Family Involvement in Psychiatric Hospitalizations: Associations With Discharge Planning and Prompt Follow-Up Care

We examined Medicaid claims and closed medical records for 179 patients who had received Medicaid and had a psychiatric hospitalization at two urban hospitals in New York State in 2012–2013. The project was part of a larger study focusing on psychiatric hospitalization discharge planning for more than 30,000 Medicaid recipients who received Social Security Disability Insurance and/or had a serious mental illness or emotional disturbance. The purpose of the initial phase of the larger study was to examine medical records as a means of assessing the reliability of information reported by Medicaid managed care organizations. We reviewed records at two high-volume urban community hospitals (each with >1,000 yearly mental health admissions) that treat a substantial number of Medicaid recipients (55% of discharges at hospital A and 95% at hospital B). In this article, we present a secondary analysis of the data extracted from medical record review.

We randomly selected 120 individuals from each hospital (N=240) by using a stratification method to ensure sampling of cases for which managed care organizations had reported incomplete discharge planning (e.g., failure to contact a current or prior outpatient provider, schedule a follow-up appointment, or forward a discharge summary to a follow-up outpatient provider). We randomly selected an additional 50 cases from each site for reviewer training; these cases were not included in the analytic sample. Institutional review boards from the research team site and from both hospitals approved the study procedures and granted waivers of consent to allow for retrospective review of closed medical records.

Demographic data were extracted from Medicaid claims. Age was categorized as youth (younger than age 21 years) or adult (age 21 and older). Clinical characteristics extracted from Medicaid claims included length of hospital stay, primary diagnosis at discharge, and co-occurring substance use disorders within the prior 12 months. We created dichotomous variables indicating whether patients attended an outpatient mental health appointment within 7 and 30 days postdischarge. A mental health appointment was defined as any visit to a clinic or specialty behavioral health service licensed by the state mental health authority or any outpatient service with a primary diagnosis of a mental disorder that was provided by a mental health practitioner or physician.

We developed a data extraction tool and guidebook (see online supplement) for reviewers to use in extracting information about family members that was documented in the medical records. Family members were most often a relative; however, our definition of “family or family member” for this study included anyone close to the patient who provided support, and therefore, the supporter was not necessarily related to the patient (e.g., significant others, friends, and foster families). The term did not include individuals who supported the patient as part of his or her paid job (e.g., parole officers and case managers). The guidebook drew from the literature (4, 5, 7) to define specific activities by family, including visiting the patient, speaking by phone with the patient, and participating in a range of interactions with inpatient staff. The variable “any involvement between family and inpatient staff” was created to denote presence of at least one interaction between a family member and inpatient staff. These interactions included discussion about the following: services (inpatient or community) available to the family, the patient’s health or mental health, discharge date, postdischarge treatment plan, postdischarge residence, warning signs of decompensation or ways to prevent readmission, or concerns regarding discharge and/or postdischarge treatment.

The guidebook and rating tool also included instructions for reviewers to document completion of three discharge planning activities performed by inpatient staff: communicating with a prior outpatient provider, scheduling an outpatient mental health appointment, and forwarding a discharge summary to the follow-up care provider. A composite variable called “comprehensive discharge planning” was created to note hospitalizations for which the inpatient staff completed all three activities.

Two reviewers were provided training in rating medical records and then independently rated nine of the records selected for training. The principal investigator and study coordinator rated the same nine records to test for interrater reliability of ratings for the three discharge planning activities. Interrater agreement (kappa statistic) was satisfactory for the three discharge planning activities: 0.77 for contacting a provider, 1.00 for scheduling an appointment, and 0.88 for forwarding a discharge summary. To ensure ongoing reliability, the study coordinator reviewed data extraction forms for all medical record reviews. If there was insufficient evidence to justify a rating, the study coordinator reviewed the case with the reviewer and developed a consensus rating. When consensus could not be obtained, the principal investigator determined the final rating. All medical record coding was performed blinded to the outpatient follow-up status of the patients.

We completed three sets of analyses. First, we created multiple logistic regression models for each family involvement variable and the composite variable any involvement between family and inpatient staff, to examine their associations with three outcomes: completion of comprehensive discharge planning, patient attending an outpatient mental health appointment within 7 days of discharge, and patient attending an appointment within 30 days of discharge. In each regression model, we controlled for hospital site and patient age, gender, and race-ethnicity. Second, we created a set of unadjusted regressions to examine associations between demographic and clinical characteristics with the three noted outcome variables as well as with any involvement between family and inpatient staff. Finally, we created two exploratory regression models to further examine the relationship between any involvement between family and inpatient staff and patient attendance at an appointment within 7 and 30 days of discharge, controlling for comprehensive discharge planning and patient demographic and clinical factors. All analyses were conducted with SAS (version 9.4).

