Self‐selection bias in eating disorders outcomes research: Does treatment response of underweight research participants and non‐participants differ?

Abstract

Observational treatment studies provide a valuable alternative to RCTs but are often criticized due to potential self-selection biases. Studies comparing those who do and do not participate in research on eating disorder treatment are scarce, but necessary to evaluate the impact of self-selection bias on outcomes. All consecutive underweight adult first admissions (N = 392) to an integrated inpatient (IP)-partial hospital (PH) behavioral specialty program were invited to participate in a longitudinal study of eating disorder treatment. Demographic and hospital course data were collected on participants (n = 234) and non-participants (n = 158). Participants and non-participants had similar BMI at admission, lengths of stay, and weight gain rates. Participants were less likely than non-participants to end treatment prematurely from IP and were discharged at a higher BMI; the effect size was small. Few differences in hospital course were observed between participants and non-participants. Although participants were more likely to transition to PH and were discharged at a higher BMI, completion of step-down to PH within this integrated IP-PH program rather than research participation status at admission was a better indicator of discharge BMI, which remains the strongest predictor of long-term weight-maintenance in eating disorders.

Discharge Body Mass Index, Not Illness Chronicity, Predicts 6-Month Weight Outcome in Patients Hospitalized With Anorexia Nervosa

“…Participants compared with non-participants gained weight more quickly [mean (SD) kg/week = 2.0 (0.87) vs. 1.8 (0.77); p = 0.017] and were more likely to attend the partial hospital component of treatment (74.4 vs. 51.79%; p < 0.001), and therefore had a slightly but significantly higher final program discharge BMI [mean (SD) kg/m 2 = 20.0 (1.84) vs. 19.3 (2.26); p = 0.046]. These results are consistent with our previously reported findings that the effect sizes of differences between participants and non-participants in our outcomes research project are small (40).…”

“…Additionally, the mean discharge BMI of non-participants was also above 19 kg/m 2 ; suggesting their hospital course, at least with respect to weight restoration, was similar to that of participants. We have previously shown that differences between participants and non-participants in our longitudinal treatment study are likely to exert at most small effects on outcome (40). The integrated inpatient-partial hospitalization program described herein is designed to achieve rapid weight restoration, and we cannot exclude that some patients actively seek this aspect of treatment, however we have previously reported high levels of perceived coercion regarding hospitalization endorsed by patients at program admission with one third denying the need for hospitalization (44,45).…”

Hospital course of underweight youth with ARFID treated with a meal‐based behavioral protocol in an inpatient‐partial hospitalization program for eating disorders

“…We chose to include all consecutive admissions rather than selecting those patients who agreed to participate in our longitudinal study and who completed self‐report questionnaires assessing these symptoms. We have, however, previously published data comparing consenters and nonconsenters to our longitudinal outcome study and found that groups did not differ on admission BMI, length of stay, or rate of weight gain (Schreyer, Redgrave, Hansen & Guarda, ). Since our sample focuses on underweight individuals with ARFID, treatment response may differ in those with ARFID who are normal weight or above and our cohort was older than those included in previous studies, which may impact generalization of results to younger children and adolescents Lastly, as multiple statistical tests were conducted, it is possible some of these findings are due to chance.…”

Autopsy of a failed trial part 2: Outcomes, challenges, and lessons learnt from the DAISIES trial

“…For example, recruiting a target sample ( n = 242) from 10 sites for the Anorexia Nervosa Treatment of Out‐Patients (ANTOP) study took 4 years (Zipfel et al., 2014). Recruitment to studies involving hospital admission might bring additional obstacles, especially in the case of anxious or ambivalent patient attitudes towards recovery (Schreyer et al., 2017). Nonetheless, two previous RCTs targeting in‐patients with AN successfully recruited 178 patients and 268 caregivers (Hibbs et al., 2018) and 371 patients plus 371 carers respectively (Cardi, personal communication), albeit prior to the pandemic.…”