Abstract
Background:
Nonaccidental trauma (NAT) affects >100,000 children in the United States every year and is associated with significant mortality and morbidity. Little is known about the financial burden of NAT, particularly in comparison to accidental trauma (AT). We sought to compare hospital charges and outcomes between children presenting with NAT and AT.
Methods:
Pediatric (<16 y) trauma hospitalizations from 2006 to 2018 were identified using the Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS) and Kid’s Inpatient Sample (KID) databases. Hospitalizations were identified as NAT or AT based on ICD codes. Discharge weights were used to obtain national estimates and standardize them across the different sampling structures. Outcomes (hospital charges, length of stay (LOS), and mortality) were compared, and multivariate regression analyses were used to assess independent predictors of hospital charges and mortality.
Results:
Fifty-eight Thousand Two Hundred Seventy-five pediatric hospitalizations were included with 17,954 (0.3%) categorized as NAT. Children with NAT were younger, more female, less likely to identify as White, and more under public insurance than those with AT. Hospital charges were significantly higher in patients with NAT ($27,100 versus $19,900, P < 0.0001). Mortality (4.9% versus 0.0%, P < 0.0001) and LOS (3.2 d versus 1.5 d, P < 0.0001) were significantly higher among patients with NAT. Multivariable regression analyses identified NAT as a predictor of higher hospital charges, mortality, and LOS.
Conclusions:
Nonaccidental trauma in pediatric patients is associated with significantly higher hospital charges, mortality, and LOS than accidental trauma. Ongoing research focused on the relative impact of known risk factors and resource utilization is needed.
Keywords: Child abuse, Cost analysis, Nonaccidental trauma, Pediatric trauma
Introduction
Traumatic injury is the leading cause of pediatric morbidity and mortality in the United States, with more than 100,000 children affected by nonaccidental trauma (NAT), now commonly referred to as child physical abuse, each year1,2 Although the number of emergency department visits for NAT has slightly decreased over the years, the yearly percentage of patients requiring hospitalization has increased, suggesting that children are presenting with more severe injuries.3 Similarly, the annual national estimate of child fatalities due to abuse and neglect continues to rise. In 2019, there were an estimated 1840 children who died from child abuse, an increase from 1710 children in 2017 and 1660 children in 2015.2 Known risk factors for child abuse include caregiver alcohol abuse, domestic violence, drug abuse, financial problems, inadequate housing, public assistance, and any caregiver disability.2
Pediatric traumas represent a significant burden to the health care system. A recent review of the Healthcare Cost and Utilization Project (HCUP) databases found that hospital-treated nonfatal injuries for patients aged 0–17 y in 2013 cost an estimated $405 billion dollars, including $22 billion in medical spending. In addition, total costs per injury were highest among patients aged less than 1 y at nearly $98,000 per injury.4 Although previous studies have shown that children who sustain NAT injuries typically present with more severe polytrauma and have increased length of stay (LOS) and mortality, less is known about how the cost of NAT compares to that of accidental trauma (AT).5–9 A recent study of pediatric traumas at a level 1 pediatric trauma center found the median initial hospital cost for NAT patients was higher than that of accidental trauma patients.7 In addition, there was a significant direct correlation between injury severity score and total costs for child abuse patients.7 Similarly, Ettaro et al. found significantly higher mean hospital charges for children who suffered abusive head trauma versus nonabusive head trauma.10 Despite these findings, very few studies have evaluated and compared costs associated with pediatric NAT, relative to accidental traumas, on a national level. Furthermore, these studies have been restricted to specific injuries such as abusive head, abdominal, or sexual trauma.11–15
We, therefore, sought to use a nationally representative database to compare hospital charges and outcomes associated with initial trauma hospitalization between NAT and AT patients. We hypothesized that hospitalizations for NAT were associated with significantly higher hospital charges than those associated with AT. We also sought to evaluate factors that were independently associated with hospital LOS, charges, and mortality.
Methods
Data source and study population
Hospitalizations of children(<16 y old) between January 1, 2006, and December 31, 2018, were identified using the HCUP National Inpatient Sample (NIS) and Kid’s Inpatient Sample (KID) databases. The KID is the largest publicly available all-payer health pediatric inpatient care database in the United States, including approximately 3 million pediatric discharges each year (roughly 7 million hospitalizations, weighted).16 The KID is currently available every 3 y. Years in this study include 2006, 2009, 2012, and 2016 (2015 was not produced due to ICD-9-CM to ICD-10-CM/PCS transition). Hospitalizations for the remaining years were identified in the NIS, the largest publicly available all-payer database in the United States and includes roughly 7 million hospitalizations (35 million weighted hospitalizations) from hospitals in the United States.17 Discharge weights are provided by HCUP and account for changes in sampling over time and the stratified sampling approach. In both databases, discharge weights were used to obtain national estimates and standardize them across the different sampling structures.
Eligible hospitalizations included children admitted due to nonaccidental or accidental trauma identified by ICD-9-CM and ICD-10-CM codes (Supplemental Table 1).
Exposure, outcomes and covariates
Nonaccidental trauma included trauma caused by child physical abuse/maltreatment, unspecified, or other child abuse and neglect or shaken infant syndrome. Accidental trauma was defined as trauma from injuries, unarmed fight or brawl, drowning, assault by firearm, sharp object, or by other/unspecified means. (See Supplemental Table 1 for ICD-9 and ICD-10 codes and descriptions). A hospitalization was coded as nonaccidental trauma if the hospitalization had both a nonaccidental trauma and accidental trauma diagnosis.
