A predictive scoring system for small diaphragmatic defects in infants with congenital diaphragmatic hernia

doi:10.21203/rs.3.rs-2003289/v1

Purpose: To develop a predictive score for small diaphragmatic defects in infants with congenital diaphragmatic hernia (CDH).

Methods: The Japanese CDH Study Group cohort was randomly divided into derivation (n = 397) and validation (n = 396) datasets. Using logistic regression, a prediction model and weighted scoring system for small diaphragmatic defects were created from derivation dataset and validated with validation dataset.

Results: Six weighted variables were selected: no hydramnios, 1 point; 1-minute Apgar score of 5–10, 1 point; apex type of the lung (left lung is detected radiographically in apex area), 1 point; oxygenation index < 8, 1 point; abdominal nasogastric tube (tip of the nasogastric tube is detected radiographically in abdominal area), 2 points; no right-to-left flow of ductus arteriosus, 1 point. In validation dataset, rates of small diaphragmatic defects for Possible (0–3 points), Probable (4–5 points), and Definite (6–7 points) groups were 36%, 81%, and 94%, respectively (p < 0.001). Additionally, sensitivity, specificity, positive predictive value, and C statistics were 0.78, 0.79, 0.88, 0.76, and 0.45, 0.94, 0.94, 0.70 for Probable and Definite groups, respectively.

Conclusion: Our simple scoring system effectively predicted small diaphragmatic defects in infants with CDH and is useful to determine thoracoscopic surgery indication.

congenital diaphragmatic hernia

congenital abnormalities

statistical models

minimally Invasive surgical procedures

diaphragm

Thoracoscopic repair of congenital diaphragmatic hernia (CDH) has been widely used in neonates since the first report of a pediatric case in 1995 [1] and an infantile case in 2001 [2]. In fact, 11% of neonates with left-sided isolated CDH underwent thoracoscopic surgery between 2006 and 2018 in the database of the Japanese CDH Study Group (JCDHSG) [3]. In more recent single-center reports, thoracoscopic surgery rates for CDH were 29% [4], 46% [5], 69% [6]. Thoracoscopic surgery for CDH has great advantages of better cosmesis, better postoperative recovery, shorter hospital stays and a lower incidence of chronic lung disease [3, 7]. Furthermore, the potential prevention of thoracic deformity and adhesive-intestinal obstruction is also expected. However, a higher rate of recurrence of diaphragmatic hernia is of great concern. According to a systematic review, the risk of recurrence in thoracoscopic surgery was threefold higher than that in open surgery [8]. Another systematic review reported that patch repair was associated with a higher risk of recurrence than direct closure in thoracoscopic surgery [9]. A similar trend was also observed in open surgery [10]. The choice of patch depends entirely on the size of the defect. Therefore, diaphragmatic defect size is an important factor in thoracoscopic surgery for CDH. Some groups limit endoscopic surgery to small defects or convert it to open surgery if the defect is too large for the use of a patch. [9,11,12]. However, we cannot observe the defect preoperatively, and there is no reliable predictive method for the size of defects. Therefore, this study aimed to establish and validate a predictive score for small diaphragmatic defects in infants with CDH.

Study design and patient selection

The JCDHSG conducted a multicenter cohort study with a retrospective design from 2006 to 2016, which is currently followed up by a prospective cohort study since 2022. Its database includes infants with CDH who were diagnosed prenatally or within 28 days after birth and were hospitalized in 15 participating centers between January 2006 and December 2020.

In the present study, we reviewed the JCDHSG database and selected patients with left CDH who underwent surgery for this condition. Patients with life-limiting chromosomal abnormalities, such as trisomy 13 or 18, or severe cardiac malformations that affect respiratory management were excluded. The cohort was randomly divided into two groups: derivation and validation. A random number between 0 and 1 was allocated for each patient, and the patients were assigned to two groups according to whether or not the random number was greater than 0.5. In all participating centers, patient management has been based on the same therapeutic protocol since 2017. The surgical approach (open or thoracoscopic surgery) was decided based on each institution’s criteria.

