One physiatrist and three speech-language pathologists with more than 5 years' experience of working in the dysphagia clinic developed this scale, which is in reference to the BROSS approach. The BROSS approach consists of six consecutive developmental levels of feeding. These include non-nutritive sucking, obligatory phase, alternating phase, intermittent sucking phase, coordinated phase, and integrated phase. In the obligatory phase, infants suck in 10- to 20-suck bursts without breathing. As the infants mature, they show an alternating pattern of sucking and breathing. This means that they suck for a burst of 3 to 5 sucks and alternate with breathing, albeit tachypneically. In this phase, initially there are frequently longer sucking bursts accompanied by mild desaturation, with recovery during the tachypneic catch up. In the next intermittent sucking phase, infants take brief catch breaths once every 2 to 3 sucks, and longer sucking bursts appear with the catch breaths imbedded. The hallmark of the coordinated phase is when the infant develops a mature and coordinated sucking pattern with sucking bursts of 20 to 30 sucks, seamlessly integrating breathing with sucking and swallowing. In this phase, many infants have a sucking-swallowing-breathing (SSB) rhythm (1:1:1), and most babies develop their own coordinated pattern with modulated suction and expression. This phase is typically observed after infants' transition from the neonatal intensive care unit to home. The integrated phase is described as having full coordination of sucking/swallowing and breathing without increased work of breathing or tachypnea, clear demands to be fed and enjoyment of eating, and unique characteristics of social interactions between the baby and the primary caregiver.

The FSSPI consists of the following six items (Table 1): (1) non-nutritive sucking, (2) nutritive sucking, (3) sucking pattern, (4) sucking burst, (5) developmental stage of SSB, and (6) regularity of rhythm. Sucking patterns are classified as immature, transitional, and mature patterns, which are in reference to the Neonatal Oral-Motor Assessment Scale (NOMAS) [9]. An immature sucking pattern involves a short burst of 3 to 5 sucks. A transitional sucking pattern involves a medium burst of 5 to 10 sucks, while a mature sucking pattern involves a long burst of 10 to 30 sucks. The developmental stages of SSB are classified as previously mentioned in BROSS. The obligatory phase involves nutritive sucking repeated more than 10 times without breathing. The alternating phase involves frequent breathing, but longer sucking bursts that lack saturation are alternated with shorter bursts with stable saturation. In the intermittent phase, the ratio of SSB is 2 to 3:1:1, with 10- to 20-suck bursts and mild incoordination. In the coordinated phase, the ratio of SSB is 1 to 2:1:1 with 20- to 30-suck bursts. In the integrated phase, the rhythm of feeding and swallowing is regular.

In the FSSPI, we subdivided the framework of six consecutive developmental levels of feeding in BROSS. First, the stage in which even non-nutritive sucking cannot be performed was added. Second, before the obligatory phase, the stage in which nutritive sucking can be performed was included. Third, the alternative phase was subdivided in two stages according to the sucking pattern because the number of sucking bursts increases with the development of feeding and swallowing skills [9]. Thus, the scale score is represented by a 9-point scale from 0 to 8. A higher score indicates poor feeding and swallowing skills. Clinicians take 10 to 15 minutes to evaluate the VFSS findings of the infant using this scale.

One hundred thirty premature infants (gestational age <34 weeks) with a corrected age of less than 3 months who had undergone VFSS at the outpatient clinic or during inpatient consultation, between January 2008 and May 2014, were included in this study. Infants with craniofacial or oral cavity malformations, known clinical syndromes (for example, Down syndrome, etc.), and grade III or IV intraventricular hemorrhage or periventricular leukomalacia were excluded. Intraventricular hemorrhage or periventricular leukomalacia was confirmed by brain ultrasound or magnetic resonance imaging. Data were collected by a physiatrist and they included gestational age (GA) at birth, birth weight, corrected age (CA) at the time of performing VFSS, changes in body weight, and CA at the time of achieving full oral feeding.

