Preoperative fasting guidelines are used to reduce the risk of rare pulmonary aspiration of gastric contents (1–4). Although gastric fluid volume (GFV) is a limited surrogate marker for pulmonary aspiration risk (5–7), studies evaluating the clinical safety of fasting guidelines rely on assuring small preoperative GFV (8–10). Ingestion of clear liquids up until 2 h before surgery results in acceptably small GFV (10), but data regarding infant formula and GFV are scant (11). Recent American Society of Anesthesiologists (ASA) guidelines suggest a 6-h fast after infant formula but recommendations in the literature range from 4 to 8 h and are largely based on expert opinion and institutional experience (2,5,7,12). We hypothesized that infants allowed formula up to 4 h before surgery would have GFVs no larger than those restricted to 8 h for formula/solids and 2 h for clear liquids. In addition, we theorized that infants allowed formula at 4 h would be less irritable and hungry, resulting in improved patient comfort and parental satisfaction.
After IRB approval and parental permission (informed consent), we randomized 97 infants scheduled for elective surgery into 1 of 2 groups. Group T (tradi-tional fast, n = 58) was allowed no solids/milk/formula <8 h before surgery, but clear liquids were permitted up to 2 h before surgery. Group L infants (liberalized fast, n = 39) were allowed formula until 4–6 h before surgery, but no solids or cow’s milk within 8 h. Cow’s milk was not allowed within 8 h because it forms solid curds when exposed to human stomach acid and because its fatty contents slow gastric transit time. Other subject inclusion criteria were age ≤9 mo at enrollment, ASA physical status I or II, and elective surgery requiring tracheal intubation. Exclusion criteria included exclusively breast-fed infants and those with gastrointestinal diseases that impede gastric emptying (pyloric stenosis, duodenal atresia), diminish gastrointestinal motility (preoperative opioid administration), or result in reflux. The procedure was randomized by giving the infant’s parent a coded envelope containing the computer-generated group assignment and a diary form to record preoperative feeds (clear liquid/formula type, quantity ingested, and time). A nurse telephoned parents the evening before surgery to give the scheduled surgery time and review feeding instructions.
On the day of surgery, a nurse confirmed feeding diary entries with the parents, but investigators remained unaware of the subject’s treatment assignment. Immediately after their infant was taken into the operating room, parents completed a short survey by marking 3 10-cm linear analog scales indicating their child’s irritability and hunger (0 = calm to 10 = extremely irritable/hungry) and their own overall satisfaction with the preoperative fasting period (0 = totally satisfied to 10 = totally dissatisfied).
Preoperative oral atropine (20 μg/kg) was administered and anesthesia induced by inhalation of nitrous oxide and either halothane or sevoflurane in oxygen. After tracheal intubation, one of the investigators blinded to infant treatment assignment removed the GFV with a 14F multiorificed orogastric tube. We have previously shown that with this blinded gastric aspiration technique, 96%–97% of GFV is recovered (13,14). Gastric fluid pH and character were recorded as was any evidence of emesis.
In our previous study, we reported that average GFV for fasted pediatric patients is 0.40 ± 0.45 mL/kg (13). We considered a 0.5 sd difference (1–2 mL absolute volume) to be both reliably measurable and clinically important. By using the Mann-Whitney test for nonparametric continuous data, we determined a sample size of 140 subjects (69 per group) to be sufficient to detect a difference of this magnitude, assuming an α = 0.05 with 80% power. An interim analysis was planned at an approximately 50% recruitment to assure protocol safety and assess significance. Kruskal-Wallis analysis of variance for ordinal data was used to determine whether GFV (mL/kg) varied by fasting interval. Spearman correlation tests were used to explore dependence of GFV (mL/kg) on age, feeding volume, and fasting duration. The Mann-Whitney U-test was used to compare differences between groups in parental assessment of subject irritability and hunger and parental satisfaction. We considered the per protocol analysis to be more clinically relevant than the intent-to-treat analysis because prolonged fasting for both groups could create bias favoring a lack of difference. For all comparisons, P = 0.05 was considered statistically significant.
