Although research studies have shown large unexplained variation in how physicians care for patients with similar medical conditions (Sirovich and Fisher, 2006; Managed Care, 2009; The Dartmouth Institute for Health Policy and Clinical Practice, 2012), there is an absence of routine measurement and reporting of individual physician performance. Such measurement could help:
- providers understand how their performance compares with peers to stimulate quality improvement
- consumers make more informed choices about providers when they need care
- researchers with understanding factors associated with variations in processes of care and health outcomes
- payers with value-based purchasing efforts.
In a 2010 white paper prepared under a grant from the California HealthCare Foundation, the National Association of Health Data Organizations (NAHDO) observed that state health data reporting systems are “well-positioned to drive [health care] system improvements by making physician-level data available for multiple purposes,” even while acknowledging that many states are falling short of this goal because they either do not collect or do not publicly release physician identifiers (Love et al., 2010). As of 2012, California remains the last of 48 states with a hospital discharge data reporting program that does not collect physician identifiers.*
A variety of stakeholders have raised the question, “Should California collect and publicly release data containing physician identifiers that could be used to profile physician performance?” To advance consideration of whether California should go forward in adding physician identifiers to the hospital discharge data set, the California HealthCare Foundation asked the RAND Corporation to explore a range of issues associated with requiring the inclusion of physician identifiers in the California hospital discharge data and the potential use of these data directly by the state and/or release of the data to others. The key issues center on (1) the potential opportunities, challenges, and considerations associated with collecting physician identifiers as part of the hospital discharge data set and (2) how to release the data, to whom, and for what purposes. The term “release” of the physician-identified data could refer to many different approaches, ranging from release of raw data to some or all parties who request the file, to the state of California analyzing the data and making publicly available comparative reports of physician performance.
Findings of This Study
Current Legal Authority to Collect
- California's Office of Statewide Health Planning and Development (OSHPD) has the statutory authority to add physician identifiers as a required data element to the hospital discharge data set after issuing a notice of proposed rulemaking and taking into account public comments in making a final ruling, in accordance with the full requirements of the California Administrative Procedures Act.
- Data elements captured on the hospital uniform billing form (UB-04)—including physician identifiers—do not count toward the statutory limit of data elements that OSHPD can add to the hospital discharge data set within a five-year period.
Feasibility of Data Collection and Data Accuracy
- According to interviews with a variety of California stakeholders, the collection of physician identifiers as part of the hospital discharge data set appears feasible and would not present an additional data collection burden on hospitals, because California hospitals already routinely capture and code physician identifiers as part of the UB-04 form required by Medicare and private payers. The UB-04 contains fields to capture the name and the National Provider Identifier (NPI) for both the attending physician and operating physician, and hospitals are required to complete these fields in order to receive reimbursement from payers.**
- Representatives for the seven states that were interviewed for this study did not report difficulties collecting physician identifiers. These representatives also reported that they did not engage in extensive verification to ensure accuracy of the data.
- Among the 47 states that do collect physician identifiers, some of the states collect only one physician identifier and some collect more than one.
- Concerns were raised about varying definitions of attending physician across hospitals and accuracy of the data; however, California stakeholders agreed that use of the data would prompt hospitals and their staffs to improve the data's accuracy. Irrespective of how attending physician is defined, there was concern about the nature and variability of practice among attending physicians that would impact on how to consistently code the attending physician across hospitals.
Benefits to Collecting Physician Identified Data and Issues for Consideration
- All California stakeholders who were interviewed for the purposes of preparing this report saw some potential benefits associated with collecting and using physician-identified data, particularly related to generating benchmarking data for providers to use in quality improvement efforts and for providing information to help consumers make informed decisions about where and from whom to receive care.
- Concerns that emerged from discussions with stakeholders included the accuracy of the physician identifiers, the need to risk-adjust outcome measures, careful consideration of defining who is accountable for the management or care of the patient, how data would be presented, and whether consumers would correctly interpret the results.
- Stakeholders agreed that identifying a responsible physician would be fairly straightforward for the operating physician but more challenging for the attending physician. Determining who was responsible for the care delivered would be particularly challenging for more medically complex patients who have longer hospital stays and multiple physicians involved in the patient's care.
- Attribution of performance measures to individual physicians highlighted the issue of defining what types of measures are truly meaningful and appropriate for reporting performance at the physician level. Providers were particularly interested in being involved in determining what measures would be reported at the physician level and for which medical conditions and procedures.
- Contextual factors can be important in correctly interpreting the data, and it was recommended that contextual factors should be considered when defining performance measures and presenting results.
- While the discussion focused on physician-identified data and holding individual physicians accountable for results, a number of stakeholders thought that the entire “care team” determines patient outcomes, not just the actions of a single physician. Some stakeholders expressed a desire to define the “care team” but recognized the challenges associated with mapping individual physicians and other health care providers to care teams as in many cases the teams are not static.
Release and Use of the Data
- If OSHPD collects physician identifiers as a data element in the hospital discharge data, the office is mandated by statute to disclose them to certain parties unless an individual patient's rights of confidentiality would be violated. OSHPD would have to decide whether to release physician identifiers in situations in which the statutes authorize but do not mandate release.
- The California Health Data and Advisory Council Consolidation Act does not specifically protect the identity of physicians. Because OSHPD is authorized to release public use data files in a standard file format, the public release files would presumably include the physician identifiers.
- There was variation among California stakeholders regarding who they thought should have access to physician-identified data and what uses of the data would be appropriate.
