- Email: [email protected]
Cost-effectiveness of laparoscopic versus open pyloromyotomy
j o u r n a l o f s u r g i c a l r e s e a r c h 1 7 8 ( 2 0 1 2 ) 3 1 5 e3 2 0
Available online at www.sciencedirect.com
journal homepage: www.JournalofSurgicalResearch.com
Cost-effectiveness of laparoscopic versus open pyloromyotomy Emma V. Carrington, MD, Nigel J. Hall, PhD, Maurizio Pacilli, MS, David P. Drake, MD, Joseph I. Curry, MD, Edward M. Kiely, MD, Paolo De Coppi, MD, Agostino Pierro, MD, and Simon Eaton, PhD* UCL Institute of Child Health and Great Ormond Street Hospital for Children, London, United Kingdom
article info
abstract
Article history:
Background: Infantile hypertrophic pyloric stenosis can be corrected by either open (OP) or
Received 18 October 2011
laparoscopic pyloromyotomy (LP). LP may provide clinical benefits of reduced time to
Received in revised form
postoperative full feeds and reduced postoperative inpatient stay, but the cost effective-
2 January 2012
ness is not known. Our aim was to compare the cost effectiveness of laparoscopic and open
Accepted 18 January 2012
pyloromyotomy.
Available online 27 March 2012
Methods: OP and LP were compared in a multicenter randomized double-blind controlled trial, for which the primary outcomes were time to full feeds and time to discharge. In
Keywords:
order to undertake a detailed cost analysis, we assigned costs, calculated on an individual
Infants
patient basis, to laboratory costs, imaging, medical staff, medication, ward, operative, and
Pyloric stenosis
outpatient appointments for 74 patients recruited from one of the participating centers.
Pyloromyotomy
Data (mean SEM) were compared using linear regression analysis, adjusting for the
Laparoscopy
minimization criteria used in the trial.
Economic analysis
Results: Operation costs were similar between the two groups ($3,276 $244 LP versus
Cost effectiveness
$3,535 $152 OP). A shorter time to full feeds and shorter hospital stay in LP versus OP patients resulted in a highly significant difference in ward costs ($2,650 $126 LP versus $3,398 $126 OP; P ¼ .001) and a small difference in other costs. Overall, LP patients were $1,263 (95% confidence interval $395e$2,130; P ¼ .005) less expensive to treat than OP patients. Sensitivity analyses of laparoscopic hardware usage and of incomplete pyloromyotomy indicated that LP was consistently less expensive than OP. Conclusions: LP is a cost-effective alternative to OP as it delivers improved clinical outcome at a lower price. ª 2012 Elsevier Inc. All rights reserved.
1.
Introduction
Pyloromyotomy for pyloric stenosis has recently been performed by the laparoscopic approach [1,2]. A number of studies have examined clinical outcomes in
open pyloromyotomy (OP) compared with laparoscopic pyloromyotomy (LP), either on a retrospective nonrandomized basis (meta-analyzed in [3]) or, more recently, in the context of prospective randomized controlled trials [4e6]. It appears from these data that LP leads to earlier time to full feeds and/
* Corresponding author. UCL Institute of Child Health and Great Ormond Street Hospital for Children, London, United Kingdom. Tel.: þ44 20 7905 2158; fax: þ44 20 7404 6181. E-mail address: [email protected] (S. Eaton). 0022-4804/$ e see front matter ª 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.jss.2012.01.031
316
j o u r n a l o f s u r g i c a l r e s e a r c h 1 7 8 ( 2 0 1 2 ) 3 1 5 e3 2 0
or discharge from hospital, although there is some concern that there may be a higher rate of incomplete pyloromyotomy following LP compared with OP [5,7,8]. In addition to clinical outcomes, evidence of cost effectiveness in patient management is becoming a fundamental part of good patient care, and is a necessary step in the evaluation of new treatments and techniques. Cost effectiveness of an intervention involves consideration not only of the monetary cost, but also of the attributable benefit derived from improved outcome. Thus, what is good for the patient is also good for the surgeon and good for the institution. A comprehensive literature review revealed that there has been no robust economic analysis of LP versus OP. A number of studies have included a basic cost analysis based on hospital charges data, some of which report higher hospital charges with LP compared with OP without statistical significance [9,10] and others of which report either no difference [11] or reduced hospital charges with LP [12]. In their randomized controlled trial of LP versus OP, St Peter et al. found no difference in cost between LP and OP, although they did not report cost data [4]. The aim of this study was to investigate the cost effectiveness of LP versus OP in patients with pyloric stenosis, using a subset of patients from a single center of a recently published multicenter randomized controlled trial [5]. We attributed costs on a per-patient basis using actual inpatient time, theater time, pathology, imaging, and medication costs. We undertook sensitivity analyses to examine the effects of laparoscopic equipment costs and different rates of incomplete pyloromyotomy on cost effectiveness.
2.
Methods
We have reported clinical outcomes from a double-blind, multicenter, international randomized controlled trial with prospective data collection [5]. The centers in this study were in five countries with great differences in healthcare systems and consequently costs could not be evaluated in a uniform fashion across all centers. In order to collect robust uniform economic data, the present evaluation used only patients recruited from one center, Great Ormond Street Hospital for Children (GOSH), UK.
