Our written questionnaire study found that 5-HT3-antagonists are the most common choice for treatment of established PONV when no prophylaxis is used. This pattern holds true following PONV prophylaxis with a regimen including a 5 HT3 antagonist regardless of the number of prophylactic antiemetics received by the patient
Overall, anesthesiologists reported administering a total of eighteen different drugs for PONV treatment and twelve different non-pharmacologic interventions. Variations in medical practice such that physicians treat similar patients differently may be created by uncertainty about efficacy of interventions, or formulary restrictions, or recent visits by the pharmaceutical sales representative, as well as differences in practitioners' residency training, judgment, and beliefs about drug acquisition costs and side-effect profiles [17–19]. Enhanced education and individualized feedback can change anesthesiologists' practice patterns [20, 21].
According to responses to a specific question on the questionnaire, almost all anesthesiologists (96%) preferred pharmacologic interventions for treatment, instead of non-pharmacologic (e.g., hydration, oxygen, acupuncture). The four anti-emetics chosen to be included for prophylaxis in the hypothetical patient vignettes – a 5-HT3 antagonist, droperidol, dexamethasone, and metoclopramide – were intended to represent major receptor systems involved in the etiology of PONV, as well as agents commonly used in clinical practice. We did not specify the doses of the four chosen antiemetics because we were mainly interested in the choice not the dose. To keep survey length reasonable, we opted not have a vignette with promethazine as a prophylaxis agent because in our outpatient practice promethazine is infrequently given for prevention.
Two-thirds of the anesthesiologists reported they would administer a 5-HT3 antagonist as first choice for PONV treatment if no prophylaxis had been given. The efficacy of the 5-HT3 antagonists may be more pronounced when a patient is vomiting than as treatment for nausea. There is weak evidence of dose-responsiveness with these drugs [22, 23]. Therefore, small doses of the 5-HT3 antagonists (ondansetron 1 mg) have been recommended for treatment. Interestingly, less than 15% of anesthesiologists reported using a combination of several agents for treatment, despite that combination of agents, or multi-modal therapy, may be increasingly being used for prophylaxis.
A majority of anesthesiologists reported they changed to a different agent for PONV treatment than the one(s) used for prophylaxis. However, 26% of practitioners would administer a second dose of the 5HT-3 antagonist (ondansetron (22%) + dolasetron (3%) + granisetron (1%)) if initial 5HT-3 antagonist prophylaxis failed. This is despite consensus guidelines, mostly derived from expert opinion not clinical trials, which suggest that if PONV occurs within six hours postoperatively, patients should not receive a repeat dose of the prophylactic antiemetic. Prescribing information for ondansetron states that a second dose does not provide additional control if the first prophylactic dose has failed. A drug from a different class should be used for treatment .
Pharmacogenomics may affect the success of a 5-HT3 antagonist because some patients have extra copies of the CYP2D6 gene, a genotype consistent with ultrarapid metabolism . A separate study of patients who failed prophylaxis with ondansetron found the complete response rate was significantly higher after treatment with promethazine (78%) than after treatment with repeat ondansetron (46%) . A third study of 428 patients (of 2,199 prophylactically treated with ondansetron) with PONV in the PACU, found that an additional dose of ondansetron was no better than placebo for reducing PONV two hours postoperatively .
Interestingly, anesthesiologists in our survey study were less likely to redose metoclopramide, dexamethasone, or droperidol for treatment (than a 5-HT3 antagonist) if any of those agents were administered for prophylaxis.
One quarter of anesthesiologists reported not having preprinted PACU orders specifically for PONV. This may increase the variability in PONV clinical practice, and make it difficult for evidence-based care to be implemented. Better mechanisms for delivering decision clinical support (e.g., evidence based guidelines) for PONV in the PACU may be possible. Four % of our sample voluntarily indicated that their group had developed their own PONV treatment guidelines.
For "older generation" antiemetics there are few data on therapeutic efficacy for established PONV. As an example, in patients who failed prophylaxis with droperidol, the complete response rate was significantly higher after treatment with promethazine (77%) than after droperidol (56%) .
It may be that anesthesiologists believe that interventions shown to be effective for prevention will be similarly effective for treatment. For example, many of our responders indicated they would use supplemental oxygen to treat PONV, but most studies of oxygen have been for PONV prevention, with varying efficacy [28, 29]. Other non-pharmacologic treatments suggested by our respondents such as IV fluid therapy, isopropyl alcohol inhalation and acupuncture/acustimulation have been studied, sometimes for prophylaxis not treatment, while others such as forced air warming have not [30, 31].
Beyond six hours, PONV can be treated a second time with any of the agents used for prophylaxis except dexamethasone and scopolamine, which are longer acting. We found that 73% of anesthesiologists reported having preprinted PACU orders for PONV at their primary practice location such that the anesthesiologist can amend the orders via checkbox, or by writing in.
To keep the questionnaire a reasonable length, we did not ask respondents why they chose the different treatments. The next study will assess if choices are based on such items as department policy, cost considerations, perceived lack of evidence or insufficient knowledge on the part of the anesthesiologist, individual patient's condition, or nursing determination.
PONV treatment research requires more precise PONV assessment
The lack of consistent assessment of PONV is an issue because studies often define endpoints differently. Nausea sometimes is defined by patient self-report, and other times as an observer asking the patient a yes/no answer. Some institutions define PONV as when actual treatment of PONV occurs, which is easily quantified, but is confounded because patients' perceptions of nausea severe enough to require intervention varies among patients, and nurses have different thresholds for initiating treatment . Often nausea and vomiting are not distinguished and the symptoms of PONV are combined into a single PONV endpoint . The challenge of multiple end-points and heterogeneity of definitions need to be addressed before aiming to establish the optimal management of PONV once it occurs in the PACU. The entire observation period should cover 24 hrs . Treatment responses we obtained might have varied between a patient developing nausea alone or having vomiting.
To control for the potential impact of biases from differing case-mix, we employed a postal questionnaire vignette methodology. The limitations of this method include that the subject sample depended on anesthesiologists' willingness to participate . While not significantly different from other national surveys of professional organizations, the response rate of 38 % is low and non-response bias may exist. This bias reflects the fraction of eligible subjects that do not respond and the difference in their answers compared to responders. Since it is unknown whether the physicians answering the questionnaire were systematically different from non-responders, there is no absolutely acceptable level of response.
The study had relatively small sample size. Determination of adequate sample size may be difficult and depends on the desired precision of the results. A larger number of respondents is always possible (to enable subgroup analyses about differences among practice types, academic vs. private practice, for example) but we obtained a reasonable sampling of current practice patterns to help design larger studies of PONV treatment.
Also, our result that a 5-HT3 antagonist is the most commonly prescribed for PONV treatment may not be applicable in other countries.
Although vignettes are suitable for comparative analyses because they control for case-mix, further studies are needed to confirm that the results from vignette-based questionnaires are in fact a valid measure of the real-life clinical care provided by anesthesiologists. The open-ended comments section in our questionnaire did not uncover any problems with anesthesiologists stating they didn't understand the questionnaire, or that key elements were missing. Since our vignettes were hypothetical, the answers provided by the anesthesiologists may not be what they actually use.