OBJECTIVE: The goal was to examine the association of gender with intraventricular hemorrhage (IVH) or periventricular leukomalacia (PVL) in a large national database.
METHODS: The National Inpatient Sample database was analyzed for the years 1998, 1999, 2001, 2002, and 2004 and its pediatrics version for the years 1997, 2000, and 2003. The National Inpatient Sample and its pediatrics version are annually collected databases for inpatient admissions from >1000 hospitals across the United States. We included all very low birth weight (VLBW) infants (birth weight [BW] of <1500 g) and excluded infants with major congenital anomalies. VLBW infants with IVH or PVL were identified by using International Classification of Diseases, Ninth Revision, diagnostic codes. We analyzed the data by using χ2 and Fisher's exact tests to calculate odds ratios (ORs) and logistic regression analysis to control for possible confounders.
RESULTS: The data included 104 847 VLBW infants. With controlling for significant confounders, IVH was associated with male gender (15.9% vs 13.6%; adjusted OR [AOR]: 1.15 [95% confidence interval: 1.11–1.19]; P < .001). More boys than girls had severe IVH (38% vs 32.7%; AOR: 1.18 [95% confidence interval: 1.06–1.32]; P < .004). The increased rates of IVH and severe IVH in boys were significant in the BW subgroups of <1000 g and 1000 to 1499 g. The association of male gender with IVH and severe IVH was more significant in infants of 1000 to 1499 g (AOR: 1.19 vs 1.14; P = .006). The incidence of PVL in VLBW infants was 0.41%, and rates did not differ between genders (0.38% vs 0.43%; P = .42).
CONCLUSIONS: Compared with girls, VLBW male newborns are at greater risk to develop IVH and severe IVH but not PVL. The association of male gender with IVH or severe IVH is stronger with higher BWs.
WHAT'S KNOWN ON THIS SUBJECT:
Male gender is associated with higher rates of death, RDS, and chronic lung disease. In a small study, male gender also was associated with IVH but not PVL.
WHAT THIS STUDY ADDS:
In a large national database, male gender was associated with a high incidence of IVH and an increased rate of severe IVH but not PVL in VLBW infants. This association was more prominent in infants of 1000 to 1499 g.
Neurovascular lesions, including intraventricular hemorrhage (IVH) and periventricular leukomalacia (PVL), are among the leading causes of death and morbidity among preterm infants. Many infants with severe IVH or PVL suffer from major impairments such as cerebral palsy and learning disabilities. The risk of IVH increases in premature infants with lower gestational age (GA) or birth weight (BW). It also increases in conditions of hemodynamic instability, such as hypovolemia, mechanical ventilation, sepsis, or pneumothorax.1,2
Boys are more susceptible to adverse outcomes of prematurity, including death,3,4 respiratory distress syndrome (RDS), and bronchopulmonary dysplasia.5,6 Preterm girls have been shown to have better survival rates7,8 and better neurodevelopmental outcomes among survivors.9,–,13 In a small cohort, IVH was shown to be more common in preterm boys.14 Data from indomethacin IVH prevention trials revealed that indomethacin reduced the incidence of IVH by one half and eliminated parenchymal hemorrhage in preterm boys but not girls.15 Reanalysis of data from the Trial of Indomethacin Prophylaxis in Preterms study showed that treated preterm boys had lower mortality and morbidity rates than did girls.16
Despite the importance of IVH in determining outcomes for preterm neonates, data on the relationship between IVH and gender are scarce. Our aim for this cohort analysis was to examine the association of IVH and PVL with male gender by using the Healthcare Cost and Utilization Project (H-CUP) databases. We hypothesized that IVH incidence and severity and PVL incidence would be greater in boys.
