Decisional support to prevent adverse drug reactions of long latency: pilot randomized controlled intervention for glucocorticoid-induced diabetes.

ABSTRACT

RATIONALE, AIMS, AND OBJECTIVES: Long-term exposure to glucocorticoids can cause adverse drug reactions of long latency (ADRLLs), including glucocorticoid-induced diabetes mellitus (GID). Providers can monitor for GID using the glycosylated haemoglobin blood (HbA1C) test. This study examined the utility of decisional support to improve HbA1C-based screening for GID. US veterans were identified as chronic users of oral glucocorticoids (>120 days of oral glucocorticoids in the last 2 years). The primary care providers caring for these patients were the target of the intervention. Providers were randomized to receive automatic HbA1C orders for their patients receiving chronic glucocorticoid or usual care. METHODS: This study was a pilot two-arm, group-randomized, controlled trial (n = 12 providers, n = 38 patients). Data collection occurred from 5 May 2013 until 10 January 2014. A pharmacist generated the order for an HbA1C through the electronic medical record. The time between the intervention start date and the date on which an HbA1C order was signed were compared using Cox proportional and hierarchical linear regression. RESULTS: The time to sign HbA1C orders (mean 12.0 days for the intervention arm; 104.0 days for control arm) was associated with significant differences favouring the intervention [HR (Hazard Ratio) 50.2, P < 0.001, confidence interval (CI) 6.3 to 398.7]. For the intervention group, 95% of orders were signed, whereas only 12% of control providers signed orders (odds ratio 150, P < 0.001, CI 12.4 to 1812.9). CONCLUSIONS: The results of this study strongly suggest that the clinical pharmacist-triggered order intervention is effective. This method of computerized decisional support may be useful in improving screening for GID and ADRLLs.