Perspectives of Parents and Providers on Reasons for Mental Health Readmissions: A Content Analysis Study

Abstract

Background and Objectives:

Pediatric hospitalizations for mental health conditions are rapidly increasing, with readmission rates for mental health conditions surpassing those for non-mental health conditions. The objective of this study was to identify reasons for pediatric mental health readmissions from the perspectives of parents and providers.

Methods:

We performed a retrospective content analysis of surveys administered to parents and providers of patients with a 14-day readmission to an inpatient pediatric psychiatry unit between 5/2017 and 8/2018. Open-ended survey items assessed parent and provider perceptions of readmission reasons. We used deductive coding to categorize survey responses into an a priori coding scheme based on prior research. We used inductive coding to identify and categorize responses that did not fit into the a priori coding scheme. All data were recoded using the revised schema and reliability of the coding process was assessed using kappa statistics and consensus building.

Results:

We had completed survey responses from 89 (64%) of 138 readmission encounters (56 parent surveys; 61 provider surveys). The top 3 readmission reasons that we identified from parent responses were: discordant inpatient stay expectations with providers (41%), discharge hesitancy (34%), and treatment plan failure (13%). Among providers, the top readmission reasons that we identified were: access to outpatient care (30%), treatment adherence (13%), and a challenging home (11%) and social environment (11%).

Conclusions:

We identified inpatient stay expectations, discharge hesitancy, and suboptimal access to outpatient care as the most prominent reasons for mental health readmissions, which provide targets for future quality improvement efforts.

INTRODUCTION

Pediatric hospitalizations for mental health conditions in the United States have increased by nearly 50% from 2006 to 2011.¹ Approximately 1 in 10 pediatric admissions have a primary mental health diagnosis, with an annual aggregate cost of $1.3 billion for depression alone.² Readmission rates are reported to be higher for mental health conditions than non-mental health conditions.³ High rates of hospital readmissions have been suggestive of poor quality of care and inefficient use of resources.⁴

Pediatric readmission rates are associated with age, race/ethnicity, insurance status, diagnosis, and geographic location.^3,5,6,7,8 Quantitative studies have also shown that pediatric mental health readmissions are associated with caregiver-child conflict, limited caregiver involvement, history of abuse and neglect, length of initial hospital stay and the receipt and volume of aftercare received.^9–14 However, despite considerable interest in understanding the reasons for pediatric mental health readmissions, prior studies have relied on claims or administrative data to identify reasons for readmissions, or are dated.^3,10–16

In general pediatric populations, qualitative research has been successfully used to identify preventable readmissions and potential targets for readmission reduction.^17–19 A similar approach for pediatric mental health readmissions may provide a nuanced and deeper understanding of the reasons for readmissions, and thus, reveal important targets for quality improvement interventions to reduce preventable readmissions. Furthermore, understanding the family experiences related to readmission will help drive patient-centered quality improvement.

The aim of this study was to identify predominant reasons for pediatric mental health readmissions to an inpatient psychiatry unit from the perspectives of parents and providers by performing a content analysis of survey data.

METHODS

Study Design, Setting and Population

We conducted a retrospective content analysis of survey data obtained from parents and/or legal guardians (henceforth referred to as parents) and attending psychiatrists (providers) of patients with mental health conditions who had an unplanned hospital readmission. Directed content analysis involves tagging a set of texts with codes that are derived from theory or prior knowledge, and as analysis progresses, revising and refining them to achieve the goal of extending or refining existing theory.^20,21

We conducted this study in the inpatient psychiatric unit at a tertiary care, university-affiliated, 403-bed children’s hospital in the northwestern United States. The 41-bed psychiatric unit is staffed by physicians, therapists, and pediatric mental health specialists. The admission process involves a psychiatry staff member, nurse, and physician who meet with the family within 24 hours of arrival. Families are given an education handout along with verbal explanation emphasizing that the average length of stay is 5 to 7 days and the goal of hospitalization is crisis stabilization. Discharge recommendations typically involve outpatient medication management, home safety proofing, and outpatient follow-up care.

