NEWS
Our prospective cohort study is well up and running with 429 included patients.
We are working on user interface with collaborators at the IT University og Copenhagen - check it out
Motivation
300 million surgical procedures each year globally - most require opioids.
Individual opioid titration guidelines are non-existent.
30-60% of surgical patients experience pain or opioid adverse events.
Demographic, genetic and surgical predictors could help develop individualised opioid algorithms.
Research program
Designed in accordance with the highest methodological standards for developing prediction models for personalised medicine. Divided into six phases:
OPI•AID Project Description Oct. 2024
The OPI•AID Algorithm extracts patient-, anesthesia- and surgical data to predict optimal opioid doses for postoperative pain management. It is an advanced therapy medicinal product (ATMP) that can be implemented in electronic patient record systems (EPR) or used in a simplified online version.
BACKGROUND: Annually, 300 million major surgical procedures are performed globally, most of which result in moderate to severe pain immediately following the surgery.1-4 Procedure-specific multimodal analgesia is the current paradigm for preventing postoperative pain.5-7 A multimodal regimen, administered during surgery, typically involves paracetamol, ibuprofen, dexamethasone, and a morphine dosage ranging from 0.1 to 0.3 mg/kg.8, 9 Hence, opioids are essential elements of acute pain management, but the evidence on individual opioid requirements during surgery and in recovery, is sparse; and opioid doses are mostly based on clinical opinion.10 Overestimating doses increases the amount of opioid-related adverse events, such as respiratory depression – potentially leading to hypoxia, atelectasis and pneumonia; constipation – potentially causing ileus; and sedation – which is associated with postoperative delirium.11 Underestimating opioid doses may lead to moderate or severe postoperative pain, causing increased stress and immobility which can lead to severe complications such as cardiovascular events.10 Thus, inaccurate dosing of opioids can have devastating consequences for patients.
Despite optimized procedure-specific analgesic regimens, a staggering 30-60% of surgical patients experience moderate-to-severe pain and opioid-related adverse events in the postanesthetic care unit and the first 24 h after surgery.4, 11-15 They decrease patient-perceived quality of care16 and impair perioperative goals with trajectories into long-term health outcomes. They delay discharge from the postanesthetic care unit and hospital, which impairs organizational efficiency.17-19 Moreover, even short-term opioid use can lead to long-term opioid abuse with devastating consequences as seen in the opioid epidemic.20-22 Thus, research on optimized opioid dosages is vital.
Precision medicine with algorithm-based models is becoming more effective in developing decisions making tools to support clinicians due to advances in machine learning technology and availability of large clinical databases.23 Evidence on how demographic and surgical exposures act together as predictors for opioid requirements is almost non-existing and individualized analgesic algorithms have not been developed yet.24, 25
AIM: Our central goal is to optimise perioperative opioid administrations to the individual need.
Beside improving the quality of life for patients, we aim at improving the following outcomes:
1. Reducing the average PACU stay
2. Reducing the need for PACU referrals
3. Reducing hospital stay length
4. Increasing Surgical Capacity
5. Reducing hospital costs.
6. Reduce persistent opioid use after surgery
HYPOTHESIS: Easily obtainable patient characteristics and perioperative factors (age, sex, physical function, BMI, co-morbidities, frailty, opioid use, anxiety, anaesthesia factors, type and dose of preventive analgesics, type of procedure and difficult/prolonged surgery) may be combined into a prediction model to create an individualised perioperative morphine dosing algorithm. This model may decrease the incidence of pain or opioid-related adverse events (OAE) after surgery and improve patient care.
NOVELTY: OPI•AID will be the first algorithm globally to assist surgery-related opioid dosing.
ORGANISTATION: OPI•AID is a collaborative effort between Bispebjerg and Frederiksberg Hospitals, Zealand University Hospital Køge, Copenhagen Trial Unit and the IT University of Copenhagen. OPI•AID has raised 4 million DKK in funding and departmental support. OPI•AID currently has 6 full-time collaborator, has weekly involvement of +30 collaborators and +10 ongoing projects.
