Blood transfusions, increased rates of surgical NEC, and lower survival: a propensity score-matched analysis

    Blood Transfusions, Increased Rates of Surgical NEC, and Survival: A Propensity-Score Matched Analysis Arash J. Sayari, Jun Tashiro, Juan E. Sola, Eduardo A. Perez PII: DOI: Reference:

S0022-3468(16)00140-8 doi: 10.1016/j.jpedsurg.2016.02.052 YJPSU 57592

To appear in:

Journal of Pediatric Surgery

Received date: Accepted date:

17 February 2016 26 February 2016

Please cite this article as: Sayari Arash J., Tashiro Jun, Sola Juan E., Perez Eduardo A., Blood Transfusions, Increased Rates of Surgical NEC, and Survival: A Propensity-Score Matched Analysis, Journal of Pediatric Surgery (2016), doi: 10.1016/j.jpedsurg.2016.02.052

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

T

Blood Transfusions, Increased Rates of Surgical NEC, and Survival: A

RI P

Propensity-Score Matched Analysis.

NU

SC

Authors: Arash J. Sayari BS, Jun Tashiro MD MPH, Juan E. Sola MD, Eduardo A. Perez MD

Author Affiliations: Division of Pediatric Surgery, DeWitt-Daughtry Department of Surgery,

Address for Correspondence:

PT

Eduardo A. Perez MD

ED

MA

Leonard M. Miller School of Medicine, University of Miami, Miami, FL

Associate Professor, Division of Pediatric Surgery

CE

DeWitt-Daughtry Family Department of Surgery

AC

Leonard M. Miller School of Medicine University of Miami

1120 NW 14th Street, Suite 450K Miami, FL 33136 Telephone: (305) 243-2247 Fax: 305-243-5731 E-Mail: [email protected].

Author Disclosure Statement: The authors have no financial affiliations to disclose.

1

ACCEPTED MANUSCRIPT

Abbreviations: BT – blood transfusion; SNEC – surgical necrotizing enterocolitis; PS –

T

propensity score; MVA – multivariate analysis; GA – gestational age; OR – odds ratio; BPD –

RI P

bronchopulmonary dysplasia; LOS – length of stay; TC – total charges; PDA – patent ductus

AC

CE

PT

ED

MA

NU

SC

arteriosus.

2

ACCEPTED MANUSCRIPT Abstract: Purpose: We sought to investigate the association between blood transfusions (BT), rates of

RI P

T

necrotizing enterocolitis requiring surgical intervention (SNEC), and survival. Methods: BT in premature infants were identified in the Kids’ Inpatient Database (2003-2009).

SC

Propensity score (PS) matched analysis compared SNEC outcomes in BT vs. non-BT groups. Multivariate analyses (MVA) were performed to determine independent predictors of outcome.

NU

Results: Overall, 663,740 cases were identified, and 4.9% received BT. SNEC occurred in 493

MA

cases in the BT group, while SNEC occurred in 1049 cases in the non-BT group. PS analysis of 20,991 BT and 20,988 non-BT cases demonstrated higher SNEC rates with BT

ED

(OR: 1.21) vs. non-BT. SNEC with BT had lower survival rates (58%) vs. non-BT (67%). SNEC with BT had lower length of stay and total charges vs. non-BT.

PT

On MVA, SNEC mortality with BT increased for lower gestational age (GA), males (OR: 45.7),

CE

African Americans (OR: 64.4), and infants with cardiac anomalies (OR: 50.8) or bronchopulmonary dysplasia (OR: 177; BPD). Non-BT SNEC demonstrated higher mortality

AC

with lower GA and infants with BPD (OR: 6.56) or sepsis (OR: 3.66). Conclusion: On PS matched analysis, SNEC occurs at higher rates after BT and is associated with lower survival and lower resource utilization vs. SNEC without BT.

