- Email: [email protected]
Cross-Sectional Analysis on Racial and Economic Disparities Affecting Mortality in Preterm Infants with Posthemorrhagic Hydrocephalus
Original Article
Cross-Sectional Analysis on Racial and Economic Disparities Affecting Mortality in Preterm Infants with Posthemorrhagic Hydrocephalus Diana L. Jin1, Eisha A. Christian1, Frank Attenello1, Edward Melamed2, Steven Cen1, Mark D. Krieger1,2, J. Gordon McComb1,2, William J. Mack1
BACKGROUND: Despite major advances in medicine, racial and socioeconomic disparities continue to affect health care outcomes. Higher overall infant mortality has been reported for black neonates compared with their Hispanic and white counterparts. The underlying basis for these differences remains unclear. A potential influencing factor is the management of premature neurologic complications in this disadvantaged group. This study examines racial and socioeconomic disparities on mortality in preterm infants with posthemorrhagic hydrocephalus (PHH).
-
METHODS: Data from the Nationwide Inpatient Sample and Kids Inpatient Database were combined from 2000 to 2010. Discharges with International Classification of Diseases, Ninth Revision, Clinical Modification codes for preterm births with intraventricular hemorrhage and PHH were included. Relative risk (RR) ratios for mortality, complications, length of stay, and hospital costs were obtained with multivariate analysis after controlling for patient-level, hospital-level, and admission-level factors.
mortality compared with those with private insurance (RR [ 1.2; P [ 0.04) after adjusting for patient and hospital factors. CONCLUSIONS: Among preterm infants with intraventricular hemorrhage and resultant PHH, black infants and those insured by Medicaid have significantly increased mortality but these 2 effects are independent. Further studies are needed to fully understand the factors affecting these racial and socioeconomic disparities.
-
-
RESULTS: When controlling for patient and hospital factors, black neonates had increased mortality compared with whites and Hispanics (RR [ 1.47; P < 0.01). This association existed despite lower rates of congenital cardiac defects (RR [ 0.84; P < 0.01), gastrointestinal complications (RR [ 0.84; P < 0.01), and general complications of prematurity (RR [ 0.95; P [ 0.04) in the black cohort. Preterm infants insured by Medicaid had increased
-
Key words Hydrocephalus - Intraventricular hemorrhage - Kids Inpatient Database - Nationwide Inpatient Sample - Preterm neonate - Preterm infant - Posthemorrhagic hydrocephalus -
Abbreviations and Acronyms IMR: Infant mortality rate IVH: Intraventricular hemorrhage KID: Kids Inpatient Database NIS: Nationwide Inpatient Sample
WORLD NEUROSURGERY 88: 399-410, APRIL 2016
INTRODUCTION
I
nfant mortality has long been considered a sensitive indicator of the impact of disparities on the health of populations. The infant mortality rate (IMR, defined as the death of an infant per 1000 live births) in the United States has declined since 19601; however, the United States ranks 31st among the Organization for Economic Cooperation and Development countries, with an IMR of 6.2 (mean for the index is 4.2). Although the United States and Turkey have the highest preterm birth rates (12%), this characteristic alone does not sufficiently explain the relatively high IMR in the United States (e.g., Austria has a preterm birth rate of 11% but an IMR of 3.8).1 Racial and socioeconomic disparities have a profound impact on infant mortality in the United States. For example, the IMR ratio for black to white infants from 1960 to 2011 has remained unchanged, and the IMR of black infants is still double that of
PHH: Posthemorrhagic hydrocephalus RR: Relative risk From the 1Department of Neurosurgery, Keck School of Medicine of USC; and 2Division of Neurosurgery, Children’s Hospital Los Angeles, Los Angeles, California, USA To whom correspondence should be addressed: Diana L. Jin, B.A. [E-mail: [email protected]] Citation: World Neurosurg. (2016) 88:399-410. http://dx.doi.org/10.1016/j.wneu.2015.12.046 Journal homepage: www.WORLDNEUROSURGERY.org Available online: www.sciencedirect.com 1878-8750/$ - see front matter ª 2016 Elsevier Inc. All rights reserved.
www.WORLDNEUROSURGERY.org
399
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
Table 1. Continued
Table 1. Cohort Demographics: Patient Factors Variable
Category
Frequency
Percentage
13,736
100
Male
8028
58
Female
5706
Total patients Gender
Missing Race
*
42
3
4890
36
01
4
5651
41
Unspecified
1518
11
Hispanic
2070
15
240
2
86
1
821
6
3346
24 01
Medicare
12
Medicaid
7449
54
Private including HMO
5477
40
No charge Other
Gestational age
8
19
Self-pay
Birth weight
1127
2610
Missing Payer status
2
Black
Other
248
2 01
* 504
4
Missing
43
1
<500 g
197
1
500e749 g
3299
24
750e999 g
3699
27
1000e1249 g
2336
17
1250e1499 g
1526
11
1500e1749 g
789
6
1750e1999 g
583
5
2000e2499 g
614
5
>2500 g
271
2
Missing
305
2
<24 weeks
670
5
24 weeks
1095
8
25e26 weeks
2514
18
27e28 weeks
2111
15
29e30 weeks
1165
9
31e32 weeks
819
6
33e34 weeks
448
3
35e36 weeks Missing Unspecified
269
2
4620
34
26
01 Continues
400
www.SCIENCEDIRECT.com
Percentage 4
33
Native American
Frequency 549
4563
Intraventricular hemorrhage grade
Category 1
White
Asian/Pacific
Variable
HMO, Health Maintenance Organization. *Frequency 10.
white infants.1 The impact of insurance status, often used as a proxy for socioeconomic status, on IMRs has also been described: in 2011, the state of Washington reported that infants insured by Medicaid had an IMR of 6.5 versus 3.9 in neonates who did not receive Medicaid.2 Moss and Carver,3 in 1998, also noted that infants covered by Medicaid had higher odds of infant mortality. Disorders related to short gestational periods and low birth weight are among the leading causes of infant mortality in the United States. The management of neurologic complications secondary to premature birth can have a substantial impact on patient outcomes and survival. This study examines outcome disparities in the setting of intraventricular hemorrhage (IVH) and posthemorrhagic hydrocephalus (PHH) in preterm infants after controlling for both patient and hospital factors. METHODS Data Source This study uses 2 of the largest public all-payer inpatient care datasets in the United States4,5 from the Healthcare Cost and Utilization Project. The Nationwide Inpatient Sample (NIS) and Kids Inpatient Database (KID) have longitudinal hospital inpatient discharge data from more than 1000 hospitals, with the NIS representing 20% of all hospital discharges and KID representing 80% of pediatric discharges in the United States. Data from the KID were available for 2000, 2003, 2006, and 2009. For the years that the KID dataset did not cover, pediatric data were extracted from the 2000e2010 NIS database. NIS also has a built-in method for obtaining national estimates of prevalence as described in the Healthcare Cost and Utilization ProjecteNIS Calculating NIS Variances Guide.6 Study Cohort The study population was limited to preterm infants with IVH and hydrocephalus. Admission data were extracted using the International Classification of Diseases, Ninth Revision, Clinical Modification codes for patients with a preterm birth weight (765.00e765.29), IVH grade (772.10e772.14), and hydrocephalus (742.3, 331.3, 331.4). To prevent double-counting, admissions that resulted in a discharge to another short-term, acute-care facility
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2015.12.046
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
Table 2. Patient Demographics: Hospital Factors Variable
Category
Table 3. Complications and Comorbidities
Frequency Percentage Event
Hospital region
Teaching status
Northeast
1704
13
Midwest
3076
22
South
5759
42
West
3197
23
Nonteaching Teaching Missing
Hospital bed size Small
Hospital location
Admission type
11,155
81
366
3 11
Medium
3175
23
Large
8723
64
Missing
366
2
Rural
122
1
Missing
Discharge
16
1472
Urban
Hospital type
2215
Not children’s hospital
13,248
96
366
3
2068
15
Children’s general or specialty
1199
9
Children’s unit
1307
9
Missing
9161
67
Nonroutine
7140
52
Routine
6594
48
Emergency
1938
14
Urgent
3538
26
Elective
993
7
Newborn
5314
39
Trauma center Other Missing
28
01 01
* 1911
14
Admission source Emergency room
96
1
another hospital
4445
32
Other facility
309
2
Routine
4944
36
Missing
3942
29
*Frequency 10.
(pretransfer hospitalizations) were excluded from the study.7 In addition, age of admission was limited to less than 3 months. Both NIS and KID include patient and hospital factors as categorical demographic variables. Discrete patient factors, including race (white, black, Hispanic, Asian, or Pacific Islander, Native American, other), payer status (Medicare, Medicaid, private insurance, self-pay, no charge), gender, length of stay, cost of hospitalization, weight, gestational age, and IVH grade, were
WORLD NEUROSURGERY 88: 399-410, APRIL 2016
Frequency Percent
95% Confidence Interval
Shunt replacement
718
5
4.2e6.2
Mechanical implant complication
437
3
2.5e3.9
Shunt removal
857
6
5.2e7.3
Shunt infection
339
2.5
1.8e3.1
67
0.5
0.3e0.7
Other implant complication Neurologic complications Meningitis
42 839
Wound breakdown
56
Wound infection
0.3
0.1e0.5
6
5.2e7
0.4
0.2e0.6
118
0.9
0.5e1.2
Seizures
1997
14.5
12.9e16.2
Necrotizing enterocolitis
1368
10
Intestinal perforation
621
Pulmonary complication
11,437
Gastrointestinal complication Hematologic complication
621 7011
4.5 83 4.5 51
8.9e11.1 3.8e5.3 81.8e84.8 3.8e5.3 48.2e53.9
Sepsis
6547
48
45e50.3
Endocrinologic complication
4960
36
33.8e38.4
Cardiac complications
76
Congenital heart defects
5606
0.6 41
0.2e0.9 38.4e43.3
encoded as categorical variables in NIS and KID. Hospital-level variables such as hospital bed size (small, <200 beds; medium: 201e400 beds; large, >400 beds), region (Northeast, Midwest, South, West), location (urban and rural), type (children’s general/ specialty hospital, children’s unit in a general hospital, general hospital), and teaching status (teaching, nonteaching) were included as categorical variables. Admission status such as admission type (emergency, elective, urgent, trauma center, newborn) and admission source (emergency department, routine, another facility, another hospital) was also included. Because the NIS dataset did not classify hospitals based on the National Association of Children’s Hospitals and Related Institutions as did the KID database, hospitals in the NIS database were identified with their hospital identification ID number and were assigned a classification based on what their KID designation was for the previous available KID year dataset. The primary outcome of interest was mortality. Secondary outcomes of interest included general complications and shunt complications. Shunt complications of interest included replacement of a shunt (2.42), mechanical implant complication (996.2), implant infection (996.63), and other implant complication (996.75). General complications and comorbidities of interest included seizures (779.0), gastrointestinal complications (necrotizing enterocolitis, 777.5e777.53; intestinal perforation, 777.6), pulmonary complications (mechanical ventilation, 96.71, 96.72; respiratory support, 93.90e96.72; respiratory distress, 769,
www.WORLDNEUROSURGERY.org
401
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
Table 4. Continued
Table 4. Multivariate Analysis of Mortality Rate Ratio
95% Confidence Interval
Gender Female
1.18
Male
1.01e1.39
0.04
Reference
Private
1.20
1.01e1.44
0.04
Reference
Medium
0.87
0.61e1.24
0.43
Large
1.09
0.77e1.55
0.62
Midwest
0.85
0.61e1.17
0.32
South
0.87
0.66e1.15
0.33
West
0.98
0.68e1.40
0.90
0.42
0.34e0.52
< 0.01
P Value
Northeast
Race Asian/Pacific
1.62
0.92e2.83
0.09
Black
1.47
1.18e1.83
< 0.01
Hispanic
0.81
0.61e1.08
0.14
Native American
0.86
0.37e2.01
0.73
White
Reference
Weight <500 g
4.47
2.33e8.56
< 0.01
500e749 g
2.55
1.55e4.21
< 0.01
750e999 g
1.66
0.97e2.65
0.06
1000e1249 g
1.10
0.64e1.92
0.73
1250e1499 g
Reference
1500e1749 g
0.73
0.31e1.75
0.48
1750e1999 g
0.44
0.13e1.51
0.19
2000e2499 g
0.75
0.29e1.95
0.56
>2500 g
1.42
0.57e3.54
0.45
Gestational age <24 weeks
3.36
1.65e6.85
0.01
24 weeks
2.06
1.02e4.14
0.04
25e26 weeks
3.49
1.78e6.83
0.01
27e28 weeks
1.29
0.63e2.65
0.49
29e30 weeks
Reference
31e32 weeks
0.80
0.30e2.12
0.65
33e34 weeks
1.17
0.38e3.58
0.78
35e36 weeks
2.3
0.86e6.16
0.10
0.88
0.44e1.76
0.73
2
1.21
0.80e1.84
0.40
3
Reference
4
2.83
2.26e3.55
< 0.01
Intraventricular hemorrhage grade 1
Hospital bed size Small Continues
402
95% Confidence Interval
Hospital region
Payer Medicaid
Rate Ratio
P Value
www.SCIENCEDIRECT.com
Teaching status Teaching Nonteaching Hospital type Children’s unit
0.73
0.50e1.09
0.12
Not children’s hospital
0.81
0.58e1.15
0.25
Children’s hospital
Reference
Bold indicates statistically significant with P < 0.05.