Outcomes of interest included mortality, hospital charges in thousands, hospital length of stay in days (LOS), and discharge disposition. Hospital charges were inflation-adjusted to 2018 dollars. Discharge disposition was defined as discharge of the patient to home, transfer to a short-term hospital or another type of facility, or discharge to home health care.
Covariates included the year of admission, patient race/ethnicity, age atadmission, sex, income, number of procedures, hospital region, type, and bed size. Age was assessed as a categorical variable with 4 categories: less than or equal to 1 y, greater than 1 y, less than or equal to 5 y, greater than 5 y and less than or equal to 10 y, and greater than 10 y and less than, or equal to 16 y. Patients were categorized in this manner given that children under the age of 5 y, especially infants, are at a significantly increased risk of childphysical abuse.18,19 Children above the age of 5 y were grouped in 5-year increments. Patient race/ethnicity was reported as White non-Hispanic, Black non-Hispanic, Hispanic, Asian or Pacific Islander, Native American, or other. Sex was reported as male or female. Non-female or male or sex that was inconsistent with diagnoses or procedures were set by HCUP to missing. The number of procedures was defined as the number of ICD-9/ICD-10 procedures on the discharge. Hospital region was defined as South, West, Midwest, or Northeast. Hospital type was defined as rural, urban nonteaching, or urban teaching. Hospital bed size was defined as small, medium, or large with assignment based on the hospital’s location and teaching status.
Statistical analyses
Wald chi-square tests were used in bivariate analyses to assess significant differences (P < 0.05) in covariates of interest between nonaccidental and accidental trauma hospitalizations. Multivariable linear regression was used to assess the impact of the type of trauma (nonaccidental versus accidental) on mean hospital charges and LOS after adjusting for covariates of interest. Multivariable logistic regression was used to assess associations between type of trauma on mortality after adjusting for covariates of interest. Regression models were run against a population of all identified hospitalizations as designed above and against a subset of this population consisting of those hospitalizations of children of 5 y of age and younger. The trend analysis for mean hospital charges was assessed by univariate linear regression.
All analyses were performed using SAS 9.4 (SAS Inc, Cary, NC). The complex sampling design and weighting were accounted for in all analyses; cluster variables included year, database (KID versus NIS), and hospital ID. The sampling stratum was assigned by HCUP [HCUP databases. HCUP. August 2018. Agency for Health care Research and Quality R, MD. In.]. Weighted results are reported unless otherwise noted. The study was deemed exempt by the University of North Carolina Institutional Review Board as it contained only deidentified data (IRB#20–1493).
Results
Study population
A total of 582,727 encounters were identified for this study (Fig. 1). Table 1 compares demographic and baseline characteristics for the study sample. The median age of the entire cohort was 6.2 y (IQR; 1.2–11.6 y). The majority of patients were male (61.3%) and identified as non-Hispanic White (53.3%). Most patients were covered by medicare or medicaid (47.7%) and private insurance (43.6%). In addition, most patients (80.3%) were treated at large, urban teaching hospitals and there was a higher percentage of patients in the Midwest and Southern United States who experienced NAT. The majority of hospitalizations were associated with injuries that were associated with a low risk of mortality (89.0%).
Fig. 1 –
Cohort selection. Figure 1 demonstrates the selection of the study population using ICD-9-CM and ICD-10-CM codes.
Table 1 –
Demographic characteristics of pediatric patients presenting with NAT and AT, in the HCUP NIS and KID databases between 2006 and 2018.
| Characteristic | Overall |
NAT |
AT |
P-value |
|---|---|---|---|---|
| N = 582,727 | N = 17,954 (3.1%) | N = 564,773 (96.9%) | ||
|
| ||||
| Age at admission, y, median (IQR) | 6.2 (1.2, 11.6) | 0.0 (0.0,1.9) | 6.4 (1.4, 11.7) | <0.0001 |
| Age group, y, n (%) | <0.0001 | |||
| ≤1 | 131,059 (22.7%) | 12,186 (68.1%) | 118,873 (21.2%) | |
| >1-≤5 | 127,801 (21.9%) | 3174 (17.7%) | 124,627 (22.0%) | |
| >5-≤10 | 128,254 (21.9%) | 1044 (5.7%) | 127,210 (22.5%) | |
| >10-<16 | 195,613 (33.5%) | 1550 (8.5%) | 194,063 (34.3%) | |
| Sex, n (%) | <0.0001 | |||
| Female | 222,008 (38.7%) | 7567 (41.7%) | 214,441 (38.4%) | |
| Male | 354,837 (61.3%) | 10,382 (58.3%) | 344,455 (61.6%) | |
