Abstract
OBJECTIVES. The objectives of this study were to describe the characteristics and morbidity of very low birth weight infants, to identify the medical intervention for these infants, and to evaluate the factors affecting the mortality of these infants among the participating hospitals.
METHODS. A large multicenter neonatal research network that included level III NICUs from throughout Japan was established. A standardized mortality rate was formulated by giving a ratio of the observed deaths and the predicted deaths based on a 100-g birth weight interval mortality. A regression model was used to predict the factors that affect neonatal mortality.
RESULTS. The network included 37 centers and 2145 infants weighing ≤1500 g, born or admitted to the centers in 2003. Gestational age and birth weight of studied infants were 28.6 ± 3.6 gestational weeks (mean ± SD) and 1025 ± 302 g, respectively. Overall, 11% of the infants died before being discharged from hospitals (range: 0%–21%). The standardized mortality rate varied among the facilities (range: 0%–30%). No association between the annual number of patients admitted and standardized mortality rate was found. Among all of the very low birth weight infants, 14% were outborn infants, 72% were delivered by cesarean sections, 27% had patent ductus arteriosus, 3% had gastrointestinal perforation, 8% had bacterial sepsis, and 13% had intraventricular hemorrhage. Medical interventions involved were: 41% antenatal corticosteroids, 54% surfactant therapy, 18% postnatal steroids for chronic lung disease, and 29% high-frequency oscillatory ventilation. We found variations in the medical interventions and the clinical outcomes among the centers.
CONCLUSIONS. The overall survival rate for very low birth weight infants among neonatal centers in Japan was ∼90%. However, differences in the morbidity and mortality were observed among these centers.
Patient outcomes and medical practices for very low birth weight (VLBW) infants may vary among hospitals for a wide range of clinical conditions.1 We established a neonatal research network database with a grant from the Ministry of Health, Labour, and Welfare, Japan. The objectives of the study were to describe the characteristics and morbidity of VLBW infants, to identify the medical interventions for these infants, and to evaluate the factors affecting the mortality of these infants among the participating hospitals.
METHODS
Patient Selection
We included infants who were born between January 1 and December 31 in 2003 with birth weight ≤1500 g in the participating neonatal centers or who were VLBW infants admitted to these facilities within 28 days of birth. We also included VLBW infants who were born alive but died in the delivery room. We collected data between December 2004 and March 2005 and recorded all information of the infants in the database. Forty-two level III perinatal centers were registered in the year 2004, and all are listed in “Acknowledgments.”
Definitions
We compiled a network database operation manual to define the patient characteristics. In the manual, the day of birth was defined as day 0. Neonatal mortality was defined as death within 28 days of birth and mortality before discharge as death occurring before discharge from a participating NICU. Gestational age (GA) was determined in the following order: obstetric examination with ultrasonography, obstetric history based on last menstrual period, and then postnatal physical examinations of neonates. Infants weighing <10th percentile of the normal birth weight curve at each GA were defined as light for GA.2 Maternal diabetes mellitus (DM) or gestational DM (GDM) and maternal hypertension were determined according to the diagnostic criteria.3 Premature rupture of membranes (PROM) was defined as rupture of membranes before an onset of labor. Clinical chorioamnionitis was diagnosed based on the clinical findings, such as maternal fever, leukocytosis, and local pain. Histologic chorioamnionitis was defined according to the criteria reported by Blanc.4 Antenatal steroid (ANS) usage was defined as the administration of any corticosteroids to accelerate fetal lung maturity. Surfactant therapy meant pulmonary surfactant (Surfacten) given during the acute phase of respiratory problems.
