Apfel Postoperative Nausea and Vomiting Risk Score
Apfel Postoperative Nausea and Vomiting Risk Score
Four binary factors give a graded predicted incidence of nausea or vomiting in the first 24 hours after general anaesthesia. The score estimates the baseline risk against which prophylaxis is considered; it is not a measure of how much prophylaxis to give.
Apfel simplified risk score
4 factors → 0–4Woman; non-smoker; no history of postoperative nausea or motion sickness; postoperative opioids expected
Scoring
0 → about 10% · 1 → 21% · 2 → 39% · 3 → 61% · 4 → 79%
- what the number is
- a predicted incidence of nausea or vomiting in the 24 hours after surgery, read off a derivation cohort. It is not a count of antiemetics, a dose, or a threshold for anything. The score establishes the baseline risk; what is done about that baseline is a clinical decision this page does not make
- the item that runs backwards
- not smoking scores the point. Reading the smoking item in the same direction as the other three gives a smoker a point they should not have and takes one from the non-smoker who should, which moves most patients by two strata
- discrimination
- modest, and the article is honest about it. Scores derived in Oulu gave an area under the ROC curve of 0.69 in Oulu and 0.69 in Würzburg; scores derived in Würzburg gave 0.75 at home and 0.65 in Oulu. The final four-factor score reached about 0.75 in the combined set. A score of around 0.7 means the ranking is useful and individual prediction is not
- the cohort, and how it differs from now
- 520 patients in Oulu, Finland and 2,202 in Würzburg, Germany, cross-validated each way, with the published score derived on a combined 1,040. Volatile anaesthesia was the norm, nitrous oxide was common and routine multimodal prophylaxis was not. A contemporary unit using total intravenous anaesthesia will see lower absolute incidences than these strata predict, while the gradient between strata has held up
- what it leaves out
- age, which falls with risk; the type and duration of surgery; volatile agent and nitrous oxide exposure; and the anaesthetic technique itself. Several of those are stronger determinants than some of the four items. The score is short on purpose, and shortness was bought with accuracy
Worked example
Woman; non-smoker; no history of postoperative nausea or motion sickness; postoperative opioids expected
Female 1 + non-smoker 1 + history 0 + opioids 1 = 3 factors
Three factors predicted nausea or vomiting within 24 hours in about 61 per cent of Apfel's combined cohort
Note that this is an ordinary patient. A non-smoking woman having an operation that will need opioids afterwards is most of a general surgical list, and she scores three of four
If the smoking item were read the way the other three are read, she would score 2 and the predicted incidence would be 39 per cent. That is the commonest error on this score
Avoiding postoperative opioids would take her to 2, 39 per cent. That is the only one of the four items a plan can change, and the score neither recommends nor discourages changing it
61 per cent is a frequency in a cohort of European surgical patients in the 1990s. It is not the chance that this woman will be sick
The four factors
| Factor | Scores 1 when | Note |
|---|---|---|
| Sex | Female | The strongest factor in the original regression |
| Smoking | The patient does NOT smoke | The item runs backwards. Smoking is associated with less postoperative nausea |
| History | Previous postoperative nausea or vomiting, OR motion sickness | Either one scores the point; both together still score one |
| Opioids | Postoperative opioids are expected | The only item that is a plan rather than a fact |
Predicted incidence by number of factors
| Factors | Predicted PONV within 24 hours | Share of a typical list |
|---|---|---|
| 0 | about 10% | Few — it needs a smoking man with no history and no opioids |
| 1 | about 21% | Common |
| 2 | about 39% | Common |
| 3 | about 61% | Common — most non-smoking women having opioid-requiring surgery |
| 4 | about 79% (the Discussion of the same paper gives 78%) | A minority |
A baseline risk, not a prescription
Apfel’s score exists because the risk models of the early 1990s were too long to be used. It has four binary items — female sex, not smoking, a history of postoperative nausea or of motion sickness, and the expectation of postoperative opioids — each worth one point, and it gives a graded predicted incidence of nausea or vomiting in the first 24 hours: roughly 10, 21, 39, 61 and 79 per cent for 0 to 4 factors. The article’s own Discussion gives 78 rather than 79 for four factors, which is a small internal inconsistency worth knowing about if you are checking a number against the source.