| Missing | 5882 | |||
| Race/Ethnicity, n (%) | <0.0001 | |||
| White | 262,790 (53.3%) | 7308 (48.1%) | 255,482 (53.5%) | |
| Black | 79,973 (16.4%) | 3755 (24.5%) | 76,218 (16.2%) | |
| Hispanic | 103,904 (21.0%) | 2797 (18.6%) | 101,107 (21.0%) | |
| Asian/Pacific Islander | 13,009 (2.7%) | 203 (1.4%) | 12,806 (2.7%) | |
| Native American | 5092 (1.0%) | 208 (1.4%) | 4884 (1.0%) | |
| Other | 27,484 (5.6%) | 911 (6.0%) | 26,573 (5.6%) | |
| Missing | 90,475 | |||
| Insurance, n (%) | <0.0001 | |||
| Medicare/Medicaid | 273,080 (47.7%) | 14,044 (78.5%) | 259,036 (46.6%) | |
| Private | 255,384 (43.6%) | 2670 (15.1%) | 252,714 (44.5%) | |
| Other/Self-pay | 51,984 (8.7%) | 1184 (6.4%) | 50,800 (8.9%) | |
| Missing | 2279 | |||
| Hospital location, n (%) | <0.0001 | |||
| Rural | 27,619 (5.0%) | 622 (3.7%) | 26,997 (5.0%) | |
| Urban nonteaching | 90,844 (14.7%) | 1208 (6.1%) | 89,636 (15.1%) | |
| Urban teaching | 451,756 (80.3%) | 15,656 (90.2%) | 436,100 (79.9%) | |
| Missing | 12,508 | |||
| Region, n (%) | <0.0001 | |||
| Northeast | 101,916 (17.4%) | 2544 (13.9%) | 99,372 (17.5%) | |
| Midwest | 128,409 (21.6%) | 4824 (26.0%) | 123,585 (21.4%) | |
| South | 209,795 (37.5%) | 7190 (41.6%) | 202,605 (37.4%) | |
| West | 142,607 (23.5%) | 3396 (18.5%) | 139,211 (23.7%) | |
| Number of procedures*, median (IQR) | 0.5 (0.0, 1.5) | 0.0 (0.0, 1.9) | 0.5 (0.0, 1.5) | 0.0002 |
| Risk of mortality, n (%) | <0.0001 | |||
| Minor | 95,347 (89.0%) | 1925 (64.3%) | 93,422 (89.7%) | |
| Moderate | 6113 (5.7%) | 427 (14.5%) | 5686 (5.4%) | |
| Major | 3398 (3.2%) | 252 (8.5%) | 3146 (3.0%) | |
| Extreme | 2354 (2.1%) | 378 (12.7%) | 1976 (1.9%) | |
| Missing | 475,515 | |||
NAT = non-accidental trauma; AT = accidental trauma; NIS=Nationwide Inpatient Sample; KID=Kids’ Inpatient Database; IQR = interquartile range.
Number of procedures were procedures that occurred on discharge record at time of discharge.
Comparison of characteristics of entire cohort
Of the patients included in the study, 17,954 (3.1%) were categorized as NAT, while 564,773 (96.9%) were categorized as AT (Table 1). Patients with NAT were younger than those with AT (median: 0 versus 6.4 y), with 68.1% of patients with NAT aged less than 1 y, compared to 21.2% of patients with AT. A higher proportion of NAT patients were female (41.7% versus 38.6%), identified as Black (24.5% versus 16.2%), or Native American (1.4% versus 1.0%), with medicare or medicaid insurance (78.5% versus 46.6%), and treated at urban teaching hospitals (90.2% versus 79.9%). Patients in the NAT group also had more injuries associated with an extreme risk of mortality (Table 1).
Hospital charges and outcomes of entire cohort
Table 2 illustrates the comparison of hospital charges and outcomes between NAT and AT groups. NAT patients had a significantly longer hospital LOS than AT patients (median: 3.2 d versus 1.5 d; P < 0.0001). Mortality was significantly higher (4.9% versus 0.0%; P < 0.0001) and the median total hospital charges associated with initial hospitalization were significantly higher in children with NAT ($27,100 versus $19,900; P < 0.0001) than those with AT.
Table 2 –
Hospital charges and outcomes of pediatric patients presenting with NAT and AT HCUP NIS and KID databases, between 2006 and 2018.
| Variable‡ | NAT | AT | P-value |
|---|---|---|---|
|
| |||
| LOS, d, median (IQR) | 3.2 (1.4, 7.7) | 1.5 (0.6, 3.3) | <0.0001 |
| Disposition, n (%) | <0.0001 | ||
| Home | 15,013 (88.6%) | 520,281 (93.1%) | |
| Home health | 556 (3.3%) | 17,093 (3.0%) | |
| Transfer | 1395 (8.1%) | 21,615 (3.9%) | |
| In-hospital death*, n (%) | 881 (4.9%) | 4915 (0.0%) | <0.0001 |
| Hospital charges†, thousands, median (IQR) | 27.1 (13.0, 68.4) | 19.9 (10.4, 39.2) | <0.0001 |
NAT = non-accidental trauma; AT = accidental trauma; NIS=Nationwide Inpatient Sample; KID=Kids’ Inpatient Database; LOS = length of stay; IQR = interquartile range.
229 individuals were excluded who were missing information on in-hospital death.
Hospital charges adjusted for inflation to 2018.
6774 individuals were excluded who were missing information on disposition.
Characteristics of NAT versus at–age 0-≤5 y
Table 3 demonstrates the comparison of demographics and baseline characteristics for patients aged 0–5 y as part of our secondary analysis. Of the 258,860 children aged 0–5 y, 15,360 (5.9%) had NAT and 234,500 (94.1%) had AT. Overall, this age group was slightly more female, less white, had more patients who had medicare/medicaid insurance and had more injuries resulting in a major or extreme risk of dying compared to the entire cohort. We found there were similar trends in the characteristics of patients in the NAT and AT groups.