Respiratory distress syndrome was diagnosed based on the clinical and radiographic findings. Chronic lung disease (CLD) was defined when an infant received supplemental oxygen on the 28th day after birth, and 36-week CLD was defined when an infant received supplemental oxygen at the 36th week postmenstrual age. Postnatal steroid (PNS) usage meant any steroids given during the hospital stay for the prevention or treatment of CLD. Symptomatic patent ductus arteriosus (PDA) was diagnosed based on both the echocardiographic findings and clinical evidence of a volume overload because of a left-to-right shunt. Persistent pulmonary hypertension of the newborn (PPHN) was defined as right-to-left shunt at the foramen ovale and/or ductus arteriosus without any anatomic malformations as detected by cardiac echocardiography. Intraventricular hemorrhage (IVH) was reported according to the classification of Papile et al.5 Necrotizing enterocolitis (NEC) was defined according to the classification of Bell et al:6 stage II or greater. Gastrointestinal perforation was diagnosed if free air was detected in the abdominal cavity by radiograph examination regardless of cause. Sepsis meant culture-proven septicemia or bacteremia at any time during the stay in the NICU. A cystic periventricular leukomalacia (PVL) diagnosis was made by using either head ultrasound or cranial MRI scans, performed either at 2 weeks of age or later. Intrauterine infection was diagnosed if any inflammatory response was detected in the infants at birth. Retinopathy of prematurity (ROP) was diagnosed if the infants were treated with laser coagulation, cryocoagulation therapy, or both. Patients were classified into adrenal insufficiency of prematurity (AOP) when any steroids were administered during the hospital stay for the treatment of a late-onset circulatory collapse of premature infants because of an impaired adrenal function. Major congenital anomalies did not include external malformation.
Standardized Neonatal Mortality Rate
The neonatal mortality rate of VLBW infants was affected by the prematurity of the patients, as well as by the quality of hospital care. We, therefore, computed the standardized neonatal mortality rate (SMR) at each NICU from the observed deaths and predicted deaths. Although several basic and clinical variables are associated with neonatal mortality, birth weight is known as a highly significant predictor for mortality among VLBW infants. Thus, we used a logistic model that included only birth weight as variable to calculate SMR. The odds ratio for the predicted probability of death for the 100-g birth weight interval before being discharged was calculated using a univariate logistic-regression model with the birth weight as an independent variable. The expected mortality rate was then obtained by summing the predicted probabilities of death before discharge for all of the infants at each NICU. Finally, SMR was calculated by multiplying the statewide mortality rate of the database by a ratio of the observed mortality rate to the expected mortality rate at each center. The validity of the regression model used for calculating the odds ratio for death before discharge was tested by both the sensitivity and specificity for predicting mortality.
Univariate Analysis
All of the participating centers were divided into 3 different groups according to SMR. To evaluate the clinical factors and interventions related to neonatal mortality, several variables were tested for differences among the groups. Univariate analyses were performed using 1-way analysis of variance for continuous variables and the χ2 tests for categorical variables to test the differences among the groups. Variables that differed significantly among the groups were considered to be candidate variables for identifying any risk factors for death before being discharged with a multivariate analysis. We further assumed that factors that could explain the variation of SMR might constantly differ among the groups. Therefore, the variables that increased or decreased across the groups were also considered to be risk factors for predicting mortality before discharge, although their differences were not significant among the groups. The variables included in the univariate analysis were GA, birth weight, male gender, multiple birth, light-for-GA weight, outborn, 1- and 5-minute Apgar scores, major congenital anomalies, maternal hypertension, maternal DM/GDM, ANS usage, PROM, clinical chorioamnionitis, histologic chorioamnionitis, intrauterine infection, cesarean section, resuscitation with endotracheal intubation, and respiratory distress syndrome. These factors were closely related to the condition or severity of VLBW infants at birth, which could not be controlled in the NICU. Furthermore, the following variables were also analyzed based on a univariate analysis: pulmonary surfactant usage, HFO usage, PNS for CLD, symptomatic PDA, air leak syndrome, pulmonary hemorrhage, PPHN, oxygen at 36 weeks, IVH, sepsis, gastrointestinal perforation, the number of infants admitted, and collected age at the time of discharge among the survivors. These factors are common morbidities or treatments for VLBW infants, and they are likely to be closely correlated with the abilities or level of participating hospitals.