As part of that study, patients were randomized with weighted minimization including center as a minimization criterion and were treated according to a standardized clinical care pathway designed specifically for the trial. This pathway stipulated all aspects of patient care including postoperative feeding regime, postoperative analgesia, and criteria for discharge and has been reported previously. Therefore all patients were treated identically, regardless of operative approach, with any treatment effect being due to different operative approach rather than other confounding factors. The feeding protocol, described fully in the previous paper [5], was 15 mL of formula or a short breast feed at 6 h postoperatively followed by (if tolerated) a full feed of either formula milk or a breast feed 1 h later. Hence the time to first tolerated full feed, by definition, could not be less than 7 h. We did not use any enteral fluid other than formula or breast milk. Economic data were derived from a combination of data collected prospectively from the original trial (inpatient and operative durations) and supplementary data collected retrospectively from individual patient records (per-patient medication and investigations). Ethical approval was obtained for this economic analysis (approval number 03SG45). The original trial was registered with clinicaltrials.gov (identification number NCT00144924). Of 84 patients recruited into the trial from this center, 10 sets of clinical notes were unavailable at the time of analysis for further retrospective review and therefore this economic analysis is based on 74 patients (41 LP, 33 OP). Demographic details and other clinical outcome measures in relation to the subgroup of infants included in this study are shown in Table 1. Economic outcome was calculated as the direct costs associated with in hospital treatment. The overall cost was subdivided into the following categories: 1) ward costs; 2) medical staff costs; 3) medication costs; 4) laboratory costs; 5) imaging costs; 6) operative costs; and 7) outpatient costs. These costs were obtained from the hospital finance department.
2.1.
Ward and staffing costs
Ward costs, representing the cost of running the neonatal surgical ward, were calculated for each patient based on
Table 1 e Demographics and clinical outcomes of patients included in the economic analysis. LP (n ¼ 41) Age (d) Weight (kg) Bicarbonate at presentation (mmol/L) Gestation at birth (wk) Feeding type Duration of symptoms <48 h Time to first tolerated feed (h) Time to full feeds (h) Time to discharge (h)
31 (27e36, 15e65) 3.7 (3.3e4.1, 2.1e59.1) 31 (27.8e36, 22e59.1) 40 (39e40, 32e42) 11 breast 30 mixed/bottle 1/41 10.2 (7.2e16.6, 5.5e31.8) 17.75 (11.6e20, 9.5e48.2) 27.6 (23.4e45.8, 13e71.5)
OP (n ¼ 33)
P value
(21e38, 13e83) (3.3e4.1, 2.4e6.6) (25.7e34.1, 19e57.7) (39e40, 36e42) 4 breast 29 mixed/bottle 0/33 13.2 (7.1e27.3, 6.2e54.7) 26.2 (16.6e44.2, 9.2e107.4) 45.5 (28.1e55, 19.2e129.2)
n.s. n.s. n.s. n.s. n.s.
33 3.7 30 40
n.s. .002 .001 .001
Data are median (interquartile range, range), or n (duration of symptoms, feed type). Data were compared using linear regression, adjusting for the minimization criteria, or Fisher exact test (duration of symptoms, feed type). A P value of <.05 was considered to be significant.
j o u r n a l o f s u r g i c a l r e s e a r c h 1 7 8 ( 2 0 1 2 ) 3 1 5 e3 2 0
length of in-patient stay (time of admission to time of discharge). This took into account most direct costs associated with ward care, including nurse staffing (but not medical staffing), stock medications (including paracetamol and codeine), indwelling devices and fluids, laundry, equipment maintenance, and staff training. This also included an attributable portion of fixed overhead costs of hospital ground rent, utility supply, and ancillary staff costs. Medical staff costs were also based on total admission time and were calculated on the basis of average salaries for consultants, and for training and other junior staff. Medical staff time per patient (and therefore cost) was divided into the fixed cost of “medical admission” (90 min of junior staff time for initial management, consent, and ordering of initial investigations) and the variable cost of “daily duties” (10 min of consultant input and 60 min of junior doctor input per 24 h). These were attributed on a per-patient basis according to length of stay on the ward.
2.2.
Operative costs
Operative costs were calculated using a combination of prospectively collected data from the trial and data from individual theater records. Costs were calculated as the total of: cost per minute of theater time, consumables, theater medical staff costs, theater reusables, and hardware. Cost per minute (calculated for each patient from induction of anesthesia to recovery) of theater time took into account most direct costs associated with theaters (anesthetic medications and equipment, salaries of nursing staff, operating department practitioners, administration staff, and theater porters). It also included maintenance of hardware (although not purchase of large hardware items) and an attributable portion of fixed overhead costs. Theater hardware (diathermy and laparoscopic equipment) purchase costs were calculated assuming a 5-y lifetime and using a straight line depreciation method. Hardware cost per operation was attributed assuming that each laparoscopic equipment set was used in theaters four times per day and each diathermy machine was used in theaters six times per day, and that diathermy was used in both LP and OP whereas the laparoscopic equipment set was used only in LP. A standardized list of theater consumables was created for both OP and LP and it was assumed that this fixed number was used in each case. It was assumed that each case (both LP and OP) involved the participation of one consultant surgeon, one junior/trainee surgeon, one consultant anesthetist, and one junior/trainee anesthetist, using the costs per minute described in Ward and staffing costs above. Theater reusable costs were calculated as the cost of sterilization and redelivery and were costed on a per-patient basis.
2.4.
Laboratory costs
Laboratory costs (microbiology, hematology, and biochemistry) were taken from NHS Reference Costs 2006e07 and were derived retrospectively based on the actual tests ordered for each patient.
2.5.
Imaging costs
Imaging costs were calculated to include purchase and maintenance of equipment, the time-attributable salary of one consultant radiologist, and a proportion of fixed overhead costs. The actual number of tests per patient was derived from the clinical records.
2.6.
Outpatient costs
Outpatient visits were assumed to take 15 min of a consultant surgeon’s time. One outpatient visit was attributed to each patient.
2.7.
Data analysis
All data were initially collected in pounds sterling (£). We converted all costs to U.S. dollars ($) at an exchange rate of £1 ¼ $1.87, which was the average (and also the median) foreign exchange rate over the time of the study. All outcomes, both clinical and economic, were compared by regression analysis, adjusting for the minimization criteria used in the trial design (age, weight, gestational age at birth, bicarbonate at presentation, feed type, and duration of symptoms), and for the randomization group on an intentionto-treat basis (SPSS version 15, SPSS Inc, Chicago, IL). P < .05 was considered significant.
3.
Results
3.1.
Clinical outcomes
The main clinical outcomes of the whole trial, demonstrating significant advantages of LP over OP, have been reported previously [5]. The primary outcomes of the subgroup of patients included in this economic analysis were similar to those from the main trial and are shown in Table 1. There were no conversions to open in the LP group and one mucosal laceration (repaired intraoperatively) in both the LP and OP groups. There were no incomplete pyloromyotomies requiring repeat operation.
