Abstract
BACKGROUND:
Because greater percentages of women deliver at hospitals without high-level NICUs, there is little information on the effect of delivery hospital on the outcomes of premature infants in the past 2 decades, or how these effects differ across states with different perinatal regionalization systems.
METHODS:
A retrospective population-based cohort study was constructed of all hospital-based deliveries in Pennsylvania and California between 1995 and 2005 and Missouri between 1995 and 2003 with a gestational age between 23 and 37 weeks (N = 1 328 132). The effect of delivery at a high-level NICU on in-hospital death and 5 complications of premature birth was calculated by using an instrumental variables approach to control for measured and unmeasured differences between hospitals.
RESULTS:
Infants who were delivered at a high-level NICU had significantly fewer in-hospital deaths in Pennsylvania (7.8 fewer deaths/1000 deliveries, 95% confidence interval [CI] 4.1–11.5), California (2.7 fewer deaths/1000 deliveries, 95% CI 0.9–4.5), and Missouri (12.6 fewer deaths/1000 deliveries, 95% CI 2.6–22.6). Deliveries at high-level NICUs had similar rates of most complications, with the exception of lower bronchopulmonary dysplasia rates at Missouri high-level NICUs (9.5 fewer cases/1000 deliveries, 95% CI 0.7–18.4) and higher infection rates at high-level NICUs in Pennsylvania and California. The association between delivery hospital, in-hospital mortality, and complications differed across the 3 states.
CONCLUSIONS:
There is benefit to neonatal outcomes when high-risk infants are delivered at high-level NICUs that is larger than previously reported, although the effects differ between states, which may be attributable to different methods of regionalization.
KEY WORDS: perinatal regionalization, premature infant, delivery hospital
What’s Known on This Subject:
Data suggest that delivery at high-volume, high-technology hospitals reduces neonatal mortality. No study has examined other complications or compared the effects in multiple states by using a study design to control for unmeasured differences in case mix.
What This Study Adds:
The survival benefit to delivering at a high-level NICU between 1995 and 2005 is larger than previously reported and varies between states. The survival benefits affect both extremely and moderately preterm infants. Complication rates were similar between hospital types.
Regionalization of health care may improve care by directing patients to facilities with the appropriate capabilities to manage a given type of illness.1,2 The regional model of perinatal care began to weaken in many areas of the United States by the 1990s,3–6 whereas several other countries attempted to increase regionalization with varying degrees of success.7,8 Although studies using data from the early to mid-1990s suggest that delivery at a high-volume, high-technology hospital reduces neonatal mortality,9–12 this issue is worth further examination. Besides mortality, there is little information on the effect of delivery hospital on other outcomes, such as complication rates.13,14 Finally, there are no studies that compare the effects of delivery hospital on the outcomes of premature infants across states, even though state regionalization policies differ.15
The goals of this study, then, are to (1) obtain unbiased measurements of the impact on mortality of delivering at a high-volume, high-level NICU in comparison with other delivery hospitals in states with different systems of regionalization and different patient populations; and (2) examine common complications of premature birth. To best replicate a randomized controlled trial of this question, we will use an instrumental variables study design. This study design is new to the perinatal literature, but it has been used in other settings where policies direct patients to a particular site of care based on illness severity.16–21
Methods
Study Design
Data Population and Sources
We obtained birth certificates from all deliveries occurring in Pennsylvania and California between January 1, 1995, and June 30, 2005 and Missouri between January 1, 1995, and December 31, 2003. Each state’s department of health linked these birth certificates to death certificates by using name and date of birth and deidentified the records. More than 98% of these linked records were then matched to maternal and newborn hospital records by using previous methods.10,22,23 More than 80% of the unmatched birth certificate records were missing hospital, suggesting a birth at home or a birthing center. The unmatched records had gestational age and racial/ethnic distributions similar to the matched records. The institutional review boards of The Children’s Hospital of Philadelphia and the departments of health in California, Missouri, and Pennsylvania approved this study.
The primary cohort for this study included infants with a gestational age between 23 and 37 weeks and a birth weight between 400 and 8000 g. A secondary cohort of infants with a birth weight between 500 and 1500 g was used to compare results with previous work.12 Birth records were excluded if the birth weight was >5 SDs from the mean birth weight for the recorded gestational age in the cohort, because of potential recording errors in either variable.24 Initially, we identified 1 362 782 birth records; 34 650 met the exclusion criteria, leaving 1 328 132 births in the final cohort.
Definition of Study Outcomes
The primary outcome for this study was in-hospital mortality, made up from 2 metrics: neonatal deaths, defined as death during the initial birth hospitalization; and fetal deaths with either a gestational age ≥23 weeks or a birth weight ≥400 g that met a previous definition of a potentially preventable fetal death by care delivered at the hospital, because management around the time of delivery could convert some neonatal deaths into fetal deaths.10 We also examined 5 sets of complications of premature birth listed in Table 1.
TABLE 1.
ICD-9CM Codes Used to Identify Complications of Premature Birth
| Comorbid Condition | ICD-9-CM Code |
|---|---|
| BPD | 770.7 |
| Necrotizing enterocolitis (NEC) | 777.5 |
| Fungal sepsis | 112.x, 771.7 |
| Bacterial sepsis | 038.x, 995.90-995.94, 041.x, 790.7 |
| ROP | 362.21 |
| ROP surgery | 14.2x, 14.34, 14.4x, 14.5x |
| Laparotomy | 45.6x, 45.7x, 45.8, 45.9x, 46.0x, 46.1x, 46.2x, 46.3x, 46.4x, 46.5x, 46.8x, 54.1 |
| Any IVH | 431, 772.1, 772.1x |
For each code, “x” represents any number at the digit location. IVH, intraventricular hemorrhage.
Definition of Covariate Variables
We included specific covariate variables in our analysis based on their association with one or more study outcomes (Table 2). The covariates included gestational age; birth weight; maternal sociodemographic factors, such as race, age, education, and insurance status; maternal residential zip code sociodemographic information, such as the percentage of inhabitants living below the United States federal poverty line; maternal comorbid conditions; and 49 congenital anomalies grouped by affected organ system, each listed in Supplemental Technical Appendix 1.10
TABLE 2.
