Highlights
– Presenting two or more appropriate treatment alternatives to primary care physicians increased the odds they would adopt an alternative instead of remaining with the current plan (adjusted OR 1.90; 95% CI, 1.09–3.30).
– The effect was larger for opioid prescribing for back pain (OR 2.95) than for surgical referral for hip osteoarthritis (OR 1.56).
– Increasing the number of alternatives beyond two (to three or four) did not produce additional benefit.
Background
Clinical decision support (CDS) and behavioral “nudges” are increasingly embedded in electronic health records (EHRs) to guide clinicians toward guideline‑concordant care. How alternatives are presented within alerts or order sets—both their number and framing—matters for decision outcomes. Classic behavioral studies in non‑clinical domains have described choice overload and a status‑quo bias, suggesting that offering many options or altering the default may paradoxically reduce the probability of change (Samuelson & Zeckhauser, 1988; Iyengar & Lepper, 2000). Whether these behavioral phenomena translate to clinician decision‑making in routine primary care and across clinical domains (e.g., opioid prescribing versus surgical referral) has important implications for CDS design, patient safety, and quality improvement.
Study design
This randomized clinical trial (May 3–8, 2024) enrolled 402 U.S. primary care physicians recruited via the Qualtrics research network. Participants were randomized 1:1 to a control arm or an intervention arm. Each physician reviewed two short clinical scenarios: one concerning a decision about surgery referral for hip osteoarthritis and one concerning opioid prescribing for back pain. In the control arm, clinicians were shown a single appropriate treatment alternative for each scenario. In the intervention arm clinicians were shown a choice set containing two, three, or four appropriate alternatives. For each scenario clinicians decided whether to remain with the current management plan or select one of the provided alternatives.
The primary outcome was the proportion of decisions in which clinicians selected a presented alternative versus staying with the current plan. A secondary analysis evaluated incremental effects of adding each additional alternative within the intervention arm (i.e., from two to three to four options).
Key findings
Overall effect
Across 804 total treatment decisions (400 in control, 404 in intervention), clinicians in the intervention group selected an alternative in 62.1% of decisions (251/404) compared with 44.0% (176/400) in the control group. After adjustment, presenting multiple alternatives was associated with higher odds of selecting an alternative (adjusted OR 1.90; 95% CI, 1.09–3.30; P = .02).
By clinical scenario
The effect magnitude varied by scenario. For opioid prescribing decisions, 56.4% of intervention‑arm decisions selected an alternative (114/202) versus 30.5% in the control arm (61/200), yielding an OR of 2.95 (95% CI, 1.96–4.45). For the surgery referral scenario, 67.8% of intervention decisions selected an alternative (137/202) versus 57.5% in control (115/200), OR 1.56 (95% CI, 1.04–2.34).
Number of alternatives
Within the intervention arm, increasing the number of suggested alternatives beyond two (to three or four) did not produce incremental increases in the likelihood of choosing an alternative. In other words, the key contrast was between one versus two or more alternatives; adding a third or fourth option did not raise adoption further.
Population and generalizability
The sample included 402 physicians from 46 states; 50% identified as male and 57.5% had under 10 years of clinical experience. Approximately half practiced in urban/metropolitan areas. Decisions were elicited in brief vignette format rather than real‑world EHR encounters.
Interpretation and implications
This trial challenges an assumption derived from consumer choice literature that offering multiple options will necessarily increase decision paralysis or reinforce status‑quo bias among clinicians. Instead, offering two or more recommended alternatives increased the probability that primary care physicians would switch from an existing plan to a guideline‑concordant alternative. Several mechanisms could explain this finding.
First, multiple alternatives may normalize change by signaling that several acceptable approaches exist and that changing management is reasonable. Second, presenting a small set of vetted options reduces the cognitive burden associated with generating alternatives independently, effectively lowering the activation energy required to change course. Third, in high‑stakes or highly scrutinized domains—such as opioid prescribing—multiple recommended options may confer perceived safety or medicolegal support for de‑escalation or non‑opioid strategies, increasing clinician willingness to change.
