Medical Decision Making最新文献

BackgroundMessaging strategies hold promise to reduce breast cancer overscreening. However, it is not known whether they may have differential effects among medical maximizers who prefer to take action about their health versus medical minimizers who prefer to wait and see.MethodsIn a randomized controlled survey experiment that included 2 sequential surveys with 3,041 women aged 65+ y from a US population-based online panel, we randomized participants to 1) no messages, 2) single exposure to a screening cessation message, or 3) 2 exposures over time to the screening cessation message. We assessed support for stopping screening in a hypothetical patient and intention to stop screening oneself on 7-point scales, where higher values indicated stronger support and intentions to stop screening. We conducted stratified analyses by medical-maximizing preference and moderation analysis.ResultsOf the women, 40.7% (n = 1,238) were medical maximizers; they had lower support and intention for screening cessation in all groups compared with the medical minimizers. Two message exposures increased support for screening cessation among medical maximizers, with a mean score of 3.68 (95% confidence interval [CI] 3.51-3.85) compared with no message (mean score 2.20, 95% CI 2.00-2.39, P < 0.001). A similar pattern was seen for screening intention. Linear regression models showed no differential messaging effect by medical-maximizing preference.ConclusionsMedical maximizers, although less likely to support screening cessation, were nonetheless responsive to messaging strategies designed to reduce breast cancer overscreening.HighlightsIt is not known if a message on rationales for stopping breast cancer screening would have differential effects among medical maximizers who prefer to take action when it comes to their health versus medical minimizers who prefer to wait and see.In a 2-wave randomized controlled survey experiment with 3,041 older women, we found that medical maximizers, although less likely to support screening cessation compared with medical minimizers, were nonetheless responsive to the messaging intervention, and the magnitude of the intervention effect was similar between maximizers and minimizers.Medical maximizers reported higher levels of worry and annoyance after reading the message compared with the minimizers, but the absolute levels of worry and annoyance were low.Our findings suggest that messaging can be a useful tool for reducing overscreening even in a highly reluctant population.

BackgroundMedication adherence is a critical factor in hypertension management, which remains a challenge for public health systems.MethodsGraded-pair questions were used to quantify the perception of how much nonadherence to antihypertensives increases the risk of serious cardiovascular events. A discrete-choice experiment was used to quantify the relative importance of medication outcomes (e.g., reduction in cardiovascular event risk and medication side effects). Rating questions were used to assess perspectives of the effect of treatment nonadherence on treatment side effects. Results were combined to assess how preferences and outcome expectations influence adherence.ResultsPatients perceived treatment adherence as the most significant contributor to cardiovascular event risk. A reduction in cardiovascular risk was the most significant consideration when choosing medication. Missing consecutive (v. alternate) doses was associated with greater perceived cardiovascular risk and fewer side effects. The differences between complete adherence and any level of nonadherence were significantly larger for side effects than for changes in the risk of cardiovascular events, suggesting that side effects are perceived to be more sensitive to nonadherence than treatment efficacy.LimitationsOur study relied on hypothetical scenarios, which may not fully capture real-world decision making. While our findings shed light on the relationship between adherence patterns and treatment perceptions, it is essential to recognize the complexity of adherence behavior.ConclusionsPatients believe that they can manage medication side effects by skipping doses without compromising the efficacy to the same degree and that they can offset compromises in efficacy by avoiding missing consecutive doses for prolonged periods.ImplicationsHealth care providers should understand the importance of patient education and counseling to address misconceptions and promote realistic expectations regarding treatment efficacy and the consequences of nonadherence.HighlightsThe average patient believes that they can manage medication side effects by skipping doses without compromising the efficacy to the same degree.There is a belief that patients can offset some of the impact of nonadherence on their cardiovascular event risk, particularly if they avoid missing consecutive doses for prolonged periods of time.This highlights the importance of patient education and counseling to address misconceptions and promote realistic expectations regarding treatment efficacy and the consequences of nonadherence.