Demographics of Patients Delivering at High-level and Other Delivery Hospitals, Pennsylvania, California, and Missouri, 1995–2005
| Pennsylvania | California | Missouri | |||||||
|---|---|---|---|---|---|---|---|---|---|
| High-level NICU | Other Delivery Hospital | Δ/SDa | High-level NICU | Other Delivery Hospital | Δ/SDa | High-level NICU | Other Delivery Hospital | Δ/SDa | |
| Differential travel time, min | 6.97 | 21.81 | −0.84 | 3.41 | 14.55 | −0.59 | 15.04 | 40.15 | −0.60 |
| Birth weight, g | 2474 | 2725 | −0.34 | 2527 | 2804 | −0.33 | 2657 | 2873 | −0.29 |
| Gestational age, wk | 34.7 | 35.7 | −0.39 | 35.3 | 35.6 | −0.17 | 34.8 | 35.4 | −0.24 |
| Race | |||||||||
| White | 64.50% | 77.90% | −0.29 | 63.52% | 65.24% | −0.04 | 77.88% | 75.61% | 0.05 |
| Black | 22.20% | 9.30% | 0.35 | 9.29% | 5.84% | 0.13 | 18.61% | 21.88% | −0.08 |
| Asian | 1.30% | 1.10% | 0.02 | 9.70% | 9.00% | 0.02 | 2.16% | 1.68% | 0.04 |
| Other | 3.00% | 3.40% | −0.02 | 15.74% | 18.40% | −0.07 | 0.72% | 0.68% | 0.00 |
| Insurance status | |||||||||
| FFS | 19.50% | 22.80% | −0.08 | 2.94% | 5.26% | −0.12 | 27.96% | 27.98% | 0.00 |
| HMO | 37.80% | 34.60% | 0.07 | 49.08% | 39.01% | 0.20 | 28.86% | 21.13% | 0.18 |
| Public | 31.80% | 29.70% | 0.05 | 44.20% | 50.83% | −0.13 | 37.17% | 41.63% | −0.09 |
| Other | 9.40% | 10.50% | −0.04 | 0.87% | 1.13% | −0.03 | 2.71% | 6.48% | −0.17 |
| Uninsured | 1.20% | 1.70% | −0.04 | 2.89% | 3.72% | −0.05 | 3.04% | 2.60% | 0.03 |
| Singleton birth | 80.60% | 86.50% | −0.16 | 88.23% | 91.94% | −0.12 | 85.70% | 90.98% | −0.17 |
| SGA | 16.90% | 15.20% | 0.05 | 11.39% | 8.69% | 0.09 | 14.14% | 10.92% | 0.10 |
| Maternal comorbid conditions and complications of pregnancy | |||||||||
| Comorbid conditions | |||||||||
| Chronic HTN | 2.00% | 1.17% | 0.07 | 1.17% | 0.75% | 0.04 | 1.81% | 1.22% | 0.05 |
| Gestational diabetes | 5.49% | 4.85% | 0.03 | 6.18% | 4.74% | 0.06 | 5.08% | 4.05% | 0.05 |
| Diabetes mellitus | 2.13% | 1.40% | 0.05 | 1.50% | 0.78% | 0.07 | 1.72% | 1.12% | 0.05 |
| Renal disease | 0.33% | 0.24% | 0.02 | 0.18% | 0.14% | 0.01 | 0.32% | 0.20% | 0.03 |
| Congenital heart disease | 0.16% | 0.05% | 0.03 | 0.06% | 0.03% | 0.01 | 0.08% | 0.05% | 0.01 |
| Complications of pregnancy | |||||||||
| Preterm labor | 48.65% | 39.79% | 0.18 | 30.34% | 21.29% | 0.21 | 37.58% | 26.50% | 0.24 |
| PIH | 12.14% | 8.25% | 0.13 | 7.83% | 5.58% | 0.08 | 10.16% | 8.33% | 0.06 |
| PROM | 20.85% | 14.86% | 0.16 | 11.14% | 8.98% | 0.06 | 16.45% | 10.45% | 0.18 |
| Oligohydramnios | 4.61% | 3.02% | 0.08 | 3.87% | 2.26% | 0.09 | 5.69% | 3.43% | 0.11 |
| Disorders of placentation | 6.57% | 4.57% | 0.09 | 4.57% | 3.42% | 0.05 | 5.69% | 3.91% | 0.09 |
FFS, fee for service; HMO, health maintenance organization; SGA, small for gestational age; HTN, hypertension; PIH, pregnancy-induced hypertension; PROM, premature rupture of membranes.
Δ/SD is the standardized difference between the high-level NICU and other delivery hospital groups for a specific variable, defined as (difference in means between 2 groups of patients) ÷ (SD of entire cohort). A value <0.20 is considered adequate balance between groups.
Hospital Definitions
Based on previous work,9,10 a specialty hospital was defined as a level III facility that delivered a minimum of 50 very low birth weight infants, on average, per year. All levels of care were obtained from the American Academy of Pediatrics perinatal survey25 and validated by using procedure codes from each hospital. The outcomes of infants were assigned to their delivery hospital, regardless of future transfers of care to other hospitals.10,11 This method assesses the primary impact of perinatal regionalization, which is the antepartum transfer of the mother, and assigns credit to hospitals that appropriately transfer patients to higher-level hospitals.
Organization of Perinatal Systems
Each state has different perinatal policies. Pennsylvania and Missouri have no legislatively defined perinatal policies, including no certificate of need requirement to open a neonatal intensive care or obstetric unit26 and no formal identification of a regional perinatal transport center. In Pennsylvania, 77% of the delivery hospitals have a level I NICU, and 21.2% of the NICUs have a level III designation. In Missouri, 86% of the delivery hospitals have a level I NICU, with 11.6% of the NICUs designated as a level III center. California has a legally codified regional perinatal system, with 23 designated regional perinatal centers, although there is a trend in California toward shifting the delivery of premature infants to lower-volume NICUs.10 Only 60.4% of the delivery hospitals have a level I NICU, with a higher percentage of hospitals with a level III NICU (32.8%) compared with the other states.
Instrument and Study Design
Previous observational studies of perinatal regionalization adjusted for differences in observed case mix variables between high-level NICUs and other delivery hospitals, usually through regression analysis. However, these methods cannot adjust for unmeasured or unrecorded factors, such as the severity of a comorbid condition or laboratory results. This results in 2 potential problems. First, sicker patients tend to deliver at high-level NICUs at higher rates than their less sick counterparts. Second, low-level hospitals may appropriately transfer high-risk mothers to more capable hospitals, and thus only deliver a woman prematurely if there is an extreme emergency. In both cases, without detailed clinical data, it may be impossible to adequately measure these differences.
