascular risk factors, and estimated cardiovascular risk—average (<15% risk of a cardiovascular event in the next 10 years), elevated (15–29% risk) or high (≥30% risk).16 Benefits and downsides were presented using natural frequencies and ordered ovals (e.g. after rows of green happy faces, red frowning faces denoted those, out of 100, who had undesirable events in 10 years of observation) (see http://kerunit.e-bm.org). The control intervention, the standard Mayo patient education pamphlet, outlined guidelines for reducing hyperlipidaemia, cholesterol, and triglycerides without consideration of patient-specific cardiovascular risk. It defined lipid disorders and provided primarily dietary guidelines for control of cholesterol along with general statements encouraging exercise and smoking cessation.
Researchers videotaped all clinical encounters. Immediately following the encounter, patients completed a 71-item survey including knowledge questions (16 questions), the decisional conflict scale (16 questions),17 and the previously validated 9-question patient trust in physician scale.11 This instrument addresses the patient’s view of the physician’s honesty, competency and agency/fidelity. Using the videotaped encounters, reviewers blinded to questionnaire result quantified encounter duration and used the OPTION scale18,19 to quantify the extent to which clinicians invited patient participation in decision making.
Individual patient scores and means on the Trust in physician scale were transformed to a 0–100 score, where 0 = total lack of trust and 100 = total trust. We dichotomized the scale to scores consistent with total trust (100 points) vs. any other score, with the idea that any score less than 100 would reveal some degree of distrust. This construction is a posteriori after noting the distribution of the data, but without knowledge of its impact on trial results. Generalized estimating equations provided parameter estimates and 95% CI using trust as a dependent variable and intervention as predictor variable with adjustment for clustering by physician.
Given data supporting age, health status, and gender-associated disparities in trust in physician,4,6,9,11,20–23 we adjusted analyses for cardiovascular risk (strongly associated with age and gender), extent of diabetes control (HbA1C), and patient–provider gender concordance; we also adjusted for whether patients were or were not using statin drugs at baseline.
Estimates of the association between total trust and knowledge, decisional conflict, OPTION scale, and time were adjusted for age, patient–provider gender concordance, education, and self-assessment of health status (using a 100-point feeling thermometer). Based on our hypothesis that long-distance travel to a referral centre implies trust, we additionally adjusted for the distance from patients’ home addresses to the Mayo Clinic.
The trial enrolled 98 patients who were randomized to either the decision aid or control groups (Fig. 1). All patients received the allocated intervention, with one patient in the decision aid group (researcher arm) failing to complete any of the survey items. Baseline cardiovascular risk factors were generally well-balanced (Table 1), although the decision aid group had significantly fewer women, greater high school completion, and a higher baseline HbA1C. Distance from place of residence to Mayo Clinic was similar for intervention and control groups.
|Statin choice decision aid (n = 52)||Control pamphlet (n = 46)|
|Age, mean (SD)||64 (12)||66 (8)|
|Women, n (%)||16 (31%)||26 (57%)|
|High school education completed, n (%)||51 (98%)||39 (87%)|
|Travel distance of US residents in miles from home to Mayo Clinic, median (range)*||71.9 (2.6–1502.8)||76.1 (2.7–1551.5)|
|Haemoglobin A1C, median (IQR)||7.6 (6.5, 8.4)||6.4 (5.9, 6.8)|
|10 or more years since DM diagnosis, n (%)||20 (38%)||16 (35%)|
|LDL cholesterol, median (IQR)||80 (76, 115)||87 (74, 110)|
|Diagnosis of CAD, n (%)||26 (50%)||20 (44%)|
|Total number of medications, mean (SD)||9 (4)||10 (4)|
|Already taking a statin||29 (56%)||27 (59%)|
The overall mean trust score was 90.7 (median 97.2, IQR 86.1–100). There was a non-significant trend towards higher trust scores when clinicians used the decision aid rather than the control pamphlet [difference in trust 3.1 out of 100, 95% CI −1.0 to 7.3, P = 0.13; (Table 2)]. In the decision aid group, 29 of 51 reported total trust (score of 100) compared with 15 of 46 patients in the control groups [odds ratio (OR) 1.77, 95% CI 0.94, 3.35, P = 0.08]. Although the proportion of patients who reported total trust was greater when the clinicians used the decision aid during the visit (17 of 26, 65%) than when the researchers used it outside the visit (12 of 25, 48%) or when patients received the control pamphlet (15 of 46, 33%), this apparent interaction between the impact of the decision aid on trust and who (clinician or researcher) delivered the intervention was not statistically significant (difference in effect size across modes of delivery, 4.7, 95% CI −3.5, 13). Patients receiving the decision aid were significantly more likely to trust their doctor to ‘tell if a mistake was made about treatment’ (mean difference vs. control 10.7, P = 0.016) and to ‘listen well so he/she understands your needs and concerns’ (mean difference vs. control 9.2, P = 0.002) (Table 2).
