Improving Hospital-to-Home Transitions for Children Entering Foster Care
- aSchool of Medicine, Tufts University, Boston, Massachusetts;
- bDartmouth-Hitchcock Medical Center, Lebanon, New Hampshire;
- cThe Dartmouth Institute for Health Policy and Clinical Practice, Lebanon, New Hampshire; and
- dDepartment of Pediatrics, The Barbara Bush Children’s Hospital, Maine Medical Center, Portland, Maine
- Address correspondence to Leah A. Mallory, MD, Department of Pediatrics, The Barbara Bush Children’s Hospital, Maine Medical Center, 22 Bramhall St, Portland, ME 04102. E-mail:
Mr DeLucia participated in the conceptualization and design of the study, designed the data collection instruments, collected data, conducted initial analysis, drafted the initial manuscript, and participated in revisions; Ms Martens participated in the conceptualization and design of the study, data collection, and manuscript revision; Dr Leyenaar participated in the conceptualization and design of the study and refinement of the data collection instruments, supervised data analysis, and revised the manuscript; Dr Mallory served as senior author and participated in the conceptualization and design of the study, data analysis, and manuscript drafting and revisions; and all authors reviewed and approved the final manuscript as submitted.
BACKGROUND AND OBJECTIVES: Hospital-to-home transitions present safety risks for patients. Children discharged with new foster caregivers may be especially vulnerable to poor discharge outcomes. With this study, our objective is to identify differences in discharge quality and outcomes for children discharged from the hospital with new foster caregivers compared with children discharged to their preadmission caregivers.
METHODS: Pediatric patients discharged from the Barbara Bush Children’s Hospital at Maine Medical Center between January 2014 and May 2017 were eligible for inclusion in this retrospective cohort study. Chart review identified patients discharged with new foster caregivers. These patients were compared with a matched cohort of patients discharged with preadmission caregivers for 5 discharge quality process measures and 2 discharge outcomes.
RESULTS: Fifty-six index cases and 165 matched patients were identified. Index cases had worse performance on 4 of 5 discharge process measures, with significantly lower use of discharge readiness checklists (75% vs 92%; P = .004) and teach-back education of discharge instructions for caregivers (63% vs 79%; P = .02). Index cases had twice the odds of misunderstandings needing clarification at the postdischarge call; this difference was not statistically significant (26% vs 13%; P = .07).
CONCLUSIONS: Hospital-to-home transition quality measures were less often implemented for children discharged with new foster caregivers than for the cohort of patients discharged with preadmission caregivers. This may lead to increased morbidity, as suggested by more frequent caregiver misunderstandings. Better prospective identification of these patients and enhanced transition improvement efforts targeted at their new caregivers may be warranted.
Hospital-to-home transitions present well-recognized safety risks, and children who are medically and socially complex are particularly vulnerable to poor outcomes.1 Although children with medical complexity are the focus of several ongoing improvement efforts in this area,2–4 there are fewer efforts aimed at children with social risk factors. More than 400 000 children are currently in foster care in the United States5; these children are, by definition, socially complex and have higher rates of developmental delay, mental illness, and chronic health conditions than children from similar socioeconomic backgrounds not in foster care.6–10 For many of these children, placement in the foster care system resulted from neglect or abuse. Less commonly, children with complex medical needs may enter the foster care system when their needs outstrip their parents’ resources and abilities.11 Hospitalization may be used as an interim step to complete a medical workup for neglect, failure to thrive, or abuse and to ensure child safety while a safe disposition plan is identified.
Children discharged into foster care may be at increased risk of adverse outcomes after hospital-to-home transitions for several reasons. In outpatient settings, communication challenges between health care providers and foster caregivers are well described, with foster caregivers frequently reporting receipt of insufficient information about the child’s medical conditions.12 At the same time, effective provider hand offs may be challenged by relocation of the child’s medical home concurrent with entry into foster care.7 However, authors of previous studies have not examined the quality of hospital-to-home transitions for this vulnerable population.
We aimed to examine differences in transitional care quality measures between children discharged to new foster caregivers and children discharged to preadmission caregivers.
Data Source and Study Population
The Improving Pediatric Patient-Centered Care Transitions (IMPACT) project is an American Academy of Pediatrics–affiliated quality improvement research collaborative aiming to implement and test a 4-element pediatric transitions bundle (including a discharge readiness checklist, teach-back education, timely and complete hand off to the primary care provider, and a postdischarge phone call).13 We used data from our site’s Project IMPACT database to conduct a single-center retrospective matched cohort study comparing discharge process and outcome measures for patients discharged with new foster caregivers compared with children discharged back to their preadmission caregivers. Patients with new foster caregivers were defined as those with Department of Health and Human Services (DHHS) involvement who were taken into custody and discharged to new foster caregivers, those in DHHS custody with a foster caregiver change between admission and discharge, and those for whom legal custody did not change, but a DHHS discharge safety plan specified discharge with a new caregiver. Patients discharged with kin were included if the caregiver had not provided care for the child before admission. All patients ages 0 to 18 years discharged from the hospital from the pediatric hospital medicine service at the Barbara Bush Children’s Hospital at Maine Medical Center between January 2014 and May 2017 were eligible for inclusion. As part of Project IMPACT, detailed chart review was performed for all discharged patients, with data entered into a Research Electronic Data Capture (REDCap) database.14
To identify index cases, manual chart review was conducted for the following 3 patient subgroups: (1) patients with “foster care” noted as the disposition in the REDCap database; (2) patients discharged with “care of DHHS” identified in the electronic health record (EHR) demographic tab; and (3) patients who had been listed as open DHHS cases during monthly meetings of hospital social workers and DHHS liaisons.