Overview of the Research Program (2024-2029)
Phase 1: Establishing Background Information
We currently gather background information from both the existing literature and current clinical practices related to opioid administration. This phase includes several studies: systematic reviews with meta-analyses, survey studies, and observational studies conducted in clinical settings. These efforts will provide a solid foundation for understanding the current evidence and practices surrounding opioid use.
Phase 2: Developing The OPI•AID Algorithm for Intraoperative Opioid Dosing
In phase 2, we develop the intraoperative opioid dosing algorithm for optimized postoperative analgesia during the first few hours after surgery, particularly while patients are recovering in the postoperative care unit. This algorithm is based on dense granular data from over one million surgical procedures, spanning every surgical specialty, conducted between 2017 and 2024 in the Capital and Zealand Regions of Denmark. Ethical approval Capital Region (R-23038250: 14.07.23; P-2023-14978: 09.02.24).
The OPI•AID database includes a wide range of demographic, anesthesia, and surgical input variables, as well as detailed information on relevant outcomes such as pain, opioid-related adverse events, and opioid consumption. Using this rich dataset, we develop the algorithm via machine-learning and conditional average dose-response curves, which allow us to predict the likelihood of pain and opioid-related adverse events based if other doses of opioids had been administered to the patients in the database.
The algorithm will be developed using best practice model-training approach, with data split into 80% for training, 15% for testing, and 5% for internal validation.26 This ensures a robust development process.27
The OPI•AID algorithm is a software that can be integrated into electronic patient record systems for automated predictions – at first targeted at- and developed for implementation in the EPR system Sundhedsplatformen (EPIC). Together with our collaborators at Team AI in SP Sundhedsdata we have a detailed implementation plan that commences during summer 2025.
We will also develop a simplified internet-based version, allowing clinicians to input key patient data directly. This flexibility broadens the potential for use in various clinical settings.
Phase 3: External Validation of the Algorithm
Following the development and internal validation, we will externally validate the OPI•AID algorithm in adherence with current guidelines.28, 29 This will be achieved through a multicenter observational study involving 700 patients with rigorously monitored pain outcomes and patient satisfaction,30 along with patient involvement in a semi-structured interview. The purpose of this phase is to ensure that the retrospective cohort used to build the database is accurate and that it has not been affected by poor data registration.
Phase 4: Clinical Testing of the Algorithm
Following the external validation, we proceed with clinical testing in a real-world setting, by conducting a randomized trial involving 330 patients, comparing the algorithm’s performance against current standard of care. Tested outcomes will be pain, opioid-related adverse events, opioid consumption and organizational and cost efficiency.
Phase 5: Clinical implementation
If the algorithm proves to be both safe and effective, it will be implemented under a quality control framework for new medical devices using the article 5.5 in-house rule. Together with our collaborators at CIMT in The Capital Region, we have a detailed plan for this application and implementation process that commences during summer 2026. Implementation of the algorithm in clinical practice, allows us to improve its performance by retraining on actual clinical responses to algorithm recommended doses. This will be the OPI•AID algorithm 2.0.
Phase 6: Large-Scale Trial
Following the initial clinical testing, we will move into a large-scale trial involving 3,000 patients testing the optimised OPI•AID algorithm 2.0. This trial will further evaluate the algorithm’s performance against standard of care, allowing us to assess its effectiveness, safety, and overall impact on clinical outcomes on a broader scale.
Phase 7: Expanding AI Models to Other Areas of Perioperative Medicine
After the initial development and clinical testing of the OPI•AID algorithm, we will move on to create parallel models for postoperative rescue opioid doses and non-opioid analgesia administration. These additional models will further enhance postoperative pain management. They will be developed following the same approach as the initial OPI•AID algorithm, incorporating the insights gained from its development to streamline the process.
Once the expanded pain management models are established, we will leverage our knowledge, database, and project framework to develop additional decision aid tools for other aspects of perioperative medicine. This expansion will allow us to apply AI-driven algorithms to broader clinical problems beyond pain management, enhancing perioperative care in general. This broader initiative has been formed in collaboration with the Danish nationwide initiative Collaboration for Evidence-based Practice and Research in Anaesthesia (CEPRA)31 and is called AI and automated CRFs in Anesthesia (Triple•A).
Conclusion
In summary, the project addresses the issue of suboptimal perioperative opioid management and evolves into a broader initiative for improving various aspects of perioperative medicine through AI-driven decision support tools.
