- Email: [email protected]
Statistical modelling of survival for babies with oesophageal atresia
Statistical modelling of survival for babies with oesophageal atresia Matthew J. Hartley, Nicholas P.M. Smith, Bruce Jaffray PII: DOI: Reference:
S0022-3468(15)00771-X doi: 10.1016/j.jpedsurg.2015.11.016 YJPSU 57491
To appear in:
Journal of Pediatric Surgery
Received date: Revised date: Accepted date:
29 July 2015 2 November 2015 21 November 2015
Please cite this article as: Hartley Matthew J., Smith Nicholas P.M., Jaffray Bruce, Statistical modelling of survival for babies with oesophageal atresia, Journal of Pediatric Surgery (2015), doi: 10.1016/j.jpedsurg.2015.11.016
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 Statistical modelling of survival for babies with oesophageal atresia
RI P
T
Matthew J Hartley Nicholas PM Smith
SC
Bruce Jaffray*
Department of Paediatric Surgery
MA NU
Address for correspondence:
The Great North Children’s Hospital
ED
Queen Victoria Road Newcastle upon Tyne NE1 4LP
PT
*Corresponding author
AC
Funding: none.
CE
[email protected] Tel: 01912829364
ORIGINAL ARTICLE This work was presented at the international congress of the British Association of Paediatric Surgeons, Cardiff, July 2015 KEYWORDS: oesophageal atresia; survival; associated anomalies; statistical model
ACCEPTED MANUSCRIPT Abstract Aim of study: We examined variables associated with survival for oesophageal
T
atresia between 1996 and 2014.
RI P
Methods: Possible explanatory variables: birth weight, gestation, cardiac anomalies (any or major), renal anomalies (any or severe), primary anastomosis, leak,
SC
secondary oesophageal surgery, tracheomalacia, aortopexy, tracheostomy, gastrostomy, fundoplication, karyotype, neurological status. Variables were
MA NU
assessed with logistic regression and a new model assessed with Kaplan-Meier graphs.
Results: 104/120 (87%) babies survived. Median gestation 37 weeks, 4 (3%) born
ED
before 28 weeks. Mean birth weight 2.3 (SD 0.7) kg, 17 (14%) less than 1500g. Frequency (%) of explanatory variables: Major cardiac anomaly 21 (18%), any
PT
cardiac anomaly 48 (40%), severe renal anomaly 10 (8%), any renal anomaly 25
CE
(21%), primary anastomosis 105 (88%), anastomotic leak 16 (13%), symptomatic tracheomalacia 28 (23%), aortopexy 17 (14%), tracheostomy 12 (10%), neurological
AC
anomaly 7 (6%), fundoplication 15 (13%), gastrostomy 30 (25%), secondary oesophageal surgery 8 (7%), abnormal karyotype 6 (5%). Multivariate analysis showed only renal (OR 0.04, 0.007 0.2) p = 0.001, cardiac (OR 0.1, 0.002 0.6) p = 0.01 and a primary anastomosis (OR 12.2, 1.8 81.6) p = 0.01 (R2 = 0.48) , or major cardiac (OR 0.04, 0.007 0.29) p = 0.001 and severe renal anomalies (OR 0.009, 0.001 0.12) p <0.001 alone were significant (R2 = 0.57). Conclusions: Survival is dependent on cardiac and renal anomalies. Birth weight is not significant. We propose a new classification system: 1: neither severe renal nor major cardiac anomaly, 2: either severe renal or major cardiac anomaly, 3: severe renal and major cardiac anomaly.
ACCEPTED MANUSCRIPT Introduction The survival rate of babies born with an oesophageal atresia with an associated
T
tracheo-oesophageal fistula in a recent national survey is now 97%1. Waterston,
RI P
reporting a 50% mortality in 1962, noted three variables which were associated with death: birth weight, additional congenital anomalies and presence of significant
SC
pneumonia2. The classification of babies into three categories based on these
MA NU
factors allowed a calculation of prognosis and comparison of outcome between centres. Advances in surgery and neonatal care made this classification redundant 4
and it was replaced in 1994 by the Spitz classification based more simply on birth
weight and presence of major cardiac anomalies5.
ED
The aim of this study was to identify those variables associated with mortality in a
PT
contemporary population and to develop an alternative classification of babies born
AC
CE
with oesophageal atresia to better stratify risk of mortality
3,
ACCEPTED MANUSCRIPT Methods We identified a consecutive series of babies treated for oesophageal atresia over 18 We included all cases of oesophageal atresia
T
years in our tertiary centre institution.
RI P
with a tracheal fistula, and all cases of pure atresia without fistula, but excluded H type fistulas without atresia, since we believe these cases should have no mortality.
SC
Case note review was performed and the outcome was classified as survival or
MA NU
death.
Variables which we thought might be associated with survival were as follows: birth weight, gestation, major cardiac anomaly (which we defined as any congenital cardiac anomaly requiring surgery, including patent ductus arteriosus), minor cardiac
ED
anomaly (which we defined as any cardiac anomaly), severe renal anomaly (which
PT
we defined as either bilateral structural renal anomalies, or unilateral structural with elevated serum creatinine within one week of birth), any renal anomaly, neurological
CE
anomaly, chromosomal anomaly, the occurrence of anastomotic leak, whether a primary anastomosis was performed (which we defined as the performance of an
AC
oesophageal anastomosis at the first procedure, secondary oesophageal surgery (which we define as any procedure other than an initial oesophago-oesophageal anastomosis usually either oesophageal replacement or resection of strictured anastomosis), the presence of symptomatic tracheomalacia, the need for aortopexy, the use of a tracheostomy, the need for fundoplication and the use of a gastrostomy. We examined the effects of birth weight both as a continuous variable and as a dichotomous variable above or below 1500g. We similarly examined gestation as a continuous and dichotomous variable, above and below 28 weeks. For each patient we calculated both the Spitz criteria and the modified Spitz criteria 6.
ACCEPTED MANUSCRIPT Data were recorded on an Access database, which we programed to calculate Spitz criteria as follows: Original Spitz criteria: 1 Birth weight >1500g and either no or
T
minor cardiac anomaly, 2 Birth weight < 1500g or major cardiac anomaly, 3 birth
RI P
weight< 1500g and cardiac anomaly major. Modified Spitz : 1 Birth weight >1500 and cardiac anomaly either absent or minor, 2.1 Birth weight < 1500g and cardiac
MA NU
weight < 1500g and cardiac anomaly major.
SC
anomaly absent or minor, 2.2 birth weight >1500g and cardiac anomaly major, 3 birth
Statistical Analysis
We constructed a statistical model using logistic regression analysis of the listed variables. Covariates with a Wald’s p value ≤ 0.05 on univariate analysis were
ED
entered into multivariate analysis using a forced entry blockwise design. Variables
PT
which have previously been shown to be of prognostic significance (Major cardiac anomalies and birth weight) were entered as the first block. Because of the
CE
collinearity between cardiac anomaly and major cardiac anomaly, and between renal anomaly and severe renal anomaly, the multivariate analysis was performed twice,
AC
using renal and cardiac anomalies, then severe renal and major cardiac. The two models so produced were compared using Nagelkerke’s R2 using this to decide on the best fit. Variables which were significant on multivariate analysis were then used to construct a final model. The model was assessed for outliers and observations with unusual influence by examination of standardised residuals and calculation of Cooks distance. Collinearity was assessed by calculation of tolerance and variance inflation factor.
ACCEPTED MANUSCRIPT We then constructed further models using the Spitz and modified Spitz criteria, comparing them to the new model again using Nagelkerke’s R2.
T
Finally we constructed Kaplan Meier survival graphs using the Spitz, modified Spitz
RI P
and the new model, comparing survival with the log rank test.
SC
Significance was set at 5%. Statistical analysis used IBM SSPS statistical software, version 21.Categorical data are presented as frequencies (%), continuous data are
AC
CE
PT
ED
MA NU
presented as either median (range) or mean (SD).
