A prospective randomized comparison of laparoscopic appendectomy with open appendectomy: Clinical and economic analyses

Surgical outcomes research A prospective randomized comparison of laparoscopic appendectomy with open appendectomy: Clinical and economic analyses Kirsten Hall Long, PhD, Michael P. Bannon, MD, Scott P. Zietlow, MD, Eva R. Helgeson, BA, William S. Harmsen, MS, C. Daniel Smith, MD, Duane M. Ilstrup, MS, Yvonne Baerga-Varela, MD, Michael G. Sarr, MD, and the Laparoscopic Appendectomy Interest Group, Rochester, Minn

Background. Previous randomized studies of laparoscopic appendectomy produced conflicting recommendations, and the adequacy of sample sizes is generally unknown. We compared clinical and economic outcomes after laparoscopic and open appendectomy in a sample of predetermined statistical power. Methods. A pre-study power analysis suggested that 200 randomized patients would yield 80% power to show a mean decrease of 1.3 days’ hospitalization. One hundred ninety-eight patients with a preoperative diagnosis of acute appendicitis were randomized prospectively to laparoscopic or open appendectomy. Economic analysis included billed charges, total costs, direct costs, and indirect costs associated with treatment. Results. Laparoscopic appendectomy took longer to perform than open appendectomy (median, 107 vs 91 minutes; P < .01) and was associated with fewer days to return to a general diet (mean, 1.6 versus 2.3 days; P < .01), a shorter duration of parenteral analgesia (mean, 1.6 versus 2.2 days; P < .01), fewer morphine-equivalent milligrams of parenteral narcotic (median, 14 mg versus 34 mg; P = .001), a shorter postoperative hospital stay (mean, 2.6 versus 3.4 days; P <.01), and earlier return to full activity (median, 14 versus 21 days; P < .02). However, operative morbidity and time to return to work were comparable. Billed charges and direct costs were not significantly different in the 2 groups ($7711 versus $7146 and $5357 versus $4945, respectively), but total costs (including indirect costs) of laparoscopic appendectomy were, on average, nearly $2400 less, given the shorter length of stay and abbreviated recuperative period ($11,577 versus $13,965). Subgroup analyses suggested the benefit of a laparoscopic approach for uncomplicated appendicitis and for patients with active lifestyles. Conclusions.While laparoscopic appendectomy is associated with statistically significant but clinically questionable advantages over open appendectomy, a laparoscopic approach is relatively less expensive. The estimated difference in total costs of treatment (direct and indirect costs) was at least $2000 in more than 60% of the bootstrapped iterations. The economic significance and implications favoring a laparoscopic approach cannot be ignored.(Surgery 2001;129:390-400.) From the Departments of Surgery and Health Sciences Research, Mayo Clinic and Mayo Foundation, Rochester, Minn Presented in part at the Society for Surgery of the Alimentary Tract Meeting, May 11-14, 1997, Washington, DC, and published in abstract form in Gastroenterology (Supplement) 112(4):A1429, April 1997. The economic analysis was presented at the 16th Annual Meeting of the International Society of Technology Assessment in Health Care on June 21, 2000, The Hague, Netherlands. Accepted for publication December 27, 2000. Reprint requests: Michael P. Bannon, MD, Mayo Clinic, Department of Surgery, 200 First St SW, Rochester, MN 55905. Copyright © 2001 by Mosby, Inc. 0039-6060/2001/$35.00 + 0 11/60/114216 doi:10.1067/msy.2001.114216

390 SURGERY

ENCOURAGED BY THE SUCCESS of laparoscopic cholecystectomy, general surgeons are now defining other applications for minimal access surgery. Appendectomy, essentially an extirpative procedure of a small diverticulum, lends itself to a minimal access approach. Laparoscopic appendectomy has emerged as a safe procedure,1-4 and its potential advantages have led some authors to advocate this approach as the procedure of choice for uncomplicated appendicitis.5,6 However, these recommendations were made on the basis of retrospective data. Most later prospective randomized studies were inherently flawed because they failed

Surgery Volume 129, Number 4

to determine sample size with a pre-study power analysis.7 Economic analyses that have been conducted comparing laparoscopic and open appendectomy are additionally flawed given small sample sizes, inappropriate statistical testing in the presence of skewed cost data, the use of billed charges versus economic cost data, and the complete omission of important indirect costs associated with treatment.8-13 No economic analysis to date, in fact, meets the widely accepted criteria set forth by the U.S. Panel in Cost Effectiveness in Health and Medicine as constituting a valid study.14 The indications for laparoscopic appendectomy in general surgical practice remain ill-defined, and the minimal morbidity and excellent results of open appendectomy are well-established.15-19 This prospective randomized study was undertaken without preconceived biases to explore potential advantages of laparoscopic appendectomy for patients presenting to an emergency general surgical practice with a preoperative diagnosis of acute appendicitis. Our primary aim was to compare laparoscopic and open appendectomy in a sample of predetermined, statistically derived size to address clinically important issues involving hospital course, morbidity, time until return to work, and reliable economic outcomes based on total charges and estimated total costs. Only 1 other randomized study of laparoscopic appendectomy has been based on a statistically sound power analysis20; however, because patients without appendicitis were excluded after randomization, the results of this study do not directly reflect the application of laparoscopic appendectomy in clinical practice. PATIENTS AND METHODS Sample size determination. Sample size was determined before beginning the study by power analysis. A pre-study mean hospital length of stay for open appendectomy (January 1990 through September 1990, n = 125 patients) was 4.2 days with an SD of 2.9 days. Assuming normally distributed data and a 2-tailed, 2-sample t test if 100 patients each were randomized to the open and laparoscopic groups, there should be greater than 80% power to detect a 1.3 day or greater decrease in mean hospital stay associated with laparoscopic appendectomy. This difference in hospital stay is considered to be between a small and medium effect size.21 Patient selection and randomization. Beginning in July 1992, Emergency Department patients with a diagnosis of acute appendicitis were randomized between laparoscopic and open appendectomy. Randomization by a random numbers table was

