Results

OESOPHAGEAL pH MONITORING RESULTS

The mean duration of pH recordings was 27.1 (0.49) hours (range 21.3 to 30.2 hours). Five of the 26 infants (19.2%) had an abnormal fractional reflux time of more than 10%20; seven (26.9%) had a value of between 5% and 10%, and 14 (53.8%) had a value below 5%.

CLINICAL PRESENTATION AND GOR

Parents of 12 infants (46.2%) gave a history of frequent vomiting or posseting after most feeds. The remaining infants had infrequent (n = 9) or no vomiting (n = 5). Infants with frequent vomiting had a significantly higher fractional reflux time than infants with infrequent or absent vomiting (p = 0.03). Four of the 12 infants with frequent vomiting or posseting had pathological reflux. The positive predictive value of frequent vomiting/posseting for pathological GOR was only 33.3%, and the χ² test was not significant (4/12 v 1/13, p = 0.09). However, the negative predictive value for frequent vomiting was 92.9%. There was no significant difference for the other reflux indices (table 1).

Seven infants (26.9%) were perceived to be irritable by their parents. Three of these infants had pathological GOR (positive predictive value 57.1%, χ² test 3/7 v 4/17, p = 0.06). There was no significant difference in reflux indices (fractional reflux time, number of reflux episodes, and duration of reflux episodes) between infants who were irritable and those who were not (table 1). None of the infants in the study presented with haematemesis. Two infants had apnoeic episodes in the neonatal period, which were thought to be unrelated to reflux. Four infants (15.4%) presented with frequent wheeze, only one of whom had pathological GOR.

FEEDING DETAILS AND GOR

Twenty of the 26 (76.9%) infants were breast fed at the time of enrolment. Only two infants presented with feeding difficulties as perceived by the parents. There was no significant age difference between breast and bottle fed infants (2.0 v 2.5 months, p = 0.38). Reflux indices were not statistically different between the two groups (table 1).

NUTRITIONAL ASSESSMENT AND CORRELATION WITH GOR

At birth, z scores were normally distributed for weight for age (mean 0.15 (0.20)) and weight for length for age (mean −0.29 (0.20)). Mean z scores at birth were not significantly different from zero (one sample t test: p = 0.46 and p = 0.16, respectively). At the time of pH monitoring, both z scores had fallen and were now significantly different from zero (z weight for age −1.00 (0.20), p = 0.0001; z weight for length for age −0.68 (0.18), p = 0.001). On paired t test there was a highly significant fall in z scores for weight for age (p < 0.0001), but not for weight for length for age (p = 0.22).

At the time of pH monitoring, eight infants (30.8%) were failing to thrive as evidenced by either a z score for weight for age below 2 or a fall in z scores for weight for age or weight for length for age by more than 2 SD between birth and pH monitoring. For the eight patients with failure to thrive, z scores for weight for age (−1.84 (0.29)v −0.58 (0.21); p = 0.0031) and weight for length for age (−1.35 (0.38) v −0.38 (0.18); p = 0.032) were significantly lower than in the remaining 18 infants. There was no significant difference in reflux indices (number of episodes per 24 hours, duration of reflux episodes, and fractional reflux time) between infants with and without failure to thrive (table 1).

MECONIUM ILEUS AND ILEAL ATRESIA

Six infants (23.1%) presented with intestinal obstruction at birth (five with meconium ileus, one with ileal atresia). The remaining 20 infants were found by routine neonatal screening at about 6 weeks of age. Although the infants who had presented with intestinal obstruction had slightly higher mean values for number of episodes per 24 hours and fractional reflux time, this did not reach statistical significance (table 1).

PANCREATIC STATUS AND GOR

Twenty two (84.6%) of the infants had signs of pancreatic insufficiency at the time of pH monitoring and were started on enzyme supplements. Only four infants (15.4%) had not developed steatorrhoea and were thriving well. Three of these had formal 72 hour faecal fat clearance measured, which was normal (⩾ 96%) in all three cases. Pancreatic status at the time of pH monitoring was not predictive of pathological reflux (table 1).

GENOTYPE AND GOR

Of the 26 infants, 16 (61.5%) were homozygous for the common deletion ΔF₅₀₈. Seven infants (26.9%) were compound heterozygous, and three (11.1%) had other mutations. The genotypes are detailed in table 2. All reflux indices were slightly more frequent in the group of homozygous infants. This reached statistical significance for the mean duration of reflux episodes. There was no significant difference for number of reflux episodes per 24 hours or the fractional reflux time (table 1).

CHEST RADIOGRAPHS AND CORRELATION WITH GOR

Of the 26 patients, 15 (57.7%) had a normal chest xray at the time of pH monitoring, and 11 (42.3%) had abnormal films. Of these, four had focal changes (consolidation or collapse) and seven had diffuse changes (hyperinflation and/or peribronchial thickening). Infants with normal x rays were slightly younger than infants with radiological abnormalities, although this did not reach statistical significance (1.8 (0.12) v 2.4 (0.45); p = 0.23, NS). The infants with a normal chest x ray had a significantly higher fractional reflux time than those with an abnormalx ray (table 1). All infants with abnormal chestx rays had a normal fractional reflux time < 10% (χ² test, p = 0.03).