Table 3 –
Demographic characteristics of pediatric patients, between 0 and 5 y of age, presenting with NAT and AT, in the HCUP NIS and KID databases between 2006 and 2018.
| Overall |
NAT |
AT |
P-value | |
|---|---|---|---|---|
| N = 258,860 | N = 15,360 (5.9%) | N = 243,500 (94.1%) | ||
|
| ||||
| Age at admission, y, median (IQR) | 0.9 (0.0, 2.5) | 0.0 (0.0, 0.7) | 1.0 (0.0, 2.6) | <0.0001 |
| Age group, y, n (%) | <0.0001 | |||
| ≤1 | 131,059 (50.9%) | 12,186 (79.3%) | 118,873 (49.0%) | |
| >1-≤5 | 127,801 (49.1%) | 3174 (20.7%) | 124,627 (51.0%) | |
| Sex, n (%) | 0.027 | |||
| Female | 107,881 (41.7%) | 6291 (40.6%) | 101,590 (41.8%) | |
| Male | 150,188 (58.2%) | 9066 (59.4%) | 141,122 (58.2%) | |
| Race/Ethnicity, n (%) | <0.0001 | |||
| White | 106,800 (48.5%) | 6291 (48.4%) | 100,509 (48.5%) | |
| Black | 37,794 (17.4%) | 3134 (24.0%) | 34,660 (17.0%) | |
| Hispanic | 52,555 (23.5%) | 2434 (18.8%) | 50,121 (23.8%) | |
| Asian or Pacific Islander | 6615 (3.0%) | 176 (1.4%) | 6439 (3.1%) | |
| Native American | 2418 (1.1%) | 175 (1.4%) | 2243 (1.1%) | |
| Other | 14,289 (6.5%) | 775 (6.0%) | 13,514 (6.5%) | |
| Missing | 38,389 | |||
| Insurance, n (%) | <0.0001 | |||
| Medicare/Medicaid | 145,500 (56.9%) | 12,116 (79.2%) | 133,384 (55.4%) | |
| Private | 90,593 (34.8%) | 2167 (14.3%) | 88,426 (36.2%) | |
| Other/Self-pay | 21,848 (8.3%) | 1030 (6.5%) | 20,818 (8.4%) | |
| Missing | 919 | |||
| Hospital location, n (%) | <0.0001 | |||
| Rural | 10,298 (4.1%) | 362 (2.5%) | 9936 (4.3%) | |
| Urban nonteaching | 35,708 (12.9%) | 989 (5.7%) | 34,719 (13.3%) | |
| Urban teaching | 207,205 (83.0%) | 13,574 (91.8%) | 193,631 (82.4%) | |
| Missing | 5649 | |||
| Region, n (%) | <0.0001 | |||
| Northeast | 44,501 (17.0%) | 2242 (14.3%) | 42,259 (17.2%) | |
| Midwest | 56,021 (21.3%) | 4239 (26.7%) | 51,782 (20.9%) | |
| South | 95,098 (38.1%) | 5827 (39.6%) | 89,271 (38.0%) | |
| West | 63,240 (23.6%) | 3052 (19.4%) | 60,188 (23.9%) | |
| Number of procedures*, median (IQR) | 0.4 (0.0, 1.5) | 0.2 (0.0, 2.2) | 0.4 (0.0, 1.5) | <0.0001 |
| Risk of mortality | <0.0001 | |||
| Minor | 41,035 (87.0%) | 1608 (62.2%) | 39,427 (88.5%) | |
| Moderate | 2940 (6.2%) | 356 (14.0%) | 2584 (5.7%) | |
| Major | 1832 (3.9%) | 236 (9.3%) | 1596 (3.6%) | |
| Extreme | 1357 (2.9%) | 373 (14.5%) | 984 (2.2%) | |
| Missing | 211,696 | |||
NAT = nonaccidental trauma; AT = accidental trauma; NIS=Nationwide Inpatient Sample; KID=Kids’ Inpatient Database; IQR = interquartile range.
Number of procedures were procedures that occurred on discharge record at time of discharge.
Hospital charges and outcomes of NAT versus at–age 0-≤5y
Table 4 compares outcomes in children between 0 and 5 y of age. NAT patients had significantly longer median hospital LOS than AT patients (3.0 d versus 1.4 d, P < 0.001). Similar to the entire cohort, mortality was significantly higher in NAT patients compared to AT patients (5.6% versus 1.1%; P < 0.001). NAT hospitalizations were associated with significantly higher initial hospital charges (median: $28,500 versus $16,600; P < 0.001).
Table 4 –
Hospital charges and outcomes of pediatric patients 0–5 y old, presenting with NAT and AT trauma, in the HCUP NIS and KID databases between 2006 and 2018.
| NAT | AT | P-value | |
|---|---|---|---|
|
| |||
| LOS, d, median (IQR) | 3.0 (1.3, 7.6) | 1.4 (0.5, 3.0) | <0.0001 |
| Disposition*, n (%) | <0.0001 | ||
| Home | 12,789 (88.8%) | 225,147 (93.6%) | |
| Home health | 503 (7.8%) | 8124 (3.4%) | |
| Transfer | 1120 (3.4%) | 7276 (3.0%) | |
| In-hospital death†, n (%) | 868 (5.6%) | 2605 (1.1%) | <0.0001 |
| Hospital charges‡, thousands, median (IQR) | 28.5 (13.7, 73.3) | 16.6 (8.3, 34.3) | <0.0001 |
LOS = lenght of stay; IQR = interquartile range.