To address both of these potential issues, our study uses a matched-pair instrumental variables design, referred to as “near-far matching.”27 An instrument is a variable that encourages patients to deliver at a particular hospital, in essentially a randomized fashion. A strong and valid instrument varies where a mother delivers, while controlling for both measured and unmeasured differences in case mix between types of hospitals, similar to a randomized study. Our instrument is the difference in travel times from the mother’s residential zip code to the nearest high-level NICU and the mother’s residence to the nearest other delivery hospital. We calculated travel times by using ArcView software (ESRI, Inc) as in previous work.28 Women with lower differential travel times lived in residential zip codes that were closer to a high-level NICU. Differential travel time satisfies the 3 characteristics of an instrument29: (i) Association with treatment: previous studies suggest that women tend to deliver at hospitals near their residential zip code30,31; (ii) Independence from unmeasured confounding: women do not expect to have a premature delivery, and, conditional on measured socioeconomic variables, women do not choose where to live based on distance to a high-level NICU; (iii) No direct effect: the marginal travel time to either facility should not directly affect outcomes.30,31
To ensure that patients with higher and lower values of the instrument are comparable, we matched patients on 59 measured covariates while maximizing the difference in the instrument. This design parallels a matched-pair randomized controlled trial of patients encouraged to deliver at a high-level NICU versus patients not encouraged to deliver at a high-level NICU. By including both an instrumental variables approach and this matched-pairs design, we improved the equality of the 2 study groups, which improved the accuracy of the results (Supplemental Technical Appendix 2).27
Data Analysis
We first assessed the strength of the instrument. A strong instrument would find that women living closer to high-level NICUs would deliver at high-level NICUs at higher rates than women living further away. We then assessed the validity of the instrument and the matched pairs by calculating the standardized difference of each measured covariate. A valid instrument should distribute measured covariates equally across both the quartiles of the instrument and the matched pairs. A value <0.20 for the standardized difference is considered adequate balance.32,33 For the instrument, this statistic equals (largest pairwise difference in means across quartiles of the instrument) ÷ (SD of entire prematch cohort). For the matched pairs, this statistic equals (difference in means between matched patients) ÷ (SD of entire prematch cohort).
Three analyses will be presented. First, we present a naïve analysis by using unadjusted differences in each of the 9 outcome measures between patients delivering at a high-level NICU and other delivery hospitals. Second, we present the appropriate analysis that controls for measured and unmeasured differences by using a matched-pairs instrumental variables analysis. Finally, as a secondary analysis, we present results after controlling for measured differences in case mix with a matched-paired propensity score analysis, to allow the comparison of our results with previous work. Risk differences and risk ratios are presented for each analysis. Confidence intervals for risk differences were calculated by standard inversion of a pivot-based test of the null, at an α-error rate of .05.30 Confidence intervals for risk ratios were calculated by using bootstrap methods. All data are presented separately by state and for the 2 infant cohorts.
Results
Women who delivered at a high-level NICU were more likely to have either a preexisting comorbid condition, such as diabetes mellitus, or a complication of pregnancy, such as preterm labor (Table 2). Infants delivered at high-level NICUs had a younger gestational age.
Strength and Validity of Instrument
The instrument was strong. In Pennsylvania, 79.8% of the pregnancies in the first quartile of the instrument delivered at a high-level NICU, compared with 23.9% in the fourth quartile (Supplemental Results Appendix 1). Similar instrument strengths were seen in California (79.6% vs 38.3%, respectively) and Missouri (55.7% vs 10.1%, respectively). In all 3 states, women in the middle 2 quartiles delivered at high-level NICUs at rates between the 2 extremes. The analysis also balanced all measured covariates between those patients encouraged to deliver at a high-level NICU versus patients not encouraged to deliver at a high-level NICU (Table 3). There were differences between states in the prevalence of complications of pregnancy, such as preterm labor, and in the racial/ethnic distributions of the mothers.
TABLE 3.
Improved Balance of Measured Covariates Between High-level NICUs and Other Delivery Hospitals After Use of Instrument and Matching, Pennsylvania, California, and Missouri, 1995–2005
| Pennsylvania | California | Missouri | |||||||
|---|---|---|---|---|---|---|---|---|---|
| High-level NICU | Other Delivery Hospital | Δ/SDa | High-level NICU | Other Delivery Hospital | Δ/SDa | High-level NICU | Other Delivery Hospital | Δ/SDa | |
| Birth weight, g | 2598 | 2597 | 0.00 | 2849 | 2849 | 0.00 | 2818 | 2817 | 0.00 |
| Gestational age, wk | 35.2 | 35.2 | 0.00 | 35.2 | 35.2 | 0.00 | 35.2 | 35.2 | 0.00 |
| Race | |||||||||
| White | 85.0% | 85.9% | −0.02 | 68.4% | 71.9% | −0.07 | 91.6% | 91.7% | 0.00 |
| Black | 5.1% | 4.7% | 0.01 | 5.9% | 5.8% | 0.01 | 7.3% | 7.3% | 0.00 |
| Asian | 1.1% | 0.4% | 0.06 | 8.0% | 6.8% | 0.04 | 0.6% | 0.5% | 0.01 |
| Other | 2.3% | 1.3% | 0.05 | 16.5% | 13.9% | 0.06 | 0.4% | 0.4% | 0.00 |
| Insurance status | |||||||||
| FFS | 23.9% | 25.1% | −0.03 | 3.3% | 5.4% | −0.09 | 27.8% | 25.6% | 0.05 |
| HMO | 37.0% | 30.9% | 0.13 | 45.7% | 44.2% | 0.03 | 21.1% | 19.7% | 0.03 |
| Public | 28.7% | 33.4% | −0.10 | 47.2% | 45.6% | 0.03 | 45.5% | 49.1% | −0.07 |
| Other | 9.1% | 9.0% | 0.00 | 0.9% | 1.5% | −0.06 | 3.3% | 3.6% | −0.01 |
| Uninsured | 1.0% | 1.2% | −0.01 | 3.0% | 3.3% | −0.02 | 2.3% | 2.0% | 0.01 |
| Singleton birth | 84.4% | 83.9% | 0.01 | 89.7% | 88.9% | 0.03 | 90.6% | 90.1% | 0.02 |
| SGA | 16.0% | 16.4% | −0.01 | 10.7% | 10.9% | −0.01 | 11.9% | 11.7% | 0.01 |
| Maternal comorbid conditions and complications of pregnancy | |||||||||
| Comorbid conditions | |||||||||
| Chronic HTN | 1.1% | 1.2% | 0.00 | 0.9% | 1.0% | −0.01 | 1.0% | 1.0% | 0.00 |
| Gestational diabetes | 4.7% | 4.7% | 0.00 | 5.2% | 5.3% | 0.00 | 3.8% | 3.4% | 0.02 |
| Diabetes mellitus | 1.4% | 1.8% | −0.03 | 1.0% | 1.1% | 0.00 | 1.0% | 1.1% | −0.01 |
| Renal disease | 0.2% | 0.3% | −0.01 | 0.1% | 0.2% | −0.01 | 0.2% | 0.2% | −0.01 |
| Congenital heart disease | 0.1% | 0.1% | 0.00 | 0.0% | 0.1% | −0.01 | 0.1% | 0.0% | 0.01 |
| Complications of pregnancy | |||||||||
| Preterm labor | 45.2% | 45.2% | 0.00 | 28.7% | 28.4% | 0.01 | 30.0% | 29.9% | 0.00 |
| PIH | 9.7% | 10.4% | −0.03 | 6.6% | 7.6% | −0.04 | 8.0% | 8.3% | −0.01 |
| PROM | 18.3% | 17.7% | 0.02 | 10.2% | 11.4% | −0.04 | 11.5% | 11.6% | 0.00 |
| Oligohydramnios | 3.3% | 3.1% | 0.01 | 3.0% | 2.9% | 0.00 | 3.8% | 3.5% | 0.01 |
| Disorders of placentation | 4.2% | 5.0% | −0.03 | 3.7% | 4.2% | −0.03 | 3.2% | 3.7% | −0.02 |
FFS, fee for service; HMO, health maintenance organization; SGA, small for gestational age; HTN, hypertension; PIH, pregnancy-induced hypertension; PROM, premature rupture of membranes.
