2.1 Study design, approvals and consent

The manuscript was prepared according to the Consensus-Based Checklist for Reporting of Survey Studies (CROSS)²³ (Appendix 1) and conducted via the secure web application Research Electronic Data Capture (REDCap, Vanderbilt, Nashville, TN, USA)²⁴ hosted by the Capital Region of Denmark. The study was registered in the controller's (the Capital Region of Denmark) record of processing activities (reference no: p-2023-15040, approved 5 December 2023). Ethical approval was unnecessary as no patient data were collected, and participants were anonymous. A protocol was uploaded to https://opiaid.dk/survey before the survey activation (9 January 2024). Informed consent from participants was obtained as the first item of the questionnaire.

2.2 Population

All anaesthesiologists, nurse anaesthetists and post-anaesthesia care unit (PACU) nurses who are involved in the perioperative management of adult patients within the Danish public hospital system.

2.3 Danish healthcare and organisation

The Danish healthcare system provides free healthcare services to approximately six million residents²⁵ distributed across five regions, which collectively house 47 public anaesthesia departments. According to self-reported departmental employee records for 2024, 1699 anaesthesiologists, 1196 nurse anaesthetists and 1292 PACU nurses are employed across the departments. In Denmark, nurse anaesthetists are responsible for maintenance and intraoperative analgesic administrations in collaboration with anaesthesiologists. See Appendix 2 for additional details regarding anaesthesia education in Denmark.

2.4 Survey design

This cross-sectional survey consisted of 37 questions sub-grouped into four sections: demographics, pain management guidelines, intra- and postoperative opioid management practices in different clinical scenarios (Appendix 3). Anaesthesiologists answered both intra- and postoperative questions. Nurse anaesthetists were only presented with questions regarding the intraoperative phase, and PACU nurses were only presented with questions regarding the postoperative phase. Respondents could exit the survey before completion, resulting in section-specific missing data. The survey was pilot-tested by anaesthesia personnel at three different hospitals from 29 September 2023 to 29 January 2024 (Appendix 2).

2.5 Survey distribution

The site participation rate was 98% (46 of 47 sites). At each site, a designated local representative was responsible for identifying eligible participants, distributing the survey via email with an online link and sending at least three reminder emails. Before survey activation, site investigators received a brief study protocol, a demo version of the survey and a project video presentation (https://opiaid.dk) from the coordinating site. Each site investigator also provided a list of the total number of invited participants from each profession, which was used to calculate the overall and profession-specific response rates. The survey was open for data collection from 5 February to 30 April 2024, with a minimum of three reminders sent to all participating sites before survey closure.

2.6 Statistical analysis

Categorical data were presented as numbers and percentages followed by nominator and denominator and continuous as either mean (95% confidence intervals [CI]) or median (IQR), depending on the distribution of data. Additional details are presented in the full statistical analysis report (Appendix 4). The anaesthesia personnel's specific choice of opioids and doses were presented along with conversion values into i.v. morphine equivalents (MEQ)/kg using the opioid conversion²⁶ presented in Appendix 2. Missingness was reported for each survey section. All analyses were conducted as complete case analyses without imputation. We used a convenience sample with no formal sample size estimation and did not use any measures to adjust for non-representativeness in the survey sample. Data were handled and analysed using R version 4.2.1 (R Core Team, Vienna, Austria).²⁷

3.1 Response rates and demographics

We invited 4187 anaesthesia personnel to participate, of whom 2850 accessed the survey link. Among these, 2025 answered at least one clinical question, defining them as the overall respondents. The overall response rate was 48%, with a median site response rate of 50% (IQR 43–63) (Appendix 2). The geographical distribution is detailed in Table 1.

