Practice variation in gastroschisis: Factors influencing closure technique Jennifer Stanger, Noosheen Mohajerani, Erik D. Skarsgard PII: DOI: Reference:
S0022-3468(14)00166-3 doi: 10.1016/j.jpedsurg.2014.02.066 YJPSU 56719
To appear in:
Journal of Pediatric Surgery
Received date: Accepted date:
2 February 2014 13 February 2014
Please cite this article as: Stanger Jennifer, Mohajerani Noosheen, Skarsgard Erik D., Practice variation in gastroschisis: Factors influencing closure technique, Journal of Pediatric Surgery (2014), doi: 10.1016/j.jpedsurg.2014.02.066
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Practice Variation in Gastroschisis: Factors Influencing Closure
SC
RI P
T
Technique
Jennifer Stanger, Noosheen Mohajerani, Erik D. Skarsgard,
MA
NU
And the Canadian Pediatric Surgery Network (CAPSNet)
ED
Department of Surgery, Division of Pediatric Surgery, BC Children’s Hospital, University
AC
CE
PT
of British Columbia, Vancouver, BC, Canada
Corresponding Author: Erik D. Skarsgard, MD Division of Pediatric Surgery, BCCH K0-110 ACB, 4480 Oak Street Vancouver, BC V6J 4K7 (604) 875-2548; fax: (604) 875-2721 E-mail:
[email protected] 1
ACCEPTED MANUSCRIPT Abstract Background: Little is known about the factors influencing surgical practice variation in
T
newborns with gastroschisis. The purpose of this study was to correlate prognostic
RI P
variables with the intended and actual abdominal closure technique and assess related outcomes.
SC
Methods: GS cases were abstracted from a national database. Variables evaluated
NU
included GA, BW, bowel injury severity (GPS), neonatal illness severity (SNAP-II), inborn status, center volume and training status, and admission time. Evaluated
MA
outcomes by closure method included duration of TPN, LOS, and complications. Descriptive, univariate, and analyses were conducted.
ED
Results: The cohort consisted of 679 patients. 372 (55%) underwent attempted PR, of
PT
which 300 (81%) were successful, while 307 (45%) had a silo placed intentionally. Patients undergoing attempted PR were more likely to be inborn, have daytime
CE
admissions, and higher SNAP-II scores. Successful PR was predicted by low risk GPS
AC
and high volume center. With the exception of higher rates of SSI in the planned silo group, outcomes in the successful PR and planned silo groups were comparable. Conclusion: Practice variation related to type of closure is predicted by situational and institutional factors (outborn, nightime admission, and center volume), while outcome variation is attributable to patient factors rather than practice variation.
Keywords: Gastroschisis, Surgical Management, Practice Variation, Risk-Adjusted Outcome variation
2
ACCEPTED MANUSCRIPT
T
Introduction
RI P
Surgical decision-making regarding the timing and type of abdominal closure in newborns with gastroschisis (GS) is potentially influenced by patient, situation and
SC
hospital-specific factors. In an era when primary repair was the preferred method of
NU
abdominal closure, a hand sewn silo followed by delayed closure was recognized as an important rescue technique when abdominal wall closure could not be obtained due to
MA
viscero-abdominal disproportion and the risk of development of abdominal compartment syndrome.[1] With the advent of the spring-loaded silo, routine, non-selective silo
ED
placement has been advocated as a technique yielding comparable outcomes with the
PT
added advantage of avoiding out of hours surgery.[2,3] Little is known about the factors
achieved.
CE
which determine choice of closure technique, and the comparability of the outcomes
AC
Significant practice variation exists in the peri-natal care of newborns with GS which may be a contributing factor to outcome variation.[4] Protocol – driven patient care in gastroschisis is rare.[5] While most modern series report survival rates of greater than 90%, the associated morbidities are less well reported.[6,7,8] Efforts to standardize patient care are limited by a lack of appropriately risk-adjusted studies to predict which management strategies are associated with superior outcomes. With the rates of GS increasing and the potential increased burden of disease and associated costs of care, improved predictive models leading to standardized care pathways are needed. [1]
3
ACCEPTED MANUSCRIPT The purpose of this study was to use the population-based CAPSNet database to analyze the factors involved in the surgical management of newborns with GS, with an
SC
Methods
RI P
factors (if any), that contribute to outcome variation.