Δ/SD is the standardized difference between the high-level NICU and other delivery hospital groups for a specific variable, defined as (difference in means between 2 groups of patients) ÷ (SD of entire cohort). A value <0.20 is considered adequate balance between groups.
Primary Cohort: 23 to 37 Weeks’ Gestation
Association of Delivery Hospital and Mortality
In the unadjusted analysis, delivering at a high-level NICU was associated with higher mortality rates in all 3 states (Table 4). After adjusting for both measured and unmeasured case mix differences between hospital types, delivering at a high-level NICU was associated with lower in-hospital mortality rates (Table 5). The risk differences ranged from 2.7 fewer deaths/1000 deliveries in California (95% confidence interval [CI] 0.9–4.5) to 12.6 fewer deaths/1000 deliveries in Missouri (95% CI 2.6–22.6). The risk ratios for in-hospital mortality at high-level NICUs ranged from 0.35 in Pennsylvania to 0.82 in California. Pennsylvania and Missouri showed a 2-fold reduction in neonatal mortality rates with delivery at a high-level NICU, whereas California showed a reduction in the rate of preventable fetal deaths.
TABLE 4.
Unadjusted Rates of Mortality and Complications at High-level NICUs and Other Delivery Hospitals, Pennsylvania, California, and Missouri 1995–2005
| Pennsylvania | California | Missouri | ||||
|---|---|---|---|---|---|---|
| RDa | RRb | RDa | RRb | RDa | RRb | |
| In-hospital death | 11.4 | 1.99 | 6.9 | 1.50 | 7.3 | 1.43 |
| Neonatal death | 9.5 | 1.93 | 7.3 | 1.77 | 8.0 | 1.59 |
| Preventable fetal death | 1.9 | 2.40 | −0.4 | 0.92 | −0.6 | 0.82 |
| BPD | 18.6 | 3.83 | 8.5 | 2.55 | 8.2 | 1.68 |
| NEC | 6.3 | 2.98 | 4.1 | 2.82 | 6.1 | 3.10 |
| Fungal sepsis | 7.7 | 2.97 | 3.9 | 2.22 | 7.6 | 2.04 |
| Bacterial sepsis | 24.8 | 1.94 | 18.7 | 1.85 | 47.7 | 2.92 |
| ROP | 7.7 | 3.28 | 14.4 | 3.96 | 22.3 | 3.16 |
| Surgery for ROP | 1.2 | 3.79 | 2.4 | 4.78 | 2.4 | 2.27 |
| Laparotomy | 1.8 | 1.97 | 2.2 | 2.03 | 4.0 | 2.44 |
| Any IVH | 19.2 | 3.47 | 10.9 | 3.34 | 13.6 | 2.18 |
RD, risk difference.
RD between groups. A positive RD indicates a higher rate at high-level NICUs compared with other delivery hospitals. A negative RD indicates a lower rate at high-level NICUs compared with other delivery hospitals.
A RR >1 indicates a higher rate at high-level NICUs compared with other delivery hospitals. A RR <1 indicates a lower rate at high-level NICUs compared with other delivery hospitals.
TABLE 5.
