Simplified Algorithm for the Prevention of Post-operative Nausea and Vomiting: A Before and After Study

Dewinter G, Staelens W, Veef E et al.

British Journal of Anaesthesia 2018: 120 (1): 156-163.

Question

Would a simplified algorithm for PONV prophylaxis reduce the incidence of PONV in adult patients undergoing elective non-cardiac non-day-case surgery under general anaesthesia?

Study design

Quasi-experimental design

Single-centre, before and after study

2 x 5-working day audits (excl. Sat/Sun)

Initial audit using original algorithm Jan 12-18, 2016

Follow-up audit using simplified algorithm Nov 28 – Dec 2, 2016

University Hospitals Leuven, Belgium

Adult patients (Audit 1: 257 screened, 46 included, n=211. Audit 2: 231 screened, 30 excluded, n = 201) undergoing general anaesthesia

Inclusion criteria:

All adult patients >18 years old

Admitted to our post-anaesthesia care unit (PACU)

Undergoing elective, non-cardiac, non-day-case surgery, under general anaesthesia

Exclusion criteria:

Emergency procedures

Admitted to ICU/HDU

Patients expected to require overnight ventilation

Statistics

Fisher’s exact and Manne Whitney U tests were used to compare groups

Uni and multivariable logistic-regression models with a binomial logit link function were used for all binary outcomes

Poisson regression was used for count data

Primary outcome a p-value of 0.025

24 h PONV incidence, the multivariable logistic regression was risk adjusted

Logistic-regression analyses, the 10% and 20% Apfel risk strata were combined

 

Interventions

Initial algorithm

Simplified algorithm

Primary outcome

PONV incidence 1h and 24 h post-op

Secondary outcomes

• Incidence of PON & POV
• Accurate Apfel score
• Compliance with guidelines
• Anti-emetics administered
• Use of a rescue medication

Results

211 (1^st) v 201 (2^nd)

Primary outcome

Reduction in PONV over 24h (33% (69/211) vs 22% (45/201) P=0.02) RRR 33%, ARR 11%

Statistically significant when controlling for different confounders

Secondary outcomes

• No difference in PON & POV over 1h or POV over 24h
• Apfel score predicts PONV risk
  • But was poorly calculated (correct in 36% v 42%)
• Simplified algorithm improved compliance with guidelines from 18% to 46%
• Use of rescue medication reduced during 2^nd audit
  • 58 pts (102 anti-emetics doses) v 30 (61), P=0.002

Discussion

• Low adherence to PONV guidelines is a well-known problem
• Uncertain how best to improve compliance with guidelines
• Electronic prompts/reminders rely on accurate calculation of risk – which was demonstrably poorly done – and have been shown not to work
• Simplification is effective
• Some patients will be over-treated but risk of harm is low

Strengths

• Relevant question
• PONV an issue
• Non-compliance with guidance is recognised issue
• Sample reflective of KH
• Reflective of KH practice
  • Similar drugs, similar groups
• Well matched groups with bias minimised by lack of awareness of audit
• Significant difference in PONV rates over 24h

Limitations

Stated

• Short observation periods
• Study sample might not reflect the population incidence
• Clinical guidelines are introduced by means of a lecture given to the residents in their preparation year; the first audit occurred 5 months, and the second audit occurred 4 months, after their commencement.
• Poor calculation of PONV risk may have caused “non-compliance”
  • Correct calculation and knowledge/reminder of guidelines may have improved performance
• Unclear cause of the reduction of the PONV incidence
  • Introduction of the simplified algorithm
  • Improved compliance with algorithm
  • Increased number of anti-emetics given

Not-stated

• The beta-error (power) of the study is not specified
• No data on ethnicity – ?generalisable
• Majority of patients non-smokers and planned post-op opioid use
• No mention of N₂O
• Significant number given no antiemetic (35% in initial audit)
• No mention of cost analysis or SEs of anti-emetics
• Un-blinded study with no control group
  • Risk of Reporting validity – measurement method/accuracy changed between audits
  • Risk of Regression-to-the-mean – particularly good or bad weeks
  • Risk that audit itself improved awareness and improved performance
  • Risk that introduction of novel guidelines (irrespective of content) improved awareness and improved performance

Study conclusions

The introduction of a radically simplified and mainly sex-centred PONV algorithm for the prophylaxis of PONV resulted in a significant reduction of PONV incidence and in a better compliance with the institutional algorithm.

Summary by Dr A Feben. Journal Club Meeting 27 June 2019.