ACCEPTED MANUSCRIPT Results 120 babies with OA born between 1996 and 2014 were studied. Median gestation
T
was 37 weeks (range 25-42), with 4 (3%) born before 28 weeks. Mean birth weight
RI P
was 2.3 (0.7) kg, and 17 (14%) were less than 1500g. 16 babies died (Table 1) and
SC
survival was 87%.
The frequency (%) of explanatory variables is as follows: Major cardiac anomaly 21
MA NU
(17%), any cardiac anomaly 48 (40%), severe renal anomaly 10 (8%), any renal anomaly 25 (21%), primary anastomosis 105 (88%), anastomotic leak 16 (13%), symptomatic tracheomalacia 28 (23%), aortopexy 17 (14%), tracheostomy 12 (10%), neurological anomaly 7 (6%), fundoplication 15 (13%), gastrostomy 30 (25%),
ED
secondary oesophageal surgery 8 (7%), abnormal karyotype 6 (5%).
PT
The results of univariate analysis are shown in table 2. On univariate analysis the following variables were significantly related to survival: any cardiac anomaly, major
CE
cardiac anomaly, any renal anomaly, severe renal anomaly, performance of a
AC
primary anastomosis, presence of a gastrotomy. Birth weight as a categorical variable of greater or less than 1500g was not significantly associated with survival, but birth weight as a continuous variable was. Multivariate analysis showed that only renal and cardiac and a primary anastomosis (Table 3), or major cardiac and severe renal anomalies alone (Table 4), were significantly associated with probability of survival. Birth weight ceased to be significant. Nagelkerke’s R2 for the model using any cardiac, any renal anomaly and a primary anastomosis was 0.48, while the R2 for the model using severe renal and major cardiac was 0.57, suggesting a better model. The final model using only major cardiac anomaly and severe renal anomaly as covariates is described in table 5.
ACCEPTED MANUSCRIPT The probability of survival for a baby born with oesophageal atresia = 4.01 – 3.26 (presence of major cardiac anomaly) – 4.6 (presence of severe renal anomaly).
T
There was no evidence of collinearity of the covariates.
RI P
Using these variables, we constructed a new scoring system thus: grade 1, no major cardiac or severe renal anomaly, grade 2, presence of either major cardiac or severe
SC
renal, grade 3 presence of major cardiac and severe renal anomaly. We term this the
MA NU
Newcastle classification.
Univariate analysis comparing the Spitz criteria, the modified Spitz criteria and the Newcastle criteria are presented in table 6. Kaplan Meir survival graphs for the Spitz classification and the Newcastle classification are figures 1 and 2. The log rank
ED
statistic for the Spitz criteria are C2 = 20.7 p <0.001, and the Newcastle criteria are C2
AC
CE
PT
= 65.3 p <0.001.
ACCEPTED MANUSCRIPT Discussion The survival for babies born with oesophageal atresia in the current series was 87%.
T
Attempts to improve this statistic should be directed to those variables which
RI P
influence survival. We have shown that the subdivision of birth weight by a value of 1500g, one of the variables used to categorise babies using the Spitz classification,
SC
is no longer a significant factor in determining survival. In contrast, renal anomalies,
MA NU
particularly severe renal anomalies have a profound effect on the likelihood of a baby surviving.
The loss of influence of birth weight on survival is unsurprising. Advances in neonatal care have brought survivability down to 24 weeks gestation, and babies born
ED
prematurely with birth weights below 1kg are now routinely operated on for such
PT
conditions as necrotising enterocolitis with expectation of survival7. A number of other studies of mortality in oesophageal atresia have shown no influence of birth
CE
weight 4, 8-10.
AC
Unlike low birth weight, renal compromise in the new born remains a complex therapeutic dilemma, and even with aggressive and novel dialysis techniques, mortality is appreciable11, 12. The particularly poor prognosis of end stage renal failure in early life when additional anomalies are present has been documented, although oesophageal atresia did not feature as a notable co-morbidity in a national survey13. The relative risk for congenital heart disease among babies with oesophageal atresia compared to the general population is 2314. All published evidence is in agreement that severe congenital heart disease is invariably associated with increased mortality in this population.
ACCEPTED MANUSCRIPT Our findings may not be generalizable. In recent studies of survival using health insurance coding data from multiple centres, birth weight was still found to be
T
strongly associated with outcome15-17. This may reflect variability in the outcome of
RI P
small babies within differing units with differing levels of expertise. Alternatively, the health insurance coding did not detect variables such as the achievement of a
SC
primary anastomosis or occurrence of a leak, and this may affect the regression
MA NU
model derived by these authors.
For a prognostic scoring system to have utility it must clearly distinguish patients at different risks and the degrees of difference must be of a magnitude to be clinically important. Empirically, we suggest that a difference in survival rate would be at least
ED
20% to be clinically meaningful in discussions with parents. It must also have relatively few categories, which select meaningful numbers of patients. Ideally it
PT
should be simple with unambiguous criteria and easily memorised. Lastly, it must be
CE
generalizable; a scoring system which produces a perfect model, but only in one hospital is of little use.
AC
A review of published outcomes where scores have been presented in the last 2 decades shows existing scoring systems to be inadequate (Table 7). Either failing to correctly predict progressively poorer outcome using Waterston
10, 18-22
, Spitz 23-25 or
Bremen 21, 22, or producing categories with less than 20% difference in survival rates between them 9, 15, 18-21, 24-29. We find the Montreal classification difficult to apply18. Our proposed classification consists of three groups: 1, neither severe renal nor major cardiac anomaly. 2, either severe renal or major cardiac anomaly. 3, both severe renal and major cardiac anomaly. In our population this produced mortality rates of 2%, 43% and 100%. We suggest this classification has the merits of
ACCEPTED MANUSCRIPT simplicity, unambiguous criteria and near perfect splitting of survival probabilities. Whether our classification is generalizable will require it to be tested in other
T
populations.
RI P
In conclusion, we have found that survival of babies born with oesophageal atresia can be best described using a model with major cardiac anomalies and severe renal
SC
anomalies as covariates. Improvements in survival will be dependent on advances in
MA NU
renal replacement in the new born and better survival of congenital cardiac
AC
CE
PT
ED
anomalies.
ACCEPTED MANUSCRIPT References 1. Burge DM, Shah K, Spark P, Shenker N, Pierce M, Kurinczuk JJ, et al.
T
Contemporary management and outcomes for infants born with oesophageal atresia.
RI P
Br J Surg. 2013;100:515-21.
2. Waterston DJ, Carter RE, Aberdeen E. Oesophageal atresia: tracheo-
SC
oesophageal fistula. A study of survival in 218 infants. Lancet. 1962;1:819-22. 3. Beasley SW, Myers NA. Trends in Mortality in Esophageal Atresia. Pediatr Surg
MA NU
Int. 1992;7:86-9.
4. Engum SA, Grosfeld JL, West KW, Rescorla FJ, Scherer LRT. Analysis of morbidity and mortality in 227 cases of esophageal atresia and/or
ED
tracheoesophageal fistula over 2 decades. Arch Surg. 1995;130:502-8. 5. Spitz L, Kiely EM, Morecroft JA, Drake DP. Oesophageal atresia: at-risk groups
PT
for the 1990s. J Pediatr Surg. 1994;29:723-5. 6. Malakounides G, Lyon PC, De Agustin JC, Cross K, Drake D, Pierro A, et al.
CE
Oesophageal atresia: Improved outcomes in high risk groups? -revisited. British
AC
Association of Paediatric Surgeons International Congress. Bournmouth2013. 7. Lemons JA, Bauer CR, Oh W, Korones SB, Papile LA, Stoll BJ, et al. Very low birth weight outcomes of the National Institute of Child Health and Human Development Neonatal Research Network, January 1995 through December 1996. Pediatrics. 2001;107:art. no.-e1. 8. Calisti A, Oriolo L, Nanni L, Molle P, Briganti V, D'Urzo C. Mortality and long term morbidity in esophageal atresia: the reduced impact of low birth weight and maturity on surgical outcome. J Perinat Med. 2004;32:171-5.