Long et al 391

performed in blocks of 4 patients stratified by surgeon. Assignments were made by means of sealed sequenced envelopes. Patients with complicated appendicitis presenting as a palpable right lower quadrant mass, known phlegmon or abscess, diffuse peritonitis, or septic shock were not considered for randomization. Additional exclusion criteria were pregnancy, age younger than 15 years, and onset of pain after hospitalization. This clinical study was approved by the Institutional Review Board of the Mayo Foundation on June 19, 1992, with additional approval on April 6, 1999, for the retrospective chart review and access to billing data for economic analysis. Consenting patients were informed of randomization assignment before operation; no patient subsequently withdrew. During the study period, laparoscopic appendectomy was not offered for acute appendicitis except through randomization. Antibiotics. Preoperatively, patients received cefoxitin (1 g) intravenously; those allergic to penicillin were given clindamycin (600 mg) and aztreonam (1 g). Postoperative antibiotics were not administered to those patients with a healthy appendix, simple appendicitis, or suppurative appendicitis, but they were continued for those with gangrenous or perforative appendicitis until the patient’s temperature was less than 38°C for 24 hours and the white blood cell count was less than 10,000 cells/mm3. Appendectomy techniques. Laparoscopic appendectomy was performed after nasogastric and bladder catheterization. Techniques of laparoscopic cannulation and appendiceal mobilization and resection were not standardized. All laparoscopic wounds were closed primarily. Conversion of laparoscopic to open appendectomy was left to the discretion of the operating surgeon. Open appendectomy was performed through a muscle-splitting right lower quadrant celiotomy. Appendiceal mobilization and stump closure were not standardized; nasogastric and urinary catheters were not routinely placed. All wounds were closed primarily. In patients with gangrenous or perforative appendicitis, a subcutaneous catheter was irrigated postoperatively 3 times daily with 10 mL antibiotic solution (containing 0.5 g neomycin sulfate, 0.1 g of polymyxin B sulfate, and 80 mg of gentamicin sulfate per liter of 0.9% sodium chloride) and removed on the fourth postoperative day. All appendices found to be clinically noninflamed were resected except in 1 patient in the open appendectomy group who underwent sigmoidectomy for perforated diverticulitis after negative appendiceal exploration. All procedures were

392 Long et al

performed within the context of a surgical residency training program. Thirteen faculty surgeons participated in the study; all were proficient with laparoscopic cholecystectomy and had well-established laparoscopic practices. The 2 primary surgeons (M. P. B. and S. P. Z.) completed a pre-study series of 22 laparoscopic appendectomies and supervised 47% of the study appendectomies. Postoperative care. Morphine sulfate was administered intravenously by a patient-controlled infusion pump; for patients with a morphine allergy or intolerance, meperidine hydrochloride was used. As soon as patients began to drink liquids, they were offered propoxyphene and acetaminophen orally; patients intolerant of this were offered acetaminophen with codeine. Acetaminophen alone was offered for mild pain. Orally administered analgesics were continued in preference to parenterally administered analgesics if they were efficacious. A general diet was given as soon as the patient’s appetite returned and was considered tolerated if no nausea, emesis, abdominal distention, or abdominal cramping occurred. Patients were dismissed when pain was controlled with orally administered analgesics and a general diet was tolerated. Both total hospital stay (interval in days between presentation at the Emergency Department and dismissal) and postoperative hospital stay (interval between operation and dismissal) were tabulated. Throughout the study, patients undergoing laparoscopic appendectomy were instructed to resume usual activities immediately after operation. Patients undergoing open appendectomy after February 1, 1994, were given identical instructions; however, before February 1, 1994, these patients were given a 6-week disability from strenuous activity. Because of this initial policy, only patients entered into the study after February 1, 1994 (n = 43 laparoscopic group and n = 47 open group), were evaluated for return to work or school and return to full activity end points. Follow-up. Outpatient postoperative evaluation was performed 2 weeks after dismissal. Patients were contacted by telephone at the end of postoperative weeks 1, 2, and 3 and at the end of postoperative months 1, 3, and 12. Patients who were lost to telephone follow-up were sent a questionnaire 1 year postoperatively. All telephone and written surveys addressed resolution of symptoms, return to work or school, return to full activity, symptoms of wound infection and hernia, potential related reoperations, mortality, and satisfaction with operative procedure. Complete follow-up was obtained for 195 patients (98.5% [100% open group, 96.8% laparoscopic group]).

Surgery April 2001

Clinical data collection. Preoperative and operative data were collected prospectively by the surgical team. Two registered nurse registrars performed postoperative data collection and follow-up surveys. All data were entered into a SAS program for collation and analysis. Economic analysis. Patient medical resource utilization for each treatment modality was tracked to assess the direct and indirect costs associated with intervention. This economic analysis was facilitated by the use of the Olmsted County Healthcare Expenditure and Utilization Database (OCHEUD), a costing system developed by researchers at the Mayo Clinic. OCHEUD provides a standardized inflation-adjusted estimate of the cost of each service or procedure in 1995 constant dollars. Specifically, using a micro-costing approach, OCHEUD groups resource utilization into the Medicare Part A and B classification system; Part A billed charges are adjusted by using hospital costto-charge ratios, and Part B physician services are adjusted by using 1995 Medicare reimbursement rates. The use of standardized unit costs is desirable because of the well-known discrepancies between billed charges, which are directly available in the utilization database, and true resource use—the “opportunity” costs in health care.14,23 OCHEUD provides an estimated economic cost for each line item in the billing record. Patient-specific cost data such as this clearly provide the most accurate picture of medical resource utilization for technology assessment purposes. An economic analysis was conducted from the societal perspective to compare the billed charges, total costs, direct costs, and indirect costs associated with treatment. The total costs associated with treatment included direct hospital costs (physician services, room and board, supplies, operating room expenses, inpatient medications, etc) and the indirect costs associated with lost productivity during the hospitalization and subsequent recuperative period. Indirect costs were valued by using standard methods based on sex- and age-specific average hourly wage rates from the Bureau of Labor Statistics Current Population Survey.24 To accurately reflect the costs associated with laparoscopic appendectomy in clinical practice, the analysis was based on intention to treat. However, a “pure” analysis was also conducted for comparison purposes, excluding those patients from the analysis who were initially randomized to laparoscopic appendectomy but ended up having the open procedure. Given the desire to capture disease-specific costs, we imputed costs for those

