A System-Wide Hospital Child Maltreatment Patient Safety Program

This is a retrospective study of a child maltreatment patient safety program in a tertiary pediatric health care system in a Midwestern city with a metropolitan population of >2 million. The health care system includes 2 free standing children’s hospitals, each with dedicated pediatric EDs and inpatient units, a PICU, 3 UC sites, 2 primary care offices, and >30 specialty clinics. Subjects included children who received care within the system and had an identified concern for child maltreatment during care between December 2015 and April 2019.

By health care system policy, whenever any staff member had concerns for child maltreatment, a patient at risk (PAR) (Table 1) form was completed by hospital social workers (SWs) in the electronic medical record. PAR forms served as the location for documentation by SWs after an in-person meeting with the family. Completion of a PAR form did not automatically result in a report to CPS because a concern for maltreatment may not have risen to the level of suspicion requiring a report. Information gathered by SWs to complete the PAR was used to further inform decision-making regarding filing a report. The filing of reports with CPS was at the discretion of staff involved in each specific case on the basis of their determination of the level of suspicion for abuse in that case. By state laws, all staff are mandated reporters of suspected child maltreatment. A report to LE was made when staff believed a crime may have been committed, but this is not mandated by state laws. A potential victim of maltreatment could be identified at any location in the system (ED or UC, inpatient, or clinic) and at any time while under the care of the hospital system. All children with completed PAR forms were subsequently engaged in the child maltreatment safety program, as described below, and serve as the subjects for this study. Subjects were excluded if the PAR form was completed because of a CAP consult because the patient safety program was not necessary to identify these children. Hospital medical personnel were free to consult or discuss cases with the CAP medical providers outside of the safety program.

The hospital policy on child maltreatment contained a list of questions describing situations in which staff should consider maltreatment (Supplemental Information); however, this existed only in the hospital policy and was not used as a formal checklist. Staff could refer to it as desired, but the list was not otherwise formalized within the safety program. As such, all staff members used their own judgment regarding the possibility of abuse in any given situation. SW consultation for completion of a PAR form was based on the concerns of any staff member. SW staff classified the reason for the PAR form, and a single PAR form could include concerns for >1 type of maltreatment. PAR forms categorized as “other” included patients with intimate partner violence and food insecurity, among other concerns.

Daily, an Excel file including names, medical record numbers, location within the hospital system, and PAR text field entries, was sent to the hospital CAP team. CAP providers reviewed the PAR form and medical documentation for the encounter (including interventions, history taking, test results, photographs [if present], and diagnoses, as well as documentation regarding mandated reporting, interactions with CPS and/or LE, and safety planning at discharge, if appropriate). Hospital policy allowed CAP providers to make further recommendations and interventions, when necessary, on an individual case level, including contacting inpatient teams to discuss the possible need for a CAP consult. All completed PAR forms with associated medical documentation were reviewed within 24 hours by the CAP team and assigned ≥1 subsequent categories of intervention by the CAP team:

1. Immediate callback: immediate return to the hospital to correct medical error (eg, a child with a potential sentinel injury discharged from the hospital back into a potentially abusive home or young child needing acute occult injury screening).

2. Need for inpatient consult: managing inpatient team contacted by a CAP to discuss the need for possible consult.

3. Contact CPS or LE: the CAP contacted CPS or LE for any reason (eg, the CAP recommends further scene investigation to clarify the history provided regarding a child’s injury and findings).

4. Follow-up appointment: the CAP arranged for a follow-up medical appointment regarding the possibility of abuse (eg, a follow-up appointment with the CAP needed to reassess findings and/or a repeat skeletal survey).

5. Other: all other actions not encompassed by the above.

6. No action taken: no errors or need for CAP intervention.

The above interventions were used to address any possible type of error. For instance, an infant with a sentinel injury (eg, bruise) could have been identified through the program with any of the following errors: lack of report to CPS, lack of appropriate testing for occult injury, discharge to an unsafe setting (even in the context of a CPS report being made), and/or lack of appropriate testing for a medical condition. In cases involving immediate callbacks, the previous managing physician was contacted by the CAP to discuss the case and reasons for the immediate callback. There were no other maltreatment-related decision rules or algorithms in use in the health care system during the study period.

Subjects were identified through a review of the PAR database encompassing the study period. For research purposes, authors collected the following on each subject: the location of care (inpatient, clinic, or ED or UC), age, reason for PAR form, CPS and/or LE involvement, and CAP interventions using the 6 categories, above. Medians, interquartile ranges (IQRs), and proportions were used to characterize the data. χ² and Wilcoxon rank tests were used to compare the groups on the basis of categorical and continuous variables, respectively. Statistical analyses were conducted by using SAS version 9.4 (SAS Institute, Inc, Cary, NC) software.