This study used the National Inpatient Sample (NIS) database and its pediatric version, the Kids' Inpatient Database (KID). These data sets were produced by the H-CUP through sponsorship of the federal Agency for Healthcare Research and Quality. H-CUP databases include data for millions of hospital stays collected from ∼1000 hospitals across the United States each year. The NIS includes >100 data elements for each hospital stay, such as primary and secondary diagnoses (International Classification of Diseases, Ninth Revision [ICD-9] codes), primary and secondary procedures (Current Procedural Terminology codes), admission and discharge status, patient demographic features, expected payment source, and total charges. NIS data represent a 10% sample of all hospital admissions during any given year for patients of all ages. The KID is a database of hospital inpatient stays that are available only for the years 1997, 2000, and 2003. It includes samples of pediatric discharges from >2500 to 3500 US community general and specialty hospitals. The KID includes data elements similar to those included in the NIS data set.17
We analyzed KID data for the years 1997, 2000, and 2003. For the years for which KID data were not available, we used NIS data when available (1998, 1999, 2001, 2002, and 2004). To identify very low BW (VLBW) infants, we used ICD-9 diagnostic codes 76515, 76514, 76503, 76502, and 76501 for BW categories of 1250 to 1499, 1000 to 1249, 750 to 999, 500 to 749, and <500 g, respectively. We used ICD-9 diagnostic codes 7721, 77210, 77211, 77212, 77213, and 77214 for the occurrence of IVH and diagnostic codes 77211 to 77214 for the grade of IVH.1,–,4 We considered IVH grades 3 and 4 to be severe IVH. Diagnostic codes for the grade of IVH were not available for all patients. Therefore, severe IVH was compared between groups as proportions of the infants who had codes for IVH grade. The diagnostic code 7797 was used to identify PVL. We excluded preterm newborns with central nervous system anomalies, congenital heart disease except patent ductus arteriosus (PDA), lung anomalies, congenital abdominal wall defects, multiple congenital anomalies, and chromosomal disorders. These associated diagnoses may contribute directly to the occurrence of IVH or PVL or may affect the clinical courses or outcomes of these patients. The study was approved by the institutional review board at the George Washington University Medical Center.
Data Management and Analysis
We classified infants into 2 groups (boys and girls) according to their gender. The 2 groups were further divided into 2 subgroups (<1000 g and 1000–1499 g) according to their BW. We used χ2 and Fisher's exact tests (SAS 8.2; SAS Institute, Cary, NC) to calculate odds ratios (ORs) of IVH, severe IVH, and PVL in association with male gender. In logistic regression models, we calculated adjusted ORs (AORs) after controlling for demographic and clinical confounding variables. ICD-9 diagnostic codes were used to identify the following confounders: race, mode of delivery, birth asphyxia, fetal acidemia, apnea of prematurity, RDS, persistent pulmonary hypertension of the newborn (PPHN), pneumothorax, pulmonary hemorrhage, PDA, sepsis, necrotizing enterocolitis, maternal hypertension, maternal infection or chorioamnionitis, placental infarction or insufficiency, prepartum hemorrhage, cord prolapse, nuchal or abnormal cord, breech presentation, precipitous delivery, instrumental delivery, and neonatal transport.
The study included 104 847 VLBW infants. The distribution of the study population in groups and subgroups is presented in Fig 1. Perinatal and neonatal risk factors associated with IVH are presented in Table 1. Infants with IVH had greater incidences of asphyxia, RDS, PPHN, PDA, pulmonary hemorrhage, pneumothorax, necrotizing enterocolitis, and neonatal sepsis. Differences between genders with regard to clinical and demographic variables are presented in Table 2.
The overall incidence of IVH in VLBW infants was 14.8%. With controlling for significant confounding variables, IVH was associated with male gender, compared with female gender (15.9% vs 13.6%; AOR: 1.15 [95% confidence interval [CI]: 1.11–1.19]; P < .001). Of all cases of IVH, 43.9% had diagnostic codes for the grade of IVH. Of those, 35.64% cases were severe IVH. More boys than girls had severe IVH (38% vs 32.7%; AOR: 1.18 [95% CI: 1.06–1.32]; P < .004). The association of IVH and severe IVH with male gender was significant in the BW subgroups of <1000 g and 1000 to 1499 g (Fig 2).
The incidence and severity of IVH decreased with increasing BW in both genders. The association of male gender with IVH was more significant in infants of 1000 to 1499 g, compared with those of <1000 g (AOR: 1.19 vs 1.14; P = .006). The AORs for preterm boys with severe IVH in the 2 BW subgroups were 1.25 and 1.15, respectively (P = .054) (Fig 3).
The overall incidence of PVL in VLBW infants was 0.41%, and rates did not differ between boys and girls (0.38% vs 0.43%; P = .42). In the BW subgroup of <1000 g, PVL was increased in girls (0.37% vs 0.53%; P = .035 in bivariate analysis). This difference lost its significance in the logistic regression analysis (Fig 4).
We repeated the same analysis with a sample of patients selected from the same data sets on the basis of GA of <28 weeks. The AORs for IVH and severe IVH in association with male gender were 1.14 and 1.21, respectively. These figures were similar to the AORs for IVH and severe IVH in boys of <1500 g (1.15 and 1.18, respectively).