When stratified by profession, the response rates were 56% (665/1196) for anaesthesiologists, 47% (797/1699) for nurse anaesthetists and 44% (563/1292) for PACU nurses. The respondents pool consisted of 33% (665/2025) anaesthesiologists, 39% (797/2025) nurse anaesthetists and 28% (563/2025) PACU nurses (Table 1). The median anaesthesia-related work experience was 12 years (IQR 5–20) for anaesthesiologists, 13 years (IQR 5–20) for nurse anaesthetists and 7 years (IQR 2.5–16) for PACU nurses (Appendix 4). The respondents most frequently provided anaesthesia and pain management to orthopaedic-, abdominal- and/or gynaecology-obstetric patients (Table 1).

Proportionate missingness for each of the 37 questions ranged between 0% and 22% (see Appendix 2 for further details).

3.2 Guidelines for intraoperative pain management

Intraoperative guidelines for pain management were reported to be available by 89% (1304/1460) of respondents. Guidelines were followed ‘often’ or ‘always’ by 84% (1108/1320) and mainly consisted of combined general and procedure-specific recommendations (Table 2). Guidelines were deemed to cover individual analgesic needs to an ‘acceptable’, ‘good’ or ‘really good’ degree by 65% (824/1260) of the respondents, with specific recommendations for chronic pain (38%), kidney insufficiency (15%), age (13%) and substance abuse (11%). Factors less commonly covered in the guidelines included body weight (8%), comorbidities in general (6%), cancer (4%), alcohol abuse (4%) and dementia (1%) (Appendix 4).

3.3 Guidelines for postoperative pain management

Postoperative guidelines for pain management were reported to be available by 88% (1024/1159) of respondents. Guidelines were followed ‘often’ or ‘always’ by 88% (910/1038) and mainly consisted of combined general and procedure-specific recommendations (Table 3). Guidelines were deemed to cover individual analgesic needs to an ‘acceptable’, ‘good’ or ‘really good’ degree by 74% (730/992) of respondents, with specific recommendations for chronic pain (42%), kidney insufficiency (30%), acute surgery (33%), age (24%), substance abuse (13%) and body weight (10%). Less common factors included in the guidelines were comorbidities in general (6%), alcohol abuse (5%), cancer (5%) and dementia (2%) (Appendix 4).

3.4 Intraoperative opioid administration strategies

The preferred intraoperative opioids for postoperative analgesia were fentanyl (44%; 643/1460) and morphine (36%; 519/1460) (Table 2). The most important factors in intraoperative opioid dose adjustment were deemed to be age (82%), chronic pain (69%), pre-operative opioid use (54%), type of surgery (52%) and body weight (51%) (Figure 1).

In the four intraoperative cases (variables: 30 or 65 years old; male or female; ASA score 1 or 3; anterior cruciate ligament- or laparoscopic cholecystectomy surgery), the median intraoperative opioid doses for postoperative analgesia ranged from 0.12 to 0.38 mg i.v. MEQ/kg (Figure 2). Overall, fentanyl and morphine were the most preferred opioids. Among those who preferred fentanyl, 72% (407/568) initiated the treatment during or shortly after induction and subsequently titrated dosages during surgery to provide both intra- and postoperative analgesia (Appendix 4). Median MEQ/kg doses were consistently higher for fentanyl, followed by sufentanil, oxycodone and lastly, morphine (Figure 2). MEQ/kg doses were comparable between sex and type of surgery (cholecystectomy or anterior cruciate ligament reconstruction) while being elderly or ASA 3 would lead to a dose reduction (Figure 2). This pattern was consistent regardless of which opioid respondents chose. The interquartile ranges (25-75%) of chosen doses were between 10.4 and 15.2 mg, which reflects a large inter-clinician variability in preferred opioid dosages for patients with identical baseline characteristics and surgical procedures.

3.5 Postoperative opioid administration strategies

Respondents selected the four most important factors for dose adjustments in the postoperative setting, which were chronic pain (77%), age (66%), preoperative opioid use (65%), body weight (50%) and type of surgery (38%) (Figure 1).