770.6e770.8, 770.89), cardiac complications (cardiac arrest, 779.85; hypotension, 458.9; hypovolemia, 785.59), hematologic complications (disseminated intravascular coagulation, 776.2; jaundice, 774.2, 774.6; hematologic disorder, 776.0e776.9), sepsis (V29.0, 771.8, 771.81, 771.83, 771.89), endocrine complications (775.1e775.9), and congenital heart defects (745.0e747.9). Length of stay was also included as a dichotomized outcome: high length of stay was defined as greater than 120 days (90th percentile). Daily cost was dichotomized into high and low, with high daily cost being more than $12,654 (90th percentile). High total cost was defined as greater than $807,627 (90th percentile). For categorical variables, missing code was generated to represent missing data when fitting the model. Statistical Analysis The primary goal was to assess socioeconomic, demographic, and hospital factors associated with mortality. Outcomes of interest were 1) mortality; 2) general and specific complications as listed above; and 3) length and cost of stay. The predicting factors included patient-level factors, hospital-level factors, and admission statuselevel factors. Hospital location (urban vs. rural) was excluded as a predicting factor because nearly all patients went to urban hospitals. A univariate model (unadjusted model) was initially generated for all patient and hospital factors for the sample, after which a multivariate logistic regression model was created that adjusted for patient-level factors (race, payment type, comorbidity, sex, gestational age, birth weight, and IVH grade), hospital-level factors (region, teaching status, bed size, children’s hospital type), admission type (emergency, elective, urgent, trauma center,
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2015.12.046
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
Table 5. Continued
Table 5. Multivariate Analysis of General Complications Rate Ratio
95% Confidence Interval
Gender Female Male
1.00
0.97e1.04
0.80
Private
95% Confidence Interval
Medium
1.19
0.71e2.01
0.51
Large
1.48
0.87e2.52
0.15
1.28e4.41
< 0.01
Reference
Hospital region Midwest
2.38
1.03
Northeast
Reference
Payer Medicaid
Rate Ratio
P Value
0.99e1.07
0.15
Reference
Race
South
1.61
0.90e2.87
0.11
West
1.28
0.59e2.78
0.53
2.57
1.44e4.58
< 0.01
Asian/Pacific
0.99
0.87e1.12
0.84
Teaching status
Black
0.95
0.90e1.00
0.04
Teaching Nonteaching
Hispanic
0.97
0.91e1.02
0.23
Native American
0.94
0.74e1.20
0.63
White Weight <500 g
1.07
0.91e1.27
0.40
500e749 g
1.10
1.02e1.19
< 0.01
750e999 g
1.11
1.03e1.19
< 0.01
1000e1249 g
1.06
0.99e1.14
0.10
1250e1499 g
Reference
1500e1749 g
0.98
0.89e1.07
0.65
1750e1999 g
0.85
0.76e0.96
< 0.01
2000e2499 g
0.93
0.83e1.05
0.25
>2500 g
0.87
0.74e1.02
0.10
Gestational age <24 weeks
1.06
0.95e1.18
0.32
24 weeks
1.08
0.98e1.19
0.13
25e26 weeks
1.10
1.01e1.19
0.03
27e28 weeks
1.05
0.96e1.13
0.28
29e30 weeks
Reference
31e32 weeks
0.90
0.81e1.00
0.05
33e34 weeks
0.89
0.78e1.02
0.09
35e36 weeks
0.89
0.74e1.05
0.17
0.98
0.90e1.08
0.73
2
1.04
0.97e1.11
0.25
3
Reference
4
0.98
0.94e1.02
0.36
Intraventricular hemorrhage grade 1
Reference
Hospital type Children’s hospital
Reference
Hospital bed size Small Continues
WORLD NEUROSURGERY 88: 399-410, APRIL 2016
P Value
Reference
Children’s unit
0.99
0.56e1.75
0.97
Not children’s hospital
0.82
0.47e1.42
0.48
Bold indicates statistically significant with P < 0.05.
newborn), and admission source (emergency department, routine, another facility, another hospital). The association between predicting factors and outcome was assessed by the relative risk (RR) ratio, and statistical significance was predetermined as P < 0.05. All descriptive, univariate, and multivariate analyses were performed using SAS 9.4 (SAS Institute Inc., Cary, North Carolina, USA). RESULTS Patient Demographics A total of 13,736 patients had IVH and PHH. Patient and hospital demographics are presented in Tables 1 and 2. Most patients were white (33%), covered by Medicaid (54%), had a birth weight between 750 and 999 g (27%), had a gestational age of 25e26 weeks (18%), and presented with grade 4 IVH (41%). Most patients were admitted to urban, nonchildren’s teaching hospitals with a large bed size. Average length of stay for the entire cohort was 62 days, and average hospital cost $359,158 per admission. Mortality was 10.6%. Complications and comorbidities are presented in Table 3. The most common complications were pulmonary (83%) and hematologic (51%). Variables Affecting Inpatient Mortality Multivariable analyses examining associations between patient demographics/hospital-level factors and inpatient mortality are presented in Table 4. Black patients had a 47% increased risk of inpatient death compared with white patients (RR ¼ 1.47; P < 0.01). Medicaid patients had a 20% increased rate of
www.WORLDNEUROSURGERY.org
403
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
Table 6. Multivariate Analysis of Congenital Heart Anomalies Rate Ratio
95% Confidence Interval
Table 6. Continued
P Value
Rate Ratio
95% Confidence Interval
P Value
Hospital bed size
Gender Female
1.11
Male
1.02e1.22
0.02
Private
Reference
0.93
0.78e1.11
0.43
Large
1.01
0.85e1.21
0.90
1.03
Hospital region Midwest
1.02
0.86e1.22
0.78
South
1.02
0.88e1.19
0.81
West
0.96
0.79e1.18
0.71
1.01e1.32
0.04
0.94e1.14
0.50
Reference
Race Asian/Pacific
1.00
0.73e1.38
1.00
Black
0.84
0.73e0.95
< 0.01
Northeast
Hispanic
0.91
0.79e1.05
0.22
Teaching status
Native American
1.02
0.57e1.82
0.95
Teaching
White
Reference
Medium
Payer Medicaid
Small
Nonteaching
Reference
Reference
1.15 Reference
Hospital type
Weight <500 g
1.31
0.88e1.96
0.18
Children’s unit
0.92
0.76e1.12
0.43
0.95
0.8e1.14
0.57
500e749 g
1.20
0.98e1.48
0.08
Not children’s hospital
750e999 g
1.20
0.99e1.45
0.07
Children’s hospital
1000e1249 g
1.19
0.98e1.45
0.07
1250e1499 g
Reference
1500e1749 g
1.07
0.83e1.38
0.58
1750e1999 g
0.87
0.62e1.20
0.39
2000e2499 g
0.96
0.69e1.34
0.82
>2500 g
1.07
0.70e1.63
0.76
<24 weeks
1.5
1.13e1.98
< 0.01
24 weeks
1.42
1.11e1.82
< 0.01
25e26 weeks
1.43
1.15e1.78
< 0.01
27e28 weeks
1.23
1.00e1.52
0.05
29e30 weeks
Reference
Reference
Bold indicates statistically significant with P < 0.05.
inpatient mortality compared with those with private insurance (RR ¼ 1.20; P ¼ 0.04). Female gender, weight <750 g, gestational age <27 weeks, and patients with grade 4 IVH had increased inpatient mortality. Teaching hospitals had lower mortality compared with nonteaching hospitals.
Gestational age
31e32 weeks
0.88
0.67e1.17
0.38
33e34 weeks
0.86
0.59e1.26
0.44
35e36 weeks
1.09
0.7e1.69
0.70
1.04
0.81e1.33
0.75
2
1.17
0.99e1.38
0.07
3
Reference
4
1.04
0.94e1.15
0.46
Intraventricular hemorrhage grade 1
Continues
404
www.SCIENCEDIRECT.com
Variables Affecting Perinatal Complications Multivariable analyses examining associations between patient demographics/hospital-level factors and perinatal complications are presented in Tables 5e8. Compared with their white counterparts, black infants had a lower rate of general complications (RR ¼ 0.95; P ¼ 0.04, Table 4), fewer cardiac anomalies (RR ¼ 0.84; P < 0.01), and fewer gastrointestinal complications (RR ¼ 0.59; P ¼ 0.01). Medicaid patients had a higher likelihood of overall shunt complications (RR ¼ 1.16; P ¼ 0.02). Gestational age less than 26 weeks was associated with an increased rate of cardiac defects, and birth weight less than 1000 g was associated with an increased risk of gastrointestinal complications, including necrotizing enterocolitis and intestinal perforation. Length of Stay Multivariable analyses examining associations between patient demographics/hospital-level factors and length of stay are presented in Table 9. Blacks and Asians had increased length of stay compared with their white counterparts (RR ¼ 1.37; P ¼ 0.02 and
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2015.12.046
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
Table 7. Multivariate Analysis of Gastrointestinal Complications Rate Ratio
95% Confidence Interval
Male
0.63
0.47e0.84
< 0.01
Reference
Private
1.06
95% Confidence Interval
Medium
1.19
0.71e2.01
0.51
Large
1.48
0.87e2.52
0.15
Midwest
2.38
1.28e4.41
< 0.01
South
1.61
0.90e2.87
0.11
West
1.28
0.59e2.78
0.53
1.44e4.58
< 0.01
0.79e1.42
0.68
Reference
Northeast
Race Asian/Pacific
0.91
0.33e2.53
0.85
Teaching status
Black
0.59
0.39e0.90
0.01
Teaching
Hispanic
0.86
0.55e1.35
0.52
Native American White
0.73
P Value
Hospital region
Payer Medicaid
Rate Ratio
P Value
Gender Female
Table 7. Continued
0.10e5.36
0.75
Reference
Weight <500 g
5.23
1.42e19.33
0.01
500e749 g
4.93
2.04e11.89
< 0.01
750e999 g
3.82
1.60e9.15
< 0.01
1000e1249 g
2.47
1.00e6.12
0.05
1250e1499 g
Reference
1500e1749 g
0.28
0.03e2.35
0.24
1750e1999 g
0.45
0.05e3.86
0.46
2000e2499 g
N/A
N/A
N/A
>2500 g
N/A
N/A
N/A
<24 weeks
1.92
0.59e6.28
0.28
24 weeks
2.03
0.66e6.23
0.21
25e26 weeks
1.67
0.57e4.94
0.35
27e28 weeks
1.02
0.33e3.16
0.97
29e30 weeks
Reference
Reference
2.57
Nonteaching
Reference
Hospital type Children’s unit
0.99
0.56e1.75
0.97
Not children’s hospital
0.82
0.47e1.42
0.48
Children’s hospital
Reference
Bold indicates statistically significant with P < 0.05.