Matching was used to identify up to 3 cohort patients for each index case. Match criteria were determined a priori on the basis of attributes that were perceived to be the most important potential confounders in this study population and included the following: age at discharge (<1 month, 1 month to 1 year, 1–3 years, or 3–5 years), medical complexity (1–3 rating determined by using the Pediatric Medical Complexity Algorithm),15 discharge medical technology, payor, and language spoken by the primary caregiver (English or non-English) as identified in the EHR. A Microsoft Excel Visual Basic for Applications algorithm was used to identify matching patients from the REDCap database. Eligible cohort patients were selected sequentially in chronological order after each index case discharge date. Cohort patients matched cases exactly on all match criteria. In addition, we extracted data regarding patient sex, race, length of stay, whether home nursing was arranged, and whether a patient’s primary care physician (PCP) was in-network (with shared EHR).
The Maine Medical Center Institutional Review Board determined the study was exempt from review.
We evaluated 5 hospital-to-home discharge quality improvement bundle elements and 2 discharge outcome measures, all reported as binary outcomes. Discharge quality bundle elements included the following: (1) the use of a discharge transition readiness checklist, (2) documentation of teach-back education of discharge instructions for caregivers, (3) new medications filled before discharge, (4) discharge information communicated with the PCP within 1 day of discharge, and (5) completion of follow-up phone calls with caregivers. Discharge outcome measures included the following: (1) 30-day all-cause hospital readmission rate and (2) caregiver-reported misunderstanding of discharge instructions needing clarification at the postdischarge phone call. All quality measures were operationalized as previously published, including the postdischarge phone call script.13,16
Additional chart review of index cases was conducted to gather data for the 2 additional descriptive analyses of assessing whether contact information for new caregivers was listed in the EHR and whether patients were assigned new PCPs at discharge.
For the 5 process measures, comparisons between index cases and matched cohort patients were performed by using conditional logistic regressions. For the 2 outcome measures, data only existed for a subset of patients and are as follows: (1) whether there was a misunderstanding needing clarification on the follow-up phone call could only be assessed for the subset of patients where the follow-up phone call was completed and (2) whether new medications were filled before discharge could only be assessed for the subset of patients where new medications were prescribed at discharge. Because data for these 2 outcome variables only existed for a subset of patients, we did not have completely paired data and could not use a matched analysis. Therefore, for these 2 outcome measures, unconditional logistic regressions with the matching criteria included as additional explanatory variables were performed to compare index cases to patients discharged back to their preadmission caregivers.
Five variables beyond our match criteria were included in our multivariable models (patient sex, race, length of stay, home nursing, and in-network PCP). These variables were included in our final regression models if they were significantly different (P < .10) between case and cohort groups and if inclusion of each variable in a 2-variable conditional logistic regression changed the odds ratio for the index case status variable by ˃10%. Statistical analyses were performed in Stata version 12.1 (Stata Corp, College Station, TX).
Fifty-six index cases were identified, along with 165 matched cohort patients. At least 1 matched cohort patient was identified for each index case. Demographic and other key characteristics for participants are shown in Table 1. Index cases were more frequently male patients and had significantly longer length of stay in the hospital.
With Table 2, we summarize primary admission diagnoses and reasons for discharge into foster care. Children discharged into foster care were most frequently admitted for trauma, whereas respiratory illnesses were the most common primary admission diagnosis in matched cohort patients. Just over one-third (39%) of children discharged into foster care were placed with kin.
With Table 3, we report discharge quality measures for index cases, compared with matched cohort patients. Four of the 5 discharge quality improvement bundle elements were less often implemented for index cases, and this difference was statistically significant for (1) the use of discharge transition readiness checklists and (2) teach-back education of discharge instructions. Compared with patients discharged back to their preadmission caregivers, index cases had twice the odds of misunderstandings needing clarification at the postdischarge phone call; this difference was not statistically significant. No difference was detected between cases and the matched cohort for 30-day readmission rates.
Additional descriptive measures collected for children discharged with new foster caregivers revealed that 32 (57%) did not have contact information for the foster caregiver listed in the discharge summary or discharge transitions checklist. Six (11%) had a change in PCP noted in the EHR between admission to hospital discharge.