Keywords: blood transfusion, necrotizing enterocolitis, neonatal prematurity

3

ACCEPTED MANUSCRIPT Introduction Necrotizing enterocolitis (NEC) is one of the most common gastrointestinal medical

T

emergencies in newborns, and the incidence is inversely related to birth weight and gestational

RI P

age. The incidence of NEC has been reported to range from 1 to 7.7% of admissions to the neonatal intensive care unit (NICU), almost exclusively in premature infants.1 Appropriate

SC

measures should be taken to minimize the risk of developing NEC, because when NEC

NU

necessitates surgery (SNEC), the mortality reaches higher than 50%.2,3 Anemia of prematurity is another common entity in the NICU and is often seen in

MA

premature infants weighing less than 1000 g.4 When neonates require blood transfusions (BT), there is a reported risk of NEC. Since the first suggested association between packed red blood

ED

cell (pRBC) transfusions and NEC in 1987,5 a multitude of case-control and cohort studies have

PT

emerged supporting this notion, primarily in the last decade.6-16 In their meta-analysis, Mohamed and Shah observed that there was an increased risk of

CE

NEC within 48 hours of a BT.17 The meta-analysis only included 12 studies, and all but two were

AC

single or double center studies. Large population based studies are needed to provide multicenter data that can better estimate the overall incidence. To the best of our knowledge, the present study is the largest to evaluate the association between BT and NEC, and the first study to examine this relationship in cases of NEC that require surgery. We hypothesized that there would be a higher rate of SNEC in premature infants who received a BT.

Methods We used the Kids’ Inpatient Database (KID) to search for cases of packed red blood cell transfusions (BT) in premature (<37 weeks’ gestation) infants admitted at less than 8 days of life.

4

ACCEPTED MANUSCRIPT The KID samples national pediatric admissions, with information on up to 7.6 million weighted cases per year. For this study, we used data releases from 2003, 2006, and 2009. Diagnoses and

T

procedures were coded using the International Classification of Diseases, 9th revision, Clinical

RI P

Modification (ICD-9-CM). Cases were weighted to project national estimates, and all analyses were limited to available data.

SC

We defined cases as surgical NEC (SNEC) in this analysis where infants with NEC (ICD-

NU

9-CM 777.5x) required surgical intervention. A patient with NEC was considered to receive surgical intervention (SNEC) when they underwent laparotomy, bowel resection/ostomy, or

MA

peritoneal drain placement, whereas all other cases of NEC were considered to be medically managed.18 For the BT (ICD-9-CM 99.04) group, only SNEC cases occurring within 7 days

ED

post-BT were included, as these were cases of SNEC likely to be associated with BT. Cases of

PT

SNEC occurring outside of this window (prior to BT or ≥ 8 days post-BT) were excluded from analyses. For the non-BT group, all cases of SNEC were considered. Cases with disposition

CE

coded as ‘transfer to short term hospital’ or ‘other transfers, including skilled nursing facility,

AC

intermediate care, and another type of facility’ were excluded from analyses. Propensity score (PS)-matched analyses BT vs. non-BT were performed using the 1:1 nearest neighbor method. For each comparison group, a dataset containing a total of 58 variables were matched according to 47 covariates to construct a 1:1 fixed ratio matched cohort to compare cases of premature infants who received BT vs. those who did not receive BT. Between these matched cohorts, we compared rates of SNEC and outcomes following this diagnosis. PS values were assigned via multivariate logistic regression, according to demographic (gender, race), socioeconomic (payer status, median income quartile), hospital characteristics (bed size, location/teaching status, region, type), birth weight (in grams), gestational age (in weeks), as well

5

ACCEPTED MANUSCRIPT as common comorbidities, specifically cardiac anomalies, bronchopulmonary dysplasia, intraventricular hemorrhage, sepsis, respiratory distress syndrome, and multiple gestation, as

T

performed in a previous analysis.18 Additional risk-adjustment was performed using the

RI P

Elixhauser method, which is a standard set of comorbid conditions that has been validated in previous retrospective outcome studies.18-21 PS value assignment, case sorting, and matching

SC

were performed using MatchIt, version 2.4-20 (Cambridge, MA), a supplemental module for R,

NU

version 2.14.2 (R foundation for Statistical Computing; Vienna, Austria). After constructing two risk-adjusted comparison groups (BT vs. non-BT), we compared rates and outcomes (mortality,