The final sample included 179 unique individuals and discharges: 93 from hospital A and 86 from hospital B. Of the 240 records selected for review, 225 were rated (15 from hospital B were not rated because of time constraints). Forty-six of the remaining 225 records were excluded from the analyses: 22 did not meet inclusion criteria for the main reliability study after review; 14 were readmissions of participants already in the sample; and for 10, the records indicated there were no family members available to engage in care. The demographic and clinical characteristics of the final sample are described in Table 1. Ninety-six (54%) and 139 (78%) of the patients attended an appointment within 7 and 30 days postdischarge, respectively.

The frequencies of specific involvement by family are listed in Table 2. Inpatient staff contacted a family member for 134 (75%) patients. Staff were unsuccessful in attempting to contact a family for another two (1%) patients. There were no documented attempts to contact a family member for the remaining 43 (24%) patients. Seven (4%) had no mention of a family member in their medical records, 19 (11%) had evidence in their records that a family member or other support person existed but no specified reason inpatient staff did not attempt to contact them, 14 (8%) patients refused to involve a family member in their treatment or refused to provide permission for staff to contact them, and three (2%) were unable to provide contact information and/or the staff could not obtain contact information.

The first set of regression analyses is reported in Table 2. Family member interactions with the patient, visits with the patient, attendance at a family therapy session, and communication with inpatient staff about services available to families were significantly associated with patients’ receiving comprehensive discharge planning. Inpatient staff contacting a support person, communicating about the patient’s health or mental health, and discussing discharge-related topics before or after discharge were significantly associated with the patient’s attending a follow-up appointment within 7 and 30 days of discharge.

The second set of regression analyses showed that being under 21 was significantly associated with having more involvement from family (odds ratio [OR]=0.07, 95% confidence interval [CI]=0.01–0.56), as was a length of stay greater than 14 days compared with patients who had stays of 0–6 days (OR=2.49, 95% CI=1.03–5.97). Patients with a length of stay of 7–13 days were significantly more likely to receive comprehensive discharge planning compared with the reference group who had stays of 0–6 days (OR=2.65, 95% CI=1.15–6.11). Co-occurring substance use disorder was significantly associated with no family involvement (OR=0.39, 95% CI=0.2–0.77) and lower likelihood of receiving comprehensive discharge planning (OR=0.46, 95% CI=0.24–0.86). Patients at hospital A (N=45, 48%) were also significantly more likely to receive comprehensive discharge planning than patients at hospital B (N=22, 26%; OR=2.73, 95% CI=1.45–5.13). No demographic or clinical characteristics were significantly associated with attending an appointment within 7 or 30 days of discharge.

The results of the regression model examining the association between any involvement between family and inpatient staff and follow-up care attendance are shown in Table 3. Any involvement between family and inpatient staff was the only variable that was significantly associated with patient attendance at an outpatient follow-up appointment 7 and 30 days after discharge, after analyses controlled for discharge planning and other patient characteristics.

Despite widespread recognition of the importance of family involvement, there is a paucity of information about the types of involvement typically seen during psychiatric hospitalizations. In this study, we examined hospital records to assess frequencies of inpatient staff involving family members in treatment and discharge plans. We found percentages of family members visiting patients, attending a family meeting or therapy session, and discussing the patient’s postdischarge treatment plan with inpatient staff that were similar to those found by Boyer et al. (4). Both studies also showed a significant association between family involvement and linkage to follow-up care, suggesting the validity of these results (4). Previous reports of support by family members elaborated on a range of hospital- and discharge-related topics that families deem important (7) but did not address the frequency with which the topics are addressed by inpatient staff. The most frequently occurring activities by families in our study involved discussions of the patient’s health, mental health, and logistics of discharge planning (discharge date and postdischarge treatment plan and residence). These activities, along with the composite variable any involvement between family and inpatient staff, were positively associated with the patient attending an outpatient mental health appointment within 7 and 30 days of discharge. This finding lends support to the benefits of family involvement in mental health care (1) and adds new evidence that these benefits extend to the psychiatric hospitalization setting. Furthermore, family involvement was associated with patients’ receiving more comprehensive discharge planning, underscoring the importance of family involvement and its impact on treatment.