ACCEPTED MANUSCRIPT 9. Choudhury SR, Ashcraft KW, Sharp RJ, Murphy JP, Snyder CL, Sigalet DL. Survival of patients with esophageal atresia: Influence of birth weight, cardiac
T
anomaly, and late respiratory complications. J Pediatr Surg. 1999;34:70-3.
RI P
10. Deurloo JA, Ekkelkamp S, Schoorl M, Heij HA, Aronson DC. Esophageal atresia: Historical evolution of management and results in 371 patients. Ann Thorac Surg.
SC
2002;73:267-72.
11. Everdell NL, Coulthard MG, Crosier J, Keir MJ. A machine for haemodialysing
MA NU
very small infants. Pediatr Nephrol. 2005;20:636-43.
12. Coulthard MG, Crosier J, Griffiths C, Smith J, Drinnan M, Whitaker M, et al. Haemodialysing babies weighing < 8 kg with the Newcastle infant dialysis and
ED
ultrafiltration system (Nidus): comparison with peritoneal and conventional haemodialysis. Pediatr Nephrol. 2014;29:1873-81.
PT
13. Coulthard MG, Crosier J, British Assoc Paediat N. Outcome of reaching end stage renal failure in children under 2 years of age. Arch Dis Child. 2002;87:511-7.
CE
14. Leonard H, Barrett AM, Scott JES, Wren C. The influence of congenital heart
AC
disease on survival of infants with oesophageal atresia. Arch Dis Child. 2001;85:F204-F6.
15. Turner B, Dasgupta R, Brindle ME. A contemporary prediction rule for esophageal atresia (EA) and tracheo-esophageal fistula (TEF). J Pediatr Surg. 2014;49:1758-61. 16. Wang B, Tashiro J, Allan BJ, Sola JE, Parikh PP, Hogan AR, et al. A nationwide analysis of clinical outcomes among newborns with esophageal atresia and tracheoesophageal fistulas in the United States. J Surg Res. 2014;190:604-12.
ACCEPTED MANUSCRIPT 17. Sulkowski JP, Cooper JN, Lopez JJ, Jailcherla Y, Cuenot A, Mattei P, et al. Morbidity and mortality in patients with esophageal atresia. Surgery. 2014;156:483-
T
91.
RI P
18. Poenaru D, Laberge JM, Neilson IR, Guttman FM. A new prognostic classification for esophageal atresia. Surgery. 1993;113:426-32.
SC
19. Okada A, Usui N, Inoue M, Kawahara H, Kubota A, Imura K, et al. Esophageal atresia in Osaka: A review of 39 years' experience. J Pediatr Surg. 1997;32:1570-4.
MA NU
20. Dunn JCY, Fonkalsrud EW, Atkinson JB. Simplifying the Waterston's stratification of infants with tracheoesophageal fistula. Am Surg. 1999;65:908-10. 21. Yagyu M, Gitter H, Richter B, Booss D. Esophageal atresia in Bremen, Germany
ED
- Evaluation of preoperative risk classification in esophageal atresia. J Pediatr Surg. 2000;35:584-7.
PT
22. Konkin DE, O'Hali WA, Webber EM, Blair GK. Outcomes in esophageal atresia and tracheoesophageal fistula. J Pediatr Surg. 2003;38:1726-9.
CE
23. Driver CP, Shankar KR, Jones MO, Lamont GA, Turnock RR, Lloyd DA, et al.
AC
Phenotypic presentation and outcome of esophageal atresia in the era of the Spitz classification. J Pediatr Surg. 2001;36:1419-21. 24. Lilja HE, Wester T. Outcome in neonates with esophageal atresia treated over the last 20 years. Pediatr Surg Int. 2008;24:531-6. 25. Koivusalo AI, Pakarinen MP, Rintala RJ. Modern outcomes of oesophageal atresia: Single centre experience over the last twenty years. J Pediatr Surg. 2013;48:297-303. 26. Teich S, Barton DP, GinnPease ME, King DR. Prognostic classification for esophageal atresia and tracheoesophageal fistula: Waterston versus Montreal. J Pediatr Surg. 1997;32:1075-9.
ACCEPTED MANUSCRIPT 27. Lopez PJ, Keys C, Pierro A, Drake DP, Kiely EM, Curry JI, et al. Oesophageal atresia: improved outcome in high-risk groups? J Pediatr Surg. 2006;41:331-4.
T
28. Okamoto T, Takamizawa S, Arai H, Bitoh Y, Nakao M, Yokoi A, et al.
RI P
Esophageal atresia: Prognostic classification revisited. Surgery. 2009;145:675-81. 29. Niramis R, Tangkhabuanbut P, Anuntkosol M, Buranakitjaroen V, Tongsin A,
SC
Mahatharadol V. Clinical Outcomes of Esophageal Atresia: Comparison Between the Waterston and the Spitz Classifications. Annals Academy of Medicine Singapore.
MA NU
2013;42:297-300.
30. Turner B, Dasgupta R, Brindle ME. A contemporary prediction rule for esophageal atresia (EA) and tracheo-esophageal fistula (TEF). J Pediatr Surg.
AC
CE
PT
ED
2014;49:1758-61.
AC
CE
PT
ED
MA NU
SC
RI P
T
ACCEPTED MANUSCRIPT
Figure 1
AC
CE
PT
ED
MA NU
SC
RI P
T
ACCEPTED MANUSCRIPT
Figure 2
ACCEPTED MANUSCRIPT Table 1 of anomalies in babies who died.
Renal
Cardiac anomaly
1
Cause of death
anomaly
RI P
number
Other anomalies
T
Child
Tracheomalacia
Severe
Pulmonary
with intrarenal
neurological
hypertension
reflux
impairment.
VSD
Down’s syndrome,
Nil
ASD. Small right ventricle. Dextrocardia. Abnormal drainage of
PT
PDA ligated
5
AC
4
PDA. VSD
AVSD. PDA.
Multiple vertebral
and rib anomalies
Bilateral renal
Cloaca.
dysplasia
Hypothyroid. Long
Not known
gap atresia with
CE
3
Imperforate anus.
ED
SVC to left atrium
Single kidney
MA NU
2
SC
duodenal atresia
oesophagostomy Nil
ATN
Necrotising
Klebsiella
enterocolitis
septicaemia
Gastric perforation
Sepsis
Hypoplastic aortic arch 6
Nil
Right
Pulmonary
hydronephrotic
hypoplasia
kidney drained in utero at 25
ACCEPTED MANUSCRIPT weeks. Left multi-cystic dysplastic
Vesico-
double outlet right
ureteric reflux
8
Tetralogy of Fallot
Nil
PT
Nil
Renal failure
CE
10
cardiac surgery Long gap atresia
Pulmonary
with
hypoplasia
oesophagostomy Long gap atresia.
Renal failure
Fistula ligated only. Severe
AC 11
Tracheomalacia
Died during
Nil
ED
9
Hypoplastic left heart
MA NU
ventricle, pulmonary stenosis
RI P
Tetralogy of Fallot,
SC
7
T
kidney
hydrocephalus. Ascites
Severe pulmonary
Bilateral renal
Pulmonary
trunk hypoplasia. VSD.
dysplasia
hypoplasia. Imperforate anus. Limb anomalies. Vertebral anomalies
Care withdrawn
ACCEPTED MANUSCRIPT 12
Single
Imperforate anus.
dysplastic
Cleft palate.
kidney
Structurally
Nil
Care withdrawn
VSD
Nil
Diffuse lymphatic
Uncontrollable
leak
lymphatic
14
Tetralogy of Fallot
MA NU
SC
13
RI P
T
abnormal brain.
Nil
losses
Limb and vertebral
Post mortem
anomalies. Failed
inconclusive.
initial anastomosis Delayed primary anastomosis
PDA. Coarctation.
