Tuesday, September 06, 2011

Yet more health IT articles based on functionalist and determinist assumptions of the general format “what is the impact of technology X on outcome Y"

The article "Tensions and Paradoxes in Electronic Patient Record Research: A Systematic Literature Review Using the Meta-narrative Method" by Greenhalgh, Potts, Wong, Bark and Swinglehurst at University College London appeared in the Dec. 2009 Milbank Quarterly. I wrote about it extensively and quoted it at this post. A key statement:

... This review has also identified some areas where more research does not appear to be needed ... [including] simplified experimental studies based on functionalist and determinist assumptions of the general format “what is the impact of technology X on outcome Y?” or variations thereof ... the circumstances in which they add value are more limited than is often assumed.

We [also] believe that surveys of attitudes of patients or staff towards ‘the EPR’ or ‘computerization’ which are not adequately contextualized have almost no enduring value.

So guess what was just published in the NEJM?

An article based on functionalist and determinist assumptions of the general format “what is the impact of technology X on outcome Y."

In a special article entitled "Electronic Health Records and Quality of Diabetes Care", NEJM August 31, 2011 (link), a study was performed in which the researchers:

... compared EHRs with paper-based records in a long-term regional collaborative that seeks to improve care and outcomes for patients with chronic conditions.

They found that:

... EHR sites were associated with higher levels of achievement of and improvement in regionally vetted standards for diabetes care and outcomes. Our findings focus on composite standards, although the results were similar for virtually all component standards.

This is not really news. I had the same results in a more limited EHR data-based study of diabetics ... in 1997.

As I've often written, health IT can be of great benefit...but only if done well. (I have to frequently repeat that there is massive, perhaps wicked complexity behind those simple two words "done well.") When not done well, disaster can strike.

There are no statistics in in the NEJM article regarding complications, "close calls", patient injuries, or patient deaths due to the implementation of health IT. I sincerely doubt the incidence was zero. Their dismissal or lack of mention is common in the medical and health IT literature and seems to reflect an amoral, pervasive paternalism in medicine. The amoral paternalism in turn seems to be a repeat of the attitudes towards experimentation that led to the many human subjects protections that apply everywhere else in biomedicine (link) - except computing - e.g.:

45 CFR 46 Protection Of Human Subjects

Guidelines for Conduct of Research Involving Human Subjects at NIH (Gray Booklet) (pdf file)

The Belmont Report Ethical Principles and Guidelines for the Protection of Human Subjects of Research

Nuremberg Code Directives for Human Experimentation

World Medical Association Declaration Of Helsinki


To their credit, the NEJM authors did issue several caveats:

... we compared sites that had sophisticated EHR systems with paper-based organizations that, as safety-net practices, care for a vulnerable patient population and may have fewer quality-related resources than other paper-based practices ... Our results, showing accelerated improvement in care and outcomes, should encourage those concerned that the quality of ambulatory care may fail to improve with increased adoption of EHRs [per numerous past studies - ed.] ... our study did not determine changes in achievement after the conversion from paper to electronic records, which would provide more compelling evidence of the benefits of EHRs.

One might wonder how the tens or hundreds of millions of dollars spent on EMR's might compare, with regard to disease management, with the results achieved by hundreds of dedicated people who could be hired for that purpose for far less money. In other words, the ROI issues of the health IT investment vs. alternatives are not addressed (they rarely are), and truly robust RCT's were not performed comparing the two alternatives.

Greenhalgh et al. also wrote in the aforementioned Milbank article:

... as a cross-cutting theme in all the above areas, the realpolitik of EPR projects within and between organizations and interest groups should be more explicitly explored ... Orlikowski and Yates have called for more research on the “messy, dynamic, contested, contingent, negotiated, improvised, heterogeneous, and multi-level character of ICTs [information & communications technologies - ed.] in organizations” (page 132) (Orlikowski and Yates 2006).

We suggest that sponsors and publishers eschew sanitized accounts of successful projects and instead invite studies of the EPR in organizations that “tell it like it is” – perhaps using the critical fiction technique to ensure anonymity (Winter 1986).

There's no trace of that in the new NEJM article. Where health IT is concerned, that's where the money is (no pun intended) in learning how to "do health IT well."

-- SS

2 comments:

Anonymous said...

In addition, the report's conclusions are flawed due to the selection bias that entails the poor and less educated being cared for in the paper record practices and the wealthy and educated in the EMR practices.

Even Dr. Cebul stated, he was "not surprised by the findings", in the press release. So, did he write the conclusion before the study was done?

Finally, the ONC publication, "Buzz Blog" does not publish comments that contradict its ra ra ra HIT cheer leading mantra. Is that any way to treat the taxpayers who are funding that operation?

Anonymous said...

I mean really, what does either X or Y have to do with EMRs?