3.2. 2.3.
317
Costs
Medication costs
Simple analgesia (i.e., paracetamol and codeine phosphate) was taken from the ward medication cost and therefore accounted for in the ward costs. Discharge medication was costed on a per-patient basis separately according to the BNF for Children 2008 manual [13] and derived retrospectively from clinical discharge records.
Overall cost of LP was on average $1,263 cheaper than OP (adjusted mean difference; Table 2, Fig. 1); this was highly significant (P ¼ .005). A regression analysis, adjusting for the minimization criteria (age, weight, gestational age at birth, bicarbonate at presentation, feed type, and duration of symptoms), used in the main trial demonstrated that none of these was found to significantly affect overall cost.
318
j o u r n a l o f s u r g i c a l r e s e a r c h 1 7 8 ( 2 0 1 2 ) 3 1 5 e3 2 0
Table 2 e Costs for each category. Costs Ward Medical staff Medication Pathology Imaging Operative Outpatientsa Overall
LP cost $ 2,650 501 0.4 28 115 3,276 45 6,615
126 13 0.1 2 20 244 269
OP cost $ 3,398 580 1.9 28 128 3,535 45 7,715
Adjusted mean difference (95% CI) $ 932 98 1.5 1 15 215
126 28 1.5 4 24 152
(e1,481, e384) (e156, e41) (e4.4, 1.3) (e11, 7) (e79, 49) (e847, 418)
1,263 (e2,130, e395)
346
P value .001 .001 .29 .74 .65 .5 Not applicable .005
Data are mean SEM; P values were calculated using linear regression, adjusting for the minimization criteria. A P value of <.05 was considered to be significant. a Defined as a fixed cost per patient.
Ward costs (average 42% in LP and 43% in OP) and operative costs (average 47% in both LP and OP) were the main contributors to overall costs. Ward costs were significantly lower in the LP group when compared to the OP group, reflecting the shorter length of stay (adjusted mean difference ¼ $932, P ¼ .001). The shorter length of stay also resulted in significantly cheaper medical staff costs (adjusted mean difference $98, P ¼ .001). No significant difference was found in imaging or laboratory costs (P ¼ .65 and P ¼ .74 respectively) between LP and OP. Medication costs were minimal and similar between groups (P ¼ .29). No significant difference was found in the operative costs (P ¼ .5) even though significantly more consumable sterile sets were used per LP (P ¼ .007). The attributable cost of the laparoscopic hardware was found to be $6.83 per procedure, which was negligible when compared to a median total operative cost of $3,276 per LP.
Fig. 1 e Costs for open and laparoscopic pyloromyotomy. Graph shows overall costs for open pyloromyotomy (OP, n [ 33) and laparoscopic pyloromyotomy (LP, n [ 41). In addition, the overall costs were divided into operative, ward, and other costs (medical staff costs plus medication costs, pathology costs, imaging costs, and outpatient costs). Data are shown as mean ± SEM, and the P values shown are for linear regression after adjusting for the minimization criteria. n.s. [ not significant.
4.
Discussion
Using a robust economic analysis technique, we have found that laparoscopic pyloromyotomy is significantly less expensive than open pyloromyotomy. The major contributory factor to the lower cost was the decreased length of stay and, therefore, lower ward costs and medical staff costs. Although laparoscopy is usually assumed to be more expensive on the basis of hardware costs and potentially longer operative time, we found no significant difference in operative costs between the two procedures. Economic analysis necessitates making assumptions in order to derive and attribute costs. It is therefore important to perform a sensitivity analysis to test the effect of these assumptions and, therefore, the robustness of the conclusions. The large initial capital outlay for laparoscopic equipment renders the economic analysis sensitive to the usage of laparoscopic equipment for other procedures. For example, if a set of laparoscopic equipment were to be purchased solely for use in pyloromyotomies, the operative cost per procedure would be considerably greater. In a hospital treating approximately 30 patients per year laparoscopically for pyloromyotomy, this would increase the cost per operation by approximately $374. In such a scenario, LP would still be cheaper overall than OP, but only by $888, and the difference between the procedures would only just be significant (P ¼ .045). Clearly, the economic effectiveness of laparoscopic equipment depends on its frequency of use. One of the main concerns regarding laparoscopic pyloromyotomy is the potential for an increased risk of incomplete pyloromyotomy [3,6,8]. Although the risk of incomplete pyloromyotomy has not so far been shown to be significantly greater in any of the three published randomized controlled trials comparing LP and OP [4e6], there was a tendency for more in the laparoscopic group in two of these trials [5,6]. In the single-center cohort studied for the present analysis (from the larger cohort of Hall et al. [5]), there were no incomplete pyloromyotomies. It is therefore of interest to undertake a sensitivity analysis to try to determine the potential impact that a higher incidence of incomplete pyloromyotomy would have in the economic analysis. For this sensitivity analysis, we chose to model the effect of a fairly extreme difference in the rate of incomplete pyloromyotomy:
j o u r n a l o f s u r g i c a l r e s e a r c h 1 7 8 ( 2 0 1 2 ) 3 1 5 e3 2 0
a 5% rate of incomplete pyloromyotomy in the laparoscopic group, which is very high in comparison with the published rate of 3.5%, and a rate of zero in the open group. For each patient with this complication, we assumed an additional ward stay of 5 d, and an additional operation with associated imaging, pathology, and medical staff costs. On the basis of these assumptions, each incomplete pyloromyotomy would cost $9,500. Despite these additional costs, laparoscopic pyloromyotomy is still, on average, cheaper, with a mean difference in cost of $636, compared with an adjusted mean difference of $1,263 in the actual analysis. However, economic effectiveness is a combination of clinical outcome and the financial cost of achieving that clinical outcome. Whether a 5% rate of incomplete pyloromyotomy can be considered economically effective is a subjective matter. There were no wound infections in the patients reported in this study, although there is a potential difference in the rate of wound infection between LP and OP. If wound infections, or other wound complications, did occur, then this could impact length of stay and cost of medication