Absent right,
Trisomy 20
Right ventricular
hydronephrotic
PT
dysplasia
CE
16
Bilateral renal
Nil
ED
15
hypertrophy. Aortic
Renal failure
Care withdrawn
left kidney
AC
stenosis.
PDA: patent ductus arteriosus, VSD: ventricular septal defect, SVC: superior vena cava AVSD: atrioventricular septal defect,
ACCEPTED MANUSCRIPT Table 2. Univariate analysis of possible influential variables on survival of babies with oesophageal atresia
No cardiac anomaly
68/72 (94.5%)
Major cardiac anomaly
11/21 (52.4%)
No major cardiac anomaly
93/99 (93.9%)
Birth weight >1500g
91/103 (87%)
Birth weight< 1500g
13/17 (76.5%)
Birth weight as a continuous variable
0.17 (0.053 0.587)
0.071 (0.022 0.233)
0.005
<0.001
0.429 (0.12 1.52)
0.19
0.33 (0.14 0.77)
0.010
0.1 (0.032 0.32)
<0.001
0.02 ( 0.004 0.1)
<0.001
9.3 (2.74 31.7)
<0.001
1 (0.1 9.4)
0.9
0.9 (0.26 3.0)
0.86
0.67 (0.17 2.66)
0.57
0.41 (0.09 1.71)
0.22
2.33 (0.28 19.07)
0.42
15/25 (60%)
ED
Any renal anomaly
p value*
T
36/48 (75%)
RI P
Any cardiac anomaly
Odds ratio (95% CI)
SC
Survival
MA NU
Variable
No renal anomaly
89/95 (94%) 2/10 (20%)
PT
Severe renal anomaly
Primary anastomosis
AC
No primary anastomosis
CE
No severe renal anomaly
102/110 (93%) 96/105 (91%) 8/15 (53%)
Secondary oesophageal surgery
15/16 (94%)
No secondary oesophageal surgery
97/104 (94%)
Symptomatic tracheomalacia
24/28 (86%)
No symptomatic tracheomalacia
80/92 (87%)
Aortopexy
14/17 (82%)
No aortopexy
90/103 (87%)
Tracheostomy
9/12 (75%)
No tracheostomy
95/108 (88%)
Fundoplication
14/15 (93%)
No fundoplication
90/105 (86%)
ACCEPTED MANUSCRIPT
Abnormal karyotype
4/6 (67%)
Normal karyotype
100/114 (88%)
Abnormal neurology
5/7 (71%)
Normal neurology
99/113 (88%)
Anastomotic leak
15/16 (94%)
No anastomotic leak
89/104 (86%)
Gestation <=28 weeks
2/4 (50%)
Gestation > 28 weeks
102/116 (88%)
Gestation as a continuous variable
CE
PT
ED
* Wald p value
AC
0.26 (0.09 0.79)
T
82/90 (91%)
0.28 (0.05 1.65)
RI P
No gastrostomy
0.35 (0.06 2)
SC
22/30 (73%)
2.5 (0.3 20.5)
MA NU
Gastrostomy
0.018
0.16
0.24
0.3
0.13 (0.01 1.05)
0.05
0.8 (0.7 0.9)
0.005
ACCEPTED MANUSCRIPT Table 3. Multivariate blockwise entry analysis of variables related to survival, using
Odds ratio (95% ci)
Any cardiac anomaly
0.1 (0.017 0.6)
0.01
Any renal anomaly
0.04 (0.007 0.2)
0.001
Birth weight as a continuous variable
0.46 (0.6 3.4)
0.4
0.54 (0.09 3.0)
0.4
0.9 (0.6 1.3)
0.7
12.2 (1.8 81.6)
0.01
Gastrostomy
ED
Gestation as a continuous variable
CE
PT
Primary anastomosis
AC
MA NU
RI P
T
Variable
SC
any cardiac anomaly and any renal anomaly as variables.
p value
ACCEPTED MANUSCRIPT Table 4. Multivariate blockwise entry analysis of variables related to survival, using major cardiac anomaly and severe renal anomaly as variables.
Odds ratio (95% ci)
Major cardiac anomaly
0.044 (0.007 0.29)
Severe renal anomaly
0.009 (0.001 0.12)
<0.001
Birth weight as continuous variable
1.7 (0.2 14.2)
0.6
0.39 (0.05 2.9)
0.36
0.7 (0.47 1.1)
0.1
4.8 (0.5 44)
0.16
RI P
SC
ED
Gestation as a continuous variable
MA NU
Gastrostomy
CE
PT
Primary anastomosis
AC
T
Variable
p value
0.001
ACCEPTED MANUSCRIPT Table 5. Final statistical model of survival for oesophageal atresia
Odds ratio (95% ci)
Constant
4.01 (1.18)
T
B (SE)
RI P
Variable
Major cardiac
0.038 (0.007 0.2)
SC
-3.26 (0.86) * anomaly
MA NU
Severe renal
-4.6 (1.09) * anomaly
AC
CE
PT
ED
* Wald p < 0.001. Nagelkerke’s R2 0.57
0.01 (0.001 0.08)
ACCEPTED MANUSCRIPT Table 6. Univariate analysis of different prognostic scoring systems assessed as ordinal scales. Survival
Odds ratio (95% ci)
p value
Spitz
RI P
T
Scoring system
<0.001
4.1 (1.9 9)
<0.001
39 (8.1 186)
<0.001
4.6 (2 10.7) 81/86 (94%)
2
20/28 (71%)
MA NU
SC
1
3
3/6 (50%)
Modified Spitz
81/86 (94%)
ED
1
11/12 (92%)
PT
2.1
3
AC
Newcastle model
CE
2.2
8/15 (53%) 4/7 (57%)
1 Neither cardiac nor renal
91/93 (98%)
2 Either cardiac or renal
10/23 (43%)
3 Cardiac and renal
0/4 (0%)
ACCEPTED MANUSCRIPT Table 7. Published survival and risk stratification of oesophageal atresia
95
Waterston *+
C 73%
A 100%
B 100%
C 50%
I 92%
II 50%
III 0%
A 100%
B 91%
C 43%
I 91%
II 53%
A 100%
B 100%
C 43%
I 89%
II 64%
III 50%
A 100%
B 100%
C 66%
I 96%
II 57%
III 0%
No complications
I 100%
II 50%
III 0%
Complications
I 33%
II 40%
III 0%
134
Spitz * +
I 92%
II 69%
III 100%
357
Waterston * +
A 86%
B 88%
C 51%
144
Waterston * +
A 100%
B 100%
C 80%
Spitz +
I 99%
II 84%
III 43%
Montreal
I 92%
II 71%
No complications
I 99%
II 84%
III 43%
Complications
I 100%
II83%
III 25%
Spitz +
I 99%
II 82%
III 50%
1993 [18]
Montreal
Okada
112
Waterston * +
1997 [19]
Spitz 94
Waterston +
1997 [26]
Montreal
Dunn
64
Waterston * +
Choudhury 1999 [9]
240
Spitz +
Yagu
133
Waterston * +
2000 [21]
ED
1999 [20]
A 93%
B 94%
I 93%
II 31%
MA NU
Teich
Classification grade and survival
T
Poenaru
Classifications reported
RI P
Year of publication
Number of patients
SC
Author
PT
Spitz
2001 [23] Deurloo
AC
Driver
CE
Bremen * +
2002 [10] Konkin 2003 [22]
Bremen * +
Lopez
188
ACCEPTED MANUSCRIPT 2006 [27] Lilja
65
Spitz * +
I 100%
II 78%
III 100%
52
Spitz +
I 84%
II 76%
III 66%
Revised Spitz +
I 100%
II 88%
III 83%
130
Spitz * +
I 99%
II 94%
III No patients
132
Waterston +
A 100%
B 92%
C 49%
I 97%
II 64%
III 27%
II 87%
III 54%
2009[28] Koivusalo
2013 [29] Turner
Spitz 1219
Spitz +
2014 [30]
MA NU
Niramis
SC
2012 [25]
RI P
Okamoto
I 99%
T
2008 [24]
IV 40%
ED
* Classification where outcomes are inconsistent with progressively poorer grading.