Long et al 393

Surgery Volume 129, Number 4

patients who had additional procedures performed during the appendectomy hospitalization. While it is typical for cost data to be highly skewed, we checked data outliers and line item charges in detail for possible billing errors by using the medical record as verification of resource utilization. Statistical analysis Clinical end points. Quantitative variables proved to have nonGaussian distributions and thus were compared with the Wilcoxon rank sum test. The central tendencies of these variables are described with medians unless otherwise specified. Qualitative variables were compared with a chi-square or Fisher exact test. All analyses were based on the intention to treat; thus, patients whose procedures were converted from laparoscopic to open were analyzed as being in the laparoscopic group. Economic end points. Although median values provide important descriptive information, mean costs per patient are the appropriate statistics from an economic and budgetary perspective.25 We, therefore, compared mean total costs, direct costs, indirect costs, and total billed charges using standard t tests. The robustness of these standard tests of significance in the presence of skewed data and the variability in estimated costs were determined through the use of nonparametric bootstrapping techniques.26,27 Multiple linear regression methods were also performed to control for possible confounding factors on estimated costs including age, sex, type of appendicitis (uncomplicated or complicated), and activity level. RESULTS Between July 18, 1992, and August 9, 1995, the necessary 200 patients were randomized. However, 2 randomized patients did not meet entry criteria and were eliminated from the study before operation. One hundred ninety-eight patients were evaluable: 93 randomized to laparoscopic appendectomy and 105 to open appendectomy. Clinical characteristics, including duration of symptoms, white blood cell count, temperature, time from emergency department to operating room, age, or sex did not differ aside from a borderline difference in body mass index (BMI) (open versus laparoscopic, 24.9 versus 26.1; P < .05). Similarly, occupations (student or office worker, laborer, no occupation or unemployed, and retired) and activity levels (based on the presence or absence of regular exercise or manual labor) were evenly distributed (data not shown). Clinical analyses. The histopathologic diagnosis was acute appendicitis in 78 patients (84%) and 88 patients (84%) in the laparoscopic and

Table I. Reasons for conversion of laparoscopic to open appendectomy Reason Perforative appendicitis Iatrogenic cecal perforation Other appendiceal characteristics Appendiceal diameter too wide for laparoscopic closure Unable to mobilize appendix safely Celiotomy performed for oophorectomy (ruptured cyst) Technical difficulties with laparoscopic equipment Randomization violation Total

Patients (No.) 8 1 1 2 1 1 1 15

open groups, respectively. There were no differences in the number of patients with simple (44% versus 47%), suppurative (28% versus 26%), perforative (10% versus 10%), or gangrenous (3% versus 4%) appendicitis (P = .994, chi-square test). The final diagnoses for the patients with a histologically normal appendix and the sex distribution of patients with indeterminate right lower quadrant pain also did not differ. Procedures other than appendectomy were performed in 2 patients (salpingo-oophorectomy and femoral hernia repair) in the laparoscopic group and 8 patients (3 right hemicolectomies, 2 ovarian cystectomies, sigmoidectomy, ileocecectomy, and resection of infarcted omentum) in the open group (P = .11). Despite the slight preponderance of nonappendiceal procedures in the open group, there was no evidence of missed pathologic conditions in the laparoscopic group. Laparoscopic appendectomy was converted to open appendectomy in 15 patients (16%) (Table I). Successful laparoscopic appendectomies were performed with 3 cannulas in 14 patients (18%), 4 cannulas in 63 (81%), and 5 cannulas in 1 (1%). No procedure was converted to open appendectomy for bleeding. The appendiceal stump was closed with a linear stapling device in 74 patients (80%) and with loop ligatures in 5 (5%). In no patient did the appendiceal stump closure dehisce intraoperatively or postoperatively. The appendix was placed into a laparoscopic bag before removal from the peritoneal cavity in 57 patients (73%), withdrawn through a laparoscopic cannula in 20 (26%), and removed unprotected in 1 (1%). Clinical end points. A comparison of clinical end points is summarized in Table II. The median duration of the operation was greater for laparoscopic appendectomy (107 versus 91 minutes;

394 Long et al

Surgery April 2001

Table II. Comparison of major clinical end points in all patients assigned by intention-to-treat Laparoscopic (n = 93) End point Duration of operation (minutes) Days to tolerance of general diet Days parenteral analgesia Milligrams morphine-equivalent parenteral analgesia Days oral analgesia Total hospital stay (days) Postoperative hospital stay (days) Wound infection (n) Days to work or school* Days to full activity†

Open (n = 105)

Median

Mean

SD

Median

Mean

SD

P value

107 1 1 14

110.5 1.6 1.6 33.3

34.6 2.5 1.8 56.1

91 1 2 34

95.8 2.3 2.2 53.5

33 2.3 1.6 63.6

< .001 .002 < .001 < .001

2 2 2 17 9 14

2.1 2.9 2.6

1.9 3.3 3.3

2.5 3.5 3.4

2.6 3.3 3.3

14.6 18.9

11.1 16

2 2 2 17 12 21

14.6 24.2

9.8 14.2

.324 .057 .004 .697 .716 .015

*n = 35 laparoscopic group, n = 39 open group. †n = 38 laparoscopic group, n = 47 open group.

