Doubting the cost savings of health information technology

A recent Health Affairs article casts doubt on the cost-saving prospects of health information technology despite the significant investments made by hospitals, physicians, and the federal government to promote electronic records.

Reduced ordering of imaging and other diagnostic studies by physicians is often cited as a likely mechanism for estimated cost savings due to health information technology. Possible mechanisms include reduction in redundant (duplicated) tests secondary to better access to information or due to point-of-order decision support that helps providers  rethink the appropriateness of testing. However, these mechanisms and the purported savings are merely assumptions.

Harvard researchers assessed a large outpatient data set to evaluate the hypothesis that health IT (computerized availability of imaging or use of full EMR) reduces image test ordering. The study design was that of an observational cross-sectional survey. The primary outcome variable, or dependent variable, was the presence or absence of physician-ordered imaging (CT, MRI, any radiology imaging). The predictor variable, or independent variable, was the availability of computerized reporting of test results.

The data were derived from the National Ambulatory Medical Survey which was collected in 2008 (prior to the passage of the 2009 HITECH legislation that dramatically expanded federal financial incentives for adoption of health IT). The data include a nationally representative sample with 28,742 patient visits to 1,187 physicians scattered across the United States. Multivariate logistic regression modeling was performed to account for multiple confounding variables that affect a physician’s decision process to order imaging tests: computerized order entry, physician specialty (excluded orthopedics, neurology, cardiology which order disproportionately more imaging tests), physician affinity for technology, physician financial incentives, physician practice business profiles (owner or employee/contractor, prepaid practices, hospital-owned practices, solo practitioner, urban), and also patient factors (age, sex, ethnicity, insurance type, residence in poverty-level zip code, whether seen by physician before, and presence of chronic diseases: diabetes, heart disease, cerebrovascular disease, cancer). Patient outcomes, disease acuity, and ICD-9 disease-specific reasons for the images ordered were not measured. Additionally, whether or not specific images were duplicated in an individual patient was not measured.

Overall, the prevalence of imaging orders by physicians with computerized systems for accessing imaging results was 40-70 percent higher than without computerized systems. Of the 28,741 physician visits, 4,335 resulted in an order for imaging. Computerized systems for accessing imaging results were available for 13,401 visits (in 6,458 the actual images were viewable). Computerized order entry did not affect the likelihood of image ordering. Patient and physician practice characteristics were not associated with likelihood of image ordering. Physician affinity for technology and physician financial incentives did not change the likelihood of image ordering.


Computerized order entry did not affect the likelihood of image ordering but access to computerized imaging results was associated with increased rather than decreased image ordering. However, the original assumption of the effect of health IT on reducing unnecessary imaging duplication was not evaluated; duplicate imaging per patient was not actually measured. It is the duplication of imaging per patient that is most likely the culprit for wasteful spending rather than the total orders of images within a health care system.

Several important factors that affect a physician’s decision to order images in the first place were not addressed: disease acuity, disconnected health IT systems, and defensive medicine.

Perhaps a better research question: does health IT reduce the duplication of image ordering in a population with matched disease acuity?

Researchers should also investigate how disconnected health IT systems (even within a single hospital) might affect image ordering duplication. For example, although a chest X-ray might have been ordered in the emergency department (ED), an inpatient physician without access to the ED health IT system may order another chest X-ray on hospital day 1 so as to have a desired clinical baseline for comparison. In this scenario, connected health IT systems within one hospital (not to mention across myriad sites of care) could likely reduce unnecessary duplication of imaging. For example, the inpatient physician could be liable in a malpractice case if the original ED X-ray could not be accessed or evaluated. Thus, the unnecessary duplication of a baseline X-ray is done only to prove that it was evaluated by the inpatient physician. This provides protection in the mind of the physician worried about liability and likely drives the escalating costs of defensive medicine.

Perhaps it is not enough just to have a health IT system but rather it is the quality of connectivity between health IT systems (coupled with a less litigious environment) that produces the anticipated cost-saving advantages of health IT and the true effects on physician behavior?

Jennifer Shine Dyer, a pediatric endrocrinologist, is Founder, EndoGoddess LLC and Chief Medical Advisor, Eproximiti She blogs at EndoGoddess Musings and Policy Prescriptions.

Submit a guest post and be heard on social media’s leading physician voice.

View 11 Comments >

Most Popular

✓ Join 150,000+ subscribers
✓ Get KevinMD's most popular stories