Selection of recommendations
Our search on websites of international Choosing Wisely initiatives resulted in 147 reliable recommendations. We excluded 44 duplicates (identical recommendations from various medical societies), 24 recommendations with similar content, and seven recommendations that were already covered by the top‑5 list of the Society of Geriatrics and Gerontology . From the remaining pool of 72 recommendations, a team of ten experts from the Austrian Society of General Practice and Family Medicine selected those that were most relevant for general practitioners. A total of 49 recommendations were excluded from the pool due to their relevance for medical specialities other than general practice (Fig. 1). Finally, a pool of 23 recommendations was available for further assessment and for the selection of the top 5. Three of the recommendations related to antibiotic therapy for respiratory tract infections and were therefore combined, so that 21 recommendations were available for the final Delphi survey. Fig. 1 illustrates the selection of recommendations for the pool of the top‑5 list. The top‑5 list was created in a two-step Delphi process, which took place from July 2018 to September 2018.
First Delphi round
Eleven experts assessed the pool of 21 recommendations on a Likert scale from 1 (least important) to 5 (most important). Not performing imaging studies for nonspecific low back pain achieved the highest mean value among the eleven experts (4.5 ± 1.0), with eight of eleven rating it most important. Three recommendations contained the advice to avoid antibiotics for different indications (Table 1). Avoid prescribing antibiotics for upper respiratory tract infections was rated a 4 or a 5 (very or most important) by eight of the eleven experts. The other two indications for which antibiotics should not be used routinely were otitis media in children aged 2 to 12 years and asymptomatic bacteriuria. The fifth recommendation was indirectly linked to the unnecessary prescription of antibiotics and was about not obtaining a urine culture if there are no symptoms indicating a urinary tract infection. There was a broad consensus within the team on the importance of the first five recommendations with mean scores above 4.0 (4.1–4.5; standard deviation, SD: 0.7–1.2). The recommendation to avoid antibiotics for upper respiratory tract infections achieved a mean score of 4.2; however, the standard deviation was 1.2 as one expert rated this item as a 1.
Second Delphi round
In order to establish a consensus regarding the top‑5 recommendations, a second survey was necessary. All the experts from the first Delphi round reevaluated the 21 recommendations. The four first-ranked recommendations were each identified twice (in the first and second Delphi round) by a majority as the four most important recommendations with mean scores of 4.3–4.6 (standard deviations 0.5–1.0). The fifth recommendation varied in the two rounds. In the second round, the experts chose antihypertensive treatment in older persons as the fifth recommendation. Since this recommendation was subject to international discussions, the team agreed that the sixth recommendation concerning routine PSA screening should take precedence, which was rated as 4 or 5 by nine of eleven experts (mean score: 4.1; SD: 0.9).
Survey: reasons for the selection of recommendations and expectations for the campaign
Ten experts were asked to rate five blocks of selection criteria for each of the five recommendations according to their importance on a scale of 1 (very good reason) to 5 (no reason at all). Two questions about the expectations of the campaign were agreed upon. The survey was performed from October to November 2019 among all experts involved in the Delphi process, excluding R. S., one of first authors of this study.
Table 2 illustrates the results. In the following section we present the mean values and the corresponding standard deviations.
Frequent topic in daily practice
The argument Frequent topic in daily practice and therefore particularly important was considered relevant for the selection process by all ten experts for all recommendations in the top‑5 list. This concern yielded the highest overall agreement, ranging from a weighted average of 1.4 (± 1.2) for Avoid prescribing antibiotics for upper respiratory tract infections, which had achieved the highest mean value in the Delphi process, to 1.9 (± 0.9) for the recommendation not to perform routine screening to detect prostate cancer.
The criterion Improper decisions have a special impact on patients’ health met with the second highest overall approval (range: 1.4 ± 0.7 to 2.5 ± 1.2). For all three recommendations concerning unnecessary use of antibiotics, experts judged possible harm as a good or very good reason for the selection. Potentially negative impact on health was considered a major concern by eight of ten experts (mean value: 1.8 ± 1.0) for routine screening for prostate cancer.
The argument Improper decisions have a special impact on costs was altogether attributed less importance in choosing the recommendations for the top‑5 list. The highest impact of costs was seen for the recommendation concerning imaging studies for non-specific low back pain, with nine experts out of ten agreeing (mean value: 1.6 ± 0.7).
Three of the recommendations on the top‑5 list were also chosen by experts because medical uncertainties could be a reason for non-adherence to the guidelines. Seventy percent (7 of 10) suspect that family doctors were unsure (rating 1 or 2) about whether or when to recommend imaging studies for non-specific low back pain (mean value: 2.3 ± 1.0). Sixty percent (6 of 10) also stated that uncertainty could lead to more antibiotics being prescribed unnecessarily for children with otitis media (mean value: 2.3 ± 0.9) or for asymptomatic bacteriuria (mean value: 2.5 ± 0.7).
Pressure on doctors through requests from other parties
The majority of our experts (8 of 9) agreed or strongly agreed that patient pressure has a role in driving decisions against the recommendation not to perform imaging studies for nonspecific low back pain (mean 1.4 ± 0.7). Consistently, nine out of ten experts agreed that patients often expect physicians to routinely screen for prostate cancer using PSA testing (mean value: 1.7 ± 0.6). Also, the prescription of antibiotics for upper respiratory tract infections or otitis media in children is often demanded by patients or parents, which may contribute to overuse (Table 2).
Pressure from specialists influenced decisions related to performing imaging studies for non-specific low back pain (mean value: 2.1 ± 1.6) against recommendations. Seven out of nine experts agreed or strongly agreed with this, while two experts disagreed. Four of the experts explicitly mentioned pressure from physiotherapists. Nine of ten voters agreed or strongly agreed that specialists enhance decisions to deviate from the recommendation not to routinely screen for prostate cancer (mean value: 1.7 ± 0.9).
Reducing the time required for education
The greatest agreement of the potential of the Choosing Wisely campaign to save time in educating patients was achieved by the recommendation not to perform routine screening for prostate cancer (8 of 9). Eighty percent of the experts thought that the campaign would save time when explaining why antibiotics should not be prescribed for upper respiratory tract infections (the recommendation ranked first in the Delphi process). Seventy percent expected this for the second-ranked recommendation no imaging should be performed for non-specific low back pain (7 of 10) and 56% (5 of 9) for the recommendation not to administer antibiotics for children with non-severe otitis media.
Reduction of external pressure
Nine voters expected reduced pressure on decision-making processes as a result of the campaign for the recommendation not to prescribe antibiotics for non-severe otitis media in children above 2 years, and eight of those nine experts expected less pressure with respect to the recommendation not to routinely perform PSA screening.
Legal support was an issue for all the selected recommendations (Table 2). For example, eight of the ten experts agreed or strongly agreed that the Choosing Wisely campaign will provide legal protection concerning two of the recommendations regarding antibiotics: not to prescribe antibiotics for upper respiratory tract infections (mean value: 2.1 ± 1.1) and for otitis media in children aged 2–12 years with non-severe symptoms (mean value: 1.8 ± 0.8).