Variables

The following variables were extracted: sex, gestational age, birth weight, Apgar score at 1-minute, location of birth (inborn or outborn), fetal diagnosis of CDH, history of hydramnios and fetal hydrops, liver-up, which was defined as prenatally detected liver herniation occupying more than one-third of the thoracic space [13], thoracic position of stomach, international classification of the diaphragmatic defect [14], closure of the diaphragmatic defect (direct closure or not), the lowest oxygenation index (OI) within 24 h after birth which was calculated using the best PaO₂ and ventilator settings at the time of blood sampling, the need for extracorporeal membrane oxygenation (ECMO), and life prognosis at 90 days of age. The left lung type in postnatal X-ray was also used as a variable. “Apex type” was defined as radiographical detection of the left lung in the apex area, but not in the hilar area [15]. Left-to-right dominant flow of ductus arteriosus (DA) at the first ultrasound examination after birth was used as a variable. The location of the nasogastric tube tip was also assessed. An “abdominal nasogastric tube” was defined as a radiographic detection of the nasogastric tube tip in the abdominal area, but not in the thoracic area.

Endpoints

The primary outcome of this study was “small diaphragmatic defect”, defined by the international classification of diaphragmatic defect as class A or B. “Large diaphragmatic defect” was defined by the international classification of diaphragmatic defect as class C or D [14].

Derivation of the model

We developed a prediction model using a derivation dataset. We conducted a multivariate logistic regression analysis using the potential independent variables. The variables were selected from the postnatal and preoperative patients’ information, and prenatal information which was available in patients without prenatal diagnosis of CDH, such as history of hydramnios and fetal hydrops. These variables had significant difference between small and large diaphragmatic defects in univariate analysis and had missing data less than 15% in the JCDHSG database. Continuous variables were dichotomized as follows: gestational week, < 37 weeks and ≥ 37 weeks; birth weight, ≤ 2,500 g and > 2,500 g; Apgar score at 1 minute, 1–4 and 5–10; OI, < 8 and ≥ 8 [16]. The variables for the model were selected by a forward-backward stepwise selection method with Akaike’s information criterion (AIC) [17].

Risk score development

A weighted scoring system for small diaphragmatic defects was created based on logistic regression analysis. The selected variables were weighted proportionally to their adjusted parameter estimates. Each parameter estimate, divided by the standard error, was divided by the minimum value and rounded to the nearest integer. Patients were separated into three categories, “Possible,” “Probable” and “Definite” based on the score distribution.

Validation

Subsequently, we evaluated the generalization ability of the best models using a validation dataset. We calculated the area under the curve (AUC) (e.g., the C-index) to assess the discriminatory power of the models. The calibration curve of the model was plotted using the Platt scaling method [18]. The predictive ability of the scoring system was assessed by its sensitivity, specificity, positive predictive value, and negative predictive value.

Statistical analysis

Summary statistics were expressed as frequencies and proportions for categorical data and as median and interquartile range (IQR) for continuous variables. Statistical comparisons between the two groups were performed using the Wilcoxon signed-rank test for continuous variables and Chi-squared or Fisher’s exact test for categorical variables. Areas under the receiver operating characteristic (ROC) curve (e.g., the C-index) were constructed to assess the discriminatory power of each risk factor. Statistical significance was defined as a two-sided p-value < 0.05. All statistical analyses were performed using JMP software (version 12.01; SAS Institute, Inc., Cary, NC, USA).

Ethical approval

This study was approved by the Ethics Committee of Chiba University Hospital (approval number: 509). The requirement for signed informed consent was waived because of the retrospective study design and use of de-identified data. Details of the study were published on an institutional website, and individuals had the right to decline their participation. The study was performed in accordance with the principles of the Declaration of Helsinki and the ethical guidelines for medical and health research involving human subjects.

Baseline characteristics

A total of 1,034 patients were included in the JCDHSG database during the study period. Of these, 793 left-isolated patients who underwent surgery were included in the study. The patients were randomly divided into two groups: derivation (n = 397) and validation (n = 396) datasets. There were no significant differences in variables between the two datasets.

