PAIN. Editor s key points

Transcription

1 British Journal of Anaesthesia 108 (6): (2012) Advance Access publication 30 March doi: /bja/aes091 PAIN Cost-effectiveness analysis comparing epidural, patient-controlled intravenous morphine, and continuous wound infiltration for postoperative pain management after open abdominal surgery P. Tilleul 1,2, M. Aissou 3,4, F. Bocquet 1, N. Thiriat 1, O. le Grelle 1, M. J. Burke 5, J. Hutton 5,6 and M. Beaussier 3,4 * 1 Department of Pharmacy, Assistance Publique Hopitaux de Paris, St Antoine Hospital Paris, France 2 University Paris Descartes Paris 5, France 3 Department of Anaesthesia and Intensive Care, Assistance Publique Hopitaux de Paris, St Antoine Hospital Paris, France 4 University Pierre and Marie Curie Paris 6, France 5 York Health Economics Consortium and 6 Department of Health Sciences, University of York, York, UK * Corresponding author: Département d Anesthésie-Réanimation chirurgicale, Hôpital St Antoine, 184 rue du Fbg St Antoine, Paris Cédex 12, France. marc.beaussier@sat.aphp.fr Editor s key points A range of analgesic techniques are commonly used after laparotomy. There has been no comparison of the cost-effectiveness of some of these commonly used techniques. This analysis combines prospective observational data with published clinical trial results. Continuous wound infiltration may provide equally effective analgesia for a lower cost than epidural analgesia. Background. Continuous wound infiltration (CWI), i.v. patient-controlled analgesia (i.v.- PCA), and epidural analgesia (EDA) are analgesic techniques commonly used for pain relief after open abdominal surgery. The aim of this study was to evaluate the costeffectiveness of these techniques. Methods. A decision analytic model was developed, including values retrieved from clinical trials and from an observational prospective cohort of 85 patients. Efficacy criteria were based on pain at rest (VAS 30/100 mm at 24 h). Resource use and costs were evaluated from medical record measurements and published data. Probabilistic sensitivity analysis (PSA) was performed. Results. When taking into account all resources consumed, the CWI arm (E6460) is economically dominant when compared with i.v.-pca (E7273) and EDA (E7500). The proportion of patients successfully controlled for their postoperative pain management are 77.4%, 53.9%, and 72.9% for CWI, i.v.-pca, and EDA, respectively, demonstrating the CWI procedure to be both economically and clinically dominant. PSA reported that CWI remains cost saving in 70.4% of cases in comparison with EDA and in 59.2% of cases when compared with PCA. Conclusions. Device-related costs of using CWI for pain management after abdominal laparotomy are partly counterbalanced by a reduction in resource consumption. The costeffectiveness analysis suggests that CWI is the dominant treatment strategy for managing postoperative pain (i.e. more effective and less costly) in comparison with i.v.-pca. When compared with EDA, CWI is less costly with almost equivalent efficacy. This economic evaluation may be useful for clinicians to design algorithms for pain management after major abdominal surgery. Keywords: decision tree; economics; epidural analgesia; laparotomy; pain, postoperative Accepted for publication: 28 December 2011 Abdominal laparotomy is associated with intense pain in the immediate postoperative period. This pain has a major impact on patients satisfaction, 1 and may negatively interfere with the postoperative recovery course. 23 In the setting of abdominal surgery, postoperative i.v. morphine patient-controlled analgesia (PCA) continues to be frequently used, 4 although it has been clearly identified as contributing to the delay of postoperative recovery due to side-effects such as postoperative nausea and vomiting (PONV), prolongation of gastric ileus, sedation, and dizziness. 5 After abdominal laparotomy, epidural analgesia (EDA) using local anaesthetics provides better analgesia than i.v. morphine PCA, especially for pain at mobilization. 67 However, there are some limitations to the use of EDA in daily practice such as contra-indications, technical failure, and side-effects. 689 & The Author [2012]. Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved. For Permissions, please journals.permissions@oup.com

