The use of a nasogastric (NG) tube to prevent postoperative nausea and vomiting (PONV) has long been suggested in the literature. Postulated mechanisms for an effect have included decompressing the stomach and decreasing acidity. Given that the experience of the person ventilating the lungs with a face mask has been described as influencing PONV1 and that use of histamine-antagonists can reduce PONV,2 the routine use of a NG tube to prevent PONV appears plausible. The effect of a gastric tube reported in the literature is very heterogeneous,3 but individual studies may be underpowered to detect a small but still clinically relevant difference.
Using a dataset of more than 1000 patients, we tested the hypothesis that routine intraoperative or perioperative use of a NG tube would not affect the incidence of PONV. The primary endpoint in this analysis was incidence of PONV during the first 24 h postoperatively.
Our comparative study used data from the previously published International Multicenter Protocol to Assess the Single and Combined Benefits of Antiemetic Strategies in a Controlled Clinical Trial of Factorial Design (IMPACT)4 (Appendix).
In the IMPACT trial, patients were randomized in double-blind fashion and assigned to several antiemetic strategies. The insertion of the NG tube was not randomized and left to the discretion of the anesthesiologist. In patients with an intraoperative NG tube, the tube was placed after intubation, suctioned, capped, and removed under suction immediately before extubation, whereas in patients with perioperative use, it was left in place, suctioned, and capped, with intermittent suctioning for more than 24 h after surgery.
In the postanesthesia care unit, the time, severity, and characteristics of PONV were recorded on standardized forms. PONV was defined as the occurrence of nausea (using a severity score 0–10), vomiting/retching, or both during the first 24 h after surgery.
Associations between NG tube use and three 24-h outcomes (nausea, emesis, and overall PONV) were assessed using propensity score analysis. Any baseline variable even remotely predictive of NG tube use, defined as P < 0.30, was included in the calculation of the propensity scores, including such factors as experience of the anesthesiologist, patient age, PONV risk score, location, surgery type, surgical approach (open versus laparoscopic), anesthetic regimen, and clinical center.
Our analysis for each exposure (NG tube use versus nonuse) thus consisted of two stages. In the first stage, all available baseline factors were used in a model to predict NG tube use (yes/no), from which each patient was assigned a predicted probability of having received a NG tube. Each patient who actually did receive a NG tube was then matched on that probability to a nonuse patient using the greedy matching algorithm5 with a matching criterion of 0.05 propensity score units.
In the second stage, we compared the matched NG tube groups (yes/no) on the outcome(s) of interest, PONV, using logistic regression analyses. Multivariable models included any covariates significant at the 0.05 level, further adjusting for any remaining imbalance on available potential confounders. Note that our multivariable analysis is based on the propensity-matched patients only and is quite distinct from a traditional multivariable model using all patients in the dataset, regardless of distribution of baseline variables. The significance level for the two-tailed χ2 test was 0.05.
For each analysis, we performed the usual two-tailed test for superiority of one treatment versus the other. We also performed a nonsuperiority analysis in which we tested the null hypothesis that NG tube use is beneficial. We defined “beneficial” as a reduction in the odds of having the outcome by at least 5% with NG use, corresponding to an odds ratio (OR) of 0.95 or lower. The alternative hypothesis in this one-tailed test was that the OR is ≥0.95, i.e., that NG tube use is either worse than nontube use (OR >1) or that it reduces the odds of the outcome no more than 5% (OR ≥0.95). A significant test result would thus be interpreted as NG tube use being not superior to nonuse (i.e., either equivalent or worse). The significance level for each hypothesis was 0.05. No adjustment was made for assessing the three primary outcomes. SAS statistical software (Cary, NC, version 9.1) was used for all analyses.
A total of 4055 patients were initially considered for analysis: 2743 patients did not receive a NG tube, 1185 received a NG tube intraoperatively, and 127 received one intra- and postoperatively for 24 h. This initial grouping demonstrated imbalance on important baseline predictors of morbidity. Propensity scores were then used to compile a subgroup of matched NG tube use and control patients for intraoperative and 24-h postoperative use. Balance was achieved for all variables used in the propensity score matching and, innate to the methodology, also for variables that influence the risk for PONV (Tables 1 and 2).
Results comparing propensity-matched intraoperative NG tube use versus controls are shown in Figure 1 and with more detail in Table 3. Intraoperative use of the NG tube use was not associated with a reduction in nausea (multivariable OR of 1.23, P = 0.14), vomiting (0.92, P = 0.64), or PONV (1.22, P = 0.16). The 24-h PONV incidence was 44.4% in patients with an intraoperative NG tube use versus 41.5% in controls, for a difference of 2.9% (95% CI −3.2%, 9.1%).
Perioperative NG tube use propensity score results are displayed in Table 4. There was no evidence of an association between perioperative NG tube use and reduction in nausea (0.85, P = 0.65), emesis (0.90, P = 0.83), or overall PONV (0.84, P = 0.64). The 24-h PONV incidence was 27.8% in patients with perioperative NG tube use versus 31.3% in controls, for a difference of −2.4% (95% CI −16.1%, 11.1%).