Adjusted Difference in Mortality and Complications of Prematurity for Infants Between 23 and 37 Weeks’ Gestation Delivering at a High-level NICU Compared With Other Delivery Hospitals, Pennsylvania, California, and Missouri 1995–2005
| Outcome | Statistic | Pennsylvania | California | Missouri |
|---|---|---|---|---|
| Mortality measures | ||||
| All hospital death | RD per 1000 deliveriesa | −7.8 (−11.5 to −4.1)b | −2.7 (−4.5 to −0.9)b | −12.6 (−22.6 to −2.6)b |
| RRc | 0.35 (0.09–0.61)b | 0.82 (0.70–0.94)b | 0.50 (0.26–0.82)b | |
| Neonatal death | RD per 1000 deliveries | −7.2 (−10.7 to −3.7)b | −0.5 (−2.0 to 1.0) | −8.4 (−17.5 to 0.7) |
| RR | 0.27 (0–0.59)b | 0.94 (0.79–1.07) | 0.56 (0.27–1.07) | |
| Preventable fetal death | RD per 1000 deliveries | −0.6 (−2.0 to 0.8) | −2.2 (−3.1 to −1.2)b | −4.2 (−9.1 to 0.7) |
| RR | 0.72 (0.22–1.46) | 0.60 (0.48–0.74)b | 0.32 (0–1.30) | |
| Complication measures | ||||
| BPD | RD per 1000 deliveries | 0 (−3.5 to 3.6) | 1.0 (−0.3 to 2.4) | −9.5 (−18.4 to −0.7)b |
| RR | 1.02 (0–2.53) | 1.21 (0.96–1.53) | 0.05 (0–1.00)b | |
| NEC | RD per 1000 deliveries | 1.6 (−0.9 to 4.1) | 1.7 (0.7–2.6)b | −4.5 (−9.8 to 0.7) |
| RR | —d | 1.98 (1.46–3.04)b | 0.28 (0–1.20) | |
| Fungal sepsis | RD per 1000 deliveries | 4.9 (2.3–7.6)b | 0.7 (−0.4 to 1.8) | 3.2 (−4.3 to 10.7) |
| RR | 3.67 (1.88–11.6)b | 1.28 (0.95–1.87) | 1.67 (0–17.7) | |
| Bacterial sepsis | RD per 1000 deliveries | 10.1 (4.6–15.6)b | 15.9 (13.4–18.3)b | 10.6 (−4.4 to 25.7) |
| RR | 2.37 (1.50–4.51)b | 1.92 (1.69–2.05)b | 1.29 (0.92–2.02) | |
| ROP | RD per 1000 deliveries | −0.7 (−3.2 to 1.9) | 3.0 (1.6–4.4)b | 3.5 (−5.6 to 12.7) |
| RR | 0.38 (0–6.34) | 2.52 (1.52–3.33)b | 1.31 (0.60–3.52) | |
| Surgery for ROP | RD per 1000 deliveries | −1.1 (−2.3 to 0)b | 1.6 (0.9–2.3) | 1.3 (−2.8 to 5.3) |
| RR | —d | —d | —d | |
| Laparotomy | RD per 1000 deliveries | −0.6 (−2.3 to 1) | −1.3 (−2.1 to −0.4)b | 0 (−4.6 to 4.6) |
| RR | —d | 0.16 (0–0.70)b | 1.00 (0–11.5) | |
| Any IVH | RD per 1000 deliveries | −0.8 (−4.6 to 3.1) | 1.1 (−0.3 to 2.6) | −4.9 (−14.0 to 4.3) |
| RR | —d | —d | —d |
All values in parentheses indicate 95% confidence intervals for a given statistic.
RD between groups. A positive RD indicates a higher rate at high-level NICUs compared with other delivery hospitals. A negative RD indicates a lower rate at high-level NICUs compared with other delivery hospitals.
All results statistically significant at a P < .05 level.
A RR >1 indicates a higher rate at high-level NICUs compared with other delivery hospitals. A RR <1 indicates a lower rate at high-level NICUs compared with other delivery hospitals.
RR calculations were unstable because of the limited number of events occurring at other delivery hospitals or at high-level NICUs.
Association of Delivery Hospital and Neonatal Complications
In unadjusted analyses, there were higher rates of all studied complications at high-level NICUs regardless of state. After accounting for case mix differences, few of these differences remained (Table 5). Delivering at a high-level NICU in Missouri was associated with lower rates of bronchopulmonary dysplasia (9.5 fewer cases/1000 deliveries, 95% CI 0.7–18.4), whereas Pennsylvania and California showed smaller, statistically nonsignificant changes. Rates of other complications were similar between the high-level NICU and other delivery hospital group, with the exception of infection rates, where the risk difference decreased from 5 to 45 extra infections at high-level NICUs/1000 deliveries in unadjusted analyses to 0 to 14 cases/1000 deliveries in adjusted analysis. The risk difference and risk ratios for most complications showed variation between states.
Secondary Cohort: 500 to 1500 g Birth Weight
Similar results were found when we analyzed the 500- to 1500-g cohort separately. Delivering at a high-level NICU was associated with lower mortality rates in all 3 states (Table 6). Complication rates were similar, in general, between the 2 types of NICUs, except that, as with the 23- to 37-week cohort, rates of bronchopulmonary dysplasia (BPD) were lower in Missouri high-level NICU hospitals, and rates of bacterial sepsis were higher in Pennsylvania high-level NICUs. The relative risk for the improvement in mortality in the 500- to 1500-g cohort was somewhat smaller for each state in comparison with the 23- to 37-week cohort.
TABLE 6.
Adjusted Difference in mortality and complications of prematurity for infants between 500 and 1500 g birth weight delivering at a high-level NICU compared with other delivery hospitals, Pennsylvania, California, and Missouri 1995–2005
| Outcome | Statistic | Pennsylvania | California | Missouri |
|---|---|---|---|---|
| Mortality measures | ||||
| All hospital death | RD per 100 deliveriesa | −5.7 (−11.0 to −0.4)b | −5.3 (−8.5 to −2.0)b | −0.4 (−10.7 to 9.8) |
| RRc | 0.68 (0.44–0.96)b | 0.77 (0.66–0.91)b | 0.98 (0.60–1.76) | |
| Neonatal death | RD per 100 deliveries | −5.9 (−11.0 to −0.8)b | −2.6 (−5.6 to 0.4) | 0.5 (−9.2 to 10.1) |
| RR | 0.63 (0.40–0.94)b | 0.85 (0.71–1.04) | 1.03 (0.61–2.06) | |
| Preventable fetal death | RD per 100 deliveries | 0.2 (−1.7 to 2.1) | −2.7 (−4.4 to −0.9)b | −0.9 (−5.7 to 3.9) |
| RR | 1.13 (0.27–5.40) | 0.56 (0.38–0.78)b | 0.75 (0–8.0) | |
| Complication measures | ||||
| BPD | RD per 100 deliveries | 4.0 (−2.6 to 10.5) | 3.5 (0.2–6.8) | −21.9 (−32.9 to −10.9)b |
| RR | 1.22 (0.88–1.70) | 1.33 (1.03–1.81) | 0.29 (0.12–0.52)b | |
| NEC | RD per 100 deliveries | 3.9 (0.1–7.6)b | 0.7 (−1.4 to 2.7) | −2.1 (−8.0 to 3.8) |
| RR | —d | 1.12 (0.79–1.67) | 0.78 (0.32–1.67) | |
| Fungal sepsis | RD per 100 deliveries | 0.7 (−2.1 to 3.5) | −2.1 (−4.1 to 0)b | −4.1 (−11.3 to 3.0) |
| RR | 1.47 (0–27.1) | 0.52 (0.22–0.96)b | 0.51 (0.07–2.06) | |
| Bacterial sepsis | RD per 100 deliveries | 13.6 (7.1–20.9)b | −1.7 (−5.5 to 2.1) | −8.1 (−20.1 to 3.8) |
| RR | 2.64 (1.65–6.03)b | 0.93 (0.79–1.11) | 0.79 (0.56–1.13) | |
| ROP | RD per 100 deliveries | −1.4 (−5.9 to 3.1) | 12.0 (8.4–15.5)b | −6.2 (−17.8 to 5.4) |
| RR | 0.68 (0–3.67) | 2.46 (1.82–3.67)b | 0.84 (0.60–1.19) | |
| Surgery for ROP | RD per 100 deliveries | 1.1 (−1.1 to 3.3) | 3.5 (1.8–5.1)b | −3.9 (−9.4 to 1.7) |
| RR | —d | —d | 0.50 (0.08–1.47) | |
| Laparotomy | RD per 100 deliveries | 0 (−2.2 to 2.2) | −1.5 (−3 to 0) | −1.6 (−6. to 2.7) |
| RR | 1 (0–8.0) | 0.65 (0.39–1.06) | 0.63 (0–2.75) | |
| Any IVH | RD per 100 deliveries | 12.5 (6.5–18.5) | 3.0 (0–6.1) | −7.8 (−18.3 to 2.5) |
| RR | —d | —d | —d |
All values in parentheses indicate 95% confidence intervals for a given statistic.
