421 Order-linked, Automatic Computer Alert Dramatically Reduces Time-to-Treatment and Isolation in Clostridium difficile Infections

Sunday, April 3, 2011
Trinity Ballroom (Hilton Anatole)
Kathleen V. Dhundale, RN, ANP-BC, MSN , Stony Brook University Medical Center, Stony Brook, NY
Melinda Monteforte, PharmD , Stony Brook University Medical Center, Stony Brook, NY
Katherine Holzmacher, RN-BC, MS, NP , Stony Brook University Medical Center, Stony Brook, NY
Francina Singh, RN, BScN, MPH , Stony Brook University Medical Center, Stony Brook, NY
Lori Escallier, PhD, RN, CPNP , Stony Brook University Medical Center, Stony Brook, NY
Background: Clostridium difficile infections (CDI) have increased in incidence and severity over the last decade. Prompt recognition is important for treatment and prevention. Information technology may speed recognition, initiation of treatment and institution of isolation.

Objective: Create an order-linked alert to providers, within the Electronic Health Record (EHR) for positive stool tests. Determine if this decreases time to initiation of anticlostridial medications and institution of isolation.

Methods: An alert, triggered instantaneously by lab posting of positive stool for C. difficile PCR results in the EHR, “pops up” upon opening patient chart, informs provider of positive results and availability of CDI order set. The order set, based on SHEA practice guidelines 1) automatically fires task for isolation in the EHR  and 2) provides choice of order sentences for metronidazole and/or oral vancomycin based on severity of illness, with preselected high pharmacy priority. Retrospective EHR data was examined for all adult inpatients  with PCR-positive CDI  for one month before initiation of alert (n=41), and one month after alert (n=42). Setting is a 571 bed academic medical center with a hybrid medical record. Endpoints were time from positive lab results to 1) ordering of anticlostridial medication 2) receipt of first dose of medication and 3) posting of isolation (paper signage posted by Healthcare Epidemiology or automatic computer task). Independent t-tests were used for statistical analysis.

Results: When alert resulted in use of order set, significant improvement was seen in all end points. Mean time to order decreased from 256 to 91 minutes (m.) (p=.04), time to first dose decreased from 408 m. to 176 m. (p=.01), and time to isolation decreased from 1908 m. to 291 m. (p=<.0001). Effect of alert without order set use also decreased time to treatment but changes were not statistically significant: mean time to ordering of anticlostridial medications decreased by 89.2 m. (p=.229), mean time to first dose decreased by 129.4 m. (p=0.147), and mean time to isolation decreased by 306.6 m.  (p=.1225).

 

Time to Order

Time to 1st Dose

Time to Isolation

Before (n=41)

311.0

489.1

1741.4

After : alert only (n=42)

221.8

359.7

1434.8

          p

.2294

.1474

.1225

After alert: order set not used

 256.3 (n=19)           

408.1 (n=19)

1907.5 (n=26)

                    order set used

   91.0 (n=5)

176.0 (n=5)

 291.1 (n=12)

          p

.0410

.0105

<.0001

Conclusions: A computer alert, triggered by lab results, and linked to an order set significantly decreased time to treatment.  Effect of alert alone, while decreasing mean time in all endpoints, did not produce statistically significant change. Provider use of order set was inconsistent; barriers to order set use could not be explored in limited time frame of the study. The intervention changes process- its impact on clinical outcomes needs study in a larger sample over a longer time period.