Freedman, Stephen B.*; Tung, Connie†; Cho, Dennis†; Rumantir, Maggie*; Chan, Kevin J.*
Acute gastroenteritis accounts for >1.5 million outpatient visits annually in the United States and 13% of all of the hospitalizations among young children (1,2). Despite clinical trial evidence that ondansetron use reduces the frequency of vomiting, intravenous rehydration, and hospitalization (3–8), guidelines recommend against the pharmacologic treatment of vomiting (1), citing a variety of reasons including safety concerns, adverse effects, and a shift away from appropriate fluid, electrolyte, and nutritional therapy (1,9,10).
These concerns have not dissuaded physicians from prescribing antiemetics. In 2005, in the United States, 23% of outpatients younger than 20 years old received a prescription for an antiemetic (11), most commonly promethazine, which now has a black-box warning against its use in children younger than 2 years old (12). More recently, American pediatric emergency medicine physicians have indicated that ondansetron has become their agent of choice (13).
Because ondansetron has entered into widespread use (14), we sought to evaluate how its usage correlated with clinical outcomes. We hypothesized that increasing ondansetron usage would result in a corresponding decrease in the administration of intravenous rehydration, shorter length of stay, and reduced hospitalization and emergency department (ED) revisit rates.
Study Setting and Design
The Hospital for Sick Children is a tertiary care referral hospital in downtown Toronto. The ED treats approximately 60,000 children annually. The ondansetron oral disintegrating tablet became ward stock in July 2005. The administration of a single oral dose of ondansetron to children with evidence of dehydration and ongoing vomiting in the ED is recommended in our institution's clinical practice guideline. The recommended dose provided is within the previously evaluated dose range of 0.13 to 0.26 mg/kg (4,15). This retrospective report includes data from a consecutive series of children younger than 18 years old who presented to the ED between July 1, 2003 and June 30, 2008 with acute gastroenteritis.
Potentially eligible children were identified by searching the ED database for relevant International Classification of Diseases (ICD) -9-Clinical Modification (CM) (2003–2007) and -10 (2007–2008) codes (003.0, 007, 007.9, 008.5, 008.6, 008.69, 008.8, 009.3, 276.51, 787.0, 787.01, 787.03, 787.91). Acute gastroenteritis was defined by the presence of either vomiting or diarrhea for <14 days (9) and the absence of an alternative diagnosis (eg, inflammatory bowel disease exacerbation). Twenty percent of visits identified as potentially eligible were selected at random for chart review. All of the visits associated with a visit by the same child during the preceding 7-day period had the earlier visit coded as the index visit.
Data were extracted by 3 data abstractors who were blinded to the study's hypothesis. Medical records were reviewed from our Electronic Patient Chart system using a set of precise operational definitions for the key variables and uniform procedures for missing, conflicting, or ambiguous data. A standardized data collection instrument was used to record demographics, history and physical examination findings, intravenous fluid and medication administration, laboratory investigations, disposition, adverse events, and revisit data. Weekly meetings occurred during the data abstraction period to resolve disputes and review coding rules. A subset of randomly selected charts (10%) was independently abstracted by one of the investigators (M.R.) to evaluate inter-observer agreement.
Key Outcome Measures
Our primary outcome was the correlation over time between the proportion of children administered ondansetron and the proportion of children who received intravenous rehydration. Secondary outcomes included the correlation between ondansetron administration and the need for ED revisits, hospitalization, and length of stay.
Medical history was considered as a 3-level categorical variable: none, mild systemic disease, and severe systemic disease (ie, likely would affect treatments administered). The Canadian Triage Acuity Scale score, a clinical variable, is assigned by the triage nurse and measures the acuity of the child's illness (16). Recorded temperatures were adjusted for location of measurement (17). Children with an adjusted temperature ≥38.0°C were considered to have a fever (17). General appearance was classified for all of the children by the data abstractor as well (“well appearing,” “no apparent distress,” “alert,” “normal mental status,” “interactive,” “smiling”) or unwell (“sick,” “toxic,” “shocky,” “decreased mental status,” “lethargic,” “unresponsive,” “irritable,” “fussy,” “inconsolable,” “not looking well,” “poor or decreased perfusion,” “decreased pulses”) (18). Descriptors that did not meet the above definitions were labeled “unclear.”