- The states with which we held discussions differed in their approaches to making data available to potential end users, ranging from limiting release to contributing hospitals and their providers to making the data widely available with data use agreements. In a few cases, states collect but do not publicly report physician-level data. Not all of the 47 states that collect hospital discharge data release the data, and states that do release hospital discharge data vary in how and to whom they release the data.
- A small number of states—namely New York, Pennsylvania, New Jersey, California, and Massachusetts—produce and publicly release physician-identified risk-adjusted outcome reports (RAORs) for a limited set of procedures, and in these situations physicians are provided an opportunity to review the results. Other states report the volume of procedures performed by physicians.
- OSHPD currently has statutory authority to produce RAORs based on the hospital discharge data. If OSHPD had physician identifiers and wanted to produce physician-level RAORs on conditions other than coronary bypass graft surgery (CABG), then the agency would be required to convene a clinical panel, and the panel would need to approve the risk-adjustment model in order to enable production of additional RAORs. If OSHPD decided to pursue collecting additional clinical data, then OSHPD would have to follow the statutory limits on how many data elements can be added in total. The impact or burden on hospitals would depend on the number of additional data elements that would need to be collected and the ease of retrieving the data elements, and could range from minimal to significant impact.
The two central policy questions are (1) whether California should go forward in adding physician identifiers to the hospital discharge data set and, as highlighted by our discussions with stakeholders, (2) how California should proceed with both the collection and release of those data.
The collection of physician identifiers as part of the hospital discharge data set represents an opportunity for California to generate performance data at the physician level that could be used by many stakeholders for a variety of purposes. Almost all other states currently collect physician identifiers and do so without substantial burden to hospitals, and the data have been released and/or used without major problems or incident that should cause California pause. Based on our review, we recommend that:
- OSHPD should move forward without delay to add physician identifiers to the list of data elements it routinely collects as part of the hospital discharge data.
- Because there was consensus among those we interviewed that attributing responsibility for a procedure to the operating physician was straightforward, this minimally represents a place where OSHPD should begin the process of including a physician identifier to the hospital discharge data. Even basic information on the number of procedures a physician performed annually could be useful to consumers, and understanding differences in the outcomes across physicians performing like surgeries could be extremely helpful in reducing variation and improving outcomes for patients.
- Because the identity of the attending physician also is routinely collected by most states without problem and is part of the UB-04 billing form requirements such that hospitals are already routinely capturing these data, California should collect the identity of the attending physician as part of the hospital discharge data.
- OSHPD should ask hospitals to use the standard definitions for operating physician and attending physician that are already used by hospitals in reporting these identifiers to the Centers for Medicare and Medicaid Services (CMS) as part of the standard UB-04 billing process.
- As OSHPD and the California stakeholders gain experience, the type of physician identifiers that are captured could be expanded to include, for example, physicians associated with secondary procedures and for other care settings—emergency departments (ED) and ambulatory surgery (AS) centers.
- Stakeholders expressed a desire to be engaged in the process for determining the specifics for how this is done in California and, from the provider's perspective, in helping disseminate results back to providers to help them improve. Given genuine concerns about how the data will be analyzed and used once collected, we recommend that the stakeholders should come together to forge a blueprint for appropriate data use that could be used to guide the actions of the state and end users of the data. The blueprint could serve to address a variety of concerns about how the data could be used and misused, for example, what types of measures are appropriate at the physician level, which physician(s) are appropriate to attribute responsibility for the care delivered to the patient, and how the data should be analyzed and interpreted. The development of the blueprint should happen in parallel with the regulatory process.
It is clear that many stakeholders are interested in this topic, and nearly all, including physicians, see benefits associated with collecting and using this information. The findings from this study highlight some of the key issues and questions we believe will need to be addressed as OSHPD and the California stakeholders consider how best to advance inclusion of physician identifiers in the hospital discharge data set.
California Health and Safety Code § 128745(c) and (d), 2012.
The Dartmouth Institute for Health Policy and Clinical Practice, The Dartmouth Atlas of Health Care, 2012. As of November 27, 2012:
Department of Health and Human Services, Centers for Medicare and Medicaid Services, CMS Manual System, Pub 100-04 Medicare Claims Processing, November 2, 2006.
Love, Denise, Emily Sullivan, Robert Davis, et al., The Collection and Release Practices of Physician Identifiers in Statewide Hospital Discharge Data Reporting: A NAHDO White Paper, National Association of Health Data Organizations, 2010. As of November 27, 2012:
Managed Care, “A Conversation with the Dartmouth Atlas Project's Elliott Fisher, MD, MPH: National Data, National Impact,” August 2009. As of November 27, 2012:
Sirovich, Brenda, and Elliott Fisher, “Regional Variations in Health Care Intensity and Physicians' Perceptions of Care Quality,” Annals of Internal Medicine, Vol. 145, No. 10, 2009, p. 789.
* The only condition for which public reporting at the physician level is mandated in the statute in California is coronary artery bypass graft surgery, and this is done under a separate reporting effort from the hospital discharge data reporting (see California Health and Safety Code § 128745[c] and [d], 2012).
** The hospital uniform billing form (UB-04) has standard definitions for the operating physician and attending physician. The attending physician is defined as the individual who has overall responsibility for the patient's medical care and treatment reported in a claim/encounter. The operating physician is defined as the individual with primary responsibility for performing the surgical procedure(s) and is required when a surgical procedure is listed (Department of Health and Human Services, 2006).
The research described in this report was sponsored by the California HealthCare Foundation and was conducted within RAND Health, a division of the RAND Corporation.