A Trial of the Cost Effectiveness of IV Acetaminophen in Bariatric Surgery

Overview

Bariatric surgery is the only proven and effective long term treatment for morbid obesity. In an attempt to reduce patients' post-operative hospital stay, lower associated health care costs, and improve satisfaction scores, St Luke's University Health Network (SLUHN) recently adopted a fast track bariatric surgery (FTBS) protocol. Findings to date show that FTBS is safe and effective when performed in a Center of Excellence (COE) such as ours. However, post-operative pain control remains a challenging issue, with only intravenous (IV) or PO (by mouth) narcotics appropriate for bariatric surgery patients. IV acetaminophen, which has been used successfully in Europe, was recently approved by the Federal Drug Administration (FDA) for use in the US. However, no data exist regarding the use of IV acetaminophen in bariatric surgery patients, nor are there any data assessing its cost effectiveness. Therefore, our study will investigate the economic impact of administering IV acetaminophen to bariatric surgery patients, as well as its effect on clinical outcomes such as patients' post-operative length of stay, self-reported pain, readmissions, emergency room (ER) visits and complications. The study design will be a randomized, double-blind, parallel-group, controlled trial in a single bariatric Center of Excellence (COE) that is part of the St. Luke's University Health Network (SLUHN). Patients will consist of up to 200 morbidly obese adult bariatric surgery candidates \> 18 years of age undergoing either laparoscopic Roux-en-Y Gastric Bypass (LRYGB) or laparoscopic Sleeve Gastrectomy (LSG). Group 1 (treatment) will receive IV acetaminophen plus IV narcotics for the first 6 hours post-surgery followed by IV/ PO narcotics for the remaining 18 hours. Group 2 (control) will receive IV normal saline plus IV narcotics for the first 6 hours post-surgery followed by IV/PO narcotics for the remaining 18 hours. Data analysis will include quantile regression, mixed randomized-repeated analysis of covariance (ANCOVA) and selected univariate comparisons, with p \< .05 denoting statistical significance for all outcomes.