This study indicates that, with controlling for demographic and clinical confounders, male gender was associated with increased rates of IVH and severe IVH in VLBW infants. These findings were consistent across different BW categories. There was no gender difference in the incidence of PVL. Apart from a small retrospective study,14 this is the first study, to our knowledge, to examine such an association in a large national cohort. These results may be helpful in predicting outcomes during prenatal and postnatal bedside consultations with parents.
Several mechanisms have been proposed to explain differences between preterm boys and girls in response to brain insults. Apoptotic mechanisms controlling cell death have been shown to differ between boys and girls at early developmental stages in animal models.18,–,22 Preterm boys were found to be more susceptible than girls to brain injury secondary to altered catecholamine levels in their brains.23 Preterm girls were found to have lower cerebral blood flow.24 Because IVH is considered to be a hypoperfusion-reperfusion injury, high cerebral blood flow in preterm boys might be a possible explanation for the increased incidence of IVH. Cytokines such as interleukin 1 receptor antagonist and tumor necrosis factor α are associated with common perinatal complications. The formation of interleukin 1 receptor antagonist is dependent on both genotypic and nongenotypic factors, and there is a gender difference in interleukin 1 receptor antagonist gene polymorphism.25,26 Genotypic analysis of the tumor necrosis factor α promoter gene showed that premature boys with IVH had a higher incidence of the homozygous genotype for the tumor necrosis factor β2 allele.27 This difference could be a cause of the gender difference in newborns' responses to perinatal diseases.
Our study showed that more-mature preterm infants demonstrated more-prominent associations of male gender with IVH and severe IVH (Fig 3). This association possibly reflects improved female fetus neurovascular maturation and better regulatory mechanisms. This association may give more weight to hormonal influences on the maturity of the female brain, compared with the male brain. Estrogen has been shown to reduce brain injury in vivo and in vitro, and progesterone seems to protect against ischemic or traumatic injury in animal models.28,29 Sex hormones are produced at different concentrations in male and female fetuses and thus may have preferential protective effects.30,31 The effects of estrogen are mediated through estrogen receptor α. In preterm infants, estrogen receptor α polymorphism is associated with postnatal morbidity in boys but not girls.32
In our study, the incidence of PVL in preterm boys did not differ from that in girls. A possible explanation may be that the pathophysiolgic features of PVL are different from those of IVH. Whereas IVH is explained on the basis of reperfusion injury, PVL is primarily a hypoperfusion injury.1,2 Both white and gray matter masses were smaller in preterm infants, compared with term infants, as measured through quantitative analysis of MRI data at the age of 8 years. Although boys in general have larger volumes of white matter than do girls, the white matter in preterm boys was smaller, and boys were more vulnerable to the complications of prematurity.33 Another explanation is that the vulnerability to risk factors for PVL is continuous during the entire NICU course but the risk for IVH is mainly during the first 2 to 3 days of life. This reflects the greater impact of the NICU environment and management strategies, compared with genetic predisposition, on the occurrence of PVL.
This study demonstrated that male vulnerability extends beyond IVH. Preterm boys had more infants in the BW category of <1000 g and more deaths and cases of RDS, pulmonary hemorrhage, and sepsis (Table 2). The increased incidence of IVH in male infants could be secondary to their generalized immaturity, compared with female infants. Male gender was found to be associated with slower intrauterine pulmonary development, a significant predictor of increased morbidity and mortality rates.34,35 Our study showed that, even when we analyzed a sample of infants with GA of <28 weeks at birth, the male associations with IVH and severe IVH remained significant. Of note, the mortality rate was increased with IVH, compared with that without IVH (22.4% vs 21.9%). With controlling for all confounding factors, it is interesting to note that the AOR for death was decreased in the IVH group (OR: 0.78). This finding puts great emphasis on confounding variables (eg, male gender) that are associated with IVH, rather than IVH itself, as determining factors for death.
The NIS and KID data sets are collected through the H-CUP, sponsored by the federal Agency for Healthcare Research and Quality.17 Using the NIS and KID data set allowed us to obtain a large sample size (N = 104 847) that truly represented the entire nation. In this comprehensive data sample, we were able to identify many clinical and demographic variables that may contribute to postnatal outcomes, and we were able to control for those variables in the logistic regression models.