A positive RD indicates a higher rate at high-level NICUs compared with other delivery hospitals. A negative RD indicates a lower rate at high-level NICUs compared with other delivery hospitals.
All results statistically significant at a P < .05 level.
A RR >1 indicates a higher rate at high-level NICUs compared with other delivery hospitals. A RR <1 indicates a lower rate at high-level NICUs compared with other delivery hospitals.
RR calculations were unstable because of the limited number of events occurring at other delivery hospitals or at high-level NICUs.
Secondary Analysis
In the propensity score analysis (Supplemental Results Appendix 2), which accounts for measured differences in case mix between the hospitals, neonatal mortality rates at both types of hospitals were statistically similar in Missouri (relative risk [RR] 1.01, 95% CI 0.92–1.01), Pennsylvania (RR 0.95, 95% CI 0.85–1.05), and California (RR 0.96, 95% CI 0.93–1.01). As with the unadjusted analyses, there were higher rates of all studied complications at high-level NICUs regardless of state in the propensity score analysis. For both cohorts, the results showed no difference over the 11 years of the study.
Discussion
Determining the true impact of a policy intervention such as perinatal regionalization is critical to accurately weighing the benefits and costs of the intervention. In perinatal regionalization, specialty hospitals usually treat sicker patients in ways that may not be measurable.4 Also, perinatal policies may vary between geographic areas.15 Regardless of geographic region, our study suggests that there is a continued survival benefit to delivering premature infants at high-level NICUs that is much larger than the effect shown in previous studies. These effects differ across the 3 states. The largest benefit in mortality was found in the cohort that included both very low birth weight infants and moderately preterm infants, suggesting that choice of delivery hospital may influence the outcomes of all preterm infants, not just the extremely premature infant.
Studies use an instrumental variables approach when patients with certain characteristics are more likely to receive a given treatment.16–21 With this approach we found statistically significant results that were much larger than shown in previous studies.4–6,9–11,34–37 This difference may arise because, as more infants are delivered at lower-volume, lower-level NICUs, gestational age and maternal indications for premature delivery cannot adequately control for the differences in case mix between types of NICUs. Increased access to clinical information, such as data available in electronic health records, would be 1 solution to obtain more accurate estimates of the impact of health policies.
The strongest effect of delivery hospital was seen in the improvement in mortality rates. The 100% to 300% higher risk-adjusted mortality rates at hospitals without a high-level NICU suggest that the delivery, resuscitation, and initial management of a premature infant is highly important to the infant’s survival. The use of these therapies varies between hospitals.38 Our work also examined the association between delivery hospital and neonatal complication rates. After adjusting for differences in case mix, rates of important complications, such as BPD, necrotizing enterocolitis (NEC), and retinopathy of prematurity (ROP),39 were statistically similar in the 2 hospitals. This similarity occurred despite lower mortality rates at high-level NICUs, which should increase the rate of these complications that are diagnosed several weeks or months after birth.
Our study also suggests that the association between delivery hospital and neonatal outcomes differs between states. Although our study design may adjust for unmeasured differences in case mix between types of hospitals, systematic differences between states still exist, including the characteristics of women delivering in various types of hospitals in each state. Additionally, differences in regionalization legislation or financial incentives to hospitals may change which hospitals build a low- or high-level NICU.2,15 For example, California has stronger regionalization legislation, which reduced the number of level I delivery hospitals in comparison with other states. Thus, the “other delivery hospital” group is made up of hospitals with different characteristics in California in comparison with the other states. Each of these factors could contribute to the state differences observed in this study.
One complication that remains elevated at high-level NICUs in all 3 states was bacterial infections. Studies suggest that organizational factors of the NICU, such as increased patient-nurse ratios and fewer sinks, are associated with higher infection rates.40,41 Although increased volume was not associated with increased infections in those studies, high-level NICUs may also have periods of crowding or increased occupancy, which are also associated with higher infection rates.42–45 In contrast, recent data from multiple NICUs in New York City suggests that infection rates could be lower in high-level NICUs with the use of standardized guidelines.46 Determining hospital characteristics and policies associated with lower infection rates is important to optimize the care of premature infants.
There are several limitations to this study. We used International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes to detect complications of pregnancy and of premature birth. Thus, there may be some heterogeneity in how different hospitals code for these conditions. We included 2 surgical conditions, which should be coded more accurately than medical diagnoses,47 and found similar patterns to the medical complications. There are no codes for the administration of surfactant or antenatal corticosteroids in our dataset, so the impact of these therapies on individual patients could not be measured. The instrumental variables approach cannot estimate what would happen to those women who, because of preexisting maternal or fetal conditions, would always deliver at a high-level NICU. Given the specialized conditions of these pregnancies, the effect on mortality is likely larger than we have reported.
In conclusion, our work suggests that the survival benefit to delivering at a high-level NICU between 1995 and 2005 is larger than previously reported and appears to benefit both extremely preterm and moderately preterm infants. Complication rates were similar between hospital types. These benefits vary between states, suggesting that the effect of delivery hospital may depend on the organization of perinatal services or the types of populations served. Assessments of perinatal policies that only use variables available in administrative databases may not adequately adjust for actual case mix differences between hospitals.