No risk factors and treatments known to be associated with increased neurodevelopmental morbidity rather than mortality were added to the analysis. Risks and treatments not included in the model were CLD at 28 days, PVL, and treatment for ROP.
Multivariate Analysis
Two-step logistic-regression models were developed consecutively to identify the risk factors for predicting mortality before discharge. In the first model, only perinatal and early neonatal factors were included to evaluate the risk factors that predicted neonatal mortality at birth. The second model included not only factors that had a significant correlation with death before discharge in the first model but also common complications and interventions in VLBW infants. Using the 2-step model, we could evaluate the prenatal and postnatal factors for neonatal mortality by avoiding interferences among factors, because some of the interventions during the neonatal period were related to prenatal risk factors. The risk factors used for the first models were, thus, closely related to the conditions of VLBW infants at birth rather than the quality of cares in the NICU. A second regression model to evaluate the effect of common complications in VLBW infants and hospital-level characteristics was then estimated by adding other risk factors. The stepwise backward elimination method was used for each regression model, whereas parameters at a significant level in the first model were compulsorily included in the second model.
All of the statistical analyses were performed using the SPSS 14.0J (SPSS, Inc, Chicago, IL). Significant difference was set at P = .05.
RESULTS
Characteristics of Participating Centers
Forty-two level III perinatal centers were registered in the neonatal research network database in 2004 (see “Acknowledgments” section). Five institutions were excluded from the final analysis because of incomplete data. The characteristics of the remaining 37 participating centers were as follows (mean ± SD): total number of beds for sick neonates was 35 ± 6.1; number of beds for NICU was 12.8 ± 2.1; number of doctors belonging to NICU was 5.7 ± 0.7; and number of nurses belonging to NICU was 43.9 ± 5.1. Psychologists were assigned to 18 (48.6%) facilities among all of the hospitals. Pediatric surgeons were available in 34 facilities (91.9%), pediatric cardiac surgeons in 11 (29.7%), and neurosurgeons in 5 (13.5%).
Basic Characteristics of Infants
A total of 2145 VLBW infants from the 37 centers were included. The mean birth weight was 1024 ± 302 g, and the mean GA was 28.9 ± 3.4 weeks (Table 1). These infants represented ∼30% of all VLBW infants born in Japan in 2003. Three centers received only outborn infants, because they were solely functioning as children's hospitals. As the result, 14% of the total infants were outborn infants. Multiple births composed 30% of total infants: 24% were twins and 6% were triplets or more. Major congenital anomalies were detected in 7% of the infants, and 53% of these infants were associated with chromosomal anomalies, whereas nearly 70% (23 of 36) of the infants with chromosomal anomalies had 18 trisomy.
Basic Characteristics of All Infants
Perinatal and Early Neonatal Characteristics
Table 2 lists perinatal and early neonatal information. Maternal hypertension occurred in 18% of the mothers, whereas only 1% of the mothers had DM or GDM. Thirty percent of mothers had PROM. Nineteen percent of all of the mothers had chorioamnionitis before delivery, whereas 24% of the 314 placentas examined histologically by pathologists actually had chorioamnionitis. Twenty-one percent of those placentas were classified as grade I, 35% of them were grade II, and 44% were grade III, respectively.
Perinatal and Early Neonatal Characteristics
ANS treatment was given to 41% of the mothers. The frequency of ANS use, however, varied markedly among the participating institutions (5%–100%). Seventy-two percent of the infants were delivered by cesarean section. Fifty-six percent of the infants were ventilated with an endotracheal tube at delivery, whereas 90% were resuscitated with oxygen.
