Applying best-worst scaling to identify health-outcome preferences among patients with diabetes or hypertension and multiple chronic conditions

2017 Cape Town [Global Evidence Summit]

Aschmann HE¹, Puhan MA¹, Armacost K², Bayliss EA³, Bennett WL⁴, Chan WV⁵, Glover C⁶, Leff B⁷, Maslow K⁸, Mintz S⁹, Mularski RA⁵, Robbins CW¹⁰, Sheehan OC⁷, Wilson R¹¹, Yu T¹², Boyd CM⁴

¹Epidemiology, Biostatistics and Prevention Institute, University of Zurich

²Patient Caregiver, Informing Patient-Centered Care for People with Multiple Chronic Conditions Patient and Caregiver Partners

³Kaiser Permanente, Institute for Health Research, Denver, CO

⁴The Johns Hopkins University School of Medicine & The Johns Hopkins University School of Public Health, Baltimore, MD

⁵Kaiser Permanente National Guideline Program

⁶Patient Caregiver, Informing Patient-Centered Care for People with Multiple Chronic Conditions Patient and Caregiver Partners & Sought Out Redeemed Hope Center

⁷The Johns Hopkins University School of Medicine, Baltimore, MD

⁸Gerontological Society of America, Washington, DC

⁹Founder Family Caregiver Advocacy

¹⁰Kaiser Permanente National Guideline Program & Kaiser Permanente, Institute for Health Research, Denver, CO

¹¹The Johns Hopkins University School of Public Health, Baltimore, MD

¹²Department of Public Health, China Medical University (ROC), Taichung

Background:

Treatment decisions and guideline development need to consider patient values and preferences. There is little evidence in the literature on patient preferences for health outcomes in people with hypertension or diabetes and multiple chronic conditions.

Objectives:

To elicit patient preferences for clinically relevant and patient-important outcomes to guide decision making for treatment of hypertension and diabetes among people with multiple chronic conditions.

Methods:

In collaboration with patient and caregiver focus groups, we determined clinically relevant and patient-important outcomes for two questions, one on second-line treatments for diabetes and the other on blood-pressure targets in hypertension. We designed the surveys as best-worst scaling tasks (case 1) based on the balanced, incomplete-block design. We sent both surveys to Kaiser Permanente Colorado patients with multiple chronic conditions and a Quan score of at least 3 and who have diabetes or hypertension, respectively. The analysis used best minus worst scores (BMWS) based on a preliminary dataset (N=154 (diabetes)/148 (hypertension)). BMWS reflect how many times an outcome was selected as best or worst, averaged across respondents. The range of scores depends on the design, i.e. how many times the outcome can be selected. The range is [-4,4] for diabetes and [-5,5] for hypertension outcomes.

Results:

Our response rate was 46 per cent. BMWS are shown in Figure 1 for diabetes, and Figure 2 for hypertension. In diabetes loss of vision was considered the most worrisome outcome, followed by stroke and myocardial infarction. The least worrisome events were nausea or diarrhoea, mild depression and weight gain. In the hypertension survey, stroke was considered the most worrisome health outcome, followed by heart failure and myocardial infarction. The least worrisome were treatment burden, injurious falls and hypotension or dizziness. In both cases, mean scores did not go to the extremes.

Conclusions:

The best-worst scaling allowed good discrimination between the importance of health outcomes among people with multiple chronic condititions.