P < .001, rank-sum test). When stratified by a BMI of ≥28 or <28, the duration of the operation was longer for a BMI <28 in the open group but not different between groups for a BMI > 28. Although the distribution of time until a general diet was tolerated was shorter in the laparoscopic group (mean, 1.6 versus 2.3 days, P < .01), the median duration was 1 day in each group, emphasizing the small magnitude of the difference. In the laparoscopic group, 66 patients (71%) versus 55 (52%) in the open group tolerated a general diet the day of or the day after operation (P < .01, chi-square test). In the laparoscopic group, 20 patients (22%) versus 3 patients (3%) in the open group did not require parenteral analgesia (P < .0001, chi-square test), and the median number of days of parenteral analgesia was smaller for the laparoscopic group: 1 versus 2 days (P < .001). However, of those who required parenteral analgesics, the median duration was 2 days in each group (P = .10). When converted to equivalent dosages of morphine by using the conversion 7.5 mg meperidine per milligram morphine, the median total morphine-equivalent dose was less for the laparoscopic group (14 mg versus 34 mg, P < .01). Considering only those patients who received parenteral narcotics, morphine equivalent doses were 24 mg and 35 mg, respectively (P = .02). The percentage of patients who did not use oral analgesics was greater in the laparoscopic group: 22% versus 9% (P = .01, chi-square test), but the median number of days that oral analgesics were used was 2 in both groups (P = .32). Three patients in the laparoscopic group never required analgesics; all in the open group required analgesics (P = .10, Fisher exact test). The percentage of patients requiring postoperative antiemetic medications did not differ (32% in each group); the median duration was 1 day for both groups.

Wound infections, defined as suppuration or clinically significant erythema or cellulitis, occurred in 17 patients in each group (18% and 16% in the laparoscopic and open groups, respectively; P = .70). Eight of these infections (24%) were diagnosed after hospital dismissal. After laparoscopic appendectomy, wound infections occurred more often in the larger cannula sites (umbilical or lower midline) than in the 5-mm cannula sites (right upper and left lower quadrants). When stratified by type of appendectomy and BMI (BMI < 28 or ≥ 28), a BMI ≥ 28 was associated with an increased incidence of wound infection in both groups. This association was significant in the laparoscopic group, P < .001 (35.1% versus 5.6%), but did not reach statistical significance in the open appendectomies, P = .24 (24.1% versus 13.2%). An intra-abdominal abscess developed in 4 patients (4%) in the laparoscopic group and in 1 (1%) in the open group (P = .19). These abscesses complicated simple appendicitis (n = 2) and suppurative appendicitis (n = 2) in the laparoscopic group and perforative appendicitis in the open group. Laparoscopic complications occurred in 3 patients (3%), including intra-abdominal hematoma, abdominal wall hematoma, and iatrogenic cecal perforation (1 each). The hematomas resolved with conservative management, but the cecal perforation required immediate conversion to open appendectomy. There was no further complication. One additional patient in the laparoscopic group whose procedure was converted to open appendectomy had dehiscence of the right lower quadrant celiotomy wound and required reoperative closure. A deep wound infection with consequent ventral hernia was the only technical complication in the open group (P = .19).

Long et al 395

Surgery Volume 129, Number 4

Postoperative fever (temperature > 37.5°C), ileus, urinary retention, urinary tract infection, and other postoperative complications did not differ between groups. There were no postoperative hemorrhages or pneumonia. Overall, the number of complications did not differ (Table III). Reoperations and related operations were few and did not differ between groups (data not shown). Although the distribution of total hospital stay did not differ significantly between groups (P = .06), the distribution of postoperative hospital stay was less for the laparoscopic group (P < .01). However, median total hospital stay and median postoperative stay were 2 days for both groups, the estimated means of total and postoperative stay were lower (2.9 versus 3.5 days and 2.6 versus 3.4 days, respectively). The median time until return to work or school did not differ (9 versus 12 days, P = .72), but the median time to full activity was less for the laparoscopic group (14 versus 21 days, P < .02). At follow-up, all patients were alive with normal dietary patterns and bowel function. One incisional hernia developed in a patient after an open celiotomy but there were no bowel obstructions. The percentages of patients satisfied were 97% and 96% in the laparoscopic and open groups, respectively. An analysis of clinical end points compared by the type of appendectomy actually performed yielded findings similar to those based on the above intention-to-treat comparison. In addition, both the duration of oral analgesic use and the total hospital stay were slightly but significantly less in the laparoscopic group (data not shown). Because 5 patients in the open group but none in the laparoscopic group underwent colon resection, there is a potential bias against the open group, especially in terms of hospital stay. However, an analysis after the exclusion of patients who underwent procedures other than appendectomy led to no new findings (data not shown); specifically, the laparoscopic advantage in postoperative stay (mean, 2.6 versus 3.2 days; median, 2.0 versus 2.0 days, P = .015) persisted. Subgroup analyses. Analysis after stratification of patients by type of appendicitis (complicated or uncomplicated) and by preoperative activity level yielded potentially important findings. However, the study sample size was not calculated to support subgroup analyses: a risk of both false-positive and false-negative error is increased and conclusions must be drawn cautiously. Comparison of clinical end points within the uncomplicated appendicitis subgroup (simple appendicitis, suppurative appendicitis, negative

Table III. Summary of postoperative complications after appendectomy Complication General postoperative* (n) Wound infection (n) Abscess (n) Other technical (n) Total (n) Total patients (n) (%)

Open

Laparoscopic

20 17 1 1 39 28 (27)