VFSSs were conducted as described by Logemann [10] with some modifications [1112]. To evaluate infants, the fluoroscope table was tilted vertically and a feeder seat was placed on the ledge in a semi-reclined position at approximately 45°. Careful attention was paid to the stability of the positioning of the infant on the chair or footplate. Images of the swallowing process were taken in the lateral view (and occasionally in the anteroposterior view) using fluoroscopic equipment Shimavision 3200 HG (Shimadzu, Kyoto, Japan). In infants who were fed only milk, VFSSs were performed with bottle-feeding by mixing an undiluted liquid barium solution, Solotop solution 140 barium sulfate (Taejoon Pharm, Seoul, Korea) into the milk. The ratio of milk and barium solution was set at 2:1. The infants' own bottles and nipples were used. To monitor radiation exposure during examination, Dose Area Product (DAP) meter PD8100 (Toreck, Yokohama, Japan) was used. Radiation exposure was maintained at 400 µGycm² or less per examination based on the criteria of the Health Protection Agency (HPA), 2005. To minimize radiation exposure, examination time was limited to 2 minutes. Also, collimation of radiation field and pulsed radiation exposure technique were applied. VFSS was carried out when the pulse oximeter N-560 (Nellcor, Minneapolis, MN, USA) had been applied. Tolerance of oxygen saturation and heart rate were reduced to 90% and baseline heart rate ±10% of baseline heart rate, respectively.

For the verification of intra-rater reliability, one physiatrist and three speech-language pathologists with more than 5 years of experience in VFSS evaluated 18 randomly extracted VFSS videos. The same videos were evaluated again 2 weeks later. For the verification of inter-rater reliability, four evaluators assessed these 18 randomly extracted VFSS videos.

For the verification of validity, the correlation between the FSSPI and clinical indicators that reflect the maturation of feeding and swallowing, such as GA at birth, birth weight, CA at the time of performing VFSS, changes in body weight, and CA at the time of achieving full oral feeding, was analyzed [1314].

The Cronbach's alpha coefficient was calculated to examine the reliability of the FSSPI. To examine the normality of clinical indicators, we applied the Shapiro-Wilk normality test and the Kolmogorov-Smirnov normality test. For the verification of validity, we used the Pearson correlation coefficient in parametric analysis and the Spearman correlation coefficient in nonparametric analysis. The level of significance was set at p<0.05. All analyses were performed using SPSS ver. 23.0 (IBM, Armonk, NY, USA).

Eighty-three infants were included in the final analysis. Forty-seven infants who did not meet the inclusion criteria were excluded: 8 infants had a craniofacial or oral cavity malformation, 6 infants were suspected of having a clinical syndrome, and 33 infants had grade III or IV intraventricular hemorrhage or periventricular leukomalacia. Of the 83 infants included, 45 were boys and 38 were girls. The mean GA was 27.3±2.8 weeks. The mean body weight at birth was 1.1±0.4 kg. The mean CA at the time of performing VFSS was 2.1±54.4 days and the mean chronological age when VFSS was performed was 3.6±1.9 months. The corrected age at the time of achieving full oral feeding was recorded in 51 out of the 83 infants (mean age, 1.3±0.7 months) (Table 2). Information on weight gain was available in 23 infants.

The FSSPI showed a high degree of intra-rater reliability (Cronbach's alpha coefficient=0.963, p<0.01) and inter-rater reliability (Cronbach's alpha coefficient=0.970, p<0.01).

There was a significant negative correlation between the FSSPI scale score and CA at the time of performing VFSS (Spearman correlation coefficient=-0.706, p<0.01) (Fig. 1A). Further, a significant positive correlation was observed between the FSSPI scale score and CA at the time of achieving full oral feeding (Pearson correlation coefficient=0.514, p<0.01) (Fig. 1B). Additionally, there was a significant negative correlation between the FSSPI scale score and weight gain: between the 1st and 2nd month after birth (Pearson correlation coefficient=-0.437, p<0.05) (Fig. 2A) and between the 2nd and 3rd month after birth (Pearson correlation coefficient=-0.432, p<0.05) (Fig. 2B), respectively.