MA

length of stay [LOS], and total charges [TC]) of SNEC in the BT vs. non-BT group. In a separate analysis, we used risk-adjusted multivariate analysis (MVA) to identify

ED

predictors of in-hospital mortality and higher resource utilization (LOS, TC) among cases of

PT

SNEC within BT or non-BT categories. Demographic, clinical, and hospital characteristic variables were used to construct logistic regression models. The binary, backward step-wise

CE

method was used to identify determinants of mortality, whereas ordinal regression models were

AC

constructed for determinants of resource utilization endpoints. Adjustment for comorbid risk factors were performed in a manner similar to the PS analysis, including variables for demographic, socioeconomic, hospital, and neonatal and Elixhauser comorbidities. TC values were standardized to 2009 US dollars (USD), according to inflation rates provided by the US Bureau of Labor Statistics.22 We set the significance level at α = 0.05, and SPSS Statistics, version 21.0 (IBM; Armonk, NY) was used to analyze data. This retrospective analysis was exempt from Institutional Review Board review.

6

ACCEPTED MANUSCRIPT Results Overall, 663,740 premature infants were identified, of which 32,241 (4.9%) received BT.

T

SNEC occurred in 493 (1.53%) cases in the BT group, while 1049 (0.17%) developed SNEC in

RI P

the non-BT group. For detailed demographic characteristics, see Table 1.

On 1:1 PS matched analysis of 41,979 cases, comprised of 20,991 BT (50.0%) and

SC

20,988 non-BT (50.0%) cases. The BT group demonstrated a significantly higher odds of SNEC

NU

(odds ratio [95% confidence interval]: 1.21 [1.03, 1.42]) vs. the non-BT group, p=0.017. SNEC in the BT group (58.1%) had a lower survival rate vs. SNEC in the non-BT group (66.7%),

MA

p=0.026. SNEC in the BT group had lower length of stay (median [interquartile range]: 47 [74]) vs. SNEC in the non-BT group (88 [70]), p<0.001. SNEC in the BT group (351,381 [578,577]

ED

USD) had lower total charges vs. SNEC in the non-BT group (557,481 [535,191] USD),

PT

p<0.001. For a tabular representation of these results, see Table 2. On MVA, SNEC mortality for BT group increased for lower gestational age (GA), male

CE

gender (OR: 45.7), African Americans (OR: 64.4), and associated comorbidities (cardiac

AC

anomalies [OR: 50.8], bronchopulmonary dysplasia [OR: 177; BPD]), p<0.05. Non-BT SNEC demonstrated higher mortality with lower GA and infants with BPD (OR: 6.6) or sepsis (OR: 3.7), p<0.05. For a graphic representation of these results, see Figure 1.

Discussion BT in premature neonates are not without risks, and there has been recent awareness of the risk of NEC after BT. Overall, 4.9% of the 663,740 premature infants in our study received a BT. This is significantly lower than a previous randomized trial in 2005 that reported transfusion

7

ACCEPTED MANUSCRIPT rates as high as 90%, although they only included patients less than 1300 g.23 A separate trial by Strauss et al. found that 5 to 6% of premature infants will receive at least one blood transfusion.24

T

In our study, of the premature infants who received a BT, 1.53% developed SNEC,

RI P

versus 0.17% in the group that did not receive a BT, which is consistent with only one previous study.12 Paul et al.12 found that when compared to infants not receiving a BT, 1.4% of premature

SC

infants developed NEC within 48 hours of a PRBC transfusion (OR: 2.3). In contrast to the low

NU

incidence in our study, a retrospective study by El-Dib et al.10 found that 56% of premature infants with NEC had received a BT within 48 hours preceding their diagnosis. Other studies

MA

reported rates of NEC after BT to be between 15.5% and 38%, all of which were found to be significantly higher than cases of NEC that did not involve a preceding BT.6,8,9,11,14-16 These

ED

studies examined rates of NEC, and we do not expect all patients with NEC to necessitate