Morbidity and Interventions
Table 3 shows the incidences of selected morbidities and interventions. Respiratory distress syndrome was diagnosed in 54% (range: 27%–85%) of the infants, and all of those infants were treated with surfactant replacement. Among those infants who were on a ventilator, high-frequency oscillatory ventilator was introduced in 29% of them (range: 0%–100%). CLD, defined as an oxygen requirement at 28 days after birth or 36 weeks of postmenstrual age, affected 33% (range: 4%–76%) and 28% (range: 0%–100%) of the infants, respectively. Eighteen percent (range: 0%–85%) of the infants were given postnatal corticosteroid to treat CLD, whereas 4% of the infants were discharged requiring oxygen supplementation. Other pulmonary morbidities, including pneumothorax and pulmonary hemorrhage, were observed in 3% and 5% of the infants, respectively.
Morbidities, Interventions, and Length of Stay
Cranial sonograms were obtained in 91% of the infants at least once before discharge or death. Thirteen percent (range: 0%–42%) of those infants had IVH, and severe hemorrhage (grade III or IV) was diagnosed in 7% of the infants. Eighty-six percent of the infants had cranial ultrasonography or MRI at ≥2 weeks of age, and among these infants PVL was noted in 4%. PDA was diagnosed in 27% (range: 7%–100%) of the infants. Among the infants with this diagnosis, indomethacin was administered to 92% and a surgical ligation was performed in 6%. The incidence of sepsis was 8%, whereas 80% of the infants received antibiotics during their stay at NICU. The incidence of NEC was 1%, but gastrointestinal perforation was observed in 3% of the infants. Treatment for ROP was performed in 17% (range: 3%–100%) of the infants. During the stay at NICU, corticosteroid administration was given to 4% of the infants who had acute circulatory collapse.
Neonatal Mortality, Mortality Before Discharge, and Length of Hospital Stay Among Survivors and Nonsurvivors
Of the 2154 VLBW infants, 89% survived until they were transferred to another institution or were discharged to their home. Eighty-nine percent of all of the survivors were discharged to their home, whereas 23% of the remaining infants were transferred back to their previous hospitals; 29% were transferred to affiliate hospitals, and 45% were transferred to pediatric wards before discharge to their home. The remaining 3% were transferred to local orphanages because of social and economic reasons. The average duration of hospitalization for survivors was 97 days. The average postmenstrual age at discharge was 42 weeks, body weight was 2591 ± 849 g, and head circumference was 34.1 ± 3.8 cm. Among the infants who died, the average length of stay was 29 days.
The birth weight-specific neonatal mortality and mortality before discharge for all of the infants at 100-g birth weight interval are shown in Fig 1. The GA-specific mortalities are shown in Fig 2.
Birth weight-specific neonatal mortality and mortality before discharge. Data expressed as mortality and 95% CIs for each 100-g birth weight interval with number of infants above bars.
and 
, neonatal mortality and mortality before discharge, respectively.
GA-specific neonatal mortality and mortality before discharge. Data expressed as mortality and 95% CIs for each GA group with number of infants above bars.
and 
, neonatal mortality and mortality before discharge, respectively.
The mortality rates before discharge at each institution ranged from 0% to 21%. The odds ratio of death before discharge for 100-g birth weight interval calculated by the regression model was 0.643 (95% confidence interval [CI]: 0.605–0.683; P < .0001). The sensitivity of the odds ratio, which calculated the predicted number of deaths divided by observed number of deaths, was 1.17. On the other hand, the specificity of the odds ratio, which calculated the predicted number of survivors divided by observed number of survivors, was 0.98. These data suggest that the overall fitness of the model was sufficiently accurate for predicting SMR among the infants.
The SMR at each centers also varied from 0% to 30%. The 2 sites that had a mortality rate of 0% had small sample sizes (21 and 23 infants per hospital, respectively). The variation in SMR among institutions was greater than we had expected. In addition, a significant variation in the length of stay among survivors and nonsurvivors among NICUs remained after it was adjusted for the patients' birth weights.