Our sampling frame included patients who were discharged from the psychiatric unit during their index encounter and readmitted to the unit within 14 days between May 1^st 2017 and August 1^st 2018. We only included encounters that had qualitative survey data from either the parent or provider. If a patient was readmitted more than once during the study period, we counted each subsequent readmission and corresponding survey responses as separate observations, as the reasons for each readmission may have been independent from one another (n=22 with more than one readmission). We did not exclude any patients based on age or admission or discharge diagnoses, and we did not apply any other exclusion criteria. The study was approved by the hospital’s Institutional Review Board.

Data Collection

We used existing survey data that was initially collected to increase provider awareness of potentially preventable readmissions and to inform future quality improvement work. Survey items were administered to parents by a psychiatric unit support staff member in-person or by phone after the patient was readmitted. The staff member recorded parent responses either verbatim or by paraphrasing in a Research Electronic Data Capture (REDCap) database; therefore, the database contained a mix of responses from a first and third person point of view.^22,23 Providers received an email invitation to complete the survey within 1–3 days of the readmission and they were given 14 days to complete it. Both the admitting attending of record for the readmission and the discharge attending of record for the index admission were invited to complete the provider survey. Providers independently completed REDCap surveys via an email link. Surveys were not required and no incentives were provided.

Survey items between the parent and provider surveys were slightly different. Responses to the following items on the parent survey were used for this study: (1) “Why do you think that your child was readmitted?”; (2) “Could anything have been done to prevent his/her readmission?”; and (3) “If yes, then how?” Responses to the following items on the provider survey were used for this study: (1) “Could this readmission have been prevented?”; (2) “If unable to determine, please explain why” and (3) “What (if anything) could have been done differently to prevent this readmission?” If both the admitting and discharging providers completed a survey for the same readmission (n=10), we only included responses from the admitting provider in our analyses. We made this decision to avoid overrepresentation of certain themes, with the assumption that the admitting provider would have better insight into the range of reasons for the readmission. All survey responses were de-identified and downloaded in a separate data file to allow for the research team to conduct a blinded qualitative assessment.

We used hospital administrative data to collect the following demographic information: age, sex, race/ethnicity, preferred language for medical communication, primary diagnoses (e.g. depressive disorder, disruptive disorder), proximity of patients’ home to the hospital (urban core, suburban, large town, and small town/rural), and patients’ insurance type (Medicaid, private, and other).

Data Analysis

We used content analysis techniques to categorize reasons for readmission from parent and provider responses using a combination of deductive and inductive coding.^21,24 Consistent with a directed content analysis approach, we first conducted deductive coding using the following codes determined a priori from prior literature and expert opinion: discharge hesitancy, access to mental health care, coordination of care, discordant parental expectations from providers, and treatment adherence (Table 1).^{11–13,15,18,25–35}

TABLE 1:

Reason for Readmission Codes and Definitions

Codes	Code Definition
Access to outpatient care	Describes parents’ (and providers’) perception of the availability and accessibility of outpatient psychiatry services (clinic or home-related).
Challenging home environment	Describes parents’ (and providers’) perception of inconsistencies or severe stressors in the home environment. It may relate to a child who is a ward of the state, or in foster care or it may relate to joint custody, involvement of different caretakers.
Challenging social environment	Describes parents’ (and providers’) perception of a social environment trigger (family member, social media, interpersonal relationships, school environment).
Chronic illness	Describes parents’ (and providers’) perception of chronic mental illness or history of trauma.
Coordination of care	Describes parents’ (and providers’) perception of a consistent care plan communicated with their child’s various health care providers. It may relate to disposition planning. It can be specifically related to communication occurring between inpatient provider teams or related specifically to inpatient to outpatient provider communication (i.e. a warm handoff).
Discharge hesitancy	Describes parents’ (and providers’) perception of the time/duration spent in the hospital during the child’s first inpatient stay and their readiness for discharge.
Discordant expectations	Describes discrepancies between parents’ (patients’) and providers’ perception of the patient’s overall inpatient care and/or discharge process. It may also relate to parents’ expectations regarding outpatient behavioral escalation.
Treatment adherence	Describes parents’ (and providers’) perception of A) their own ability or their child’s ability to follow-through on planned treatment recommendations (for inpatient, includes willingness to engage in treatment planning) or B) failure to safety proof items such as guns, ropes, knives, or medications.
Treatment plan failure	Describes parents’ (and providers’) perception of the planned treatment: A) medication failure or B) outpatient therapy concerns.
Unspecifiable/Ambiguous	Includes parent or provider responses that do not fit in any other category or which provide no illustrative comments or are not specific.

