The following is a guest blog post by Andy Oram, writer and editor at O’Reilly Media.
If you want to see a blueprint for real health reform, take the time to read through the white paper, “A Robust Health Data Infrastructure,” written by an independent set of experts in various areas of health and information technology. They hone in, more intently than any other official document I’ve seen, on the weaknesses of our health IT systems and the modernizations required to fix them.
The paper fits very well into the contours of my own recent report, The Information Technology Fix for Health. I wish that my report could have cited the white paper, but even though it is dated November 2013, it was announced only last week. Whether this is just another instance of the contrasting pace between technologists and a government operating in a typically non-agile manner, or whether the paper’s sponsor (the Agency for Healthcare Research and Quality) spent five months trying to figure out what to do with this challenging document, I have no way of knowing.
The Robert Wood Johnson Foundation played an important role organizing the white paper, and MITRE, which does a lot in the health care space, played some undescribed role. The paper’s scope can almost be described as sprawing, with forays into side topics such as billing fraud, but its key points concern electronic health records (EHRs), patient ownership of information, and health data exchange.
Why do I like this white paper so much? Two reasons. First, it highlights current problems in health information technology. The authors:
Decry “the current lack of interoperability among the data resources for EHRs” as leading to a “crippled” health data infrastructure (p. 2), and demand that “EHR software vendors should be required to develop and publish APIs for medical records data, search and indexing, semantic harmonization and vocabulary translation, and user interface applications” (p. 44).
Report with caution that “The evidence for modest, but consistent, improvements in health care quality and safety is growing.” Although calling these “encouraging findings,” the authors can credit only “the potential for improved efficiency” (p. 2 of the paper).
Warn that the leading government program to push health care providers into a well-integrated health care system, Meaningful Use, fails to meet its goals “in any practical sense.” Data is still not available to most patients, to biomedical researchers, or even to the institutions that currently exchange it except as inert paper-based documents (p. 6). The authors recommend fixes to add into the next stage of Meaningful Use.
Lament the underpopulated landscape of business opportunities for better interventions in patient care. “Current approaches for structuring EHRs and achieving interoperability have largely failed to open up new opportunities for entrepreneurship and innovation” (p. 6).
Second, the paper lays out eminently feasible alternatives. The infrastructure they recommend is completely recognizable to people who have seen how data exchange works in other fields: open standards, APIs, modern security, etc. There is nothing surprising about the recommendations, except that they are made in the context of our current disfunction in handling health information.
A central principle in the white paper is that “the ultimate owner of a given health care record is the patient him/herself” (p. 4), a leading demand of health reformers and a major conclusion in my own report. Patient control solves at one stroke the current abuse of patient data for marketing, and allows patients to become partners in research instead of just subjects.
The principle of patient control leads to data segmentation, a difficult but laudable attempt to protect the patient from bias or exploitation. Patients may want to “restrict access to certain types of information to designated individuals or groups only (e.g., mental health records, family history, history of drug abuse) while making other types of information more generally available to medical personnel (e.g., known allergies, vaccination records, surgical history)” (p. 33).
This in turn leads to the most novel suggestion in the paper, the notion of a “patient privacy bundle.” Because most people have trouble deciding how to protect sensitive parts of their records, and don’t want to cull through all their records each time someone asks for research data, the health care field can define privacy policies that
meet common needs and let patients make simple choices. Unfortunately, a lot of hurdles may make it unfeasible to segment data, as I have pointed out.
Other aspects of the white paper are also questionable, such as their blithe suggestion that patients offer deidentified data to researchers, although this does appeal to some patients as shown by the Personal Genome Project. (By the way, the authors of the white paper mischaracterized that project as anonymous.) Deidentification expert Khaled El Emam (author of O’Reilly’s Anonymizing Health Data) pointed out to me that clnical and administrative data involves completely different privacy risks from genomic data, but that the white paper fails to distinguish them.
I was a bit disappointed that the paper makes only brief mentions of patient-generated data, which I see as a crucial wedge to force open a provider-dominated information system.
The paper is very research-friendly, though, recognizing that EHRs “are already being supplemented by genomic data, expression data, data from embedded and wireless sensors, and population data gleaned from open sources, all of which will become more pervasive in the years ahead” (p. 5). Several other practical features of health information also appear. The paper recognizes the strains of storing large amounts of genomics and related “omics” data, pointing out that modern computing infrastructures can scale and use cloud computing in a supple way. The authors also realize the importance of provenance, which marks the origin of data (p. 28).
Technologists are already putting in place the tools for a modern health IT system. The white paper did not mention SMART, but it’s an ideal API–open source, government-sponsored, and mature–through which to implement the white paper’s recommendations. The HL7 committee is working on a robust API-friendly standard, FHIR, and there are efforts to tie SMART and FHIR together. The Data Distribution Service has been suggested as a standard to tie medical devices to other data stores.
So the computer field is rising to its mission to support better treatment. The AHRQ white paper can reinforce the convictions of patient advocates and other reformers that better computer systems are feasible and can foster better patient interventions and research.