2 Economic analysis of postoperative pain management BJA Patient eligible No intra-operative failure Patients treated with favoured CWI and EDA Successful pain relief Unsuccessful pain relief CWI / EDA Intra-operative failure Patients now treated with i.v. PCA Successful pain relief Successful pain relief Unsuccessful pain relief Patient not eligible Patients now treated with i.v. PCA Unsuccessful pain relief Successful pain relief i.v. PCA Unsuccessful pain relief Fig 1 Decision analytic model. Continuous wound infiltration (CWI) is another analgesic technique of administering local anaesthetic directly into the surgical incision by way of a multi-holed catheter placed by the surgeon at the end of the procedure. Compared with i.v. morphine PCA, CWI provides better analgesia and demonstrates beneficial effects for postoperative recovery. 10 The technique is easy to perform, with a high success rate, and good tolerance. 11 Compared with conventional parenteral analgesia, CWI allows for a reduction in the duration of hospitalization. Efficacy, tolerance, ease of use, and potential benefit for recovery represent the main criteria for medical staff. However, all the mentioned analgesic strategies could have different economic consequences relating to the acquisition costs of drugs and devices, nursing and medical time, management of side-effects, and duration of hospital stay. For example, it has been demonstrated that opioid-related adverse events increase the length of hospital stay, resulting in significant additional costs; 5 and staff costs represent more than 50% of the total costs associated with EDA management. 14 EDA has been identified as the more expensive treatment strategy, but it was shown to be more effective in comparison with i.v. morphine PCA in terms of pain-free days. 15 After open nephrectomy, Forastiere and colleagues 16 showed that the additional acquisition cost of CWI could be offset by a reduction in the duration of hospitalization when compared with i.v. morphine PCA alone. Until now, no cost-effectiveness analysis has been conducted comparing CWI with i.v. morphine PCA and with EDA. In this study, we aimed to develop a decision-tree model, based on peer-reviewed literature and clinical observational data, in order to assess total costs of the techniques and to generate an incremental cost-effectiveness ratio (ICER) to assist healthcare providers in their decision-making process. Methods A decision tree was developed using Microsoft Excel 2007 to model the clinical pathways followed by patients receiving each of the three postoperative pain management techniques (Fig. 1). Decision trees are useful for treatments which can be measured as a one-off outcome (e.g. success or failure ). 17 The model estimates the costs and outcomes associated with pain management after open major abdominal surgery. As usually recommended, 18 the model has been populated from a variety of sources including a cohort of 85 observational patients (pain management, duration of hospital stay, and micro-costing evaluations) and randomized controlled trial/meta-analysis (incidence of side-effects). The observational study was conducted in the St Antoine University Hospital, where the three analgesic techniques under investigation are all commonly performed. The study received IRB approval (ethical committee 5 Paris no ). Patients were informed and gave their consent to participate in the study. Patients were prospectively assessed during a predefined period of evaluation. According to usual practices, availability of the devices, anaesthetists /patients preferences, and organizational parameters, they received one of the analgesic strategies under evaluation, in exact accordance with the usual clinical practice. In order to estimate a potential selection bias, patient characteristic data (i.e. age, weight, gender, ASA status, types of surgery, duration of the procedures) were checked in all subgroups. 999