For study aim 1, the frequency of subjects with a PAR form who needed further intervention (eg, categorized as any of 1 to 5 in the daily PAR review) was tallied and compared with the total number of PARs performed. For aim 2, the need for further intervention was evaluated for association with the subject age and location of PAR. For aim 3, subjects requiring an immediate callback (category 1 in the program) were identified, and study authors performed an individual chart review and collected the following information: the recommended immediate interventions, the reasons for the immediate callback (categorized by the primary presenting injury or concern), whether the recommended intervention was completed, and diagnostic impressions related to abuse by the initial medical provider and, if seen by the CAP team, that from the CAP team as well. The institutional review board of the authors’ institution approved this study.

During the 30-month study period, 1 497 117 hospital visits occurred, generating 7698 PAR forms (0.5% of all visits), resulting in an average of 8.6 PAR forms per day. The distribution of PAR forms across the first (33.4%), second (33.1%), and third (33.5%) 10-month periods in the study was similar. The median age of the subjects was 5.4 years (IQR 2–12.3). Subject sex, ethnicity, and socioeconomic status were not collected in the PAR form to preclude their potential to induce bias in the safety program and thus are not available as study data points.

Table 2 details the results of the program. One-third of PARs were completed outside of the ED or UC centers. Physical abuse was the most common single reason for a PAR form, but two-thirds of PAR forms were done for reasons other than physical abuse. Physical abuse was the most common reason for PAR forms requiring further intervention in younger ages, whereas sexual abuse was predominate in teenagers (Fig 1). For study aim 1, of the 7698 PAR forms reviewed, the safety program resulted in some intervention for 2061 (26.8%). Excluding follow-up appointments, 619 (8.0%) of PAR forms resulted in an intervention. For aim 2, the median age of the subjects for whom an intervention occurred (median age: 5.6 years; IQR 2.3–10.8) was not statistically different from that of those for whom no intervention occurred (median age: 5.2 years; IQR 1.9–12.7). Subjects requiring an immediate callback (median age: 1.5 years; IQR 0.6–4.3) were significantly younger than those requiring no intervention (median age: 5.2 years; IQR 1.9–12.7; P < .0001) and those requiring any other intervention (median age: 5.7 years; IQR 2.4–10.9; P < .0001).

Interventions were more common in the ED or UC settings than inpatient or clinic settings (Table 3). Although the odds of an immediate callback were higher in the ED or UC setting than in the inpatient setting, there was no significant difference between ED or UC and clinic, and 15% of the immediate callbacks occurred because of care in the clinic setting. It should be noted that a single PAR form could be performed for multiple maltreatment concerns (eg, physical abuse and neglect) and multiple different interventions could be performed on the basis of a single PAR form. Of the 5171 ED or UC subjects with a PAR, 1687 (32.6%) required an intervention, most commonly for physical abuse. Of the 1228 inpatients with a PAR form, 123 (10%) required some intervention, most commonly for neglect. Finally, for clinic patients, 251 (19.3%) required an intervention, again, most commonly for neglect. Of the 2154 total interventions by the CAP team through the safety program, 398 (18.5%) occurred on the basis of inpatient or clinic visits.

When evaluating the need for interventions based on PAR form reason (Table 4), physical abuse was more likely than sexual abuse (odds ratio [OR]: 2.0; 95% confidence interval [CI] 1.7–2.2) or neglect (OR: 5.1; 95% CI 4.4–6.0), whereas sexual abuse was more likely than neglect (OR: 2.6; 95% CI 2.2–3.1). Immediate callbacks were more likely to occur in PAR forms done for physical abuse than in those done for sexual abuse (OR: 8.2; 95% CI 2.9–23) or neglect (OR: 5.9; 95% CI 2.3–14.9). No significant difference was seen when comparing sexual abuse and neglect (OR: 0.7; 95% CI 0.2–.7).

For aim 3, Table 5 details the reasons for the 53 immediate callbacks. Over 40% involved subjects with bruising. Immediate follow-up in the CAP clinic occurred in 87% (46 of 53) of cases. There was CPS involvement in the remaining 7 cases: 5 declined to follow-up in the clinic, and 2 could not be located. The 46 cases seen in immediate follow-up are detailed in Table 6. For 17% (8 of 46) of cases, a diagnosis regarding abuse was not provided by the non-CAP but was subsequently provided by the CAP. In 39% (18 of 46), the diagnosis regarding abuse was changed by the CAP from the initial non-CAP diagnosis. The diagnostic change was from not abuse to abuse in 94% (17 of 18) of cases. Overall, CAP assessment resulted in a new or changed diagnosis in 57% of emergency call back cases. Of the 17 cases initially diagnosed as not abuse and ultimately diagnosed as abuse by the CAP, the concern was related to physical abuse for 16 subjects and sexual abuse for 1 subject. Their ages ranged from 9 days to 15 years, with more than one-half being >3 years old.