In all postoperative cases (variables: 30 or 65 years old; male or female; ASA score 1 or 3; moderate or severe pain), morphine i.v. was the preferred rescue opioid for patients with moderate postoperative pain (NRS 4–7), whereas fentanyl i.v. was preferred for patients with severe pain (NRS 8–10) (Table 3, Figure 3). MEQ/kg doses were highest for respondents using sufentanil, followed by oxycodone, morphine, and fentanyl.

When assessing all opioids, young, healthy patients would receive a median of 0.09 mg i.v. MEQ/kg (IQR 0.06–0.12) for moderate pain and 0.12 mg MEQ/kg (IQR 0.06–0.12) for severe pain (postoperative cases 1 and 3). If the patient was elderly and classified as ASA 3, median doses were reduced to 0.06 mg MEQ/kg (IQR 0.03–0.06) for moderate pain and 0.08 mg MEQ/kg (IQR 0.06–0.08) for severe pain (postoperative case 2 and 4; Appendix 4). Lastly, in a young, healthy patient with moderate pain who had received an initial rescue morphine dose and experienced no pain relief after 20 min, 81% (926/1148) would administer more morphine immediately, 9% (103/1148) would administer another opioid immediately, 9% (101/1148) would wait another 10 min and 1% (11/1148) would wait another 20 min (postoperative case 5).

3.6 Impact of pain and ORADEs

When scoring the negative impact on the recovery of pain and ORADEs (0–10 numeric rating scale), PACU nurses considered the most important factors to be severe pain (NRS 9; IQR 7–10), severe nausea (NRS 9; IQR 7–10), hallucinations (NRS 8; IQR 7–10) and vomiting (NRS 8; IQR 5–9) (Figure 4). Similar severity degrees of pain and nausea (mild, moderate or severe) were regarded to have a similar negative impact on patient recovery.

4.1 Factors influencing opioid administration

We found that anaesthesia personnel predominantly focused on chronic pain, age, preoperative opioid use, body weight and type of surgery to guide perioperative opioid dosing. This is consistent with findings from previous studies.^{16, 28-31} These studies also suggest that other factors may impact opioid requirements: female sex, psychological distress, anxiety and smoking, although the impact of female sex seems to be controversial.^{32, 33} While preoperative anxiety or psychological distress were not on the list of factors in our survey, anaesthesia personnel rarely added them under ‘other predictors’.

4.2 Choice of opioids

Danish anaesthesia personnel preferred fentanyl and morphine intraoperatively for postoperative analgesia. For postoperative rescue opioid analgesia, morphine i.v. was preferred for moderate postoperative pain, fentanyl i.v. or morphine i.v. were most preferred for severe pain, and fentanyl i.v. or oxycodone i.v. were preferred for severe pain with ORADEs. In the literature, it is debatable whether fentanyl, oxycodone, and morphine differ in their impact on the risk of ORADEs. However, it has been suggested that oxycodone or fentanyl may reduce ORADE incidences compared with morphine.^{9, 34} Due to its short-acting pharmacodynamics, fentanyl may require larger MEQ doses to provide adequate analgesia over time.³⁵ Longer treatment with large MEQ doses combined with potential likability from short-acting opioids may increase the risk of addiction and abuse.^36-39 Whether this is the case for short-term opioid treatment during the recovery phase is unclear. We observed that only a small number of anaesthesia personnel used methadone as first- or second-line treatment intraoperatively. A recent study suggests that methadone may reduce the need for rescue opioids without increased risk of ORADEs.⁴⁰

4.3 Opioid dosing variations

In our survey, we found considerable variation in opioid dosing, which has also previously been demonstrated among anaesthesia personnel in the intraoperative phase,^41-43 during recovery³⁶ and among surgeons at discharge.⁴⁴ These findings can partly be explained by limited evidence, random variation, and individual emphasis on various factors.