Two investigators (SC, TT) coded 30% of the responses (a combination of parent and provider responses) using “yes/no” if the response either fully or partially pertained to an a priori identified code. The investigators used an “other” category for content that did not apply to one of these a priori codes. Responses could be codified with more than one code.

Next, these two investigators inductively coded the “other” category and identified three additional codes: challenging home environment, treatment plan failure, and unspecifiable/ambiguous (Table 1). The full research team reviewed these codes and resolved coding discrepancies in cases when the investigators did not agree on which code to apply. Two investigators (SC, TT) then applied this revised coding scheme to the remaining 70% of the data, with a third coder (KA) resolving coding discrepancies.

After all responses were coded, two investigators (SC, TT) repeated the process of inductive coding of the “other category” and added two additional codes to the coding scheme: chronic illness and challenging social environment (Table 1). A challenging social environment while similar to home environment was added when the team noted repeat mentions of social circumstances (e.g. social media, school) that remained untagged even after challenging home environment was accounted for. We then finalized code definitions as presented in Table 1 and three investigators (SC, TT, KA) subsequently recoded all the responses using this final coding scheme.

Reliability of the coding process was assessed using kappa statistics and consensus building. Initial overall interrater reliability of the first round of coding was adequate (kappa = 0.77; range for each of the original eight codes: 0.48–0.95). During the final review, validity of the results was addressed through reasoning and by stepwise reanalysis by all three investigators, a process that was repeated twice to ensure reproducibility.

Once our review process was complete, we counted codes and used descriptive statistics to identify the frequency of reported reasons for readmission among parents and providers. All statistical analyses were performed in SAS version 9.4 and R version 11.

RESULTS

Of the 138 readmissions, 89 (64%) encounters had survey data (56 [41%] parent survey responses and 61 [44%] provider responses). Of the provider surveys, 36 encounters had responses from an admitting attending (+/− discharge attending) and 25 surveys from a discharge attending only. For encounters with survey data, the mean age of patients at first discharge was 14 years (range: 6 through 18 years old), slightly less than half of the encounters were for female patients, and 43% of encounters had a primary diagnosis of a depressive disorder (Table 2). Compared to encounters with parent survey responses, encounters with provider survey responses had less representation from Black or African American patients, privately-insured patients, and patients from a large town (Table 2). Demographic characteristics of patients who had encounters with survey responses compared to all readmission encounters during the study period are presented in Table 2.

TABLE 2:

Descriptive Statistics at the Encounter Level

	Parent Respondents (n = 56)	Provider Respondents (n = 61)	Sampling Framea (n = 138)
	No. (%)b	No. (%)b	No. (%)b
Mean age in years (SD)	13.6 (3)	13.8 (3)	13.8 (3)
Female	27 (48)	30 (49)	75 (54)
Race/Ethnicityc
White	34 (61)	38 (62)	79 (57)
Black or African American	5 (9)	3 (5)	11 (8)
Hispanic	7 (13)	8 (13)	20 (15)
Asian	1 (2)	0 (0)	1 (1)
Other	9 (16)	12 (20)	27 (20)
Language
English	56 (100)	61 (100)	134 (97)
Spanish	0 (0)	0 (0)	2 (1)
Other	0 (0)	0 (0)	2 (1)
Insurance Type
Medicaid-insured	24 (43)	33 (54)	69 (50)
Privately-insured	32 (57)	28 (46)	68 (49)
Other	0 (0)	0 (0)	1 (1)
Primary Diagnosis
Depressive disorder	24 (43)	26 (43)	59 (43)
Disruptive disorder	16 (29)	15 (25)	39 (28)
Primary Residenced
Urban core	46 (82)	52 (85)	112 (81)
Suburban	5 (9)	6 (10)	16 (12)
Large town	5 (9)	3 (5)	9 (7)
Small town/rural	0 (0)	0 (0)	1 (1)

^aSampling frame refers to all 14-day readmission encounters during the study period.

