Factors Associated with Readmission in Patients with Eating Disorders

Abstract

Purpose:

This study examined predictors of readmission to inpatient/residential settings in patients with eating disorders. We hypothesized that readmitted patients would report worse transitional care and continuity of care compared to those who were not readmitted.

Methods:

We conducted a cross-sectional, online survey of 80 individuals from the US who reported that they were diagnosed with an eating disorder requiring inpatient/residential treatment. Participants completed questionnaires on demographic and clinical characteristics, the Care Transition Model-15, and the Continuity of Care Questionnaire. We analyzed data using univariate statistics and a series of logistic regression models.

Results:

Participants who reported better transitional care (adjusted OR (AOR)=1.14; p<0.001), continuity of care during their inpatient or residential stay, including greater transfer of information (AOR=6.39, p=0.002), relationships in the hospital (AOR=6.83, p=0.03), management of follow up (AOR=3.41, p=0.02), management of communication (AOR=8.74, p=0.001) and forms (AOR=9.61, p=0.01), reported increased odds of being readmitted to an inpatient or residential treatment facility. Use of nasogastric (NG) tube feedings was significantly associated with being readmitted.

Conclusions:

Contrary to our hypotheses, we found that better transitional care and continuity of care were associated with higher odds of readmission.

Level of Evidence:

Level III, case-control analytic study

1. Introduction

The lifetime prevalence of anorexia nervosa (AN), bulimia nervosa (BN), and binge eating disorder (BED) are 0.9%, 1.5% and 3.5% among women, respectively [1]. Patients can have devastating consequences if they are not treated properly. These include short- and long-term physical, mental, emotional, and financial sequelae such as sudden cardiac death, osteoporosis, neurocognitive changes, and long-term cardiac structure changes [2].

Inpatient hospitalization or residential treatment is required for patients who are medically or psychiatrically unstable. Patients may be referred to in-patient or residential treatment by a clinician such as their primary care provider, psychiatrist, dietitian, or therapist. Unfortunately, relapse is common. One of every four patients with AN who are hospitalized will be readmitted, [3] primarily due to failure to maintain weight after discharge or reoccurrence of severe eating disorder psychopathology [4]. Few studies have identified factors related to readmission for eating disorders. One study identified younger age, higher abnormal eating attitudes, and lower rate of weight gain as risk factors related to readmission [5]. Additional factors that may contribute to readmission include “longer length of hospitalization, psychiatric comorbidity, purging behavior, earlier age of onset, age at admission, duration of AN, degree of malnutrition, family problems, perceptual body image distortion, and body dissatisfaction” [6]. It is also noted that the failure to reach the established target weight in full during an inpatient stay increased the risk for relapse and subsequent readmission [6]. Knowledge of factors related to readmission can help to improve care.

Discharge is a crucial process to help with the transition from a residential or inpatient setting back to the community. There is little literature exploring transitional care and continuity of care for those diagnosed with eating disorders or other mental health conditions [7, 8] and the effects on readmission. Transitional care refers to “a broad range of services and environments designed to promote the safe and timely passage of patients between levels of health care and across care settings” [9]. Continuity of care is “the patient’s experience of a coordinated and smooth progression of care” [10]. In previous studies in patients with mental health conditions, some studies have shown that improved continuity of care and transitional care in patients discharged from a psychiatric hospital were associated with reduced risk of rehospitalization. [11, 12, 13] However, not all studies have demonstrated this relationship [8, 14, 15]. Increased service utilization has also been associated with increased risk of readmission for mental health conditions [12]. These processes have yet to be explored in patients with eating disorders. Discharge criteria for eating disorders vary between levels of treatment, but typically center around physiological parameters and medical stability, attainment of a target weight or weight restoration, and family education and being able to select meals [16]. However, there is a lack of evidence and guidance available about the discharge process in patients with eating disorders and how transitions in care influence patient outcomes. Even though patients with eating disorders likely need continued support after they are discharged, little is known about services they receive after they return to the community. The purpose of this exploratory study is to evaluate patients’ views of the transitional process from hospital to home for patients with eating disorders and how this relates to hospital readmission.

2. Methods

2.1. Study Design and Participants

This was a cross-sectional study that used an online survey of individuals who reported being diagnosed with an eating disorder requiring inpatient or residential treatment. Participants who were eligible for the study needed to be 18 years of age or older, reside in the US, and be formally diagnosed with an eating disorder that required an inpatient or residential treatment stay within the past three years. A total of 80 individuals were recruited for the study. We stratified the sample such that 40 participants had not ben readmitted and 40 had been readmitted. The sample had a mean±SD age of 31.3±7.5 years and BMI of 24.4±6.8 kg/m². Forty-four participants were female (55%), thirty-five were male (43.75%), and one individual identified as non-binary (1.25%).