Our second set of analyses highlighted differences in family involvement, discharge planning, and connection to follow-up care in at-risk populations. Youths were more likely to have family involvement, which may indicate that younger individuals have more contact with family or family is more readily available. Inpatient staff may be more likely to involve families of young patients; previous literature has shown that youths are at high risk for disengagement in care, and the family-therapeutic alliance is a critical protective factor for engagement (8). Individuals with a co-occurring substance use disorder were less likely to have family involvement and to receive comprehensive discharge planning. Family involvement may be lower in this population because of the strained relationships often experienced by individuals with substance use. However, given that co-occurring substance use disorders are a strong predictor of failed transition to outpatient care (9) and the protective factors of family and supportive others’ involvement identified in this study, inpatient staff should consider alternative strategies in order to engage families of individuals with co-occurring substance use disorders. Patients with longer hospital stays were more likely to have family involvement and comprehensive discharge planning than those with shorter stays. This finding is understandable, given that inpatient staff have more time to complete these activities for patients with longer stays. Previous literature has shown that longer stays are associated with increased linkage to outpatient care (10). These findings suggest the need for sustained efforts to provide comprehensive discharge planning and to engage families before discharging patients.

Although medical records may not provide complete details of all events that occur during hospitalization, our reviews indicated the presence of significant family involvement, occurring in 40%−75% of the sample. Olfson et al. (5) examined family involvement during hospitalization by interviewing patients and reported that 16% of families participated in therapy. In this retrospective review of medical records, we found that 40% of families participated in family therapy. Nevertheless, some types of involvement were less frequently reported in the medical records, such as expressions of concern about discharge or follow-up care. This finding suggests that such conversations may be underdocumented and that inpatient staff may need to make an effort to ask about and facilitate problem solving regarding the patient’s transition to outpatient care. A primary barrier to such engagement may be that inpatient lengths of stay are typically brief. However, given the high rates of readmission and failed care transitions in this population, and given the evidence that family involvement is associated with better outcomes, inpatient staff should consider implementing processes to better educate and elicit feedback from family or other supportive persons regarding discharge plans.

Some caveats should be noted and addressed in future research. Medical record documentation should not be considered as fully representing actual activities and interactions, especially given the pace of inpatient care and the lack of uniform documentation standards. For example, phone calls between families and patients were not significantly associated with our main outcomes, but these conversations are likely not reliably reported in medical records. In addition, we do not know the impact of urban versus rural hospital settings. Because families in urban settings may live closer to hospitals and have greater access to alternative transportation, our results may not generalize to rural settings. In addition, this sample was composed of patients receiving Medicaid, and the findings may not generalize to commercially insured patients.

Our regression models suggesting an impact of family involvement beyond discharge planning did not account for other known predictors of successful care transitions, such as housing stability, persistent symptoms, and engagement in care prior to admission. Additionally, patients who have contact with a family member may have unmeasured characteristics that make them more likely to engage in follow-up care; we did not control for those factors in this study. Although there was a correlation between family involvement during hospitalization and follow-up care attendance, we cannot assume causation. Certain families may have a strong influence on aftercare adherence independent of any contact with inpatient staff.

It would be helpful to understand why 14 patients (8% of the sample) refused to involve a family member. It may sometimes be appropriate to limit family involvement. On the other hand, prior research has emphasized that inpatient staff should consider both the patient’s and family members’ points of view regarding treatment and discharge planning when patients refuse to involve their families (5). The potential for positive involvement by family may also depend on which family member is available and the patient’s living situation; this study was limited in that it did not collect data on these two factors. We must also acknowledge the 10 people who had no family available to contact (excluded from analyses) and the three (2%) who could not provide contact information. In cases such as these, hospital staff could elicit support from professionals, such as case managers, and engage with prior and/or future outpatient providers to establish a support network for the patient. Research has shown that patients who meet outpatient providers and/or start outpatient programs before discharge more than triple their odds of successful linkage to follow-up care (4). For individuals without family available or who refuse to involve family, this alternative could be a useful strategy for hospital staff. Finally, for 19 individuals (10% of the sample), there was documentation that family existed but no mention in the record that inpatient staff had tried to contact them and no reason given for not doing so, and for another seven (5%), no family was mentioned in the record at all. These findings underscore the need for specific procedures and expectations regarding staff engagement with patients’ families or other support persons.

This study demonstrated that involving families (or other supportive persons close to the patient) during inpatient care was significantly associated with comprehensive discharge planning and follow-up care attendance at 7 and 30 days postdischarge. This study suggests that even a low threshold of involvement is significant. Future studies should further examine key family involvement activities and associations with improved outcomes to determine whether a higher threshold of involvement is needed for these activities. Involvement of patients’ families is an inexpensive intervention, and although it is standard care for hospital staff to contact and involve families, such interactions do not always occur. Hospitals should formalize efforts to educate staff about the benefits of involving families in treatment and should implement standard procedures requiring contact and communication. These procedures should include specific steps to facilitate family engagement beyond contact and to establish alternative approaches for at-risk patients, including individuals with substance use disorders and those with no family available. Further research could then examine the impact of implementing such requirements.