T
intent of understanding the determinants of surgical practice variation and identifying
NU
The Canadian Pediatric Surgery Network (CAPSNet) consists of 17 perinatal / surgical centers that provide population-based care for GS in Canada. The CAPSNet
MA
database captures prospective data that describe prenatal diagnosis and postnatal treatment and outcome up to the point of hospital discharge. The database was
ED
designed specifically for outcomes research and contains fields that allow discrimination
PT
of risk variables and treatment, timing and type of surgical management, as well as relevant clinical outcomes. Details of data collection, privacy protection, and use of
CE
aggregate data for research purposes have been described previously [6]. For the
AC
purpose of this study, the CAPSNet GS database was interrogated for all “completed” cases (i.e. those who survived or died to index hospitalization discharge) accrued between May 2005 and December 2011. Data abstracted included neonatal risk factors (gender, gestational age {GA}, birth weight {BW}), delivery factors (mode of delivery, inborn status, time of admission) and surgical factors (closure intent/success, GS case volumes of the treating center, and whether the treating center hosts a pediatric surgical training program). Patient risk stratification was also performed using the validated Score for Neonatal Acute Physiology-II (SNAP-II) and the Gastroschisis Prognostic Score (GPS). [9, 10] GPS
4
ACCEPTED MANUSCRIPT uses a standardized visual assessment of the intestine by the surgeon at initial consultation to categorize the baby as low or high risk for adverse outcome. Time of
T
day of admission was classified as daytime if the baby arrived between 0800h and
RI P
2000h. Outcomes of interest included length of stay (LOS), timing of enteral feeding, duration of total parenteral nutrition (TPN), duration of mechanical ventilation and
SC
complications including culture-proven bacteremia and surgical site infection (SSI).
NU
The primary study cohort compared patients that underwent primary repair (PR) (defined as patients undergoing attempted abdominal closure within 6 hours) to those
MA
for whom a silo was placed to facilitate delayed closure. This comparison sought factors that were predictive of choice of one closure technique over the other, as well as
ED
a comparison of outcomes. A subset analysis was conducted on patients in whom PR
PT
was attempted: Patients undergoing successful PR were compared to those for whom attempted PR was unsuccessful, to see if there were variables which predicted success,
CE
and to observe outcome differences between groups.
AC
Data were analysed using SPSS version 21 for Windows software (IBM software Inc, Chicago, IL). Descriptive, univariate and multivariable regression analyses were performed. Student’s t-test was employed for continuous variables, and the chi-squared test for categorical variables. A double sided p-value of < 0.05 was considered significant.
Results Between May 2005 and December 2011, 694 babies with gastroschisis had “completed” entries in the CAPSNet database, of which 679 were included in the study
5
ACCEPTED MANUSCRIPT cohort. 15 patients were excluded due to missing data. The average GA of all patients was 35.8 weeks and the average BW 2557 grams. Males comprised 52.1% (n = 360) of
T
the study population. Over half of the patients were classified as inborn (n = 398,
RI P
57.6%) and 52.8% (n = 365) were managed in high volume centers (defined as a center that treats 9 or more patients with gastroschisis each year). Time of day of admission
SC
was only available for patients treated between 2005 and 2010 (n = 540), with 54.4% (n
NU
= 294) being born during the daytime.
The 679 patient study cohort was classified according to the intended and actual
MA
surgical technique used to achieve abdominal closure, as summarized in Figure 1. Closure intent was PR in 372 patients, and of these, closure was successful in 300
ED
(81%). A total of 379 patients underwent silo placement and delayed abdominal closure,
PT
and of these, 72 had failed initial attempts at PR. Comparison of the groups according to actual closure technique demonstrated similarities in GA, BW, gender, mode of
CE
delivery and GPS (Table 1). Patients undergoing PR had a higher mean SNAP-II score
AC
(11.1 vs. 7.21, p 90% are typically reported, there remains
T
significant short and occasionally, long term morbidity in some patients.[6,7] An
RI P
evolution in surgical treatment has been the introduction of the pre-formed ring silo which offers the ease and simplicity of bedside placement as well as the avoidance of
SC
out of hours surgery. Unfortunately, there is little existing evidence to guide surgical
NU
decision-making which likely contributes to the observed, widespread variation in surgical practice within and between pediatric surgical centers. [4, 5]
MA
Previous reports have sought to address this question of optimal timing and technique of GS defect closure, and the results are conflicting. Several authors have
ED
found that silo placement and delayed closure as a preferred closure technique yield
PT
comparable outcomes to PR [2,3], while others have identified an association between delayed closure, prolonged LOS and increased rates of complication.[13,14]. Very few
CE
studies have sought to identify predictors of treatment. Our findings are supported by
AC
the previous studies by Nasr et al. and Quirk et al. that demonstrated location of delivery as a significant predictor of outcome in newborns with GS. [14, 15]. In our study, NICU admission time was a predictor of surgical management with patients being born at night (8pm to 8am) more likely to undergo delayed closure. SNAP-II is an aggregate neonatal illness severity score measured within 12 hours of NICU admission that reflects the hemodynamic, oxygenation, perfusion and neurologic profile of an infant admitted to the NICU. [9] The observation of higher SNAP-II scores amongst infants undergoing PR seems paradoxical, and may be an issue of timing of measurement rather than a true reflection of the illness severity of a newborn with GS.