3 BJA Tilleul et al. Analgesic techniques and efficacy outcomes All the patients were operated on under general anaesthesia. According to the usual practice in our institution, i.v. sufentanil was used intraoperatively in all patients. The patients received one of the following postoperative analgesic techniques described below. Group i.v.-pca (n¼43) A PCA device (Graseby 9300, Watford Herts, UK) was connected in the post-anaesthesia care unit (PACU) to an i.v. infusion set to deliver a 1 mg dose of morphine with a 5 min lock-out time. Before surgery, each patient had received information on the PCA device. PCA was maintained throughout the first 48 postoperative hours. Group EDA (n¼22) An epidural catheter (Vygon, Ecouen, France) was inserted at the thoracic level (between T7 and T10) before the anaesthetic induction. The correct placement of the catheter was checked with a test dose of 4 ml 2% lidocaine. EDA was not used during the surgery. At the end of the procedure, EDA was started using a bolus of 4 6 ml of a mixture of 1.25 mg ml 21 levobupivacaine and 0.5 mg ml 21 sufentanil followed by a patient-controlled epidural system (4 6 ml h 21 continuous infusion rate with 2 4 ml bolus with a 20 min lock-out time). Group CWI (n¼20) At the end of the surgery, after closure of the parietal peritoneal membrane with running sutures, the surgeon inserted a 20 G multiholed (fenestrated) catheter (Painfusor Baxter Healthcare Corp., Maurepas, France) into the wound through an introducer needle 3 cm from the lower end of the midline incision. The catheter was positioned in a preperitoneal position as previously described by our group. 10 Once the wound was closed, a 10 ml bolus of 0.2% ropivacaine was administered through the catheter. Thereafter, a pre-filled elastomeric pump, set to deliver a 10 ml h 21 rate for 48 h, was connected immediately. In addition, all patients received i.v. paracetamol 1 g at the end of the procedure and every 6 h for the first 2 postoperative days. Pain management efficacy was derived from the observational cohorts. Pain intensity was measured using the numerical verbal scale (NVS: 0 10) for both pain at rest and at mobilization (while moving from a lying to a sitting position) at the 24th postoperative hour. Analgesia success (controlled patients) was defined as the percentage of patients experiencing NVS 3/10 for pain at rest at the 24th postoperative hour. Side-effects Incidence of side-effects, duration of hospitalization, and failure to provide the chosen technique were derived from published meta-analyses and major clinical trials. The inclusion of appropriate side-effects was made based on their incidence, their clinical relevance, and their documentation in the literature. We therefore focused on PONV, the incidence of which was derived from meta-analyses comparing the analgesic strategies Values selected to populate the model were 36%, 24%, and 18% for the i.v.-pca, CWI, and EDA subgroups, respectively. Several other side effects were not taken into account, such as urinary retention (most of the patients have a urinary bladder catheter) and respiratory depression (rare and highly variable according to definitions). Major complications, such as permanent disability, were also not accounted for in the model as they are extremely rare and their costs difficult to estimate. Cost of pain management The total costs for providing each postoperative pain management technique included those relating to medical devices, drugs, and time (medical and nursing) spent to manage postoperative pain. Nursing protocols used for the micro-costing evaluation have included the direct measurement of times corresponding to different steps in the clinical pathway: (i) preparation of the drugs (ordering both drugs and devices, dilution, filling the Infusor, adjustment to the patient, replacement of empty Infusor, preparation of drugs for rescue analgesia, if needed) and (ii) monitoring/supervision, response to patient s solicitation, and corrective actions if necessary. These different times have been recorded during three distinct settings (operating theatre, PACU, and ward) for each of the analgesic techniques (Fig. 2). The corresponding costs were calculated from the hospital administration for manpower costs (mean value 25.8E h 21 for nurses and 57.6E for medical staff) and from the pharmacy database for drugs and medical devices. Min Operating theatre Post-anaesthesia care unit Ward CWI Epidural i.v. PCA Analgesic technique Fig 2 Mean time spent by healthcare professionals to implement and supervise the analgesic technique. 1000

