Emerging Issues in the Anthropology of Policy
Did you notice your health care provider keying information into a computer during a recent office visit? If so, there is a good chance that you observed the data entry process of an electronic health record (EHR) that is exchanging information with national, state and regional level databases, creating a particular form of Big Data – “datasets whose size is beyond the ability of typical database software tools to capture, store, manage, and analyze” (Manyika et al, 2011).
An EHR is the digital version of a patient’s medical chart, designed to contain and share information electronically with other health care providers and agencies involved in a patient’s care (eg, payers, public health officials). EHRs facilitate Big Data because they draw from millions of patient records, and they can be captured, stored, managed and analyzed using software tools created for this purpose.
The volume of EHRs is growing because they are mandated in the United States by the 2009 American Recovery and Reinvestment Act (ARRA), particularly a section called the HITECH Act (Health Information Technology for Economic and Clinical Health). That’s right – not the Patient Protection and Affordable Care Act (PPACA), or so-called Obama Care), but legislation growing out of the 2008 financial crisis.
The new EHR policy requires the meaningful use of EHRs by all physicians and hospitals no later than 2015. If health care providers use these records as required, they receive economic incentives from the Centers for Medicare and Medicaid Services (CMS); failure to do so imposes an economic disincentive (see athenahealth 2009).
Big Data from EHRs are intended to improve health care quality and reduce costs; eg, providers receive access to patient outcomes data that can be linked to payer reimbursement levels. The aggregation of individual EHRs and its analysis by algorithms may be viewed as an aspect of evidence based medicine – treating patients based on the best scientific findings available (Timmermans and Berg 2003).
Some physicians have expressed reservations about EHRs and have not yet registered to purchase one, leading to delays in the federal government’s timetable for implementation. Physicians’ concerns include problems and errors in data entry and software, the need to change workflow, additional time on system maintenance and administrative issues (see Karsh, et al 2010).
A problem for Big Data in health care is the quality of data in EHRs. The problem goes beyond data entry errors, which are difficult to eliminate, to include more systemic concerns. For example, reimbursement policies require extensive documentation, and health care providers may respond with cut-and-paste approaches that propagate mistakes or obsolete information. Further, EHR documentation can be difficult to interpret as a result of differences in physicians’ practices. What does it mean if a medication list is empty in a patient’s chart? Is the patient not on any medications, not on medications prescribed by that provider, or were medications documented elsewhere? (See Adler-Milstein and Jha 2013). With variation in physicians’ practices, aggregation will require both advances in software for analyzing natural languages and changes in the way care is documented.
Another problem for Big Data and EHRs concerns the fragmented health care system in the United States. Patients visit multiple health care providers where EHRs purchased from different vendors are not interoperable (ie, do not communicate with each other). The ARRA of 2009 created state-level Health Information Exchanges (HIEs), which are platforms that allow “medical providers to electronically exchange, in a secure environment, an individual’s health information” (Pennsylvania eHealth Partnership Authority, 2011).2 Such developments allow providers to access a range of clinical information to support quality and cost goals. However, the present state of HIEs does not provide a comprehensive picture of patients’ care.
Kaelber et al (2012) developed an approach to aggregating EHRs from multiple health care systems based on medical informatics. They demonstrated that data from nearly one million patients from different health care systems with distinct EHRs could be pooled, searchable through a HIPAA-compliant, patient de-identified web application that standardized and normalized the data using common ontologies (ie, formal representations of knowledge for clinical information that exist in EHRs). This study also showed that patient demographics (eg, race and ethnicity) could be linked to disease incidence (eg, thromboembolic events). This is significant because federal regulations require that EHRs collect data on race and ethnicity for purposes of addressing health disparities, meaning that researchers will be able to impute racial or ethnic tendencies with respect to various types of disease incidence using very large sample sizes; other parties (eg, pharmaceutical companies) may be interested in this information.
Yet, the collection of data on race and ethnicity in medical settings is subject to questions of validity on numerous grounds (eg, What do the chosen fields mean to the respondent? Is the meaning of the field constant across time and place? Who provided the original data represented in the system?) (see Epstein 2004 and Hunt and Megyesi 2008). Anthropologists should attend to how these issues are accounted for when such data are entered into the EHR, aggregated and analyzed by third parties, with interpretation often rendered less transparent by the use of algorithms (Maxwell 2013).
Anthropologists also can address issues such as: patient confidentiality in de-identified databases; specific actors’ influences on meaningful use guidelines embedded in EHRs; the cost of mandatory EHR implementation for small, independent physician practices in rural areas; and patient representation and inclusion in the decision-making processes that are institutionalized through EHRs and Big Data. The time has come for anthropologists to engage the intersection of health care, public policy and information technology, because that is where public policy issues are moving.
Acknowledgements: Thanks go to Emily Altimare, Christine LaBond, Natalie Petersen-Menefee and Efrem Silverman for invaluable research support, critical commentary and collegial advice, and to Samantha Solimeo for suggesting this project and guiding it along the way.
Marietta L Baba is Dean of the College of Social Science, Professor of Anthropology, and Professor of Human Resources and Labor Relations, at Michigan State University. Marietta is Co-President Elect of the Association for the Anthropology of Policy, a section of the American Anthropological Association.
Sarah Ono, Heather Schacht Reisinger and Samantha L Solimeo are contributing editors of Anthropology in the Public Sector. The views expressed in this column are those of the authors and do not represent the US government or the Department of Veterans Affairs.