possibility that serial endoscopies to evaluate the progression of disease might be useful in the management of ischemic colitis.

2087

Timing of Prophylactic Surgery for Diverticulitis: A Coct-Effectivaness Analysis Robert J. Richards, Univ of Kansas Medical Ctr, Kansas City, KS; James K. Hammiff, Ctr for Risk Analysis - Harvard Sch of Public Health, Boston, MA

Purpose: Surgery is recommended after 2 or more established attacks of diverticulitis. Our aim was to determine the most favorable time to perform prophylactic surgery for the prevention of diverticulitis recurrence. Methods: A cost-utfectiveness analysis was performed using a Markov model from a societal perspective. Three strategies wore examined: 1) surgery after the 1 ~ attack; 2) surgery after the 2 "d attack; or 3) surgery after the 3 'dattack of diverticulitis. Cost, life expectancy (LE), and quality adjusted life years (OALYs) were calculated for bypnthoti- cal 60-yr-old cohorts. Transition state probabilities and utilities were assigned values using literature-based estimates and expert opinion. Costs were estimated from Medicare reimburse- ment rates, A sensitivity analysis was performed for all clinical events. Costs, life years, and OALYs were discounted at a 3% rate in the baseline analysis. Results: The baseline results are given in the table below. The 3 ~ attack strategy is preferred and is cost saving over the other 2 strategies. The results were not sensitive to changes in the discount rate (0-5%). The model's results were stable for most variables tested with sensitivity analysis. The model was sensitive to the recurrence rate: when the 5-yr risk was increased from 26% (baseline) to 33%, the 1 ~ attack option yielded more QALYs but was not cost-atfectJve ($490,000 per additional OALY saved). We increased the recurrence rate of the 3'd attask relative to the 2 = attack. When the RR was > 2 the 2=aftack strategy was preferred but was not cost-effective vs. the 3'dattack option ($95,000 per additional QALY saved). The model was also sensitive to the risk of diverticular complications and the mortality rate of urgent surgery but the 3 m attack option remained the most cost-effective. Conclusions: Prophylactic surgery after the 3 rd attack is the preferred strategy, yielding higher life expectancy and QALYs, at a lower cost. This strategy remains the most cost-effective with sensitivity analysis.

Baseline results of 60-yr-old cohort using a discount rate of 3%

Strategy Avg. Cost Avg. Life Avg. I n c r s e ~ (1998 U.S.$) Yrs. QAJ.Ys ~ Y

t=Attack 8183 14.154 14.143 Dominated 2~Attack 3621 14.158 14.153 Dominated 3'~Attack 2507 14.164 14.160

2088

Meta.Analysis Analysis Of Medical Therapies For Chronic Anal Seth Hoftman, Northwestern Univ, Chicago, IL; Grigoris I. Leontiadis, G Papenikolaou Hosp, Thessaloniki Greece; Patrick Okolo III, Colin W. Howden, Northweatem gniv, Chicago, IL