Supplementary Material
Glossary
- BPD
bronchopulmonary dysplasia
- CI
confidence interval
- ICD-9-CM
International Classification of Diseases, Ninth Revision, Clinical Modification
- NEC
necrotizing enterocolitis
- ROP
retinopathy of prematurity
- RR
risk ratio
Footnotes
Drs Lorch, Baiocchi, and Small were responsible for study concept and design and drafting of the manuscript; Dr Lorch was responsible for acquisition of the data and obtaining funding; Drs Lorch and Baiocchi, Corinne Ahlberg, and Dr Small were responsible for analysis and interpretation of the data, critical revision of the manuscript for important intellectual content, statistical analysis, administrative, technical, or material support, and final approval of the version to be published; Drs Lorch and Small were responsible for study supervision; and Dr Lorch, principal investigator, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: This study was funded by AHRQ grant R01 HS 015696, “Perinatal Regionalization and Quality of Care”. AHRQ was not responsible for the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.
References
- 1.Institute of Medicine Emergency Medical Services at the Crossroads. Committee on the Future of Emergency Care in the United States Health System, Board on Health Care Services. Washington, DC: National Academies Press; 2007 [Google Scholar]
- 2.Lorch SA, Myers S, Carr B. The regionalization of pediatric health care. Pediatrics. 2010;126(6):1182–1190 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Meadow W, Kim M, Mendez D, et al. Which nurseries currently care for ventilated neonates in Illinois and Wisconsin? Implications for the next generation of perinatal regionalization. Am J Perinatol. 2002;19(4):197–203 [DOI] [PubMed] [Google Scholar]
- 4.Gould JB, Marks AR, Chavez G. Expansion of community-based perinatal care in California. J Perinatol. 2002;22(8):630–640 [DOI] [PubMed] [Google Scholar]
- 5.Bode MM, O’shea TM, Metzguer KR, Stiles AD. Perinatal regionalization and neonatal mortality in North Carolina, 1968-1994. Am J Obstet Gynecol. 2001;184(6):1302–1307 [DOI] [PubMed] [Google Scholar]
- 6.Powell SL, Holt VL, Hickok DE, Easterling T, Connell FA. Recent changes in delivery site of low-birth-weight infants in Washington: impact on birth weight-specific mortality. Am J Obstet Gynecol. 1995;173(5):1585–1592 [DOI] [PubMed] [Google Scholar]
- 7.Rautava L, Lehtonen L, Peltola M, et al. PERFECT Preterm Infant Study Group . The effect of birth in secondary- or tertiary-level hospitals in Finland on mortality in very preterm infants: a birth-register study. Pediatrics. 2007;119(1). Available at: www.pediatrics.org/cgi/content/full/119/1/e257. [DOI] [PubMed] [Google Scholar]
- 8.Neto MT. Perinatal care in Portugal: effects of 15 years of a regionalized system. Acta Paediatr. 2006;95(11):1349–1352 [DOI] [PubMed] [Google Scholar]
- 9.Phibbs CS, Bronstein JM, Buxton E, Phibbs RH. The effects of patient volume and level of care at the hospital of birth on neonatal mortality. JAMA. 1996;276(13):1054–1059 [PubMed] [Google Scholar]
- 10.Phibbs CS, Baker LC, Caughey AB, Danielsen B, Schmitt SK, Phibbs RH. Level and volume of neonatal intensive care and mortality in very-low-birth-weight infants. N Engl J Med. 2007;356(21):2165–2175 [DOI] [PubMed] [Google Scholar]
- 11.Cifuentes J, Bronstein J, Phibbs CS, Phibbs RH, Schmitt SK, Carlo WA. Mortality in low birth weight infants according to level of neonatal care at hospital of birth. Pediatrics. 2002;109(5):745–751 [DOI] [PubMed] [Google Scholar]
- 12.Lasswell SM, Barfield WD, Rochat RW, Blackmon L. Perinatal regionalization for very low-birth-weight and very preterm infants: a meta-analysis. JAMA. 2010;304(9):992–1000 [DOI] [PubMed] [Google Scholar]
- 13.Synnes AR, Macnab YC, Qiu Z, et al. Canadian Neonatal Network . Neonatal intensive care unit characteristics affect the incidence of severe intraventricular hemorrhage. Med Care. 2006;44(8):754–759 [DOI] [PubMed] [Google Scholar]
- 14.Rautava L, Häkkinen U, Korvenranta E, et al. PERFECT Preterm Infant Study Group . Health and the use of health care services in 5-year-old very-low-birth-weight infants. Acta Paediatr. 2010;99(7):1073–1079 [DOI] [PubMed] [Google Scholar]
- 15.Nowakowski L, Barfield WD, Kroelinger CD, et al. Assessment of state measures of risk-appropriate care for very low birth weight infants and recommendations for enhancing regionalized state systems. Matern Child Health J. 2012;16(1):217–227 [DOI] [PubMed] [Google Scholar]
- 16.McClellan M, McNeil BJ, Newhouse JP. Does more intensive treatment of acute myocardial infarction in the elderly reduce mortality? Analysis using instrumental variables. JAMA. 1994;272(11):859–866 [PubMed] [Google Scholar]
- 17.Stukel TA, Fisher ES, Wennberg DE, Alter DA, Gottlieb DJ, Vermeulen MJ. Analysis of observational studies in the presence of treatment selection bias: effects of invasive cardiac management on AMI survival using propensity score and instrumental variable methods. JAMA. 2007;297(3):278–285 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Wang PS, Schneeweiss S, Avorn J, et al. Risk of death in elderly users of conventional vs. atypical antipsychotic medications. N Engl J Med. 2005;353(22):2335–2341 [DOI] [PubMed] [Google Scholar]
- 19.Meara E, Kotagal UR, Atherton HD, Lieu TA. Impact of early newborn discharge legislation and early follow-up visits on infant outcomes in a state Medicaid population. Pediatrics. 2004;113(6):1619–1627 [DOI] [PubMed] [Google Scholar]
- 20.Schneeweiss S, Seeger JD, Landon J, Walker AM. Aprotinin during coronary-artery bypass grafting and risk of death. N Engl J Med. 2008;358(8):771–783 [DOI] [PubMed] [Google Scholar]