for antibiotics, and also potentially lead to additional out-of-hospital costs if identified after discharge. Medical staff costs were based on assumptions concerning daily duties, and we acknowledge that not knowing accurately how many times each patient was reviewed is a potential shortcoming of this study. However, the cost of medical staffing calculated as described in the methods was only about 7% of the total cost; therefore, only very large differences in actual staffing time between OP and LP would significantly alter the conclusion that LP is cheaper than OP. This calculation was based only on time at the bedside, which is merely one component of medical staff duties. Other medical staff duties (including review of investigations, audit, discussion with relatives, teaching and training, etc.) are less easily attributable on a per-patient basis but could also influence economic decision making. In order to assess the potential impact of these activities, we estimated the cost associated with these other duties. Assuming an additional 2 h of consultant time and 2 h of junior staff time per patient, a cost of $507 per patient additional to the costs described in Table 2 would be incurred. This may give a more accurate reflection of total cost (i.e., $7,122 per patient LP versus $8,222 per patient OP). However, this amount is the same for OP and LP and therefore does not alter the finding that LP is less expensive. This analysis has the advantage of being conducted at a single center, allowing costs to be estimated consistently for each patient using the same data sources and assumptions. The disadvantage of this approach is that the conclusions are not necessarily applicable in other centers due to potential differences between institutions. For example, Great Ormond Street Hospital is located in central London, so fixed costs (associated with location) and staffing costs may be significantly higher than in other more regional centers in the UK. Although absolute costs at other institutions are inevitably going to be different, unless relative expense between categories is markedly different (e.g., theater costs are proportionally higher than ward costs at a given institution), it is likely that the principal findings are generalizable across other centers.
319
In the main report of our trial, a limited cost analysis was undertaken in which there appeared to be no significant difference in cost between the OP and LP groups [5]. However, in that analysis, all costs were estimated rather than taken on a per-patient basis, and costs were combined from centers in different continents and, therefore, very different healthcare systems. We believe that the conclusions derived from this study, in which a much more detailed analysis was made on patients from a single center, are much more robust than those previously reported. The costs associated with LP and OP have previously been investigated in several papers. Kim et al., in 2005, using a nonrandomized cohort of patients receiving LP, right-upper quadrant OP, or circum-umbilical OP, reported that LP was significantly associated with lower operative costs than circum-umbilical OP; however, costs associated with the length of hospital stay were not reported [14]. Campbell et al. [9] and Bufo et al. [10], both reporting hospital charges on nonrandomized patients from single centers, found no significant difference in costs between LP and OP. Campbell et al., studying a multi-institutional, nonrandomized group of patients, reported a significant difference in hospital charges in centers that performed LP compared with centers performing OP [11]. However, this may reflect a difference in hospital billing policies or in fixed institutional costs rather than a real difference between surgical approaches. We believe that the analysis we have undertaken is more rigorous, and the conclusions more robust, than these previous reports.
5.
Conclusions
We conclude that in a center with suitable laparoscopic experience, performing a variety of operations using a laparoscopic approach, LP is significantly less expensive and therefore more economically effective than OP. If experience is less (possibly leading to a much higher rate of incomplete pyloromyotomy) or laparoscopic equipment is not frequently used, these conclusions may not be valid.
Acknowledgments The Sir Halley Stewart Trust is thanked for a grant supporting this work. No authors have any conflict of interest.
references
[1] Alain JL, Grousseau D, Terrier G. Extramucosal pylorotomy by laparoscopy. J Pediatr Surg 1991;26:1191. [2] Najmaldin A, Tan HL. Early experience with laparoscopic pyloromyotomy for infantile hypertrophic pyloric stenosis. J Pediatr Surg 1995;30:37. [3] Hall NJ, Van Der Zee J, Tan HL, et al. Meta-analysis of laparoscopic versus open pyloromyotomy. Ann Surg 2004; 240:774. [4] St Peter SD, Holcomb III GW, Calkins CM, et al. Open versus laparoscopic pyloromyotomy for pyloric stenosis: A prospective, randomized trial. Ann Surg 2006;244:363.
320
j o u r n a l o f s u r g i c a l r e s e a r c h 1 7 8 ( 2 0 1 2 ) 3 1 5 e3 2 0
[5] Hall NJ, Pacilli M, Eaton S, et al. Recovery after open versus laparoscopic pyloromyotomy for pyloric stenosis: a double-blind multicentre randomised controlled trial. Lancet 2009;373:390. [6] Leclair MD, Plattner V, Mirallie E, et al. Laparoscopic pyloromyotomy for hypertrophic pyloric stenosis: A prospective, randomized controlled trial. J Pediatr Surg 2007; 42:692. [7] Eaton S, Hall NJ, Pierro A. Zero-total event trials and incomplete pyloromyotomy. J Pediatr Surg 2009;44:2434. [8] Sola JE, Neville HL. Laparoscopic vs open pyloromyotomy: A systematic review and meta-analysis. J Pediatr Surg 2009;44: 1631. [9] Campbell BT, McLean K, Barnhart DC, et al. A comparison of laparoscopic and open pyloromyotomy at a teaching hospital. J Pediatr Surg 2002;37:1068.
[10] Bufo AJ, Merry C, Shah R, et al. Laparoscopic pyloromyotomy: A safer technique. Pediatr Surg Int 1998;13:240. [11] Campbell BT, McVay MR, Lerer TJ, et al. Ghosts in the machine: a multi-institutional comparison of laparoscopic and open pyloromyotomy. J Pediatr Surg 2007; 42:2026. [12] Kim SS, Lau ST, Lee SL, et al. Pyloromyotomy: A comparison of laparoscopic, circumumbilical, and right upper quadrant operative techniques. J Am Coll Surg 2005; 201:66. [13] BNF for children. London: British National Formulary Publications; 2008. [14] Kim SS, Lau ST, Lee SL, et al. The learning curve associated with laparoscopic pyloromyotomy. J Laparoendosc Adv Surg Tech A 2005;15:474.