Background: Chronic anal fissure (AF) is associated with considerable discomfort and morbid° ity. Surgical management may be complicated by fecal incontinence. Different medical treat- ments have been examined in randomized controlled trials (RCTs). Aims: By meta-analysis of RCTs, to evaluate the efficacy of different medical treatments. Methods: Recuraive literature search for RCTs of medical treatments for chronic AF in adults. Homogeneity assessed according to Breslow-Day method; RCTs pooled and overall healing rates calculated. Derivation of relative and absolute benefit increase (RBI, ABI) and number needed to treat (NNT). Calculation of pooled ManteI-Haenszel odds ratios (OR~) for healing rates on different treatments. Results: 5 RCTs compared nitroglycerin (NTG) ointment with placebo (Pla). Significant heterogeneity among RCTs (P = 0.001). Pooled healing rates were 56.8% for NTG and 31.0% for Pla; RBI = 82%; ASI = 25.5% (95% CI = 14.6%, 36.4%); NNT = 3.9 (95% CI = 2.7, 6.8); OR~ = 2.8 (95% CI = 1.7, 4.5). 2 RCTs compared BoTox wh'ti Pla; significant heterogeneity between them (P < 0.02). Pooled healing rates were 51.4% for BoTox and 24.3% for Pla; RBI = 111%; ABI = 27% (95% Cl = 5.8%, 48.3%); NNT = 3.7 (95% CI = 2.1, 17.2); OR~., = 3.0 (95% Cl = 1.1, 8.0). 1 RCT compared BoTox with NTG; healing rates were, respectively, 96% and 60%; RBI = 60%; A81 = 36% (95% Cl = 15.3%, 56.7%); NNT = 2.8 (95% CI = 1.8, 6,5); OR = 16.0 (95% CI = 1.9, 138.0). Conclusions: Treatment of chronic AF with NTG or BoTox is superior to placebo. BoTox was superior to NTG in one RCT. The cost-utilty of 8oTox in the treatment of chronic AF should be formally evaluated.

2089

Long-Term Hospitalization Outcomes and Economic Impact of GasMu r d ~ l Stimulation (GES) in Patients with Drug Refractory Gaatroparesls Thomas L Abell, Univ of Arkansas for Medical Science, Little Rock, AR; Michael Hocking', Univ of Florida, Gainesville, FL; Richard McCallum, Univ of Kansas, Kansas City, KS; Kenneth Koch, Penn State Univ, Hershey, PA; Thomas Nowak, St Vincent's Hosp, Indianapolis, IN; John K. Dibaise, Univ of Nebraska Medical Ctr, Omaha, NE

Introduction: The objective of the Worldwide Anti-Vomiting Electrical Stimulation Study was to evaluate the effectiveness of gastric electrical stimulation (GES) in the treatment of drug refractory gastroparesis. Reported here are hospitalization outcomes for patients enrolled in US centers and followed for one year. Methods: Patients with chronic drug refractory gastrupar- esis were enrolled [11 diabetic (5 M, 6 F) and 16 idiopathic (16 F)]; mean overall age 39.7 yrs.). All patients received an implanted GES system via laparotomy. Entry criteria and electrical stimulation parameters have been reported previously (Gastruenterol 118(4):A393)~ Hospital days for the year before surgery (H-Prior) was determined from discharge summaries and hospital stay for implant surgery (H-Surg) was noted; cumulative hospital days in the year after implant surgery discharge (H-Post) was determined from adverse events reporting. Results: Average H-Surg was 5.7 days. During the 12 month follow up period, 3 patients

were terminated (1 due to pregnancy @ 10 rag., 1 lost to follow up after 6 rag., 1 patient died of unrelated causes @ 9 rag.) from the study and are excluded from this analysis. 4 patients had the device removed [3 due to infection (2 @ 11 mo., 1 @ 9 rag.), 1 due to lead perforation of the stomach wall @ 3 mo.] and are included. Mean H-Prior was 49 compared to 20 for H-Post (n=24). H-Post for the 4 patients whose implants were removed was 27 days. The major reasons for hospitalization after implant surgery included recurrent symptoms, dehydration, infections and feeding tube complications. Total days for the year after implant surgery (H-Surg + H-Post) declined by 15.4 days compared to H-Prior. Assuming an average dally hospital cost of $1500, the average savings in the first year is approximately $23,000 per patient. Conclusions: We conclude that long-term GES can result in a meaningful reduction in hospital utilization in the first year after implant for this medically challenging patient group.

