Ethical clearances were obtained from the Metro South Hospital and Health Service Human Research Ethics Committee at Princess Alexandra Hospital and the Human Research Ethics Committee at The University of Queensland (Australia). It was a two-group quasi-randomised controlled trial, conducted from May to November 2016. Outcome measures were assessed at baseline and 12 weeks after intervention. Participants with T1D were randomly allocated to the Tai Chi intervention, or to the usual care control group, based on recruitment date. Due to logisitic constraints, the first six participants were assigned to the intervention group. The control group was offered the Tai Chi program at the end of the study. We recruited participants through referral from endocrinologists at the Department of Diabetes and Endocrinology of the Princess Alexandra Hospital, as well as letters of invitation to members of Diabetes Australia Queensland. Inclusion criteria were: type 1 diabetes with HbA1c of 7.5% - 10%, on stable insulin therapy; aged 18 - 70 years; no health or injury problems that would prevent doing the exercises; able to attend intervention sessions 2 times per week for 12 weeks; and living in Brisbane. From 39 potential participants, 13 were eligible to participant in the study (see Figure 1). Before attending the baseline assessment, all participants were cleared by their Endocronologist/General Practitioner, and signed the consent form to participate in the study.

During the 12 week study, the Tai Chi group attended group training twice a week, conducted by an experienced Tai Chi instructor. The intervention group also practiced at home using the Tai Chi program DVD provided by the study, on days when they did not attend the group training. The exercise employed in this study was KaiMai Tai Chi style [13] . Each training session consisted of

warm up, practice and cool down and lasted for around 1.5 hours. The intensity of the training was individualized according to each participant’s health condition.

All participants provided written informed consent prior to baseline assessment. The baseline assessment was conducted at the Princess Alexandra Hospital by an independent research assistant (including blood sample collection, questionnaire and physical measures), followed by the same assessment at 12 weeks.

The primary outcomes included HbA1c (assessed from a fasting venous blood sample) and depressive symptoms (using the short form of the CES-D Depressive Scale 10) [14] . The secondary outcomes included physical measures and quality of life (QOL). Waist circumference, height and weight were measured using standard protocols [15] . Body mass index (BMI) was calculated as weight (kg) divided by height squared (m²). Resting blood pressure was measured using standard protocols [16] [17] . QOL was assessed using the Medical Outcomes Study 36-Item Short-Form Health Survey (SF-36) [18] . Leg strength was assessed using a chair-stand test (number of stands completed in 30 seconds) [19] .

The questionnaire asked about age, gender, country of birth, family history of diabetes, language spoken at home, depressive symptoms, and health-related quality of life. The acceptability and feasibility of the Tai Chi program were also assessed using open-ended questions at post-intervention including perceived benefits of and barriers to participating in the program, and comments on the Tai Chi program DVD provided for at-home practice.

The group training instructor recorded the group session attendance, and reasons for non-attendance at each class. The at-home practice was recorded by the participants themselves using a diary during the intervention.

All the statistical analyses were conducted in SPSS (Version 23). 13 participants participated in this study, with six assigned to the intervention group and seven to the control group. One from the intervention group and two from the control group were lost to follow up. Primary analyses were conducted using the Expected Maximization (EM) method to estimate missing values. Secondary analyses were also conducted using treatment-received analyses. Descriptive statistics were used to characterize participants at baseline and follow up. Independent samples t-tests were used to assess between group differences at baseline in outcome variables. One way repeated measures ANOVA was used to assess differences between groups in changes over time in each of the outcome variables.

Thirteen participants (6 men and 7 women, aged 24 - 63 years old) participated in this pilot study. All except three were born in Australia, but all spoke English at home. The average age of diagnosis with T1D was 26 (range 5 - 48) years, with an average history of T1D of 16 (2 - 31) years. HbA1c levels ranged from 7.5 to 9.2%.

Adherence to the program was good, with all intervention participants except one retained during the 12 week program (Figure 1). On average participants attended 83% of the group classes, with absence mainly attributed to competing family or work commitments. Two participants in the control group dropped out during the study (Figure 1).

Participants reported both physical and psychological benefits from the program. Perceived physical benefits included “knees feeling better” and “getting sore less”, “starting to get thigh muscles”, “more energy”, “sleeping better”, “improved flexibility” and “improved wellbeing”. Reported psychological benefits included feeling “happier”, “more calm”, “better concentration”, and “more confidence in managing and talking about T1D”. Participants reported “feeling more positive and more active”, “improved coping” and “less stress” (particularly at work). All participants reported that the DVD was useful for their learning and training, with comments indicating that it assisted with memorising the movements.

There were no between group differences in the demographic or outcome variables at baseline.

Changes in HbA1c and depressive symptoms from baseline to post-intervention in each group are shown in Table 1 and Figure 2. There was no change in HbA1C in either group. There was however a small but significant improvement in depressive symptoms in the intervention group (p < 0.01) and a concomitant worsening in the control group (p < 0.05) (see Table 1). Individual data showed an improvement in depressive symptoms in 4 of the 6 intervention participants and slightly worsening symptoms in all but one of the control group (see Figure 2).

Changes in physical measures and quality of life from baseline to post-intervention are also shown in Table 1 and Figure 2. The only between group differences in changes over time were in waist circumference (p = 0.059), mental components summary score (p = 0.051) and leg strength (p < 0.05). Although there was no significant improvement in waist circumference in the intervention group, this measure increased significantly (worsened) in the control group (p < 0.05). There were improvements in the mental health summary score (p < 0.05) and leg strength measure in the intervention group (p < 0.01) but not in the control group (Table 1).

^#Results from Expected Maximization method of analysis; * Significant or borderline significant differences in changes over time.

Individual data indicated variable changes in BMI in both groups, with small improvements in 4 of 6 intervention, and 2 of 7 control participants (Figure 2). Similarly, while there were improvements in leg strength in 5 of the 6 participants in the intervention group, leg strength also improved slightly in 4 of the control group participants. However, the improvement was statistically significant in the intervention group, but not in the control group (Table 1).

The results described above were based on the primary Expected Maximization analysis method. These findings did not change with “treatment received” analyses, except that the latter demonstrated significant between-group effects in favour of the intervention group in HbA1c (p < 0.05) and waist circumference (p < 0.05).

Three participants made changes to their medication during the study. One participant in the intervention group changed one brand of insulin (Humalog) to another (Novorapid), but with the same dose. Two control participants made changes to their insulin regimens, by changing the dose or type of insulin used.