The 97 subjects were enrolled between February 1998 and November 2000. Many parents did not wake infants early enough for feeding and, combined with changes in scheduled start times, 30 infants had prolonged fasts (formula >6.5 h or clear liquids >5.5 h) as defined by our entry criteria. The intent-to-treat analysis consisted of all 97 subjects and the per protocol analysis only 67 (Group T, n = 36; Group L, n = 31). The interim analysis revealed that our infant population had smaller GFV (0.18 ± 0.34 mL/kg) than the population of our previous study. In addition, during the conduct of the study, we found we could reliably quantitate a minimum of 1.5 mL. Using the interim data, we then reestimated the required sample size to detect a 1.5-mL GFV difference. The subsequent reduced sample size allowed earlier termination of the trial. We found, using a one-sided t-test for independent samples at the 0.05 level of significance, with 80% power to detect a 0.62 sd (approximately 1.5 mL) difference, that the GFVs of the two populations following protocol (n = 67) were not different.
Mean age, weight, sex distribution, and feeding volume were similar (Table 1). Ages ranged from 0.7 to 10.5 mo. Several subjects were older than the ≤9-mo inclusion criterion because of the delay between patient enrollment at the surgical office visit and the date of surgery itself. Formula types included: Alimentum, Carnation, Enfamil, Isomil, Neosure, Prosobee, and Similac. GFV and pH were not different between groups (Table 2, Fig. 1). Sixty-four percent of Group T and 55% of Group L had GFV = 0. The largest measured GFV was 1.27 mL/kg in Group T and 1.87 mL/kg in Group L (the latter 6.4-mo-old infant was fed 3.9 h before the induction of anesthesia.) Across all subjects, GFV (mL/kg) increased with age (Spearman correlation coefficient of +0.23, P = 0.03) (Fig. 2) but did not correlate with either fasting interval (Fig. 1) or volume of last feed. There was evidence of formula in the residual GFV of 9 infants, including 1 fasted 10 h (Fig. 3). In these instances, the GFV appeared as a thin white-tinged fluid, with the exception of one infant in Group L who had several small particles in an otherwise clear gastric aspirate. No subject had evidence of regurgitation of gastric contents or pulmonary aspiration. Infant irritability and hunger and parental satisfaction were not significantly different between treatment groups (Table 3).
Whereas our ultimate goal in restructuring fasting guidelines is to reduce, or at least not to increase, the incidence of pulmonary aspiration, demonstrating that different fasting protocols change the risk of this rare event would require a very large subject population. Assuming an overall pediatric perioperative pulmonary aspiration incidence of 1:3000 (3,4), a comparative fasting study would need to include a sample size of >30,000 to assess differences between protocols. Because clinical outcomes in the majority of pulmonary aspiration events are benign, evaluating incidence differences in the rare cases of severe morbidity and mortality associated with aspiration would require an even larger sample size. In the absence of such a national study garnering the magnitude of subjects needed to address differences in pulmonary aspiration risk, investigators continue to rely on GFV in assessing relative safety (8–10).
We found that average GFV was small in both traditional and liberalized fasting groups. The 0.18 ± 0.34 mL/kg average GFV for protocol subjects was smaller than GFV of historical controls (0.3 ± 0.9; 0.4 ± 0.45 mL/kg) associated with what has become the traditional fast—clear liquids allowed up until 2 hours before surgery and solids/formula 6–8 hours before (10,13). We did not allow the liberalized formula-fed group to also ingest clear liquids two hours before surgery because we were concerned about the possibility of increased GFV with two large-volume feedings only two hours apart. In the present study, GFV (in mL/kg) increased with age and it is possible that the larger GFVs reported in earlier studies may be related to the inclusion of older infants and children. It is also possible that our routine preoperative use of the muscarinic antagonist, atropine, decreased gastric secretions to a greater extent than it may have decreased gastrointestinal motility, leaving, in balance, smaller GFV.
Although average GFV was demonstrably small, some subjects had larger residual GFV. With regard to formula feeding in particular, van der Walt et al. (11) were concerned by the occasional infant who had >0.4 mL/kg GFV 3–4 hours after ingesting formula and recommended a more prolonged fast. We believe there to be no basis for this historical concern about GFV >0.4 mL/kg because large GFV occasionally does occur, even after longer fasts. The 95th percentile for GFV in one reported series of traditionally fasted healthy pediatric patients was 1.25 mL/kg with an upper bound of 4.1 mL/kg (13), and in a study limited to infants <1 year of age, 17% had GFV ≥0.4 mL/kg and 4% had GFV ≥1 mL/kg (10). The latter study concluded that “formula-fed infants under one year of age who ingest clear liquids up to two hours prior to surgery are at no greater risk for pulmonary aspiration of gastric contents than are older children reported in previous studies.” It would seem then that after either liberalized or longer traditional fasting intervals, occasional healthy subjects will have GFV between 1–4 mL/kg, and that further restricting ingestion is not guaranteed to empty the stomach completely.