Available online at www.sciencedirect.com
journal homepage: www.JournalofSurgicalResearch.com
Cost-effectiveness of laparoscopic versus open pyloromyotomy Emma V. Carrington, MD, Nigel J. Hall, PhD, Maurizio Pacilli, MS, David P. Drake, MD, Joseph I. Curry, MD, Edward M. Kiely, MD, Paolo De Coppi, MD, Agostino Pierro, MD, and Simon Eaton, PhD* UCL Institute of Child Health and Great Ormond Street Hospital for Children, London, United Kingdom
article info
abstract
Article history:
Background: Infantile hypertrophic pyloric stenosis can be corrected by either open (OP) or
Received 18 October 2011
laparoscopic pyloromyotomy (LP). LP may provide clinical benefits of reduced time to
Received in revised form
postoperative full feeds and reduced postoperative inpatient stay, but the cost effective-
2 January 2012
ness is not known. Our aim was to compare the cost effectiveness of laparoscopic and open
Accepted 18 January 2012
pyloromyotomy.
Available online 27 March 2012
Methods: OP and LP were compared in a multicenter randomized double-blind controlled trial, for which the primary outcomes were time to full feeds and time to discharge. In
Keywords:
order to undertake a detailed cost analysis, we assigned costs, calculated on an individual
Infants
patient basis, to laboratory costs, imaging, medical staff, medication, ward, operative, and
Pyloric stenosis
outpatient appointments for 74 patients recruited from one of the participating centers.
Pyloromyotomy
Data (mean SEM) were compared using linear regression analysis, adjusting for the
Laparoscopy
minimization criteria used in the trial.
Economic analysis
Results: Operation costs were similar between the two groups ($3,276 $244 LP versus
Cost effectiveness
$3,535 $152 OP). A shorter time to full feeds and shorter hospital stay in LP versus OP patients resulted in a highly significant difference in ward costs ($2,650 $126 LP versus $3,398 $126 OP; P ¼ .001) and a small difference in other costs. Overall, LP patients were $1,263 (95% confidence interval $395e$2,130; P ¼ .005) less expensive to treat than OP patients. Sensitivity analyses of laparoscopic hardware usage and of incomplete pyloromyotomy indicated that LP was consistently less expensive than OP. Conclusions: LP is a cost-effective alternative to OP as it delivers improved clinical outcome at a lower price. ª 2012 Elsevier Inc. All rights reserved.
1.
Introduction
Pyloromyotomy for pyloric stenosis has recently been performed by the laparoscopic approach [1,2]. A number of studies have examined clinical outcomes in
open pyloromyotomy (OP) compared with laparoscopic pyloromyotomy (LP), either on a retrospective nonrandomized basis (meta-analyzed in [3]) or, more recently, in the context of prospective randomized controlled trials [4e6]. It appears from these data that LP leads to earlier time to full feeds and/
* Corresponding author. UCL Institute of Child Health and Great Ormond Street Hospital for Children, London, United Kingdom. Tel.: þ44 20 7905 2158; fax: þ44 20 7404 6181. E-mail address: [email protected] (S. Eaton). 0022-4804/$ e see front matter ª 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.jss.2012.01.031
316
j o u r n a l o f s u r g i c a l r e s e a r c h 1 7 8 ( 2 0 1 2 ) 3 1 5 e3 2 0
or discharge from hospital, although there is some concern that there may be a higher rate of incomplete pyloromyotomy following LP compared with OP [5,7,8]. In addition to clinical outcomes, evidence of cost effectiveness in patient management is becoming a fundamental part of good patient care, and is a necessary step in the evaluation of new treatments and techniques. Cost effectiveness of an intervention involves consideration not only of the monetary cost, but also of the attributable benefit derived from improved outcome. Thus, what is good for the patient is also good for the surgeon and good for the institution. A comprehensive literature review revealed that there has been no robust economic analysis of LP versus OP. A number of studies have included a basic cost analysis based on hospital charges data, some of which report higher hospital charges with LP compared with OP without statistical significance [9,10] and others of which report either no difference [11] or reduced hospital charges with LP [12]. In their randomized controlled trial of LP versus OP, St Peter et al. found no difference in cost between LP and OP, although they did not report cost data [4]. The aim of this study was to investigate the cost effectiveness of LP versus OP in patients with pyloric stenosis, using a subset of patients from a single center of a recently published multicenter randomized controlled trial [5]. We attributed costs on a per-patient basis using actual inpatient time, theater time, pathology, imaging, and medication costs. We undertook sensitivity analyses to examine the effects of laparoscopic equipment costs and different rates of incomplete pyloromyotomy on cost effectiveness.
2.
Methods
We have reported clinical outcomes from a double-blind, multicenter, international randomized controlled trial with prospective data collection [5]. The centers in this study were in five countries with great differences in healthcare systems and consequently costs could not be evaluated in a uniform fashion across all centers. In order to collect robust uniform economic data, the present evaluation used only patients recruited from one center, Great Ormond Street Hospital for Children (GOSH), UK.
As part of that study, patients were randomized with weighted minimization including center as a minimization criterion and were treated according to a standardized clinical care pathway designed specifically for the trial. This pathway stipulated all aspects of patient care including postoperative feeding regime, postoperative analgesia, and criteria for discharge and has been reported previously. Therefore all patients were treated identically, regardless of operative approach, with any treatment effect being due to different operative approach rather than other confounding factors. The feeding protocol, described fully in the previous paper [5], was 15 mL of formula or a short breast feed at 6 h postoperatively followed by (if tolerated) a full feed of either formula milk or a breast feed 1 h later. Hence the time to first tolerated full feed, by definition, could not be less than 7 h. We did not use any enteral fluid other than formula or breast milk. Economic data were derived from a combination of data collected prospectively from the original trial (inpatient and operative durations) and supplementary data collected retrospectively from individual patient records (per-patient medication and investigations). Ethical approval was obtained for this economic analysis (approval number 03SG45). The original trial was registered with clinicaltrials.gov (identification number NCT00144924). Of 84 patients recruited into the trial from this center, 10 sets of clinical notes were unavailable at the time of analysis for further retrospective review and therefore this economic analysis is based on 74 patients (41 LP, 33 OP). Demographic details and other clinical outcome measures in relation to the subgroup of infants included in this study are shown in Table 1. Economic outcome was calculated as the direct costs associated with in hospital treatment. The overall cost was subdivided into the following categories: 1) ward costs; 2) medical staff costs; 3) medication costs; 4) laboratory costs; 5) imaging costs; 6) operative costs; and 7) outpatient costs. These costs were obtained from the hospital finance department.
