P0275 - Training the Dubowitz Gestational Age Assessment in Low- and Middle-Income Countries: Feasibility and Inter-rater Agreement

Background
Globally, prematurity is the leading cause of under-five mortality. Differentiating premature infants from term-growth restricted infants can be challenging in low- and middle-income countries (LMICs). The Dubowitz exam uses 21 neuromuscular and physical signs to estimate gestational age. It is the most accurate neonatal assessment compared to ultrasound dating; however, the exam is relatively long and requires specialized training which may inhibit widespread use in LMICs.

Objective
To evaluate a protocol for teaching the Dubowitz exam by assessing the agreement of trainee-trainer scores and to explore implementation barriers to the Dubowitz exam in two LMIC settings.

Design/Methods
We taught the Dubowitz exam during two half-day workshops in Malawi and India. We used didactic presentations followed by small group skills training. We compared agreement of scoring between the trainer (neonatologist) and trainees by calculating the percent agreement for individual signs. We used Bland Altman analysis to compare the gestational age determined by the full Dubowitz score by trainer and trainees. Trainees completed pre- and post-training surveys regarding their perception of the Dubowitz exam and the feasibility of implementation in their clinical setting.

Results
Almost half of the trainees had previous experience using either the Ballard or Dubowitz exams. On average, trainees took 14 minutes (SD 4.5 min) to complete the exam. Amongst 21 individual exam components, certain signs, such as skin texture, ear recoil, foot surface and leg recoil, were more reliable than others, such as testes and edema, that had wider inter-rater variation. When comparing the agreement of calculated gestational age between trainee and trainer, the bias on Bland Altman plot was -0.47 weeks with 95% limits of agreement of -3.56 and 2.63 weeks (Figure). After the training, all participants responded that they felt either moderately or very confident in their ability to perform the exam.

Respondents identified a number of potential barriers to implementation of the Dubowitz. They expressed concerns about the time to learn and perform the exam. They also thought that parents may perceive the exam to be painful or cause harm (Table).

Conclusion
Adequate training of the Dubowitz exam may be feasible in a one-day workshop. The trainee estimates of gestational age (as compared to the trainer) were within ± 3 weeks of the trainer, a value similar to what has been seen in studies validating the Dubowitz to the best obstetric estimate. A few signs had poor trainee-trainer agreement, warranting focus during subsequent trainings. While the Dubowitz exam may be one of the more accurate neonatal methods to assign gestational age in settings with limited ultrasound access, barriers such as time to learn and perform the exam may hinder widespread implementation.