Discussion

The first clinical description of a high incidence of GOR in cystic fibrosis as evidenced by barium study was published by Feigelson and Sauvegrain in 1975.4 Since then several studies have been published on this topic.5-11 16 Reported incidence figures for GOR vary widely between 25% and 100%.4 6 7 10 In our study, about one fifth of newly diagnosed infants with cystic fibrosis had pathological GOR. This incidence is considerably lower than in most studies. Vandenplaset al provided normal values for GOR in a large study of 509 healthy infants.18 They found that in the first 12 months of life a fractional reflux time of 10% was normal (95th centile), decreasing from about 13% at birth to 8% at 12 months. Using this study as a historical control, the risk of pathological GOR in cystic fibrosis was about four times the risk in healthy infants. Our study was conducted prospectively and is close to a population based study. The high incidence figures in other studies are most likely to be due to selection bias, enrolling mainly patients with symptomatic reflux. An alternative explanation is that the incidence of pathological reflux in cystic fibrosis may increase with age.

Frequent vomiting was associated with a significantly increased fractional reflux time, although the positive predictive value for frequent vomiting was only 33%. In contrast, the high negative predictive value of 93% suggests that absence of frequent vomiting is a useful negative predictor of significant reflux. Similarly, in a recent study on GOR in otherwise healthy irritable infants, the absence of frequent vomiting was also found to be a useful negative predictor.21

Failure to thrive is a common presentation in cystic fibrosis and may be compounded by GOR and loss of nutrients. Compared with z scores at birth we found a significant fall in z scores for weight for age at the time of pH monitoring, which coincided with diagnosis and start of enzyme supplementation in many of the infants. Other authors have found evidence of growth retardation at the time of diagnosis in infants with cystic fibrosis found at neonatal screening.3 This emphasises the need for early diagnosis in order to start early nutritional supplementation. Eight infants (30.7%) fulfilled our criteria for failure to thrive, with z scores 2 SD below the 50th centile for weight. There was no increased incidence of pathological reflux in these infants.

Sixteen infants were homozygous for ΔF₅₀₈, and seven were compound heterozygotes. There was no difference in clinical presentation between the two groups. Except for four (15.4%), all infants had developed pancreatic insufficiency at the time of pH monitoring. This is in keeping with a recent multicentre study that found similar rates of pancreatic insufficiency in infants with cystic fibrosis who were homozygous or compound heterozygous for ΔF₅₀₈.22 The mean duration of reflux episodes was significantly longer for infants who were ΔF₅₀₈ homozygous. This may indicate delayed oesophageal clearance in these patients. The significance of this finding, however, is uncertain and interpretation is hampered by small numbers, although phenotypic variations have been reported among different cystic fibrosis genotypes. There was no statistical difference in the remaining reflux indices for genotype or pancreatic status.

Six (23%) of the study infants presented with meconium ileus or ileal atresia. This is comparable with the incidence rate of 21% found for the cystic fibrosis cohort in the state of Victoria between 1973 and 1992.23 Infants with meconium ileus have a high morbidity secondary to abdominal surgery, often requiring prolonged hospital admission. They often need parenteral nutrition and are at particular risk of failure to thrive. Despite these concerns we did not find an increased incidence of pathological reflux in the six infants that presented with intestinal obstruction.

The mechanism of GOR in cystic fibrosis is not clear. Some investigators initially suggested that reflux may occur secondary to lung disease. They proposed that hyperinflation and stenting of the diaphragm caused widening of the angle of His and impairment of gastro-oesophageal competence.4 16 In our study, we found no correlation between severity of lung disease and severity of GOR. All infants with significant reflux had normal chest xrays, and pathological reflux appears to be present before lung disease is established.

Cucchiara et al showed that most reflux episodes in cystic fibrosis occurred during transient lower oesophageal sphincter relaxations and were not due to impaired lower oesophageal sphincter tone.9 24 Transient sphincter relaxations are increased during distension of the gastric fundus, and hyperalimentation of cystic fibrosis infants may be a predisposing factor. Delayed gastric emptying has been recognised as a risk factor for pathologically increased GOR. The information on gastric emptying in cystic fibrosis is not conclusive. Gastric emptying of liquids was initially thought to be delayed in cystic fibrosis.25 Two more recent studies found that gastric emptying of non-homogenised fat26 and of solids27 is faster in patients with cystic fibrosis than in normal individuals. Breast fed infants may have lower rates of GOR and more rapid gastric emptying than bottle fed infants.28 Although most infants in our study were breast fed, we did not find a difference in reflux indices compared with the bottle fed infants. The presence of pancreatic enzyme supplements may alter gastric emptying. However, infants receiving pancreatic enzyme supplements had no significant differences in GOR compared with infants not on enzyme replacement.

Besides its peptic complications, GOR may adversely affect pulmonary function and worsen early failure to thrive. Postural drainage chest physiotherapy is commonly prescribed in these infants and increases the risk of GOR.15 Absence of frequent vomiting was the only useful negative predictor for pathological GOR. We found no significant association for irritability, meconium ileus, pancreatic status, breast feeding, failure to thrive, genotype, and chest x ray findings. In the absence of clear clinical predictors, clinicians need a high degree of suspicion to make a diagnosis of pathological GOR in young infants with cystic fibrosis. The role of GOR as a contributor to development of lung disease clearly needs further study.