All of the variables were analyzed based on their group assignment, which was determined by the timing of the ED visit, with groups running from July 1 to June 30 for each of the 5 time periods (2003–2008). Frequency counts and percentages are given for discrete variables, means with standard deviations are provided for continuous variables. The χ2 test was used for discrete variables, with Fisher exact test used when appropriate. Continuous variables were compared between time periods using analysis of variance. We assessed interobserver agreement for 6 outcome variables (intravenous fluid administration, ondansetron administration, hospitalization, return visit within 7 days, return visit requiring intravenous rehydration, return visit requiring hospitalization) with the κ statistic.
A regression analysis of an interrupted time-series dataset was conducted using segmented logistic regression, which divides a time series into pre- and postintervention segments (19). Such analyses allow an assessment of how much an intervention affected outcomes immediately and over time (19). Because our data revealed that ondansetron usage gradually increased during the period July 2005 to June 2006, we chose this interval as the intersection between segments (ie, the intervention). A regression model has 2 key parameters: a level and a slope. The difference between the 2 segments can be quantified by testing the change in these parameters. A change in level indicates an absolute change between the last measurement of the number of intravenous insertions in the preintervention segment and the first measurement of the postintervention segment; a change in slope indicates a change in the rate of intravenous insertions within each segment.
Because events closer together in time tend to be more similar than events further apart (autocorrelation) (20), the model residuals may not be independent (21) and may result in biased standard deviations, which can affect tests of significance (22). We evaluated and detected the presence of autocorrelation using the Durbin-Watson statistic. Autocorrelation was then corrected for by including a term in the regression model. Similarly, seasonal autocorrelation (annual trends) was evaluated and incorporated into the model.
The outcomes of interest in our time-series analysis were change in the level pre- and postintervention, change in the slope pre- and postintervention, and estimation of monthly average intervention effect considering the outcome (ie, intravenous rehydration rate) had the intervention (ie, increasing ondansetron usage) not occurred (23). This is estimated by comparing the intravenous insertion rate in the absence of ondansetron administration and the outcome with its use (Fig. 1).
All of the statistical tests were 2-sided and evaluated at the 5% level of significance. Statistical analysis was conducted with the use of SPSS (Windows version 16.0; SPSS Inc, Chicago, IL). The hospital's research ethics board approved the present study, and given the retrospective nature of the study, they waived the requirement for written informed consent.
During the 5-year study period, there were 22,125 potentially eligible visits identified. A total of 4064 charts of eligible children were reviewed representing 4425 patient visits. A total of 3508 visits were included in our final analysis (Fig. 2). Baseline features are reported by time period (Table 1) and ondansetron administration status (Table 2).
ED Interventions and Outcomes
During the 5-year study period, there was a significant reduction in the use of intravenous rehydration from 26% (312/1209) in the preintervention period to 14% (248/1735) in the postintervention period (P < 0.001). During the same time periods, ondansetron use rose from 1% (9/1209) to 18% (317/1735) (Fig. 3A). The reduction in intravenous rehydration between the first and the final year of the study was most notable in children 1 year old or older (99/390 (24%)–124/862 (14%); difference −11.0%; 95% confidence interval [CI] −16 to −6) compared with those <1 year old (27/149 (18%)–36/283 (13%); difference −5.4%; 95% CI −13 to 2). Similarly, a larger increase in ondansetron administration occurred among children 1 year old or older (0/390 (0%)–181/862 (21%); difference 21%; 95% CI 18–24) compared with those younger than 1 year old (0/149 (0%)–25/283 (9%); difference 9%; 95% CI 6–12).
Time-series analysis showed a statistically significant downward-level break (ie, a change in the number of absolute intravenous insertions between the pre- and postlevel intervention; P = 0.03) after ondansetron introduction. After adjustment, the intravenous insertion rate fell from 21.2 to 10.2 intravenous insertions per 100 gastroenteritis cases. The trend (ie, the change in intravenous insertions per month before and after the introduction of ondansetron) was not significant (P = 0.59). There was also evidence of first-degree autocorrelation (P < 0.001); thus, a regression parameter was inserted. There was no statistical significant evidence of seasonal autocorrelation.
