IMplementation of a Prediction rule in Anesthesia practice to improve...


Update content

Title IMplementation of a Prediction rule in Anesthesia practice to improve Cost-effectiveness of Treatment of postoperative nausea and vomiting: the IMPACT trial
Period 12 / 2005 - unknown
Status Completed
Research number OND1317996
Data Supplier ZONMW


Background and objectives. So-called prediction rules (risk scores) have become increasingly popular in all medical disciplines. This will only rise with the introduction of electronic patient records as these will enhance their use. However, effects of implementation of such rules in daily care has hardly been studied. Also not in anesthesiology. We developed and validated an accurate rule to preoperatively predict the risk of postoperative nausea and vomiting (PONV) in surgical inpatients. PONV causes extreme patient discomfort and occurs in even 30%-50% of all surgical inpatients. As routine administration of PONV prophylaxis is not cost-effective, a risk-tailored approach using an accurate prediction rule is widely advocated. Before large-scale implementation, we aim to study whether such implementation indeed changes physician behavior and improves patient outcome. Given the increased interest in prediction rules, another aim is to study general causes of successful/poor implementation of prediction rules in health care. Design. Cluster, randomized study in which 60 anesthesiologists of the UMC Utrecht will be randomized to either the intervention or usual care group. Health care problem and implementation problem. For many adverse outcomes preventive strategies exist. Risk-benefit analyses taught us that they often should be limited to high risk patients - see e.g. the latest cholesterol consensus. This requires tools that accurately predict a patient?s risk of adverse outcome. Prediction rules or risk scores are algorithms to objectively estimate the risk of an outcome using the patient profile. With prediction rules, risk estimations and treatment indications become more objective and accurate than using clinical experience only.(1-4) Development of prediction rules has sharply increased in all medical disciplines, which will continue with the introduction of electronic patient records (EPR). To improve quality of care, prediction rules must not only predict accurately, practitioners must also use them. Hence, before large scale implementation a phased approach is recommended.(1-3,5-8) After development, a rule's predictive accuracy must first be validated in other populations. Then, implementation studies need to quantify whether risk estimations by the rule indeed change physician behavior in terms of treatment decisions. Finally, implementation studies are needed to quantify whether they improve patient outcome and cost-effectiveness. If so, widespread implementation is indicated. Unfortunately, many rules are being developed, yt whether their use truly changes physician attitudes, let alone patient outcome and cost-effectiveness of care, is rarely studied.(9) All this also applies to the prediction and management of post-operative nausea and vomiting (PONV). PONV seems a 'minor' complication, but results in extreme patient discomfort, may delay discharge from the postoperative anesthesiology care unit (PACU) and even hospital, and (if sustained)lead to serious complications.(10-14) PONV is also a big problem in terms of occurrence; worldwide it occurs in 30%-50% of all inpatient surgeries.(10,12-20) Pre-emptive PONV strategies exist, notably using intravenous propofol - rather than inhalation - anesthesia and anti-emetic prophylaxis. Where patients undergoing chemotherapy always receive prophylactic anti-emetics, in surgical patients they are underused.(10,12-20) Routine administration of anti-emetics is not cost-effective, but should be based on individual risks.(10,12-22) Two prediction rules to preoperatively estimate the risk of PONV have been developed(10,23), but showed poor accuracy in e.g. The Netherlands.(15) We thus developed and validated an improved prediction rule.(16) Following methodological guidelines, we will quantify the ability of our PONV rule to improve physicians' behavior, patient outcome and cost-effectiveness of current PONV management. Overall aim To quantify the clinical value of implementing in daily care a tool that facilitates timely prediction of the risk of postoperative nausea and vomiting (PONV) allowing for tailoring of preemptive management. Contribution of project to resolution health care problem. Although their popularity is sharply increasing, prediction rules will only improve quality of care if doctors are able and willing to use them in their day-to-day decision making. These latter two implementation aspects of prediction rules have scarcely been studied, in anaesthesiology nor in general. This study will therefore provide a template for implementation of prediction rules in general, and may serve all disciplines where clinical decision making guided by prediction rules is undertaken. More specifically, our results will improve the cost-effectiveness of current PONV management which currently shows too much variation resulting in too high PONV incidences worldwide. Additionally, similar studies could be undertaken to improve,for example, pre-emptive postoperative pain management for which we also have developed and validated an accurate prediction rule.(30) The same applies to various other adverse outcomes that perhaps occur less frequently, but surely have a much higher impact on quality of life and health care costs, e.g. myocardial infarction after non-cardiac surgery, poor outcome after intensive care stay, and cognitive decline after coronary bypass surgery, for which also prediction rules exist. In conclusion, for each adverse outcome for which prediction rules are/will be developed to improve risk-tailored care, the proposed study can become an example study. Experience with implemented tool/focus of implementation. The investigators have extensive experience with development and validation of prediction rules, both in the discipline of anesthesiology as well as in other medical disciplines such as general practice and pediatrics (see publications) The rule to predict the risk of PONV, the tool to be implemented in the proposed study, is the latest one. In brief, due to the high incidence of PONV even though pre-emptive PONV strategies exist, we have studied for years how to improve the management of PONV and whether an accurate prediction rule could be developed and implemented to enhance physicians decisions and thus patient outcome. We started with the validation and application of two previously developed PONV rules(10, 23) on a large sample (n=1500) of adult, non-emergency, non-ambulatory Dutch patients undergoing all types of surgery. We found, however, that both rules showed poor accuracy in our patients.(15) Prompted by these results, we aimed to update and improve the existing rules to predict the occurrence of PONV and developed and validated an enhanced rule.(16) To do so, all potential risk factors of PONV and the occurrence of PONV until 24 hours post-surgery were systematically documented in 1500 adult, non-emergency, non-cardiac, non-intracranial and non-ambulatory surgical patients in the AMC Amsterdam. Using multivariable modeling and validation techniques we developed a simple rule enabling anesthesiologists to timely -i.e. preoperatively - discriminate between high and low risk patients in order to enable preventive PONV management.

Related organisations

Related people

Researcher T.H. Kappen
Project leader Prof.dr. K.G.M. Moons


D21100 Bioinformatics, biomathematics
D23310 Surgery

Go to page top
Go back to contents
Go back to site navigation