3901 individuals were excluded who were missing information on disposition.
229 individuals were excluded who were missing information on in-hospital death.
Hospital charges adjusted for inflation to 2018.
Regression analyses
Multivariable linear regression analysis of LOS (Table 5) identified NAT (ß: 2.83; P < 0.0001), age< 1 and number of procedures as significant predictors of LOS.
Table 5 –
Multivariable linear regression model for predictors of length of stay in pediatric patients presenting with NAT and AT, in the HCUP NIS and KID databases between 2006 and 2018.
| Overall |
|||
|---|---|---|---|
| Coefficient | (95% CI) | P | |
|
| |||
| NAT (ref = AT) | 2.83 | (1.79, 3.86) | <0.0001 |
| Age group (ref ≥10-<16yo) | |||
| ≤1 | 1.32 | (1.10, 1.54) | <0.0001 |
| >1-≤5 | −0.94 | (−1.07, −0.82) | <0.0001 |
| >5-≤10 | −0.83 | (−0.92, −0.74) | <0.0001 |
| Number of procedures* | 2.02 | (1.95, 2.09) | <0.0001 |
| Race/Ethnicity (ref = white)† | |||
| Black | 0.01 | (−0.12, 0.14) | 0.8616 |
| Hispanic | −0.19 | (−0.29, −0.09) | 0.0004 |
| Asian or Pacific Islander | 0.02 | (−0.17, 0.21) | 0.8581 |
| Native American | 0.62 | (0.18, 1.06) | 0.0057 |
| Other | 0.12 | (−0.10, 0.33) | 0.2905 |
| Insurance (ref = private)‡ | |||
| Medicare/Medicaid | 0.75 | (0.65, 0.85) | <0.0001 |
| Other/Self-pay | 0.06 | (−0.08, 0.20) | 0.4127 |
| Hospital location (ref = urban teaching)§ | |||
| Rural | −0.14 | (−0.83, 0.54) | 0.6864 |
| Urban nonteaching | 0.05 | (−0.12, 0.21) | 0.5610 |
CI = confidence interval; NAT = non-accidental trauma; AT = accidental trauma.
Number of procedures were procedures that occurred on discharge record at time of discharge.
90,475 individuals were excluded who were missing information on race/ethnicity.
2279 individuals were excluded who were missing information on insurance type.
12,508 individuals were excluded who were missing information on region.
NAT (OR 4.09, 95% CI 3.59–4.65), age less than 1 and number of procedures were significant predictors of mortality, as well as self-pay insurance status (Table 6).
Table 6 –
Multivariable logistic regression model for predictors of mortality for pediatric patients with NAT and AT, in the HCUP NIS and KID databases between 2006 and 2018.
| Overall |
|||
|---|---|---|---|
| OR | (95% CI) | P | |
|
| |||
| NAT (ref = AT) | 4.09 | (3.59, 4.65) | <0.0001 |
| Age group (ref ≥10-<16yo) | |||
| ≤1 | 1.50 | (1.35, 1.67) | <0.0001 |
| >1-≤5 | 1.14 | (1.02, 1.28) | 0.0258 |
| >5-≤10 | 0.69 | (0.61, 0.79) | <0.0001 |
| Number of procedures* | 1.25 | (1.24, 1.26) | <0.0001 |
| Race/Ethnicity (ref = white)† | |||
| Black | 1.07 | (0.96, 1.20) | 0.2156 |
| Hispanic | 0.82 | (0.73, 0.92) | 0.0005 |
| Asian or Pacific Islander | 0.89 | (0.68, 1.16) | 0.3713 |
| Native American | 0.86 | (0.59, 1.26) | 0.4445 |
| Other | 1.20 | (1.02, 1.41) | 0.0325 |
| Insurance (ref = private)‡ | |||
| Medicare/Medicaid | 0.96 | (0.87, 1.06) | 0.4106 |
| Other/Self-pay | 1.71 | (1.50, 1.96) | <0.0001 |
| Hospital location (ref = urban teaching) § | |||
| Rural | 0.28 | (0.18, 0.42) | <0.0001 |
| Urban nonteaching | 0.40 | (0.34, 0.48) | <0.0001 |
CI = confidence interval; NAT = non-accidental trauma; AT = accidental trauma.
Number of procedures were procedures that occurred on discharge record at time of discharge.
90,475 individuals were excluded who were missing information on race/ethnicity.
2279 individuals were excluded who were missing information on insurance type.
12,508 individuals were excluded who were missing information on region.
Table 7 demonstrates the linear regression model for hospital charges. On average NAT was associated with US$ 11,750 (95% CI US$ 8600-US$ 15,000) higher cost compared to AT cases. Age less than 1 y was associated with higher hospital charges ($9040; 95% CI: US$ 6740–11,340), and a unit increase in the number of procedures was associated with higher hospital charges (US$ 30,260; 95% CI: US$ 28,630-US$ 31,900). Black or Hispanic race/ethnicity were associated with higher hospital charges, as well as holding nonprivate insurance. Treatment at a rural hospital was a predictor of significantly less cost (US$ —7560; 95% CI: US$ —10,000 to US$ —5120).