References
1 Weiser TG, Haynes AB, Molina G, et al. Estimate of the global volume of surgery in 2012: an assessment supporting improved health outcomes. Lancet 2015; 385 Suppl 2: S11
2 Gasbjerg KS, Hagi-Pedersen D, Lunn TH, et al. Effect of dexamethasone as an analgesic adjuvant to multimodal pain treatment after total knee arthroplasty: randomised clinical trial. BMJ 2022; 376: e067325
3 Thybo KH, Hagi-Pedersen D, Dahl JB, et al. Effect of Combination of Paracetamol (Acetaminophen) and Ibuprofen vs Either Alone on Patient-Controlled Morphine Consumption in the First 24 Hours After Total Hip Arthroplasty: The PANSAID Randomized Clinical Trial. JAMA 2019; 321: 562-71
4 Gerbershagen HJ, Aduckathil S, van Wijck AJ, Peelen LM, Kalkman CJ, Meissner W. Pain intensity on the first day after surgery: a prospective cohort study comparing 179 surgical procedures. Anesthesiology 2013; 118: 934-44
5 Wick EC, Grant MC, Wu CL. Postoperative Multimodal Analgesia Pain Management With Nonopioid Analgesics and Techniques: A Review. JAMA Surg 2017; 152: 691-7
6 Wainwright TW, Gill M, McDonald DA, et al. Consensus statement for perioperative care in total hip replacement and total knee replacement surgery: Enhanced Recovery After Surgery (ERAS(®)) Society recommendations. Acta Orthop 2020; 91: 3-19
7 Memtsoudis SG, Cozowicz C, Bekeris J, et al. Peripheral nerve block anesthesia/analgesia for patients undergoing primary hip and knee arthroplasty: recommendations from the International Consensus on Anesthesia-Related Outcomes after Surgery (ICAROS) group based on a systematic review and meta-analysis of current literature. Reg Anesth Pain Med 2021; 46: 971-85
8 https://esraeurope.org/prospect/, The PROSPECT (PROcedure-SPECific postoperative pain managemenT) initiative. 2024)
9 Steiness J, Hagi-Pedersen D, Lunn TH, et al. Non-opioid analgesic combinations following total hip arthroplasty (RECIPE): a randomised, placebo-controlled, blinded, multicentre trial. Lancet Rheumatol 2024; 6: e205-e15
10 Aubrun F, Mazoit JX, Riou B. Postoperative intravenous morphine titration. Br J Anaesth 2012; 108: 193-201
11 Benyamin R, Trescot AM, Datta S, et al. Opioid complications and side effects. Pain Physician 2008; 11: S105-20
12 Halawi MJ, Grant SA, Bolognesi MP. Multimodal Analgesia for Total Joint Arthroplasty. Orthopedics 2015; 38: e616-25
13 Smith HS, Laufer A. Opioid induced nausea and vomiting. Eur J Pharmacol 2014; 722: 67-78
14 Rasmussen AM, Toft MH, Awada HN, et al. Waking Up in Pain: a prospective unselected cohort study of pain in 3702 patients immediately after surgery in the Danish Realm. Reg Anesth Pain Med 2021; 46: 948-53
15 Aubrun F, Zahr N, Langeron O, et al. Opioid-related genetic polymorphisms do not influence postoperative opioid requirement: A prospective observational study. Eur J Anaesthesiol 2018; 35: 496-504
16 Gan TJ, Lubarsky DA, Flood EM, et al. Patient preferences for acute pain treatment. Br J Anaesth 2004; 92: 681-8
17 Dahl JB, Nielsen RV, Wetterslev J, et al. Post-operative analgesic effects of paracetamol, NSAIDs, glucocorticoids, gabapentinoids and their combinations: a topical review. Acta Anaesthesiol Scand 2014; 58: 1165-81
18 Kehlet H, Dahl JB. Anaesthesia, surgery, and challenges in postoperative recovery. Lancet 2003; 362: 1921-8
19 Kehlet H, Wilmore DW. Evidence-based surgical care and the evolution of fast-track surgery. Ann Surg 2008; 248: 189-98
20 Kelly MA. Current Postoperative Pain Management Protocols Contribute to the Opioid Epidemic in the United States. Am J Orthop (Belle Mead NJ) 2015; 44: S5-8
21 Raji Y, Strony JT, Trivedi NN, et al. Effects of opioid-limiting legislation on postoperative opioid use in shoulder arthroplasty in an epidemic epicenter. J Shoulder Elbow Surg 2022; 31: 269-75