Univariate Analyses
To evaluate the variations in SMR among the participating centers, the perinatal and neonatal characteristics were compared among the different mortality groups. The participating hospitals were divided into 3 different groups according to SMR. The low mortality group consists of 11 hospitals with the lower one third of SMR. In the same way, the middle mortality group consists of 12 hospitals with the middle one third of SMR, and the high mortality group consists of 14 hospitals with the higher one third of SMR. The perinatal and neonatal characteristics that were analyzed by a univariate analysis are shown in Table 4. Among the factors strongly correlated with prenatal and perinatal conditions, birth weight, outborn, 5-minute Apgar score, major congenital anomalies, maternal hypertension, maternal DM/GDM, PROM, clinical chorioamnionitis, histologic chorioamnionitis, cesarean section, resuscitation with endotracheal intubation, and respiratory distress syndrome were significantly different among the groups. In addition to these risk factors, GA and male gender were also candidate variables for the regression model regarding constant decreases across of the groups. All of these factors were closely related to the condition and severity of VLBW infants at birth, which could not be modulated in NICU, and selected as independent variables for the first logistic-regression analysis. Furthermore, HFO usage, PNS for CLD, PDA, PPHN, gastrointestinal perforation, AOP, and the number of infants admitted were significantly different among the groups. These variables were closely related to hospital levels rather than to the clinical condition of VLBW infants at birth and thus were added to the second regression model in addition to the variables that had been selected in the first model.
Univariate Analysis of Perinatal and Neonatal Characteristics Among Low-, Middle-, and High-Mortality Hospitals
Logistic-Regression Analysis
The analysis of the first logistic-regression models showed an increasing birth weight (100-g increment) and 5-minute Apgar score to be associated with a decreased risk of death with an odds ratio of 0.67 (95% CI: 0.63–0.72; P < .001) and 0.7 (95% CI: 0.65–0.75; P < .001), respectively. On the other hand, major congenital anomalies were strongly associated with an increased risk of death with an odds ratio of 11.58 (95% CI: 6.82–19.68; P < .001). Other risk factors, such as GA, male gender, outborn, maternal hypertension, maternal DM/GDM, PROM, clinical chorioamnionitis, histologic chorioamnionitis, cesarean section, resuscitation with endotracheal intubation, and respiratory distress syndrome, were not significantly associated with the risk of mortality of VLBW infants using the stepwise backward elimination method.
The results of the second logistic-regression model, including common complications in VLBW infants and hospital-level characteristics, are shown in Table 5. The analysis was again performed using the stepwise backward elimination method, whereas 3 factors, birth weight, 5-minute Apgar score, and major congenital anomalies, which were found to be significantly associated with the risk of mortality before discharge in the first model, were compulsorily included. Finally, these factors remained significantly independent variables in the second model. Furthermore, PNS for CLD was associated with a decreased risk in mortality before discharge, whereas PPHN and gastrointestinal perforation were associated with an increased risk. On the other hand, HFO usage, PDA, AOP, and the number of infants admitted were not significantly associated with the risk of mortality before discharge.
Logistic-Regression Analysis to Evaluate Factors Affecting Early Perinatal and Neonatal Mortality Before Discharge
DISCUSSION
Our study is the first multicenter survey of VLBW infants in Japan using a network database. The data were collected from 37 centers across Japan, and they showed the general morbidity and mortality of VLBW infants born in 2003. However, our results should be interpreted with caution for the following 3 reasons. First, our sample could be biased, because we collected data of high-risk infants only from centers that were recognized as perinatal centers from local governments as well as the national government. Although the number of infants registered to the network represented ∼30% of the total VLBW infants born in 2003 in Japan, there are still many other centers with very large units for VLBW infants that were not government recognized perinatal centers. Second, because the network represented major NICUs, it excluded smaller units. However, we doubt whether the mortality and morbidity of either large or small units outside our network would be extraordinary high. Finally, our data did not include infants who died after they were transferred to another hospital. Because all of the registered hospitals were level III NICUs and most of the hospitals that received transferred infants from level III NICUs were local community hospitals, it was unlikely that the infants transferred to other hospitals would need further intensive care and die thereafter. Taken together, the present results should closely represent the VLBW infants born throughout Japan in 2003.