The stronger effect in the opioid prescribing scenario is noteworthy and clinically relevant. Opioid prescribing has been a major target for stewardship and guideline implementation; decision aids that present multiple non‑opioid or multimodal pain‑management alternatives may be particularly effective in reducing inappropriate opioid initiation or continuation.
From a CDS design perspective, the results suggest that alerts and order sets might be more persuasive when they present clinicians with two appropriate options (e.g., a recommended non‑opioid strategy plus a referral to physical therapy) rather than a single recommended option or a long menu of choices. However, adding more than two options did not increase adoption in this study, so simple, limited choice sets appear sufficient.
Expert commentary and situating the results in the literature
Behavioral economics and decision science have long emphasized defaults and choice architecture. Samuelson and Zeckhauser’s work on status‑quo bias highlighted a tendency to stick with current choices under uncertainty; Iyengar and Lepper’s choice‑overload research demonstrated that very large choice sets can reduce selection in consumer contexts. Translating these findings to clinicians, however, requires accounting for differences in expertise, accountability, and the normative context of medical decision‑making.
Previous evidence supports that CDS can improve practitioner performance when it provides actionable, timely, and specific recommendations (Garg et al., 2005). The current trial refines that understanding by isolating the effect of the number of suggested alternatives and demonstrating that a modest expansion from one to two or more appropriate options can increase clinician uptake.
Limitations
Several limitations temper the findings. The study used vignette‑based decisions rather than observation of behavior in routine clinical workflows; real‑world EHR integration, time pressures, workflow interruptions, and patient preferences may alter effects. Participants were recruited from an online panel and may not fully represent the broader population of U.S. primary care physicians; selection bias is possible. The outcomes measured were choices in hypothetical scenarios rather than patient‑level outcomes such as symptom improvement, adverse events, or downstream health care utilization. The trial also did not evaluate long‑term adherence to changes or unintended consequences such as increased workload or alert fatigue in deployed CDS systems.
Practical recommendations and future research
For clinical informaticists, quality leaders, and policymakers designing decision support: consider presenting a small set (two) of evidence‑based alternatives within alerts or order sets rather than a single option. When feasible, tailor the alternatives to the clinical context and include succinct rationale to reinforce appropriateness. Avoid presenting long, uncurated menus of choices that may add cognitive burden without benefit.
Future research should test these choice‑set effects in live EHR environments across diverse clinical domains and measure downstream patient outcomes, clinician workflow consequences, and persistence over time. Investigations into how framing (e.g., comparative risks, normative statements, or defaults) interacts with option number would further refine optimal CDS design.
Conclusion
This randomized clinical trial among primary care physicians demonstrates that offering two or more appropriate treatment alternatives increases the likelihood clinicians will choose an alternative to their current management plan, with the largest effect observed for opioid prescribing decisions. The study suggests that modestly expanding curated, evidence‑based option sets in clinical decision support may enhance adoption of recommended care without the penalty of choice overload—challenging simple extrapolations from consumer choice psychology to clinical decision‑making.
Funding and trial registration
Trial registration: ANZCTR Identifier: ACTRN12625001025426. Funding information was not reported in the provided summary.
References
Altinger G, Maher CG, Jones CMP, Collins J, Linder JA, Bell KJL, Lin CC, Tracy M, Boroumand F, Traeger AC. Multiple Suggested Care Alternatives and Decision‑Making of Primary Care Physicians: A Randomized Clinical Trial. JAMA Netw Open. 2025 Nov 3;8(11):e2542949. doi:10.1001/jamanetworkopen.2025.42949. PMID: 41231470; PMCID: PMC12616461.
Samuelson W, Zeckhauser R. Status quo bias in decision making. Journal of Risk and Uncertainty. 1988;1(1):7–59.
Iyengar SS, Lepper MR. When choice is demotivating: can one desire too much of a good thing? Journal of Personality and Social Psychology. 2000;79(6):995–1006.
Garg AX, Adhikari NKJ, McDonald H, Rosas‑Arellano MP, Devereaux PJ, Beyene J, Sam J, Haynes RB. Effects of computerized clinical decision support systems on practitioner performance and patient outcomes: a systematic review. JAMA. 2005;293(10):1223–1238.