8
ACCEPTED MANUSCRIPT If measurements are made after PR, it is possible that what is being measured are the acute physiologic derangements resulting from operative intervention, which may
T
represent a source of type 1 error in the study. It should be noted that even though the
RI P
SNAP-II scores were significantly elevated in the PR group this did not translate into inferior outcomes compared to the delayed surgery group.
SC
The secondary aim of this study was to identify factors that predict successful
NU
PR. We observed that low risk GPS, inborn status and treatment in a pediatric surgical training center all predicted successful PR on analysis. Furthermore those patients that
MA
were successfully closed primarily had significantly better outcomes, with reduced mechanical ventilation requirements, improved enteral nutritional recovery, decreased
ED
infectious complication. Length of stay approached statistically significant difference,
PT
with an absolute difference of 20 days. Our study provides further validation of the prognostic value of GPS in that it not only predicts outcome but may predict best
CE
surgical management for patients in the high risk group.
AC
While the patient cohort used in this study is sizeable, our study reflects the limitations of an observational dataset. Although we have a datafield that captures the consulting surgeon’s “treatment intent”, the surgeon out of necessity has seen the baby before making that declaration. It is impossible to know what factors might bias a surgeon’s decision to attempt PR or place a silo. It is for this reason that we conducted our analysis based on the actual rather than planned approach. Time of NICU admission was only available for the cohort born before 2010. While statistically significant, the lower end of the confidence interval of the OR for time of day does approach 1. A larger sample size would make this result more robust. Although GPS
9
ACCEPTED MANUSCRIPT can be used to predict the success of planned PR, it is predictive only as a binary (low risk, high risk) rather than continuous variable and therefore it lacks quantitative
T
prediction for both success of treatment and likelihood of adverse outcome. Finally, the
RI P
utility of SNAP-II as a predictor of treatment is suspect, due to the uncertainty of whether it is measuring innate patient risk or the immediate impact of surgical closure
SC
on neonatal physiology.
NU
In this study we did not evaluate the duration of silo placement prior to successful delayed closure. It is possible that the comparable outcomes observed between the PR
MA
and silo group result from the heterogeneity in the silo group; arising from variability in situational factors (time of day of admission, outborn status etc.) rather than true patient
ED
factors (birth weight, GPS score etc.). Future work should attempt to stratify patients
PT
based on duration of silo placement prior to primary closure in order to further define the role of patient, institutional and situational factors in gastroschisis management.
CE
Practice variation in the surgical management of newborns with GS appears to
AC
be influenced by patient factors (GPS), situational factors (time of NICU admission), and by institutional factors (outborn status and GS case volumes). GPS is a valid marker of those infants who are most at risk of adverse outcomes and may have utility in predicting those patients who are most amenable to PR. Future risk-adjusted studies targeting outcome improvement in GS should seek to establish a relationship (if any) between variations in practice and outcome variation, with the intention of developing evidence-based “best practice” guidelines for surgical treatment and other essential GS care pathways.
10
ACCEPTED MANUSCRIPT References
T
1. Holland AJ, Walker K, Badawi N. Gastroschisis: an update. Pediatr Surg Int
RI P
2010; 26: 871-8.
2. Schlatter M, Norris K, Uitvlugt N, et al. Improved outcomes in the treatment of
SC
gastroschisis using a prefprmed silo and delayed repair approach. J Pediatr
NU
Surg 2003; 38: 459-64.
