Improving antibiotic prescribing for adults with community acquired pneumonia: Does a computerised decision support system achieve more than academic detailing alone?--A time series analysis

From Clinfowiki
Jump to: navigation, search

This paper addresses the issue of improving clinician compliance with known reliable and applicable clinical guidelines. The authors compared concordance with empiric antibiotic prescribing guideline recommendations for the treatment of community acquired pneumonia (CAP) in adults in three phases of a time series; baseline, academic detailing only and with a computerized decision support system.

Buising KL et al. BMC Medical Informatics and Decision Making 2008, 8:35

Introduction

Concordant therapy was defined in a binary manner. Examiners determined whether or not each patient received the recommended empiric antibiotic therapy to cover both typical and atypical pathogenic bacteria. A minimum standard approach to empiric antibiotic coverage was used, therefore it is possible that patients deemed concordant actually received broader empiric coverage or additional antibiotics beyond what was recommended, or both.

Methods

Patients were prospectively identified from an established database tool in the emergency department (ED). Exclusion criteria were used including age (< 18) and complicating co-morbidities.

Randomization of patients to an intervention and a control group was not done.

Instead of comparing two like groups with the exception of an intervention, a time-series approach was used. As a time-series analysis, this study assumes that antibiotic prescribing patterns for CAP would slowly improve over time in the absence of interventions due to a so-called learning effect. This assumption of similar, but improving, prescribing patterns may be reasonable for short periods of time yet, for the three-and-one-half year time duration of this study, may be a confounding factor.

Another weakness is that the academic detailing period’s interventions were not formally stopped but instead assumed to become dormant.

One other potentially confounding factor not accounted for is the organizational behavior of ED and hospital leadership. If leadership ever set an organizational expectation for CAP guideline use, these researchers may not have known.

Finally, the authors acknowledge that this study design did not address the sustainability of any impacts from the academic detailing or computerised decision support interventions.

The three phases of the time series

The baseline period was much longer than the other two periods (13 months, n=392). During this first time period, prospective assessments of ED CAP patients’ antibiotic regimen concordance were made but no attempt was made by the researchers to improve guideline concordance.

For the second phase of the study, 215 patients were included during the period of academic detailing. Selected clinicians provided one-on-one education to colleagues. This academic detailing was reinforced in the ED through the distribution of laminated cards with information about CAP assessment and antibiotic treatment.

The third and final phase of the study relied on ED clinicians to voluntarily use an available, computerised, stand-alone, clinical decision support system (CDSS) exclusively for CAP by clicking a shortcut on the desktop of the ED’s computers. Only 133 patients needed to be studied to achieve the statistical power goal of the researchers.

Results

The authors concluded that the academic detailing period improved guideline concordance for CAP over baseline (odds ratio 2.79, p < 0.01). This represented an improvement from a baseline concordance of 62% to a new, higher level of 69% concordance.

The authors also concluded that the stand-alone CDSS system improved guideline concordance for CAP over the academic detailing period (odds ration 1.99, p = 0.02). The CDSS was associated with a period of 90% CAP guideline concordance.

While not statistically significant, the use of the CDSS was also associated with a notable decline in prescribing an antibiotic for which the patient had a documented allergy.

Conclusion

  1. This study demonstrates that a stand-alone, disease-specific CDSS can improve concordance with established prescribing guidelines for a period measured in months.
  2. This study suggests that the more structured nature of a CDSS may better address secondary measures and intended outcomes beyond binary concordance with established practice guidelines.
  3. This study suggests that clinician time spent learning to do academic detailing to improve guideline concordance may be better spent developing CDSS tools.
  4. This study challenges the wisdom of relying solely on enterprise electronic medical record systems with computerized provider order entry and clinical decision support by suggesting that highly targeted, disease-specific, web-based CDSS systems may have a unique role in guideline adherence to improve care outcomes.
  5. This study did not detail the costs, in time and effort, of building and maintaining the CAP CDSS.