AC
CE
PT
+ Classification where groups show less than 20% difference in survival rates
S0022-3468(15)00771-X doi: 10.1016/j.jpedsurg.2015.11.016 YJPSU 57491
To appear in:
Journal of Pediatric Surgery
Received date: Revised date: Accepted date:
29 July 2015 2 November 2015 21 November 2015
Please cite this article as: Hartley Matthew J., Smith Nicholas P.M., Jaffray Bruce, Statistical modelling of survival for babies with oesophageal atresia, Journal of Pediatric Surgery (2015), doi: 10.1016/j.jpedsurg.2015.11.016
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 Statistical modelling of survival for babies with oesophageal atresia
RI P
T
Matthew J Hartley Nicholas PM Smith
SC
Bruce Jaffray*
Department of Paediatric Surgery
MA NU
Address for correspondence:
The Great North Children’s Hospital
ED
Queen Victoria Road Newcastle upon Tyne NE1 4LP
PT
*Corresponding author
AC
Funding: none.
CE
[email protected] Tel: 01912829364
ORIGINAL ARTICLE This work was presented at the international congress of the British Association of Paediatric Surgeons, Cardiff, July 2015 KEYWORDS: oesophageal atresia; survival; associated anomalies; statistical model
ACCEPTED MANUSCRIPT Abstract Aim of study: We examined variables associated with survival for oesophageal
T
atresia between 1996 and 2014.
RI P
Methods: Possible explanatory variables: birth weight, gestation, cardiac anomalies (any or major), renal anomalies (any or severe), primary anastomosis, leak,
SC
secondary oesophageal surgery, tracheomalacia, aortopexy, tracheostomy, gastrostomy, fundoplication, karyotype, neurological status. Variables were
MA NU
assessed with logistic regression and a new model assessed with Kaplan-Meier graphs.
Results: 104/120 (87%) babies survived. Median gestation 37 weeks, 4 (3%) born
ED
before 28 weeks. Mean birth weight 2.3 (SD 0.7) kg, 17 (14%) less than 1500g. Frequency (%) of explanatory variables: Major cardiac anomaly 21 (18%), any
PT
cardiac anomaly 48 (40%), severe renal anomaly 10 (8%), any renal anomaly 25
CE
(21%), primary anastomosis 105 (88%), anastomotic leak 16 (13%), symptomatic tracheomalacia 28 (23%), aortopexy 17 (14%), tracheostomy 12 (10%), neurological
AC
anomaly 7 (6%), fundoplication 15 (13%), gastrostomy 30 (25%), secondary oesophageal surgery 8 (7%), abnormal karyotype 6 (5%). Multivariate analysis showed only renal (OR 0.04, 0.007 0.2) p = 0.001, cardiac (OR 0.1, 0.002 0.6) p = 0.01 and a primary anastomosis (OR 12.2, 1.8 81.6) p = 0.01 (R2 = 0.48) , or major cardiac (OR 0.04, 0.007 0.29) p = 0.001 and severe renal anomalies (OR 0.009, 0.001 0.12) p <0.001 alone were significant (R2 = 0.57). Conclusions: Survival is dependent on cardiac and renal anomalies. Birth weight is not significant. We propose a new classification system: 1: neither severe renal nor major cardiac anomaly, 2: either severe renal or major cardiac anomaly, 3: severe renal and major cardiac anomaly.
ACCEPTED MANUSCRIPT Introduction The survival rate of babies born with an oesophageal atresia with an associated
T
tracheo-oesophageal fistula in a recent national survey is now 97%1. Waterston,
RI P
reporting a 50% mortality in 1962, noted three variables which were associated with death: birth weight, additional congenital anomalies and presence of significant
SC
pneumonia2. The classification of babies into three categories based on these
MA NU
factors allowed a calculation of prognosis and comparison of outcome between centres. Advances in surgery and neonatal care made this classification redundant 4
and it was replaced in 1994 by the Spitz classification based more simply on birth
weight and presence of major cardiac anomalies5.
ED
The aim of this study was to identify those variables associated with mortality in a
PT
contemporary population and to develop an alternative classification of babies born
AC
CE
with oesophageal atresia to better stratify risk of mortality
3,
ACCEPTED MANUSCRIPT Methods We identified a consecutive series of babies treated for oesophageal atresia over 18 We included all cases of oesophageal atresia
T
years in our tertiary centre institution.
RI P
with a tracheal fistula, and all cases of pure atresia without fistula, but excluded H type fistulas without atresia, since we believe these cases should have no mortality.
SC
Case note review was performed and the outcome was classified as survival or
MA NU
death.
Variables which we thought might be associated with survival were as follows: birth weight, gestation, major cardiac anomaly (which we defined as any congenital cardiac anomaly requiring surgery, including patent ductus arteriosus), minor cardiac
ED
anomaly (which we defined as any cardiac anomaly), severe renal anomaly (which
PT
we defined as either bilateral structural renal anomalies, or unilateral structural with elevated serum creatinine within one week of birth), any renal anomaly, neurological
CE
anomaly, chromosomal anomaly, the occurrence of anastomotic leak, whether a primary anastomosis was performed (which we defined as the performance of an
AC
oesophageal anastomosis at the first procedure, secondary oesophageal surgery (which we define as any procedure other than an initial oesophago-oesophageal anastomosis usually either oesophageal replacement or resection of strictured anastomosis), the presence of symptomatic tracheomalacia, the need for aortopexy, the use of a tracheostomy, the need for fundoplication and the use of a gastrostomy. We examined the effects of birth weight both as a continuous variable and as a dichotomous variable above or below 1500g. We similarly examined gestation as a continuous and dichotomous variable, above and below 28 weeks. For each patient we calculated both the Spitz criteria and the modified Spitz criteria 6.
ACCEPTED MANUSCRIPT Data were recorded on an Access database, which we programed to calculate Spitz criteria as follows: Original Spitz criteria: 1 Birth weight >1500g and either no or
T
minor cardiac anomaly, 2 Birth weight < 1500g or major cardiac anomaly, 3 birth
RI P
weight< 1500g and cardiac anomaly major. Modified Spitz : 1 Birth weight >1500 and cardiac anomaly either absent or minor, 2.1 Birth weight < 1500g and cardiac
MA NU
weight < 1500g and cardiac anomaly major.
SC
anomaly absent or minor, 2.2 birth weight >1500g and cardiac anomaly major, 3 birth
Statistical Analysis
We constructed a statistical model using logistic regression analysis of the listed variables. Covariates with a Wald’s p value ≤ 0.05 on univariate analysis were
ED
entered into multivariate analysis using a forced entry blockwise design. Variables
PT
which have previously been shown to be of prognostic significance (Major cardiac anomalies and birth weight) were entered as the first block. Because of the
CE
collinearity between cardiac anomaly and major cardiac anomaly, and between renal anomaly and severe renal anomaly, the multivariate analysis was performed twice,
AC
using renal and cardiac anomalies, then severe renal and major cardiac. The two models so produced were compared using Nagelkerke’s R2 using this to decide on the best fit. Variables which were significant on multivariate analysis were then used to construct a final model. The model was assessed for outliers and observations with unusual influence by examination of standardised residuals and calculation of Cooks distance. Collinearity was assessed by calculation of tolerance and variance inflation factor.
ACCEPTED MANUSCRIPT We then constructed further models using the Spitz and modified Spitz criteria, comparing them to the new model again using Nagelkerke’s R2.
T
Finally we constructed Kaplan Meier survival graphs using the Spitz, modified Spitz
RI P
and the new model, comparing survival with the log rank test.
SC
Significance was set at 5%. Statistical analysis used IBM SSPS statistical software, version 21.Categorical data are presented as frequencies (%), continuous data are
AC
CE
PT
ED
MA NU
presented as either median (range) or mean (SD).