With regard to predictably small preoperative infant GFV, further support for shortening the required formula fasting interval comes from the gastroenterology and radiology literature. The emptying half-life of formula from the stomach was reported to be 51 minutes using a simple marker-dilution technique (15). Subsequent studies with more sophisticated methodologies have shown similar results. Normal infants who ingest 99m Tc-labeled formulas empty 70% of the meal from the stomach within 1 hour (16). Using the 13C-octanoic acid breath test, the mean half-emptying time for the formula-fed infants was estimated to be 65 minutes (range, 27–98 minutes) (17). Assuming first-order kinetics and a conservative gastric formula half-life of 60 minutes, approximately 94% of the meal should leave the normal infant’s stomach by 4 hours. The results of our study are in general agreement with this: with an average last feed of 4.5 oz. = 135 mL, 8 mL of residual GFV = 1 mL/kg might be expected after 4 hours, 0.5 mL/kg at 5 hours, and 0.25 mL/kg at 6 hours.
Formula type may affect GFV because gastric emptying in infants may be influenced by osmolality, volume, protein content, energy density, pH, and curdling properties of the feedings (18). Many of the formulas used in this study (Carnation, Enfamil, Neosure, and Similac, but not Alimentum, Isomil, Prosobee) contain cow’s milk components and clinically, these formulas may curdle, but it is thought that they are less likely than cow’s milk to do so. Because of the variety of formulas used in this study, sample size for each was too small to note GFV differences based on formula type. No one formula resulted in particularly large GFVs, however. In designing our study, we did not limit formula type in order to maximize recruitment and enhance external validity. We did not allow supplementation with lipids or rice cereal.
Evidence of residual formula was found in the stomachs of 9 subjects (9%), but total volumes were scant. One formula-fed infant had a few small particles in the recovered fluid. Small amounts of residual formula were found more often in infants fasted <6 hours, but even a 10-hour fast did not preclude the possibility. Although it could be argued that practitioners should reduce the possibility of formula in the GFV by restricting formula to six hours or more before surgery, the likelihood of increased pulmonary aspiration risk of these white-tinged residua with pH not different from traditional fast GFV is, in our opinion, small.
Infant age correlated with GFV (in mL/kg) although the positive correlation was very slight and had no clear explanation. Theoretical explanations might include age-dependent variation in gastric emptying, formula preferences, feeding frequency, and patterns not fully elucidated on our feeding diary. This finding may substantiate the clinical practice of varying required fasting intervals based on age (12). After lengthy departmental discussion at our institution, this age-dependent finding was used to argue, preliminarily, for the conservative guideline of liberalizing the formula fast only in the infant group ≤6 months of age, although there was no absolute step-up in GFV above this age (P = 0.134). The data, especially in the formula-fed subjects, do demonstrate acceptably small GFV in those subjects older than six months, and other practitioners may choose to interpret that data to liberalize feedings in these older infants as well.
Contrary to our expectations, both infant irritability and hunger, and parental satisfaction were not significantly affected by allowing formula up to four hours before surgery. Interestingly, in post hoc analysis, parental scoring of infant irritability and hunger did not correlate with GFV. It may be that the surrogate measures of infant well-being, parental reports in this case, are subject to discordance and/or may not be sensitive enough to detect differences. There were trends toward improved scores in the liberalized fast group, but with positively skewed reporting in both groups (our research nurses believed that most parents were delighted simply to participate at all) and with relatively small sample sizes, the differences were not statistically significant.
Finally, this study demonstrates an important contextual message: strictly following fasting protocols is challenging in the dynamic arena of pediatric anesthesia and surgery. Although the intended fasting duration had been only 2 hours for clear liquids and 4 hours for infant formula, 30 (31%) had prolonged fasts with 19 (20%) exceeding 8 hours. Individual patient sleep cycles and surgery schedule changes will unavoidably contribute to prolonged fasting intervals. Liberalized fasting guidelines were not misinterpreted by parents as unbounded feeding allowances, and no misinterpretations interfered with anesthesia and surgery start times in this study. Our primary analysis was based on those who did follow protocol, within certain bounds, to concentrate on fasting limits.