In this retrospective matched-cohort study, children discharged with new foster caregivers less often received 4 of the 5 elements of our hospital-to-home quality improvement bundle. Additionally, caregivers more frequently had misunderstandings of key discharge instructions at the time of postdischarge phone contact. There were no differences in rates of hospital readmission in our small sample.
Although total numbers of children discharged into foster care are relatively small, these children are one of our most vulnerable populations, many of whom have already suffered from conditions of neglect or abuse. Although authors of several pediatric hospital-to-home research collaboratives are focusing on the needs and outcomes of medically complex children,2–4 to our knowledge, this is the first study used to focus on this socially complex population. With our analysis, we suggest that we are less likely to implement efforts to support hospital-to-home transitions for children discharged with new foster caregivers compared with children discharged with their preadmission caregivers, which may lead to increased morbidity and caregiver stress.
The majority of our case population were children admitted to the hospital in the custody of their parents often for a workup of suspected abuse, neglect, or failure to thrive. Their longer length of stay compared with the matched cohort is not unexpected and was likely influenced by the lack of a safe discharge plan, rather than medical indications for continued hospitalization. The process of child protective services obtaining legal custody and/or identifying a safety plan including a foster caregiver may extend beyond the time period required for resolution of medical issues.17 Thus, once the foster caregiver is identified, discharge may occur relatively quickly, often without time for foster caregivers to actively participate in patient care in the hospital with the child. In some cases, the child protective services representative may accompany the child from the hospital to the foster caregiver, such that information from the hospital care team is relayed secondhand. The decreased frequency of teach-back education of discharge instructions, as well as lower rates of filling medications before discharge may be due in part to lack of contact time between doctors, nurses, and foster caregivers in the inpatient setting. This represents a missed opportunity, because studies in which pediatric and adult patients are focused on reveal that teach-back can improve outcomes,18–20 and 2 studies regarding pediatric hospital discharge revealed that high-quality teaching provided by nurses was associated with increased parental readiness for discharge.21,22 Similarly, filling medications before discharge, and therefore enabling teaching with medications in hand, has been shown to decrease caregiver misunderstanding postdischarge, improve caregiver satisfaction, and decrease reuse in certain populations.16
Difficulties identifying the legal custodial change and postdischarge contact information for caregivers in the EHR undoubtedly also contributed to a lower rate of follow-up call completion among children discharged into foster care. Lack of clear documentation of custodial change and contact information also has implications for the hand off between the inpatient to the outpatient care teams. For the 11% of children with a documented PCP change at the time of hospital discharge, more information than simply a timely discharge summary is needed for a PCP to competently care for the child after discharge.
We acknowledge several limitations to this study. Our sample was underpowered to detect significant differences in several outcome variables (we note, however, that our finding of no difference in 30-day readmission rates is consistent with findings from other pediatric hospital-to-home improvement collaboratives).23,24 Further limitations include the low number of index cases, our single-center design, and the fact that the data in our REDCap database were manually entered by several different individuals over the study period and is therefore subject to human error. Additionally, with our small sample size, our index cases included no children over 5 years of age, although our inclusion criteria extended to age 18. This limits the ability to extrapolate our findings to older children being discharged with new foster caregivers. Before initiating the project, we asked the other Project IMPACT sites to provide the number of patients included in the overall database with a disposition of “foster care.” The total number of these patients from the other 3 sites (with over 2000 patients) was <15, likely reflecting documentation failures. Difficulties identifying the correct disposition of children entering foster care supports our concern that these children are at high risk of loss to follow-up and adverse outcomes after discharge and challenges effective hand offs to new caregivers and PCPs.
As a result of our study’s findings, our next steps will include better prospective identification of children discharged with new foster caregivers, enabling more reliable implementation of our hospital-to-home transitions improvement bundle. Clearer identification of disposition, caregiver identity and contact information, as well as allowing adequate time for discharge preparation to allow newly identified caregivers to receive in-hospital teach-back education and hands-on teaching with new medications will be key to achieving this goal.
None of the work could have occurred without support from the following members of our hospital-to-home transitions improvement team: Dannielle Allen, Kelly Anctil, Jonathan Bausman, Aggie Bellevue, Shannon Bennett, Abihijit Bhattacharyya, Jonathan Borque, Nancy Bouthot, Sarah Bunting, Danielle DiCesare, Noah Diminick, Jennifer Hayman, Jennifer Jewell, Lee Longnecker, Melanie Lord, Nicole Manchester, Lorraine McElwain, Jessica Miller, Teresa Morgan, Elizabeth Murphy, Logan Murray, Nancy Nystrom, Steve Prato, Brandy Robertson, Clare Ronan, and Susan Talbot; the Project IMPACT Pilot Site Leaders David Cooperberg, Snezana Osorio, and Sandra Gage; and Wendy Craig and Lee Lucas for data analysis support.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: Mr DeLucia received a Tufts University Medical Student Summer 2017 Research Fellowship.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- Copyright © 2018 by the American Academy of Pediatrics