ObjectiveTo study the effects of exposure to a prior diagnosis (PD) on second opinions in breast pathology.Materials and MethodsPathologists interpreted digital breast biopsy cases in 2 phases separated by a washout. Phase 2 interpretations were randomly assigned to PD or no PD. When presented, PD was always more or less severe than a participant's phase 1 diagnosis. Viewing behaviors, including zoom level, were recorded during all interpretations. Twenty pathologists yielded 556 interpretations of 32 different cases.ResultsPathologists were 71% more likely to give a less severe diagnosis when exposed to a less severe PD than with no PD (RR 1.71, 95% CI 1.33-2.20, P < 0.001). In comparison, when exposed to a more severe PD than with no PD, pathologists were 27% more likely to give a more severe diagnosis, but the effect was not significant (RR 1.27, 95% CI 0.87-1.86, P = 0.223). Compared with no PD, viewing behavior shifted toward more focus on critical image regions with exposure to a less severe PD and toward higher zoom levels with exposure to a more severe PD.DiscussionResults indicate anchoring and confirmation biases from PD exposure, such that second opinions after PD exposure are not independent assessments. Viewing behaviors illustrated how PD alters the interpretive process, including increased zooming when exposed to a more severe PD. Results have implications for best practices for computer-aided diagnosis tools.ImplicationsWhen giving a second opinion, exposure to a PD can sway diagnostic classifications and alter interpretive behavior, highlighting a need for protocols that encourage independent assessments.HighlightsIn pathology diagnosis, second opinions are systematically influenced by prior diagnostic information.Less severe prior diagnoses shift pathologists' visual attention toward clinically critical regions of a pathology image, whereas more severe prior diagnoses tend to elicit increased magnification during case interpretation.Specific viewing behaviors partially mediate the effect of prior diagnoses on second opinion diagnoses.When prior diagnoses are disclosed to pathologists, anchoring and confirmation biases undermine the independence of second opinion decisions.

Background. Awareness of shared decision making (SDM) is growing, but its integration into clinical practice guidelines (CPGs) remains challenging. We sought expert insights to identify strategies for more successfully integrating SDM and decision support tools into CPGs. Specifically, our objectives were to determine 1) how to identify CPG recommendations where SDM is most relevant and 2) what factors in CPG development hinder or facilitate the consideration of SDM and the development of decision support tools. Methods. We conducted semi-structured interviews with experts on CPGs and SDM. We analyzed the data using Mayring's qualitative content analysis. Results. The 16 interviewed participants proposed several determinants of and strategies for identifying SDM-relevant recommendations. The most frequently mentioned determinant was "multiple options with benefits and harms where choices depend on individual preferences." The most frequently mentioned strategy was prioritization, similar to the CPG scoping phase. Participants highlighted the role of patient partners in facilitating the consideration of SDM in CPG development but noted that a supportive culture toward both patient and public involvement and SDM is needed. The absence of standardized methods and inadequate resources hinder the consideration of SDM and the combined development of CPGs and decision support tools. The current format of CPGs was deemed overwhelming, while the inclusion of choice awareness in CPG recommendations could facilitate SDM. Conclusions. The identified strategies provide a starting point for CPG organizations to explore ways for integrating SDM and decision support tools into CPGs while considering context-specific barriers and facilitators. Implications. Further research is needed to assess the usefulness and feasibility of the proposed strategies. New policies and stronger collaboration between CPG and SDM communities appear to be needed to address identified barriers.HighlightsWe explored expert knowledge and experience on how to successfully integrate shared decision making (SDM) and decision support tools into clinical practice guidelines (CPGs).A combined development of CPGs and decision support tools was deemed essential; however, development processes often remain separate, with the CPG development group unaware of the decision support tool development group, and vice versa.In addition to stating choice awareness in CPGs, participants highlighted the critical role of patient partners in considering SDM in CPG development, but resource issues and a culture that neglects patient involvement and SDM remain.For CPG development groups to consider SDM and for health care professionals to practice it, things need to be as easy as possible.

BackgroundWhile machine learning (ML) models are increasingly used to predict outcomes in health care, their practical effect on health care operations, such as bed capacity management, remains underexplored. There is a variety of traditionally used evaluation metrics to analyze ML models; however, decision makers in health care settings require a deeper understanding of their implications for resource management. Traditional performance measures often fail to provide this practical insight.MethodsIn this work, we conduct a simulation study to evaluate the impact of ML-driven length-of-stay (LOS) predictions on intensive care unit (ICU) bed capacity management. Two classification models differing in terms of explainability and interpretability, logistic regression (LR) and extreme gradient boosting (XGB), are applied to predict ICU-LOS. We use the HiRID dataset containing high-frequency data of more than 33,000 patients. The predictions of the ML models are integrated into a simulation framework that replicates real-world ICU bed management, allowing for the assessment of the practical implications of using these algorithms in a clinical setting.ResultsThe application of both classification models results in improved capacity control regarding the key performance indicators in the simulation study, with XGB outperforming LR. While LR leads to slight overoccupancy in the ICU, slight underoccupancy can be observed when XGB is applied.ConclusionOur study bridges the gap between predictive accuracy and practical application by emphasizing the importance of evaluating ML models within the context of ICU capacity management. The simulation-based approach offers a more relevant assessment for health care practitioners, providing actionable insights that go beyond classical performance measures and directly address the needs of decision makers in clinical practice.HighlightsWe apply multiple classification models for ICU-LOS prediction using time-series data. This approach enables an update of the initial prediction resulting in the possibility of efficiently managing intensive care capacities.We present a simulation-based approach to evaluate ML algorithms and their impact on bed capacity management in real-world clinical settings.Our work provides in-depth insights into the impact of using ML techniques as decision support systems in the ICU and can lead to increased acceptance in practice.

