Effectiveness and implementation of a lifestyle modification intervention for women with isolated impaired fasting glucose: Study protocol for a hybrid type 2 study in Kerala, India

Background: Isolated impaired fasting glucose (i-IFG) constitutes a major group in the prediabetic spectrum among Indians, and thus it is imperative to identify effective diabetes prevention strategies. This study aims to evaluate the effects of an intensive community-based lifestyle modification program on regression to normoglycemia among women with i-IFG, compared to a control group at 24 months. The study also aims to evaluate the implementation of the intervention, via both process and implementation outcomes. Methods: We will use a hybrid design (Effectiveness-implementation hybrid type 2 trial) to test the effectiveness and implementation of the lifestyle modification intervention. Effectiveness is evaluated using a randomized controlled trial among 950 overweight or obese women, aged 30 to 60 years, with i-IFG on an oral glucose tolerance test in the Indian state of Kerala. The intervention involves an intensive lifestyle modification program through group and individually mentored sessions using behavioural determinants and behavioural change techniques. The intervention group will receive the intervention for a period of 12 months and the control group will receive general health advice through a health education booklet. Data on behavioural, clinical, and biochemical measures will be collected using standard methods at 12 and 24 months. The primary outcome will be regression to normoglycemia at 24 months, as defined by the American Diabetes Association criteria. Discussion: This study will provide the first evidence on the effects of lifestyle interventions on regression to normoglycemia in people with i-IFG among Indians. CTRI registration: CTRI/2021/07/035289 (30/07/2021)


Introduction
Type 2 diabetes mellitus (T2DM), currently affects almost 537 million adults worldwide with a projected increase of nearly 50% by 2045 1 . Low-and middle-income countries (LMICs) such as India are disproportionately affected, with the large majority (81%) of people with T2DM living in these countries 1 . India has the second-largest number of people (74 million) living with T2DM and this is projected to increase to 125 million people by 2045 1 . In addition, T2DM poses a significant economic burden in LMICs, affecting not just the health care system but also individuals and families with increased out-of-pocket spending for diabetes care 2 . Therefore, much importance has been given to diabetes prevention, with a greater research focus on individuals who are at high risk for diabetes.
Prediabetes, a high-risk metabolic state for diabetes, comprises impaired fasting glucose (IFG) and impaired glucose tolerance (IGT), wherein the glucose levels are higher than those considered to be normal, but lower than the threshold for T2DM 3 . A recent review of 19 studies globally showed that among those with prediabetes, the average proportions of isolated IFG (i-IFG), isolated IGT (i-IGT), and combined IFG and IGT were 43.9%, 41.0%, and 13.5%, respectively, in Caucasians, and 29.2%, 49.4%, and 18.2%, respectively, in Asians 4 . A large-scale study in 15 states of India showed that i-IFG is the most common form of prediabetes among adults, with a prevalence of 21% 5 . On average, nearly 5-10% of people with prediabetes develop T2DM every year, although the progression rate varies by population characteristics (including ethnicity) and the definition of prediabetes 6 . With no effective intervention, nearly 70% of people with prediabetes would eventually progress to develop T2DM 6 .
Lifestyle modification is recognized as an effective non-pharmacologic intervention to prevent or delay the onset of diabetes among individuals with prediabetes 7 . Most landmark lifestyle modification randomized controlled trials (RCTs) for diabetes prevention have been conducted in individuals with IGT 8 , with very limited evidence in the i-IFG group. Lifestyle modification in the trials for those with IGT have focused mainly on weight loss and promoting physical activity, although the intervention intensity, goals, and delivery methods have varied widely between the trials. These trials have shown that diabetes incidence could be reduced between 28.5% and 58% in the intervention group compared with the control group 8 . More importantly, a clinically meaningful reduction in diabetes incidence with lifestyle modification has been shown to persist in those with IGT even after 30 years of follow-up 9 .
Only a few RCTs have evaluated the effects of lifestyle modification on diabetes prevention in people with i-IFG 10 . RCTs conducted among 641 overweight Japanese 11 and 578 overweight Indian adults 12 showed hazard ratios of 1.17 (95% CI 0.50-2.74, n=379) and 0.88 (95% CI 0.43-1.20, n=166) respectively, in those with i-IFG at 3 years. Similarly, a trial among 880 adults with prediabetes in the UK showed a hazard ratio of 0.52 (95% CI 0.15-1.83, n=108) at 3 years in those with i-IFG 13 . The Kerala Diabetes Prevention Program (K-DPP) from India showed a relative risk of 0.95 (95% CI 0.68, 1.33) in people with i-IFG (n=579) at 2 years among 1007 high-risk individuals 14 . However, the above-mentioned findings were from the sub-group analyses, which are constrained by small sample sizes, a limited number of events, and confounding 15 .
While preventing the progression of prediabetes to diabetes is important, regression to normoglycemia is also essential to achieve, even if transient, as this has been shown to be significantly associated with reduction in the development of diabetes 10 . To our knowledge, there are no data on regression to normoglycemia with lifestyle modification in the i-IFG group among Indians, an ethnic group with i-IFG being the most common prediabetes phenotype 5,14 . In addition to being at high risk for diabetes, individuals with i-IFG are at an increased risk of developing micro-and macro-vascular complications, and of all-cause mortality 6 .
This study aims to evaluate the effects of an intensive community-based lifestyle modification program on regression to normoglycemia among women with i-IFG compared to a control group at 24 months. We will also evaluate the effects of the intervention on improving cardiometabolic risk factors. The study will use the RE-AIM framework to evaluate Reach, Adoption, Implementation and Maintenance 16 and the intervention fidelity using process measures.
This two-arm parallel group randomized controlled trial targets women as the behavioural risk factors of overweight or obesity and physical inactivity are generally higher among women, compared to men, in the Indian context 17 and the intervention outcomes reported so far are poorer among women 12 in India.

