Failure of computerized treatment suggestions to improve health outcomes of outpatients with uncomplicated hypertension: results of a randomized controlled trial

From Clinfowiki
Jump to: navigation, search

Why did not a decision support system improve the health outcomes of patients with uncomplicated hypertension?


Dr Murray et al conducted a randomised controlled trial on 712 patients to explore the effects of a computerised evidence-based treatment suggestions delivered by primary care physicians and/or pharmacists to uncomplicated hypertensive patients at an outpatient inner city setting and found that after 12 months, there was no statistically significant differences between experiment and control groups in terms of their medication concordance, health-related quality of life and blood pressure measurements. They concluded that they failed to demonstrate the effectiveness of this particular evidence-based decision support system on various health outcomes.

Both pharmacists and physicians used the electronic Regenstrief medical records system (RMRS) to deliver care for many years. Two protocols (for physicians and pharmacists) were developed, approved and integrated to the RMRS so that the system would generate automatic reminders and treatment suggestions to be offered to patients.

The authors provided some reasons to explain the failure of this trial:

  1. They suggested that the RCT did not have enough power to detect a difference of 15% in overall SF-36 scores.
  2. The potency of the intervention was too week and only one third of the suggestions were complied across all four groups.
  3. The system might have generated too many suggestions, some too complex. The intervention might have been too time consuming or intrusive for physicians.
  4. The reminder could have been easily cancelled and there were no requirement to adhere to them.
  5. Most of the above suggestions were also made for the pharmaciss. Also, there were a physical dissociation between the person who saw the reminder and the pharmacist who used to work at a separate workstation.


There are other factors rather than the intervention itself that can be accountable for the failure of this trial:

  1. The outcome measures were quite distance from the interventions in the chain of cause and effect. For example, a lifestyle or treatment advice would normally take a long time before it can show its effects on health-related quality of life, particularly for a symptom-free hypertensive patient. A feasibility study would have been much more useful in order to establish first whether physicians and pharmacists could be engaged before a full-blown RCT could have taken place.
  2. Compliance with antihypertensive medication usually decreases the quality of life of patients in short-term because of the difficulty of adherence to some medications that would usually have some side effects. Difference in quality of life can be detected after at least 10 years when the control group patients develop some complications due to their high blood pressures (i.e. stroke etc.)
  3. This randomized controlled trial had significant problems in its administration as well. From the 712 patients who entered this trial, about half were not followed up after a year. This effectively halved the sample size and significantly reduced the chance of finding any meaningful differences across groups.

Related Study

Evaluation of effectiveness and cost-effectiveness of a clinical decision support system in managing hypertension in resource constrained primary health care settings: results from a cluster randomized trial