Table 7 –
Multivariable linear regression model for predictors of hospital charges for pediatric patients, presenting with NAT and AT in the HCUP NIS and KID databases between 2006 and 2018.
| Overall |
|||
|---|---|---|---|
| Coefficient* | (95% CI) | P | |
|
| |||
| NAT (ref = AT) | 11.75 | (8.6, 15.0) | <0.0001 |
| Age group (ref ≥10-<16yo) | |||
| ≤1 | 9.04 | (6.74, 11.34) | <0.0001 |
| >1-≤5 | −6.05 | (−7.18, −4.92) | <0.0001 |
| >5-≤10 | −5.54 | (−6.54, −4.55) | <0.0001 |
| Number of procedures† | 30.26 | (28.63, 31.90) | <0.0001 |
| Race/Ethnicity (ref = white)‡ | |||
| Black | 2.04 | (0.37, 3.71) | 0.0165 |
| Hispanic | 4.38 | (2.73, 6.03) | <0.0001 |
| Asian or Pacific Islander | 3.13 | (0.06, 6.03) | 0.0542 |
| Native American | −2.65 | (−7.52, 2.22) | 0.2868 |
| Other | 2.57 | (0.65, 4.50) | 0.0089 |
| Insurance (ref = private)§ | |||
| Medicare/Medicaid | 2.18 | (1.13, 4.68) | <0.0001 |
| Other/Self-pay | 2.98 | (1.29, 4.68) | 0.0006 |
| Hospital location (ref = urban teaching)‖ | |||
| Rural | −7.56 | (−10.00, −5.12) | <0.0001 |
| Urban nonteaching | −0.16 | (−2.21, 1.90) | 0.8811 |
CI = confidence interval; NAT = Non-Accidental trauma; AT = Accidental trauma.
Coefficients are thousands of US$.
Number of procedures were procedures that occurred on discharge record at time of discharge.
90,475 individuals were excluded who were missing information on race/ethnicity.
2279 individuals were excluded who were missing information on insurance type.
12,508 individuals were excluded who were missing information on region.
Trend analysis of hospital charges
Figure 2 illustrates a trend of initial hospital charges over the duration of the study period (2006–2018) demonstrating an increase in hospital charges for both NAT and AT groups. Initial hospital charges associated with NAT were consistently higher than those for AT each year (P < 0.001).
Fig. 2 –
Trend analysis of hospital charges for NAT and AT in HCUP NIS and KID databases, from 2006 to 2018.
Discussion
Pediatric trauma is the leading cause of morbidity and mortality in children. Although nonaccidental traumas are a small proportion, they account for significant morbidity and mortality. Several studies have demonstrated the significant financial burden associated with pediatric trauma and NAT. However, there is a paucity of data evaluating and comparing the cost of NAT to AT in the pediatric population. In this study, we found that pediatric patients with NAT had significantly higher initial hospital charges compared to those with AT. This finding was consistent over the study duration. In addition, patients with NAT had longer hospital LOS and higher mortality.6,20 NAT was found to be an independent predictor of hospital charges, LOS, and mortality rates.
Overall, the demographics of the current study population were similar to other studies evaluating NAT. We found that NAT patients were significantly younger than those with AT. This was similar to other studies demonstrating that patients with NAT are significantly younger than those with AT.6–9,14 Furthermore, we found that patients who identified as Black or Native American were disproportionately represented in the NAT group.2,7,8,12,14 The majority of NAT patients were covered by medicare or medicaid-a finding that corroborates findings in other studies.7,10,14 Children in the Midwest and the Southern United States were also disproportionately represented in the NAT group.4 Although the reasons for the geographic differences in the incidence of NAT remain unclear, we believe that it is important to further evaluate and identify factors contributing to this geographic disparity in future studies. In total, these findings seem to illustrate the impact of social determinants of health in pediatric traumas and suggest there may be demographic factors that are associated with NAT compared to AT. Although the KID and HCUP databases do not have measures of injury severity, we found that there was a significantly greater proportion of NAT patients with injuries resulting in major or extreme risk of dying. This finding is similar to other studies that have shown significantly higher injury severity in NAT patients compared to those with AT.7,14 In the current study, we found that patients with NAT had worse outcomes and increased health care resource utilization than those with AT, as evidenced by significantly higher mortality and longer hospital LOS. This finding is similar to other studies that have shown longer hospital lengths of stay in NAT patients.6,7,9,10,14 It is likely that the increased hospital LOS in NAT patients is related to factors including more severe injuries, multisystem injuries requiring various specialists, and issues related to disposition. We suspect disposition contributes significantly as it may be necessary to identify a safe place to discharge NAT patients, thus prolonging hospital admission. Although the mortality rate in NAT patients in the current study was slightly lower than that reported in previous studies, we identified similar trends with higher mortality in NAT patients.6–9,14 It is likely that this lower mortality rate is related to our selection criteria as we did not include patients who died prior to arrival or in the emergency department, and our sample encompasses all types of physical NAT as opposed to specifically abdominal or head trauma.