4 Economic analysis of postoperative pain management BJA The cost of PONV, including personnel, material, and antiemetic drugs costs, was estimated at 82.5E derived from the study by Hill and colleagues. 25 The duration of hospitalization was estimated from our observational cohort. The mean cost for each day of hospitalization was directly calculated from the French Diagnostic Regimen Group (T2A) national accounting system (765E per hospital day). Economic analysis: model and assumptions Figure 1 shows a simplified version of the decision-tree model. All patients enter the model after successful open abdominal surgery. The pathways consist of a number of particular events, including a patient s willingness to benefit from a particular technique, the rate of intraoperative technical failure, successful pain relief, and potential adverse events. The uncertainty that a patient will experience a particular track is defined by a number of probabilities for each of the given events. The expected costs and outcomes are based on the sum of the track values weighted by the probabilities. The time horizon selected for the model was 48 h. This allowed for most of the postoperative events to be captured and corresponded to the time horizon of other clinical trials in postoperative pain management. This analysis was conducted from the perspective of the healthcare provider. As such, it seeks to take into account direct treatment costs but makes no attempt to capture information on indirect costs. Given this model considered only the immediate outcomes, discounting of costs and benefits was not necessary. Using the literature as mentioned above, the following values were used in the cost-effectiveness analysis model. Among the patients to whom EDA was proposed, 22% declined. 26 Once in the operating theatre, technical failure prevented the patient benefiting from EDA in 6.3% 27 to 7.2% 22 of cases. In the model, we used an average value of 6.8%. Once correctly administered, EDA does not provide satisfactory analgesia in 25 30% of cases Of the patients who experienced ineffective analgesia, 60% had EDA discontinued and were transferred into the i.v.-pca subgroup, and 40% went on to have EDA maintained with additional i.v.-pca All patients who were ineligible or experienced a technical failure after EDA received i.v.-pca. As all the patients were before operation trained in the appropriate use of i.v.-pca, they have been considered as correctly using their PCA devices. Numerous assumptions have been made in order to build the model (see Appendix). A procedure was defined as economically dominant when both cost saving and more effective. Probabilistic sensitivity analysis Probabilistic sensitivity analysis (PSA) was undertaken to estimate the level of confidence around the model s incremental cost per controlled patient. Distributions were fitted to key model parameters where the standard deviations were Table 1 Patient characteristic data, type of surgery, and duration of the procedure in the three subgroups of patients. No difference between the groups. Results in means [range], mean (SD) and number of patients. *Upper digestive surgery corresponds to gastric resection, gastro-jejunal derivation, and small intestinal resection available (Tables 2 and 3). For each model, 1000 Monte Carlo simulation iterations were performed. Statistical analysis Results on pain intensity at the 24th postoperative hours, derived from the observational cohort of patients, were compared using one-way analysis of variance (ANOVA). In the case of significantly statistical difference between subgroups, the Student Newman Keuls tests were performed for post hoc pairwise comparisons. A contingency table for multiple comparisons was used to compare results on the rate of successful analgesia. The threshold of statistical significance was set as,0.05. Results I.V.-PCA (n543) CWI (n520) EDA (n522) Age (yr) 60 (19 83) 60 (26 76) 53 (20 78) Weight (kg) 72 (14) 69 (15) 71 (16) ASA score (I/II/III) 10/25/8 7/8/5 5/10/7 Duration of surgery 206 (106) 253 (104) 268 (96) (min) Type of surgery (n) Upper digestive* Colorectal Hepato-biliary The three subgroups of patients from the observational cohort used to populate the model did not differ in patient characteristic data, type of surgery, and duration of the procedures (Table 1). Efficacy parameter Values for pain at rest reported at the 24th postoperative hour are 4.1 ( ), 2.6 ( ), and 2.4 ( ) cm for i.v.-pca, CWI, and EDA, respectively (ANOVA P,0.001, with statistically significantly better efficacy for CWI and EDA when compared with i.v.-pca). The proportion of patients who achieved successful analgesia, that is, NVS 3/10 cm, were estimated to be 41%, 65%, and 72% for i.v.-pca, CWI, and EDA, respectively (P¼0.016). Average treatment costs The cost of drugs and devices and resource consumption was recorded during the first 48 h of the postoperative period (Table 2). The duration of hospitalization was 9.3 (3.5), 7.9 (2.0), and 9.8 (2.2) days in i.v.-pca, CWI, and EDA subgroups, respectively (ANOVA P¼0.067). 1001