|How much do you trust your doctor to||Difference (95% CI)|
|Always tell you the truth?||1.5 (−2.2, 5.1)|
|Provide you with accurate, up-to-date medical information?||0.3 (−4.2, 4.8)|
|Make excellent medical judgments on your behalf?||−1.2 (−5.8, 3.3)|
|Do everything medically that should be carried out in order to ensure the best possible result?||−0.3 (−4.4, 3.9)|
|Tell you when you could benefit from seeing a specialist?||3.6 (−1.8, 9.1)|
|Tell you if a mistake was made about your treatment?||10.7 (2.0, 19.5)|
|Put your medical needs above all other considerations, including cost?||2.0 (−6.6, 10.6)|
|Listen well so he/she understands your needs and concerns?||9.2 (3.3, 15.0)|
|Never pretend to know things when he/she is not sure?||6.1 (−1.0, 13.2)|
|Overall||3.1 (−1.0, 7.2)|
As we have reported previously14, use of the decision aid resulted in higher knowledge scores (mean difference 1.6, 95% CI 0.7, 2.5) and less decisional conflict (mean difference −9.8, 95% CI −14.2, −5.4). The decision aid also resulted in greater patient participation (OPTION scale mean difference 4.4, 95% CI 2.9, 6.0). We found no significant difference in face-to-face consultation duration with the staff endocrinologist (mean difference 3.8 min longer with the decision aid, 95% CI −2.9, 10.5). The likelihood of having total trust in the physician was associated with improvement in knowledge (for every additional knowledge point, OR 1.3, 95% CI 1.1, 1.6), lower decisional conflict (for every 5-point difference in conflict, OR 1.5, 95% CI 1.2, 1.9), and greater patient participation in decision making (for each point in the OPTION scale, OR 1.1, 95% CI 1.1, 1.2); there was no significant association with encounter duration (for every 5-min increase, OR 0.9, 95% CI 0.7, 1.1). Of note, trust was not significantly associated with adherence to statins (OR 1.72, 95% CI 0.60, 4.92).
The exploratory results from this randomized trial suggest that use of decision aids does not impact negatively trust in the clinician who uses them and could enhance trust, and that this effect could be mediated through improvements in knowledge, reductions in decisional conflict, and increased patient participation in clinical decision making.
Limitations and strengths
There are several limitations to this study. This is a small single-centre trial with trust as one of several secondary outcome measures. The mean trust score in the control group of 88.8, while high, was similar to previous experiences with the Patient Trust in Physician Scale. In a primary care population, Thom reports a mean trust score of 87 (SD 16), with 29% of patients showing total trust.9 As this provided little room for improvement, we dichotomized responses into perfect and non-perfect scores to enhance interpretation. We additionally recognize that scores for trust could be lower, and the effect of the decision aid different among minority patients (who reportedly may have greater distrust of physicians)21,24, patients with established relationships with their clinician (these were all first visits) and local patients who do not travel great distances to receive care. We did not have a sizable proportion of eligible minority participants, but we did adjust all analyses for travel distance. Our study also has important strengths, including its rigorous design and conduct and the planned and focused nature of our analyses. To our knowledge, this is the first use of trust as an outcome in a decision aid trial.
Implications for clinical practice and research
These results offer an interesting hypothesis for decision-making researchers. Can decision aids not only not damage trust but improve the patient–physician conversation to the extent of furthering partnership and trust? If so, how persistent is this effect over time? If confirmed in other studies, this would be an important finding, as we have previously reported partnership-building as a key component of shared decision making in patients with chronic conditions like diabetes.25 In this study, improvements in knowledge, decisional conflict, and patient participation in decision making were directly associated with total trust. The association between these variables and trust provides empirical evidence bearing on construction of an explanatory model linking trust with decision aid design and use.12,26
The association between trust, time, and use of the decision aid is complex. Arguably, lack of trust may have changed the nature of the clinical encounter, thus making it longer. It is therefore notable that patients who received the decision aid had both the longer encounters and the highest proportion of total trust. Indeed, larger studies aiming to determine the extent to which decision aids affect trust should measure encounter duration with greater precision to better delineate whether any effect on trust is mediated by the physician spending more patient-focused time during the visit.
Use of a decision aid in patients with diabetes did not negatively impact on reported trust in the physician and may lead to increased trust in the diabetes provider through improvements in the decision-making process. Trust is a potentially useful novel outcome for decision aid trials that deserves further mechanistic exploration and replication. Our study provides further evidence that participatory decision making facilitated by a decision aid may enhance diabetes care.
The authors are very grateful to Julka Almquist, who conducted many of the videotaped observations and evaluations. The Mayo Clinic Section of Patient Education and the American Diabetes Association funded this work. These funding sources were not involved in planning or conducting this trial or in reporting its results.