The first major implication of this study is the power of examining expenditures based on how patients experience care. Health care expenditures are typically categorized based on the payer or setting; therefore, changes over time and potential policies are considered through this lens (Centers for Medicare and Medicaid Services 2009
). However, this is not how patients experience care. Patients have a chronic disease, pregnancy, trauma, or some other life event. These life events drive how patients experience the health care system, make decisions, and spend their health care dollars. MEPS is a useful data source for examining expenditures through a patient-centered lens because it is based on a household survey methodology.
We believe this is the first study to examine expenditures based on patient-centered care categories. The potential policy, payment, provider, and patient levers to influence expenditures based on these categories would be more specific and directly aligned with how patients experience care. One can further segment spending by care categories based on traditional parameters, such as setting (e.g., chronic conditions for inpatient hospital care). For example, if a state or the nation implemented a policy to attempt to decrease inpatient costs for chronic conditions or increase the percentage of expenditures overall allocated to routine preventative care, this methodology would allow tracking expenditures for the particular patient-centered care category.
The CMS NHEA data released each year receives substantial publicity within health care and to the broader public. These data are critical for overall trends in health spending; however, we believe that this viewpoint from MEPS could complement that data. CMS NHEA is primarily based on provider expenditures and includes categories such as research, structures and equipment, public health investment, and administrative costs. If one asks a question such as how much are we spending on injury, routine preventative health care, or pregnancy/birth, the CMS NHEA does not provide the answer (Sing et al. 2006
). MEPS is a valuable data source for answering these types of questions. These patient-centered expenditures may also lend more direct connections to potential policy interventions. For example, health reform included multiple policy interventions to increase use of preventative health care; this type of analysis performed annually could answer the question of whether the percentage of expenditures in this category increases over time. Correspondingly, one could examine whether the increase in routine preventative health care lowered the rate of growth and relative percentage of expenditures in categories such as chronic conditions. This data will not be able to prove causality that routine health care decreased chronic conditions through prevention of disease or better managed chronic conditions, but it could assess the associations and trends over time if performed annually. Finally, more in-depth studies of interventions at the state or local level could attempt, using these categorizations, to assess the effects of interventions (e.g., intervention to decrease inpatient hospital expenditures for chronic conditions) on these categories of care.
Another potential use of this information is for identifying priorities. For example, research priorities for comparative effectiveness research by the new Patient-Centered Outcomes Institute from the Affordable Care Act and overall biomedical research priorities should be influenced by the relative costs. This information provides a high-level look at expenditures, and expenditures could be analyzed further within a specific category (e.g., chronic conditions) to identify the specific conditions associated with high levels of expenditures. One could also ask whether the research funding allocated to categories such as injury/trauma or pregnancy/birth is aligned with their expenditure level.
This method to assess national health care expenditures can also drill down deeper into determining drivers of medical costs. This study found, as have others, that chronic conditions are the major driver of costs in the United States (Bodenheimer and Fernandez 2005
; Agency for Healthcare Research and Quality 2009
;). Unfortunately, chronic care management has a mixed history, with many intervention models failing to improve quality, control costs, or both (Bott et al. 2009
; Peikes et al. 2009
;). Models to better manage patients with chronic conditions should be a major focus of future interventions and research, with successful models being spread broadly and further tested. Comparative effectiveness evidence to inform the selection of medications for those with chronic conditions, appropriate provider and patient education, and incentives for high-value medication use are all needed (Conway and Clancy 2009
). In addition, practice-based quality improvement collaboratives may help increase the quality and efficiency of care delivered to patients with chronic conditions (American Board of Medical Specialties 2009
). The economic impact of these initiatives could be assessed using our model for tracking expenditures. The data also highlight that inpatient and outpatient hospital expenditures accounted for a significant proportion of expenditures for acute illness, trauma/injury, and chronic conditions. This may represent an opportunity to decrease expenditures through care management that prevents hospitalizations for some of these conditions.
This study has several limitations. First, MEPS data do not include institutionalized individuals or those in the military so will not capture expenditures for these populations. Institutionalized populations being excluded could result in underestimating expenditures in categories such as chronic conditions. Second, MEPS data on health care utilization, condition, and service categories are household and provider reported and therefore despite efforts to validate data, misclassification could occur. Third, MEPS will provide lower total estimates for health care expenditures than those produced by CMS due to different sampling, methods, and populations and related expenditures included. Finally, some categories have potential for overlap (e.g., chronic and acute) and assumptions had to be made on how to code expenditures into care categories, so potential for misclassification exists. For example, the assumption to place chronic above acute in the hierarchy will preferentially allocate expenses to the chronic condition if an encounter is for both chronic disease and acute exacerbation or acute illness. However, changing these assumptions did not substantially alter the overall picture of expenditures described in this manuscript. Future analyses could define even more specific care categories and evaluate the frequency that patients have expenditures in multiple categories.
In summary, this patient-centered viewpoint on our nation's health care expenditures could become an important complementary method to examine health care expenditures, identify trends, and monitor how policies affect these trends in the future.