The large sample size was sufficient to analyze mortality by 100-g birth weight subgroups and by risk factors. Eighty-nine percent of the VLBW infants survived until discharge. Survival ranged from 20% of all infants who weighed <500 g at birth to 97% of infants weighing 1001 to 1500 g at birth. The mortality of VLBW infants before being discharged varied by GA, where 74% of the infants born at 22 weeks of GA died before discharge. The mortality rate by GA then decreased steadily through 29 weeks of GA, reaching the lowest point of 3%. GA-specific mortality rate tended to increase after 30 weeks of GA, because most of the VLBW infants in that GA range were delivered with severe intrauterine growth retardation. This was an aberration caused by the 1500-g birth weight cutoff. The diversity of GA-specific mortality rates among VLBW infants was consistent with the findings in previous reports.7–10
The database showed that the mortality rate for VLBW infants was relatively low. Although it is difficult to directly compare the outcome in this study with previous reports,7–11 the birth weight-specific mortality rate was lower in the present study. One of the main reasons for the better outcome is an obvious bias of the study period. To avoid this bias, we need to compare the latest outcome among different databases. If we compare our study with the National Institute of Child Health and Human Development (NICHD) data,10 where the most recent clinical data of VLBW infants were available, some differences were found regarding the background characteristics and average incidences of common morbidities in VLBW infants between the 2 studies. The NICHD data evaluated VLBW infants with birth weight between 501 and 1500 g, and the present study included all of the infants with birth weight ≤1500 g. This difference in the inclusion criteria could explain the higher rate of light for GA in the present study. Although the subjects of both studies were not identical, some of the differences recognized between the 2 studies in the background characteristics and the rates of morbidities may be very important for evaluating the outcome of VLBW infants and for trying to find clues for further improving the outcomes. Perinatal backgrounds were markedly different in the rate of ANS use (41% in this study vs 71% in the NICHD study). Although infants in our database received ANS less frequently, the incidences of respiratory distress syndrome (54% vs 50%), CLD at 36 weeks (28% vs 23%), and PDA (27% vs 30%) were closely compatible between the 2 studies. The rates of pulmonary surfactant and steroid for CLD uses were also almost identical (54% vs 52% and 18% vs 19%, respectively), whereas the usage of indomethacin and a surgical ligation for PDA were markedly less in our study (30% vs 75% and 5% vs 15%, respectively). Although it seems that these differences between the 2 studies come from differences in clinical practices in NICUs rather than in the effect of ANS usage, further evaluations are needed to clarify the role of ANS in these morbidities. On the other hand, the rates of the 1- and 5-minute Apgar scores ≤3 were lower in our study than those of the NICHD study (27.0% vs 30% and 6.5% vs 10%, respectively). This is a very critical difference, because the low 5-minute Apgar score is one of the risk factors for mortality in our study. The rate of endotracheal intubation at birth for resuscitation was almost the same between the studies (56% vs 60%). Better clinical conditions at birth may be a benefit of the higher incidence of cesarean section (72% vs 53%). However, surgical intervention during pregnancy puts an extra risk on mothers. The combined outcomes of both the mothers and infants must be evaluated. Furthermore, the incidences of IVH (any grade, 13% vs 30%; grade III and IV, 7% vs 11%), NEC (3% vs 7%), and sepsis (8% vs 24%) were lower in the present study than in the NICHD study. These differences might also be because of the more stable condition at birth or differences in the routine practices in NICUs. We, therefore, need comparative studies among different databases in the future.