RR ¼ 2.21; P < 0.01, respectively). Furthermore, Medicaid patients had increased length of stay compared with private insurance patients (RR ¼ 1.38; P < 0.01). Patients with grade 4 IVH had longer lengths of stay and patients at teaching hospitals had longer lengths of stay compared with patients at nonteaching hospitals.
Gestational age
31e32 weeks
1.30
0.23e7.31
0.76
33e34 weeks
1.89
0.20e17.50
0.57
35e36 weeks
N/A
N/A
N/A
Intraventricular hemorrhage grade 1
0.56
0.17e1.78
0.32
2
0.60
0.30e1.21
0.16
3
Reference
4
1.02
0.74e1.39
0.92
Hospital bed size Small
Reference Continues
WORLD NEUROSURGERY 88: 399-410, APRIL 2016
Hospital Cost Multivariable analyses examining associations between patient demographics/hospital-level factors and high daily and total hospital costs were also conducted, presented in Tables 10 and 11. Fewer black patients had high daily costs compared with their white counterparts (RR ¼ 0.69; P ¼ 0.02). Yet, more Hispanic patients had higher daily costs than did white patients (RR ¼ 1.33; P ¼ 0.03). Fewer female patients (RR ¼ 0.76; P < 0.01) and medium bed size hospitals (RR ¼ 0.51; P < 0.01) had higher costs than did male patients and small bed size hospitals, respectively. Patients in the Midwest (RR ¼ 0.39; P < 0.01) and South (RR ¼ 0.51; P < 0.01), and teaching hospitals (RR ¼ 0.65; P < 0.01) had less high daily costs than did patients in the Northeast and nonteaching hospitals, respectively. DISCUSSION Racial and socioeconomic factors affect health care delivery in the United States despite widespread efforts to reduce disparities. Minorities and economically disadvantaged groups continue to experience inferior outcomes across an array of diseases and diagnoses, including cancer,8 heart disease,9 stroke,9 and
www.WORLDNEUROSURGERY.org
405
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
Table 8. Multivariate Analysis of Overall Shunt Complications Rate Ratio
95% Confidence Interval
Table 8. Continued
P Value
Rate Ratio
Female
1.05
Male
0.93e1.18
0.41
Medicaid Private
Small
Reference
0.63
0.51e0.78
< 0.01
Large
0.94
0.77e1.16
0.59
1.16
Hospital region Midwest
1.07
0.86e1.33
0.56
South
0.92
0.75e1.12
0.39
West
1.49
1.16e1.92
< 0.01
1.59e2.49
< 0.01
0.75
0.59e0.94
0.01
0.48
0.38e0.60
< 0.01
1.03e1.32
0.02
Reference
Race Asian/Pacific
0.74
0.46e1.20
0.23
Black
0.99
0.84e1.17
0.91
Northeast
Hispanic
0.96
0.79e1.16
0.67
Teaching status
Native American
0.76
0.31e1.86
0.55
Teaching
White
Reference
Medium
Payer
Nonteaching
Reference
Reference
1.99 Reference
Hospital type
Weight <500 g
0.74
0.35e1.54
0.42
500e749 g
1.41
1.08e1.83
0.01
750e999 g
1.49
1.16e1.91
< 0.01
1000e1249 g
1.23
0.95e1.59
0.12
1250e1499 g
Reference
1500e1749 g
1.52
1.12e2.07
< 0.01
1750e1999 g
0.95
0.61e1.46
0.80
2000e2499 g
0.89
0.55e1.41
0.61
>2500 g
1.21
0.70e2.09
0.49
Gestational age <24 weeks
0.98
0.68e1.42
0.93
24 weeks
0.76
0.55e1.05
0.10
25e26 weeks
1.01
0.77e1.32
0.97
27e28 weeks
0.81
0.62e1.06
0.13
29e30 weeks
Reference
31e32 weeks
0.61
0.42e0.90
0.01
33e34 weeks
0.80
0.49e1.31
0.38
35e36 weeks
0.57
0.29e1.12
0.10
0.65
0.43e0.99
0.05
2
0.57
0.43e0.77
< 0.01
3
Reference
4
1.31
1.15e1.49
< 0.01
Intraventricular hemorrhage grade
Continues
406
P Value
Hospital bed size
Gender
1
95% Confidence Interval
www.SCIENCEDIRECT.com
Children’s unit Not children’s hospital Children’s hospital
Reference
Bold indicates statistically significant with P < 0.05.
diabetes.10 Epidemiologic studies have noted that African American women exhibit a 6% lower incidence of cancer than white women, but a 16% higher mortality as a result of the disease.8 Researchers have suggested multiple factors that may affect these outcomes, including availability of proper nutrition, health insurance coverage, income, education level, access to disease screening, and physicianepatient interactions.1,11 Elimination of these disparities has been designated a major goal in Healthy People 2010.12 Most studies of health care disparities focus on the adult population, but these inequalities have also extended into the pediatric realm. African American children have decreased access to care, substantially lower adherence to preventive health measures, and decreased quality in delivery of health care.13 Black children are less likely than white children to be selected from the kidney transplant waiting list and are more likely to receive inadequate hemodialysis.13 These disparities have also been documented in the perinatal literature. El-Sayed et al.14 recently published a study examining all births in Michigan between 1989 and 2005. These investigators found that socioeconomic factors and maternal demographic, behavioral, and health access differences explained approximately 30% of disparities in infant mortality in disorders relating to preterm birth. Willis et al.1 reported an IMR of 11.4 per 1000 live births for black infants and an IMR of
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2015.12.046
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
Table 9. Continued
Table 9. Multivariate Analysis of High Length of Stay Rate Ratio
95% Confidence Interval
Gender Female Male
0.93
0.77e1.11
0.43
Reference
Private
1.38
95% Confidence Interval
Medium
0.84
0.60e1.17
0.29
Large
1.11
0.79e1.55
0.55
1.13e1.69
< 0.01
Reference
Midwest
1.48
1.04e2.1
0.03
South
1.20
0.87e1.65
0.27
West
1.17
0.75e1.83
0.50
1.15e2.25
< 0.01
Northeast
Race 2.21
1.27e3.86
< 0.01
Black
1.37
1.06e1.76
0.02
Teaching Nonteaching
Hispanic
1.25
0.91e1.71
0.17
Native American
1.02
0.32e3.24
0.98
Reference
Weight <500 g
12.85
3.93e42.07
< 0.01
500e749 g
14.92
5.41e41.13
< 0.01
750e999 g
10.63
3.86e29.27
< 0.01
5.35
1.89e15.17
< 0.01
1000e1249 g 1250e1499 g
Reference
1500e1749 g
1.78
0.44e7.14
0.42
1750e1999 g
0.71
0.08e6.46
0.76
2000e2499 g
2.05
0.43e9.74
0.37
>2500 g
1.15
0.12e11.09
0.90
<24 weeks
1.83
0.85e3.93
0.12
24 weeks
2.23
1.07e4.64
0.03
25e26 weeks
1.15
0.55e2.37
0.71
0.62
0.28e1.37
0.24
Gestational age
27e28 weeks 29e30 weeks
Reference
31e32 weeks
0.67
0.18e2.52
0.55
33e34 weeks
0.42
0.05e3.47
0.42
35e36 weeks
1.40
0.27e7.33
0.69
1.20
0.66e2.18
0.56
1.15
0.79e1.67
0.48
1.02e1.58
0.03
Intraventricular hemorrhage grade 1 2 3 4
Reference 1.27
Hospital bed size Small
Reference Continues
WORLD NEUROSURGERY 88: 399-410, APRIL 2016
Reference
Teaching status
Asian/Pacific
White
P Value
Hospital region
Payer Medicaid
Rate Ratio
P Value
1.61 Reference
Hospital type Children’s unit
0.67
0.46e0.96
0.03
Not children’s hospital
0.46
0.32e0.65
< 0.01
Children’s hospital
Reference
Bold indicates statistically significant with P < 0.05.
5.1 per 1000 live white births in 2011. Furthermore, multiple studies examining U.S.- and African-born infants have demonstrated that genetic differences alone fail to explain the disparity.15-17 When accounting for traditional confounders such as payer status, gestational age, birth weight, and IVH grade, our model demonstrates that black preterm infants with PHH have a higher mortality and increased length of hospital stay compared with their white counterparts. Paradoxically, we demonstrate that black infants with PHH have a lower rate of overall general comorbidities and complications related to prematurity such as congenital cardiac anomalies and gastrointestinal complications, including necrotizing enterocolitis and intestinal perforation. Despite these lower overall comorbidities and complication rates, black PHH infants had a 47% higher rate of mortality than had their white counterparts. By contrast, our Hispanic cohort did not demonstrate an increased risk of mortality. Our findings comport with an existing perinatal literature that describes the Hispanic Paradox.18 Gould et al.19 reported that despite a highrisk profile for Hispanic mothers (because of lack of access to prenatal care), the population did not have an increased rate of low-birth-weight infants and neonatal mortality. In this report, both black and Hispanic mothers had decreased access to prenatal healthcare and less maternal education, yet black neonates had worse outcomes than had Hispanic neonates. One potential explanation for this paradox is that Hispanic mothers are subject to lower incidences of both preterm and low-birth-weight infants. Our study cohort, for this reason, is specifically limited to low-birth-weight, preterm infants who develop IVH and PHH.
www.WORLDNEUROSURGERY.org
407
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
Table 10. Continued
Table 10. Multivariate Analysis of High Daily Cost Rate Ratio
95% Confidence Interval
Rate Ratio
95% Confidence Interval
P Value
Midwest
0.39
0.25e0.59
< 0.01
South
0.51
0.36e0.72
< 0.01
0.75
0.49e1.15
0.19
0.5e0.83
< 0.01
P Value Hospital region
Gender Female
1.00
Male
0.82e1.21
0.97
Reference
West
Payer Medicaid Private
0.87
0.7e1.07
0.18
Northeast
Reference
Teaching status
Reference
Teaching
Race
0.65
Nonteaching
Reference
Asian/Pacific
1.11
0.64e1.93
0.70
Black
0.69
0.50e0.94
0.02
Hispanic
1.33
1.03e1.72
0.03
Children’s unit
1.44
0.97e2.15
0.07
Native American
0.89
0.22e3.66
0.87
Not children’s hospital
0.87
0.61e1.24
0.44
White
Hospital type
Children’s hospital
Reference
Reference
Bold indicates statistically significant with P < 0.05.