^bSum of percentages may not add up to 100% due to rounding.

^cIndividuals reporting Hispanic ethnicity were categorized as Hispanic, and individuals reporting non-Hispanic ethnicity were classified by their reported race.

^dDetermined using Tier 4 2010 Rural Urban Commuting Area (RUCA) codes.

We provide a summary of the most common reasons for readmission that we identified based on parent and provider responses below and in Table 3. Additional illustrative quotes from parents and providers for each reason are provided in Table 4.

TABLE 3:

Parent and Provider Reasons for Readmission

Reason	Parent (n=56)	Provider (n=61)a
	No. (%)	No. (%)
Discordant expectations	23 (41)	6 (10)
Discharge hesitancy	19 (34)	4 (7)
Treatment plan failure	7 (13)	2 (3)
Access to outpatient care	6 (11)	18 (30)
Challenging social environment	6 (11)	7 (11)
Treatment adherence	3 (5)	8 (13)
Challenging home environment	2 (4)	7 (11)
Chronic illness	2 (4)	4 (7)
Coordination of care	1 (2)	2 (3)
Unspecifiable/ambiguous	12 (21)	19 (31)

^aIf both the admitting and discharging provider completed two separate surveys for the same patient encounter (n=10), only the admitting provider responses were included in this analysis.

TABLE 4:

Illustrative Quotes for Readmission Reason from Parents and Providers

Reason for Readmission	Parent Quotesa,b	Provider Quotesa
Discordant expectations	“Mom reports that she feels like nothing changed after discharge and that they needed a better plan before leaving. Mom would like migraines and sleep problems addressed during admission.”	“[Readmission might have been prevented] if we helped smooth the transition more; the team felt that they did but from parents’ feedback, we could have done more the day of versus planning days before.”
Discharge hesitancy	“[He] is not the same kid. He is so angry and sad. He is struggling and he did not feel ready last time he discharged.”	“Patient was transferred to an open residential facility bed before he had psychiatrically stabilized… and was within two days kicked out of that residential center. I believe a longer stay on the psychiatric unit prior to a residential care transfer would have prevented [readmission].”
Treatment plan failure	“The high dose of medication was not working for her OCD anymore.”	“The patient was readmitted due to current poor response to medication according to parents which could not be safely adjusted in the community due to escalating and dangerous behaviors.”
Access to outpatient care	“Getting into a [Children’s Long-term Inpatient Program (CLIP)] would have helped prevent this admission.”	“If more intensive outpatient services were available immediately (e.g. subacute residential care, in-home behavioral supports), readmission may have been prevented.”
Challenging social environment	“An outside trigger from social media [wasn’t the only reason for readmission] but it might have impacted it.”	“I think many social factors impacted admission including family visiting when leaving, patient having difficulty reengaging with school, parents’ divorce.”
Treatment adherence	“She was continuing to harm herself, refused to participate in the safety plan.”	“We recommended a dual diagnosis program for substance abuse and mental health after last admission and the patient was not willing to pursue this. He also used marijuana after admission, which we cautioned against.”
Challenging home environment	“He became aggressive again during the hotel stay overnight. The uncertainty of where he is going, multiple hotel stays, desires to be in a more stable environment and doesn’t understand how his behavior is effecting placement.”	“She is a ward of the state. Our recommendation was for a stable caregiver environment, but this was not provided. This is a systems issue. At the time of discharge, there is often no place for wards of the state to go; they may sleep in shelters or stay in a hotel each night then spend the day in the CPS office with their assigned worker. Readmission seems almost inevitable given these circumstances.”
Chronic illness	“The family does not think anything could have been done differently and [the patient was readmitted] due to chronic suicidal ideation.”	“The patient suffers from trauma and will need residential care.”
Coordination of care	“Including the therapist in a planning meeting would have helped our child with outpatient supports.”	“Closer collaboration with outpatient providers [was needed to prevent this readmission].”
Unspecifiable/Ambiguous	“Mental health issues beyond anyone’s control [resulted in readmission].”	“Apparently the patient did well at home for the first 4 days after hospitalization when he out of the blue began to become aggressive again. Therefore, it is unclear what other measures could have prevented readmission.”