3. Lansdale N, Hill R, Gull-Zamir S, et al. Staged reduction of gastroschisis using
MA
preformed silos: practicalities and problems. J Pediatr Surg 2009; 44: 2126-29. 4. Baird R, Eeson G, Safavi A, et al. Institutional practice and outcome variation in
ED
the management of congenital diaphragmatic hernia and gastroschisis in
2011; 46: 801-7.
PT
Canada: a report from the Canadian Pediatric Surgery Network. J Pediatr Surg
CE
5. Aldrink JH, Caniano DA, Nwomeh BC. Variability in gastroschisis management:
AC
a survey of North American pediatric surgery training programs. J Surg Research 2012; 176: 159-63. 6. Skarsgard E, Claydon J, Bouchard S, et al. Canadian Pediatric Surgical Network: a population-based pediatric surgery network and database for analyzing surgical birth defects. The first 100 cases of gastroschisis. J Pediatr Surg 2007; 43: 30 – 34. 7. Kassa AM, Lilja HE. Predictors of postnatal outcome in neonates with gastroschisis. J Pediatr Surg 2011; 46: 2108-2114.
11
ACCEPTED MANUSCRIPT 8. Singh SJ, Fraser A, Leditschke JF, et al. Gastroschisis: determinants of neonatal outcome. Pediatr Surg Int 2003; 19: 260 – 265.
T
9. Richardson DK, Corcoran JD, Escobar GJ, et al. SNAP-II and SNAPPE-II:
RI P
simplified newborn illness severity and mortality risk scores. J Pediatr 2001; 138: 92-100.
SC
10. Cowan KN, Puligandla PS, Laberge J-M, et al. The Gastroschisis Prognostic
NU
Score: outcome prediction in gastroschisis. J Pediatr Surg 2012; 47: 801-7. 11. Kidd Jr JN, Jackson RJ, Smith SD, et al. Evolution of staged versus primary
MA
closure of gastroschisis. Ann Surg 2003; 237: 759-64. 12. Owen A, Marven S, Johnson P, et al. Gastroschisis: a national cohort study to
ED
describe contemporary surgical strategies and outsomes. J Pediatr Surg 2010;
PT
45: 1808-1816.
13. Weinsheimer RL, Yanchar NL, Bouchard SB, et al. Gastroschisis closure – does
CE
method really matter? J Pediatr Surg 2008; 43: 874-8.
AC
14. Nasr A, Langer J, The Canadian Pediatric Surgery Network. Influence of location of delivery on outcome in neonates with gastroschisis. J Pediatr Surg 2012; 47: 2022-2025.
15. Quirk JR JG, Fortney J, Collins HB, et al. Outcomes of newborns with gastroschisis: the effects of mode of delivery, site of delivery, and interval from birth to surgery. Am J Obstet Gynecol 1996; 174: 1134-8.
12
ACCEPTED MANUSCRIPT Figure 1 Legend: Flow diagram illustrates the distribution of gastroschisis patients according to intended and
AC
CE
PT
ED
MA
NU
SC
RI P
T
actual surgical treatment received
13
ACCEPTED MANUSCRIPT Figure 1: Intended and Actual Surgical Closure of 679 Gastroschisis patients
RI P
T
679 Intended Silo
NU
SC
Intended PR
MA
372
Successful PR
PT
ED
Unsuccessful PR
72
Total Silos
379
AC
CE
300
307
14
ACCEPTED MANUSCRIPT Table 1: Clinical demographics of patients undergoing Primary Repair (< 6 hours) compared to delayed closure Delayed Closure (n=379) 35.8
Birth Weight (grams)
2572
2545
SNAP-II score (mean)
11.1
7.2
Gender: Male (N, %) Site Volume: High Volume (N, %) GPS risk score: Low (N, %) Time of Admission: Night (N, %) Delivery Type: SVD (N, %) Surgeon Status: Trainee (N, %) Location of Delivery: Outborn (N, %)
174 (46.7)
178 (46.9)
0.99
162 (43.5)
199 (52.5)
0.04
240 (63.3)
0.47
106 (28.5)
RI P
SC
MA
224 (60.2)
p-Value
T
Gestational Age (weeks)
Primary Repair (n=372) 35.8
NU
Variable
0.99 0.50 < 0.001
139 (36.7)
0.02
229 (60.4)
0.79
133 (35.8)
135 (35.6)
0.88
61 (16.4)
228 (60.2)