Biofeedback

Subjects were recruited from an obesity treatment center in a university hospital in Taiwan. The obesity treatment center personnel comprised a multi-disciplinary team, and included a surgeon, internal physician, psychiatrist, urologist, obstetrics and gynecology doctor, nurse, case manager, dietician, and physical activity director. The obesity treatments in this center included non-surgical procedures: meal replacement, pharmacotherapy, psychiatric bio-feedback treatment and intra-gastric balloon, and surgery: bariatric surgery (sleeve, band, Roux-en-Y gastric bypass). First of all, the patients made up their mind as to the treatment modality. However, the patients who wanted to receive bariatric surgery had to meet the criteria of morbid obesity. They then needed to undergo a complete pre-operation evaluation, including a psychiatric evaluation. Our hospital has a committee in charge of determining whether the patients are eligible for bariatric surgery.
Patients received a complete physical evaluation during their first visit, and also completed two questionnaires: the Taiwanese Depression Questionnaire (TDQ) and the Chinese Health Questionnaire (CHQ). The TDQ is a 0-3-point, 18-question questionnaire used to screen clinical depressive disorder.
[22 (link)]. The cut-off point in the community population is 18/19 points. The CHQ
[23 (link)] is a 12-question, 2-reverse questions, 0-1-point questionnaire for screening “minor psychiatric disorders” such as anxiety disorder. The cut-off point in community surveys screening minor mental disorders is 4/5 points.
To avoid false negative results, we lowered the cut-off points for the CHQ and TDQ in our clinical practice. Those patients with CHQ <3 and TDQ <13 were regarded as having no psychiatric disorder. If any of the two scores were above the cut-off point (i.e., CHQ ≧3 or TDQ ≧13, or both), the patients would be referred to psychiatrists for further evaluation. The lifetime psychiatric diagnosis was made based on the psychiatrist’s diagnostic interview, using the Structured Clinical Interview for the DSM-IV (SCID).
We recruited all patients that visited the obesity treatment center of E-Da Hospital from January 2007 to December 2010. The exclusion criteria were age younger than 18 years, having incomplete BMI, TDQ or CHQ data, and refusal of psychiatric interview when needed.
All analyses were performed with the Statistical Package for Social Sciences, SPSS Version 17.0. The chi-square test was used to compare differences for categorical variables and the t-test was used to compare differences for continuous variables. The level of statistical significances was 0.05, two-tailed. Logistic regression was applied to examine whether BMI was associated with a psychiatric disorder.
This study was approved by the Institutional Review Board of E-Da Hospital, Taiwan (EMPR-098-073). The study design and performance complied with the Declaration of Helsinki.

The current study was part of a larger heart rate variability biofeedback intervention study (ClinicalTrials.gov Identifier: NCT03458910)^{30 (link)}. Healthy participants without serious medical conditions participated in the seven-week study after providing informed consent approved by University of Southern California’s Institutional Review Board. The current report focuses on the 54 younger and 54 older adults who had blood samples available at both pre and post intervention. Recruiters, blind to condition assignment, allocated participants in waves of around 20 to small groups of 3–6 people such that each group could visit the lab on a common weekly schedule. Each group was randomly assigned to either the Osc+ or Osc− condition. Both conditions involved 20–40 min of daily home practice of HRV biofeedback with opposite goals. Osc+ participants were instructed to maximize their heart rate oscillations using slow paced breathing, while Osc− participants were instructed to minimize their heart rate oscillations using individualized strategies. We had Osc+ participants try five different breathing cycles from 9 to 13 s per breath and selected the pace that produced the largest amplitude oscillations at the breathing frequency (as indicated by spectral power at around that frequency), suggesting resonance between the baroreflex and breathing. For instance, if an Osc+ participant’s resonance frequency appeared to be 10 s (or 0.1 Hz) on the basis of high heart rate oscillations when breathing at that frequency, the participant was guided to inhale for 5 s and exhale for 5 s during their home practice sessions. For Osc- participants, we had them try out a set of self-generated strategies to reduce heart rate oscillations. Their proposed strategies included imagining natural scenes, listening to calming sounds, and closing eyes. Among the strategies, we selected the one whose frequency power was spread over the broad range of frequencies without a dominant frequency peak. Participants in both groups received feedback using performance scores which were calculated to reflect the opposite goals of the two conditions.
The whole study consisted of seven weekly visits. We collected baseline measurements during Week 1 and 2 visits and post-intervention measures during Week 6 and 7 visits. After Week 2 baseline measurements, participants were introduced to biofeedback training and took home a laptop computer connected with an ear sensor which measured their heartbeats and displayed real-time heart rate biofeedback on the screen. Participants were asked to practice their assigned intervention technique at home for at least 20 min every day from Week 2 until Week 7. The whole intervention lasted for 5 weeks from Week 2 through Week 7. However, we collected blood samples in Weeks 1 and 6. Since the intervention began on Week 2 and post-intervention blood draw took place on Week 6, the intervention effects on plasma Aβ and tau levels are based on 4 weeks of practice (Fig. 2). Upon completion of the study, participants were paid for their time and performance. The sample size for the intervention study was determined to detect medium effect size differences between the two groups. While we aimed for 100 younger and 100 older adults, a total of 106 younger and 56 older adults completed the whole HRV biofeedback intervention sessions lasting from Week 1 through Week 7. The number of younger participants from whom we collected blood samples was half of those who completed the whole study, because the blood collection setup was implemented halfway through the collecting of younger adult data. Due to Covid19, data collection for older adults was terminated before reaching the goal of 100. This yielded a final sample of 54 older and 54 younger participants available for plasma assays at both pre- and post-intervention (Fig. 3).Figure 2

Weekly lab visit schedule. Schedules from Week 3 to 5 were not included because the visits were irrelevant to the measures reported in the current study. Detailed descriptions for each week can be found in the main outcome report of the intervention^{30 (link)}.

Figure 3

Flow chart. The phlebotomy took place for a subset of the total participants. The whole intervention had 15 dropouts for younger adults (7 Osc+, 8 Osc−) and 16 dropouts including 6 Covid-induced study halts for older adults (9 Osc+, 7 Osc−). Among 6 Covid-halted cases (all older adults), three participants were included for plasma assays because they completed up to Week 6 sessions including phlebotomy.

The study was approved by the institutional Review Board at University of Southern California (ID: UP-17-00219, Name: HRV-biofeedback and emotion regulation). We obtained written informed consent forms from all participants of the study. All the experiments involving human blood samples were carried out in accordance with relevant guidelines and regulations. The experimental procedures were specified in the Biohazardous Use Application (BUA-18-00027), approved by the Institutional Biosafety Committee at University of Southern California.

To ensure target group participation, the intervention conception was preceded by a qualitative needs assessment of the target group health care workers. We inquired about acceptance of the biofeedback system, the chest strap, and needs and requests for app content to improve need-tailoring and individualization aspects. For the sake of simplicity, we do not report the outcomes of this assessment in this study protocol.