The following significant differences between two groups with small and large diaphragmatic defects (Table 1) were observed: gestational age at birth [38.0 (IQR 37.3 – 38.9) vs. 37.4 (36.9 – 38.1) weeks, p < 0.0001], history of hydramnios, [32 / 216 (15%) vs. 54 / 141 (38%), p < 0.0001], fetal diagnosis of CDH [183 / 236 (78%) vs. 141 / 143 (99%), p < 0.0001], liver-up [11 / 180 (6%) vs. 81 / 137 (59%), p < 0.0001], thoracic position of stomach [102 / 178 (57%) vs. 127 / 136 (93%), p < 0.0001], outborn, [48 / 237 (20%) vs. 4 / 143 (3%), p < 0.0001], birth weight [2,744 (IQR 2,46 – 3,060) vs. 2,636 (IQR 2,348 – 2,914) g, p = 0.0065], Apgar score at 1 minute [6 (IQR 4 – 8) vs. 4 (IQR 2 – 5)], p < 0.0001], apex type of the left lung [139 / 215 (65%) vs. 37 / 128 (29%), p < 0.0001], left-to-right dominant flow of DA [164 / 219 (75%) vs. 54 / 135 (40%), p < 0.0001], lowest OI [3.4 (IQR 2.8 – 4.9) vs. 6.3 (4.2 – 16.3), p < 0.0001], abdominal nasogastric tube [127 / 211 (60%) vs. 23 / 121 (19%), p < 0.0001], direct closure [206 / 237 (87%) vs. 16 / 143 (11%), p < 0.0001], ECMO requirement [5 / 237 (2%) vs. 22 / 143 (15%), p < 0.0001], death at 90 days of age [7 / 237 (3%) vs. 25 / 143 (17%), p < 0.0001].

Derivation of the model

A multiple logistic regression analysis showed that the following six predictors were independent predictors of small diaphragmatic defects: no hydramnios [adjusted odds ratio (OR), 2.9; 95% confidence interval (CI), 1.7 – 5.1; p < 0.0001); Apgar score at 1 minute of 5–10 (adjusted OR, 2.4; 95% CI, 1.5 – 3.7; p < 0.0001); apex type of the left lung (adjusted OR, 2.9; 95% CI, 1.8 – 4.6; p < 0.001); OI < 8 (adjusted OR, 2.5; 95% CI, 1.4 – 4.2; p = 0.001); left-to-right dominant flow of DA (adjusted OR, 2.1; 95% CI, 1.3 – 3.4; p = 0.001); abdominal nasogastric tube (adjusted OR, 4.2; 95% CI, 2.5 – 6.9; p < 0.001) (Table 2). The gestational week and birth weight variables were eliminated through selection. The following equation was derived:

Probability of unsatisfactory prognosis = 1 – 1/(1 + e-X)

where X = 0.20 + 0.53 (No hydramnios) – 0.44 (Apgar score at 1 minute of 5–10) – 0.52 (Apex type of the left lung) + 0.45 (OI < 8) + 0.38 (Left-to-right dominant flow of DA) + 0.72 (Abdominal nasogastric tube).

Sscore development

A predictive scoring system for small diaphragmatic defects was developed using the sum of the weighted predictors present in each case (no hydramnios, 1 point; Apgar score at 1 minute of 5–10; 1 point, apex type of the left lung; 1 point, OI < 8, 1 point; left-to-right dominant flow of DA, 1 point; and abdominal nasogastric tube, 2 points) (Figure 1). According to the observed rate of small diaphragmatic defects for each point, the patients were divided into three groups: Possible (0–3), Probable (4–5), and Definite (7–8) groups (Figure 1).

Validation of the model and the scoring system

In the validation dataset, the C statistics of the model for Probable and Definite groups (4–7 points) and Definite group (6–7 points) were 0.76 and 0.70, respectively (Table 4). The calibration plot showed that the model almost fitted, but tended to overestimate the probability of a small diaphragmatic defect (Figure 2). The observed predictors, morbidity rates, and mortality rates were compared among the Possible, Probable, and Definite groups in the validation dataset (Table 3). There were significant differences among the three groups regarding history of hydramnios (p < 0.01), liver-up (p < 0.0001), thoracic position of the stomach (p < 0.0001), Apgar score at 1 minute (p < 0.01), apex type of the left lung (p < 0.0001), left-to-right dominant flow of DA (p < 0.0001), lowest OI (p < 0.05), abdominal nasogastric tube (p < 0.0001), diaphragmatic defect class A and B (p < 0.0001), direct closure (p < 0.0001), ECMO requirement (p < 0.05), and death at 90 days of age (p < 0.0001). The sensitivity, specificity, positive predictive value, and negative predictive value for the Definite group (6–7 points) were 0.45, 0.94, 0.94, and 0.46, respectively (Table 4).

In this study, we developed a predictive scoring system for small diaphragmatic defects in patients with CDH. This system can be used for patients who had no prenatal diagnosis of CDH. In the validation dataset, the model fitted appropriately, and the prediction ability of the Definite group (6–7 points) was sufficient, particularly with high specificity and positive predictive value. This model enabled us to easily predict the size of diaphragmatic defects before surgery for CDH.