Abbreviations: OCD, obsessive compulsive disorder; CPS, child protective services.

^aQuotes were slightly modified to improve clarity.

^bParent refers to quotes that were recorded on behalf of a parent by a psychiatric unit staff member.

Among parent surveys, the readmission reason categories with the highest number of coded responses were: discordant inpatient stay expectations with providers (41%), discharge hesitancy (34%), and treatment plan failure (13%; Table 3). In terms of discordant inpatient stay expectations, parents felt that the care they received during the child’s inpatient stay was different from what they expected it would be, there was confusion around the discharge process, or the follow-up plan was unclear. For example, one parent noted: “We were expecting to meet with a therapist before he was released to go over expectations and to answer any questions that we may have had. We were discharged on a holiday and were told that therapists were not working that day.” For parents, it often appeared that the index admission should have offered more than its described purpose of acute stabilization. In terms of discharge hesitancy, parents felt that the length of stay may have been too short or their child did not feel ready to leave the hospital. One parent remarked, “Maybe if his last admission had been longer, he would not have readmitted so rapidly.” In terms of treatment failure, parents identified medication failures as the most prominent issue. As one parent described: “[His] behavioral outbursts have continued [and] seem to be worse since we switched to lithium. I just don’t know if it’s working…” For parents, we did not identify any examples in which treatment plan failure was related to recommended outpatient therapy.

Among providers, the readmission reason categories with the highest number of coded responses were: access to outpatient care (30%), treatment adherence (13%), and a challenging home (11%) and social environment (11%). Access issues included difficulty obtaining timely psychiatry, therapy, or autism clinic follow-up appointments. Providers also noted access issues related to intensive care services post-discharge such as partial hospitalization, intensive outpatient treatment, residential centers or other psychiatric needs and home services (i.e. applied behavioral analysis). One provider noted: “Readmission likely could have been prevented by the family having access to therapeutic supports in the community. [They are] on waitlists for therapy and psychiatric services…” Treatment adherence, in which the patient did not follow recommended plans of care at home or outpatient therapy, was sometimes described as a precipitating factor for readmission. For example, providers would note that prior to readmission there had been “limited adherence to psychiatric and medical treatment.” Additionally, providers often highlighted social and home environment triggers for readmission. Social environment triggers included interactions with social media, school stressors, and issues with family members. Home environment triggers included housing instability, which was frequently noted. In some cases, both types of triggers were identified to result in readmission. One provider stated: “Conflict with her mother was cited by [the] patient as the reason she experiences suicidal thoughts; she threatens to kill herself if she thinks she needs to live at home, and yet at a youth care shelter she runs away so they cannot consider her a safe resident.”

DISCUSSION

In this study of parent and provider survey responses at the time of readmission, we found that the top reasons for readmissions among parents were discordant inpatient stay expectations and discharge hesitancy; whereas suboptimal access to outpatient care was identified to be the most prevalent reason among providers.

Discordant expectations reflected the perspective of many parents who reported that the index admission should have offered more than its described purpose of acute stabilization. Defining “stabilization” for mental health conditions may be more challenging than for physical health conditions with more easily identifiable criteria to indicate a patient is ready for discharge (i.e. improved oxygen saturation).³⁶ Therefore, discordant expectations about the purpose of hospitalization might be more of an issue for children and youth with mental health conditions compared to the population of children hospitalized for physical health conditions. Mental health experts also observe that aligning expectations at the time of crisis is challenging.³⁷ Prior studies suggest that assessing and aligning care goals through explicit discussions between patients, families, and providers may help to set realistic expectations of the recovery process during and after hospitalization.^18,38 For example, the psychiatry unit set expectations upon arrival through verbal coaching combined with an educational handout. Given the existing power dynamics between patients and providers, future quality improvement efforts should engage parents and providers together in co-designing solutions to mitigate these differences in expectations and foster trusting relationships between families and their care team. Future quantitative studies are also needed to further categorize and assess the prevalence of this construct of discordant expectations as it relates to acute mental health admissions and examine associations between this construct and readmissions in response to iterative quality improvement efforts.