This was a double-blind, prospective, randomized controlled trial conducted at a single accredited bariatric center that is part of a multi-campus university teaching hospital. The Institutional Review Board approved the study. We hypothesized that administering IV acetaminophen plus IV narcotics/PO narcotics to bariatric patients will produce significant cost savings to our hospital network, as well as result in lower self-reported pain from baseline to 24 hours post-surgery; a decrease in hospital length of stay (LOS) to \< 24 hours; and higher overall patient satisfaction compared to patients receiving IV narcotics plus normal saline as placebo. Based on the smallest clinically meaningful difference in mean pain scores of \< 3 for the IV Acetaminophen group versus \> 5 for the normal saline group (representing a minimum reduction of 40%), 24 patients are required to achieve 92% power at α = .05. To ensure an adequate number of complete data points, a decision was made to recruit a total of 100 patients. Inclusion and exclusion criteria are presented in table 1. Patients were randomly assigned to either treatment or control groups in a 1:1 ratio using permuted block sizes of 8. To preserve blinding of group assignment among the surgeons, patients, nurses and other team members, randomization numbers generated by the statistician were printed out and placed on the medication infusion bags by the pharmacist. Group 1 (treatment) received IV acetaminophen (1g in 100 ml of 0.9% normal saline IV Q 6hrs for 24 hours) plus IV narcotics (2-4 mg IV Q2hrs PRN) 30 minutes before surgery, followed by medication plus IV narcotics (2-4 mg IV Q2hrs PRN)/ PO narcotics (Oxycodone 5-10 ml PO Q4 hrs PRN) for the remaining 18 hours. Group 2 (control) received IV normal saline (100 ml of 0.9% normal saline IV Q 6hrs for 24 hours) plus IV narcotics (2-4 mg IV Q2hrs PRN) 30 minutes before surgery, followed by saline plus IV narcotics (2-4 mg IV Q2hrs PRN)/ PO narcotics (Oxycodone 5-10 ml PO Q4 hrs PRN) for the remaining 18 hours. Pain was formally assessed every 2 hours for the first 24 hours post-surgery using a 10-point ordinal scale, with 0 representing no pain and 10 representing the worst pain. Rescue narcotics were given as needed depending on whether self-reported pain is moderate (4-6 on 10-point pain scale) or severe (\> 7 on 10-point pain scale). In the first 6 hours, patients were only offered IV narcotics. After the first 6 hours, patients were offered either IV or PO narcotics as deemed appropriate by the administering nurse based on the clinical condition and ability to tolerate PO intake. For moderate pain, patients received either IV morphine 2 mg IV Q2hrs PRN or PO Oxycodone 5 ml PO Q4hrs PRN. For severe pain, patients received either IV morphine 4 mg IV Q2hrs PRN or PO Oxycodone 10 ml PO Q4hrs PRN. Total hospital network costs for all relevant services were represented in dollar amounts as a continuous variable. Total costs included both direct and indirect costs. Direct costs were based on postoperative length of stay and the cost associated with acetaminophen and narcotic administration. Indirect costs included the cost of ER visits secondary to postoperative pain and also the performance of any radiographic studies postoperatively to evaluate for abdominal pain as deemed clinically necessary by the attending surgeon. Hospital length of stay (LOS) was dichotomized as \< 24 hours and \> 24 hours, given the expected relative infrequency of patients requiring a stay \> 24 hours. Patients' self-reported satisfaction was assessed using a 5-point Likert scale (from strongly disagree to strongly agree) at 10 days to 2 weeks post-surgery (see Figure 1). Descriptive outcomes are reported as means and standard deviations for normally distributed continuous variables; medians and interquartile ranges for skewed continuous variables; and frequencies and percentages for categorical variables Primary outcomes included the following variables • Total hospital network direct costs (including direct cost per adjusted patient day and costs for IV acetaminophen, morphine and narcotics administration): The original data analysis plan called for quantile regression with the covariates of treatment group (IV acetaminophen plus narcotics versus IV narcotics plus normal saline) and surgery type (LRYGB versus LSG), based on the assumption that the direct cost data would be skewed with unequal variance across the distribution of sample group outcomes. However, preliminary inspection of the data revealed that basic assumptions of normality of sampling distributions, homogeneity of variance and absence of significantly influential outliers were met. Therefore, we conducted factorial randomized-groups analysis of variance (ANOVA) with the between-group factors of treatment group and surgery type, as well as the treatment group \* surgery type interactional effect. * Mean self-reported pain scores: mixed randomized-repeated measures analysis of covariance (ANCOVA) with the randomized factors of treatment group (IV acetaminophen plus narcotics versus IV narcotics plus normal saline) and surgery type (LRYGB versus LSG); the repeated measures factor of pain scores over time (i.e., 2-hours intervals from 2 to 22 hours); and the first post-operative pain scale assessment as the covariate. Pain scores were measured on a numeric scale, with 0 representing no pain and 10 representing the worst imaginable pain. Prior to data analysis, the assumptions of normality of sampling distributions, homogeneity of variance, presence of significantly influential outliers, linearity between the baseline pain score covariate and subsequent pain scores, and homogeneity of regression for the covariate-outcomes slopes were evaluated. These assumptions were met with the exception of moderate heterogeneity of variance for the IV acetaminophen-LSG group for 10-hour pain scores and largely curvilinear relationships between the covariate and subsequent pain scores. Therefore, the covariate was logarithmically transformed, and a more conservative p-value \< .025 was used to assess pain scores over time. In order to obtain a clearer understanding of the repeated measures main effect of pain scores over time, an additional analysis of variance excluding the covariate of baseline pain scores was also conducted. * Length of stay (\< 24 hours versus \> 24 hours): Cochran-Mantel-Haenszel test for treatment group with surgery type as the stratification variable. * Patient satisfaction: a patient satisfaction survey was given to all the patients in our study to fill during their first postoperative visit between Post-Operative Day (POD) #7 and POD# 10. The questionnaires are filled by the patients and collected by our study coordinator for analysis before the patient is examined by the surgeon to avoid any bias in reporting satisfaction scores. Separate Mann Whitney rank sums tests by treatment group and surgery type * Patient rating of overall pain: pain overall score (0-10) was also collected during the first postoperative visit between POD#7 and POD#10 factorial randomized-groups ANOVA by treatment group, surgery type and treatment group\*surgery type interaction