2.1.
Ward and staffing costs
Ward costs, representing the cost of running the neonatal surgical ward, were calculated for each patient based on
Table 1 e Demographics and clinical outcomes of patients included in the economic analysis. LP (n ¼ 41) Age (d) Weight (kg) Bicarbonate at presentation (mmol/L) Gestation at birth (wk) Feeding type Duration of symptoms <48 h Time to first tolerated feed (h) Time to full feeds (h) Time to discharge (h)
31 (27e36, 15e65) 3.7 (3.3e4.1, 2.1e59.1) 31 (27.8e36, 22e59.1) 40 (39e40, 32e42) 11 breast 30 mixed/bottle 1/41 10.2 (7.2e16.6, 5.5e31.8) 17.75 (11.6e20, 9.5e48.2) 27.6 (23.4e45.8, 13e71.5)
OP (n ¼ 33)
P value
(21e38, 13e83) (3.3e4.1, 2.4e6.6) (25.7e34.1, 19e57.7) (39e40, 36e42) 4 breast 29 mixed/bottle 0/33 13.2 (7.1e27.3, 6.2e54.7) 26.2 (16.6e44.2, 9.2e107.4) 45.5 (28.1e55, 19.2e129.2)
n.s. n.s. n.s. n.s. n.s.
33 3.7 30 40
n.s. .002 .001 .001
Data are median (interquartile range, range), or n (duration of symptoms, feed type). Data were compared using linear regression, adjusting for the minimization criteria, or Fisher exact test (duration of symptoms, feed type). A P value of <.05 was considered to be significant.
j o u r n a l o f s u r g i c a l r e s e a r c h 1 7 8 ( 2 0 1 2 ) 3 1 5 e3 2 0
length of in-patient stay (time of admission to time of discharge). This took into account most direct costs associated with ward care, including nurse staffing (but not medical staffing), stock medications (including paracetamol and codeine), indwelling devices and fluids, laundry, equipment maintenance, and staff training. This also included an attributable portion of fixed overhead costs of hospital ground rent, utility supply, and ancillary staff costs. Medical staff costs were also based on total admission time and were calculated on the basis of average salaries for consultants, and for training and other junior staff. Medical staff time per patient (and therefore cost) was divided into the fixed cost of “medical admission” (90 min of junior staff time for initial management, consent, and ordering of initial investigations) and the variable cost of “daily duties” (10 min of consultant input and 60 min of junior doctor input per 24 h). These were attributed on a per-patient basis according to length of stay on the ward.
2.2.
Operative costs
Operative costs were calculated using a combination of prospectively collected data from the trial and data from individual theater records. Costs were calculated as the total of: cost per minute of theater time, consumables, theater medical staff costs, theater reusables, and hardware. Cost per minute (calculated for each patient from induction of anesthesia to recovery) of theater time took into account most direct costs associated with theaters (anesthetic medications and equipment, salaries of nursing staff, operating department practitioners, administration staff, and theater porters). It also included maintenance of hardware (although not purchase of large hardware items) and an attributable portion of fixed overhead costs. Theater hardware (diathermy and laparoscopic equipment) purchase costs were calculated assuming a 5-y lifetime and using a straight line depreciation method. Hardware cost per operation was attributed assuming that each laparoscopic equipment set was used in theaters four times per day and each diathermy machine was used in theaters six times per day, and that diathermy was used in both LP and OP whereas the laparoscopic equipment set was used only in LP. A standardized list of theater consumables was created for both OP and LP and it was assumed that this fixed number was used in each case. It was assumed that each case (both LP and OP) involved the participation of one consultant surgeon, one junior/trainee surgeon, one consultant anesthetist, and one junior/trainee anesthetist, using the costs per minute described in Ward and staffing costs above. Theater reusable costs were calculated as the cost of sterilization and redelivery and were costed on a per-patient basis.
2.4.
Laboratory costs
Laboratory costs (microbiology, hematology, and biochemistry) were taken from NHS Reference Costs 2006e07 and were derived retrospectively based on the actual tests ordered for each patient.
2.5.
Imaging costs
Imaging costs were calculated to include purchase and maintenance of equipment, the time-attributable salary of one consultant radiologist, and a proportion of fixed overhead costs. The actual number of tests per patient was derived from the clinical records.
2.6.
Outpatient costs
Outpatient visits were assumed to take 15 min of a consultant surgeon’s time. One outpatient visit was attributed to each patient.
2.7.
Data analysis
All data were initially collected in pounds sterling (£). We converted all costs to U.S. dollars ($) at an exchange rate of £1 ¼ $1.87, which was the average (and also the median) foreign exchange rate over the time of the study. All outcomes, both clinical and economic, were compared by regression analysis, adjusting for the minimization criteria used in the trial design (age, weight, gestational age at birth, bicarbonate at presentation, feed type, and duration of symptoms), and for the randomization group on an intentionto-treat basis (SPSS version 15, SPSS Inc, Chicago, IL). P < .05 was considered significant.
3.
Results
3.1.
Clinical outcomes
The main clinical outcomes of the whole trial, demonstrating significant advantages of LP over OP, have been reported previously [5]. The primary outcomes of the subgroup of patients included in this economic analysis were similar to those from the main trial and are shown in Table 1. There were no conversions to open in the LP group and one mucosal laceration (repaired intraoperatively) in both the LP and OP groups. There were no incomplete pyloromyotomies requiring repeat operation.