A 6-month-old male infant presented to the ED for concerns of wheezing and eczema. The ED provider noted the following: “bruise to the right pinna. Mom thinks the infant may have rolled off the couch.” The bruise was documented in the physical examination, but no photograph was taken. A PAR form was completed for possible medical neglect because of missing clinic appointments.

PAR form review the next day through the safety program resulted in an immediate callback to the clinic. Through a skeletal survey, a classic metaphyseal fracture of the right radius was identified. A diagnosis of child physical abuse was provided, and LE and CPS were called. The child was placed into the care of a relative.

Variation in care and misdiagnoses in child abuse have been well documented.^4–6 ,9  In particular, missed sentinel injuries and missed abusive head trauma have been shown to result in subsequent harm to young infants.^5,6  However, diagnostic and evaluation errors are not isolated to young infants and physical abuse.^3,9,10,11  Additionally, the overdiagnosis of abuse can occur in the setting of medical conditions that mimic abuse.^12,13  As such, to fully address potential medical errors regarding abuse, a more comprehensive approach is needed. This study demonstrates the impact of a child maltreatment safety program on medical errors using a similar structure to other safety programs.^7,8 

Formal screening procedures for detecting abuse and improving evaluations have been targeted at the ED setting.^14,15  ED-based screens may be a critical component of comprehensive abuse detection systems, but broader safety programs may be necessary to fully capture maltreatment-related medical errors. In this study, immediate callbacks represented the most critical interventions and demonstrate the need for safety interventions outside of the ED and UC settings. The population of immediate callbacks often fit the broad concept of a sentinel injury, with bruising in young children and diagnostic and safety issues predominating. This is unsurprising, given the known relationship between bruising in young infants and missed abuse.^5,16  However, many of the callbacks involved a lack of appropriate initial testing and sexual abuse and/or neglect, and 17% were from the clinic or inpatient setting.

The most common reason for intervention in the safety program was to arrange a follow-up appointment, followed by the CAP team contacting CPS and/or LE, a known critical element of medically based child abuse work.⁹  Because medical providers were free to consult and/or call the CAP team or arrange for follow-up appointments throughout the study, the high need for follow-up appointments may reflect the lack of understanding of when a child may benefit from a CAP clinic appointment. The need for further contact with CPS and/or LE demonstrates the vital role that CAP and CPS and/or LE collaboration in determining accurate medical diagnoses regarding abuse. This sort of collaboration is uncommon among non-CAP providers²  and is a critical benefit of the safety program.

Globally, the findings of the safety program may best be described as “we don’t know what we’re missing if we don’t look.” Because of the multiple presentations of maltreatment and the locations of presentation in the context of physician confidence and skill gaps, a hospital-wide safety program may be a necessary step to ensure patient safety from harm. Given the long-term implications and costs of child maltreatment, addressing these issues as urgently and frequently as necessary seems prudent.^17,18 

There are limitations to this study. The safety program requires time investment from SWs that may not be available in many locations. However, the core components could be adapted to the available resources (such as shortening the PAR form or simply flagging any case concerning for abuse for CAP review). Additionally, the benefits of eliminating medical errors involving child maltreatment likely outweigh the drawbacks of investments in such a program. Further study on such issues and adaptations is warranted. Data on all maltreatment concerns in the hospital system before implementation of this system are not available for comparison. It is not known how often medical providers did not take specific actions because they knew the safety program was in place. This may have resulted in more safety-based interventions in the study. However, this would also be reflective of providers’ previously documented lack of confidence in handling maltreatment.^2,3  Multiple demographic factors, including race, ethnicity, insurance status, and sex, were not recorded in the PAR form review program and, therefore, were not included in the data set. Although the program has been in existence for several years at the authors’ institution and has been well-received, there is also no objective measure of the acceptability of the child maltreatment intervention program. Finally, the safety program requires staff to have a concern for abuse. It may be best deployed with other interventions to assist in triggering the concern, such as electronic medical record based-alert systems.¹⁵  Further study is warranted.

A child maltreatment patient safety program encompassing a large pediatric health system provides significant benefits to children. Including such a program across all clinical settings is beneficial.

CAP

child abuse pediatrician

CI

confidence interval

CPS

Child Protective Services

ED

emergency department

IQR

interquartile range

LE

law enforcement

OR

odds ratio

PAR

patient at risk

SW

social worker

UC

urgent care