- 21.Xian Y, Holloway RG, Chan PS, et al. Association between stroke center hospitalization for acute ischemic stroke and mortality. JAMA. 2011;305(4):373–380 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Herrchen B, Gould JB, Nesbitt TS. Vital statistics linked birth/infant death and hospital discharge record linkage for epidemiological studies. Comput Biomed Res. 1997;30(4):290–305 [DOI] [PubMed] [Google Scholar]
- 23.Srinivas SK, Fager C, Lorch SA. Evaluating risk-adjusted cesarean delivery rate as a measure of obstetric quality. Obstet Gynecol. 2010;115(5):1007–1013 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Parker JD, Schoendorf KC. Implications of cleaning gestational age data. Paediatr Perinat Epidemiol. 2002;16(2):181–187 [DOI] [PubMed] [Google Scholar]
- 25.American Academy of Pediatrics (AAP). Elk Grove Village, IL: AAP. Available at: www.aap.org/securemoc/neoperi/content/NICU.pdf. Accessed March 28, 2012
- 26.Lorch SA, Maheshwari P, Even-Shoshan O. The impact of certificate of need programs on neonatal intensive care units. J Perinatol. 2012;32(1):39–44 [DOI] [PubMed] [Google Scholar]
- 27.Baiocchi M, Small DS, Lorch S, Rosenbaum PR. Building a stronger instrument in an observational study of perinatal care for premature infants. J Am Stat Assoc. 2010;105(492):1285–1296 [Google Scholar]
- 28.Lorch SA, Silber JH, Even-Shoshan O, Millman A. Use of prolonged travel to improve pediatric risk-adjustment models. Health Serv Res. 2009;44(2 pt 1):519–541 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Angrist JD, Imbens GW, Rubin DB. Identification of causal effects using instrumental variables. J Am Stat Assoc. 1996;91(434):444–455 [Google Scholar]
- 30.Phibbs CS, Mark DH, Luft HS, et al. Choice of hospital for delivery: a comparison of high-risk and low-risk women. Health Serv Res. 1993;28(2):201–222 [PMC free article] [PubMed] [Google Scholar]
- 31.Gould JB, Sarnoff R, Liu H, Bell DR, Chavez G. Very low birth weight births at non-NICU hospitals: the role of sociodemographic, perinatal, and geographic factors. J Perinatol. 1999;19(3):197–205 [DOI] [PubMed] [Google Scholar]
- 32.Cochran WG. The effectiveness of adjustment by subclassification in removing bias in observational studies. Biometrics. 1968;24(2):295–313 [PubMed] [Google Scholar]
- 33.Cochran WG, Rubin DB. Controlling bias in observational studies: a review. Sankhya Ser A. 1973;35(4):417–446 [Google Scholar]
- 34.Yeast JD, Poskin M, Stockbauer JW, Shaffer S. Changing patterns in regionalization of perinatal care and the impact on neonatal mortality. Am J Obstet Gynecol. 1998;178(1 pt 1):131–135 [DOI] [PubMed] [Google Scholar]
- 35.Shlossman PA, Manley JS, Sciscione AC, Colmorgen GH. An analysis of neonatal morbidity and mortality in maternal (in utero) and neonatal transports at 24-34 weeks’ gestation. Am J Perinatol. 1997;14(8):449–456 [DOI] [PubMed] [Google Scholar]
- 36.Rogowski JA, Horbar JD, Staiger DO, Kenny M, Carpenter J, Geppert J. Indirect vs direct hospital quality indicators for very low-birth-weight infants. JAMA. 2004;291(2):202–209 [DOI] [PubMed] [Google Scholar]
- 37.Rogowski JA, Staiger DO, Horbar JD. Variations in the quality of care for very-low-birthweight infants: implications for policy. Health Aff (Millwood). 2004;23(5):88–97 [DOI] [PubMed] [Google Scholar]
- 38.Howell EA, Holzman I, Kleinman LC, Wang J, Chassin MR. Surfactant use for premature infants with respiratory distress syndrome in three New York city hospitals: discordance of practice from a community clinician consensus standard. J Perinatol. 2010;30(9):590–595 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Schmidt B, Asztalos EV, Roberts RS, Robertson CM, Sauve RS, Whitfield MF, Trial of Indomethacin Prophylaxis in Preterms (TIPP) Investigators . Impact of bronchopulmonary dysplasia, brain injury, and severe retinopathy on the outcome of extremely low-birth-weight infants at 18 months: results from the trial of indomethacin prophylaxis in preterms. JAMA. 2003;289(9):1124–1129 [DOI] [PubMed] [Google Scholar]
- 40.Cimiotti JP, Haas J, Saiman L, Larson EL. Impact of staffing on bloodstream infections in the neonatal intensive care unit. Arch Pediatr Adolesc Med. 2006;160(8):832–836 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Parry GJ, Tucker JS, Tarnow-Mordi WO, UK Neonatal Staffing Study Group . Relationship between probable nosocomial bacteraemia and organisational and structural factors in UK neonatal intensive care units. Qual Saf Health Care. 2005;14(4):264–269 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Borg MA, Suda D, Scicluna E. Time-series analysis of the impact of bed occupancy rates on the incidence of methicillin-resistant Staphylococcus aureus infection in overcrowded general wards. Infect Control Hosp Epidemiol. 2008;29(6):496–502 [DOI] [PubMed] [Google Scholar]
- 43.Borg MA. Bed occupancy and overcrowding as determinant factors in the incidence of MRSA infections within general ward settings. J Hosp Infect. 2003;54(4):316–318 [DOI] [PubMed] [Google Scholar]
- 44.Cunningham JB, Kernohan WG, Sowney R. Bed occupancy and turnover interval as determinant factors in MRSA infections in acute settings in Northern Ireland: 1 April 2001 to 31 March 2003. J Hosp Infect. 2005;61(3):189–193 [DOI] [PubMed] [Google Scholar]
- 45.Tucker J, UK Neonatal Staffing Study Group . Patient volume, staffing, and workload in relation to risk-adjusted outcomes in a random stratified sample of UK neonatal intensive care units: a prospective evaluation. Lancet. 2002;359(9301):99–107 [DOI] [PubMed] [Google Scholar]
- 46.Schulman J, Stricof R, Stevens TP, et al. New York State Regional Perinatal Care Centers . Statewide NICU central-line-associated bloodstream infection rates decline after bundles and checklists. Pediatrics. 2011;127(3):436–444 [DOI] [PubMed] [Google Scholar]
- 47.Quan H, Parsons GA, Ghali WA. Validity of procedure codes in International Classification of Diseases, 9th revision, clinical modification administrative data. Med Care. 2004;42(8):801–809 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.