The average length of stay for survivors was 97 days at 42 weeks of postmenstrual age, which was much longer than those reported previously.7–11 The growth rates of VLBW infants in the NICU are particularly important, because growth retardation after birth directly affect the adverse neurologic outcomes.12 The longer length of stay among the survivors in our study, however, may be related to the difference in the local supporting system for VLBW infants after discharge from the NICU, because postnatal growth of the infants was consistent with a previous report.13
Mortality varied even within registered centers. Differences in patient mix may be an important cause for the differences in mortality rates among the NICUs, so we adjusted the number of patients by 100-g birth weight categories. However, SMR adjusted by birth weight among the facilities still varied widely from 0% to 30%. Thus, the variation of mortality among the facilities could not be explained solely by the differences of birth weights for patients that they treated. There were many reports that showed that mortality rates of newborn infants varied significantly from hospital to hospital. The Vermont Oxford Neonatal Network and NICHD Neonatal Research Network have also reported variation in practices and outcomes among their participating NICUs.1,14 Common morbidities that have been reported15 as risk factors associated with neurologic abnormalities, such as CLD and ROP treatment, also widely varied among the groups (from 5% to 75% and 2% to 38%, respectively). It is not clear at the present whether the incidence variation of these morbidities would reflect the neurologic outcome variation after follow-up.
We further investigated morbidities and treatments in the NICUs to explain the variation in the mortality rates among the centers participating in the network. Even the incidence of respiratory distress syndrome, which is one of the representative and most common morbidities among VLBW infants, ranged from 27% to 85%. The variation in the incidence of respiratory distress syndrome was also reported in a different neonatal database.14 The incidence of the disease in VLBW infants among the hospitals should not fluctuate to that degree, although the rate of ANS use in this study varied from 5% to 100%. In the present study, all of the infants who had respiratory distress syndrome were eventually treated with exogenous surfactant. We believe that this is one example of how the attitude toward diagnosis and treatment of a certain morbidity could affect the outcomes of VLBW infants.
Although it is not clear why the morbidities and treatments varied among the hospitals, we assumed that perinatal and neonatal factors could directly contribute to the outcomes. Although many factors were associated with the mortality based on a univariate analysis, a positive association with mortality was found in only major congenital anomalies, PPHN, and gastrointestinal perforation based on a multivariate logistic-regression model. Negative associations were found in birth weight, 5-minute Apgar score, and PNS for CLD. However, we did not find the relationship between ANS therapy and either the frequency of respiratory distress syndrome or the mortality of VLBW infants in this study. Furthermore, the gender and outborn status also failed to show any significance, although these factors are known to be associated with the mortality.8 Variations in practice among NICU centers have also been reported,16 and it seems to be a key issue regarding the outcome variation. It is unclear whether these features are unique to Japan or just a result of year-by-year variation. Furthermore, whether or not these features are directly associated with better outcomes remains to be determined. Further investigations are needed to explain such variations by increasing the number of centers participating in the network. Finally, the standardization of the diagnosis and intervention in neonatal care practices seems to be the next issue for the further improvement of outcome in high-risk newborn infants.
CONCLUSIONS
We evaluated survival and major in-hospital morbidities of VLBW infants in Japan. Although overall survival rate of the infants was ∼90%, we found differences in the neonatal mortality rates for VLBW infants among the NICUs. In addition, specific morbidities and treatments of VLBW infants at an NICU were not associated with differences in mortality. This suggests that the effectiveness of care varies among units. A major goal of our network is to help the participating NICUs improve their overall care and effectiveness through a coordinated program and to conduct randomized trials, outcome research, and quality improvement projects.
Acknowledgments
This study was supported in part by a grant from the Ministry of Health, Labour and Welfare, Japan.