15 17 4 4 40 28 (30)

P value .709 .697 .189 .189 — .592

*Includes fever, ileus, urinary retention, urinary tract infection, and rare others.

findings at appendectomy) yielded results similar to the whole-group analysis, with the additional finding of a slightly shorter total hospital stay after laparoscopic appendectomy (Table IV). Further, the median postoperative stay was 1 day shorter for laparoscopic appendectomy (1 versus 2 days, P < .01). Comparison within the complicated (perforative or gangrenous) appendicitis subgroup revealed no differences between laparoscopic and open appendectomy except for operative time (Table IV). The median duration of operation was 28 minutes longer for complicated versus uncomplicated appendicitis in the laparoscopic group, but it was only 4 minutes longer in the open group. When patients were stratified by activity level, laparoscopic advantages were found only within a subgroup termed “active patients” (those who were laborers or athletes or who exercised regularly) (Table V). The magnitudes of laparoscopic advantages were similar to the whole-group analysis, but the difference in median postoperative stay was 1 day, and a statistical advantage was found for total hospital stay. Conclusions drawn from this analysis are less reliable because of small numbers in each subgroup (active = 58 in the laparoscopic group and 72 in the open group). Analysis by occupation demonstrated that only laborers returned to full activity more quickly after laparoscopic versus open appendectomy (mean, 16 versus 24 days; median, 12 versus 22 days, P < .01). Economic analyses. Tables VI and VII show the estimated unadjusted costs for participants by operative technique for the intent-to-treat group and the “pure” group analysis. In both analyses, the mean total costs per patient were less for the laparoscopic group compared with the open group: $11,577 versus $13,965 (P = .09) in the intent-to-treat analysis and $9135 versus $13,965 (P = .001) in the “pure” analysis. Bootstrapped estimates of the difference in mean costs indicated that the study sample size was sufficient for robust t tests of significance.

396 Long et al

Surgery April 2001

Table IV. Subgroup analysis by type of appendicitis Laparoscopic End point

Median

Mean

Open SD

Median

Mean

SD

P value

Uncomplicated appendicitis* (n = 81) Duration of operation (minutes) Days to tolerance of general diet Days parenteral analgesia Milligrams morphine-equivalent parenteral analgesia Days oral analgesia Wound infection (n) Total hospital stay (days) Postoperative hospital stay (days) Days to work or school† Days to full activity‡

105 1 1 12

107.1 1.1 1.2 19.3

30.6 1.8 0.9 21.4

1.8

1.5

2.2 1.9 12.5 15.3

1.9 1.9 8.7 9.5

2 12 (14.8%) 2 1 9 14

P value

(n = 91) 91 1 2 26 2 12 (13.2%) 2 2 12.5 21

96.6 1.9 1.9 45.4

34.3 2.1 1.3 48.1

.005 < .001 < .001 .000

2

1.4

2.9 2.7 15 23

2.4 2.4 9.9 12.4

.387 .758 .038 .002 .204 .002

90.7 4.6 4.1 105.9

22.7 2.4 2.1 113.9

.024 .897 1.000 .316

5.7

5.2

7.7 7.5

4.9 5

.479 1.000 .624 .552

Complicated appendicitis§ (n = 12) Duration of operation (minutes) Days to tolerance of general diet Days parenteral analgesia Milligrams morphine-equivalent parenteral analgesia Days oral analgesia Wound infection (n) Total hospital stay (days) Postoperative hospital stay (days) Days to work or school Days to full activity

132.5 5 3.5 114

133.6 5.1 4.5 127.8

3 5 (41.7%) 6 6

(n = 14) 50.4 3.6 3.3 108.5

4

2.7

7.4 7.4

6.2 6.3

94.5 5 4 81 3.5 5 (35.7%) 7 7

*Simple appendicitis, suppurative appendicitis, or normal appendix. †n = 32 laparoscopic group, n = 36 open group. ‡n = 33 laparoscopic group , n = 42 open group. §Perforative or gangrenous appendicitis. n = insufficient for analysis.

As indicated by the clinical outcomes, room and board costs did not differ between the 2 groups in the intent-to-treat analysis ($1126 versus $1399, P = .21) yet the difference was highly statistically significant favoring the laparoscopic approach in the “pure” analysis ($793 versus $1399, P = .001). In both analyses, laparoscopy incurred higher supply charges because of the use of more expensive disposable medical devices ($939 versus $96, P = .00 intent-to-treat). However, this cost difference is partially offset by the reduced inpatient medication use ($281 versus $402, P = .12). Direct costs, in fact, did not differ between treatment groups ($5357 versus $4945, P = .29 intent-to-treat, $4741 versus $4945, P = .49 pure analysis). Indirect costs were, on average, $2800 less given a shorter mean length of stay and an abbreviated recuperative period in the laparoscopy versus open

group patients ($6220 versus $9020; P = .01 intentto-treat). This advantage to the laparoscopic approach was even more apparent in the “pure” group analysis, with an average difference in indirect costs of more than $4600 (P = .00). It should be noted that multivariate analyses predict $2700 lower total costs for laparoscopy (P = .02) after adjusting for confounding factors in the intent-totreat population. DISCUSSION This prospective randomized comparison made on the basis of a pre-study power analysis demonstrated statistically significant differences favoring laparoscopic over open appendectomy in analgesic requirements, days to tolerance of a general diet, postoperative hospital stay, and return to full activity. The small magnitude of these differences ren-

Long et al 397

Surgery Volume 129, Number 4 Table V. Subgroup analysis by patient activity level Laparoscopic End point

Median

Mean

Open SD

Median

Mean

SD

P value

Active patients* (n = 58) Days to tolerance of general diet Days parenteral analgesia Milligrams morphine-equivalent parenteral analgesia Days oral analgesia Total hospital stay (days) Postoperative hospital stay (days) Days to work or school† Days to full activity‡