In our nonsuperiority analyses, we rejected the null hypotheses that intraoperative NG tube use was more beneficial (i.e., superior) compared with non-NG tube use for two of the three outcomes of interest, PONV (multivariable P = 0.033) and nausea (multivariable P = 0.037), assuming that an OR between 0.95 and 1.0 represents equivalence of the two methods of care (Table 3). From these one-tailed results, we infer that the adjusted OR for perioperative NG tube use is ≥0.95 for PONV and nausea. Nonsuperiority was not demonstrated for perioperative NG tube use (Table 4).
This analysis of a large case-matched dataset with more than 1000 patients evaluating the effect of a NG tube on PONV shows no evidence of a reduction in incidence of PONV. This result seems surprising given that mechanistically every effort that reduces intragastric volume should decrease the incidence of vomiting.
A meta-analysis performed by Cheatham et al.6 identified 26 trials with 3964 patients and found no difference in the incidence of postoperative nausea but did find a decreased risk of vomiting. However, retching, which might occur instead of vomiting in the setting of an emptied stomach, was not separately accounted for in all included studies. Additionally, the effect of a gastric tube reported in the literature is so heterogeneous that no reasonable point estimates could be calculated in a Cochrane review by Nelson et al.3
Our analysis includes a significantly larger sample size than any other previous randomized controlled trial and should thus be able to detect even small effects present. The main limitation of this analysis is that the original study was not randomized for the use of a gastric tube; however, to address this drawback patients were matched using a propensity score to yield groups balanced on potential baseline confounders.7,8
In conclusion, these results provide strong evidence that the routine use of a NG tube during surgical procedures does not reduce PONV.
The authors thank Dr. Anuj Malhotra for his careful editorial assistance.
The IMPACT investigators are as follows:
- Steering Committee—C. C. Apfel, A. Biedler, and K. Korttila.
- Data Management and Monitoring—C. C. Apfel, E. Kaufmann, M. Kredel, A. Schmelzer, and J. Wermelt.
- and Data Analyses—C. C. Apfel, K.-H. Kerger, and E. Mascha.
Site Investigators—C. C. Apfel, S. Alahuhta, F. Bach, A. Bacher, H. Bartsch, H. Bause, A. Biedler, B. Book, H. Bordon, D. Buschmann, K. Danner, O. Danzeisen, D. Detzel, L. H. J. Eberhart, H. Feierfeil, H. Forst, C. Frenkel, G. Frings, B. Fritz, G. Fritz, A. Goebel, M. Hergert, C. Heringhaus, M. Hinojosa, C. Hoehne, W. Hoeltermann, H.-B. Hopf, C. Isselhorst, R. M. Jokela, E. Kaufmann, H. Kerger, T. Kangas-Saarela, P. Karjaleinen, A. Kimmich, M. Koivuranta, K. Korttila, U. Koschel, P. Kranke, M. Kredel, M. Lange, F. Liebenow, W. Leidinger, M. Lucas, C. Madler, J. N. Meierhofer, F. Mertzlufft, J. Motsch, S. Muñoz, E. Palencikova, A. Paura, S. Pohl, C. Prause, R. Rincon, N. Roewer, U. Ruppert, A. Schmelzer, I. E. Schneider, R. Sneyd, Schramm, A. Soikkeli, S. Spieth, B. Steinbrecher, K. Stoecklein, M. Trick, A. Turan, S. Trenkler, I. Vedder, P. Vila, J. Wermelt, K. Werthwein, W. Wilhelm, and C. Zernak.
1. Hovorka J, Korttila K, Erkola O. The experience of the person ventilating the lungs does influence postoperative nausea and vomiting. Acta Anaesthesiol Scand 1990;34:203–5
2. Doenicke AW, Hoernecke R, Celik I. Premedication with H1 and H2 blocking agents reduces the incidence of postoperative nausea and vomiting. Inflamm Res 2004;53(suppl 2):S154–S158
3. Nelson R, Edwards S, Tse B. Prophylactic nasogastric decompression after abdominal surgery. Cochrane Database Syst Rev 2005:CD004929
4. Apfel CC, Korttila K, Abdalla M, Biedler A, Kranke P, Pocock SJ, Roewer N. An international multicenter protocol to assess the single and combined benefits of antiemetic interventions in a controlled clinical trial of a 2 × 2 × 2 × 2 × 2 × 2 factorial design (IMPACT). Control Clin Trials 2003;24:736–51
5. Parsons LS. Reducing Bias in a Propensity Score Matched-Pair Sample Using Greedy Matching Techniques. In: Proceedings of the 26th annual SAS Users Group International (SUGI) Conference, Long Beach, California, 2001:214–26
6. Cheatham ML, Chapman WC, Key SP, Sawyers JL. A meta-analysis of selective versus routine nasogastric decompression after elective laparotomy. Ann Surg 1995;221:469–76; discussion 76–8
7. Cywinski JB, Koch CG, Krajewski LP, Smedira N, Li L, Starr NJ. Increased risk associated with combined carotid endarterectomy and coronary artery bypass graft surgery: a propensity-matched comparison with isolated coronary artery bypass graft surgery. J Cardiothorac Vasc Anesth 2006;20:796–802
8. Koch CG, Khandwala F, Nussmeier N, Blackstone EH. Gender and outcomes after coronary artery bypass grafting: a propensity-matched comparison. J Thorac Cardiovasc Surg 2003;126:2032–43