David Magid and colleagues recruited study participants who were on four or fewer drugs with persistently high blood pressures from the electronic health records of three Denver health care systems: 1) the Eastern Colorado VA Healthcare System, 2) Kaiser Permanente Colorado and 3) a county safety net system called Denver Health and Hospitals. 3083 patients were screened, 338 were chosen to participate and 283 completed the study.
Participants were randomly assigned to a control group,consisting of usual care along with a National Institutes of Health (NIH) patient information booklet , or and intervention group which got the booklet plus additional patient education, a home blood pressure monitor that was paired with an interactive voice response system (BP numbers were inputted using the phone key pad) that was, in turn, monitored and managed by a pharmacist. The pharmacist made medication changes after discussing things with the patients' physicians. The study lasted six months, the mean age of the participants was 62 years, one third were female and two thirds were white. The study was funded by the American Heart Association and the Colorado Dept. of Public Health and Environment.
The average blood pressure for the 138 patients randomized to the intervention group was 150/89, For the control patients, the average blood pressure was 144/85. That difference was not only statistically significant, it played a role in the interpretation of the observed outcomes. At six months, both groups' average blood pressures were lower and essentially the same (137/82). Because the baseline was higher in the intervention group, the authors found that the change in blood pressure was also highest in the intervention group. What's more, it appeared that the patients with blood pressures with a systolic (top number) of more than 150 experienced the greatest reduction in blood pressure and therefore derived the greatest benefit from the program.
In the discussion of their results, the authors - like all good scientists - did an excellent job of describing the limitations of their findings. From a statistical standpoint, it's possible that the greater decrease in the intervention group was partially the result of mathematical phenomenon known as "regression to the mean." The DMCB additionally wondered if the 3 subgroups (VA, Kaiser and County) had different outcomes, which would suggest that the intervention worked better in one setting than another. Last but not least, the authors did some additional analyses on medication compliance and correlating that with the outcomes.
Lessons for the DMCB It's not the pharmacist, or the IVR or the home BP monitor or the patient engaged in self care, it's the combination of all of the above. The effect is probably greater than the sum of its parts and extends the reach of the physician. This seems a far more reasonable approach than, say intrusive electronic record "prompts" during a physician appointment, pay-for-performance or quality assurance. What's more, the model fits classic remote disease management or the Patient Centered Medical Home or both.
Once again, it appears the benefit from interventions like this is not evenly distributed across a population. Patients with higher blood pressure derived the greatest benefit, suggesting that programs like this should be targeted at patients who are at greater risk or who are mostly likely to respond. In this instance, it was for patients with a blood pressure that was greater than 150.
Last but not least, while the electronic health record certainly has its limitations, one useful role is helping to find candidates for research trials.