Background: The United States is aiming to achieve nationwide adoption of electronic health records (EHRs) but lacks robust empirical evidence to anticipate the effect on health care costs.
Objective: To assess short-term cost savings from community-wide adoption of ambulatory EHRs.
Design: Longitudinal trial with parallel control group.
Setting: Natural experiment in which 806 ambulatory clinicians across 3 Massachusetts communities adopted subsidized EHRs. Six matched control communities applied but were not selected to participate.
Patients: 47 979 intervention patients and 130 603 control patients.
Measurements: Monthly standardized health care costs from commercial claims data from January 2005 to June 2009, including total cost, inpatient cost, and ambulatory cost and its subtypes (pharmacy, laboratory, and radiology). Projected savings per member per month (PMPM), excluding EHR adoption costs.
Results: Ambulatory EHR adoption did not impact total cost (pre- to postimplementation difference in monthly trend change, −0.30 percentage point; P = 0.135), but the results favored savings (95% CI, $21.95 PMPM in savings to $1.53 PMPM in higher costs). It slowed ambulatory cost growth (difference in monthly trend change, −0.35 percentage point; P = 0.012); projected ambulatory savings were $4.69 PMPM (CI, $8.45 to $1.09 PMPM) (3.10% of total PMPM cost). Ambulatory radiology costs decreased (difference in monthly trend change, −1.61 percentage points; P < 0.001), with projected savings of $1.61 PMPM (1.07% of total PMPM cost).
Limitations: Intervention communities were not randomly selected and received implementation support, suggesting that results may represent a best-case scenario. Confounding is possible.
Conclusion: Using commercially available EHRs in community practices seems to modestly slow ambulatory cost growth. Broader changes in the organization and payment of care may prompt clinicians to use EHRs in ways that result in more substantial savings.
Primary Funding Source: Massachusetts eHealth Collaborative.