This was a prospective longitudinal observational study done in a Neonatal unit of a tertiary medical college hospital in North India over a 9 month period (May 2015–January 2016).
Inborn babies admitted to the NICU were eligible for enrolment if they were born between 28 and 34 weeks of gestational age and are on full gavage feeding. Those with encephalopathy (any grade according to Levene’s classification), major congenital malformation, oro-nasal malformation, receiving any respiratory support or sedative drugs, and had undergone any surgery were excluded.
The primary outcome was “Nutritive sucking” defined as a suck to swallow ratio of 2:1 over a 3-min observation of feeding on an empty breast assessed by videography at 7, 14, 21 and/or 28 days after enrollment. There were no prior studies done to evaluate PTSR scale (Appendix), therefore sample size estimation was done as per pilot study, it was decided a-priori to enrol 40 eligible babies in the study. In all enrolled babies maternal characteristics (age, socioeconomic status (using modified Kuppuswamy scale) [8], education, parity, relevant medical and obstetric history, and mode of delivery) were recorded. Neonatal characteristics of the enrolled babies (gestational age, birth weight, intrauterine growth status (Fenton’s growth chart) [9], Clinical Risk Index for Babies [CRIB] score [10], morbidities and feeding details) was also recorded.
Assessment for sucking Readiness: All enrolled babies were assessed by a single investigator (who was trained prior to starting the study in doing PTSR) once every day at a fixed feeding session for their sucking readiness while sucking on an empty breast (non-nutritive sucking) for 3-min using the PTSR. We have used the original PTSR scale. Physiological parameters (Respiratory rate, Saturation) were recorded for the safety of neonate before assessing PTSR score. Continuous monitoring of oxygen saturation was done during the entire period of feeding. No scores were given for the physiological parameters. Thereafter, the baby was assigned a score which was a sum total of score of behavioural state just prior to feeding, score of transition between behavioural state during handling/breastfeeding and score of feeding readiness behaviour (sucking, rooting, mouthing and showing interest at the breast). As soon as the babies attained an established nutritive sucking, they were no longer assessed. No additional support was given to the enrolled mothers. The standard NICU protocols were followed.
Videographic-recording. First video recording of feeding session was performed on day 7 of enrolment and thereafter weekly. However, neonates were also considered for assessing nutritive sucking earlier than the scheduled weekly assessment if there was an appreciable increase in the daily PTSR score. This however was in addition to the scheduled weekly assessment for nutritive sucking. Video graphic recording of the feeding by the baby on the breast were done using a high quality Nikon Coolpix L840, 16 megapixel camera. However, neonates were also considered for assessing nutritive sucking earlier than the scheduled weekly assessment if there was an appreciable increase in the daily PTSR score. This however was in addition to the scheduled weekly assessment for nutritive sucking. The dynamic video-graph frame ensured the capture of the maternal breast (to monitor latching of the baby), baby’s mouth and neck (to monitor sucking and swallowing). Based on these observation the suck:swallow ratio was estimated during play back of the videorecording, first the number of sucks were counted and then video was replayed to count number of swallows, and thus suck:swallow ratio was calculated independently by two of the investigators.
Statistical analysis
The predictors of nutritive sucking at 7, 14, 21 and 28 days were assessed by univariate analysis. Multivariate analysis (ANNOVA) was done for significant variables. Categorical data were compared by Chi Square/Fisher exact test. Sensitivity and Specificity of PTSR score was done using receiver operating characteristic curve (ROC) analysis. A p-value of 0.05 was taken as significant.