In last month’s issue of The Hospitalist, members of SHM’s Performance Measurement and Reporting Committee wrote about attribution challenges. Building on that article, the use of quality measures to improve overall hospital or health system performance needs to be deliberate and smartly executed. Our colleagues discussed issues with attribution, including “gaming” and burnout. Gaming occurs when hospitalists are incentivized to prioritize improving the metrics attributed to them, potentially at the expense of actual patient-care quality that may not be sufficiently rewarded within the hospitalist group or healthcare system. Additionally, the pressures of inappropriate metrics and attribution can lead to feelings of helplessness, disillusionment, and moral injury for practitioners.
Given these challenges, hospital medicine groups need metrics that are fair, clinically meaningful, and clear enough to guide improvement rather than simply track performance. In this article, we discuss how to attribute quality metrics successfully in hospital medicine and offer a practical blueprint for choosing measures, assigning responsibility, and revising attribution methods as care models evolve.
Principles of Attribution
Even if imperfect, hospitalists need individualized feedback and a clear understanding of the quality of care they provide. Attribution decisions should begin by asking whether a metric is best assigned to an individual hospitalist, shared among clinicians involved in the hospitalization, or evaluated at the group level when individual control is limited. Multiple studies have evaluated methodologies to improve attribution, and while all have flaws, some show higher fidelity than others. Below, we offer principles to help hospital medicine leaders select attribution methods that are specific enough to be fair, practical enough to sustain, and meaningful enough to improve care.
Principle 1: A hospitalist’s metric should be as specific to their individual performance as possible.
Typically, multiple hospitalists care for patients during a hospital stay, complicating individual attribution. One way to address this issue is to evaluate the performance metrics of hospitalists based solely on cases where only one hospitalist was responsible for the patient’s care. However, doing so would considerably reduce the sample size (especially for longer hospitalizations), which would significantly limit the effectiveness of the metric. A study by Carmichael et al. compared five different methodologies for attribution of length of stay (LOS) to individual hospitalists.1 They found that their new metric was the best performing of the five in the sensitivity analysis. It is named “Intermountain Method of Provider Attributed Quality (IMPAQ)” and it attributes the patient’s LOS to the discharging clinician only if they oversaw more than 30% of the encounter. This performed better than a more traditional (standard) method, which attributes LOS to the discharging physician even if they cared for the patient for a very short amount of time. For hospitalist leaders, the practical lesson is to reserve individual attribution for measures where the clinician had meaningful influence and to use group-based attribution when responsibility is diffuse. Metrics of interest to a hospitalist group or health system that cannot reliably be attributed to an individual should use group-based attribution strategies as much as possible.
Principle 2: Appropriate data sources should be selected for a metric to ensure attribution fidelity.
Metrics are often attributed to hospitalists even when a large part of the metric applied is outside of the hospitalist’s control. A 2017 study used billing data for attribution, linking patients’ metrics to the billing physician.2 They chose different methodologies of attribution depending on the metric and found meaningful discrepancies in physicians’ scores relative to their peers. This study attributed admission-type data (venous thromboembolism prophylaxis order) to the hospitalist who billed for the history and physical while attributing the discharge-type data (discharge time, readmissions, etc.) to the hospitalist billing the discharge. They attributed shared credit according to the percentage of bills for other metrics associated with the whole hospitalization (LOS, patient satisfaction, etc.). This methodology improved attribution fidelity, tying the metric studied to the most likely responsible party. The key implication is that the data source should match the clinical moment being measured, so admission, discharge, and whole-stay metrics are not all assigned using the same blunt method.
Principle 3: Hospitalists should create and own the attribution and subsequent evaluation process.
Most important to the process of selecting and implementing a quality metric program within your group is buy-in from frontline hospitalists. Clinicians need to know that what matters to them and their patients also matters to hospital leadership. In one such example, Patel and colleagues surveyed hospitalists and conducted “design sessions” and pilots to create a dashboard of hospital medicine metrics.3 They found that hospitalists “preferred collaboration over competition and internal motivation over external incentives,” and hospitalists voted for clinical and patient-centered metrics as the most valuable and likely to change behavior. They also highlighted the importance of choosing metrics that spoke to challenges at their own institution and that hospitalists could actually control, further supporting Principle 2. In practice, attribution systems are more likely to change behavior when hospitalists help define the measures, understand how the scores are generated, and see a credible path from feedback to improvement.
Principle 4: There must be ongoing monitoring of metrics and attribution strategy, including monitoring of balancing measures.