Current evidence suggests that i.v. MEQ between 3 and 6 mg corresponds to one meaningful ORADE, around 5 mg i.v. MEQ seems reasonable to consider as a clinically relevant difference for postoperative opioid consumption.^{7, 45} A recent study reported that up to 20% of patients experience moderate to severe pain shortly after PACU arrival.⁴⁶ There is likely an optimal intraoperative opioid dose for each unique patient in terms of benefits and harm, resulting in the lowest level of pain and ORADEs after anaesthesia.²⁶

Opioid dosing ranges found in this survey come from hypothetical cases but seem to approximate opioid doses that have been reported previously.⁴⁷ We cannot make statements about how Danish anaesthesia personnel actually dose. However, our survey helps to elucidate patterns in anaesthesiologists, nurse anaesthetists and PACU nurses. Heterogeneity at the clinician level seems to contribute strongly to variation despite high adherence to guidelines. This study does not evaluate the quality or efficacy of guidelines; therefore, we cannot make statements about the most optimal opioid strategies. However, we encourage using multimodal procedure-specific pain management as the current gold standard.⁴⁸ Given the considerable variations in opioid doses, evidence-based guidelines, pain education, and individualised dosing algorithms could guide clinicians in targeting the optimal doses.

Higher intraoperative MEQ doses for short-acting opioids, such as fentanyl and sufentanil, found in our survey suggest that opioid conversions may not only depend on potency but also on duration of action.⁴⁹

4.4 Impact of ORADEs

PACU nurses regarded nausea and pain to be equally negative factors impacting recovery in the PACU, which aligns with patient perspectives on nausea.^{50, 51} Apart from being directly patient-relevant, ORADEs negatively impact health care costs and clinical outcomes,^{52, 53} and pain management strategies should emphasise reducing ORADEs.

4.5 Strengths and limitations

A strength of the survey is the nationwide coverage and the use of site investigators, which ensured acceptable overall response- and missingness rates compared with other surveys in pain management, where response rates range between 13% and 31%.^54-57 Other strengths are that the survey was pilot-tested three times in different regions before distribution, and a protocol was published before the survey launch. Our findings on opioid dosing variation do not surprise us because this phenomenon is well-documented. Nonetheless, this survey makes an earnest and novel attempt to provide, as comprehensively as possible, a multidimensional perspective of the aspects that influence dosing across the perioperative course amongst Danish anaesthesia personnel. It facilitates stronger awareness and discussion about opioid pain management, and it offers the potential for creating more user-friendly guidelines that encompass patient-specific aspects. In addition, it elucidates some aspects of opioid practice that can help inform the design of artificial intelligence-based prediction models of optimal opioid doses for patients.

Our survey has some important limitations. The clinical cases are simplified compared with daily clinical practice, and we refrained from advanced statistical analyses on these because we considered them beyond the scope of this survey. In clinical scenarios, patients received systemic multimodal analgesia but not regional anaesthesia despite its regular use in clinical practice. We excluded regional anaesthesia to avoid response fatigue given the level of detail of such procedures, for example, type of technique, concentration, dose and adjuvants. Likewise, we did not specify intraoperative remifentanil consumption nor make it clear whether local infiltration analgesia was administered in the clinical scenarios—not doing so may have increased the variance of our results. Moreover, anaesthesia personnel who do not provide anaesthesia and analgesia to abdominal and/or orthopaedic surgeons (<30% of respondents) may have chosen doses for clinical cases based on prior experience in these specialities. However, we did not inquire further into how much time had elapsed since respondents potentially held such positions. Only a small number of respondents used combinations of opioids in the postoperative cases. These were omitted from the results section even though anaesthesia personnel likely combine drugs and doses to provide short- and long-term pain relief. Further, we did not include the small number of entries such as ‘preoperative anxiety’ or ‘psychological distress’ under ‘other predictors’ regarding factors influencing opioid administrations. We sought to strike a balance between providing sufficient information and avoiding respondent fatigue, which may have affected and reduced the validity of our results. Potential sources of bias are variations in response- and missingness rates. In terms of generalisability, an overall response rate of 48% does not reflect the entire population of Danish anaesthesia personnel. To the best of our knowledge, other nationwide surveys on opioid practices amongst anaesthesia personnel in other countries do not exist. Therefore, whether our results are comparable in the broader context is uncertain.