In conclusion, we have shown residual GFV to be no larger in healthy infants formula-fasted for four to six hours than it is in infants formula-fasted for more than eight hours but allowed clear liquids until two hours. In two surveys, institutional policies were shown to be equally divided between four- and six-hour fasts for formula in young infants (5,12). ASA guidelines recommend a six-hour fast after formula or cow’s milk (2). Our study contributes further physiologic data to making decisions about appropriate fasting intervals in this patient population. Interpretation of the clinical significance of small residual GFV, which may have visible formula, will impact clinicians’ decisions to permit formula four to six hours before surgery. Although GFV increases slightly with age, there is no “step-up” GFV in older infants that would restrict applying the four-hour formula-fasting guideline only to a younger patient subpopulation.
We thank our nurse practitioners, especially Kathy Sharp, CRNP, who helped to identify and recruit subjects for this study. Also, we thank the operating room faculty and staff at the Children’s Hospital of Philadelphia for their continuing support of, and enthusiasm for, our endeavors to improve comfort and safe care of infants coming for surgery. We extend special thanks to Ronald S. Litman, DO, for his review and commentary and to Alison Cook-Sather, PhD, for her assistance in editing the manuscript.
1. Olsson GL, Hallen B, Hambraeus-Jonzon K. Aspiration during anesthesia: a computer-aided study of 185,358 anaesthetics. Acta Anaesthesiol Scand 1986; 30: 84–92.
2. Warner MA, Caplan RA, Epstein BS, et al. Practice guidelines for preoperative fasting and the use of pharmacologic agents to reduce the risk of pulmonary aspiration: application to healthy patients undergoing elective procedures—a report by the American Society of Anesthesiologists Task Force on Preoperative Fasting. Anesthesiology 1999; 90: 896–905.
3. Warner MA, Warner ME, Weber JG. Clinical significance of pulmonary aspiration during the perioperative period. Anesthesiology 1993; 78: 56–62.
4. Warner MA, Warner ME, Warner DO, et al. Perioperative pulmonary aspiration in infants and children. Anesthesiology 1999; 90: 66–71.
5. Emerson BM, Wrigley SR, Newton M. Pre-operative fasting for paediatric anaesthesia: a survey of current practice. Anaesthesia 1998; 53: 326–30.
6. Schreiner MS. Gastric fluid volume: is it really a risk factor for pulmonary aspiration? Anesth Analg 1998; 87: 754–6.
7. Splinter WM, Schreiner MS. Preoperative fasting in children. Anesth Analg 1999; 89: 80–9.
8. Schreiner MS, Triebwasser A, Keon TP. Ingestion of liquids compared with preoperative fasting in pediatric outpatients. Anesthesiology 1990; 72: 593–7.
9. Splinter WM, Schaefer JD. Unlimited clear fluid ingestion 2 hours before surgery in children does not affect volume or pH of stomach contents. Anaesth Intensive Care 1990; 18: 522–6.
10. Litman RS, Wu CL, Quinlivan JK. Gastric volume and pH in infants fed clear liquids and breast milk prior to surgery. Anesth Analg 1994; 79: 482–5.
11. van der Walt JH, Foate JA, Murrell D, et al. A study of preoperative fasting in infants aged less than three months. Anaesth Intensive Care 1990; 18: 527–31.
12. Ferrari LR, Rooney FM, Rockoff MA. Preoperative fasting practices in pediatrics. Anesthesiology 1999; 90: 978–80.
13. Cook-Sather SD, Liacouras CA, Previte JP, et al. Gastric fluid measurement by blind aspiration in paediatric patients: a gastroscopic evaluation. Can J Anaesth 1997; 44: 168–72.
14. Cook-Sather SD, Tulloch HV, Liacouras CA, et al. Gastric fluid volume in infants for pyloromyotomy. Can J Anaesth 1997; 44: 278–83.
15. Cavell B. Gastric emptying in preterm infants. Acta Paediatr Scand 1979; 68: 725–30.
16. Papaila JG, Wilmont D, Grosfeld JL, et al. Increased incidence of delayed gastric emptying in children with GER. Arch Surg 1989; 124: 933–6.
17. Van Den Driessche M, Peeters K, Marien P, et al. Gastric emptying in formula-fed and breast-fed infants measured with the 13C-octanoic acid breath test. J Pediatr Gastroenterol Nutr 1999; 29: 46–51.
18. Tolia V, Kuhns L, Kauffman R. Correlation of gastric emptying at one and two hours following formula feeding. Pediatr Radiol 1993; 23: 26–8.