The main motivation of this study was to decrease the rate of recurrence by preoperatively predicting the size of the diaphragmatic defect. A systematic review of thoracoscopic surgery for CDH showed a higher rate of recurrence in thoracoscopic surgery than in open surgery, although thoracic surgery tended to be applied to non-severe cases. Furthermore, the rate of recurrence of thoracoscopic surgery was possibly underestimated because the calculation commonly included cases with conversion to open surgery [8].

One of the most powerful risk factors of CDH recurrence is the need for patch repair owing to large diaphragmatic defect [10]. Particularly in thoracoscopic surgery, handling the patch material in the narrow thoracic space of neonates is challenging. Therefore, the size of the diaphragmatic defect may affect the outcome, with or without recurrence. Some surgeons with preference for the thoracoscopic approach may suggest that conversion to open surgery is reasonable when the defect is too large to accomplish direct closure [9]. In fact, the median rate of conversion in each facility was 25% in a systematic review [8]. Conversion to open surgery implies that a previous thoracoscopic approach was unnecessary. It is well known that thoracoscopic surgery stresses patients due to severe intraoperative hypercapnia and acidosis [19]. Furthermore, open surgery subsequent to thoracic surgery results in prolonged operative time, potentially increasing operative stress and difficulty in the postoperative management of pulmonary hypertension. Therefore, pediatric surgeons should determine the criteria for thoracoscopic repair of CDH in consideration of the risk of conversion [20].

A low recurrence rate of thoracoscopic surgery with patch usage was documented in a comparative report [5]. Strict criteria for application have enabled a low rate of recurrence even in thoracoscopic surgery [3]. It is important to consider the indication depending on the experience and skill of each surgeon and the team of institutions. This model must provide useful information regarding the preoperative size of the diaphragmatic defect. For instance, thoracoscopic surgery is applied only to the Definite group in the introduction stage, and then the indication is expanded to the Probable group in accordance with the learning curve. This model is also useful for parental counselling in choosing the surgical approach.

The size of the diaphragmatic defect is an internationally accepted, simple, and powerful risk factor in patients with CDH [14], of which the prediction ability increased further in combination with other factors [21]. However, the greatest disadvantage of this method is that the defect cannot be observed preoperatively. We cannot use this useful risk factor until surgical findings have been obtained. The present model, which was the first to predict diaphragmatic defects, can also predict the prognosis of CDH. In our study, there was no death in the Definite group ,while 15% of patients died in the Possible group.

This study had several limitations that need to be addressed in future investigations. First, owing to its multicenter design, therapeutic strategies were not uniform until 2017. Second, the study population was limited to a single race. Therefore, our results should be validated externally in a cohort of other ethnicities for broader application. Third, although the discrimination of the Definite group in the model was promising, the calibration of the model had room for improvement. Further adjustments to the model are required. Prenatal factors, which were not used in this model mainly because of the higher rate of missing values, might have an advantage for higher adjustment. Actually, necessity of patch closure was well predicted by prenatally detected liver-up and contralateral lung-to-thorax transverse area ratio [22]. Fourth, this scoring system cannot be applied to patients with right CDH because of the process of derivation.

Our simple scoring system effectively predicted small diaphragmatic defects in infants with CDH and is useful to determine thoracoscopic surgery indication.

Acknowledgments: This work was supported by a grant from the Ministry of Health, Labour, and Welfare of Japan (grant number 20FC1017).

Author contributions

K.T. designed the study and wrote the initial draft of the manuscript. K.N., M.Y., S.A., A.Y., T.F., K.M., and T.O. contributed to the analysis and interpretation of the data and assisted in the preparation of the manuscript. M.H., H.O., N.I., K.T., and N.U. served as scientific advisors and critically reviewed the study proposals. Y.S. performed statistical analyses. All other authors contributed to data collection, interpretation, and critical review of the manuscript. The final version of the manuscript was approved by all authors.

Competing interests

The authors have no relevant financial or non-financial interests to disclose.

Funding

This work was supported by a grant from the Ministry of Health, Labour and Welfare of Japan (grant number 20FC1017).

Ethics approval

This study was approved by the Ethics Committee of Chiba University Hospital (approval number: 509). The study was performed in accordance with the principles of the Declaration of Helsinki and the ethical guidelines for medical and health research involving human subjects.