Weight <500 g
2.81
1.31e6.03
< 0.01
500e749 g
2.11
1.30e3.43
< 0.01
750e999 g
1.45
0.91e2.31
0.12
1000e1249 g
1.31
0.82e2.11
0.26
1250e1499 g
Reference
1500e1749 g
1.18
0.65e2.16
0.58
1750e1999 g
1.00
0.47e2.14
1.00
2000e2499 g
0.57
0.23e1.40
0.22
>2500 g
0.93
0.37e2.34
0.89
Gestational age <24 weeks
1.62
0.88e2.97
0.12
24 weeks
1.64
0.94e2.86
0.08
25e26 weeks
1.64
0.98e2.75
0.06
27e28 weeks
1.18
0.71e1.96
0.53
29e30 weeks
Reference
31e32 weeks
0.86
0.43e1.73
0.68
33e34 weeks
1.21
0.52e2.82
0.65
35e36 weeks
3.40
1.51e7.66
< 0.01
0.80
0.45e1.41
0.44
2
0.64
0.42e0.98
0.04
3
Reference
4
1.24
1.00e1.53
0.05
Even within this selected subgroup, black infants demonstrate higher mortality. We also demonstrate that patients covered by Medicaid have an increased mortality and length of stay compared with those with private insurance after controlling for relevant confounders. This finding is consistent with results in previous socioeconomic studies in the pediatric neurosurgical population. Attenello et al.20 demonstrated that pediatric patients insured by Medicaid were 12% less likely to have routine discharges after cerebrospinal fluid shunting procedures compared with patients with private insurance. Additional studies have demonstrated that pediatric patients of lower socioeconomic status have longer hospital lengths of stay and higher infection rates.21,22 The principal limitations to this study result from the use of a large population-based administrative database. International Classification of Diseases, Ninth Revision, Clinical Modification codes are entered by hospital staff with varied training and
Intraventricular hemorrhage grade 1
Hospital bed size Small
Reference
Medium
0.94
0.64e1.38
0.77
Large
0.87
0.59e1.28
0.49 Continues
408
www.SCIENCEDIRECT.com
Figure 1. Admission frequency.
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2015.12.046
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
Table 11. Continued
Table 11. Multivariate Analysis of High Total Cost Rate Ratio
95% Confidence Interval
P Value
Male
0.76
0.63e0.93
< 0.01
Private
Small
0.51
0.36e0.73
< 0.01
Large
0.74
0.52e1.05
0.09
0.95
Hospital region Midwest
1.07
0.74e1.53
0.73
South
0.89
0.64e1.24
0.50
West
1.51
1.00e2.28
0.05
0.95e1.75
0.10
0.82
0.55e1.22
0.32
0.54
0.37e0.78
< 0.01
0.77e1.16
0.59
Reference
Asian/Pacific
0.89
0.46e1.71
0.73
Black
1.06
0.81e1.37
0.68
Northeast
Hispanic
1.34
1.00e1.78
0.05
Teaching status
Native American
1.54
0.62e3.86
0.35
Teaching
Reference
1.29
Nonteaching
Reference
Reference
Hospital type
Weight <500 g
2.08
0.78e5.51
500e749 g
3.35
1.90e5.9
0.14
Children’s unit
< 0.01
Not children’s hospital Children’s hospital
750e999 g
2.75
1.57e4.81
< 0.01
1000e1249 g
1.52
0.84e2.74
0.17
1250e1499 g
Reference
1500e1749 g
0.44
0.15e1.33
0.14
1750e1999 g
0.59
0.17e2.07
0.41
2000e2499 g
0.42
0.11e1.62
0.21
>2500 g
0.58
0.14e2.40
0.45
Gestational age <24 weeks
1.48
0.77e2.84
0.24
24 weeks
1.53
0.83e2.82
0.17
25e26 weeks
1.25
0.70e2.25
0.46
27e28 weeks
0.74
0.40e1.37
0.33
29e30 weeks
Reference
31e32 weeks
0.24
0.05e1.06
0.06
33e34 weeks
0.90
0.24e3.31
0.87
35e36 weeks
2.45
0.80e7.54
0.12
0.83
0.43e1.59
0.58
2
1.22
0.85e1.77
0.28
3
Reference
4
1.24
1.00e1.54
0.05
Reference
Bold indicates statistically significant with P < 0.05.
oversight. The potential for recording bias results in estimates of coding accuracy of approximately 80%.23 This bias can affect not only coding of diagnoses but also covariates and complications. In addition, these data sets capture only a single admission, and therefore patients who develop delayed hydrocephalus are not included. Although we used previously published methods of preventing double-counting,7 this phenomenon cannot be completely eliminated because of the lack of individual identifiers. CONCLUSIONS
Intraventricular hemorrhage grade 1
Reference
Reference
Race
White
P Value
Medium
Payer Medicaid
95% Confidence Interval
Hospital bed size
Gender Female
Rate Ratio
Continues
WORLD NEUROSURGERY 88: 399-410, APRIL 2016
The United States continues to have a high IMR compared with other Organization for Economic Cooperation and Development countries. The presence of socioeconomic disparities in health care delivery may contribute to this increased mortality. Our data, after multivariable analyses accounting for both patient and hospital factors, suggest that black preterm neonates with PHH have increased mortality despite lower overall complications of prematurity. Increased mortality is also seen in preterm neonates with PHH who have Medicaid insurance coverage independent of their racial status. Further studies are needed to understand the extent of these differences and potential causative factors.
www.WORLDNEUROSURGERY.org
409
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
10. Hedderson M, Ehrlich S, Sridhar S, Darbinian J, Moore S, Ferrara A. Racial/ethnic disparities in the prevalence of gestational diabetes mellitus by BMI. Diabetes Care. 2012;35:1492-1498.
19. Gould JB, Madan A, Qin C, Chavez G. Perinatal outcomes in two dissimilar immigrant populations in the United States: a dual epidemiologic paradox. Pediatrics. 2003;111:e676-682.
2. Wasserman C, Taylor P. Infant Mortality. Washington, DC: Washington State Department of Health; 2013.
11. Smedley BD, Stith AY, Nelson AR, eds. Unequal Treatment: Confonting Racial and Ethnic Disparities in Healthcare. Washington, DC: The National Academies Press; 2003.
20. Attenello FJ, Ng A, Wen T, Cen SY, Sanossian N, Amar AP, et al. Racial and socioeconomic disparities in outcomes following pediatric cerebrospinal fluid shunt procedures. J Neurosurg Pediatr. 2015;15:560-566.
3. Moss N, Carver K. The effect of WIC and Medicaid on infant mortality in the United States. Am J Public Health. 1998;88:1354-1361.
12. National Center for Health Statistics. Healthy People 2010 Final Review. Hyattsville: US Government Printing Office, MD. 2012.
4. Nationwide Inpatient Sample. Agency for Healthcare Research and Quality; 2007. Available at: www. hcup-us.ahrq.gov/databases.jsp. Accessed June 20, 2014.
13. Flores G, Committee On Pediatric Research. Technical reporteracial and ethnic disparities in the health and health care of children. Pediatrics. 2010;125:e979-e1020.
5. Introduction to the KID. Healthcare Cost and Utilization Project (HCUP). 2015. Available at: www.hcup-us.ahrq.gov/db/nation/kid/kid_2012_ introduction.jsp. Accessed February 20, 2015.
14. El-Sayed AM, Finkton DW Jr, Paczkowski M, Keyes KM, Galea S. Socioeconomic position, health behaviors, and racial disparities in causespecific infant mortality in Michigan, USA. Prev Med. 2015;76:8-13.
6. Final Report on Calculating Nationwide Inpatient Sample (NIS) Variances, 2001. U.S. Agency for Healthcare Research and Quality; 2005. Available at: http://www.hcup-us.ahrq.gov/reports/ methods/CalculatingNISVariances200106092005.pdf. Accessed December 10, 2014.
15. David RJ, Collins JW Jr. Differing birth weight among infants of U.S.-born blacks, African-born blacks, and U.S.-born whites. N Engl J Med. 1997; 337:1209-1214.
REFERENCES 1. Willis E, McManus P, Magallanes N, Johnson S, Majnik A. Conquering racial disparities in perinatal outcomes. Clin Perinatol. 2014;41:847-875.
7. Burke BL, Robbins JM, Bird TM, Hobbs CA, Nesmith C, Tilford JM. Trends in hospitalizations for neonatal jaundice and kernicterus in the United States, 1988-2005. Pediatrics. 2009;123: 524-532.
16. Collins JW Jr, Wu SY, David RJ. Differing intergenerational birth weights among the descendants of US-born and foreign-born Whites and African Americans in Illinois. Am J Epidemiol. 2002;155:210-216.
21. Walker CT, Stone JJ, Jain M, Jacobson M, Phillips V, Silberstein HJ. The effects of socioeconomic status and race on pediatric neurosurgical shunting. Childs Nerv Syst. 2014;30:117-122. 22. Simon TD, Hall M, Riva-Cambrin J, Albert JE, Jeffries HE, Lafleur B, et al. Infection rates following initial cerebrospinal fluid shunt placement across pediatric hospitals in the United States. J Neurosurg Pediatr. 2009;4:156-165. 23. Burns EM, Rigby E, Mamidanna R, Bottle A, Aylin P, Ziprin P, et al. Systematic review of discharge coding accuracy. J Public Health (Oxf). 2012;34:138-148.
Conflict of interest statement: The authors declare that the article content was composed in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
17. David R, Collins J Jr. Disparities in infant mortality: what’s genetics got to do with it? Am J Public Health. 2007;97:1191-1197.
Received 16 November 2015; accepted 12 December 2015
8. Siegel R, Naishadham D, Jemal A. Cancer statistics, 2013. CA Cancer J Clin. 2013;63:11-30. 9. Lloyd-Jones D, Adams RJ, Brown TM, Carnethon M, Dai S, De Simone G, et al. Executive summary: heart disease and stroke statisticse 2010 update: a report from the American Heart Association. Circulation. 2010;121:948-954.
18. Brown HL, Chireau MV, Jallah Y, Howard D. The “Hispanic paradox”: an investigation of racial disparity in pregnancy outcomes at a tertiary care medical center. Am J Obstet Gynecol. 2007;197: 197.e1-197.e7 [discussion: 197.e7-197.e9].
Journal homepage: www.WORLDNEUROSURGERY.org
410
www.SCIENCEDIRECT.com
Citation: World Neurosurg. (2016) 88:399-410. http://dx.doi.org/10.1016/j.wneu.2015.12.046
Available online: www.sciencedirect.com 1878-8750/$ - see front matter ª 2016 Elsevier Inc. All rights reserved.