During the study time period, the mean length of stay for children diagnosed as having acute gastroenteritis declined from 8.6 ± 3.4 in 2004 to 2005 to 5.9 ± 2.8 hours in 2007 to 2008, P = 0.03 (Fig. 3B). The reduction was similar among children 2 years old or younger (−2.83 hours) compared with those older than 2 years old (−2.47 hours). Among children diagnosed as having gastroenteritis there was a reduction in return visits (18% to 13%; P = 0.008; Fig. 3C). Among children diagnosed as having acute gastroenteritis at the initial visit, the proportion requiring an intravenous insertion at a subsequent visit declined (7%–4%; P = 0.02); there was no change in the need for hospitalization (P = 0.32). No adverse effects were documented that described any events construed to be related to ondansetron administration.
Interobserver agreement was assessed for 362 patients. Agreement for the outcome variables evaluated was good, with κ values ranging from 0.82 (95% CI 0.78–0.86) for return visits requiring intravenous rehydration to 0.97 for ondansetron administration (95% CI 0.96–0.97).
We report an increase in the use of ondansetron in children with acute gastroenteritis, from 1% to 18%, during a 5-year period. This corresponded with a 46% relative decrease in the use of intravenous rehydration, from 26% to 14%, with evidence of a downward-level break following the introduction of ondansetron. This reduction occurred in conjunction with a reduction in length of stay, ED revisits, and revisits requiring intravenous rehydration.
Randomized clinical trials and a meta-analysis have demonstrated that ondansetron administration can reduce the need for intravenous rehydration (3,6,8,13). The estimated absolute risk reduction in these clinical trials ranged from 15% (8) to 33% (6), with corresponding relative risk reductions of 55% (4) to 68% (8). Thus, the finding of a 14% absolute and 52% relative risk reduction following the introduction of ondansetron in our ED is in keeping with previous reports. These benefits, although slightly lower than those achieved in clinical trials, demonstrate that ondansetron can effectively be incorporated to result in improved outcomes in children with acute gastroenteritis.
Guidelines that have not supported the routine use of ondansetron cite concerns about a possible increase in the number of diarrheal stools (9,10). Although the reported increases have been statistically significant (4,8), they have not usually been clinically significant, with clinical outcomes such as length of stay and return visits favoring ondansetron administration. In our ED, all of the relevant secondary outcomes demonstrated improvement over time, thereby further enhancing the likelihood that a beneficial effect is directly derived from ondansetron administration. Although a formal cost-effectiveness analysis was not conducted, recent publications have demonstrated that the appropriate use of ondansetron has the potential to result in an annual savings of >$65 million in the United States (24).
Additional concerns have focused on the possibility that the use of a pharmacologic agent would shift the focus away from appropriate fluid, electrolyte, and nutritional therapy; that ondansetron use will result in adverse events; and that it may not be cost-effective (1). A shift in focus logically would either result in an increase in intravenous rehydration at the index visit or inadequate oral rehydration following discharge resulting in an increase in future health care provider visits; however, we found a reduction in intravenous rehydration and revisit rates. Although we do not have data related to revisits at other institutions, we have reported that because our institution has the only pediatric ED in the region, few caregivers bring their children elsewhere for care following an initial visit to our ED (25). Sturm et al (14), using logistic regression, found that ondansetron administration, although associated with a reduction in admissions (odds ratio 0.47; 95% CI 0.42–0.53), was also associated with increased revisits (odds ratio 1.74; 95% CI 1.27–1.65). These contrasting findings likely reflect the different study methodologies using and the challenge in controlling for all of the factors that may influence outcomes in the design used by Sturm et al (14). Although the present study cannot determine the frequency of adverse events following ondansetron administration, and there are reports in the literature (26), overall, ondansetron has been shown to be remarkably safe, with few severe adverse events reported and a favorable safety profile (27).