(n = 72)

1 1 11.3

1.1 1.3 29.3

1.5 1.5 58.6

1 2 35

2 2 57.5

1.8 1.4 71.8

< .001 < .001 .000

2 2 1 8 12

1.9 2.4 2.1 12.9 16

1.9 2.2 2.2 11.9 12.4

2 2 2 12 22

2.4 3.1 2.9 14.7 23.9

1.9 2.6 2.5 9.6 13.4

.081 .047 .013 .222 .008

Inactive patients§ (n = 20) Days to tolerance of general diet Days parenteral analgesia Milligrams morphine-equivalent parenteral analgesia Days oral analgesia Total hospital stay (days) Postoperative hospital stay (days) Days to work or school Days to full activity

(n = 21)

1 1 21

1.4 1.4 29.2

1.2 0.8 27.9

1 1 22

2.5 2.1 41.3

3.1 1.5 43.4

.429 .257 .611

2 2 2 17 15

1.8 2.4 2.1 18.3 18.9

1.5 1.3 1.5 10.2 8.6

2 2 2 12 19.5

1.9 3.3 3.2 14.5 20.6

1.2 3.4 3.2 10.7 8.6

.764 .543 .141 .341 .595

*Occupation: laborer and/or regular exerciser or athlete. †n = 21 laparoscopic group, n = 27 open group. ‡n = 22 laparoscopic group, n = 32 open group. §Occupation: student or office worker and no regular exercise. n = 11 laparoscopic group, n = 10 open group.

ders their clinical significance questionable. However, when indirect costs are included in the economic analysis as recommended by the U.S. Panel in Cost Effectiveness in Health and Medicine, this study indicates that laparoscopic appendectomy is relatively less expensive than open appendectomy in the treatment of acute appendicitis. Studies from the payer perspective that focus on billed charges will tend to overestimate true resource use and ignore what may be the key advantage to the laparoscopic approach— fewer indirect costs associated with treatment. Although highly statistically significant, the advantage in mean postoperative hospital stay conferred by laparoscopic appendectomy was only 0.8 days. This falls short of our minimal threshold of clinical significance: 1 entire hospital day. Matters of hours may relate to patient preferences or dismissal process efficiencies and inefficiencies; however, rarely will such issues translate into durations of 24 hours or additional calendar days. The 2 groups’ identical median postoperative stay further

illustrates the small magnitude of the laparoscopic advantage. The statistical advantage in total hospital stay unveiled when the 2 groups were controlled for age, sex, and BMI was clinically insignificant; it was likely absent from the initial whole-group analysis because differences in BMI were biased against the laparoscopic group. Our subgroup analysis showed that hospital stay data more strongly favor laparoscopic appendectomy within the uncomplicated appendicitis and active lifestyle subgroups. Here, total hospital stay significantly differed by 0.7 days in both subgroups, and the difference in postoperative stay separates the medians: 1 versus 2 days in both subgroups. This difference is arguably just large enough to be clinically meaningful. However, we are quick to acknowledge that because our sample size was not intended to allow powerful subgroup comparisons, conclusions drawn from subgroup analyses must be guarded. The laparoscopic advantage in hospital stay is comparable with that found in other randomized

398 Long et al

Surgery April 2001

Table VI. Unadjusted costs per patient associated with laparoscopic and open appendectomy for patients with acute appendicitis Intent-to-treat analysis* Laparoscopy (n = 93) Cost end point Room and board Medications Supplies Operating room Emergency room Total hospital costs Total physician costs Total direct costs Total direct charges Total indirect costs Total costs

Median $763 $143 $1019 $1321 $140 $3401 $1349 $4761 $6932 $4416 $9239

Mean

SD

$1126 $281 $939 $1366 $150 $3690 $1683 $5357 $7711 $6220 $11,577

1355 464 450 307 59 1987 842 2650 3722 7532 9726

Open (n = 105) Median $842 $168 $46 $1243 $140 $2411 $1367 $3872 $5666 $7272 $11,316

Mean $1399 $402 $96 $1264 $165 $3267 $1693 $4945 $7146 $9020 $13,965

SD 1550 571 123 215 84 2162 855 2779 3588 8253 9971

Mean difference –$273 –$120 $843 $102 –$15 $422 –$10 $412 $564 –$2800 –$2388

P value

95% CI

(–705, 159) .210 (–270, 30) .120 (735, 951) .000 (26, 178) .009 (–36, 7) .170 (–175,1020) .160 (–255, 233) .930 (–351, 1176)† .290 (–461, 1590)‡ .280 (–5026, –576)§ .010 (–5156, 380) .090

*Costs per patient in 1995 constant dollars. †Bootstrap 95% CI using the percentile method (–266 to 1194). ‡Bootstrap 95% CI using the percentile method (–335 to 1662). §Bootstrap 95% CI using the percentile method (–5101 to –485). Bootstrap 95% CI using the percentile method (–5101 to 575).