Institutions and representative physicians enrolled in the database for Neonatal Research Network, Japan include: Kushiro Red Cross Hospital: T. Nagashima; Iwate Medical University: S. Chida, S. Shimada; Sendai Red Cross Hospital: M. Yamada; Fukushima Medical University: H. Ariga; Dokkyo Medical University: H. Suzumura; Jichi Medical University: Y. Honma; Saitama Children's Medical Center: T. Ono; Saitama Medical University Saitama Medical Center: M. Tamura; Gunma Children's Medical Center: T. Koizumi; Tokyo Metropolitan Bokuto Hospital: T. Watanabe; Showa University: K. Itabashi; Tokyo Women's Medical University: I. Sakuma; Nihon University Itabashi Hospital: M. Minato, S. Hosono; Teikyo University: J. Hoshi; Toho University: N. Uga; Japan Red Cross Medical Center: Y. Kawakami; Aiiku Hospital: K. Kabe; Kanagawa Children's Medical Center: H. Itani; Yamanashi Prefectural Central Hospital: N. Mizobe; Nagaoka Red Cross Hospital: O. Nagata; Nagano Children's Hospital: T. Nakamura; Toyama Prefectural Central Hospital; K. Hatazaki; Shizuoka Children's Hospital: Y. Usukura; Seirei Hamamatsu General Hospital: S. Oki; Nagoya Red Cross First Hospital: C. Suzuki; National Mie Hospital: H. Yamamoto; Kyoto Red Cross First Hospital: N. Mitsufuji; Osaka Medical Center and Research Institute for Maternal and Child Health: J. Shiraishi, H. Kitajima; Osaka City General Hospital: H. Ichiba; Takatsuki Hospital: Y. Lee; Nara Medical University: Y. Takahashi; Hyogo Prefectural Kobe Children's Hospital: H. Nakao; Kurashiki Central Hospital: S. Watabe; Hiroshima Prefectural Hospital: R. Fukuhara; National Kagawa Children's Hospital: M. Furukawa; Ehime Prefectural Central Hospital: M. Kajiwara; St Mary's Hospital: T. Hashimoto; Kitakyushu City Municipal Medical Center: Y. Kukita; Kurume University: T. Matsuishi; Fukuoka University: H. Yukitake, S. Mori; Kumamoto City Hospital: Y. Kondo; and Okinawa Chubu Hospital: M. Kohama.
Footnotes
- Accepted May 23, 2006.
- Address correspondence to Satoshi Kusuda, MD, Maternal and Perinatal Center, Tokyo Women's Medical University, 8-1 Kawadacho, Shinjuku, Tokyo, 162-8666 Japan. E-mail: skusuda{at}boshi.twmu.ac.jp
The authors have indicated they have no financial relationships relevant to this article to disclose.
REFERENCES
- Copyright © 2006 by the American Academy of Pediatrics
In this issue
Jump to section
Related Articles
Cited By...
- Neonatal outcomes of very low birthweight infants born to mothers with hyperglycaemia in pregnancy: a retrospective cohort study in Japan
- Persistent pulmonary hypertension of the newborn in extremely preterm infants: a Japanese cohort study
- TADAFER II: Tadalafil treatment for fetal growth restriction - a study protocol for a multicenter randomised controlled phase II trial
- Variation in hospital mortality in an Australian neonatal intensive care unit network
- Changes in survival and neurodevelopmental outcomes of infants born at <25 weeks gestation: a retrospective observational study in tertiary centres in Japan
- International variations and trends in the treatment for retinopathy of prematurity
- Identification of practices and morbidities affecting the mortality of very low birth weight infants using a multilevel logistic analysis: clinical trial or standardisation?
- Individual and Center-Level Factors Affecting Mortality Among Extremely Low Birth Weight Infants
- Outcomes of Infants Born at 22 and 23 Weeks' Gestation
- Comparison of Mortality and Morbidity of Very Low Birth Weight Infants Between Canada and Japan
- Trends and centre-to-centre variability in survival rates of very preterm infants (<32 weeks) over a 10-year-period in Switzerland
- Trends and centre-to-centre variability in survival rates of very preterm infants (<32 weeks) over a 10-year-period in Switzerland
- Neonatal Care in a Japanese NICU: Notes Based on a Site Visit
- International Perspectives: Late-onset Circulatory Collapse in Very Low-birthweight Infants: A Japanese Perspective