Reduction of Prescription Errors in Neonatal Intensive Care Unit:A Quality Improvement Initiative

Errors in Antimicrobial Prescription and Administration in Very Low Birth Weight Neonates at a Tertiary South African Hospital

BACKGROUND

Antimicrobial prescription and administration-related errors occur frequently in very low birth weight (VLBW; <1,500 g) neonates treated for bloodstream infections (BSI).

METHODS

Antimicrobial prescriptions for the treatment of laboratory-confirmed BSI were retrospectively analyzed for VLBW neonates at Tygerberg Hospital, Cape Town, South Africa (1 July 2018 - 31 December 2019), describing antimicrobial type, indication, duration of therapy and BSI outcomes. The prevalence of, and risk factors for prescription (dose, interval) and administration errors (hang-time, delayed/missed doses) were determined.

RESULTS

One hundred and sixty-one BSI episodes [16 (9.9%)] early-onset, 145 [90.1%] healthcare-associated) affected 141 neonates (55% male, 25% born to mothers living with HIV, 46% <1,000 g birth weight) with 525 antimicrobial prescription episodes [median 3.0 (IQR 2-4) prescriptions/BSI episode]. The median duration of therapy for primary BSI, BSI-associated with meningitis and BSI-associated with surgical infections was 9, 22, and 28 days, respectively. The prevalence of dose and dosing interval errors was 15.6% (77/495) and 16.4% (81/495), respectively with prescription errors occurring most commonly for piperacillin-tazobactam and vancomycin given empirically. Administration errors were less frequent [3.8% (219/5,770) doses missed; 1.4% (78/5,770) delayed], however 64% had a hang-time (time from sepsis diagnosis to 1st dose of antimicrobial) exceeding 60 min. On multivariable analysis, postnatal age >7 days was associated with prescription errors (p = 0.028). The majority of neonates with BSI required escalation of respiratory support (52%) and 26% required intensive care admission. Despite fair concordance between empiric antimicrobial/s prescription and pathogen susceptibility (74.5%), BSI-attributable mortality in this cohort was 30.4%.

CONCLUSION

VLBW neonates with BSI's were critically ill and had high mortality rates. Hang-time to first antimicrobial administration was delayed in two-thirds of BSI episodes and prescription errors affected almost 1 in 6 prescriptions. Targets for intervention should include reducing hang-time, use of standardized antimicrobial dosing guidelines and implementation of antimicrobial stewardship recommendations.