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2015.12.046
Cross-Sectional Analysis on Racial and Economic Disparities Affecting Mortality in Preterm Infants with Posthemorrhagic Hydrocephalus Diana L. Jin1, Eisha A. Christian1, Frank Attenello1, Edward Melamed2, Steven Cen1, Mark D. Krieger1,2, J. Gordon McComb1,2, William J. Mack1
BACKGROUND: Despite major advances in medicine, racial and socioeconomic disparities continue to affect health care outcomes. Higher overall infant mortality has been reported for black neonates compared with their Hispanic and white counterparts. The underlying basis for these differences remains unclear. A potential influencing factor is the management of premature neurologic complications in this disadvantaged group. This study examines racial and socioeconomic disparities on mortality in preterm infants with posthemorrhagic hydrocephalus (PHH).
-
METHODS: Data from the Nationwide Inpatient Sample and Kids Inpatient Database were combined from 2000 to 2010. Discharges with International Classification of Diseases, Ninth Revision, Clinical Modification codes for preterm births with intraventricular hemorrhage and PHH were included. Relative risk (RR) ratios for mortality, complications, length of stay, and hospital costs were obtained with multivariate analysis after controlling for patient-level, hospital-level, and admission-level factors.
mortality compared with those with private insurance (RR [ 1.2; P [ 0.04) after adjusting for patient and hospital factors. CONCLUSIONS: Among preterm infants with intraventricular hemorrhage and resultant PHH, black infants and those insured by Medicaid have significantly increased mortality but these 2 effects are independent. Further studies are needed to fully understand the factors affecting these racial and socioeconomic disparities.
-
-
RESULTS: When controlling for patient and hospital factors, black neonates had increased mortality compared with whites and Hispanics (RR [ 1.47; P < 0.01). This association existed despite lower rates of congenital cardiac defects (RR [ 0.84; P < 0.01), gastrointestinal complications (RR [ 0.84; P < 0.01), and general complications of prematurity (RR [ 0.95; P [ 0.04) in the black cohort. Preterm infants insured by Medicaid had increased
-
Key words Hydrocephalus - Intraventricular hemorrhage - Kids Inpatient Database - Nationwide Inpatient Sample - Preterm neonate - Preterm infant - Posthemorrhagic hydrocephalus -
Abbreviations and Acronyms IMR: Infant mortality rate IVH: Intraventricular hemorrhage KID: Kids Inpatient Database NIS: Nationwide Inpatient Sample
WORLD NEUROSURGERY 88: 399-410, APRIL 2016
INTRODUCTION
I
nfant mortality has long been considered a sensitive indicator of the impact of disparities on the health of populations. The infant mortality rate (IMR, defined as the death of an infant per 1000 live births) in the United States has declined since 19601; however, the United States ranks 31st among the Organization for Economic Cooperation and Development countries, with an IMR of 6.2 (mean for the index is 4.2). Although the United States and Turkey have the highest preterm birth rates (12%), this characteristic alone does not sufficiently explain the relatively high IMR in the United States (e.g., Austria has a preterm birth rate of 11% but an IMR of 3.8).1 Racial and socioeconomic disparities have a profound impact on infant mortality in the United States. For example, the IMR ratio for black to white infants from 1960 to 2011 has remained unchanged, and the IMR of black infants is still double that of
PHH: Posthemorrhagic hydrocephalus RR: Relative risk From the 1Department of Neurosurgery, Keck School of Medicine of USC; and 2Division of Neurosurgery, Children’s Hospital Los Angeles, Los Angeles, California, USA To whom correspondence should be addressed: Diana L. Jin, B.A. [E-mail: [email protected]] Citation: World Neurosurg. (2016) 88:399-410. http://dx.doi.org/10.1016/j.wneu.2015.12.046 Journal homepage: www.WORLDNEUROSURGERY.org Available online: www.sciencedirect.com 1878-8750/$ - see front matter ª 2016 Elsevier Inc. All rights reserved.
www.WORLDNEUROSURGERY.org
399
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
Table 1. Continued
Table 1. Cohort Demographics: Patient Factors Variable
Category
Frequency
Percentage
13,736
100
Male
8028
58
Female
5706
Total patients Gender
Missing Race
*
42
3
4890
36
01
4
5651
41
Unspecified
1518
11
Hispanic
2070
15
240
2
86
1
821
6
3346
24 01
Medicare
12
Medicaid
7449
54
Private including HMO
5477
40
No charge Other
Gestational age
8
19
Self-pay
Birth weight
1127
2610
Missing Payer status
2
Black
Other
248
2 01
* 504
4
Missing
43
1
<500 g
197
1
500e749 g
3299
24
750e999 g
3699
27
1000e1249 g
2336
17
1250e1499 g
1526
11
1500e1749 g
789
6
1750e1999 g
583
5
2000e2499 g
614
5
>2500 g
271
2
Missing
305
2
<24 weeks
670
5
24 weeks
1095
8
25e26 weeks
2514
18
27e28 weeks
2111
15
29e30 weeks
1165
9
31e32 weeks
819
6
33e34 weeks
448
3
35e36 weeks Missing Unspecified
269
2
4620
34
26
01 Continues
400
www.SCIENCEDIRECT.com
Percentage 4
33
Native American
Frequency 549
4563
Intraventricular hemorrhage grade
Category 1
White
Asian/Pacific
Variable
HMO, Health Maintenance Organization. *Frequency 10.
white infants.1 The impact of insurance status, often used as a proxy for socioeconomic status, on IMRs has also been described: in 2011, the state of Washington reported that infants insured by Medicaid had an IMR of 6.5 versus 3.9 in neonates who did not receive Medicaid.2 Moss and Carver,3 in 1998, also noted that infants covered by Medicaid had higher odds of infant mortality. Disorders related to short gestational periods and low birth weight are among the leading causes of infant mortality in the United States. The management of neurologic complications secondary to premature birth can have a substantial impact on patient outcomes and survival. This study examines outcome disparities in the setting of intraventricular hemorrhage (IVH) and posthemorrhagic hydrocephalus (PHH) in preterm infants after controlling for both patient and hospital factors. METHODS Data Source This study uses 2 of the largest public all-payer inpatient care datasets in the United States4,5 from the Healthcare Cost and Utilization Project. The Nationwide Inpatient Sample (NIS) and Kids Inpatient Database (KID) have longitudinal hospital inpatient discharge data from more than 1000 hospitals, with the NIS representing 20% of all hospital discharges and KID representing 80% of pediatric discharges in the United States. Data from the KID were available for 2000, 2003, 2006, and 2009. For the years that the KID dataset did not cover, pediatric data were extracted from the 2000e2010 NIS database. NIS also has a built-in method for obtaining national estimates of prevalence as described in the Healthcare Cost and Utilization ProjecteNIS Calculating NIS Variances Guide.6 Study Cohort The study population was limited to preterm infants with IVH and hydrocephalus. Admission data were extracted using the International Classification of Diseases, Ninth Revision, Clinical Modification codes for patients with a preterm birth weight (765.00e765.29), IVH grade (772.10e772.14), and hydrocephalus (742.3, 331.3, 331.4). To prevent double-counting, admissions that resulted in a discharge to another short-term, acute-care facility
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2015.12.046
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
Table 2. Patient Demographics: Hospital Factors Variable
Category
Table 3. Complications and Comorbidities
Frequency Percentage Event
Hospital region
Teaching status
Northeast
1704
13
Midwest
3076
22
South
5759
42
West
3197
23
Nonteaching Teaching Missing
Hospital bed size Small
Hospital location
Admission type
11,155
81
366
3 11
Medium
3175
23
Large
8723
64
Missing
366
2
Rural
122
1
Missing
Discharge
16
1472
Urban
Hospital type
2215
Not children’s hospital
13,248
96
366
3
2068
15
Children’s general or specialty
1199
9
Children’s unit
1307
9
Missing
9161
67
Nonroutine
7140
52
Routine
6594
48
Emergency
1938
14
Urgent
3538
26
Elective
993
7
Newborn
5314
39
Trauma center Other Missing
28
01 01
* 1911
14
Admission source Emergency room
96
1
another hospital
4445
32
Other facility
309
2
Routine
4944
36
Missing
3942
29
*Frequency 10.
(pretransfer hospitalizations) were excluded from the study.7 In addition, age of admission was limited to less than 3 months. Both NIS and KID include patient and hospital factors as categorical demographic variables. Discrete patient factors, including race (white, black, Hispanic, Asian, or Pacific Islander, Native American, other), payer status (Medicare, Medicaid, private insurance, self-pay, no charge), gender, length of stay, cost of hospitalization, weight, gestational age, and IVH grade, were
WORLD NEUROSURGERY 88: 399-410, APRIL 2016
Frequency Percent
95% Confidence Interval
Shunt replacement
718
5
4.2e6.2
Mechanical implant complication
437
3
2.5e3.9
Shunt removal
857
6
5.2e7.3
Shunt infection
339
2.5
1.8e3.1
67
0.5
0.3e0.7
Other implant complication Neurologic complications Meningitis
42 839
Wound breakdown
56
Wound infection
0.3
0.1e0.5
6
5.2e7
0.4
0.2e0.6
118
0.9
0.5e1.2
Seizures
1997
14.5
12.9e16.2
Necrotizing enterocolitis
1368
10
Intestinal perforation
621
Pulmonary complication
11,437
Gastrointestinal complication Hematologic complication
621 7011
4.5 83 4.5 51
8.9e11.1 3.8e5.3 81.8e84.8 3.8e5.3 48.2e53.9
Sepsis
6547
48
45e50.3
Endocrinologic complication
4960
36
33.8e38.4
Cardiac complications
76
Congenital heart defects
5606
0.6 41
0.2e0.9 38.4e43.3
encoded as categorical variables in NIS and KID. Hospital-level variables such as hospital bed size (small, <200 beds; medium: 201e400 beds; large, >400 beds), region (Northeast, Midwest, South, West), location (urban and rural), type (children’s general/ specialty hospital, children’s unit in a general hospital, general hospital), and teaching status (teaching, nonteaching) were included as categorical variables. Admission status such as admission type (emergency, elective, urgent, trauma center, newborn) and admission source (emergency department, routine, another facility, another hospital) was also included. Because the NIS dataset did not classify hospitals based on the National Association of Children’s Hospitals and Related Institutions as did the KID database, hospitals in the NIS database were identified with their hospital identification ID number and were assigned a classification based on what their KID designation was for the previous available KID year dataset. The primary outcome of interest was mortality. Secondary outcomes of interest included general complications and shunt complications. Shunt complications of interest included replacement of a shunt (2.42), mechanical implant complication (996.2), implant infection (996.63), and other implant complication (996.75). General complications and comorbidities of interest included seizures (779.0), gastrointestinal complications (necrotizing enterocolitis, 777.5e777.53; intestinal perforation, 777.6), pulmonary complications (mechanical ventilation, 96.71, 96.72; respiratory support, 93.90e96.72; respiratory distress, 769,
www.WORLDNEUROSURGERY.org
401
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
Table 4. Continued
Table 4. Multivariate Analysis of Mortality Rate Ratio
95% Confidence Interval
Gender Female
1.18
Male
1.01e1.39
0.04
Reference
Private
1.20
1.01e1.44
0.04
Reference
Medium
0.87
0.61e1.24
0.43
Large
1.09
0.77e1.55
0.62
Midwest
0.85
0.61e1.17
0.32
South
0.87
0.66e1.15
0.33
West
0.98
0.68e1.40
0.90
0.42
0.34e0.52
< 0.01
P Value
Northeast
Race Asian/Pacific
1.62
0.92e2.83
0.09
Black
1.47
1.18e1.83
< 0.01
Hispanic
0.81
0.61e1.08
0.14
Native American
0.86
0.37e2.01
0.73
White
Reference
Weight <500 g
4.47
2.33e8.56
< 0.01
500e749 g
2.55
1.55e4.21
< 0.01
750e999 g
1.66
0.97e2.65
0.06
1000e1249 g
1.10
0.64e1.92
0.73
1250e1499 g
Reference
1500e1749 g
0.73
0.31e1.75
0.48
1750e1999 g
0.44
0.13e1.51
0.19
2000e2499 g
0.75
0.29e1.95
0.56
>2500 g
1.42
0.57e3.54
0.45
Gestational age <24 weeks
3.36
1.65e6.85
0.01
24 weeks
2.06
1.02e4.14
0.04
25e26 weeks
3.49
1.78e6.83
0.01
27e28 weeks
1.29
0.63e2.65
0.49
29e30 weeks
Reference
31e32 weeks
0.80
0.30e2.12
0.65
33e34 weeks
1.17
0.38e3.58
0.78
35e36 weeks
2.3
0.86e6.16
0.10
0.88
0.44e1.76
0.73
2
1.21
0.80e1.84
0.40
3
Reference
4
2.83
2.26e3.55
< 0.01
Intraventricular hemorrhage grade 1
Hospital bed size Small Continues
402
95% Confidence Interval
Hospital region
Payer Medicaid
Rate Ratio
P Value
www.SCIENCEDIRECT.com
Teaching status Teaching Nonteaching Hospital type Children’s unit
0.73
0.50e1.09
0.12
Not children’s hospital
0.81
0.58e1.15
0.25
Children’s hospital
Reference
Bold indicates statistically significant with P < 0.05.