Consent to participate

The requirement for signed informed consent was waived because of the retrospective study design and use of de-identified data. Details of the study were published on an institutional website, and individuals had the right to decline their participation.

Code availability

The datasets used in this study are available from the corresponding author upon reasonable request.

Silen ML, Canvasser DA, Kurkchubasche AG, Andrus CH, Naunheim KS (1995) Video-assisted thoracic surgical repair of a foramen of Bochdalek hernia. Ann Thorac Surg 60:448-450. https://doi.org/10.1016/0003-4975(95)00100-y
Becmeur F, Jamali RR, Moog R et al (2001) Thoracoscopic treatment for delayed presentation of congenital diaphragmatic hernia in the infant. A report of three cases. Surg Endosc 15:1163-1166. https://doi.org/10.1007/s004640090064
Okawada M, Ohfuji S, Yamoto M et al (2021) Thoracoscopic repair of congenital diaphragmatic hernia in neonates: findings of a multicenter study in Japan. Surg Today 51:1694-1702. https://doi.org/10.1007/s00595-021-02278-6
Romnek MJ, Diefenbach K, Tumin D, Tobias JD, Kim S, Thung A (2020) Postoperative Clinical Course and Opioid Consumption Following Repair of Congenital Diaphragmatic Hernia: Open Versus Thoracoscopic Techniques. J Laparoendosc Adv Surg Tech A 30:590-595. https://doi.org/ 10.1089/lap.2019.0510
Tyson AF, Sola R Jr, Arnold MR, Cosper GH, Schulman AM (2017) Thoracoscopic Versus Open Congenital Diaphragmatic Hernia Repair: Single Tertiary Center Review. J Laparoendosc Adv Surg Tech A 27:1209-1216. https://doi.org/10.1089/lap.2017.0298
Costerus S, Zahn K, van de Ven K, Vlot J, Wessel L, Wijnen R (2016) Thoracoscopic versus open repair of CDH in cardiovascular stable neonates. Surg Endosc 30:2818-2824. https://doi.org/10.1007/s00464-015-4560-8
Vijfhuize S, Deden AC, Costerus SA, Sloots CE, Wijnen RM (2012) Minimal access surgery for repair of congenital diaphragmatic hernia: is it advantageous?—an open review. Eur J Pediatr Surg 22:364-373. https://doi.org/10.1055/s-0032-1329532
Terui K, Nagata K, Ito M et al (2015) Surgical approaches for neonatal congenital diaphragmatic hernia: a systematic review and meta-analysis. Pediatr Surg Int 31:891-897. https://doi.org/10.1007/s00383-015-3765-1
Chan E, Wayne C, Nasr A (2014) Minimally invasive versus open repair of Bochdalek hernia: a meta-analysis. J Pediatr Surg 49:694-699. https://doi.org/10.1016/j.jpedsurg.2014.02.049
Nagata K, Usui N, Terui K et al (2015) Risk Factors for the Recurrence of the Congenital Diaphragmatic Hernia-Report from the Long-Term Follow-Up Study of Japanese CDH Study Group. Eur J Pediatr Surg 25:9-14. https://doi.org/10.1055/s-0034-1395486
Schneider A, Becmeur F (2018) Pediatric thoracoscopic repair of congenital diaphragmatic hernias. J Vis Surg 28:4:43. https://doi.org/10.21037/jovs.2018.02.03
Cioci AC, Urrechaga EM, Parreco J et al (2021) One-year outcomes of congenital diaphragmatic hernia repair: Factors associated with recurrence and complications. J Pediatr Surg 56:1542-1546. https://doi: 10.1016/j.jpedsurg.2020.09.018
Nagata K, Usui N, Kanamori Y (2013) The current profile and outcome of congenital diaphragmatic hernia: a nationwide survey in Japan. J Pediatr Surg 48:738-744. https://doi.org/10.1016/j.jpedsurg.2012.12.017
Lally KP, Lasky RE, Lally PA et al (2013) Standardized reporting for congenital diaphragmatic hernia--an international consensus. J Pediatr Surg 48:2408-2415. https://doi.org/10.1016/j.jpedsurg.2013.08.014
R Shimono, S Ibara, Y Maruyama et al (2010) Radiographic findings of diaphragmatic hernia and hypoplastic lung. J Perinatol 30:140-143. https://doi.org/10.1038/jp.2009.129
Terui K, Nagata K, Kanamori Y et al (2017) Risk stratification for congenital diaphragmatic hernia by factors within 24 h after birth. J Perinatol 37:805-808. https://doi.org/10.1038/jp.2017.11
Akaike H (1998) Information theory and an extension of the maximum likelihood principle. In: Parzen E, Tanabe K, Kitagawa G (ed). Selected papers of Hirotugu Akaike. Springer, New York, pp 199-213.
Niculescu-Mizil A, Caruana R (2005) Predicting good probabilities with supervised learning. ICML 2005 - Proc 22nd Int Conf Mach Learn 625-632, https://doi.org/10.1145/1102351.1102430
Bishay M, Giacomello L, Retrosi G et al (2013) Hypercapnia and acidosis during open and thoracoscopic repair of congenital diaphragmatic hernia and esophageal atresia: results of a pilot randomized controlled trial. Ann Surg 258:895-900. https://doi.org/10.1097/SLA.0b013e31828fab55
Inoue M, Uchida K, Kusunoki M (2017) Letter to the Editor regarding the article "Congenital diaphragmatic hernia in neonates: factors related to failure of thoracoscopic repair". Pediatr Surg Int 33:513-514. https://doi.org/10.1007/s00383-016-4055-2
Cochius-den Otter SCM, Erdem Ö, van Rosmalen J et al (2020) Validation of a Prediction Rule for Mortality in Congenital Diaphragmatic Hernia. Pediatrics 145:e20192379. https://doi.org/10.1542/peds.2019-2379
Masahata K, Usui N, Shimizu Y et al (2020) Clinical outcomes and protocol for the management of isolated congenital diaphragmatic hernia based on our prenatal risk stratification system. J Pediatr Surg 55:1528-1534. https://doi: 10.1016/j.jpedsurg.2019.10.020

