What's driving spending differences in medical groups and what might that mean for health policy.

Over the past 20 years, much attention has been paid to health care prices and the role they play in driving high health care spending in the US.1 This is in no small part due to the 2003 paper by Anderson et al. entitled “It's the Prices, Stupid: Why the United States is So Different from Other Countries”,2 and the follow-up paper in 2019 entitled “It's Still The Prices, Stupid: Why The US Spends So Much On Health Care, And A Tribute To Uwe Reinhardt”.3 In this issue, Mehrotra et al.4 take on the issue of how differences in prices at the medical group level may contribute to differences in spending using data for the non-elderly population commercially insured by the United Health Group. But as is only fitting when it comes to studying the US health care system, the answer is that it is complicated. Ultimately, the authors find that there is large spending variation by medical group even after controlling for case mix, with a difference in spending of over $1800 between patients going to medical groups in the top quartile compared to the bottom quartile of spending. Significant price variation at the provider level somewhat counterintuitively does not appear to be the primary driver of spending differences across medical groups. So, the big questions are (1) what appears to be driving this finding and (2) what does this mean for health policy? First, it is important to note that the study primarily focuses on the medical group level rather than the patient or provider level. This makes good sense in the context of health policy as this is the level often given responsibility for patients, their care, and their medical spending.4, 5 Mehrotra et al. find that risk-adjusted medical spending is 1.5 times higher in medical groups in the top quartile of spending compared with the bottom quartile. Interestingly, this difference does not change dramatically across care types, ranging from 1.35 times higher for primary care, endoscopy, and lab to 1.83 times higher for inpatient and emergency department (ED) care. So, the question they examine is how much of the difference is due to high-spending medical groups having higher prices vs. providing more care or more intensive care. On the one hand, the authors still find considerable provider-level price variation as shown in their fig. 1. However, provider price variation is not what appears to be predominantly driving the medical group level spending differences in this study. Instead, the authors find only a 21% change in the difference in spending between patients in the top quartile of medical group spending compared with the lowest quartile. In other words, nearly 80% of this difference is likely explained by differences in utilization. It is also important to note that the study by Mehrotra et al.4 does a formidable job of trying to account for the myriad complexities in the health care system. Inevitably, the authors have to make a number of analytic decisions throughout the study, but they also include a number of sensitivity analyses to examine some of these decisions. Given the breadth and complexity of this study, there are inevitably several future avenues researchers could examine to better understand what factors may be driving the primary finding that differences in utilization rather than pricing seem to drive spending differences at the medical group level. The first, and clearest, question might be how much the results extend beyond UnitedHealth Group and beyond the non-elderly population. But other factors likely warrant additional attention both in terms of how sensitive the results are to these factors but also as potential explanations for the findings. For example, (1) what might be driving differences in comorbidities across medical groups including potentially differential coding intensity? (2) Of the higher quantity of care, does it tend to be greater volume (i.e., more visits or more procedures) or greater intensity (i.e., more intensive visits or procedures)? (3) Are there some medical groups that retain more of their patients for high cost hospital, specialty, and imaging, and what does this imply for spending? (4) Are there important quality differences between medical groups at the higher or lower quartiles of medical group spending? This is less to critique the study by Mehrotra et al., but rather to note that this is likely just the tip of the iceberg of our understanding of what drives medical group spending. The study clearly has important implications for value-based models or alternative payment models. Most of these models that aim to reduce costs while providing high quality care tend to focus on primary care providers or specialists who are expected to help patients better manage chronic conditions in order to avoid costly care.5 On the one hand, the study potentially showcases the importance of this idea, namely, lower cost medical groups appear to be those that limit inpatient, specialty, and imaging spending. Yet, the big policy question is how well medical groups are really able to do this with their attributed patients. The current study shows that while hospital, specialty, and imaging services drive the spending differences, many patients not only go to a wide number of groups for inpatient, specialty, and imaging care but they go to a wide range of groups with very different prices, so wide, that relative prices just do not explain much of the spending variation (authors' Table 4). While Mehrotra et al. point to the need to control utilization, a major question is how well primary care or primary medical groups can actually control hospital, specialty, and imaging care when much of this care is outside of the home medical group. This may be easier in larger, more comprehensive health systems where there may be clearer avenues to coordinate care across the care continuum.6 But while the results also highlight that medical group spending is largely driven by inpatient and specialty care, evidence suggests it can be difficult for primary care or care coordinators to significantly impact inpatient and specialty care.7 A key question is then whether there are strategies that lower spending medical groups have implemented that might help to inform policymakers and payers. Further complicating efforts to manage spending is the unsurprising finding that much of the spending variation is driven by higher cost care, namely, hospital care, specialty care, and imaging. The focus on non-elderly enrollees in commercial plans means the sample is likely predominantly those covered by employer-sponsored insurance (ESI). With average deductibles in ESI of $1450 for single enrollees and $2540 for a family,8 much of hospital and specialty care is likely to be after the deductible has been met, meaning patients are likely to be less price sensitive.9, 10 As a result, patients may be less amenable to being nudged towards lower price providers or persuaded to avoid low-value care. This could help to explain the breadth of where patients go for inpatient and specialty care. With relatively limited cost sharing,11 patients may be using other criteria to make decisions about whether and where to get inpatient and specialty care. As a result, efforts to limit spending always circle back to the complex task of how to guide patients to the high value providers while ensuring that patients get the care they need. The authors suggest with better knowledge about prices, providers might help in this effort. But in cases of limited cost sharing, these efforts are likely to rely on trusting provider recommendations or having payers layer on often less popular efforts such as prior authorization, narrower networks, or other forms of gatekeeping.12, 13 The article by Mehrotra et al.4 shines a light on one part of the puzzle while also raising a number of important questions about what might be driving the results and what they may mean for health policy. While they highlight that utilization differences rather than prices appear to drive spending differences across medical groups, the big questions are both how consistent are these results across different payers and age groups and what are the best approaches for medical groups to achieve lower rates of utilization while still ensuring access to care and good health outcomes. No funding to report.