ACCEPTED MANUSCRIPT Results 120 babies with OA born between 1996 and 2014 were studied. Median gestation
T
was 37 weeks (range 25-42), with 4 (3%) born before 28 weeks. Mean birth weight
RI P
was 2.3 (0.7) kg, and 17 (14%) were less than 1500g. 16 babies died (Table 1) and
SC
survival was 87%.
The frequency (%) of explanatory variables is as follows: Major cardiac anomaly 21
MA NU
(17%), any cardiac anomaly 48 (40%), severe renal anomaly 10 (8%), any renal anomaly 25 (21%), primary anastomosis 105 (88%), anastomotic leak 16 (13%), symptomatic tracheomalacia 28 (23%), aortopexy 17 (14%), tracheostomy 12 (10%), neurological anomaly 7 (6%), fundoplication 15 (13%), gastrostomy 30 (25%),
ED
secondary oesophageal surgery 8 (7%), abnormal karyotype 6 (5%).
PT
The results of univariate analysis are shown in table 2. On univariate analysis the following variables were significantly related to survival: any cardiac anomaly, major
CE
cardiac anomaly, any renal anomaly, severe renal anomaly, performance of a
AC
primary anastomosis, presence of a gastrotomy. Birth weight as a categorical variable of greater or less than 1500g was not significantly associated with survival, but birth weight as a continuous variable was. Multivariate analysis showed that only renal and cardiac and a primary anastomosis (Table 3), or major cardiac and severe renal anomalies alone (Table 4), were significantly associated with probability of survival. Birth weight ceased to be significant. Nagelkerke’s R2 for the model using any cardiac, any renal anomaly and a primary anastomosis was 0.48, while the R2 for the model using severe renal and major cardiac was 0.57, suggesting a better model. The final model using only major cardiac anomaly and severe renal anomaly as covariates is described in table 5.
ACCEPTED MANUSCRIPT The probability of survival for a baby born with oesophageal atresia = 4.01 – 3.26 (presence of major cardiac anomaly) – 4.6 (presence of severe renal anomaly).
T
There was no evidence of collinearity of the covariates.
RI P
Using these variables, we constructed a new scoring system thus: grade 1, no major cardiac or severe renal anomaly, grade 2, presence of either major cardiac or severe
SC
renal, grade 3 presence of major cardiac and severe renal anomaly. We term this the
MA NU
Newcastle classification.
Univariate analysis comparing the Spitz criteria, the modified Spitz criteria and the Newcastle criteria are presented in table 6. Kaplan Meir survival graphs for the Spitz classification and the Newcastle classification are figures 1 and 2. The log rank
ED
statistic for the Spitz criteria are C2 = 20.7 p <0.001, and the Newcastle criteria are C2
AC
CE
PT
= 65.3 p <0.001.
ACCEPTED MANUSCRIPT Discussion The survival for babies born with oesophageal atresia in the current series was 87%.
T
Attempts to improve this statistic should be directed to those variables which
RI P
influence survival. We have shown that the subdivision of birth weight by a value of 1500g, one of the variables used to categorise babies using the Spitz classification,
SC
is no longer a significant factor in determining survival. In contrast, renal anomalies,
MA NU
particularly severe renal anomalies have a profound effect on the likelihood of a baby surviving.
The loss of influence of birth weight on survival is unsurprising. Advances in neonatal care have brought survivability down to 24 weeks gestation, and babies born
ED
prematurely with birth weights below 1kg are now routinely operated on for such
PT
conditions as necrotising enterocolitis with expectation of survival7. A number of other studies of mortality in oesophageal atresia have shown no influence of birth
CE
weight 4, 8-10.
AC
Unlike low birth weight, renal compromise in the new born remains a complex therapeutic dilemma, and even with aggressive and novel dialysis techniques, mortality is appreciable11, 12. The particularly poor prognosis of end stage renal failure in early life when additional anomalies are present has been documented, although oesophageal atresia did not feature as a notable co-morbidity in a national survey13. The relative risk for congenital heart disease among babies with oesophageal atresia compared to the general population is 2314. All published evidence is in agreement that severe congenital heart disease is invariably associated with increased mortality in this population.
ACCEPTED MANUSCRIPT Our findings may not be generalizable. In recent studies of survival using health insurance coding data from multiple centres, birth weight was still found to be
T
strongly associated with outcome15-17. This may reflect variability in the outcome of
RI P
small babies within differing units with differing levels of expertise. Alternatively, the health insurance coding did not detect variables such as the achievement of a
SC
primary anastomosis or occurrence of a leak, and this may affect the regression
MA NU
model derived by these authors.
For a prognostic scoring system to have utility it must clearly distinguish patients at different risks and the degrees of difference must be of a magnitude to be clinically important. Empirically, we suggest that a difference in survival rate would be at least
ED
20% to be clinically meaningful in discussions with parents. It must also have relatively few categories, which select meaningful numbers of patients. Ideally it
PT
should be simple with unambiguous criteria and easily memorised. Lastly, it must be
CE
generalizable; a scoring system which produces a perfect model, but only in one hospital is of little use.
AC
A review of published outcomes where scores have been presented in the last 2 decades shows existing scoring systems to be inadequate (Table 7). Either failing to correctly predict progressively poorer outcome using Waterston
10, 18-22
, Spitz 23-25 or
Bremen 21, 22, or producing categories with less than 20% difference in survival rates between them 9, 15, 18-21, 24-29. We find the Montreal classification difficult to apply18. Our proposed classification consists of three groups: 1, neither severe renal nor major cardiac anomaly. 2, either severe renal or major cardiac anomaly. 3, both severe renal and major cardiac anomaly. In our population this produced mortality rates of 2%, 43% and 100%. We suggest this classification has the merits of
ACCEPTED MANUSCRIPT simplicity, unambiguous criteria and near perfect splitting of survival probabilities. Whether our classification is generalizable will require it to be tested in other
T
populations.
RI P
In conclusion, we have found that survival of babies born with oesophageal atresia can be best described using a model with major cardiac anomalies and severe renal
SC
anomalies as covariates. Improvements in survival will be dependent on advances in
MA NU
renal replacement in the new born and better survival of congenital cardiac
AC
CE
PT
ED
anomalies.
ACCEPTED MANUSCRIPT References 1. Burge DM, Shah K, Spark P, Shenker N, Pierce M, Kurinczuk JJ, et al.
T
Contemporary management and outcomes for infants born with oesophageal atresia.
RI P
Br J Surg. 2013;100:515-21.
2. Waterston DJ, Carter RE, Aberdeen E. Oesophageal atresia: tracheo-
SC
oesophageal fistula. A study of survival in 218 infants. Lancet. 1962;1:819-22. 3. Beasley SW, Myers NA. Trends in Mortality in Esophageal Atresia. Pediatr Surg
MA NU
Int. 1992;7:86-9.
4. Engum SA, Grosfeld JL, West KW, Rescorla FJ, Scherer LRT. Analysis of morbidity and mortality in 227 cases of esophageal atresia and/or
ED
tracheoesophageal fistula over 2 decades. Arch Surg. 1995;130:502-8. 5. Spitz L, Kiely EM, Morecroft JA, Drake DP. Oesophageal atresia: at-risk groups
PT
for the 1990s. J Pediatr Surg. 1994;29:723-5. 6. Malakounides G, Lyon PC, De Agustin JC, Cross K, Drake D, Pierro A, et al.
CE
Oesophageal atresia: Improved outcomes in high risk groups? -revisited. British
AC
Association of Paediatric Surgeons International Congress. Bournmouth2013. 7. Lemons JA, Bauer CR, Oh W, Korones SB, Papile LA, Stoll BJ, et al. Very low birth weight outcomes of the National Institute of Child Health and Human Development Neonatal Research Network, January 1995 through December 1996. Pediatrics. 2001;107:art. no.-e1. 8. Calisti A, Oriolo L, Nanni L, Molle P, Briganti V, D'Urzo C. Mortality and long term morbidity in esophageal atresia: the reduced impact of low birth weight and maturity on surgical outcome. J Perinat Med. 2004;32:171-5.