770.6e770.8, 770.89), cardiac complications (cardiac arrest, 779.85; hypotension, 458.9; hypovolemia, 785.59), hematologic complications (disseminated intravascular coagulation, 776.2; jaundice, 774.2, 774.6; hematologic disorder, 776.0e776.9), sepsis (V29.0, 771.8, 771.81, 771.83, 771.89), endocrine complications (775.1e775.9), and congenital heart defects (745.0e747.9). Length of stay was also included as a dichotomized outcome: high length of stay was defined as greater than 120 days (90th percentile). Daily cost was dichotomized into high and low, with high daily cost being more than $12,654 (90th percentile). High total cost was defined as greater than $807,627 (90th percentile). For categorical variables, missing code was generated to represent missing data when fitting the model. Statistical Analysis The primary goal was to assess socioeconomic, demographic, and hospital factors associated with mortality. Outcomes of interest were 1) mortality; 2) general and specific complications as listed above; and 3) length and cost of stay. The predicting factors included patient-level factors, hospital-level factors, and admission statuselevel factors. Hospital location (urban vs. rural) was excluded as a predicting factor because nearly all patients went to urban hospitals. A univariate model (unadjusted model) was initially generated for all patient and hospital factors for the sample, after which a multivariate logistic regression model was created that adjusted for patient-level factors (race, payment type, comorbidity, sex, gestational age, birth weight, and IVH grade), hospital-level factors (region, teaching status, bed size, children’s hospital type), admission type (emergency, elective, urgent, trauma center,
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2015.12.046
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
Table 5. Continued
Table 5. Multivariate Analysis of General Complications Rate Ratio
95% Confidence Interval
Gender Female Male
1.00
0.97e1.04
0.80
Private
95% Confidence Interval
Medium
1.19
0.71e2.01
0.51
Large
1.48
0.87e2.52
0.15
1.28e4.41
< 0.01
Reference
Hospital region Midwest
2.38
1.03
Northeast
Reference
Payer Medicaid
Rate Ratio
P Value
0.99e1.07
0.15
Reference
Race
South
1.61
0.90e2.87
0.11
West
1.28
0.59e2.78
0.53
2.57
1.44e4.58
< 0.01
Asian/Pacific
0.99
0.87e1.12
0.84
Teaching status
Black
0.95
0.90e1.00
0.04
Teaching Nonteaching
Hispanic
0.97
0.91e1.02
0.23
Native American
0.94
0.74e1.20
0.63
White Weight <500 g
1.07
0.91e1.27
0.40
500e749 g
1.10
1.02e1.19
< 0.01
750e999 g
1.11
1.03e1.19
< 0.01
1000e1249 g
1.06
0.99e1.14
0.10
1250e1499 g
Reference
1500e1749 g
0.98
0.89e1.07
0.65
1750e1999 g
0.85
0.76e0.96
< 0.01
2000e2499 g
0.93
0.83e1.05
0.25
>2500 g
0.87
0.74e1.02
0.10
Gestational age <24 weeks
1.06
0.95e1.18
0.32
24 weeks
1.08
0.98e1.19
0.13
25e26 weeks
1.10
1.01e1.19
0.03
27e28 weeks
1.05
0.96e1.13
0.28
29e30 weeks
Reference
31e32 weeks
0.90
0.81e1.00
0.05
33e34 weeks
0.89
0.78e1.02
0.09
35e36 weeks
0.89
0.74e1.05
0.17
0.98
0.90e1.08
0.73
2
1.04
0.97e1.11
0.25
3
Reference
4
0.98
0.94e1.02
0.36
Intraventricular hemorrhage grade 1
Reference
Hospital type Children’s hospital
Reference
Hospital bed size Small Continues
WORLD NEUROSURGERY 88: 399-410, APRIL 2016
P Value
Reference
Children’s unit
0.99
0.56e1.75
0.97
Not children’s hospital
0.82
0.47e1.42
0.48
Bold indicates statistically significant with P < 0.05.
newborn), and admission source (emergency department, routine, another facility, another hospital). The association between predicting factors and outcome was assessed by the relative risk (RR) ratio, and statistical significance was predetermined as P < 0.05. All descriptive, univariate, and multivariate analyses were performed using SAS 9.4 (SAS Institute Inc., Cary, North Carolina, USA). RESULTS Patient Demographics A total of 13,736 patients had IVH and PHH. Patient and hospital demographics are presented in Tables 1 and 2. Most patients were white (33%), covered by Medicaid (54%), had a birth weight between 750 and 999 g (27%), had a gestational age of 25e26 weeks (18%), and presented with grade 4 IVH (41%). Most patients were admitted to urban, nonchildren’s teaching hospitals with a large bed size. Average length of stay for the entire cohort was 62 days, and average hospital cost $359,158 per admission. Mortality was 10.6%. Complications and comorbidities are presented in Table 3. The most common complications were pulmonary (83%) and hematologic (51%). Variables Affecting Inpatient Mortality Multivariable analyses examining associations between patient demographics/hospital-level factors and inpatient mortality are presented in Table 4. Black patients had a 47% increased risk of inpatient death compared with white patients (RR ¼ 1.47; P < 0.01). Medicaid patients had a 20% increased rate of
www.WORLDNEUROSURGERY.org
403
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
Table 6. Multivariate Analysis of Congenital Heart Anomalies Rate Ratio
95% Confidence Interval
Table 6. Continued
P Value
Rate Ratio
95% Confidence Interval
P Value
Hospital bed size
Gender Female
1.11
Male
1.02e1.22
0.02
Private
Reference
0.93
0.78e1.11
0.43
Large
1.01
0.85e1.21
0.90
1.03
Hospital region Midwest
1.02
0.86e1.22
0.78
South
1.02
0.88e1.19
0.81
West
0.96
0.79e1.18
0.71
1.01e1.32
0.04
0.94e1.14
0.50
Reference
Race Asian/Pacific
1.00
0.73e1.38
1.00
Black
0.84
0.73e0.95
< 0.01
Northeast
Hispanic
0.91
0.79e1.05
0.22
Teaching status
Native American
1.02
0.57e1.82
0.95
Teaching
White
Reference
Medium
Payer Medicaid
Small
Nonteaching
Reference
Reference
1.15 Reference
Hospital type
Weight <500 g
1.31
0.88e1.96
0.18
Children’s unit
0.92
0.76e1.12
0.43
0.95
0.8e1.14
0.57
500e749 g
1.20
0.98e1.48
0.08
Not children’s hospital
750e999 g
1.20
0.99e1.45
0.07
Children’s hospital
1000e1249 g
1.19
0.98e1.45
0.07
1250e1499 g
Reference
1500e1749 g
1.07
0.83e1.38
0.58
1750e1999 g
0.87
0.62e1.20
0.39
2000e2499 g
0.96
0.69e1.34
0.82
>2500 g
1.07
0.70e1.63
0.76
<24 weeks
1.5
1.13e1.98
< 0.01
24 weeks
1.42
1.11e1.82
< 0.01
25e26 weeks
1.43
1.15e1.78
< 0.01
27e28 weeks
1.23
1.00e1.52
0.05
29e30 weeks
Reference
Reference
Bold indicates statistically significant with P < 0.05.
inpatient mortality compared with those with private insurance (RR ¼ 1.20; P ¼ 0.04). Female gender, weight <750 g, gestational age <27 weeks, and patients with grade 4 IVH had increased inpatient mortality. Teaching hospitals had lower mortality compared with nonteaching hospitals.
Gestational age
31e32 weeks
0.88
0.67e1.17
0.38
33e34 weeks
0.86
0.59e1.26
0.44
35e36 weeks
1.09
0.7e1.69
0.70
1.04
0.81e1.33
0.75
2
1.17
0.99e1.38
0.07
3
Reference
4
1.04
0.94e1.15
0.46
Intraventricular hemorrhage grade 1
Continues
404
www.SCIENCEDIRECT.com
Variables Affecting Perinatal Complications Multivariable analyses examining associations between patient demographics/hospital-level factors and perinatal complications are presented in Tables 5e8. Compared with their white counterparts, black infants had a lower rate of general complications (RR ¼ 0.95; P ¼ 0.04, Table 4), fewer cardiac anomalies (RR ¼ 0.84; P < 0.01), and fewer gastrointestinal complications (RR ¼ 0.59; P ¼ 0.01). Medicaid patients had a higher likelihood of overall shunt complications (RR ¼ 1.16; P ¼ 0.02). Gestational age less than 26 weeks was associated with an increased rate of cardiac defects, and birth weight less than 1000 g was associated with an increased risk of gastrointestinal complications, including necrotizing enterocolitis and intestinal perforation. Length of Stay Multivariable analyses examining associations between patient demographics/hospital-level factors and length of stay are presented in Table 9. Blacks and Asians had increased length of stay compared with their white counterparts (RR ¼ 1.37; P ¼ 0.02 and
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2015.12.046
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
Table 7. Multivariate Analysis of Gastrointestinal Complications Rate Ratio
95% Confidence Interval
Male
0.63
0.47e0.84
< 0.01
Reference
Private
1.06
95% Confidence Interval
Medium
1.19
0.71e2.01
0.51
Large
1.48
0.87e2.52
0.15
Midwest
2.38
1.28e4.41
< 0.01
South
1.61
0.90e2.87
0.11
West
1.28
0.59e2.78
0.53
1.44e4.58
< 0.01
0.79e1.42
0.68
Reference
Northeast
Race Asian/Pacific
0.91
0.33e2.53
0.85
Teaching status
Black
0.59
0.39e0.90
0.01
Teaching
Hispanic
0.86
0.55e1.35
0.52
Native American White
0.73
P Value
Hospital region
Payer Medicaid
Rate Ratio
P Value
Gender Female
Table 7. Continued
0.10e5.36
0.75
Reference
Weight <500 g
5.23
1.42e19.33
0.01
500e749 g
4.93
2.04e11.89
< 0.01
750e999 g
3.82
1.60e9.15
< 0.01
1000e1249 g
2.47
1.00e6.12
0.05
1250e1499 g
Reference
1500e1749 g
0.28
0.03e2.35
0.24
1750e1999 g
0.45
0.05e3.86
0.46
2000e2499 g
N/A
N/A
N/A
>2500 g
N/A
N/A
N/A
<24 weeks
1.92
0.59e6.28
0.28
24 weeks
2.03
0.66e6.23
0.21
25e26 weeks
1.67
0.57e4.94
0.35
27e28 weeks
1.02
0.33e3.16
0.97
29e30 weeks
Reference
Reference
2.57
Nonteaching
Reference
Hospital type Children’s unit
0.99
0.56e1.75
0.97
Not children’s hospital
0.82
0.47e1.42
0.48
Children’s hospital
Reference
Bold indicates statistically significant with P < 0.05.