A major issue in choosing metrics and attribution is that, often, the measure does not positively impact the outcome in question. Our colleagues use the example of LOS and discharge before noon (DBN). Studies have shown no correlation or even a negative correlation between DBN and LOS.4 Even LOS-reduction metrics can be harmful if too aggressive. Many hospital medicine groups still track the DBN metric despite these data. The reasoning for this is likely complex, but among these is the issue that a measure that positively impacts LOS has been elusive and difficult to monitor regularly. Monitoring any metric may require time, people, money, and a robust electronic health record. Any hospital deciding to implement a quality-based program for incentive or other performance evaluation needs to have a sound monitoring system prior to implementation. Leaders should also identify balancing measures in advance so that improvement in one metric does not unintentionally worsen another outcome, such as readmissions, patient experience, or clinician workload.
Conclusion
We posit that no hospitalist wrote about reducing LOS in their personal statement for medical school. The ongoing challenge for leaders in hospital medicine is to connect measurement to the clinical “why” that motivates hospitalists: improving care for the patients in front of them.
Without consideration of both the frontline practitioner and the health system in which they operate, quality metrics can feel punitive and lead to burnout, moral injury, dissatisfaction with work, and attrition. But quality-based evaluation will likely continue to expand in hospital medicine, making it even more important to choose measures that identify and spread high-quality care rather than reward documentation artifacts or arbitrary targets.
Data suggest that more experienced hospitalists have better quality outcomes.5,6 When determining which metrics matter for one’s own group of hospitalists, it may be useful to focus on what those experienced or otherwise high-functioning clinicians do very well. Sharing these best practices through educational initiatives may provide a “free” resource for many hospitalist groups to improve quality without creating a culture of unfriendly competition or questionable incentive practices. In fact, a study by Yurso et al. showed improvement in quality metrics through the implementation of educational modules, which could also be coupled with grand rounds series, mentorship programs, and other educational interventions.7
In conclusion, we recommend meaningful quality measures that are focused on the individual organization’s goals and environment and that are created by hospitalists for hospitalists. Any metric considered should involve outcomes or processes within the hospitalist’s meaningful control, be attributed at the right level—individual, shared, or group-based—and be designed to improve care rather than track performance for its own sake. Hospital medicine groups should use the principles outlined above to choose the metric and attribution method, monitor the strategy over time, and adjust or abandon it when the measure no longer advances fair and clinically meaningful improvement.
Dr. Bruti
Dr. Humes
The authors are members of SHM’s Performance Measurement and Reporting Committee, which created this series to explore quality measures common in hospital medicine. Dr. Bruti is an associate professor of internal medicine and pediatrics and the chief of the division of hospital medicine in the department of internal medicine at Rush Medical School at Rush University in Chicago. Dr. Humes is an associate professor of clinical medicine at The Ohio State University Wexner Medical Center in Columbus, Ohio.
References
- Carmichael, HL, et al. Comparing methods for attributing hospital quality metrics to individual physicians: impact on performance assessment and outlier identification. J Gen Intern Med. 2026. doi:10.1007/s11606-026-10391-w.
- Herzke CA, et al. A method for attributing patient-level metrics to rotating providers in an inpatient setting. J Hosp Med. 2018;13(7):470-475. doi:10.12788/jhm.2897.
- Patel S, et al. Using participatory design to engage physicians in the development of a provider-level performance dashboard and feedback system. Jt Comm J Qual Patient Saf. 2022;48(3):165-172. doi:10.1016/j.jcjq.2021.10.003.
- Rajkomar A, et al. The association between discharge before noon and length of stay in medical and surgical patients. J Hosp Med. 2016;11(12):859-861. doi:10.1002/jhm.2529.
- Goodwin JS, et al. Association of hospitalist years of experience with mortality in the hospitalized Medicare population. JAMA Intern Med. 2018;178(2):196-203. doi:10.1001/jamainternmed.2017.7049.
- Sharda M, et al. The impact of hospitalist experience on patient outcomes: a retrospective cohort analysis at an academic medical center. Hosp Pract (1995). 2025;53(1):2602423. doi:10.1080/21548331.2025.2602423.
- Yurso M, et al. Reducing unneeded clinical variation in sepsis and heart failure care to improve outcomes and reduce cost: a collaborative engagement with hospitalists in a multistate system. J Hosp Med. 2019;14(9):541-546. doi:10.12788/jhm.3220.