Similar to other large cohort studies, this study encountered certain limitations. There is no direct linkage between infants' data and some relevant maternal data in the data sets. Although many maternal diagnostic codes were documented under the infants' hospital records and were used as possible confounders, we were unable to control for variables such as prenatal maternal administration of steroids, magnesium sulfate, antibiotics, or any other medications. In addition, we could not control for postnatal catastrophic events such as prolonged resuscitation, incidents of difficult endotracheal intubation, or acute hypotensive episodes. The aforementioned variables either did not have ICD-9 diagnostic codes (eg, multiple intubation trials) or were underreported (eg, hypotensive episodes). Despite these limitations, controlling for known associated outcomes of the aforementioned “uncontrolled” factors, such as RDS, PPHN, or pulmonary hemorrhage, did not change our findings.
The overall incidence of IVH in our sample was 14.5% and that of PVL was 0.41%, compared with 24.6% to 27.4% and 1% to 3%, respectively, in the Vermont Oxford Network (1990–1999). Data from other national sources reported rates of IVH of ≤20%, whereas the incidence of PVL was ∼1% in most recent years.1,2 Data sets used for this study were collected from many hospitals of different care levels (primary to tertiary), with possibly different acuity, clinical practices, and diagnostic techniques (especially important for PVL, which may be missed in routine ultrasound examinations). In this study, we calculated proportions and ORs among those proportions; AORs remain constant in comparisons of proportions. Another limitation of these data was the fact that not all cases of IVH had a diagnostic code for severity; 6773 IVH cases (43.9% of the total) had a known grade for IVH.
ICD-9 codes for GA categories were introduced into the data late in 2002. With the incidences of unsure dates of pregnancy, poor or no prenatal care, or no ultrasound examination, exact dates are difficult to follow. In contrast, BW is a precise parameter and was included in the data for all patients through the analyzed years. We analyzed a sample of preterm infants with GA of <28 weeks at birth for the possible association of male gender with IVH. We obtained similar results and concluded that there was no significant statistical difference in using GA versus BW. In this study, BW was used as our inclusion criterion, to obtain larger sample size.
Our study showed significant increases in IVH and severe IVH in preterm male infants, compared with female infants. The link of IVH with male gender was more prominent in more-mature infants. Controlling for clinical and demographic variables revealed consistency of these findings. Further research is needed to determine the effects of these gender differences on long-term outcomes, as well as to reveal the pathophysiologic basis of these differences.
We thank Tahmina Ahmad from George Washington University for editing this manuscript.
- Accepted August 18, 2009.
- Address correspondence to Mohamed A. Mohamed, MD, MPH, George Washington University Medical Center, Newborn Services, 900 23rd St NW, Suite G-2092, Washington, DC 20037. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
- IVH =
- intraventricular hemorrhage •
- PVL =
- periventricular leukomalacia •
- GA =
- gestational age •
- BW =
- birth weight •
- VLBW =
- very low birth weight •
- RDS =
- respiratory distress syndrome •
- PPHN =
- persistent pulmonary hypertension of the newborn •
- PDA =
- patent ductus arteriosus •
- NIS =
- National Inpatient Sample •
- KID =
- Kids' Inpatient Database •
- ICD-9 =
- International Classification of Diseases, Ninth Revision •
- H-CUP =
- Healthcare Cost and Utilization Project •
- OR =
- odds ratio •
- AOR =
- adjusted odds ratio •
- CI =
- confidence interval
- Martin RJ,
- Fanaroff AA,
- Walsh MC
- de Vries LS
- MacDonald MG,
- Mullett MD,
- Seshia MMK
- Hill A
- Stevenson D,
- Verter J,
- Fanaroff A,
- et al
- Zeitlin J,
- Saurel-Cubizolles MJ,
- Mouzon J,
- et al
- Perelman RH,
- Palta M,
- Kirby R,
- Farrell PM
- Van Marter LJ,
- Leviton A,
- Kuban KC,
- Pagano M,
- Allred EN
- Hoffman EL,
- Bennett FC
- Vohr BR,
- Wright LL,
- Dusick AM,
- et al
- Brothwood M,
- Wolke D,
- Gamsu H,
- Benson J,
- Cooper D
- 17.↵Healthcare Cost and Utilization Project. Overview of H-CUP. Available at: www.hcup-us.ahrq.gov/overview.jsp. Accessed February 1, 2009
- Du L,
- Bayir H,
- Lai Y,
- et al
- Heep A,
- Schueller AC,
- Kattner E,
- et al
- Shaheen SO,
- Barker DJP
- Copyright © 2010 by the American Academy of Pediatrics