Table 1. Demographic data of the validation and derivation datasets

	Small diaphragmatic defect (n = 237)	Large diaphragmatic defect (n = 143)	P value
Male sex, n (%)	209 / 397 (53)	214 / 396 (54)	1.000
Gestational age at birth (week)	38.0 (IQR 37.3 – 38.9)	37.4 (IQR 36.9 – 38.1)	< 0.0001
History of hydramnios, n (%)	32 / 216 (15)	54 / 141 (38)	< 0.0001
History of fetal hydrops, n (%)	3 / 216 (1)	1 / 142 (1)	1.000
Fetal diagnosis of CDH, n (%)	183 / 236 (78)	141 / 143 (99)	< 0.0001
Liver-up, n (%)	11 / 180 (6)	81 / 137 (59)	< 0.0001
Thoracic position of stomach, n (%)	102 / 178 (57)	127 / 136 (93)	< 0.0001
Outborn, n (%)	48 / 237 (20)	4 / 143 (3)	< 0.0001
Birth weight (g)	2,744 (IQR 2,460 – 3,060)	2,636 (IQR 2,348 – 2,914)	0.0065
Apgar score at 1 minute	6 (IQR 4 – 8)	4 (IQR 2 – 5)	< 0.0001
Apex type of the left lung	139 / 215 (65)	37 / 128 (29)	< 0.0001
Left-to-right dominant flow of DA, n (%)	164 / 219 (75)	54 / 135 (40)	< 0.0001
Lowest oxygenation index	3.4 (IQR 2.8 – 4.9)	6.3 (IQR 4.2 – 16.3)	< 0.0001
Abdominal nasogastric tube	127 / 211 (60)	23 / 121 (19)	< 0.0001
Direct closure, n (%)	206 / 237 (87)	16 / 143 (11)	< 0.0001
ECMO requirement, n (%)	5 / 237 (2)	22 / 143 (15)	< 0.0001
Death at 90 days of age, n (%)	7 / 237 (3)	25 / 143 (17)	< 0.0001
Abbreviations: CDH; congenital diaphragmatic hernia; DA, ductus arteriosus; ECMO, extracorporeal membrane oxygenation; IQR, interquartile range.