5 BJA Tilleul et al. Table 2 Resource use for different procedures and their unit costs. Results in means (SD). Assistance Publique Hôpitaux de Paris (APHP) 2010 list price for drugs and devices. OT, operating theatre; PACU, post-anaesthesia care unit. *n, number of units (stated if.1 unit). The average wage per minute was 0.43E for nurses and 0.96E for medical times I.V.-PCA n* CWI n* EDA n* Medical devices Peridural set 21.85E Surgical gowns 1.78E Sterile gloves 0.37E Syringe 5 ml 0.03E E E Syringe 20 ml 0.07E Syringe 50 ml 0.20E 0.20E 0.20E PCA device 1.37E 2 Tube for EDA 12.86E Tube for PCA 6.10E 6.10E Other tube (Infusor) 0.55E Needle 0.01E 0.01E E EDA (amortization) 2.80E Scope (amortization) 3.20E CWI catheter 50.00E Infusor set 37.65E Subtotal med. dev. 9.11E 94.66E 43.10E Drugs Cutaneous disinfectant 1.24E Lidocaine 10 mg ml E Morphine 50 mg 0.35E 0.35E 2 L-bupivacaine bags 16.30E Sufentanil 50 mg 10ml E 2 Droperidone 2.98E Ropivacaine bags 18.16E 2 Ropivacaine 40 mg 20 ml E NaCl 0.9% bags 0.67E 0.67E Subtotal drugs 4.00E 40.22E 19.20E Staff cost for preparation Medical Nurse 13.18E (4.0) 26.55E (6.4) 28.30E (7.8) Follow-up cost Medical time (OT) E (1.8) 8.10E (2.4) Nurse time (OT) E(2.4) Nurse time (PACU) 3.80E (0.7) 4.30E (1.2) 2.10E (0.5) Nurse time (Ward) 14.62E (6.1) 9.03E (4.9) 47.73E (10.7) Total costs (E) 44E (9.0) 181E (7.1) 158E (12.3) When only considering the costs of the medical devices, drug, and staff time, CWI (E181) is more costly than both EDA (E158) and i.v.-pca (E44) (Table 2). When taking into account the management of side-effects and hospitalization costs, the average treatment cost for the CWI arm was lower (E6460) than for both the i.v.-pca (E7273) and the EDA (E7500) arms, representing an average cost saving of E753 when compared with i.v.-pca and E1040 with EDA per treated patient (Table 3). Proportion of patients whose pain was controlled When the proportion of controlled patients were adjusted to reflect patient eligibility for the techniques and unsuccessful pain relief from the first analgesia, the economic model estimated that 77.4%, 53.9%, and 72.9% of patients achieve successful pain relief from CWI, i.v.-pca, and EDA, respectively (Table 3). CWI was found to be the dominant strategy when compared with i.v.-pca (both more effective and less costly). When CWI is compared with EDA, it has a lower treatment cost, but the proportion of patients who experience successful pain relief are almost equivalent. One-way sensitivity analysis Table 4 shows the relative effects of various changes to key parameters used in each of the decision-tree models. 1002

6 Economic analysis of postoperative pain management BJA CWI compared with i.v.-pca The most sensitive data point was the length of stay in hospital. CWI remained the dominant strategy except when the length of stay was reduced by 20% in the i.v.-pca group and increased by 20% in the CWI arm. In Table 3 Cost per controlled patient (base case) at rest. *Control at rest is defined as NVS 3 Total cost of technique (per pt) Arithmetic mean NVS achieved (per pt) Proportion of patients controlled Incremental cost per controlled patient Total cost of technique (per pt) Arithmetic mean NVS achieved (per pt) Proportion of patients controlled Incremental cost per controlled patient CWI I.V. PCA EDA Incremental E6460 E7213 2E % 53.9% 23.5% E7500 Dominant 2E % 4.5% Dominant these both cases, CWI was still more effective but also more costly (Table 4). CWI compared with EDA The key drivers were the proportion of patients controlled for each technique. CWI remained the dominant strategy except when the proportion of patients controlled was investigated with a +20% change and 220% change in the EDA and CWI treatment arms, respectively. In these cases, CWI was less costly but also less clinically effective (Table 4). Probabilistic sensitivity analysis CWI is cost saving 59.2% of the time when compared with i.v.-pca (i.e. the probability for CWI to be cost-effective at a willingness-to-pay of E0 per controlled patient is 59.2%). CWI is cost saving 70.4% of the time compared with EDA. Discussion In this economic evaluation, we aimed to provide physicians with information on costs and outcomes associated with pain management after open major abdominal surgery. The lack of previous comparative clinical evaluation between CWI, EDA, and i.v.