Table VII. Unadjusted cost per patient associated with laparoscopic and open appendectomy for patients with acute appendicitis: Pure analysis* Laparoscopy (n = 78) Cost end point

Median

Mean

Room and board Medications Supplies Operating room Emergency room Total hospital costs Total physician costs Total direct costs Total direct charges Total indirect costs Total costs

$723 $134 $1101 $1321 $140 $3338 $1322 $4600 $6695 $4191 $8606

$793 $155 $1030 $1334 $152 $3277 $1474 $4741 $6853 $4389 $9135

SD

Open ( n = 105) Median

506 $842 94 $168 415 $45 232 $1243 65 $140 908 $2411 461 $1368 1105 $3872 1584 $5665 2783 $7272 5848 $11,316

Mean $1399 $402 $96 $1264 $165 $3267 $1693 $4945 $7146 $9020 $13,965

SD 1550 571 123 215 84 2162 855 2779 3588 8253 9971

Mean difference –$606 –$247 $934 $70 –$13 $10 –$220 –$204 –$293 –$4631 –$4830

95% CI

P value

(–947, –265) (–363, –131) (824, 1044) (4, 137) (–36, 11) (–509, 530) (–420, –20) (–793, 385)† (–1071, 484)‡ (–6342, –2920)§ (–6887, –2773)

.001 .000 .000 .040 .280 .970 .030 .490 .450 .000 .000

*Costs per patient in 1995 constant dollars. †Bootstrap 95% CI using the percentile method (–772 to 363). ‡Bootstrap 95% CI using the percentile method (–992 to 336). §Bootstrap 95% CI using the percentile method (–6362 to –3117). Bootstrap 95% CI using the percentile method (–6916 to –2888).

studies,8-13,20,27-36 which on balance suggest that laparoscopic appendectomy does not appreciably shorten length of stay. The trauma of peritoneal access is not likely the primary determinant of hospital stay. Laparoscopic appendectomy eliminated the need for any postoperative analgesics in only 3 patients (3%). A comparison of morphine-equivalent doses of narcotics given parenterally revealed

an advantage of 20 mg, or roughly 60% of the total open group dose, for the laparoscopic group. This decrement in analgesic requirement represents the laparoscopic advantage of greatest in-hospital clinical significance detected by our study; it appeared to be conferred only to the uncomplicated appendicitis and active lifestyle subgroups. Our study found similar incidences of wound infections, abscesses, technical complications, gen-

Long et al 399

Surgery Volume 129, Number 4

eral postoperative complications, and total complications. The occurrence of wound infection was similar even after complicated appendicitis, but the small number of relevant patients increases the risk of a type II statistical error. Laparoscopic appendectomy has not increased the wound infection rate in any study and has diminished it in 3 studies. Although laparoscopic appendectomy was not associated with an increased incidence of postoperative intra-abdominal abscess, it is a concern that all 4 laparoscopy-associated abscesses occurred after uncomplicated appendicitis. There may be a laparoscopic predisposition to abscess that our study was not powerful enough to detect. Overall, however, no study8-13,20,28-35 associated laparoscopic appendectomy with more complications; 1 study30 demonstrated a decreased incidence. Laparoscopic appendectomy thus appears to be a safe procedure. Another laparoscopic advantage of both clinical and social importance is the patient’s return to full activity 1 week earlier than those in the open group. Subgroup analysis suggests that active, but not inactive, lifestyle patients realize this benefit of laparoscopic appendectomy, which somewhat diminishes the impact of the earlier return to full activity. Five30,31,33-35 of the 1212,20,28-37 other randomized studies found that patients undergoing laparoscopic appendectomy returned to full activity 5 to 13 days sooner, but only 1 found that they returned to work 6 days sooner.28 Our economic analysis relied heavily on administrative claims data to track the resource utilization associated with the treatment modalities. As noted previously, we checked all data outliers and the line item charges in detail for possible billing errors by using medical records as verification of utilization. Significant billing errors were found in 3 cases, and adjustments were made to the cost calculations to correct for these discrepancies. For example, 1 patient was missing a charge for operating room use despite the fact that the operation was verified in both the trial and medical records. Another patient had 7-fold the average operating room charges although operating room times were typical, and there were no indications of operative complications in the medical or clinical trial record. Finally, another patient was billed as having bilateral carpal tunnel operations in addition to the appendectomy, yet the medical record showed no evidence of such a procedure. While it is likely that these billing errors were eventually corrected in the claims records, they occurred outside of the “episode of care window” used to identify line item charges related to the appendectomy. Further-

more, despite the fact that these billing errors occurred in only 1% of the claims, these errors resulted in high charge and cost variability, potentially obfuscating treatment effects. This highlights the importance of this careful detailed review with attention to verifying line item charges when using claims data to track medical resource use. While the difference in mean unadjusted total costs per patient may only be mildly significant in the intent-to-treat analysis, the difference in total costs was highly significant when patients were stratified by the procedure actually performed and after adjusting for confounding factors. Furthermore, in approximately 94% of the bootstrapped iterations, the average cost was lower for laparoscopic than for open appendectomy. More than 60% of the time, this cost difference was at least $2000. When one considers that more than 260,000 appendectomies were performed in the United States in 1997, it is hard to ignore the economic significance of this global cost difference in favor of a laparoscopic approach. The Laparoscopic Appendectomy Interest Group: Alan L. Beal, MD, Richard M. Devine, MD, John H. Donohue, MD, David R. Farley, MD, Michael B. Farnell, MD, Clive S. Grant, MD, Heidi Nelson, MD, Florencia G. Que, MD, and Geoffrey B. Thompson, MD. REFERENCES 1. Semm K. Endoscopic appendectomy. Endoscopy 1983; 15:59-64. 2. Schreiber JH. Early experience with laparoscopic appendectomy in women. Surg Endosc 1987;1:211-6. 3. Pier A, Gotz F, Bacher C. Laparoscopic appendectomy in 625 cases: from innovation to routine. Surg Laparosc Endosc 1991;1:8-13. 4. Ludwig KA, Cattey RP, Henry LG. Initial experience with laparoscopic appendectomy. Dis Colon Rectum 1993; 36:463-7. 5. Nowzaradan Y, Westmoreland J, McCarver CT, Harris RJ. Laparoscopic appendectomy for acute appendicitis: indications and current use. J Laparoendosc Surg 1991;1:247-57. 6. Nowzaradan Y, Barnes JP Jr, Westmoreland J, Hojabri M. Laparoscopic appendectomy: treatment of choice for suspected appendicitis. Surg Laparosc Endosc 1993;3:411-6. 7. Saueland S, LeFering R, Holthausen U, Neugebauer EAM. Laparoscopic vs conventional appendectomy—a metaanalysis of randomized trials. Langenbecks Arch Surg 1998;383:289-95. 8. McCahill LE, Pellegrini CA, Wiggins T, Helton WS. A clinical outcome and cost analysis of laparoscopic versus open appendectomy. Am J Surg 1996;171:533-7. 9. Fritts LL, Orlando R III. Laparoscopic appendectomy. A safety and cost analysis. Arch Surg 1993;128:521-4. 10. Buckley RC, Hall TJ, Muakkassa FF, Anglin B, Rhodes RS, Scott-Conner CE. Laparoscopic appendectomy: is it worth it? Am Surg 1994;60:30-4. 11. Cohen MM, Dangleis K. The cost-effectiveness of laparoscopic appendectomy. J Laparoendosc Surg 1993;3:93-7.