PT

surgery,1 which may explain the lower rate of SNEC seen in our study. Premature infants with SNEC in the BT group had a significantly lower survival rate than

CE

the SNEC cases in the non-BT group. Christensen et al.7 reported a large, but non-significant

AC

difference in mortality rate in infants with NEC following BT versus those with NEC unrelated to a BT (40% versus 28%, respectively). Another study by Couselo et al.25 reported a mortality rate of 50% in infants with SNEC after within 48 hours of a BT versus those with SNEC who did not receive a BT, supporting the notion that survival is lower in patients with SNEC who had previously received a BT. Patients who developed SNEC in the BT group had lower length of stay and total charges vs. the non-BT group, p<0.001. This may reflect the differing pathophysiological manifestations of NEC in those who previously received a BT versus those who had not. Surgically reported findings in infants with NEC revealed an odds ratio of 0.5 for developing advanced disease in the

8

ACCEPTED MANUSCRIPT SNEC after BT versus SNEC without BT group.25 This supports the notion that the patients with SNEC after BT would have a shorter hospital stay. The same study also explained that more

T

ventilation support was provided to the SNEC group that received BT, which may hasten

RI P

recovery.25 However, the database used in this study precludes us from confirming a causative association.

SC

The suggestion that mortality increases with a lower gestational age and associated

NU

comorbidities has been extensively reported, and our results confirmed this in the SNEC after BT group. Interestingly, mortality was higher in the male gender (OR: 45.7) and in African

MA

Americans (OR: 64.4), associations not previously reported in this population. The MVA for the BT SNEC group also showed higher mortality in patients with associated comorbidities, such as

ED

cardiac anomalies. The relationship between an existing patent ductus arteriosus (PDA) and the

PT

development of NEC after a BT was reported as a case series by Sellmer et al.,26 which revealed a mortality rate of 75%. Furthermore, Baxi et al.27 examined rates of NEC in infants with

CE

congenital heart disease (CHD). They reported a higher death rate in the CHD and NEC group

AC

when compared to matched controls (p = 0.04), and found that 84.4% had received a BT prior to the diagnosis of NEC. BPD is another recognized complication of BTs in premature infants, and the present study found higher mortality in premature infants in both BT and non-BT groups, which suggests that BPD is associated with SNEC. There are limitations in the present study that should be addressed. There are obvious biases in a retrospective study that are avoided in a prospective, randomized study. However, there are ethical concerns with randomization of neonates,28 and retrospective studies provide valuable information on rare complications. Secondly, the data in the KID is made up by ICD-9 codes, which is dependent on appropriate billing and limits our inclusion and exclusion criteria,

9

ACCEPTED MANUSCRIPT such as gestational age or birth weight. Regardless, database studies are essential in pooling national data, and results may be more generalizable to the population as a whole. The KID

T

specifically provides detailed data regarding pediatric disease, and its advantages have been

RI P

previously reported.18,29,30

The complex pathophysiology of NEC leaves no surprise of the significant morbidity and

SC

mortality. With improving guidelines on transfusion practices, reports have suggested a

NU

relationship between PRBC transfusions and the development of NEC. Using a national database and using a PS matched analysis, we found that SNEC occurs at higher rates with BT and

MA

confers lower survival and worse outcomes, but results in lower resource utilization when

ED

compared to cases of SNEC without BT.

3.

4. 5.

6.

7.

8.

CE

2.