2.2. Procedures

The advertisement and link to the survey were posted to eating disorder support pages and advocacy sites, with permission from the sites, as well as to online Facebook groups. Interested participants were first provided with an electronic informed consent form. Those who agreed to participate were then asked screening questions. The screening questions included asking about participants’ ages, whether they lived in the US, if they had been diagnosed with an eating disorder that required inpatient or residential treatment, the year of their first admission, and if they were readmitted. Eligibility requirements included: being 18 years or older at the time of completing the survey, residing in the US, having been diagnosed with an eating disorder requiring inpatient or residential care with their first admission occurring in or after 2015. Eligible participants continued to the rest of the survey. “Catch” questions were included throughout the survey and accurate responses were required to be counted in the results. Participants’ responses were screened prior to inclusion based on consistency of responses, the time taken to complete the survey (roughly 20–30 minutes), missing information, scrutinization of provided email address, and completeness of the survey. Surveys that were completed in abnormally short periods of time, had apparent patterns, responses that provided answers only for mandatory questions, and where individuals submitted multiple responses under the same email address were excluded. Participants were compensated with a $25 Amazon gift card for their time. This study was approved by the University of Pennsylvania’s Institutional Review Board.

2.3. Measures

2.3.1. Demographic information

Demographic information was collected from participants including age, height, gender, and race. Self-reported weight was also collected.

2.3.2. Eating disorder hospitalization and discharge information.

Participants were asked questions to retrospectively recall information regarding their primary eating disorder diagnosis at time of hospitalization; age of eating disorder onset; type of treatment (inpatient versus residential); duration of stay; insurance coverage; age at time of hospitalization; and admission status (voluntary versus involuntary). We also asked about what treatments or therapies participants received while in treatment (i.e., nasogastric (NG) feeding tube, cognitive behavioral therapy, dialectic behavioral therapy, family therapy, group therapy, and/or nutrition services). Participants were asked whether they felt they received adequate treatment during their stay on a 5-point scale from “Strong Agree” to “Strong Disagree”.

2.3.3. Care Transition Model-15 (CTM-15).

Transitional care was measured using an adapted version of the Care Transition Model-15 (CTM-15). The CTM-15 is a fifteen-question survey asking patients about their experiences during discharge and the quality of their transition in care from the inpatient to outpatient setting [17]. We asked participants to refer to the first time they were hospitalized for an eating disorder to standardize across individuals. The survey asked about patients’ knowledge and understanding of their illness, follow-up appointments with providers, and an understanding about medications being taken. Ratings were made on a 5-point response scale (i.e., strongly disagree, disagree, neutral, agree, and strongly agree) or “Don’t Know/Don’t Remember/Not Applicable” [18]. A single total score ranging from 0 to 100 can be calculated with higher scores indicating a better care transition. The questionnaire has been well-validated [18] and used in various patient populations such as elderly adults with complex care needs [19] and joint arthroplasty patients. [20]. The Cronbach’s alpha in this sample was 0.93.

2.3.4. Continuity of Care Questionnaire.

The Patient Continuity of Care questionnaire was used to assess the continuity of information, relationships, and care received after their first hospitalization [21]. This questionnaire included six subscales that measure perceptions of relationships with providers in the hospital; information transfer to patients; relationships with providers in the community; management of written forms; management of follow-ups; and management of communication among providers. In previous studies, the instrument had good internal reliability and validity for assessing patients’ perceptions of factors crucial to continuity of care [18]. However, the Cronbach’s alphas for this study for each subscale were: relationships with providers in the hospital=0.76; information transfer to patients=0.79; relationships with providers in the community=0.57; management of written forms=0.65; management of follow-ups=0.49; and management of communication among providers=0.53.

2.3.5. Eating Disorder Examination Questionnaire (EDE-Q 6.0).

The EDE-Q provided a measurement of the overall severity and range of the features of eating disorders [22]. It measured the frequency of eating disorder behaviors and psychopathology, including global eating disorder psychopathology, restraint, and weight and shape concerns. The Cronbach’s alphas for each subscale were: restraint=0.89; eating concern=0.72; shape concern=0.86; weight concern=0.76; global eating disorder psychopathology=0.95.

2.3.6. Transitional care outcomes.

Participants were asked whether they received transitional care or supportive measures after their only or most recent discharge. Measures included referrals to outpatient treatment, support groups, nutritionists, nutritional support, group psychotherapy, individual therapy, and medication, as well as an option to fill in any additional supports or measures they received. They were also asked to rate how helpful they found each transitional care measure they received on a scale from “Very helpful” to “Not helpful at all.”