BackgroundDecision aids (DAs) are evidence-based tools to improve patient-centered care, but their use in routine care is limited. The purpose of this project was to work with orthopedic practices to deliver DAs.MethodsEligible sites needed to identify an administrative and clinical champion and have access to DAs for treatment of hip, knee, and/or spine conditions. The implementation strategies included an Orthopaedic Learning Collaborative (OLC), external facilitation, and audit and feedback. The project was conducted over 15 mo with 5 OLC sessions, individual monthly meetings, and monthly data reports. Clinicians and staff completed a baseline survey prior to the start of the project. Sites provided details on their DA workflow and number of DAs delivered. We calculated adoption (the number of specialists who used DAs) and estimated reach (percentage of eligible patients who received DAs). We calculated descriptive statistics and explored predictors of reach.ResultsTwelve participating sites had an average annual orthopedic surgical volume of 550, half were academic medical centers, and some (4/13, 30.7%) had prior experience with orthopedic DAs. Adoption was 76% (60/79 physicians). Sites distributed 9,626 DAs and reached 44% of eligible patients (range 7%-100%). Sites that indicated at baseline that DA delivery was a high priority for staff had higher reach (60% reach for high v. 47% for moderate v. 9% for low priority, P = 0.21). Sites with no prior experience with DAs had higher reach than those with prior experience did (60% v. 38%, P = 0.26, d = 0.71).ConclusionsParticipating sites were able to implement workflows that reached about half of eligible patients. Establishing DA delivery as a priority for staff at the outset appears important for reach, while prior experience does not.HighlightsThe 12 sites were able to reach, on average, 44% of eligible patients with decision aids in routine care demonstrating feasibility of distribution.The study and associated implementation toolkit provide concrete examples of workflows for orthopedic practices interested in incorporating decision aids into routine care.A bundle of implementation strategies, including a learning collaborative, external facilitation, and audit and feedback, helped most sites meet targets for decision aid implementation.

In health or medical studies, participants can often experience the outcome(s) of interest multiple times during the observation period, creating recurrent event data. Depending on the primary research objective, advanced statistical methods are required to correctly analyze this special type of data. This tutorial discusses 4 general frameworks, appropriate for analyzing recurrent events data: 1) extended Cox, 2) parametric survival, 3) longitudinal, and 4) multistate models. We present in detail the implementation of these methods, including a description of the required dataset structure, R code, and interpretation of results, using data from the CTN-0051 study, a randomized clinical trial comparing the effectiveness of opioid use disorder treatments. The objectives of 3 use case scenarios exemplify the usage and relevance of the methods for the analysis of recurrent events: 1) estimate adjusted effects, 2) make individual-level predictions, and 3) model a complicated process involving multidirectional transitions between disease states. We compare the methods, comment on their strengths and limitations, and make recommendations on the preferred method depending on the primary research objective.HighlightsRecurrent events are a common phenomenon in experimental research settings, and their analysis requires advanced survival modeling approaches. This tutorial aims to explain and make these approaches more accessible with code and detailed instructions.We compare a detailed list of statistical methods for analyzing recurrent events and make suggestions on which one should be used depending on the study objective.This tutorial will enable researchers to make better use of recurrent events data.

BackgroundResource scarcity during large-scale crises, such as pandemics, can increase the emphasis on efficiency in medical decision making. However, it remains unclear whether such shifts are primarily driven by the direct experience of scarcity or by the way in which ethical principles for health care priority setting are expressed in the context of a crisis. This study investigates whether a national crisis affects public support for health care priority-setting principles and whether abstract versus concrete formulations of these principles shape that support.DesignWe conducted a preregistered online experiment (N = 1,404) to examine public attitudes toward three ethical principles formalized in the Swedish ethical platform-human dignity, needs-solidarity, and cost-effectiveness-in both crisis and noncrisis contexts. We also manipulated how the principles were presented, using either abstract or concrete formulations.ResultsIn the crisis condition, support for the human dignity and cost-effectiveness principles decreased, while support for the needs-solidarity principle increased. However, these effects were small, and the overall ranking of the principles remained stable. Notably, the level of abstractness had a stronger impact than the crisis context did: support for needs solidarity was higher when described abstractly, whereas support for cost-effectiveness increased when it was presented in a more concrete, action-oriented way. Support for the human dignity principle was unaffected by the abstractness manipulation.ConclusionThe findings suggest that people's moral views are relatively stable in the face of crisis. Rather than the crisis context itself, the way ethical principles are formulated-abstractly or concretely-may be a more powerful driver of shifts in public support for different moral values in health care priority setting.HighlightsPublic support for ethical principles remained largely stable during a simulated national crisis.The level of abstraction in how principles were presented strongly influenced support.Support for needs-solidarity increased in a crisis, while cost-effectiveness support declined.The way ethical principles were formulated had a greater impact than the presence of a crisis.