Kosloske AM. Epidemiology of necrotizing enterocolitis. Acta Paediatr Suppl. 1994;396:2-7. Gephart SM, McGrath JM, Effken JA, Halpern MD. Necrotizing enterocolitis risk: state of the science. Adv Neonatal Care. Apr 2012;12(2):77-87; quiz 88-79. Derienzo C, Smith PB, Tanaka D, Bandarenko N, Campbell ML, Herman A, et al. Feeding practices and other risk factors for developing transfusion-associated necrotizing enterocolitis. Early Hum Dev. May 2014;90(5):237-240. Widness JA. Pathophysiology of Anemia During the Neonatal Period, Including Anemia of Prematurity. Neoreviews. Nov 1 2008;9(11):e520. McGrady GA, Rettig PJ, Istre GR, Jason JM, Holman RC, Evatt BL. An outbreak of necrotizing enterocolitis. Association with transfusions of packed red blood cells. Am J Epidemiol. Dec 1987;126(6):1165-1172. Mally P, Golombek SG, Mishra R, Nigam S, Mohandas K, Depalhma H, et al. Association of necrotizing enterocolitis with elective packed red blood cell transfusions in stable, growing, premature neonates. Am J Perinatol. Nov 2006;23(8):451-458. Christensen RD, Lambert DK, Henry E, Wiedmeier SE, Snow GL, Baer VL, et al. Is "transfusion-associated necrotizing enterocolitis" an authentic pathogenic entity? Transfusion. May 2010;50(5):1106-1112. Josephson CD, Wesolowski A, Bao G, Sola-Visner MC, Dudell G, Castillejo MI, et al. Do red cell transfusions increase the risk of necrotizing enterocolitis in premature infants? J Pediatr. Dec 2010;157(6):972-978 e971-973.

AC

1.

PT

References

10

ACCEPTED MANUSCRIPT

15. 16.

17. 18.

19.

20.

21. 22. 23.

24. 25.

T

RI P

SC

NU

14.

MA

13.

ED

12.

PT

11.

CE

10.

Singh R, Visintainer PF, Frantz ID, 3rd, Shah BL, Meyer KM, Favila SA, et al. Association of necrotizing enterocolitis with anemia and packed red blood cell transfusions in preterm infants. J Perinatol. Mar 2011;31(3):176-182. El-Dib M, Narang S, Lee E, Massaro AN, Aly H. Red blood cell transfusion, feeding and necrotizing enterocolitis in preterm infants. J Perinatol. Mar 2011;31(3):183-187. Blau J, Calo JM, Dozor D, Sutton M, Alpan G, La Gamma EF. Transfusion-related acute gut injury: necrotizing enterocolitis in very low birth weight neonates after packed red blood cell transfusion. J Pediatr. Mar 2011;158(3):403-409. Paul DA, Mackley A, Novitsky A, Zhao Y, Brooks A, Locke RG. Increased odds of necrotizing enterocolitis after transfusion of red blood cells in premature infants. Pediatrics. Apr 2011;127(4):635-641. Holder GL, Dohert D, Patole S. Elective red cell transfusion for anemia of prematurity and development of necrotizing enterocolitis in previously well preterm neonates: incidence and difficulties in proving a cause-effect association. J Neonat Perinat Med. 2009;2(3):181-186. Stritzke AI, Smyth J, Synnes A, Lee SK, Shah PS. Transfusion-associated necrotising enterocolitis in neonates. Arch Dis Child Fetal Neonatal Ed. Jan 2013;98(1):F10-14. Valieva OA, Strandjord TP, Mayock DE, Juul SE. Effects of transfusions in extremely low birth weight infants: a retrospective study. J Pediatr. Sep 2009;155(3):331-337 e331. Wan-Huen P, Bateman D, Shapiro DM, Parravicini E. Packed red blood cell transfusion is an independent risk factor for necrotizing enterocolitis in premature infants. J Perinatol. Oct 2013;33(10):786-790. Mohamed A, Shah PS. Transfusion associated necrotizing enterocolitis: a meta-analysis of observational data. Pediatrics. Mar 2012;129(3):529-540. Tashiro J, Perez EA, Sola JE. Reduced Hospital Mortality With Surgical Ligation of Patent Ductus Arteriosus in Premature, Extremely Low Birth Weight Infants: A Propensity Score-matched Outcome Study. Ann Surg. Mar 27 2015. Stone ML, Lapar DJ, Kane BJ, Rasmussen SK, McGahren ED, Rodgers BM. The effect of race and gender on pediatric surgical outcomes within the United States. J Pediatr Surg. Aug 2013;48(8):1650-1656. Stone ML, LaPar DJ, Mulloy DP, Rasmussen SK, Kane BJ, McGahren ED, et al. Primary payer status is significantly associated with postoperative mortality, morbidity, and hospital resource utilization in pediatric surgical patients within the United States. J Pediatr Surg. Jan 2013;48(1):81-87. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. Jan 1998;36(1):8-27. Consumer Price Index Inflation Calculator. 2014; http://www.bls.gov/data/inflation_calculator.htm. Accessed June 1, 2015. Bell EF, Strauss RG, Widness JA, Mahoney LT, Mock DM, Seward VJ, et al. Randomized trial of liberal versus restrictive guidelines for red blood cell transfusion in preterm infants. Pediatrics. Jun 2005;115(6):1685-1691. Strauss RG. Anaemia of prematurity: pathophysiology and treatment. Blood Rev. Nov 2010;24(6):221-225. Couselo M, Fonseca R, Mangas L, Vila JJ. Surgical Findings in Transfusion-Associated Necrotizing Enterocolitis. Eur J Pediatr Surg. May 26 2015.