ACCEPTED MANUSCRIPT 9. Choudhury SR, Ashcraft KW, Sharp RJ, Murphy JP, Snyder CL, Sigalet DL. Survival of patients with esophageal atresia: Influence of birth weight, cardiac
T
anomaly, and late respiratory complications. J Pediatr Surg. 1999;34:70-3.
RI P
10. Deurloo JA, Ekkelkamp S, Schoorl M, Heij HA, Aronson DC. Esophageal atresia: Historical evolution of management and results in 371 patients. Ann Thorac Surg.
SC
2002;73:267-72.
11. Everdell NL, Coulthard MG, Crosier J, Keir MJ. A machine for haemodialysing
MA NU
very small infants. Pediatr Nephrol. 2005;20:636-43.
12. Coulthard MG, Crosier J, Griffiths C, Smith J, Drinnan M, Whitaker M, et al. Haemodialysing babies weighing < 8 kg with the Newcastle infant dialysis and
ED
ultrafiltration system (Nidus): comparison with peritoneal and conventional haemodialysis. Pediatr Nephrol. 2014;29:1873-81.
PT
13. Coulthard MG, Crosier J, British Assoc Paediat N. Outcome of reaching end stage renal failure in children under 2 years of age. Arch Dis Child. 2002;87:511-7.
CE
14. Leonard H, Barrett AM, Scott JES, Wren C. The influence of congenital heart
AC
disease on survival of infants with oesophageal atresia. Arch Dis Child. 2001;85:F204-F6.
15. Turner B, Dasgupta R, Brindle ME. A contemporary prediction rule for esophageal atresia (EA) and tracheo-esophageal fistula (TEF). J Pediatr Surg. 2014;49:1758-61. 16. Wang B, Tashiro J, Allan BJ, Sola JE, Parikh PP, Hogan AR, et al. A nationwide analysis of clinical outcomes among newborns with esophageal atresia and tracheoesophageal fistulas in the United States. J Surg Res. 2014;190:604-12.
ACCEPTED MANUSCRIPT 17. Sulkowski JP, Cooper JN, Lopez JJ, Jailcherla Y, Cuenot A, Mattei P, et al. Morbidity and mortality in patients with esophageal atresia. Surgery. 2014;156:483-
T
91.
RI P
18. Poenaru D, Laberge JM, Neilson IR, Guttman FM. A new prognostic classification for esophageal atresia. Surgery. 1993;113:426-32.
SC
19. Okada A, Usui N, Inoue M, Kawahara H, Kubota A, Imura K, et al. Esophageal atresia in Osaka: A review of 39 years' experience. J Pediatr Surg. 1997;32:1570-4.
MA NU
20. Dunn JCY, Fonkalsrud EW, Atkinson JB. Simplifying the Waterston's stratification of infants with tracheoesophageal fistula. Am Surg. 1999;65:908-10. 21. Yagyu M, Gitter H, Richter B, Booss D. Esophageal atresia in Bremen, Germany
ED
- Evaluation of preoperative risk classification in esophageal atresia. J Pediatr Surg. 2000;35:584-7.
PT
22. Konkin DE, O'Hali WA, Webber EM, Blair GK. Outcomes in esophageal atresia and tracheoesophageal fistula. J Pediatr Surg. 2003;38:1726-9.
CE
23. Driver CP, Shankar KR, Jones MO, Lamont GA, Turnock RR, Lloyd DA, et al.
AC
Phenotypic presentation and outcome of esophageal atresia in the era of the Spitz classification. J Pediatr Surg. 2001;36:1419-21. 24. Lilja HE, Wester T. Outcome in neonates with esophageal atresia treated over the last 20 years. Pediatr Surg Int. 2008;24:531-6. 25. Koivusalo AI, Pakarinen MP, Rintala RJ. Modern outcomes of oesophageal atresia: Single centre experience over the last twenty years. J Pediatr Surg. 2013;48:297-303. 26. Teich S, Barton DP, GinnPease ME, King DR. Prognostic classification for esophageal atresia and tracheoesophageal fistula: Waterston versus Montreal. J Pediatr Surg. 1997;32:1075-9.
ACCEPTED MANUSCRIPT 27. Lopez PJ, Keys C, Pierro A, Drake DP, Kiely EM, Curry JI, et al. Oesophageal atresia: improved outcome in high-risk groups? J Pediatr Surg. 2006;41:331-4.
T
28. Okamoto T, Takamizawa S, Arai H, Bitoh Y, Nakao M, Yokoi A, et al.
RI P
Esophageal atresia: Prognostic classification revisited. Surgery. 2009;145:675-81. 29. Niramis R, Tangkhabuanbut P, Anuntkosol M, Buranakitjaroen V, Tongsin A,
SC
Mahatharadol V. Clinical Outcomes of Esophageal Atresia: Comparison Between the Waterston and the Spitz Classifications. Annals Academy of Medicine Singapore.
MA NU
2013;42:297-300.
30. Turner B, Dasgupta R, Brindle ME. A contemporary prediction rule for esophageal atresia (EA) and tracheo-esophageal fistula (TEF). J Pediatr Surg.
AC
CE
PT
ED
2014;49:1758-61.
AC
CE
PT
ED
MA NU
SC
RI P
T
ACCEPTED MANUSCRIPT
Figure 1
AC
CE
PT
ED
MA NU
SC
RI P
T
ACCEPTED MANUSCRIPT
Figure 2
ACCEPTED MANUSCRIPT Table 1 of anomalies in babies who died.
Renal
Cardiac anomaly
1
Cause of death
anomaly
RI P
number
Other anomalies
T
Child
Tracheomalacia
Severe
Pulmonary
with intrarenal
neurological
hypertension
reflux
impairment.
VSD
Down’s syndrome,
Nil
ASD. Small right ventricle. Dextrocardia. Abnormal drainage of
PT
PDA ligated
5
AC
4
PDA. VSD
AVSD. PDA.
Multiple vertebral
and rib anomalies
Bilateral renal
Cloaca.
dysplasia
Hypothyroid. Long
Not known
gap atresia with
CE
3
Imperforate anus.
ED
SVC to left atrium
Single kidney
MA NU
2
SC
duodenal atresia
oesophagostomy Nil
ATN
Necrotising
Klebsiella
enterocolitis
septicaemia
Gastric perforation
Sepsis
Hypoplastic aortic arch 6
Nil
Right
Pulmonary
hydronephrotic
hypoplasia
kidney drained in utero at 25
ACCEPTED MANUSCRIPT weeks. Left multi-cystic dysplastic
Vesico-
double outlet right
ureteric reflux
8
Tetralogy of Fallot
Nil
PT
Nil
Renal failure
CE
10
cardiac surgery Long gap atresia
Pulmonary
with
hypoplasia
oesophagostomy Long gap atresia.
Renal failure
Fistula ligated only. Severe
AC 11
Tracheomalacia
Died during
Nil
ED
9
Hypoplastic left heart
MA NU
ventricle, pulmonary stenosis
RI P
Tetralogy of Fallot,
SC
7
T
kidney
hydrocephalus. Ascites
Severe pulmonary
Bilateral renal
Pulmonary
trunk hypoplasia. VSD.
dysplasia
hypoplasia. Imperforate anus. Limb anomalies. Vertebral anomalies
Care withdrawn
ACCEPTED MANUSCRIPT 12
Single
Imperforate anus.
dysplastic
Cleft palate.
kidney
Structurally
Nil
Care withdrawn
VSD
Nil
Diffuse lymphatic
Uncontrollable
leak
lymphatic
14
Tetralogy of Fallot
MA NU
SC
13
RI P
T
abnormal brain.
Nil
losses
Limb and vertebral
Post mortem
anomalies. Failed
inconclusive.
initial anastomosis Delayed primary anastomosis
PDA. Coarctation.
Absent right,
Trisomy 20
Right ventricular
hydronephrotic
PT
dysplasia
CE
16
Bilateral renal
Nil
ED
15
hypertrophy. Aortic
Renal failure
Care withdrawn
left kidney
AC
stenosis.