-pca justified the development of a decision-tree model. When considering only costs for the implementation of the technique (micro-costing for staff, drug, and devices), CWI was the most costly, mainly Table 4 Results of one-way sensitivity analysis. (The mean NVS score achieved for each technique is based on successful analgesia is provided; therefore no rescue strategy has been implemented. Controlled patient means patients who successfully achieve pain relief defined as NVS 3 at rest) Parameter Base value Low value High value For CWI vs i.v.-pca (at rest) Incremental cost per controlled patient Dominant Length of stay (CWI) % Dominant +20% E993 Length of stay (i.v.-pca) % E % Dominant Rate of successful analgesia (CWI) 65.0% 220% Dominant +20% Dominant Rate of successful analgesia (i.v.-pca) 41.9% 220% Dominant +20% Dominant Cost per hospital day E % Dominant +20% Dominant Technique cost (CWI) E % Dominant +20% Dominant Technique cost (i.v.-pca) E % Dominant +20% Dominant Proportion of patients controlled (CWI) 77.4% 220% Dominant +20% Dominant Proportion of patients controlled (i.v.-pca) 53.9% 220% Dominant +20% Dominant For CWI vs EDA (at rest) Incremental cost per controlled patient Dominant Length of stay (CWI) % Dominant +20% Dominant Length of stay (EDA) % Dominant +20% Dominant Rate of successful analgesia (CWI) 65.0% 220% Dominant +20% Dominant Rate of successful analgesia (EDA) 72.7% 220% Dominant +20% Dominant Cost per hospital day E % Dominant +20% Dominant Technique cost (CWI) E % Dominant +20% Dominant Technique cost (EDA) E % Dominant +20% Dominant Proportion of patients controlled (CWI) 77.4% 220% E % Dominant Proportion of patients controlled (EDA) 72.9% 220% Dominant +20% E

7 BJA Tilleul et al. due to the acquisition cost of the device. However, when taking into account the costs relating to all the healthcare resource consumption associated with a pain treatment strategy, the estimated total costs of CWI are lower than i.v.-pca and EDA. When applying PSA, the dominance of CWI remains 59.2% of the time when compared with i.v.-pca and in 70.4% of the time when compared with EDA. The economic model has been developed in accordance with common recommendations to take into account data coming from different sources, depending on their robustness, and also their availability. 18 The cost data and analgesia measurements were retrieved from an observational study. As stated by Baltussen and colleagues, 18 this observational approach allows the study to relate more closely to actual clinical practice and therefore gives it credibility. For rare events (like side-effects), values should be derived from large multicentre studies, meta-analysis, or both. 18 In our study, it was decided to use this approach in order to improve the robustness and the external validity of these parameters. Sensitivity analysis was performed to capture the uncertainty of the results under different assumptions. 30 It was decided to use data on analgesic efficacy and duration of hospitalization from our observational cohort. Noteworthy, pain values in the three subgroups were consistent with those reported in the literature Rates of failure for EDA procedure inducing switch to another analgesic technique has both economic and clinical consequences. 28 Costs related to the management of sideeffects have also to be included in an economic approach. The duration of hospitalization is a major determinant of the overall cost. Opioid-related side-effects may prolong the duration of hospitalization. 5 Every analgesic technique that may reduce parenteral morphine consumption, such as EDA and CWI, is able to accelerate the course of recovery. CWI has been shown to reduce the duration of hospitalization in comparison with i.v.-pca. 11 Two recent studies in the setting of abdominal surgery led to the same observations The duration of hospitalization in the EDA and i.v.-pca subgroups appeared very similar in our analysis, a finding consistent with data from a comparative meta-analysis. 624 Several other studies have previously investigated the cost associated with pain management. However, comparisons are difficult because there are many confounding factors that may potentially drive the total cost of an analgesic technique. In the current study, the staff costs were obtained from a direct micro-costing measurement performed by nurses caring for the patients. Bartha and colleagues 15 suggest that EDA is not a cost-effective use of healthcare resources in comparison with i.v.