Research design and methods
We conducted a mixed-methods study using semistructured in-person interviews with adult patients diagnosed with type 2 diabetes and hypertension who received primary care at an academic urban hospital-based clinic. Interviews were conducted in English using a semistructured guide, which included scaled-response questions and follow-up exploratory questions. This study was approved by the University of Chicago Institutional Review Board.
Participant recruitment
We prescreened patients seen between 1 August 2012 and 31 July 2013 using data from the Clinical Research Data Warehouse at the University of Chicago. We included patients who were between 40 and 70 years old with both type 2 diabetes and hypertension, with duration of diabetes <10 years, and who were taking oral medications for both conditions. We excluded patients who were on insulin, pregnant, had severe visual impairment, or deafness. We also excluded patients who may have had difficulty considering the future in their decisions due to mental disability (ie, a history of stroke or cognitive impairment) or limited life expectancy (ie, on dialysis, active cancer, liver failure). A research assistant telephoned eligible participants to confirm data, screen for cognitive impairment,16 and schedule interviews.
After in-person consent was obtained, participants were interviewed by one of two trained interviewers (PCF and AN). Interviewers recorded responses to scaled questions and digitally recorded interviews. We interviewed participants until we reached our a priori sample of 60 completed interviews. We used stratified purposeful sampling in order to obtain a distribution in race/ethnicity.17 Recruitment and interviews occurred between January and September 2014.
Measures
Our main outcomes were change in likelihood of starting an oral medication for diabetes and hypertension after being provided information about its (1) time requirements and (2) legacy effects.
Prior to informing participants about the medications' time requirements and legacy effects, participants were asked how likely they would be to start an additional medication if recommended by their doctor, using a response scale of 1 (not at all likely) to 10 (very likely). Patients were told that the short-term benefits of the medication were to lower their sugars (or blood pressure) and the long-term benefits of lower blood sugars (or blood pressure) was a moderately lower risk of complications, like amputation, heart attack, stroke, kidney disease, blindness, and numbness (or heart disease, stroke, kidney disease, eye disease, and vascular disease). Patients were told that the medication would be taken once a day and would be easy to swallow and affordable. Patients were provided information on the potential side effects based on metformin and hydrochlorothiazide. For diabetes, they were advised that the medication could cause low blood sugar in a very small number of people and could cause a very small amount of weight gain. If asked, the interviewer was able to quantify moderately lower risk (12% relative risk), very small number (<3%) and weight gain (4 lbs). For hypertension, they were told that the medication could cause muscle cramps, aches, and irregular heartbeats in a very small number of people. If patients asked how much a moderately lower risk or very small number was, the interviewer quantified moderately lower risk as a 24% relative risk and very small number as 4%. As their likelihood of starting a medication may be associated with their self-efficacy18 and outcome expectancy for taking additional medications,19 ,20 participants were also asked, on a scale of 1–10, how confident they were that they could take the medication and how likely they believed they would obtain its benefits.
Then, to ascertain how information about time requirements and legacy effects impacts decision-making, the interviewer told participants how long they would need to take the additional medication in order to get its benefits and how long the benefits would persist after stopping the medicine. We used data about the time requirements and legacy effects of intensive control from the UKPDS. The UKPDS demonstrated a lower risk of complications after 10 years of intensive glycemic control and 3 years of intensive blood pressure control, and a legacy effect of 10 years with intensive glycemic control and no legacy effect with intensive blood pressure.3 ,4 ,15 ,21 ,22 We used the starting of a medication as a proxy for intensive control.
For example, for diabetes, the interviewer informed participants that it would take 10 years to get the long-term benefits of taking the additional medication and then asked if their likelihood of taking it would change. Then the interviewer informed participants that the long-term benefits of taking the medication would last an additional 10 years, even after they stopped taking it, and asked if their likelihood of starting it would change. A similar scenario was described for hypertension, except the time requirement was 3 years and there was no benefit of taking the medication after stopping it. To control for sequencing effects, participants were randomized to either receive information about diabetes or hypertension first. Interviewers also asked participants how certain they were that they would be alive in 10 or 20 years, on a scale of 1 (absolutely no chance) to 10 (absolute certainty).
Finally, participants were asked whether they were interested in learning about the time requirements of medications, and why. Participants reported their sociodemographics; electronic health records were reviewed for their most recent glycated hemoglobin and blood pressure values.
Data analysis
On scale-based responses, participant responses were categorized into high (≥7), medium,4–6 and low (≤3). Interviews were transcribed verbatim, and a modified template approach was used for qualitative analysis. We created a codebook with predefined codes and iteratively updated it to capture new information and identify themes.23 Each transcript was reviewed and coded by two or more trained coders (PCF, AN, NL, CL, NS, or DG). Codes were compared and discrepancies discussed until agreement. Data were collected and managed using REDCap electronic data capture tools hosted at the University of Chicago.24 We used SAS V.9.3 to conduct quantitative analyses and atlas.TI (V.7.5) to manage qualitative data.