2.4. Statistical analysis

Data analyses were conducted using univariate statistics and a series of logistic regression models. Logistic regression models were employed to compare differences in transitional care outcomes and continuity of care during their treatment stay between those who were readmitted versus those who were not readmitted. Analyses were adjusted for demographic and clinical characteristics including age, gender, race (white vs non-white), age at hospitalization/residential treatment, change in weight during treatment (computed from self-reported pre and post-discharge weights) and eating disorder diagnosis. Significance was set at a p-value of <0.05.

3. Results

3.1. Participant characteristics

The sample had a mean±SD age of 31.3±7.5 years and BMI of 24.4±6.8 kg/m². Of the sample, 55% were female and 75% were white, and 38.8% were diagnosed with AN, 22.5% with BED, 20.0% with BN, and 18.7% with other specified feeding or eating disorder (OSFED). (Table 1) The average age at eating disorder onset was 27.1±8.0 years and the average age at first admission was 28.9±7.2 years. Most participants reported being admitted to an inpatient unit (87.5%) and being admitted voluntarily (85.0%).

Table 1.

Demographic, clinical, and treatment characteristics

	Total (N=80)	Readmitted (N=40)	Not Readmitted (N=40)	p-value
Age, mean ± SD, years	31.3±7.5	32.53±8.9	29.98±5.6	0.13
Gender, N (%)
Female	44 (55%)	23 (57.5%)	21 (52.5%)	0.57
Male	35(43.75%)	17 (42.5%)	18 (45%)
Non-Binary	1 (1.25%)	0 (0%)	1 (2.5%)
Race/Ethnicity, N (%)				0.10
White	60 (75%)	29 (72.5%)	31 (77.5%)
Hispanic/Latino	13 (16.25%)	10 (25%)	3 (7.5%)
Black/African American	2 (2.5%)	0 (0%)	2 (5%)
Asian/Pacific Islander	1 (1.25%)	0 (0%)	1 (2.5%)
Native American/American Indian	0 (0%)	0 (0%)	0 (0%)
More than One Race	4 (5%)	1 (2.5%)	3 (7.5%)
Weight, mean ± SD, lbs	158.9±42.7	160.9±45.9	156.82±39.8	0.67
Body mass index, mean ± SD, kg/m2	24.4±6.8	24.8±7.5	24.0±6.0	0.60
Eating Disorder Diagnosis				0.05
Anorexia Nervosa	31 (38.75%)	16 (40%)	15 (37.5%)
Bulimia Nervosa	16 (20%)	11 (27.5%)	5 (12.5%)
Binge Eating Disorder	18 (22.5%)	10 (25%)	8 (20%)
†OSFED	9 (11.25%)	3 (7.5%)	6 (15%)
Orthorexia	6 (7.5%)	0 (0%)	6 (15%)
ǂCTM Scores, mean ± SD
	68.3±15.7	77.1±11.7	59.4±14.1	<0.001
ǂǂEDE-Q Scores, mean ± SD	2.3±1.1
Global eating disorder psychopathology		2.4±1.0	2.3±1.1	0.86
Restraint		1.9±1.4	2.0±1.2	0.72
Eating concern		2.2±0.9	2.1±1.1	0.71
Weight concern		2.7±1.1	2.7±1.2	0.97
Shape concern		2.4±1.2	2.5±1.2	0.45
Continuity of Care Scores, mean ± SD
Relationships with providers in hospital	3.8±0.6	3.9±0.6	3.7±0.5	0.05
Information transfer	3.8±0.6	4.0±0.6	3.6±0.7	0.01
Relationship with providers in community	3.8±0.6	3.9±0.6	3.5±0.5	0.00
Management of forms	3.9±0.7	4.1±0.7	3.8±0.7	0.04
Management of follow-up	3.9±0.7	3.9±0.6	3.7±0.7	0.10
Management of communication among providers	3.9±0.6	4.1±0.6	3.6±0.6	0.00

^*Mean±SD or N(%)

^†Other Specified Feeding and Eating Disorder

^ǂCare Transition Model-15

^ǂǂEating Disorder Examination Questionnaire

3.2. Univariate analyses

In univariate analyses of participants who were and were not readmitted demonstrated, there were no significant differences in age, gender, race/ethnicity, weight, eating disorder diagnosis, or eating disorder psychopathology (ps>0.05). Compared to participants who were not readmitted, those who had been readmitted reported significantly better care transition scores and continuity of care. (Table 1) Participants who were readmitted were more likely to have received NG feeds while in an inpatient or residential facility (p=0.002). However, those who were or were not readmitted did not differ on other treatments received (Supplement 1). Relative to participants who were not readmitted, those who were readmitted were more likely to strongly agree (10% vs 27.5%) or agree (65% vs 50%) that they received adequate treatment during their inpatient stay (p=0.004).