A novel aspect of the current study is the comparison of hospital charges associated with NAT and AT. We found that hospitalizations associated with NAT had significantly higher charges than those for AT. This finding is similar to a previous review of the KID database for abusive abdominal trauma in which the authors estimated that the median total hospital charges for abusive abdominal trauma were $28,570 compared to $17,911 for patients with noninflicted injury.14 Although a direct comparison of exact hospital charges across studies is difficult due to variability in reporting charges, inflation, and patient selection, it is evident from this study and others that abusive trauma care is associated with higher costs. The higher costs associated with NAT are likely due to the more severe multisystem injuries and longer lengths of stay children with NAT experience. Additionally, given the increased awareness regarding child physical abuse, it is likely that some of the costs associated with NAT are related to the diagnostic workup that is often necessary to rule out other injuries when there is a suspicion of child physical abuse. Furthermore, the cost of caring for children with NAT injuries is not limited to the initial hospitalization. A study by Peterson et al. using the Truven Health Market Scan data from 2003 to 2011 determined the estimated total medical cost related to abusive head trauma in the 4 y following diagnosis averaged nearly $48,000, estimating the need for health care resources and services that go beyond the initial hospitalization.12 The majority of these costs were attributed to continued outpatient visits and prescription drug needs, which does not account for costs related to the legal services, special education, loss of work, quality of life, and long-term social and psychological effects many child survivors and caregivers/families experience.11
In our regression analysis, we found that NAT, age <1, number of procedures, Black and Hispanic race and ethnicity, and nonprivate insurance were all independent predictors of significantly higher hospital charges. It is likely that NAT, young age, and the number of procedures are correlated to injury severity. A previous study by Shahi et al. found that injury severity was significantly correlated with median initial hospital charges [r = 0.40 (0.33, 0.47); P < 0.001], while there was no association between injury severity score and charges in patients with AT.7 Similarly, in a study using KID data from 2006 to 2012 comparing NAT to other types of child maltreatment, the authors found that in NAT patients, non-White race and ethnicity, age less than 7 y, and medicaid insurance were independent predictors of higher treatment cost.15 Though insurance status is often believed to be a surrogate for socioeconomic status and associated with child abuse, indirect measurements of income including median household income by zip code were found to be negative predictors of increased hospital charges in NAT patients in that study.15 Interestingly, we found that treatment at a rural hospital was associated with significantly lower charges on regression analysis. Although it is difficult to ascertain factors that are related to this finding, it is possible that the lower charges seen in rural hospitals could be attributed to transfers to larger, urban hospitals. We speculate that rural hospitals may not have the resources to care for complex, pediatric traumas that would be associated with higher hospital charges. As such, those complex patients are transferred to urban facilities where they can be further managed.
As highlighted, there are several factors suspected to be drivers of health care costs associated with NAT. This points to the need for more research on the relative impact of risk factors for NAT and how hospital resources are utilized in treating NAT patients. According to the United States Prevention Services Task Force, not all primary prevention strategies are equally effective at preventing child abuse, particularly across diverse populations.21 Studies that help delineate the influence of insurance coverage as opposed to household income on the incidence of NAT, or that follow healthcare dollars spent on medical treatment as opposed to in-hospital social services, for example, can help us better determine how to allocate prevention and postresponse resources.
Despite the findings of this study, the results are limited by its retrospective nature and the use of an administrative database. Given the administrative nature of the HCUP and KID databases, it is likely that our results are subject to information bias, variability in coding and data input, and missing variables that may have affected the determination of NAT versus AT status. Moreover, since the databases are not trauma databases, it is likely that not all pediatric traumas were captured based on our coding. We were also unable to obtain data regarding important diagnostic tests, procedures performed, comorbidities, and commonly used markers of injury severity, which were not available in these databases. These aforementioned factors are significant as they limit our ability to obtain information regarding patients who were initially thought to have NAT and then found to have AT and vice-versa. It also limits the ability to understand clinical factors that informed the final diagnoses that were reported.
Another significant limitation of the study relates to our inability to ascertain the true relationship between hospital LOS and NAT, and its ultimate impact on health care charges. In the current study, we found that patients with NAT had significantly longer hospital LOS than those with AT. We surmised that this increase in LOS is partly due to the need for a safe disposition plan for NAT patients. In patients with suspected or confirmed cases of NAT, health care providers are required to inform child protective services or social services. The determination of the safest place to discharge the patient in such instances is based on child protective services investigation and discretion. It is, therefore, not uncommon to have instances in which a child with NAT is medically cleared, but has to remain in the hospital pending a safe discharge plan. In those instances, hospital charges, despite the absence of medical needs, continue to be incurred until the patient is discharged. Given that the current study is based on data from an administrative database, we are unable to ascertain the proportion of patients in whom the need for a safe disposition plan prolonged the hospital LOS and therefore, increased hospital charges, as compared to NAT patients with prolonged LOS due to medical needs. We acknowledge this as a limitation of the current database given its administrative nature, which may be addressed with prospective studies.
Given the limitations highlighted above, there is a need for multiinstitutional, prospective studies to evaluate the cost of NAT in comparison to AT. Such studies will likely be able to address more details than can be elucidated from an administrative database. Despite these limitations, we demonstrated findings consistent with the larger body of evidence that shows poorer outcomes and higher hospital charges for NAT patients.
Conclusions
In this study, utilizing an administrative, national database, we found that nonaccidental trauma is associated with significantly longer hospital LOS, higher mortality rate, and higher hospital charges compared to accidental trauma. Standardized child abuse screening implementation, as well as ongoing research comparing the relative impact of known risk factors on the incidence of nonaccidental trauma and resource utilization, are needed to reduce the devastating consequences of this public health problem.
Supplementary Material
Funding
This work was supported by funding from the National Institutes of Health (UNC Integrated Translational Oncology Program T32-CA244125 to UNC/yourinitials).
Footnotes
Study Type
Retrospective study.
Level of Evidence
Level III.
Supplementary Materials
Supplementary data related to this article can be found at https://doi.org/10.1016/j.jss.2022.08.045.
Disclosure
None declared.
Meeting Presentation
Seventeenth Annual Academic Surgical Congress.
references
- 1.Borse N, Sleet DA. CDC childhood injury report: patterns of unintentional injuries among 0- to 19-year olds in the United States, 2000–2006. Fam Community Health. 2009;32:189. [DOI] [PubMed] [Google Scholar]
- 2.U.S. Department of Health & Human Services, Administration for Children and Families, Administration on Children, Youth and Families, Children’s Bureau. (2021). Child Maltreat; 2019. Available at: https://www.acf.hhs.gov/cb/data-research/child-maltreatment. Accessed November 25, 2022.
- 3.Swedo E, Idaikkadar N, Leemis R, et al. Trends in U.S. Emergency department visits related to suspected or confirmed child abuse and neglect among children and adolescents aged <18 years before and during the COVID-19 pandemic - United States, January 2019-September 2020. MMWR Morb Mortal Wkly Rep. 2020;69:1841–1847. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Zonfrillo MR, Spicer RS, Lawrence BA, Miller TR. Incidence and costs of injuries to children and adults in the United States. Inj Epidemiol. 2018;5:37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Naik-Mathuria B, Akinkuotu A, Wesson D. Role of the surgeon in non-accidental trauma. Pediatr Surg Int. 2015;31:605–610. [DOI] [PubMed] [Google Scholar]
- 6.Livingston JK, Grigorian A, Kuza CM, et al. Non-accidental trauma increases length of stay and mortality in pediatric trauma. Pediatr Surg Int. 2019;35:779–784. [DOI] [PubMed] [Google Scholar]
- 7.Shahi N, Phillips R, Meier M, et al. The true cost of child abuse at a level 1 pediatric trauma center. J Pediatr Surg. 2020;55:335e340. [DOI] [PubMed] [Google Scholar]
- 8.Estroff JM, Foglia RP, Fuchs JR. A comparison of accidental and nonaccidental trauma: it is worse than you think. J Emerg Med. 2015;48:274–279. [DOI] [PubMed] [Google Scholar]
- 9.DiScala C, Sege R, Li G, Reece RM. Child abuse and unintentional injuries: a 10-year retrospective. Arch Pediatr Adolesc Med. 2000;154:16–22. [PubMed] [Google Scholar]
- 10.Ettaro L, Berger RP, Songer T. Abusive head trauma in young children: characteristics and medical charges in a hospitalized population. Child Abuse Negl. 2004;28:1099–1111. [DOI] [PubMed] [Google Scholar]
- 11.Miller TR, Steinbeigle R, Lawrence BA, et al. Lifetime cost of abusive head trauma at ages 0–4, USA. Prev Sci. 2018;19:695–704. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Peterson C, Xu L, Florence C, et al. The medical cost of abusive head trauma in the United States. Pediatrics. 2014;134:91–99. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Letourneau EJ, Brown DS, Fang X, Hassan A, Mercy JA. The economic burden of child sexual abuse in the United States. Child Abuse Negl. 2018;79:413–422. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Lane WG, Lotwin I, Dubowitz H, Langenberg P, Dischinger P. Outcomes for children hospitalized with abusive versus noninflicted abdominal trauma. Pediatrics. 2011;127:e1400–e1405. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Nunez Lopez O, Hughes BD, Adhikari D, Williams K, Radhakrishnan RS, Bowen-Jallow KA. Sociodemographic determinants of non-accidental traumatic injuries in children. Am J Surg. 2018;215:1037–1041. [DOI] [PubMed] [Google Scholar]
- 16.Health care cost and utilization Project (HCUP). Available at: https://www.ahrq.gov/data/hcup/index.html. Accessed November 25, 2022.
- 17.Healthcare cost and utilization Project (HCUP). KID overview. Available at: https://www.hcup-us.ahrq.gov/kidoverview.jsp. Accessed February 9, 2021.
- 18.Centers for Disease Control and Prevention. Child Abuse and Neglect: Risk and Protective Factors. Available at: https://www.cdc.gov/violenceprevention/childabuseandneglect/riskprotectivefactors.html. Accessed February 9, 2021.
- 19.Jones RE, Babb J, Gee KM, Beres AL. An investigation of social determinants of health and outcomes in pediatric nonaccidental trauma. Pediatr Surg Int. 2019;35:869–877. [DOI] [PubMed] [Google Scholar]
- 20.Litz CN, Amankwah EK, Danielson PD, Chandler NM. Implications of non-accidental trauma on resource utilization and outcomes. Pediatr Surg Int. 2018;34:635–639. [DOI] [PubMed] [Google Scholar]
- 21.U. S. Preventive Services Task Force, Curry SJ, Krist AH, Owens DK, et al. Interventions to prevent child maltreatment: US preventive services Task Force recommendation statement. JAMA. 2018;320:2122–2128. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.