The questionnaire data and relevant data of outcome indicators will be collected by the investigator by using a unified guidance. The following six collection times will be used: baseline (T0), 3 months (mid-intervention: T1), 6 months (end of intervention: T2), 12 months (T3), 24 months, (T4) and 36 months (T5). At T0, health managers will explain the purpose, procedure, possible benefits and risks of participating in the study. Moreover, they will also obtain written informed consent (blood collection and scales) for participating in the study.

Sociodemographic data will include information on sex, age, marital status, education, employment, region of residence, monthly income level, smoking status and drinking status.

Diabetes-related medical data will include diagnosis, disease duration in years, illness information, treatment information, alcohol use and tobacco use.

The primary outcome will be the incidence of newly diagnosed diabetes at the end of the study (T2, T3, T4 and T5), as defined by the following international criteria: abnormal FBG (≥7.0 mmol/L) and abnormal 2-hPG (OGTT; ≥11.1 mmol/L). The conversion rate will be calculated using the following formula: conversion rate=(cases converted to newly diagnosed diabetes/total cases of each group)×100%.

The following secondary outcomes will be evaluated: HbA1c, waist circumference, hip circumferences, weight, BMI, dietary compliance, physical activity (PA) compliance, self-efficacy, social support and quality of life.

First, InBody520 body composition analysis will be used to measure the current weight, target weight, weight control, fat control, muscle control and BMI. Waist circumference and hip circumference will be measured by professionally trained medical staff with a soft tape measure. Blood will be collected and assessed for HbA1c, blood lipids and serum insulin levels.

Second, dietary compliance will be assessed using a dietary behaviour questionnaire for diabetes and risk groups. The questionnaire consists of 18 items and four dimensions, including eating habits, cooking habits, self-regulation, adverse preferences, slowing sugar increase rate, preventing hypoglycaemia and controlling consumption of vegetables, fruits and meat. Higher Likert scores indicate higher patient dietary behaviour compliance, and the Cronbach’s α coefficient of this scale is 0.763.

Third, PA compliance will be measured using the Long Form of International Physical Activity Questionnaire (IPAQ-L), which comprises five domains (work or study, transportation, household duties, sports leisure and sitting). The respondents will be asked to recall their PA in the last week, and based on the survey results, the total energy expenditure of PA for 7 consecutive days will be calculated and expressed in the form of a metabolic equivalent (MET). Each MET represents the amount of energy consumed per minute for one person at rest. According to the energy consumption evaluation of each PA obtained by the survey, the total energy consumption of the PA of the subject for 7 consecutive days will be calculated using the following formula: MET of different types of PA×activity time (min)×frequency per week=total energy expenditure (MET-min) of patients for 7 consecutive days. The PA level of the respondents will be divided into the following three groups: PA-deficiency group, moderate-PA group and high-PA group.25 Based on the translation of the IPAQ-L into a Chinese version by Qu et al,26 the retest reliability of the Chinese version IPAQ-L is 0.689–0.934. The criterion for the validity of acceleration is 0.50, and the correlation coefficient with energy expenditure recorded in PA is 0.821, indicating that the questionnaire has good reliability and validity.

Fourth, the self-efficacy of patients with pre-diabetes will be evaluated using the General Self-Efficacy Scale (GSES), which includes 10 items rated on a 4-point Likert scale, ranging from 1 (not true at all) to 4 (exactly true), with a total score ranging from 10 to 40.27 Higher scores indicate higher levels of general self-efficacy. The Cronbach’s α coefficient of the scale ranges from 0.75 to 0.94, and the Cronbach’s α coefficient of the Chinese version is 0.87. In addition, the retest reliability is 0.83, and the broken half reliability is 0.90.

Fifth, the level of social support in patients with pre-diabetes will be evaluated using the Social Support Rating Scale (SSRS), which was designed by Xiao Shuiyuan and is the most commonly used scale to evaluate the level of social support.28 The SSRS has three categories and 10 items in total. The scale consists of three factors, namely, objective support, subjective support and support utilisation. The items will be graded on a 4-point Likert scale, with scores ranging from 1 (strongly disagree) to 4 (strongly agree). The total score for the 10 items will be used to calculate the subject’s current overall social support status, and the score for each category will be calculated by using the scores for the relevant items. The Cronbach’s α coefficient of this scale is 0.91.

Finally, the quality of life of pre-diabetic individuals in the past 4 weeks will be evaluated using the 12-item Short-Form health survey (SF-12). This scale consists of 12 items with eight dimensions, namely, physical functioning (PF), role-physical (RP), bodily pain (BP), general health (GH), energy/fatigue (VT), social functioning (SF), role-emotional (RE) and mental health (MH).29 This scale will report two summary scores, namely, the physical component summary (PCS) score and the mental component summary (MCS) score, which will be used to measure the physical health-related quality of life (HRQOL) and mental HRQOL, respectively. Higher scores reflect better mental or physical functioning. For comparison, scores will be calculated using the norm-based scoring algorithm. The crude score will be calculated first, and the standardised score will be calculated subsequently using the following formula: standardised score=(actual score-possible minimum score)/(possible maximum score and possible minimum score)×100. The Cronbach’s α coefficient of the Chinese version of the SF-12 scale ranges from 0.845 to 0.877, showing good reliability and validity of the scale.29