In our analysis, discharge hesitancy was often coded concurrently with discordant expectations; parents often felt that their child was not stable for discharge and their child needed more help. While the relationship between length of initial inpatient stay and mental health readmission has been previously studied with mixed results, there is limited information on its role from the parent perspective.^12,25–27 Our study findings build on prior research related to the association between readiness for hospital discharge and lower readmission rates.^35,39 However, although psychiatry staff advocate on behalf of patients, waiting for parents to feel ready for discharge may not be practical considering payment structures for mental health admissions and hospital capacity.⁴⁰ Alternate payment models which consider high quality of care across the continuum (versus episodic reimbursement) may be needed to ensure adequate resources are available for both inpatient and outpatient care of these patients. Even when the purpose of the hospitalization is clear, extensive discharge education provided, and a robust follow-up plan put in place, parents may still feel anxious about assuming primary responsibility for their child’s illness. Efforts to promote parental self-efficacy during the hospitalization, such as parental safety planning classes (which since the time of this study have become mandatory), crisis management training, and parent peer support groups, may help to reduce feelings of anxiety at the time of discharge.⁴¹ In situations where discharge hesitancy is still present, providers could advocate on behalf of families to support more time to promote self-efficacy and consider opportunities to connect families with outpatient provider teams while the patient is still hospitalized to promote comprehensive and culturally sensitive transitions.

Providers cite suboptimal access to outpatient care as the most prominent reason for readmission, which may represent a different perspective on the same issue of parental discordant expectations about the purpose of the hospitalization or parental discharge hesitancy. As with other conditions requiring hospitalization, providers are more aware and accepting that acute stabilization is the primary goal of these admissions; therefore, they may rely on outpatient providers to take primary responsibility for managing the patient’s chronic health condition. However, suboptimal access to outpatient care could create a desire for parents to want more from the inpatient stay than the hospital is designed to absorb or provide. Studies examining the association between access to outpatient care and pediatric mental health readmissions have produced mixed results.^13,14,16 One reason for conflicting reports regarding the effectiveness of follow-up care in preventing pediatric mental health readmissions is lack of uniformity amongst the studies in terms of type and amount of outpatient care.¹⁴ Future studies should consider examining whether patients who report lower discordant expectations or discharge hesitancy also report better access to high-quality outpatient follow-up care.

Our findings elucidate several targets for quality improvement work to potentially reduce the rate of mental health readmission. First, written or verbal communication strategies to clarify the purpose of an inpatient admission may be beneficial, but the timing of delivery should also be taken into account.^38,42 Efforts to promote parental self-efficacy, consistent communication at the time of discharge, and clear follow-up plans may help to ease hospital-to-home transitions. Given the dearth of outpatient mental health providers,⁴³ access to outpatient mental health care will continue to remain challenging, and time frames for reasonable and possible follow-up should be discussed with families. Innovative interventions are needed to expand mental health care access, including telemedicine, mental health services provided through the internet and related technologies (eMental Health services), and better integration within and between the healthcare system and other key players (i.e the juvenile justice system and the education system). Alternate payment structures, which support high quality care across the continuum, may also play a larger role in these future interventions.

This study has several limitations. Our study was conducted at a single institution; therefore, our findings may not be generalizable to other hospitals or geographic areas. Response rates for both parents and providers were low and readmission encounters with survey data represented more male, White, and English-speaking patients compared to our sampling frame, which resulted in a biased sample. Additional studies with underrepresented groups will be important to conduct in the future. Larger quantitative survey studies would also be able to determine associations between readmission reasons and race, education status, and insurance type. Parent responses were recorded by psychiatry unit staff, but were not transcribed verbatim, which may have introduced some bias, missing information, or misinterpretations.

CONCLUSIONS

Among parents, discordant inpatient stay expectations and discharge hesitancy were the predominant reasons for mental health readmission. Providers perceived suboptimal access to outpatient care as a key driver of these readmissions. Parental perspectives may also reflect limited access to outpatient mental healthcare; thus, parents expected more from the inpatient system than it is currently designed to provide. Quality improvement strategies to clarify and deliver information on the purpose of an inpatient admission and optimize inpatient to outpatient care transitions may help to reduce readmissions for pediatric patients with mental health conditions. Access to effective and possibly novel forms of mental healthcare delivery post-inpatient hospitalization should be further evaluated.