RR ¼ 2.21; P < 0.01, respectively). Furthermore, Medicaid patients had increased length of stay compared with private insurance patients (RR ¼ 1.38; P < 0.01). Patients with grade 4 IVH had longer lengths of stay and patients at teaching hospitals had longer lengths of stay compared with patients at nonteaching hospitals.
Gestational age
31e32 weeks
1.30
0.23e7.31
0.76
33e34 weeks
1.89
0.20e17.50
0.57
35e36 weeks
N/A
N/A
N/A
Intraventricular hemorrhage grade 1
0.56
0.17e1.78
0.32
2
0.60
0.30e1.21
0.16
3
Reference
4
1.02
0.74e1.39
0.92
Hospital bed size Small
Reference Continues
WORLD NEUROSURGERY 88: 399-410, APRIL 2016
Hospital Cost Multivariable analyses examining associations between patient demographics/hospital-level factors and high daily and total hospital costs were also conducted, presented in Tables 10 and 11. Fewer black patients had high daily costs compared with their white counterparts (RR ¼ 0.69; P ¼ 0.02). Yet, more Hispanic patients had higher daily costs than did white patients (RR ¼ 1.33; P ¼ 0.03). Fewer female patients (RR ¼ 0.76; P < 0.01) and medium bed size hospitals (RR ¼ 0.51; P < 0.01) had higher costs than did male patients and small bed size hospitals, respectively. Patients in the Midwest (RR ¼ 0.39; P < 0.01) and South (RR ¼ 0.51; P < 0.01), and teaching hospitals (RR ¼ 0.65; P < 0.01) had less high daily costs than did patients in the Northeast and nonteaching hospitals, respectively. DISCUSSION Racial and socioeconomic factors affect health care delivery in the United States despite widespread efforts to reduce disparities. Minorities and economically disadvantaged groups continue to experience inferior outcomes across an array of diseases and diagnoses, including cancer,8 heart disease,9 stroke,9 and
www.WORLDNEUROSURGERY.org
405
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
Table 8. Multivariate Analysis of Overall Shunt Complications Rate Ratio
95% Confidence Interval
Table 8. Continued
P Value
Rate Ratio
Female
1.05
Male
0.93e1.18
0.41
Medicaid Private
Small
Reference
0.63
0.51e0.78
< 0.01
Large
0.94
0.77e1.16
0.59
1.16
Hospital region Midwest
1.07
0.86e1.33
0.56
South
0.92
0.75e1.12
0.39
West
1.49
1.16e1.92
< 0.01
1.59e2.49
< 0.01
0.75
0.59e0.94
0.01
0.48
0.38e0.60
< 0.01
1.03e1.32
0.02
Reference
Race Asian/Pacific
0.74
0.46e1.20
0.23
Black
0.99
0.84e1.17
0.91
Northeast
Hispanic
0.96
0.79e1.16
0.67
Teaching status
Native American
0.76
0.31e1.86
0.55
Teaching
White
Reference
Medium
Payer
Nonteaching
Reference
Reference
1.99 Reference
Hospital type
Weight <500 g
0.74
0.35e1.54
0.42
500e749 g
1.41
1.08e1.83
0.01
750e999 g
1.49
1.16e1.91
< 0.01
1000e1249 g
1.23
0.95e1.59
0.12
1250e1499 g
Reference
1500e1749 g
1.52
1.12e2.07
< 0.01
1750e1999 g
0.95
0.61e1.46
0.80
2000e2499 g
0.89
0.55e1.41
0.61
>2500 g
1.21
0.70e2.09
0.49
Gestational age <24 weeks
0.98
0.68e1.42
0.93
24 weeks
0.76
0.55e1.05
0.10
25e26 weeks
1.01
0.77e1.32
0.97
27e28 weeks
0.81
0.62e1.06
0.13
29e30 weeks
Reference
31e32 weeks
0.61
0.42e0.90
0.01
33e34 weeks
0.80
0.49e1.31
0.38
35e36 weeks
0.57
0.29e1.12
0.10
0.65
0.43e0.99
0.05
2
0.57
0.43e0.77
< 0.01
3
Reference
4
1.31
1.15e1.49
< 0.01
Intraventricular hemorrhage grade
Continues
406
P Value
Hospital bed size
Gender
1
95% Confidence Interval
www.SCIENCEDIRECT.com
Children’s unit Not children’s hospital Children’s hospital
Reference
Bold indicates statistically significant with P < 0.05.
diabetes.10 Epidemiologic studies have noted that African American women exhibit a 6% lower incidence of cancer than white women, but a 16% higher mortality as a result of the disease.8 Researchers have suggested multiple factors that may affect these outcomes, including availability of proper nutrition, health insurance coverage, income, education level, access to disease screening, and physicianepatient interactions.1,11 Elimination of these disparities has been designated a major goal in Healthy People 2010.12 Most studies of health care disparities focus on the adult population, but these inequalities have also extended into the pediatric realm. African American children have decreased access to care, substantially lower adherence to preventive health measures, and decreased quality in delivery of health care.13 Black children are less likely than white children to be selected from the kidney transplant waiting list and are more likely to receive inadequate hemodialysis.13 These disparities have also been documented in the perinatal literature. El-Sayed et al.14 recently published a study examining all births in Michigan between 1989 and 2005. These investigators found that socioeconomic factors and maternal demographic, behavioral, and health access differences explained approximately 30% of disparities in infant mortality in disorders relating to preterm birth. Willis et al.1 reported an IMR of 11.4 per 1000 live births for black infants and an IMR of
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2015.12.046
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
Table 9. Continued
Table 9. Multivariate Analysis of High Length of Stay Rate Ratio
95% Confidence Interval
Gender Female Male
0.93
0.77e1.11
0.43
Reference
Private
1.38
95% Confidence Interval
Medium
0.84
0.60e1.17
0.29
Large
1.11
0.79e1.55
0.55
1.13e1.69
< 0.01
Reference
Midwest
1.48
1.04e2.1
0.03
South
1.20
0.87e1.65
0.27
West
1.17
0.75e1.83
0.50
1.15e2.25
< 0.01
Northeast
Race 2.21
1.27e3.86
< 0.01
Black
1.37
1.06e1.76
0.02
Teaching Nonteaching
Hispanic
1.25
0.91e1.71
0.17
Native American
1.02
0.32e3.24
0.98
Reference
Weight <500 g
12.85
3.93e42.07
< 0.01
500e749 g
14.92
5.41e41.13
< 0.01
750e999 g
10.63
3.86e29.27
< 0.01
5.35
1.89e15.17
< 0.01
1000e1249 g 1250e1499 g
Reference
1500e1749 g
1.78
0.44e7.14
0.42
1750e1999 g
0.71
0.08e6.46
0.76
2000e2499 g
2.05
0.43e9.74
0.37
>2500 g
1.15
0.12e11.09
0.90
<24 weeks
1.83
0.85e3.93
0.12
24 weeks
2.23
1.07e4.64
0.03
25e26 weeks
1.15
0.55e2.37
0.71
0.62
0.28e1.37
0.24
Gestational age
27e28 weeks 29e30 weeks
Reference
31e32 weeks
0.67
0.18e2.52
0.55
33e34 weeks
0.42
0.05e3.47
0.42
35e36 weeks
1.40
0.27e7.33
0.69
1.20
0.66e2.18
0.56
1.15
0.79e1.67
0.48
1.02e1.58
0.03
Intraventricular hemorrhage grade 1 2 3 4
Reference 1.27
Hospital bed size Small
Reference Continues
WORLD NEUROSURGERY 88: 399-410, APRIL 2016
Reference
Teaching status
Asian/Pacific
White
P Value
Hospital region
Payer Medicaid
Rate Ratio
P Value
1.61 Reference
Hospital type Children’s unit
0.67
0.46e0.96
0.03
Not children’s hospital
0.46
0.32e0.65
< 0.01
Children’s hospital
Reference
Bold indicates statistically significant with P < 0.05.
5.1 per 1000 live white births in 2011. Furthermore, multiple studies examining U.S.- and African-born infants have demonstrated that genetic differences alone fail to explain the disparity.15-17 When accounting for traditional confounders such as payer status, gestational age, birth weight, and IVH grade, our model demonstrates that black preterm infants with PHH have a higher mortality and increased length of hospital stay compared with their white counterparts. Paradoxically, we demonstrate that black infants with PHH have a lower rate of overall general comorbidities and complications related to prematurity such as congenital cardiac anomalies and gastrointestinal complications, including necrotizing enterocolitis and intestinal perforation. Despite these lower overall comorbidities and complication rates, black PHH infants had a 47% higher rate of mortality than had their white counterparts. By contrast, our Hispanic cohort did not demonstrate an increased risk of mortality. Our findings comport with an existing perinatal literature that describes the Hispanic Paradox.18 Gould et al.19 reported that despite a highrisk profile for Hispanic mothers (because of lack of access to prenatal care), the population did not have an increased rate of low-birth-weight infants and neonatal mortality. In this report, both black and Hispanic mothers had decreased access to prenatal healthcare and less maternal education, yet black neonates had worse outcomes than had Hispanic neonates. One potential explanation for this paradox is that Hispanic mothers are subject to lower incidences of both preterm and low-birth-weight infants. Our study cohort, for this reason, is specifically limited to low-birth-weight, preterm infants who develop IVH and PHH.
www.WORLDNEUROSURGERY.org
407
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
Table 10. Continued
Table 10. Multivariate Analysis of High Daily Cost Rate Ratio
95% Confidence Interval
Rate Ratio
95% Confidence Interval
P Value
Midwest
0.39
0.25e0.59
< 0.01
South
0.51
0.36e0.72
< 0.01
0.75
0.49e1.15
0.19
0.5e0.83
< 0.01
P Value Hospital region
Gender Female
1.00
Male
0.82e1.21
0.97
Reference
West
Payer Medicaid Private
0.87
0.7e1.07
0.18
Northeast
Reference
Teaching status
Reference
Teaching
Race
0.65
Nonteaching
Reference
Asian/Pacific
1.11
0.64e1.93
0.70
Black
0.69
0.50e0.94
0.02
Hispanic
1.33
1.03e1.72
0.03
Children’s unit
1.44
0.97e2.15
0.07
Native American
0.89
0.22e3.66
0.87
Not children’s hospital
0.87
0.61e1.24
0.44
White
Hospital type
Children’s hospital
Reference
Reference
Bold indicates statistically significant with P < 0.05.
