The first survey (hereafter, the U.S. Survey) was mailed in November 2013 to the work addresses of a nationally representative random sample of 2000 internal medicine, family medicine, and geriatrics physicians in the United States identified from the AMA Masterfile, which includes almost all doctors of medicine (MD) and doctors of osteopathic medicine (DO). Participants received a $5 incentive in their initial survey packet. Non-respondents received up to two additional surveys over the following 6 months. This design was approved by the University of Michigan’s Health Sciences and Behavioral Sciences Institutional Review Board.
For the second survey (hereafter, the VA Survey), using data from the VA Corporate Data Warehouse, we identified all non-resident primary care providers (physicians, nurse practitioners, and physician assistants) with at least 1 day of direct patient care per week. During October–December 2014, a national random sample of 2500 providers received a pre-notification letter, followed by an email containing a link to an anonymous online survey using REDCap [Research Electronic Data Capture] tools hosted at the VA.
14 We sent follow-up reminders (maximum of 5) to non-responders by REDCap (email) and UPS overnight mail (surveys included). Respondents were entered into a lottery for a chance to win one of thirty $100 Amazon gift cards. The Ann Arbor VA Human Studies Committee approved this study.
Survey Instrument Design
We conducted a focused review of all CW recommendations available as of July 2013 to identify those we believed most critically relevant to primary care for adult patients. Recommendations related to pediatrics, obstetrics, or hospital-based care were excluded. Four PCPs (the three physician authors and one other physician) evaluated each of the remaining 84 (out of 135) recommendations for its relevance to primary care for adults. Raters met after the initial ratings, discussed any disagreements, and for each recommendation came to consensus on whether it should be included or not. In particular, raters agreed to include (a) recommendations related to services a PCP might provide or order without active specialty involvement (even if some PCPs might just refer to specialist for an evaluation), (b) recommendations related to services that would be provided whenever a PCP ordered them (i.e., specialty involvement is not required), and (c) recommendations related to a referral that a PCP would initiate. We excluded recommendations for which a PCP would lack sufficient information or expertise to act upon independently.
The raters then graded each recommendation on its likelihood that overuse could cause harm (to either patients or society) and the degree of societal cost (based on prevalence as well as immediate and downstream costs). Differences in ratings were discussed and resolved by consensus. We selected 12 of the set of 41 highest-ranking recommendations (Table
1)—four diagnostic testing, four screening, and four medications—for inclusion in both surveys based on these ratings, eliminating duplicates from different societies.
Table 1
The 12 Choosing Wisely® Recommendations Studied
Recommendations related to diagnostic testing |
• Don’t do imaging for low back pain within the first 6 weeks unless red flags are present. |
• In the evaluation of simple syncope and a normal neurological examination, don’t obtain brain imaging studies (CT or MRI). |
• Don’t image for suspected pulmonary embolism (PE) without moderate or high pre-test probability. |
• Avoid cardiovascular testing for patients undergoing low-risk surgery. |
Recommendations related to screening |
• Do not repeat colorectal cancer screening (by any method) for 10 years after a high-quality colonoscopy is negative in average-risk individuals. |
• Don’t screen for carotid artery stenosis (CAS) in asymptomatic adult patients. |
• Don’t use dual-energy x-ray absorptiometry (DEXA) screening for osteoporosis in women younger than 65 or men younger than 70 with no risk factors. |
• Do not repeat colonoscopy for at least 5 years for patients who have one or two small (<1 cm) adenomatous polyps, without high-grade dysplasia, completely removed via a high-quality colonoscopy. |
Recommendations related to medications |
• Avoid using medications to achieve hemoglobin A1c <7.5 % in most adults age 65 and older; moderate control is generally better. |
• Don’t use antimicrobials to treat bacteriuria in older adults unless specific urinary tract symptoms are present. |
• Don’t use benzodiazepines or other sedative-hypnotics in older adults as first choice for insomnia, agitation, or delirium. |
• Don’t routinely prescribe antibiotics for acute mild-to-moderate sinusitis unless symptoms last for 7 or more days, or symptoms worsen after initial clinical improvement. |
The U.S. Survey included a brief description of the Choosing Wisely initiative and a focal section with the 12 selected CW recommendations. For each recommendation, respondents answered two Yes/No questions: (a) “Will it be difficult for YOU to follow this recommendation most of the time?” and (b) “Will it be difficult for MOST PATIENTS to accept this recommendation in practice?” Respondents also (a) rated their overall familiarity with the CW initiative on a five-point scale from “not at all familiar” to “extremely familiar”; (b) answered a set of questions on practice characteristics; and (c) rated eight possible barriers to efforts to reduce overuse of inappropriate services in general, rating each as a major barrier, a minor barrier, or not a barrier.
For the similarly structured VA Survey, the key questions were worded “How easy or difficult will it be for YOU to follow this recommendation most of the time?” and “How easy or difficult will it be for MOST PATIENTS to accept the recommendation in practice?”, with a four-point response scale labeled “very easy,” “somewhat easy,” “somewhat difficult,” and “very difficult.” We collapsed “somewhat” and “very” responses to create a dichotomous variable for our analyses. In addition, we (a) adjusted the practice characteristic questions to reflect relevant categories of clinicians within the Veterans Health Administration, (b) eliminated a question about payment policies that reward ordering more services, and (c) changed the response scale on the question about familiarity with the CW initiative to a three-point scale (“not at all familiar,” “somewhat familiar,” and “very familiar”).
We report the proportion of clinicians in each sample (unadjusted) reporting particular CW recommendations as difficult to follow, difficult for patients to accept, or both. We also report the proportion endorsing particular barriers to reducing overuse. Based on the observed distributions, we then grouped recommendations with similar response patterns. We also examined correlations between PCPs’ reporting of difficulty following each CW recommendation and their ratings of each of the eight general barriers to reducing overuse.