3.2. 2.3.
317
Costs
Medication costs
Simple analgesia (i.e., paracetamol and codeine phosphate) was taken from the ward medication cost and therefore accounted for in the ward costs. Discharge medication was costed on a per-patient basis separately according to the BNF for Children 2008 manual [13] and derived retrospectively from clinical discharge records.
Overall cost of LP was on average $1,263 cheaper than OP (adjusted mean difference; Table 2, Fig. 1); this was highly significant (P ¼ .005). A regression analysis, adjusting for the minimization criteria (age, weight, gestational age at birth, bicarbonate at presentation, feed type, and duration of symptoms), used in the main trial demonstrated that none of these was found to significantly affect overall cost.
318
j o u r n a l o f s u r g i c a l r e s e a r c h 1 7 8 ( 2 0 1 2 ) 3 1 5 e3 2 0
Table 2 e Costs for each category. Costs Ward Medical staff Medication Pathology Imaging Operative Outpatientsa Overall
LP cost $ 2,650 501 0.4 28 115 3,276 45 6,615
126 13 0.1 2 20 244 269
OP cost $ 3,398 580 1.9 28 128 3,535 45 7,715
Adjusted mean difference (95% CI) $ 932 98 1.5 1 15 215
126 28 1.5 4 24 152
(e1,481, e384) (e156, e41) (e4.4, 1.3) (e11, 7) (e79, 49) (e847, 418)
1,263 (e2,130, e395)
346
P value .001 .001 .29 .74 .65 .5 Not applicable .005
Data are mean SEM; P values were calculated using linear regression, adjusting for the minimization criteria. A P value of <.05 was considered to be significant. a Defined as a fixed cost per patient.
Ward costs (average 42% in LP and 43% in OP) and operative costs (average 47% in both LP and OP) were the main contributors to overall costs. Ward costs were significantly lower in the LP group when compared to the OP group, reflecting the shorter length of stay (adjusted mean difference ¼ $932, P ¼ .001). The shorter length of stay also resulted in significantly cheaper medical staff costs (adjusted mean difference $98, P ¼ .001). No significant difference was found in imaging or laboratory costs (P ¼ .65 and P ¼ .74 respectively) between LP and OP. Medication costs were minimal and similar between groups (P ¼ .29). No significant difference was found in the operative costs (P ¼ .5) even though significantly more consumable sterile sets were used per LP (P ¼ .007). The attributable cost of the laparoscopic hardware was found to be $6.83 per procedure, which was negligible when compared to a median total operative cost of $3,276 per LP.
Fig. 1 e Costs for open and laparoscopic pyloromyotomy. Graph shows overall costs for open pyloromyotomy (OP, n [ 33) and laparoscopic pyloromyotomy (LP, n [ 41). In addition, the overall costs were divided into operative, ward, and other costs (medical staff costs plus medication costs, pathology costs, imaging costs, and outpatient costs). Data are shown as mean ± SEM, and the P values shown are for linear regression after adjusting for the minimization criteria. n.s. [ not significant.
4.
Discussion
Using a robust economic analysis technique, we have found that laparoscopic pyloromyotomy is significantly less expensive than open pyloromyotomy. The major contributory factor to the lower cost was the decreased length of stay and, therefore, lower ward costs and medical staff costs. Although laparoscopy is usually assumed to be more expensive on the basis of hardware costs and potentially longer operative time, we found no significant difference in operative costs between the two procedures. Economic analysis necessitates making assumptions in order to derive and attribute costs. It is therefore important to perform a sensitivity analysis to test the effect of these assumptions and, therefore, the robustness of the conclusions. The large initial capital outlay for laparoscopic equipment renders the economic analysis sensitive to the usage of laparoscopic equipment for other procedures. For example, if a set of laparoscopic equipment were to be purchased solely for use in pyloromyotomies, the operative cost per procedure would be considerably greater. In a hospital treating approximately 30 patients per year laparoscopically for pyloromyotomy, this would increase the cost per operation by approximately $374. In such a scenario, LP would still be cheaper overall than OP, but only by $888, and the difference between the procedures would only just be significant (P ¼ .045). Clearly, the economic effectiveness of laparoscopic equipment depends on its frequency of use. One of the main concerns regarding laparoscopic pyloromyotomy is the potential for an increased risk of incomplete pyloromyotomy [3,6,8]. Although the risk of incomplete pyloromyotomy has not so far been shown to be significantly greater in any of the three published randomized controlled trials comparing LP and OP [4e6], there was a tendency for more in the laparoscopic group in two of these trials [5,6]. In the single-center cohort studied for the present analysis (from the larger cohort of Hall et al. [5]), there were no incomplete pyloromyotomies. It is therefore of interest to undertake a sensitivity analysis to try to determine the potential impact that a higher incidence of incomplete pyloromyotomy would have in the economic analysis. For this sensitivity analysis, we chose to model the effect of a fairly extreme difference in the rate of incomplete pyloromyotomy:
j o u r n a l o f s u r g i c a l r e s e a r c h 1 7 8 ( 2 0 1 2 ) 3 1 5 e3 2 0
a 5% rate of incomplete pyloromyotomy in the laparoscopic group, which is very high in comparison with the published rate of 3.5%, and a rate of zero in the open group. For each patient with this complication, we assumed an additional ward stay of 5 d, and an additional operation with associated imaging, pathology, and medical staff costs. On the basis of these assumptions, each incomplete pyloromyotomy would cost $9,500. Despite these additional costs, laparoscopic pyloromyotomy is still, on average, cheaper, with a mean difference in cost of $636, compared with an adjusted mean difference of $1,263 in the actual analysis. However, economic effectiveness is a combination of clinical outcome and the financial cost of achieving that clinical outcome. Whether a 5% rate of incomplete pyloromyotomy can be considered economically effective is a subjective matter. There were no wound infections in the patients reported in this study, although there is a potential difference in the rate of wound infection between LP and OP. If wound infections, or other wound complications, did occur, then this could impact length of stay and cost of medication for antibiotics, and also potentially lead to additional out-of-hospital costs if identified after discharge. Medical staff costs were based on assumptions concerning daily duties, and we acknowledge that not knowing accurately how many times each patient was reviewed is a potential shortcoming of this study. However, the cost of medical staffing calculated as described in the methods was only about 7% of the total cost; therefore, only very large differences in actual staffing time between OP and LP would significantly alter the conclusion that LP is cheaper than OP. This calculation was based only on time at the bedside, which is merely one component of medical staff duties. Other medical staff duties (including review of investigations, audit, discussion with relatives, teaching and training, etc.) are less easily attributable on a per-patient basis but could also influence economic decision making. In order to assess the potential impact of these activities, we estimated the cost associated with these other duties. Assuming an additional 2 h of consultant time and 2 h of junior staff time per patient, a cost of $507 per patient additional to the costs described in Table 2 would be incurred. This may give a more accurate reflection of total cost (i.e., $7,122 per patient LP versus $8,222 per patient OP). However, this amount is the same for OP and LP and therefore does not alter the finding that LP is less expensive. This analysis has the advantage of being conducted at a single center, allowing costs to be estimated consistently for each patient using the same data sources and assumptions. The disadvantage of this approach is that the conclusions are not necessarily applicable in other centers due to potential differences between institutions. For example, Great Ormond Street Hospital is located in central London, so fixed costs (associated with location) and staffing costs may be significantly higher than in other more regional centers in the UK. Although absolute costs at other institutions are inevitably going to be different, unless relative expense between categories is markedly different (e.g., theater costs are proportionally higher than ward costs at a given institution), it is likely that the principal findings are generalizable across other centers.