AC

9.

11

ACCEPTED MANUSCRIPT

T

RI P

SC

NU

MA

30.

ED

29.

PT

28.

CE

27.

Sellmer A, Tauris LH, Johansen A, Henriksen TB. Necrotizing enterocolitis after red blood cell transfusion in preterm infants with patent ductus arteriosus: a case series. Acta Paediatr. Dec 2012;101(12):e570-572. Baxi AC, Josephson CD, Iannucci GJ, Mahle WT. Necrotizing enterocolitis in infants with congenital heart disease: the role of red blood cell transfusions. Pediatr Cardiol. Aug 2014;35(6):1024-1029. Modi N. Ethical pitfalls in neonatal comparative effectiveness trials. Neonatology. 2014;105(4):350-351. Sola JE, Bronson SN, Cheung MC, Ordonez B, Neville HL, Koniaris LG. Survival disparities in newborns with congenital diaphragmatic hernia: a national perspective. J Pediatr Surg. Jun 2010;45(6):1336-1342. Tashiro J, Wang B, Sola JE, Hogan AR, Neville HL, Perez EA. Patent ductus arteriosus ligation in premature infants in the United States. J Surg Res. Feb 12 2014;190(2):613622.

AC

26.

12

ACCEPTED MANUSCRIPT Figure Legend

T

Figure 1. Prognostic indicators of mortality in premature infants with surgical necrotizing

RI P

enterocolitis (SNEC) by comorbidity and transfusion, Kids’ Inpatient Database, 2003-2009. NEC – necrotizing enterocolitis. Plots represent selected mortality odds ratio (OR) values

SC

derived from binary logistic regression models for mortality; error bars represent 95%

NU

confidence intervals for ORs. OR plots not shown were not significant determinants at α = 0.05. Area under the curve (95% confidence interval) demonstrating overall model performance: 0.982

AC

CE

PT

ED

MA

(0.962, 0.999) for transfusion, 0.940 (0.911, 0.968) for no blood transfusion group.

13

ACCEPTED MANUSCRIPT

Baseline Comparison p <0.001 0.002

287 (58%) 206 (42%)

RI P

T

SNEC with no blood transfusion n=1049 (0.17% of non-BT) 81 (88) 458,626 (582,048)

<0.001 Ref

259 (25%) 348 (33%) 442 (42%)

<0.001 0.013 Ref

628 (60%) 420 (40%)

NS Ref

182 (41%) 98 (22%) 103 (23%) 25 (6%) * (< 2%) 29 (6%)

311 (44%) 175 (25%) 140 (20%) 22 (3%) * (< 2%) 52 (7%)

Ref NS NS 0.028 * NS

240 (49%) 216 (44%) * (< 2%) 26 (5%)

585 (56%) 399 (38%) * (< 2%) 43 (4%)

0.015 Ref * NS

169 (35%) 92 (19%) 111 (23%) 105 (22%)

333 (32%) 276 (27%) 253 (25%) 164 (16%)

Ref 0.006 NS NS

52 (11%) 121 (26%) 295 (63%)

93 (10%) 156 (17%) 680 (73%)

NS <0.001 Ref

* (< 2%) 54 (11%) 415 (89%)