PDA: patent ductus arteriosus, VSD: ventricular septal defect, SVC: superior vena cava AVSD: atrioventricular septal defect,
ACCEPTED MANUSCRIPT Table 2. Univariate analysis of possible influential variables on survival of babies with oesophageal atresia
No cardiac anomaly
68/72 (94.5%)
Major cardiac anomaly
11/21 (52.4%)
No major cardiac anomaly
93/99 (93.9%)
Birth weight >1500g
91/103 (87%)
Birth weight< 1500g
13/17 (76.5%)
Birth weight as a continuous variable
0.17 (0.053 0.587)
0.071 (0.022 0.233)
0.005
<0.001
0.429 (0.12 1.52)
0.19
0.33 (0.14 0.77)
0.010
0.1 (0.032 0.32)
<0.001
0.02 ( 0.004 0.1)
<0.001
9.3 (2.74 31.7)
<0.001
1 (0.1 9.4)
0.9
0.9 (0.26 3.0)
0.86
0.67 (0.17 2.66)
0.57
0.41 (0.09 1.71)
0.22
2.33 (0.28 19.07)
0.42
15/25 (60%)
ED
Any renal anomaly
p value*
T
36/48 (75%)
RI P
Any cardiac anomaly
Odds ratio (95% CI)
SC
Survival
MA NU
Variable
No renal anomaly
89/95 (94%) 2/10 (20%)
PT
Severe renal anomaly
Primary anastomosis
AC
No primary anastomosis
CE
No severe renal anomaly
102/110 (93%) 96/105 (91%) 8/15 (53%)
Secondary oesophageal surgery
15/16 (94%)
No secondary oesophageal surgery
97/104 (94%)
Symptomatic tracheomalacia
24/28 (86%)
No symptomatic tracheomalacia
80/92 (87%)
Aortopexy
14/17 (82%)
No aortopexy
90/103 (87%)
Tracheostomy
9/12 (75%)
No tracheostomy
95/108 (88%)
Fundoplication
14/15 (93%)
No fundoplication
90/105 (86%)
ACCEPTED MANUSCRIPT
Abnormal karyotype
4/6 (67%)
Normal karyotype
100/114 (88%)
Abnormal neurology
5/7 (71%)
Normal neurology
99/113 (88%)
Anastomotic leak
15/16 (94%)
No anastomotic leak
89/104 (86%)
Gestation <=28 weeks
2/4 (50%)
Gestation > 28 weeks
102/116 (88%)
Gestation as a continuous variable
CE
PT
ED
* Wald p value
AC
0.26 (0.09 0.79)
T
82/90 (91%)
0.28 (0.05 1.65)
RI P
No gastrostomy
0.35 (0.06 2)
SC
22/30 (73%)
2.5 (0.3 20.5)
MA NU
Gastrostomy
0.018
0.16
0.24
0.3
0.13 (0.01 1.05)
0.05
0.8 (0.7 0.9)
0.005
ACCEPTED MANUSCRIPT Table 3. Multivariate blockwise entry analysis of variables related to survival, using
Odds ratio (95% ci)
Any cardiac anomaly
0.1 (0.017 0.6)
0.01
Any renal anomaly
0.04 (0.007 0.2)
0.001
Birth weight as a continuous variable
0.46 (0.6 3.4)
0.4
0.54 (0.09 3.0)
0.4
0.9 (0.6 1.3)
0.7
12.2 (1.8 81.6)
0.01
Gastrostomy
ED
Gestation as a continuous variable
CE
PT
Primary anastomosis
AC
MA NU
RI P
T
Variable
SC
any cardiac anomaly and any renal anomaly as variables.
p value
ACCEPTED MANUSCRIPT Table 4. Multivariate blockwise entry analysis of variables related to survival, using major cardiac anomaly and severe renal anomaly as variables.
Odds ratio (95% ci)
Major cardiac anomaly
0.044 (0.007 0.29)
Severe renal anomaly
0.009 (0.001 0.12)
<0.001
Birth weight as continuous variable
1.7 (0.2 14.2)
0.6
0.39 (0.05 2.9)
0.36
0.7 (0.47 1.1)
0.1
4.8 (0.5 44)
0.16
RI P
SC
ED
Gestation as a continuous variable
MA NU
Gastrostomy
CE
PT
Primary anastomosis
AC
T
Variable
p value
0.001
ACCEPTED MANUSCRIPT Table 5. Final statistical model of survival for oesophageal atresia
Odds ratio (95% ci)
Constant
4.01 (1.18)
T
B (SE)
RI P
Variable
Major cardiac
0.038 (0.007 0.2)
SC
-3.26 (0.86) * anomaly
MA NU
Severe renal
-4.6 (1.09) * anomaly
AC
CE
PT
ED
* Wald p < 0.001. Nagelkerke’s R2 0.57
0.01 (0.001 0.08)
ACCEPTED MANUSCRIPT Table 6. Univariate analysis of different prognostic scoring systems assessed as ordinal scales. Survival
Odds ratio (95% ci)
p value
Spitz
RI P
T
Scoring system
<0.001
4.1 (1.9 9)
<0.001
39 (8.1 186)
<0.001
4.6 (2 10.7) 81/86 (94%)
2
20/28 (71%)
MA NU
SC
1
3
3/6 (50%)
Modified Spitz
81/86 (94%)
ED
1
11/12 (92%)
PT
2.1
3
AC
Newcastle model
CE
2.2
8/15 (53%) 4/7 (57%)
1 Neither cardiac nor renal
91/93 (98%)
2 Either cardiac or renal
10/23 (43%)
3 Cardiac and renal
0/4 (0%)
ACCEPTED MANUSCRIPT Table 7. Published survival and risk stratification of oesophageal atresia
95
Waterston *+
C 73%
A 100%
B 100%
C 50%
I 92%
II 50%
III 0%
A 100%
B 91%
C 43%
I 91%
II 53%
A 100%
B 100%
C 43%
I 89%
II 64%
III 50%
A 100%
B 100%
C 66%
I 96%
II 57%
III 0%
No complications
I 100%
II 50%
III 0%
Complications
I 33%
II 40%
III 0%
134
Spitz * +
I 92%
II 69%
III 100%
357
Waterston * +
A 86%
B 88%
C 51%
144
Waterston * +
A 100%
B 100%
C 80%
Spitz +
I 99%
II 84%
III 43%
Montreal
I 92%
II 71%
No complications
I 99%
II 84%
III 43%
Complications
I 100%
II83%
III 25%
Spitz +
I 99%
II 82%
III 50%
1993 [18]
Montreal
Okada
112
Waterston * +
1997 [19]
Spitz 94
Waterston +
1997 [26]
Montreal
Dunn
64
Waterston * +
Choudhury 1999 [9]
240
Spitz +
Yagu
133
Waterston * +
2000 [21]
ED
1999 [20]
A 93%
B 94%
I 93%
II 31%
MA NU
Teich
Classification grade and survival
T
Poenaru
Classifications reported
RI P
Year of publication
Number of patients
SC
Author
PT
Spitz
2001 [23] Deurloo
AC
Driver
CE
Bremen * +
2002 [10] Konkin 2003 [22]
Bremen * +
Lopez
188
ACCEPTED MANUSCRIPT 2006 [27] Lilja
65
Spitz * +
I 100%
II 78%
III 100%
52
Spitz +
I 84%
II 76%
III 66%
Revised Spitz +
I 100%
II 88%
III 83%
130
Spitz * +
I 99%
II 94%
III No patients
132
Waterston +
A 100%
B 92%
C 49%
I 97%
II 64%
III 27%
II 87%
III 54%
2009[28] Koivusalo
2013 [29] Turner
Spitz 1219
Spitz +
2014 [30]
MA NU
Niramis
SC
2012 [25]
RI P
Okamoto
I 99%
T
2008 [24]
IV 40%
ED
* Classification where outcomes are inconsistent with progressively poorer grading.
AC
CE
PT
+ Classification where groups show less than 20% difference in survival rates