400 Long et al

12. Minne L, Varner D, Burnell A, Ratzer E, Clark J, Haun W. Laparoscopic vs open appendectomy. Arch Surg 1997; 132:708-12. 13. Hiekkinen TJ, Haukipuro K, Hulkko A. Cost-effective appendectomy. Open or laparoscopic? A prospective randomized study. Surg Endosc 1998;1:1204-8. 14. Gold M, Siegel J, Russel L, Weinstein M, editors. Cost effectiveness in health and medicine. New York: Oxford University Press, 1996. 15. Lewis FR, Holcroft JW, Boey J, Dunphy JE. Appendicitis. A critical review of diagnosis and treatment in 1,000 cases. Arch Surg 1975;110:677-83. 16. Silberman VA. Appendectomy in a large metropolitan hospital. Retrospective analysis of 1,013 cases. Am J Surg 1981;142:615-8. 17. Pieper R, Kager L, Nasman P. Acute appendicitis: a clinical study of 1018 cases of emergency appendectomy. Acta Chir Scand 1982;148:51-62. 18. Berry J Jr, Malt RA. Appendicitis near its centenary. Ann Surg 1984;200:567-75. 19. Hale DA, Molloy M, Pearl RH, Schutt DC, Jaques DP. Appendectomy: a contemporary appraisal. Ann Surg 1997;225:252-61. 20. Reiertsen O, Larsen S, Trondsen E, Edwin B, Faerden AE, Rosseland AR. Randomized controlled trial with sequential design of laparoscopic vs conventional appendicectomy. Br J Surg 1997;84:842-7. 21. Cohen J. Statistical power analysis for the behavioral sciences. New York: Academic Press; 1977. p. 25-6. 22. Farnell MB, Worthington-Self S, Mucha P Jr, Ilstrup DM, McIlrath DC. Closure of abdominal incisions with subcutaneous catheters. A prospective randomized trial. Arch Surg 1986;121:641-8. 23. Lave JR, Pashos CL, Anderson GF, Brailer D, Bubolz T, Conrad D, et al. Costing medical care: using Medicare administrative data. Med Care 1994;32:JS77-89. 24. United States Department of Labor Bureau of Labor Statistics. Usual weekly earnings of wage and salary workers: fourth quarter 1995. News release of Thursday, February 15, 1996. 25. Briggs A, Gray A. The distribution of health care costs and their statistical analysis for economic evaluation. J Health Service Res Policy 1998;3:233-45.

Surgery April 2001 26. Efron B, Tibshirani RJ, editors. An introduction to the bootstrap. New York: Chapman & Hall/CRC; 1998. 27. Briggs AH, Wonderling DE, Mooney CZ. Pulling cost-effectiveness analysis up by its bootstraps: A non-parametric approach to confidence interval estimation. Health Econ 1997;6:327-40. 28. Attwood SE, Hill AD, Murphy PG, Thornton J, Stephens RB. A prospective randomized trial of laparoscopic vs open appendectomy. Surgery 1992;112:497-501. 29. Tate JJ, Dawson JW, Chung SC, Lau WY, Li AK. Laparoscopic vs open appendicectomy: prospective randomised trial. Lancet 1993;342:633-7. 30. Kum CK, Ngoi SS, Goh PM, Tekant Y, Isaac JR. Randomized controlled trial comparing laparoscopic and open appendicectomy. Br J Surg 1993;80:1599-1600. 31. Frazee RC, Roberts JW, Symmonds RE, Snyder SK, Hendricks JC, Smith RW, et al. A prospective randomized trial comparing open vs laparoscopic appendectomy. Ann Surg 1994;219:725-8. 32. Martin LC, Puente I, Sosa JL, Breslaw R, McKenney MG, Ginzburg E, et al. Open vs laparoscopic appendectomy. A prospective randomized comparison. Ann Surg 1995;222: 256-61. 33. Ortega AE, Hunter JG, Peters JH, Swanstrom LL, Schirmer B. A prospective, randomized comparison of laparoscopic appendectomy with open appendectomy. Am J Surg 1995;169:208-12. 34. Hansen JB, Smithers BM, Schache D, Wall DR, Miller BJ, Menzies BL. Laparoscopic vs open appendectomy: prospective randomized trial. World J Surg 1996;20:17-20. 35. Cox MR, McCall JL, Toouli J, Padbury RTA, Wilson TG, Wattchow DA, et al. Prospective randomized comparison of open vs laparoscopic appendectomy in men. World J Surg 1996;20:263-6. 36. Mutter D, Vix M, Bui A, Evrard S, Tassetti V, Breton JF, et al. Laparoscopy not recommended for routine appendectomy in men: results of a prospective randomized study. Surgery 1996;120:71-4. 37. Rohr S, Thiry CL, de Manzini N, Perrauid Y, Meyer C. Laparoscopic vs open appendectomy in men: a prospective randomized study (abstract). Br J Surg 1994;81(Suppl 1):6-7.