* (< 2%) 95 (10%) 827 (89%)

* NS Ref

180 (39%) 116 (25%) 171 (37%)

283 (31%) 215 (24%) 414 (45%)

0.001 NS Ref

149 (30%) 84 (17%) 63 (13%) 197 (40%)

123 (12%) 387 (37%) 223 (21%) 316 (30%)

<0.001 <0.001 <0.001 Ref

SC

727 (69%) 322 (31%)

CE

PT

ED

MA

287 (58%) 206 (42%)

NU

91 (19%) 147 (30%) 255 (52%)

AC

Length of stay, days Total cost, US $ Survival Alive Dead Year 2003 2006 2009 Sex Male Female Race White African American Hispanic Asian/PI Native American Other Payer Medicaid PI/HMO Self-pay Other Median income First quartile Second quartile Third quartile Fourth quartile Hospital bed size Small Medium Large Location/teaching Rural Urban nonteaching Urban teaching NACHRI type NIACH CGH CUGH Region Northeast Midwest South West

SNEC with blood transfusion n=493 (1.53% of BT) 51 (76) 329,911 (550,834)

14

ACCEPTED MANUSCRIPT Gestational age, week <24 24 25-26 27-28 29-30 31-32 33-34 35-36 Birth weight, grams <500 500-749 750-999 1000-1249 1250-1499 1500-1749 1750-1999 2000-2499 ≥ 2500 Cardiac anomalies BPD IVH Sepsis RDS Multiple gestation

0.001 NS Ref NS NS 0.037 0.027 0.011

19 (2%) 225 (23%) 236 (24%) 135 (14%) 122 (13%) 63 (6%) 57 (6%) 86 (9%) 37 (4%) 10 (1%) 272 (26%) 226 (22%) 620 (59%) 658 (63%) 177 (17%)

NS NS Ref NS NS NS NS 0.006 0.004 <0.001 0.001 <0.001 0.005 NS NS

RI P SC NU

PT

ED

MA

12 (2%) 113 (23%) 103 (21%) 48 (10%) 46 (9%) 40 (8%) 22 (5%) 65 (13%) 34 (7%) 19 (4%) 91 (18%) 160 (32%) 254 (52%) 299 (61%) 68 (14%)

33 (3%) 138 (13%) 257 (25%) 208 (20%) 150 (14%) 102 (10%) 101 (10%) 60 (6%)

T

34 (7%) 48 (10%) 111 (22%) 70 (14%) 52 (10%) 66 (13%) 67 (14%) 46 (9%)

CE

Table 1. Characteristics of surgical necrotizing enterocolitis (SNEC) occurring in blood transfusion (BT) vs. non-blood transfusion (non-BT) cases. Continuous variables are presented

AC

as median (interquartile range), whereas categorical variables are presented as n (%). Baseline comparison analyses were performed; NS – not significant at α = 0.05. Asterisks (*) represent values censored in accordance with HCUP Data Use Agreement. Asian/PI – Asian or Pacific Islander; PI/HMO – private insurance / health maintenance organization; NACHRI – National Association of Children’s Hospitals and Related Institutions, NIACH – not identified as a children’s hospital, CGH – freestanding children’s general hospital, CUGH – children’s unit in a general hospital; BPD – bronchopulmonary dysplasia, IVH – intraventricular hemorrhage, RDS – respiratory distress syndrome.

15

ACCEPTED MANUSCRIPT

p 0.017 0.026 <0.001 <0.001

RI P

SNEC rate Survival LOS, days TC, US $

No blood transfusion (n = 20,988) n % 291 1.39% 194/291 66.7% 88 (70) 557,481 (535,191)

T

Blood transfusion (n = 20,991) n % 351 1.67% 204/351 58.1% 47 (74) 351,381 (578,577)

SC

Table 2. Results of propensity score-matched analyses of surgical necrotizing enterocolitis occurring in blood transfusion vs. non-blood transfusion cases. SNEC – surgical necrotizing

AC

CE

PT

ED

MA

NU

enterocolitis, LOS – length of stay, TC – total charges.

16