-pca, with an ICER of E5625 per pain-free day. These costs including staff, material, and pharmacy were higher than in the present study, mainly because personnel costs were more highly valued and their time frame longer. 15 However, it is noteworthy that the ratio between EDA and i.v.-pca was in the same range of values between our two analyses (approximately one-third for i.v.-pca when compared with EDA). In our analysis, when considering the total average costs of the pain management procedures, CWI is less costly than the two comparators, with savings ranging from E753 to E1040, respectively, for CWI vs i.v.-pca and CWI vs EDA. In conclusion, when compared with i.v.-pca and EDA, the global reduction in resource consumption outweighed the additional cost of using CWI devices, from the perspective of the healthcare provider. When compared with i.v.-pca, CWI appears to be a dominant treatment strategy. Results were less clear when comparing CWI with EDA. CWI was less costly than EDA but with almost equivalent efficacy, offering an alternative for pain management after abdominal surgery. Declaration of interest The grants and honoraria were for the work as follows: P.T. received educational grants or honoraria from Sanofi Aventis, GSK, Ipsen, and Celgene but without any relation to the subject of this study. M.B. received educational grants or honoraria from Astra Zeneca, Baxter, and Gamida, for advice as expert on CWI. Funding This study was supported by a research grant from Baxter SA. The research grant paid for the health economic modelling expertise from the York Health Economics Consortium. Appendix: Assumptions made to build the model Patients who travel down the unsuccessful pain relief pathways are assumed to achieve successful pain relief after the second-treatment strategy has been administered. Resources utilized for the patients who are treated with a number of postoperative analgesia techniques (as a consequence of failing for technical reasons and/or are not benefiting from any pain relief) are summed for each technique. Resources utilized for the patients incur all the resource/ treatment costs of the first-line technique plus the ward element of the resource/treatment costs if any subsequent techniques are administered. In cases when the technique has failed in the intraoperative period, the first technique is only valued for the resource used in the operating theatre. In cases where patients are eligible for CWI or epidural but fail the technique during the intraoperative period, patients are transferred to the i.v.pca group. The rate of adverse events is based on the last analgesia technique administered. In circumstances where the patients receive two techniques simultaneously, i.e. epidural and i.v.-pca, the costs of hospital stay, and NVS outcomes are weighted 50:50%. 1004

8 Economic analysis of postoperative pain management BJA In circumstances where the patients receive two techniques simultaneously (i.e. epidural and i.v.-pca), the rate of adverse events is based on the higher value of the two techniques administered. For example, patients treated with epidural and i.v.-pca have a 36% chance of suffering PONV (based on 18% and 36% rates of PONV for epidural and i.v.-pca, respectively). For the purpose of this analysis, it was conservatively assumed that the adverse event rates are based on the last analgesia the patient received. References 1 Myles PS, Williams DL, Hendrata M, Anderson H, Weeks AM. Patient satisfaction after anaesthesia and surgery: results of a prospective survey of patients. Br J Anaesth 2000; 84: White PF, Kehlet H, Neal JM, Schricker T, Carr DB, Carli F. The role of the anesthesiologist in fast-track surgery: from multimodal analgesia to perioperative medical care. Anesth Analg 2007; 104: Bonnet F, Marret E. Influence of anaesthetic and analgesic techniques on outcome after surgery. Br J Anaesth 2005; 95: Fletcher D, Fermanian C, Mardaye A, Aegerter P. A patient-based national survey on postoperative pain management in France reveals significant achievements and persistent challenges. Pain 2008; 137: Oderda GM, Said Q, Evans RS, et al. Opioid-related adverse drug events in surgical hospitalizations: impact on costs and length of stay. Ann Pharmacother 2007; 41: Marret E, Remy C, Bonnet F. Meta-analysis of epidural analgesia versus parenteral opioid analgesia