Weight <500 g
2.81
1.31e6.03
< 0.01
500e749 g
2.11
1.30e3.43
< 0.01
750e999 g
1.45
0.91e2.31
0.12
1000e1249 g
1.31
0.82e2.11
0.26
1250e1499 g
Reference
1500e1749 g
1.18
0.65e2.16
0.58
1750e1999 g
1.00
0.47e2.14
1.00
2000e2499 g
0.57
0.23e1.40
0.22
>2500 g
0.93
0.37e2.34
0.89
Gestational age <24 weeks
1.62
0.88e2.97
0.12
24 weeks
1.64
0.94e2.86
0.08
25e26 weeks
1.64
0.98e2.75
0.06
27e28 weeks
1.18
0.71e1.96
0.53
29e30 weeks
Reference
31e32 weeks
0.86
0.43e1.73
0.68
33e34 weeks
1.21
0.52e2.82
0.65
35e36 weeks
3.40
1.51e7.66
< 0.01
0.80
0.45e1.41
0.44
2
0.64
0.42e0.98
0.04
3
Reference
4
1.24
1.00e1.53
0.05
Even within this selected subgroup, black infants demonstrate higher mortality. We also demonstrate that patients covered by Medicaid have an increased mortality and length of stay compared with those with private insurance after controlling for relevant confounders. This finding is consistent with results in previous socioeconomic studies in the pediatric neurosurgical population. Attenello et al.20 demonstrated that pediatric patients insured by Medicaid were 12% less likely to have routine discharges after cerebrospinal fluid shunting procedures compared with patients with private insurance. Additional studies have demonstrated that pediatric patients of lower socioeconomic status have longer hospital lengths of stay and higher infection rates.21,22 The principal limitations to this study result from the use of a large population-based administrative database. International Classification of Diseases, Ninth Revision, Clinical Modification codes are entered by hospital staff with varied training and
Intraventricular hemorrhage grade 1
Hospital bed size Small
Reference
Medium
0.94
0.64e1.38
0.77
Large
0.87
0.59e1.28
0.49 Continues
408
www.SCIENCEDIRECT.com
Figure 1. Admission frequency.
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2015.12.046
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
Table 11. Continued
Table 11. Multivariate Analysis of High Total Cost Rate Ratio
95% Confidence Interval
P Value
Male
0.76
0.63e0.93
< 0.01
Private
Small
0.51
0.36e0.73
< 0.01
Large
0.74
0.52e1.05
0.09
0.95
Hospital region Midwest
1.07
0.74e1.53
0.73
South
0.89
0.64e1.24
0.50
West
1.51
1.00e2.28
0.05
0.95e1.75
0.10
0.82
0.55e1.22
0.32
0.54
0.37e0.78
< 0.01
0.77e1.16
0.59
Reference
Asian/Pacific
0.89
0.46e1.71
0.73
Black
1.06
0.81e1.37
0.68
Northeast
Hispanic
1.34
1.00e1.78
0.05
Teaching status
Native American
1.54
0.62e3.86
0.35
Teaching
Reference
1.29
Nonteaching
Reference
Reference
Hospital type
Weight <500 g
2.08
0.78e5.51
500e749 g
3.35
1.90e5.9
0.14
Children’s unit
< 0.01
Not children’s hospital Children’s hospital
750e999 g
2.75
1.57e4.81
< 0.01
1000e1249 g
1.52
0.84e2.74
0.17
1250e1499 g
Reference
1500e1749 g
0.44
0.15e1.33
0.14
1750e1999 g
0.59
0.17e2.07
0.41
2000e2499 g
0.42
0.11e1.62
0.21
>2500 g
0.58
0.14e2.40
0.45
Gestational age <24 weeks
1.48
0.77e2.84
0.24
24 weeks
1.53
0.83e2.82
0.17
25e26 weeks
1.25
0.70e2.25
0.46
27e28 weeks
0.74
0.40e1.37
0.33
29e30 weeks
Reference
31e32 weeks
0.24
0.05e1.06
0.06
33e34 weeks
0.90
0.24e3.31
0.87
35e36 weeks
2.45
0.80e7.54
0.12
0.83
0.43e1.59
0.58
2
1.22
0.85e1.77
0.28
3
Reference
4
1.24
1.00e1.54
0.05
Reference
Bold indicates statistically significant with P < 0.05.
oversight. The potential for recording bias results in estimates of coding accuracy of approximately 80%.23 This bias can affect not only coding of diagnoses but also covariates and complications. In addition, these data sets capture only a single admission, and therefore patients who develop delayed hydrocephalus are not included. Although we used previously published methods of preventing double-counting,7 this phenomenon cannot be completely eliminated because of the lack of individual identifiers. CONCLUSIONS
Intraventricular hemorrhage grade 1
Reference
Reference
Race
White
P Value
Medium
Payer Medicaid
95% Confidence Interval
Hospital bed size
Gender Female
Rate Ratio
Continues
WORLD NEUROSURGERY 88: 399-410, APRIL 2016
The United States continues to have a high IMR compared with other Organization for Economic Cooperation and Development countries. The presence of socioeconomic disparities in health care delivery may contribute to this increased mortality. Our data, after multivariable analyses accounting for both patient and hospital factors, suggest that black preterm neonates with PHH have increased mortality despite lower overall complications of prematurity. Increased mortality is also seen in preterm neonates with PHH who have Medicaid insurance coverage independent of their racial status. Further studies are needed to understand the extent of these differences and potential causative factors.
www.WORLDNEUROSURGERY.org
409
ORIGINAL ARTICLE DIANA L. JIN ET AL.
FACTORS AFFECTING DEATH IN NEONATES WITH PHH
10. Hedderson M, Ehrlich S, Sridhar S, Darbinian J, Moore S, Ferrara A. Racial/ethnic disparities in the prevalence of gestational diabetes mellitus by BMI. Diabetes Care. 2012;35:1492-1498.
19. Gould JB, Madan A, Qin C, Chavez G. Perinatal outcomes in two dissimilar immigrant populations in the United States: a dual epidemiologic paradox. Pediatrics. 2003;111:e676-682.
2. Wasserman C, Taylor P. Infant Mortality. Washington, DC: Washington State Department of Health; 2013.
11. Smedley BD, Stith AY, Nelson AR, eds. Unequal Treatment: Confonting Racial and Ethnic Disparities in Healthcare. Washington, DC: The National Academies Press; 2003.
20. Attenello FJ, Ng A, Wen T, Cen SY, Sanossian N, Amar AP, et al. Racial and socioeconomic disparities in outcomes following pediatric cerebrospinal fluid shunt procedures. J Neurosurg Pediatr. 2015;15:560-566.
3. Moss N, Carver K. The effect of WIC and Medicaid on infant mortality in the United States. Am J Public Health. 1998;88:1354-1361.
12. National Center for Health Statistics. Healthy People 2010 Final Review. Hyattsville: US Government Printing Office, MD. 2012.
4. Nationwide Inpatient Sample. Agency for Healthcare Research and Quality; 2007. Available at: www. hcup-us.ahrq.gov/databases.jsp. Accessed June 20, 2014.
13. Flores G, Committee On Pediatric Research. Technical reporteracial and ethnic disparities in the health and health care of children. Pediatrics. 2010;125:e979-e1020.
5. Introduction to the KID. Healthcare Cost and Utilization Project (HCUP). 2015. Available at: www.hcup-us.ahrq.gov/db/nation/kid/kid_2012_ introduction.jsp. Accessed February 20, 2015.
14. El-Sayed AM, Finkton DW Jr, Paczkowski M, Keyes KM, Galea S. Socioeconomic position, health behaviors, and racial disparities in causespecific infant mortality in Michigan, USA. Prev Med. 2015;76:8-13.
6. Final Report on Calculating Nationwide Inpatient Sample (NIS) Variances, 2001. U.S. Agency for Healthcare Research and Quality; 2005. Available at: http://www.hcup-us.ahrq.gov/reports/ methods/CalculatingNISVariances200106092005.pdf. Accessed December 10, 2014.
15. David RJ, Collins JW Jr. Differing birth weight among infants of U.S.-born blacks, African-born blacks, and U.S.-born whites. N Engl J Med. 1997; 337:1209-1214.
REFERENCES 1. Willis E, McManus P, Magallanes N, Johnson S, Majnik A. Conquering racial disparities in perinatal outcomes. Clin Perinatol. 2014;41:847-875.
7. Burke BL, Robbins JM, Bird TM, Hobbs CA, Nesmith C, Tilford JM. Trends in hospitalizations for neonatal jaundice and kernicterus in the United States, 1988-2005. Pediatrics. 2009;123: 524-532.
16. Collins JW Jr, Wu SY, David RJ. Differing intergenerational birth weights among the descendants of US-born and foreign-born Whites and African Americans in Illinois. Am J Epidemiol. 2002;155:210-216.
21. Walker CT, Stone JJ, Jain M, Jacobson M, Phillips V, Silberstein HJ. The effects of socioeconomic status and race on pediatric neurosurgical shunting. Childs Nerv Syst. 2014;30:117-122. 22. Simon TD, Hall M, Riva-Cambrin J, Albert JE, Jeffries HE, Lafleur B, et al. Infection rates following initial cerebrospinal fluid shunt placement across pediatric hospitals in the United States. J Neurosurg Pediatr. 2009;4:156-165. 23. Burns EM, Rigby E, Mamidanna R, Bottle A, Aylin P, Ziprin P, et al. Systematic review of discharge coding accuracy. J Public Health (Oxf). 2012;34:138-148.
Conflict of interest statement: The authors declare that the article content was composed in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
17. David R, Collins J Jr. Disparities in infant mortality: what’s genetics got to do with it? Am J Public Health. 2007;97:1191-1197.
Received 16 November 2015; accepted 12 December 2015
8. Siegel R, Naishadham D, Jemal A. Cancer statistics, 2013. CA Cancer J Clin. 2013;63:11-30. 9. Lloyd-Jones D, Adams RJ, Brown TM, Carnethon M, Dai S, De Simone G, et al. Executive summary: heart disease and stroke statisticse 2010 update: a report from the American Heart Association. Circulation. 2010;121:948-954.
18. Brown HL, Chireau MV, Jallah Y, Howard D. The “Hispanic paradox”: an investigation of racial disparity in pregnancy outcomes at a tertiary care medical center. Am J Obstet Gynecol. 2007;197: 197.e1-197.e7 [discussion: 197.e7-197.e9].
Journal homepage: www.WORLDNEUROSURGERY.org
410
www.SCIENCEDIRECT.com
Citation: World Neurosurg. (2016) 88:399-410. http://dx.doi.org/10.1016/j.wneu.2015.12.046
Available online: www.sciencedirect.com 1878-8750/$ - see front matter ª 2016 Elsevier Inc. All rights reserved.
WORLD NEUROSURGERY, http://dx.doi.org/10.1016/j.wneu.2015.12.046