Effoiogy t ~ Days: H-Prior H-Su~ H-Post

70.5 6.4 41.1 MlqlalM¢ 40.9 5.5 27.6 All ~ 48.8 5.7 27.8

Coat Amlyds of F.Noesopi¢ Ultrmmud in the Evaluation of Pancreatic Admmmmdam Gavin C. Harewood, Maurits J. Wiersema, Mayo Clin, Rochester, MN

Background: Endoscopic ultrasound (EUS) guided fine needle aspiration biopsy (FNA) of non- peritumoral (NPT) lymph nodes (LN) can be helpful in preoperative staging of pancreatic a d e ~ n o m a . The economic impact of this staging strategy has not yet been described. Aim: To apply a decision analysis model to compare the costs of 3 approaches to the management of non-metastatic pancreatic edenocarcinoma: EUS FNA vs computerized tomng- raphy (CT)-Ouided FNA vs surgery. A cost-minimization approach was employed, as viewed from the perspective of the payer. Methods: A decision analysis model with 3 management arms was designed using DATA Version 3.5 TM, taking entry criteria as "resectable" pancreatic adenocarcinoma (no vascular invasion or distant metastases) as determined by helical CT. Detectk~ of metastatic NPT LN by FNA signified unresectability and prompted palliative biliary stenting rather than surgery in jaundiced patients. Baseline probabilities, obtained from the published literature, were varied through plausible ranges using sensitivity analysis (table). Prevalence of NPT LN involvement (20%) was based on our data from resected specimens. Cost inputs were based on Medicare professional plus facility fees. The endpoint was cost of management per patient. Results: Use of EUS FNA avoided 12 unnecessary surgeries/lOg patients evaluated; CT FNA avoided 6/100 patients. Sensitivity analysis revealed that EUS FNA remained the least costly option provided the frequency of NPT LN involvement was >2%; below this value, surgery became the most economical strategy. Conclusion: EUS FNA is the least costly staging strategy in the work-up of patients with non-mataststic pancreatic adenocarcinoma primarily due to confirmation of NPT LN involvement avoiding unnecessary surgery. These results support performing EUS in patients thought to be resectable on helical CT in order to enhance NPT LN assessment.

B a ~ Prob~iUes of Model and Final Costs of Management per pa'dent:

Vadable EUS CT FNA FNA Sims ~ a

Baseline 80% 40% Range 60-95 20-70

NPT Surg Strategy LN cx rate

prey 20% 15% EUS FNA 1-40 2-30 CT FNA

Surgery

Sons = sens~vity; prev = prevalence; surg = surgical; cx = complication

Cost

$21,718 $23,034 $23,543

Cost Analysis of Endoscopic Ultrasound in the Evaluation of Esophageal Cancer Gavin C. Harawood, Maurits J. Wiersema, Mayo Clin, Rochester, MN

Background: The use of endoscopic ultrasound (EUS) with guided fine needle aspiration biopsy (FNA) of suspicious lymph nodes has become an important aid in the staging of esophageal carcinoma, The economic impact of this staging strategy has not yet been described. Aim: To apply a decision analysis model to compare the costs of ELlS FNA vs computerized tomography (CT)-guided FNA vs surgical resection (thoracotomy) (TRC) in the management of esophageal carcinoma. A cost-minimization approach was employed, as viewed from the perspective of the payer. Methods: A decision analysis model with 3 management arms was designed using DATA Version 3.5", taking entry criteria as esophageal carcinoma without evidence of distant metastases as determined by CT. Detection of tumor on celiac lymph node (CLN) FNA signified unresectability and prompted palliative endoscopic esophageal stenting rather than surgery. Baseline probabilities, obtained from published literature, were varied through plausible ranges using sensitivity analysis (table). Cost inputs were based on Medicare professional fees plus Medicare facility fees. The endpoint was the cost of manage- ment per patient. Results: Use of EUS FNA avoided 18 unnecessary TRC/IO0 patients; CT- FNA avoided 8 unnecessary TRC/IO0 patients. Sensitivity analysis revealed that EUS FNA remained the least costly option provided the prevalence of CLN involvement was >5%; below this value, TRC became the most economical strategy (table). Conclusion: By minimizing unnecessary surgery, primarily by detecting CLN involvement, EUS FNA is the least costly staging strategy in the work-up of patients with non-metastatic esophageal cancer.

A - 4 0 8