Public Proponents Proliferating

I was recently reviewing research on the costs of recruiting subjects for population studies. I found that even with several million dollars of subscription costs paid by the University per year, about 1/3 of the articles I had identified were not covered by our subscriptions. These articles then required an additional $20 to $35 personal investment each if I wanted to see their details. For less well-endowed institutions, the challenge is even more significant for their faculty and students. In this context, it is heartening to read this report from the Public Library of Science organization. Not only have they published high quality articles but they have 5.4 million readers, 26,000 authors, 13,000 peer reviewers, over 11,000 articles submitted in 2008 and they project that 90% of operating expenses will be covered by the PLoS funding model by the end of 2009. Hats off to a daring and creative act of social and academic organization.

A statistical cure for the voodoo

A recent issue of Perspectives in Psychological Science demonstrates the kind of full-throated give and take that makes for better science. An article by Ed Vul et al, questions the results of functional magnetic resonance imaging (fMRI) studies that relate imaged changes to personality traits, emotional states, and social interactions. The heart of the critique is how the correlations are calculated and how their statistical significance is reported. Their analysis is backed up with a very substantial survey of the investigators of the studies in question. In the same issue, the supporters and detractors of the critique are given their say. Most impressive is the statistical perspective of Lindquist and Gelman. It is even-handed and informative and should be required reading for any student about to engage in research of the increasingly familiar "high dimensionality" data sets (e.g. in genomics and imaging) that allow a multiplicity of questions to be asked or hypotheses to be tested.

Update: 7/7/2009: Nice piece on this topic from NPR.


Minting a new medical record

Early on in the development of personal health records, we had many discussions about how to jump-start the critical first step to adoption: populating that personally controlled record with clinical data. We recognized that, especially in the early years (i.e. 1994) healthcare institutions might be loathe to share data of their patients with these same patients (in electronic form) and even less so with other institutions delivering health care. So, it was with a slightly evil gleam that we considered the following opportunity when some of the local healthcare institutions developed patient portals to their healthcare systems (e.g. PatientSite and Patient Gateway). If patients could access their own medical record at one or more institutions, then a program working on their behalf (using their credentials) could then automatically log in and scrape up whatever HTML-formatted data was available and then reformat it into a standardized data model within a truly portable and personally controlled health record. This solution would not require that any institution hew to a particular data standard or communication protocol (beyond HTTP and HTML) but it would require us to be very nimble to update our translation programs. Specifically, if one of the hospitals would change the formatting of the screens on the patient portal, we would have to change the program that transforms those screens into a useable personal health database. That would certainly have been possible and legal but we decided not to proceed out of a sense of collegiality with our medical informatics colleagues and instead worked to cultivate more explicit data sharing governance and agreements.

Recently, however, I was reminded by my colleague Jeff Behrens, of a very successful model of data sharing in a personally controlled financial record: Mint.com. It's a remarkable site that allows anybody (and without additional cost) to develop a cohesive view of their finances drawn from myriad information sources including: your bank account(s), your retirement accounts (e.g. 401K plan), your investment brokerage account, your credit/debit card(s), and your mortgage account. To just name a few. This in the absence of any data interchange model across this multiplicity of institutions and functions. Once this aggregation is performed for you, many services are available such as alerts for over-budget spending, due dates of credit card bills, changes in your spending mix, and decision support for improved financial performance (e.g. switching to lower interest credit cards—Mint.com knows the interest rates on each card). There are a few remarkable properties of this site that may be quite instructive for healthcare information technology.

  • Some of us are quite comfortable passing along all our account information, user names, and passwords to a third party given a basic level of confidence in their privacy policy, as backed up by their explicitly state policies.
    • Essential to this comfort however, is the immediate value that the user obtains by participating. Not in the distant future but today.
  • Hundreds of data sources of relevance are unified without a single standard data model (unless you consider Mint.com's model the standard).
  • The data source institutions did not enter into an agreement for data sharing. Only the consumer had to agree to let Mint.com automatically access these data sources with their borrowed credentials.
  • A national data sharing infrastructure was created in 1 year from a startup-sized technology development and marketing budget.
Mint.com now has over a million users and is tracking over $50 billion in transactions. Can healthcare organizations, or the government or healthcare entrepreneur catalyze an analoguous success?