What the output is gets misread constantly. The score estimates the baseline risk of a patient who receives no prophylaxis. It is not a count of antiemetics, not a dose, and not a threshold at which something happens. A reader who treats the number of risk factors as the number of drugs to give has turned a prediction into a protocol, and the paper that derived the score says nothing of the kind. Published guidance does use baseline risk to frame the prophylaxis question, but the framing and the arithmetic are separate things, and this page does the arithmetic.
The smoking item is the other reliable source of error. It runs backwards: not smoking scores the point. Smoking is reproducibly associated with less postoperative nausea, and reading the item in the same direction as the other three both adds a point the smoker should not have and removes one the non-smoker should — a two-stratum error in opposite directions on two different patients.
The score’s discrimination is modest and the paper is candid about it: areas under the ROC curve of 0.65 to 0.75 depending on which centre derived and which validated, with the final score near 0.75. That is enough to rank a list and not enough to predict a person. The derivation cohort was two European hospitals in the 1990s, when volatile anaesthesia and nitrous oxide were routine and multimodal prophylaxis was not, so a contemporary unit using total intravenous anaesthesia will see lower absolute incidences than these strata suggest while the gradient between them holds. The score also leaves out age, the surgery, the duration and the anaesthetic technique, several of which matter more than one or two of the items it keeps. A score’s output is a cohort frequency, not this patient’s probability: a stratum in which 9 per cent had an event describes that stratum, not which 9 per cent. This page reports what a stratum predicted in a named study. It recommends no action.
Frequently asked questions
What are the four Apfel risk factors?
Female sex, being a non-smoker, a history of postoperative nausea and vomiting or of motion sickness, and the expectation of postoperative opioids. Each scores one point and all four weigh the same, so the score counts how many apply rather than which.
Why does not smoking score a point?
Because smoking is reproducibly associated with less postoperative nausea and vomiting, probably through hepatic enzyme induction and altered emetic sensitivity, so being a non-smoker predicts more. The item runs in the opposite direction to the other three and misreading it is the commonest error on this score. It is a prediction, not an argument for anything.
Does the Apfel score say how many antiemetics to give?
No. It gives a predicted baseline incidence of nausea or vomiting in a patient receiving no prophylaxis. Treating the number of risk factors as the number of drugs is a widespread misreading with no basis in the paper that derived the score. What is done about a given baseline risk is a clinical decision.
How accurate is the Apfel score?
Modestly. The original cross-validation gave areas under the ROC curve of 0.69 in both centres for the Oulu-derived score, and 0.75 and 0.65 for the Würzburg-derived one, with the final four-factor score around 0.75 in the combined set. That ranks patients usefully and predicts individuals poorly.
Do the published incidences still apply?
The gradient does; the absolute numbers are from 1990s European practice with volatile anaesthesia, common nitrous oxide and little routine prophylaxis. A unit using total intravenous anaesthesia and multimodal prophylaxis will see lower incidences at every stratum than the derivation cohort did.
Related calculators
References
- Apfel CC, Läärä E, Koivuranta M, Greim CA, Roewer N. A simplified risk score for predicting postoperative nausea and vomiting: conclusions from cross-validations between two centers. Anesthesiology. 1999;91(3):693–700.
- Halvorsen S, Mehilli J, Cassese S, et al. 2022 ESC Guidelines on cardiovascular assessment and management of patients undergoing non-cardiac surgery. Eur Heart J. 2022;43(39):3826–3924.
- American Society of Anesthesiologists Committee on Standards and Practice Parameters. Practice guidelines for preoperative fasting and the use of pharmacologic agents to reduce the risk of pulmonary aspiration. Anesthesiology. 2017;126(3):376–393.
Not medical advice. For healthcare professionals and education. Reference intervals vary by laboratory and assay — always use your own laboratory's. Never base a dose or a treatment decision on this page alone. Full disclaimer at calcengines.com/disclaimer/