22 Judd D, King CR, Galke C. The Opioid Epidemic: A Review of the Contributing Factors, Negative Consequences, and Best Practices. Cureus 2023; 15: e41621
23 Johnson KB, Wei WQ, Weeraratne D, et al. Precision Medicine, AI, and the Future of Personalized Health Care. Clin Transl Sci 2021; 14: 86-93
24 Papadomanolakis-Pakis N, Uhrbrand P, Haroutounian S, Nikolajsen L. Prognostic prediction models for chronic postsurgical pain in adults: a systematic review. Pain 2021; 162: 2644-57
25 Verwoerd MJ, Wittink H, Maissan F, van Kuijk SMJ, Smeets R. A study protocol for the validation of a prognostic model with an emphasis on modifiable factors to predict chronic pain after a new episode of acute- or subacute nonspecific idiopathic, non-traumatic neck pain presenting in primary care. PLoS One 2023; 18: e0280278
26 Pedersen NK, Andersen JV. The Right Tool for the Job: The Effects of Data Volume and Machine Learning Model Complexity on Decision Accuracy in Business Analytics. ECIS: European Conference on Information Systems, 2023
27 Collins GS, Dhiman P, Ma J, et al. Evaluation of clinical prediction models (part 1): from development to external validation. BMJ 2024; 384: e074819
28 Riley RD, Archer L, Snell KIE, et al. Evaluation of clinical prediction models (part 2): how to undertake an external validation study. BMJ 2024; 384: e074820
29 Riley RD, Snell KIE, Archer L, et al. Evaluation of clinical prediction models (part 3): calculating the sample size required for an external validation study. BMJ 2024; 384: e074821
30 Karlsen APH, Tran TXM, Mathiesen O, et al. The OPI*AID Zone Tool as a composite outcome for postoperative pain management quality-A protocol for an observational pilot study. Acta Anaesthesiol Scand 2024
31 Norskov AK, Jakobsen JC, Afshari A, et al. Collaboration for Evidence-based Practice and Research in Anaesthesia (CEPRA): A consortium initiative for perioperative research. Acta Anaesthesiol Scand 2023; 67: 804-10
Since the foundation of OPI•AID, we have:
- Finalized the OPI•AID research plan and our collaboration.
- Funded 900,000 DKK and raised >1,000,000 DKK in departmental support.
- Employed a PhD-student and four master students for a year.
- Ensured chair Anders Karlsen an associate professor position.
- Designed and almost done cleaning the 895,000 patient database.
- Collaborated with IT University Copenhagen in cleaning data and machine learning model development.
- Finalized a national survey on opioid treatment for anaesthesia personnel (2025 respondents).
- Finalized a 145 patient pilot observational cohort for the algorithm validation study.
- Published three studies, while having ten upcoming submissions during 2024.
- Partnered with The Collaboration of Anaesthesia Researchers in Denmark-CEPRA for site recruitment in our algorithm test trials.
- Started a new CEPRA-cluster: The Data Automation Cluster - using the OPI•AID database for perioperative epidemiological research and data retrieval in prospective studies.
- Designed multiple side-studies.
- Applied and been granted important ethical approvals regarding extending the database to end 2026, making data retrieval in our validation study and clinical test trial much easier.
.jpeg?etag=%22400573-6578cd56%22&sourceContentType=image%2Fjpeg&rotate=90&ignoreAspectRatio&resize=374%2B499&extract=7%2B18%2B342%2B414&quality=85)
The OPI•AID Project group is situated in Denmark at:
- Dept. of Anaesthesia and Intensive Care Medicine, Bispebjerg and Frederiksberg University Hospital
- Dept. of Anaesthesiology, Zealand University Hospital, Køge
- Copenhagen Trial Unit, Rigshospitalet
- The IT-University of Copenhagen
- In collaboration with CEPRA (cepra.nu)