319
In the main report of our trial, a limited cost analysis was undertaken in which there appeared to be no significant difference in cost between the OP and LP groups [5]. However, in that analysis, all costs were estimated rather than taken on a per-patient basis, and costs were combined from centers in different continents and, therefore, very different healthcare systems. We believe that the conclusions derived from this study, in which a much more detailed analysis was made on patients from a single center, are much more robust than those previously reported. The costs associated with LP and OP have previously been investigated in several papers. Kim et al., in 2005, using a nonrandomized cohort of patients receiving LP, right-upper quadrant OP, or circum-umbilical OP, reported that LP was significantly associated with lower operative costs than circum-umbilical OP; however, costs associated with the length of hospital stay were not reported [14]. Campbell et al. [9] and Bufo et al. [10], both reporting hospital charges on nonrandomized patients from single centers, found no significant difference in costs between LP and OP. Campbell et al., studying a multi-institutional, nonrandomized group of patients, reported a significant difference in hospital charges in centers that performed LP compared with centers performing OP [11]. However, this may reflect a difference in hospital billing policies or in fixed institutional costs rather than a real difference between surgical approaches. We believe that the analysis we have undertaken is more rigorous, and the conclusions more robust, than these previous reports.
5.
Conclusions
We conclude that in a center with suitable laparoscopic experience, performing a variety of operations using a laparoscopic approach, LP is significantly less expensive and therefore more economically effective than OP. If experience is less (possibly leading to a much higher rate of incomplete pyloromyotomy) or laparoscopic equipment is not frequently used, these conclusions may not be valid.
Acknowledgments The Sir Halley Stewart Trust is thanked for a grant supporting this work. No authors have any conflict of interest.
references
[1] Alain JL, Grousseau D, Terrier G. Extramucosal pylorotomy by laparoscopy. J Pediatr Surg 1991;26:1191. [2] Najmaldin A, Tan HL. Early experience with laparoscopic pyloromyotomy for infantile hypertrophic pyloric stenosis. J Pediatr Surg 1995;30:37. [3] Hall NJ, Van Der Zee J, Tan HL, et al. Meta-analysis of laparoscopic versus open pyloromyotomy. Ann Surg 2004; 240:774. [4] St Peter SD, Holcomb III GW, Calkins CM, et al. Open versus laparoscopic pyloromyotomy for pyloric stenosis: A prospective, randomized trial. Ann Surg 2006;244:363.
320
j o u r n a l o f s u r g i c a l r e s e a r c h 1 7 8 ( 2 0 1 2 ) 3 1 5 e3 2 0
[5] Hall NJ, Pacilli M, Eaton S, et al. Recovery after open versus laparoscopic pyloromyotomy for pyloric stenosis: a double-blind multicentre randomised controlled trial. Lancet 2009;373:390. [6] Leclair MD, Plattner V, Mirallie E, et al. Laparoscopic pyloromyotomy for hypertrophic pyloric stenosis: A prospective, randomized controlled trial. J Pediatr Surg 2007; 42:692. [7] Eaton S, Hall NJ, Pierro A. Zero-total event trials and incomplete pyloromyotomy. J Pediatr Surg 2009;44:2434. [8] Sola JE, Neville HL. Laparoscopic vs open pyloromyotomy: A systematic review and meta-analysis. J Pediatr Surg 2009;44: 1631. [9] Campbell BT, McLean K, Barnhart DC, et al. A comparison of laparoscopic and open pyloromyotomy at a teaching hospital. J Pediatr Surg 2002;37:1068.
[10] Bufo AJ, Merry C, Shah R, et al. Laparoscopic pyloromyotomy: A safer technique. Pediatr Surg Int 1998;13:240. [11] Campbell BT, McVay MR, Lerer TJ, et al. Ghosts in the machine: a multi-institutional comparison of laparoscopic and open pyloromyotomy. J Pediatr Surg 2007; 42:2026. [12] Kim SS, Lau ST, Lee SL, et al. Pyloromyotomy: A comparison of laparoscopic, circumumbilical, and right upper quadrant operative techniques. J Am Coll Surg 2005; 201:66. [13] BNF for children. London: British National Formulary Publications; 2008. [14] Kim SS, Lau ST, Lee SL, et al. The learning curve associated with laparoscopic pyloromyotomy. J Laparoendosc Adv Surg Tech A 2005;15:474.