The most significant potential limitation to our data is that other changes may have occurred during the study period; however, during the 5-year time period, there were no major changes at our institution (eg, ED layout, nursing or physician staffing, protocols, pathways, guidelines), nor were there major advances in the management of acute gastroenteritis aside from evidence supporting the use of ondansetron. Furthermore, in a similar cohort of children (tertiary care Canadian pediatric ED) during the same period (2004–2007), only a 2.6% absolute reduction in the rate of intravenous rehydration was reported (28). This occurred despite the implementation of a formal oral rehydration clinical pathway. Moreover, ondansetron use at this institution remained low even in the final year of the study (3.5% in 2007).
Rotavirus vaccination status theoretically may be a confounding variable in our model; however, it is not covered at present by provincial health plans. In winter 2009, only 2% of caregivers reported that their child had received the rotavirus vaccine (S.B.F., unpublished data, 2010). Thus, rotavirus vaccine is unlikely to explain the change in outcomes that we reported and the use of an interrupted time-series model is justified to conduct our analysis.
Although a case-control design is an alternative method of assessing the clinical effect of ondansetron administration, the potential effect of bias was believed to outweigh the potential advantages of this design. Our concern focused on sources of bias, particularly in the identification of controls with similar baseline likelihoods of experiencing the outcomes of interest (29). Given that the outcome of intravenous rehydration is subjective and heavily influenced by patient (ie, clinical, physiological, social), physician (ie, inter- and intraphysician), and environmental (ie, time of day, ED volume, and wait times) factors, we could not satisfactorily match cases with controls. By the very nature of the decision to administer ondansetron, such patients have a priori been identified at high risk for needing intravenous rehydration.
An additional limitation was the switch from ICD-9 to ICD-10 coding during the final year of our study. Studies of the comparability between revisions have found that for some diagnoses, there are substantial changes because of the coding (30). Thus, we saw a significant increase in the frequency of the coding of diagnoses that corresponded to our diagnostic list when our institution switched to ICD-10 coding. It is possible that in the earlier years, potentially eligible subjects were not included in the analysis, whereas in the final year, some ineligible children were actually included. Our data indicate, however, that the subjects included in our dataset had similar baseline characteristics over time (Table 1).
In conclusion, this analysis provides evidence from a large cohort at a single institution that the use of ondansetron is associated with clinically significant reductions in the use of intravenous rehydration, length of stay, and need for revisits. The present study highlights that in addition to evidence of efficacy, there exists evidence of effectiveness regarding the administration of ondansetron to children with gastroenteritis. The incorporation of ondansetron into the treatment of pediatric gastroenteritis appears to result in improved clinical outcomes.
1. King CK, Glass R, Bresee JS, et al. Managing acute gastroenteritis among children: oral rehydration, maintenance, and nutritional therapy. MMWR Recomm Rep 2003; 52:1–16.
2. Malek MA, Curns AT, Holman RC, et al. Diarrhea- and rotavirus-associated hospitalizations among children less than 5 years of age: United States, 1997 and 2000. Pediatrics 2006; 117:1887–1892.
3. DeCamp LR, Byerley JS, Doshi N, et al. Use of antiemetic agents in acute gastroenteritis: a systematic review and meta-analysis. Arch Pediatr Adolesc Med 2008; 162:858–865.
4. Freedman SB, Adler M, Seshadri R, et al. Oral ondansetron for gastroenteritis in a pediatric emergency department. N Engl J Med 2006; 354:1698–1705.
5. Stork CM, Brown KM, Reilly TH, et al. Emergency department treatment of viral gastritis using intravenous ondansetron or dexamethasone in children. Acad Emerg Med 2006; 13:1027–1033.
6. Roslund G, Hepps TS, McQuillen KK. The role of oral ondansetron in children with vomiting as a result of acute gastritis/gastroenteritis who have failed oral rehydration therapy: a randomized controlled trial. Ann Emerg Med 2008; 52:22e6–29e6.
7. Reeves JJ, Shannon MW, Fleisher GR. Ondansetron decreases vomiting associated with acute gastroenteritis: a randomized, controlled trial. Pediatrics 2002; 109:e62.
8. Ramsook C, Sahagun-Carreon I, Kozinetz CA, et al. A randomized clinical trial comparing oral ondansetron with placebo in children with vomiting from acute gastroenteritis. Ann Emerg Med 2002; 39:397–403.