Table 2. Univariate and multivariate analyses of selected factors for the prediction model of small defect

Variable	Univariate analysis		Multivariate analysis		Integer score assigned
Variable	Adjusted OR (95% CI)	P value	Adjusted OR (95% CI)	P value	Integer score assigned
No hydramnios	3.3 (2.2 – 4.7)	< 0.001	2.9 (1.7 – 5.1)	< 0.001	1
Apgar score (1 minute) of 5 – 10	4.4 (3.2 – 6.1)	< 0.001	2.4 (1.5 – 3.7)	< 0.001	1
Apex type of the left lung	5.1 (3.6 – 7.2)	< 0.001	2.9 (1.8 – 4.6)	< 0.001	1
Oxygenation index < 8	5.2 (3.6 – 7.6)	< 0.001	2.5 (1.4 – 4.2)	0.001	1
Left-to-right dominant flow of DA	4.2 (3.0 – 5.9)	< 0.001	2.1 (1.3 – 3.4)	0.001	1
Abdominal nasogastric tube	7.4 (5.0 – 10.9)	< 0.001	4.2 (2.5 – 6.9)	< 0.001	2
Abbreviations: CI, confidence interval; DA, ductus arteriosus; OR, odds ratio;

Table 3. Comparison of the scoring system of the three groups in the validation dataset

	Possible group (n = 111)	Probable group (n = 72)	Definite group (n = 83)	P value
Male sex, n (%)	61 / 111 (55)	36 / 72 (50)	42 / 83 (51)	0.7551
Gestational age at birth (weeks)	37.7 (IQR 37.1 – 38.4)	38.1 (IQR 37.4 – 38.7)	38.0 (IQR 37.3 – 38.9)	0.0432^§
History of hydramnios, n (%)	28 / 111 (25)	10 / 72 (14)	7 / 83 (8)	0.0062
History of fetal hydrops, n (%)	1 / 108 (1)	0 / 72 (0)	0 / 83 (0)	0.4866
Fetal diagnosis of CDH, n (%)	109 / 111 (98)	66 / 72 (92)	74 / 83 (89)	0.0286
Liver-up, n (%)	53 / 108 (49)	7 / 66 (11)	5 / 71 (7)	< 0.0001
Thoracic position of stomach, n (%)	102 / 108 (94)	44 / 63 (70)	19 / 69 (28)	< 0.0001
Outborn, n (%)	4 / 111 (4)	6 / 72 (8)	8 / 83 (10)	0.2096
Birth weight (g)	2,694 (IQR 2,372 – 2,990)	2,736 (IQR 2,333 – 3,088)	2,920 (IQR 2,506 – 3,160)	0.0052^§
Apgar score at 1 minute	3 (IQR 2 – 4)	5 (IQR 3 – 7)	6 (IQR 5 – 8)	< 0.0001^§† 0.0093^*
Apex type of the left lung	29 / 111 (26)	39 / 72 (54)	69 / 83 (83)	< 0.0001
Left-to-right dominant flow of DA, n (%)	36 / 111 (32)	49 / 72 (68)	74 / 83 (89)	< 0.0001
Lowest oxygenation index	5.5 (IQR 3.7 – 13.8)	3.4 (IQR 2.7 – 6.2)	3.2 (IQR 2.4 – 4.0)	< 0.0001^§† 0.0374^*
Abdominal nasogastric tube	4 / 111 (4)	35 / 72 (49)	83 / 83 (100)	< 0.0001
Defect size of the diaphragm, A+B, n (%)	38 / 106 (36)	56 / 69 (81)	76 / 81 (94)	< 0.0001
Direct closure, n (%)	32 / 111 (29)	51 / 72 (71)	74 / 83 (89)	< 0.0001
ECMO requirement, n (%)	9 / 111 (8)	2 / 72 (3)	0 / 83 (0)	0.0155
Death at 90 days of age, n (%)	17 / 111 (15)	1 / 72 (1)	0 / 83 (0)	< 0.0001
Abbreviations: CDH, congenital diaphragmatic hernia; DA, ductus arteriosus; ECMO, extra corporeal membrane oxygenation; IQR, interquartile range. §, Possible vs Definite group; †, Possible vs Probable group ; *, Probable vs Definite group.

Table 4. Prediction ability of the scoring system in the validation dataset

	Probable and Definite group (4-7 points)	Definite group (6-7 points)
C statistics	0.76	0.70
Sensitivity	0.78	0.45
Specificity	0.79	0.94
Positive predictive value	0.88	0.94
Negative predictive value	0.64	0.46

No competing interests reported.

A predictive scoring system for small diaphragmatic defects in infants with congenital diaphragmatic hernia

Status:

Version 1

Abstract

Figures

Introduction

Methods

Results

Discussion

Conclusion

Declarations

References

Tables

Additional Declarations

Status:

Version 1