Readmitted participants had higher rates of support group referrals (n=17), nutritionist referrals (n=23), nutritionist support (n=15), individual psychotherapy (n=12), and prescriptions for medications (n=13). Referrals to outpatient programs upon discharge were equal across both groups (n=9), while those who were not readmitted reported higher rates of receiving group psychotherapy (n=11) than those requiring readmission (n=5).

3.3. Multivariate analyses

After adjusting for demographic and clinical characteristics, increased readmission odds were associated with better transitional care (adjusted OR (AOR)=1.14; p<0.001; Table 2), and better continuity of care during their hospitalization (i.e., relationships in hospital (AOR=6.83, p=0.003) and community (AOR=31.04, p<0.001), information transfer (AOR=6.39, p=0.002), management of forms (AOR=9.61, p=0.002), management of follow ups (AOR=3.41, p=0.02) and management of communication (AOR=8.74, p=0.001)).

Table 2.

Logistic regression results of factors related to readmission for eating disorders

Factors	AOR*	95% CI	p
Transitional care	1.14	1.07, 1.22	<0.001
Relationships in hospital	6.83	1.96, 23.77	0.003
Relationships in community	31.04	5.77, 166.97	<0.001
Information transfer	6.39	2.00, 20.45	0.002
Management of forms	9.61	2.35, 39.36	0.002
Management of follow up	3.41	1.21, 9.64	0.02
Management of communication	8.74	2.45, 31.17	0.001

^*Note. Analyses were adjusted for age, gender, race (white vs non-white), weight change during treatment, age at eating disorder diagnosis, and eating disorder diagnosis.

4. Discussion

In this study, our aim was to determine factors that may predict readmission to inpatient or residential treatment facilities. Based on previous literature demonstrating that better transitional care and continuity of care are associated with decreased readmissions in other chronic diseases, [11–13, 23] we hypothesized that individuals who reported poorer transitional care outcomes would have higher rates of readmission for an eating disorder. Our hypotheses were not supported. We found that better transitional care outcomes were associated with higher odds of being readmitted to either an inpatient or residential treatment facility. This was a paradoxical effect that we were not expecting. These findings suggest that it is possible that individuals who have positive experience once they are admitted may be more likely to seek help, or be encouraged to seek help, for a relapse in symptoms. It is possible that that presence of better transitional outcomes may indicate that there is greater availability for funding treatment and support programs, which may in turn, also predict a higher likelihood of funding for individuals requiring additional hospitalizations. However, further research is needed to test this hypothesis.

There are several strengths to this study including the nationally representative sample of participants and individuals with different types of eating disorders. Despite the strengths of this study, there are limitations that must be addressed. Given the three-year period in which participants could have been admitted, it is possible that information regarding their weights, experiences, lengths of stay, and received services could have been impacted by recall bias, as we were asking participants to recall these events retrospectively. People who had better experiences with treatment or those who were more engaged with services may have been more inclined to complete the survey. In addition, the current sample size is small, and the confidence intervals are wide, likely, in part, due to the heterogeneity of eating disorders included in this sample. Further replication of these results in larger samples and with specific eating disorder types is needed.

We did find that the use of NG tube feedings had a greater association of participants being readmitted than those who did not receive feedings via NG tubes. It is possible that it is due to these patients having lower BMIs and/or more severe eating disorder symptoms at admission. Previous research has shown that those who presented with low BMIs and a higher eating disorder severity rating, as indicated by the Diagnostic Statistical Manual (DSM)-5, tended to have a greater need and utilization of services, including an increase in the number of hospitalizations [24]. In addition to lower BMIs, those who reported a family history of AN, the presence of paternal alcoholism, an eating disorder in infancy, episodes of periodic overactivity, and a low weight increase during an individual’s first admission were at an increased risk of being readmitted [25]. However, additional research would need to be conducted to further understand why these patients are at a greater risk of requiring additional admissions to inpatient or residential treatment facilities.

In conclusion, we found that participants who were readmitted had higher scores among all continuity of care measures, as well as transitional care measure scores. In respect to transitional care supports and services, those who were readmitted generally had higher levels of services and treatments provided than those who were not readmitted across most categories. Future research that looks more closely at the different transitional care supports at pre-discharge, post-discharge, and bridging components and the influence of those services on readmission rates as well as differences between eating disorder diagnoses would be valuable.

5. What is already known on the subject?

There has been little research conducted evaluating transitional care from residential or inpatient treatment to the home environment for patients with eating disorders. Given the importance of continued support upon discharge, little is known about the services patients receive upon discharge.

What does this study add?

As a result of this study, we now know that many individuals who reported positive experiences had greater incidences of readmission and that there is a disconnect between patients receiving referrals for outpatient support services and their ability to access them. Many factors may contribute to this disconnect, such as a lack of funding, insurance type and coverage, employment that either has hours that are not conducive to seeking outpatient treatment or that may not provide benefits, and lack of family support.