9. Guarino A, Albano F, Ashkenazi S, et al. European Society for Paediatric Gastroenterology, Hepatology, and Nutrition/European Society for Paediatric Infectious Diseases evidence-based guidelines for the management of acute gastroenteritis in children in Europe. J Pediatr Gastroenterol Nutr 2008; 46 (suppl 2):S81–122.
10. National Collaborating Centre for Women's and Children's Health. Diarrhoea and Vomiting Caused by Acute Gastroenteritis: Diagnosis, Assessment and Management in Children Younger Than 5 Years. London: National Institute for Health and Clinical Excellence; 2009.
11. Pfeil N, Uhlig U, Kostev K, et al. Antiemetic medications in children with presumed infectious gastroenteritis—pharmacoepidemiology in Europe and northern America. J Pediatr 2008; 153:659–662.
12. Hampton T. Promethazine warning. JAMA 2005; 293:921.
13. Freedman SB, Sivabalasundaram V, Bohn V, et al. The treatment of pediatric gastroenteritis: a comparative analysis of pediatric emergency physicians’ practice patterns. Acad Emerg Med 2011:18;38–45.
14. Sturm JJ, Hirsh DA, Schweickert A, et al. Ondansetron use in the pediatric emergency department and effects on hospitalization and return rates: are we masking alternative diagnoses? Ann Emerg Med 2010; 55:415–422.
15. Freedman SB, Powell EC, Nava-Ocampo AA, et al. Ondansetron dosing in pediatric gastroenteritis: a prospective cohort, dose-response study. Paediatr Drugs 2010; 12:405–410.
16. Warren DW, Jarvis A, LeBlanc L, et al. Revisions to the Canadian Triage and Acuity Scale paediatric guidelines (PaedCTAS). Can J Emerg Med 2008; 10:224–243.
17. Alpern ER, Henretig FM. Fever. In: Fleisher GR, Ludwig S, eds. Textbook of Pediatric Emergency Medicine, 6th ed. Philadelphia: Lippincott Williams & Wilkins; 2010:266–75.
18. Schnadower D, Kuppermann N, Macias CG, et al. Febrile infants with urinary tract infections at very low risk for adverse events and bacteremia. Pediatrics 2010; 126:1074–1083.
19. Wagner AK, Soumerai SB, Zhang F, et al. Segmented regression analysis of interrupted time series studies in medication use research. J Clin Pharm Ther 2002; 27:299–309.
20. Chatfield C. The Analysis of Time Series. An Introduction, 4th ed. London, UK: Chapman & Hall; 1989.
21. Altman D. Practical Statistics for Medical Research. Boca Raton, FL: Chapman & Hall; 1999.
22. McDowall D, McGleary R, Meidinger E, et al. Interrupted Time Series Analysis. Beverly Hills, CA: Sage Publications; 1980.
23. Cook TD, Campbell DT. Quasi Experimentation: Design and Analysis Issues for Field Settings. Boston: Houghton Mifflin; 1979.
24. Freedman SB, Steiner MJ, Chan KJ. Oral ondansetron administration in emergency departments to children with gastroenteritis: an economic analysis. PLoS Med 2010; 7:e1000350.
25. Freedman SB, Thakkar VA. Easing the strain on a pediatric tertiary care center: use of a redistribution system. Arch Pediatr Adolesc Med 2007; 161:870–876.
26. Gener B, Burns JM, Griffin S, et al. Administration of ondansetron is associated with lethal outcome. Pediatrics 2010; 125:e1514–e1517.
27. George M, Al-Duaij N, O’Donnell KA, et al. Obtundation and seizure following ondansetron overdose in an infant. Clin Toxicol (Phila) 2008; 46:1064–1066.
28. Doan Q, Chan M, Leung V, et al. The impact of an oral rehydration clinical pathway in a pediatric emergency department. Paediatr Child Health 2010; 15:503–507.
29. Schlesselman JJ, Stolley PD. Sources of bias. In: Schlesselman JJ, ed. Case-Control Studies: Design, Conduct, Analysis. New York: Oxford University Press; 1982:124–43.
30. Anderson RN, Minino AM, Hoyert DL, et al. Comparability of cause of death between ICD-9 and ICD-10: preliminary estimates. Natl Vital Stat Rep 2001; 49:1–32.