after colorectal surgery. Br J Surg 2007; 94: Wu C, Cohen S, Richman J, et al. Efficacy of postoperative patientcontrolled and continuous infusion epidural analgesia versus intravenous patient-controlled analgesia with opioids. Anesthesiology 2005; 103: Dolin S, Cashman J. Tolerability of acute postoperative pain management: nausea, vomiting, sedation, pruritus and urinary retention. Evidence from published data. Br J Anaesth 2005; 95: Popping DM, Zahn PK, Van Aken HK, Dasch B, Boche R, Pogatzki-Zahn EM. Effectiveness and safety of postoperative pain management: a survey of consecutive patients between 1998 and 2006 (2nd revision): a database analysis of prospectively raised data. Br J Anaesth 2008; 101: Beaussier M, El Ayoubi H, Schiffer E, et al. Continuous preperitoneal infusion of ropivacaine provides effective analgesia and accelerates recovery after colorectal surgery: a randomized, double-blind, placebo-controlled study. Anesthesiology 2007; 107: Liu S, Richman J, Thirlby R, Wu C. Efficacy of continuous wound catheter delivering local anesthetic for postoperative analgesia: a quantitative and qualitative systematic review of randomized controlled trials. J Am Coll Surg 2006; 203: Liu SS, Wu CL. The effect of analgesic technique on postoperative patient-reported outcomes including analgesia: a systematic review. Anesth Analg 2007; 105: Liu SS, Wu CL. Effect of postoperative analgesia on major postoperative complications: a systematic update of the evidence. Anesth Analg 2007; 104: Schuster M, Standl T. Cost drivers in anesthesia: manpower, technique and other factors. Curr Opin Anaesthesiol 2006; 19: Bartha E, Carlsson P, Kalman S. Evaluation of costs and effects of epidural analgesia and patient-controlled intravenous analgesia after major abdominal surgery. Br J Anaesth 2006; 96: Forastiere E, Sofra M, Giannarelli D, Fabrizi L, Simone G. Effectiveness of continuous wound infusion of 0.5% ropivacaine by On-Q pain relief system for postoperative pain management after open nephrectomy. Br J Anaesth 2008; 101: Drummond M, Manca A, Sculpher M. Increasing the generalizability of economic evaluations: recommendations for the design, analysis, and reporting of studies. Int J Technol Assess Health Care 2005; 21: Baltussen R, Leidl R, Ament A. Real world designs in economic evaluation. Bridging the gap between clinical research and policy-making. Pharmacoeconomics 1999; 16: Liu SS, Allen HW, Olsson GL. Patient-controlled epidural analgesia with bupivacaine and fentanyl on hospital wards: prospective experience with 1030 surgical patients. Anesthesiology 1998; 88: McLeod GA, Dell K, Smith C, Wildsmith JA. Measuring the quality of continuous epidural block for abdominal surgery. Br J Anaesth 2006; 96: Motamed C, Farhat F, Remerand F, Stephanazzi J, Laplanche A, Jayr C. An analysis of postoperative epidural analgesia failure by computed tomography epidurography. Anesth Analg 2006; 103: Popping DM, Elia N, Marret E, Remy C, Tramer MR. Protective effects of epidural analgesia on pulmonary complications after abdominal and thoracic surgery: a meta-analysis. Arch Surg 2008; 143: Walder B, Schafer M, Henzi I, Tramer MR. Efficacy and safety of patient-controlled opioid analgesia for acute postoperative pain. A quantitative systematic review. Acta Anaesthesiol Scand 2001; 45: Werawatganon T, Charuluxanun S. Patient controlled intravenous opioid analgesia versus continuous epidural analgesia for pain after intra-abdominal surgery (review). Cochrane Database Syst Rev 2005; 25: CD Hill RP, Lubarsky DA, Phillips-Bute B, et al. Cost-effectiveness of prophylactic antiemetic therapy with ondansetron, droperidol, or placebo. Anesthesiology 2000; 92: Ochroch EA, Troxel AB, Frogel JK, Farrar JT. The influence of race and socioeconomic factors on patient acceptance of perioperative epidural analgesia. Anesth Analg 2007; 105: de Leon-Casasola O, Parker B, Lema M, Harrison P, Massey J. Postoperative epidural bupivacaine-morphine therapy. Experience with 4227 surgical cancer patients. Anesthesiology 1994; 81: Ready L. Acute pain: lessons learned from patients. Reg Anesth Pain Med 1999; 24: Scott J, Beilby D, Mcclymont C. Postoperative analgesia using epidural infusions of fentanyl with bupivacaine. Anesthesiology 1995; 83: Baltussen RM, Hutubessy RC, Evans DB, Murray CJ. Uncertainty in cost-effectiveness analysis. Probabilistic uncertainty analysis and stochastic league tables. Int J Technol Assess Health Care 2002; 18: