National Risk Factors for Child Maltreatment after Trauma: Failure to Prevent

Abstract

The purpose of this study was to identify the risk factors for hospital readmission for child maltreatment after trauma, including admissions across different hospitals nationwide. The Nationwide Readmissions Database for 2010–2014 was queried for all patients younger than 18 years admitted for trauma. The primary outcome was readmission for child maltreatment. The secondary outcome was readmission for maltreatment presenting to a hospital different than the index admission hospital. A subgroup analysis was performed on patients without a diagnosis of maltreatment during the index admission. Multivariable logistic regression was performed for each outcome. There were 608,744 admissions identified and 44,569 (7.32%) involved maltreatment at the index admission. Readmission for maltreatment was found in 1,948 (0.32%) patients and 368 (18.89%) presented to a different hospital. The highest risk for readmission for maltreatment was found in patients with maltreatment identified at the index admission (odds ratios (OR) 9.48 [8.35–10.76]). The strongest risk factor for presentation to a different hospital was found with the lowest median household income quartile (OR 3.50 [2.63–4.67]). The subgroup analysis identified 647 (0.11%) children with readmission for maltreatment that was missed during the index admission. The strongest risk factor for this outcome was Injury Severity Score > 15 (OR 3.29 [2.68–4.03]). This study demonstrates that a significant portion of admissions for trauma in children and teenagers could be misrepresented as not involving maltreatment. These index admissions could be the only chance for intervention for child maltreatment. Identifying these at-risk individuals is critical to prevention efforts.

Despite increasing efforts at surveillance and prevention,^1–4 over the last few decades, the epidemic of child maltreatment has been worsening. In 1998, there were 3.1 child deaths per day due to abuse and neglect, in 2010 that number rose to 5.⁵ The estimated number of child protective service investigations has also been increasing. Between 2012 and 2016, the number rose by 9.5 per cent, from 3.2 to 3.5 million per year.⁶ This could be explained by the difficulty in identifying child maltreatment. The clinical indicators can vary widely and hidden child abuse may present with no risk factors.⁷

Victims of child abuse often show evidence of previous injuries in addition to those first detected. Successfully identifying abuse from the pattern of these injuries has been widely reported,^8–10 yet abuse has been initially missed in 20–30 per cent of cases.^{8, 11} These missed instances of child abuse are typically only identified when a subsequent abusive injury occurs.¹² Repeated abusive injury occurs in 35 per cent of all cases and 5–10 per cent of children die as a result of missed intervention.¹¹ Despite these alarming statistics, there have been no national studies of readmissions for child abuse after trauma admissions, including readmissions to different hospitals. Previous studies of hospitalization for child abuse have been limited by an inability to track prior admissions to different hospitals.^12–14 The purpose of this study was to identify the risk factors for hospital readmission for child maltreatment after hospitalization for trauma including admissions across different hospitals nationwide.

Material and Methods

The largest Federal-State-Industry partnership providing national encounter-level health-care data in the United States is the Healthcare Utilization Project (HCUP). The project provides researchers with a family of databases, including the Nationwide Inpatient Sample and the recently released, Nationwide Readmissions Database (NRD). The NRD extends the capability of the Nationwide Inpatient Sample by providing the unique capability of tracking patients across hospitalizations at different hospitals. The NRD data are also extensively processed to ensure that readmissions are counted as accurately as possible. This is especially important for trauma readmission studies that use population-level data because it is possible that admissions related to the index injury could be miscoded as a new injury.^{15, 16} The NRD also greatly reduces the likelihood of this type of miscoding by collapsing multiple records into one if they involve a transfer or same-day event such as discharge and admission between different hospitals.¹⁷

For this study, the NRD for 2010–2014 was queried for all admissions with an age less than 18 years and an ICD-9 Clinical Modification diagnosis code corresponding to trauma.¹⁸ Admissions with maltreatment were identified by ICD-9 diagnosis code corresponding to abuse or neglect (99550, 99551, 99552, 99553, 99554, 99555, 99559, 99580, 99581, 99582, 99583, 99584, 99585, and V7181) or an E-code corresponding to assault.^{19, 20} The primary outcome was readmission within one year for maltreatment. The secondary outcome was readmission for maltreatment at a different hospital. To identify possible missed maltreatment, a subgroup analysis was performed on patients without a diagnosis of maltreatment during the index admission. Univariable analysis was performed for each outcome using all variables during the index admission. Categorical variables were compared using a chi-squared test and continuous variables were compared using Student’s t test. Multivariable logistic regression was performed for each outcome using all significant variables, defined as a P value less than 0.10 on univariable analysis. Univariable analysis was used to limit variable selection bias and logistic regression was used to control for confounding factors.^{21, 22} The results of logistic regression are reported as odds ratios (OR) with [95% confidence interval (CI)]. Results were weighted for national estimates according to HCUP standards.¹⁷ The Institutional Review Board of the University of Miami waived the requirement for approval of this study because the NRD contains deidentified, publicly available data and is not considered human subjects research. The Abbreviated Injury Scale (AIS), Injury Severity Score (ISS), and the Charlson Comorbidity Index (CCI) were calculated for each patient using the ICDPIC version 3.0 software package implemented in Stata/SE version 12.0, StataCorp, College Station, TX.^{23, 24} Statistical analyses were performed using SPSS Statistics version 24, IBM Corporation, Armonk, NY.

Results

There were 608,744 admissions of children and teenagers for trauma during the study period and 44,569 (7.32%) involved maltreatment. Readmission for maltreatment was found in 1,948 (0.32%) and from these patients, 368 (18.89%) presented to a different hospital. The patient characteristics overall and for each outcome are shown in Table 1. The results of multivariable logistic regression are shown in Table 2. The most common diagnoses on index admission for maltreatment are shown in Table 3, whereas the most common diagnoses on readmission for maltreatment are shown in Table 4. The comparison of mean maximum AIS severity by body region is shown in Table 5, both with and without maltreatment at the index admission and readmission for maltreatment.

Table 1.

Overall Patient Characteristics for Readmission for Maltreatment and Readmission for Maltreatment at a Different Hospital

		Total	Readmission for Maltreatment		Readmission for Maltreatment at a Different Hospital
Characteristic		n (%)	n (%)	P Value	n (%)	P Value
Total		608,744 (100)	1,948 (0.32)		368 (18.89)
Age group (years)	12–17	299,730 (49.24)	709 (0.24)	<0.001	158 (22.28)	<0.001
	6–11	128,168 (21.05)	62 (0.05)		*
	0–5	180,847 (29.71)	1,177 (0.65)		210 (17.84)
Abuse at index admission		44,569 (7.32)	1,301 (2.92)	<0.001	173 (13.30)	<0.001
Female		217,817 (35.78)	518 (0.24)	<0.001	162 (31.27)	<0.001
Mechanism	Fall	174,522 (28.67)	221 (0.13)	<0.001	115 (52.04)	<0.001
	Penetrating	54,945 (9.03)	389 (0.71)		62 (15.94)
	Transportation	143,046 (23.50)	48 (0.03)		11 (22.92)
	Others	236,230 (38.81)	1,289 (0.55)		179 (13.89)
Self-inflicted		32,063 (5.27)	82 (0.26)	0.036	31 (37.80)	<0.001
ISS > 15		79,061 (12.99)	976 (1.23)	<0.001	106 (10.86)	<0.001
Control/ownership of hospital	Public	113,513 (18.65)	365 (0.32)	<0.001	50 (13.70)	<0.001
	Not for profit	450,121 (73.94)	1,512 (0.34)		290 (19.18)
	Investor owned	45,110 (7.41)	71 (0.16)		28 (39.44)
Teaching status of urban hospitals	Metropolitan nonteaching	107,613 (17.68)	285 (0.26)	<0.001	159 (55.79)	<0.001
	Metropolitan teaching	470,776 (77.34)	1,648 (0.35)		206 (12.50)
	Nonmetropolitan	30,354 (4.99)	14 (0.05)		*
Medicaid		287,197 (47.18)	1,486 (0.52)	<0.001	306 (20.59)	0.001
Lowest household income quartile		196,859 (32.34)	790 (0.40)	<0.001	264 (33.42)	<0.001
CCI > 0		62,221 (10.22)	179 (0.29)	0.132	18 (10.06)	0.002

^*n < 11; redacted according to HCUP Data Use Agreement.

Table 2.

Overall Patient Characteristics for Readmission for Maltreatment and Readmission for Maltreatment at a Different Hospital

		Readmission for Maltreatment		Readmission for Maltreatment at a Different Hospital
Characteristic		P Value	OR [95% CI]	P Value	OR [95% CI]
Age group (years)	12–17		Reference		Reference
	6–11	<0.001	0.52 [0.40–0.69]	*	*
	0–5	<0.001	2.28 [2.01–2.58]	<0.001	0.43 [0.29–0.64]
Abuse at index admission		<0.001	9.48 [8.35–10.76]	0.59	1.11 [0.76–1.62]
Female		<0.001	0.69 [0.62–0.77]	0.28	1.20 [0.87–1.66]
Mechanism	Fall		Reference		Reference
	Penetrating	<0.001	1.61 [1.30–1.99]	<0.001	0.11 [0.06–0.21]
	Transportation	<0.001	0.28 [0.20–0.38]	0.09	0.48 [0.20–1.13]
	Other	<0.001	1.38 [1.17–1.63]	<0.001	0.30 [0.18–0.48]
Self-inflicted		<0.001	2.30 [1.78–2.97]	<0.001	2.89 [1.60–5.23]
ISS >15		<0.001	3.88 [3.51—4.30]	0.25	0.82 [0.58–1.16]
Control/ownership of hospital	Public		Reference		Reference
	Not for profit	0.01	1.17 [1.04–1.31]	0.42	1.16 [0.81–1.67]
	Investor owned	<0.001	0.62 [0.47–0.81]	0.39	1.34 [0.69–2.60]
Teaching status of urban hospitals	Metropolitan nonteaching		Reference		Reference
	Metropolitan teaching	<0.001	0.69 [0.60–0.79]	<0.001	0.23 [0.16–0.33]
	Nonmetropolitan	<0.001	0.21 [0.13–0.37]	0.09	0.32 [0.09–1.18]
Medicaid		<0.001	1.93 [1.73–2.15]	0.10	1.33 [0.95–1.86]
Lowest household income quartile		<0.001	1.22 [1.11–1.34]	<0.001	3.50 [2.63–4.67]
CCI > 0		*	*	<0.001	0.25 [0.14–0.45]

^*Not significant on univariable comparison.

Table 3.

Most Common Injury Diagnoses on Index Admission with Maltreatment

ICD-9 Code	Description	n
920	Contusion of face, scalp, and neck except eye(s)	5,890
85220	Subdural hemorrhage after injury without mention of open intracranial wound, unspecified state of consciousness	2,390
9100	Abrasion or friction burn of face, neck, and scalp except eye, without mention of infection	1,996
85221	Subdural hemorrhage after injury without mention of open intracranial wound, with no loss of consciousness	1,708
8605	Traumatic pneumohemothorax with open wound into thorax	1,304
82101	Closed fracture of shaft of femur	1,248
92231	Contusion of back	1,123
80225	Closed fracture of mandible, angle of jaw	1,090
8020	Closed fracture of nasal bones	1,061
9221	Contusion of chest wall	1,049

Table 4.

Most Common Injury Diagnoses on Readmission with Maltreatment.

ICD-9 Code	Description	n
85220	Subdural hemorrhage after injury without mention of open intracranial wound, unspecified state of consciousness	451
81342	Other closed fractures of distal end of radius (alone)	251
80706	Closed fracture of six ribs	247
82101	Closed fracture of shaft of femur	129
85221	Subdural hemorrhage after injury without mention of open intracranial wound, with no loss of consciousness	127
85226	Subdural hemorrhage after injury without mention of open intracranial wound, with loss of consciousness of unspecified duration	96
9248	Contusion of multiple sites, not elsewhere classified	84
9100	Abrasion or friction burn of face, neck, and scalp except eye, without mention of infection	84
920	Contusion of face, scalp, and neck except eye(s)	72
80020	Closed fracture of vault of skull with subarachnoid, subdural, and extradural hemorrhage, unspecified state of consciousness	67

Table 5.

Mean Maximum AIS Severity by Body Region

Maximum AIS Severity Body Region	Maltreatment at Index Admission			Readmission for Maltreatment
	No	Yes	P Value	No	Yes	P Value
Head or neck	0.65 ± 1.18	1.08 ± 1.57	<0.001	0.67 ± 1.21	2.09 ± 1.87	<0.001
Face	0.11 ± 0.43	0.22 ± 0.60	<0.001	0.12 ± 0.45	0.15 ± 0.52	0.01
Chest	0.23 ± 0.78	0.51 ± 1.10	<0.001	0.25 ± 0.81	0.70 ± 1.25	< 0.001
Abdomen	0.24 ± 0.76	0.37 ± 0.90	< 0.001	0.25 ± 0.78	0.29 ± 0.87	0.02
Extremities	1.04 ± 1.17	0.62 ± 1.07	< 0.001	1.01 ± 1.17	0.85 ± 1.15	<0.001
External	0.40 ± 0.60	0.47 ± 0.77	< 0.001	0.41 ± 0.61	0.32 ± 0.58	< 0.001
All ISS body regions	2.07 ± 0.89	2.36 ± 1.12	< 0.001	2.09 ± 0.91	2.98 ± 1.16	< 0.001

Age Group

The age group with the most patients in this study was 12–17 years with 299,730 (49.24%) patients. The age group with the highest rate of readmission for maltreatment was 0–5 years with 1,177 (0.65%, P < 0.001). The rate of presentation to a different hospital with maltreatment was highest in the age group of 12–17 years (22.28%, P < 0.001). After controlling for confounding factors through multivariable logistic regression, the age group 0–5 years had the highest risk for readmission for maltreatment (OR 2.28 [2.01–2.58], P < 0.001). This youngest age group also had a decreased risk of presenting to a different hospital (OR 0.43 [0.29–0.64], P < 0.001, Table 2) compared with the reference group, 12–17 years.

Mechanism and Intent

Penetrating injury accounted for 54,945 (9.03%) patients in this study. The rate of readmission for maltreatment in these patients was elevated at 0.71 per cent (P < 0.001), but the rate of presentation to a different hospital was decreased at 15.94 per cent (P < 0.001). Multivariable logistic regression revealed increased risk for readmission for maltreatment with penetrating injury (OR 1.61 [1.30–1.99], P < 0.001). There were 32,063 (5.27%) patients suffering self-inflicted injury in this study. There was a decreased rate of readmission for maltreatment for self-inflicted injury (0.26%, P = 0.036) and an increased rate of readmission to a different hospital (37.80%, P < 0.001). Multivariable logistic regression revealed an increased risk of readmission for maltreatment with self-inflicted injury (OR 2.30 [1.78–2.97], P < 0.001) and for readmission to a different hospital (OR 2.89 [1.60–5.23], P < 0.001).

Hospital Characteristics

The majority of patients in this study were treated in not-for-profit (73.94%) metropolitan teaching (77.34%) hospitals. The highest rate of readmission for maltreatment occurred in not-for-profit (0.34%, P < 0.001) and metropolitan teaching (0.35%, P < 0.001) hospitals. However, the highest rate of readmission to a different hospital occurred with investor-owned (39.44%, P < 0.001) and metropolitan nonteaching (55.79%, P < 0.001) hospitals. Multivariable logistic regression revealed a decreased risk of readmission for maltreatment at investor-owned hospitals (OR 0.62 [0.47–0.81], P < 0.001) and for metropolitan teaching hospitals (OR 0.69 [0.60–0.79], P < 0.001).

Injury Severity and Comorbidities

There were 79,061 (12.99%) patients with an ISS > 15, the rate of readmission for maltreatment for these patients was increased at 1.23 per cent (P < 0.001) and the rate of readmission to a different hospital was decreased at 10.86 per cent (<0.001). Controlling for risk factors for multivariable logistic regression revealed that patients with an ISS > 15 were at increased risk of readmission for maltreatment (OR 3.88 [3.51–4.30], P < 0.001). Patients with a CCI > 0 comprised 10.22 per cent (62,221) of patients in this study, and these patients were at decreased risk for readmission at a different hospital (OR 0.25 [0.14–0.45], P < 0.001).

Primary Payer and Median Household Income

Medicaid was the primary payer for 47.18 per cent of patients in this study and 32.34 per cent were in the lowest median household income quartile. The rate of readmission for maltreatment was higher with both Medicaid (0.52%, P < 0.001) and the lowest median household income quartile (0.40%, P < 0.001). There was a higher rate of readmission to a different hospital with Medicaid (20.59%, P = 0.01) and the lowest median household income quartile (33.42%, P < 0.001). An increased risk for readmission for maltreatment was found with Medicaid (OR 1.93 [1.73–2.15], P < 0.001); however, there was no difference in risk of readmission at a different hospital with Medicaid (P = 0.10). The lowest median household income quartile was associated with increased risk for both readmission for maltreatment (OR 1.22 [1.11–1.34], P < 0.001) and readmission to a different hospital (OR 3.50 [2.63–4.67], P < 0.001).

Subgroup Analysis for Missed Maltreatment

After excluding patients with maltreatment on index admission (44,569), there were 564,175 children and teenagers admitted for trauma. Readmission for maltreatment was identified in 647 (0.11%) patients. Multivariable logistic regression revealed that the strongest risk factors for readmission for maltreatment were ISS > 15 (OR 3.29 [2.68–4.03], P < 0.001), a primary payer of Medicaid (OR 3.20 [2.64–3.88], P < 0.001), penetrating injury (OR 2.86 [2.16–3.79], P < 0.001), lowest median household income (OR 1.89 [1.61–2.21], P < 0.001), age group 0–5 years (OR 1.80 [1.50–2.16], P < 0.001), and female (OR 1.46 [1.25–1.72], P < 0.001).

Discussion

This study represents the first nationwide evaluation of readmissions for maltreatment after trauma, including readmissions at a different hospital. The finding that 1,948 (0.32%) children were readmitted for maltreatment after trauma suggests that at least a portion of the index admissions were due to unrecognized child maltreatment. This missed maltreatment is even more likely with the subgroup that only included children without maltreatment on the index admission. This analysis found 647 children who suffered injuries highly suggestive of maltreatment. The strongest risk factor for missed maltreatment was ISS > 15 (OR 3.29 [2.68–4.03], P < 0.001). This is consistent with other studies that have found that an ISS > 15 in a child is much more likely to be due to child maltreatment than accidental injury.^{25, 26}

This study presents the novel finding of an 18.89 per cent rate of readmissions for maltreatment at a different hospital. This is similar to previously reported rates of readmission to a different hospital after trauma from the NRD (10–26%).^{27, 28} This fragmentation of care represents an important area for clinical improvements.²⁹ There are other important implications for prevention through identifying the most common patterns of maltreatment from this study, particularly the finding that maltreatment at the index admission had a much higher rate of readmission for maltreatment (2.92%, P < 0.001, Table 1). Even when controlling for confounding factors through logistic regression, prior maltreatment was still a risk factor for readmission for maltreatment (OR 9.48 [8.35–10.76], P < 0.001). Other studies have shown that child maltreatment is up to six times more likely to occur in children with prior maltreatment.^{30, 31}

Unrecognized maltreatment was also prevalent in the youngest age group (0–5 years) in this study. These vulnerable children were at increased risk for readmission for maltreatment (OR 2.28 [2.01–2.58], P < 0.001). Yet, these youngest children were at decreased risk for readmission for maltreatment at a different hospital (OR 0.43 [0.29–0.64], P < 0.001, Table 2). Because there was no attempt to hide maltreatment by presenting to a different hospital, this could indicate that the maltreatment was unacknowledged by both the home caregivers and health-care providers. These findings are consistent with previous studies of mothers of neglected children that found that these women tended to believe that they had been caring for their children well and were thus surprised when their child required hospitalization.^{32, 33}

Penetrating injuries also fit a similar pattern where they were associated with an increased risk for readmission for maltreatment, but these children tended to return to the same hospital (Table 2). However, self-inflicted injuries showed a different pattern where they were at high risk of both readmission for maltreatment (OR 2.30 [1.78–2.97], P < 0.001) and readmission to a different hospital (OR 2.89 [1.60–5.23], P < 0.001, Table 2). Self-harm has been a well reported indicator of child maltreatment.^{34, 35} The findings of this study suggest that the caregivers of self-harming children may be attempting to avoid scrutiny from health-care providers by presenting to a different hospital for maltreatment.

The connection between subdural hematomas and child abuse has been reported since 1939.³⁶ In this study, subdural hemorrhage was the second most common diagnosis code on index admission with maltreatment (Table 3) and was the most common diagnosis on readmission with maltreatment (Table 4). This is further represented by the AIS body regions in this study. The body region for head or neck had a much higher mean maximum AIS for patients with maltreatment at the index admission and readmission for maltreatment (Table 5). These worse AIS severities for the head and neck region have been previously shown to correlate with abuse in children. Davies et al.²⁵ showed that in children with injuries and suspected abuse, the body region that was most severely injured was the head with 57 per cent having an AIS 3+ injury. In a separate later study, Davies et al.³⁷ also showed that the extremities were the most severely injured body region in children with accidental trauma. This is another finding that is common with the pattern of maltreatment in this study. The mean maximum AIS for extremities was lower in children with maltreatment at the index admission (P < 0.001) and in children with readmission for maltreatment (P < 0.001, Table 5).

The majority of children in this study were treated in metropolitan teaching hospitals (77.34%, Table 1). These hospitals were associated with a decreased risk of readmission for maltreatment and readmission for maltreatment at a different hospital (Table 2). This could be explained by the likelihood that these hospitals are better equipped to both identify child maltreatment at the index admission and to ensure readmissions are to the index hospital.^{38, 39} These hospitals also treat more patients with a lower socioeconomic status, and it has been widely reported that both abuse and neglect are more prevalent in these patients.^40–42 This is consistent with this study where the rate of readmission for maltreatment was higher with both Medicaid (0.52%, P < 0.001) and the lowest median household income quartile (0.40%, P < 0.001, Table 1). Both Medicaid and the lowest median household income were risk factors for readmission for maltreatment, whereas the strongest predictor of readmission for maltreatment at a different hospital was found with the lowest median household income quartile (OR 3.50 [2.63–4.67], P < 0.001, Table 2).

The limitations of this study include those of retrospectively collected administrative databases such as the NRD, e.g., coding error. In addition, the NRD cannot follow up patients across years or if the patients are readmitted across state lines. HCUP estimates that less than 5 per cent of readmissions occur across state lines.¹⁷ Despite these limitations, the NRD is the largest database that allows for following up readmissions across different hospitals in the United States. This is particularly important in victims of child maltreatment who often present to nonindex hospitals.^{27, 28}

Conclusions

The recently released NRD provides the unique capability of tracking patients across hospital admissions in the United States. This study leverages this distinctive database to reveal the pattern of risk factors that could lead to early identification and prevention of child maltreatment. A significant portion of readmissions for child maltreatment was at a different hospital, indicating the paramount need for a standardized medical record easily transferable between hospitals. Furthermore, the characteristics elucidated in this study should prove useful to guide future efforts at preventing missed opportunities for intervention on behalf of the vulnerable victims of child maltreatment.