Study design and setting
The study uses a hybrid design (Hybrid type 2) 18 that tests the effectiveness and implementation of a lifestyle modification program. A randomized controlled trial will evaluate the effectiveness of the intervention, and the intervention implementation and fidelity will be evaluated using implementation outcomes and process measures, respectively.
The study was registered with Clinical Trials Registry, India (CTRI/2021/07/035289, on 30 th July 2021). The trial will be reported in accordance with the Consolidated Standards of Reporting Trials (CONSORT) guidelines 19 .
The study will be conducted in Kasaragod district of Kerala ( Figure 1). Kerala, the southernmost Indian state, has a higher prevalence of T2DM (nearly 20%) and a greater burden of several cardiometabolic risk factors than most other Indian states 17 . Further, the state is in the most advanced stage of epidemiological transition in the country 20 and said to be a harbinger of the future burden of diabetes and other chronic diseases in India 20 . Thus, Kerala provides an ideal place to implement and evaluate lifestyle modification programs. Figure 2 shows the CONSORT diagram of the trial. Kasaragod is the northernmost district of Kerala, with a population of 1,307,375, a sex ratio of 1079 women for every 1000 men, and a literacy rate of 90% 21 . Kasaragod has four taluks (sub-district) with 777 wards (lowest administrative division with approximately 1300 individuals in each ward) 21 . From these four taluks, two were randomly selected, namely Hosdurg taluk and Kasaragod taluk. There are 465 wards in these two taluks, amongst which those wards within 20km distance from the institute were included (269 wards) considering the logistics and feasibility. Out of 269 wards, 25 wards with at least 1000 individuals in each ward were randomly selected. From these selected wards, individuals with i-IFG will be identified and recruited to the trial with a minimum of 23 participants per ward.
Exclusion criteria: 1) Women with known T2DM; 2) having GDM; 3) prior history of GDM; 4) breastfeeding women; 5) having major chronic illnesses including mental disorders, that are likely to impede consenting and participation in the intervention program; 6) taking medications that could alter glucose metabolism (e.g., corticosteroids); and 7) diagnosed with normoglycemia, IGT, and T2DM, as per the American Diabetes Association criteria 3 on the OGTT.

Sample size and randomization
We assumed that the cumulative incidence of regression to normoglycemia at two years would be 18% in the control group (based on unpublished data from the K-DPP trial) and that there would be a 50% relative risk in conversion to normoglycemia with the intervention. The sample size required in each study group was 475 with a type 1 error of 5%, at least 80% power, a contamination rate of 15%, and a 10% loss to follow-up. The sample size required with same estimates and 90% power in each group is 625, and will be recruited, if logistically feasible.
The estimated sample size of 950 participants will be randomly allocated in a 1:1 ratio to the control group and the intervention group using a computer-generated randomization sequence using Microsoft excel by an independent statistician not involved in the trial. The independent statistician will generate the randomized sequence and allocate the individuals to the randomized groups. The participants and the Principal Investigator (EM) will be blinded to the allocation sequence until informed written consent for study participation is obtained. The outcome assessors, data entry personnel and statisticians who analyse the data will be blinded throughout the study as they are not part of the team that delivers intervention or manage the project.
This study uses individual randomization in order to avoid the methodological challenges of cluster randomization, including the clustering effect on statistical power and selection bias, possibility of imbalanced study groups, and the dilution effect 23 . Trial conduct solutions will be put in place to address potential contamination, if any, between the study groups with individual randomization. This include using different staff for each group, education of the participants against contamination 23 , and getting signed nondisclosure agreement from the intervention participants regarding the type of  intervention being received. Furthermore, analytical methods that adjust for contamination will also be used. We will also quantify the contamination using a treatment fidelity framework based on the criteria advocated by the Behaviour Change Consortium 24 and NICE guidance on behaviour change 25 .

Recruitment of participants and data collection
Screening and recruitment of participants and data collection will be conducted in 2 stages as described below: Stage 1: Home visits. All women in the age group of 30-60 years will be selected from the voters' list of the selected wards. Eligibility for participation in the study will be ascertained during home visits using a screening questionnaire (includes demographics and eligibility criteria) by trained staff. Only one participant per household (randomly selected) will be screened. Those who meet the eligibility criteria and provide consent for study participation will be screened with a point-of-care glucometer (OneTouch Select Plus, LifeScan Inc.) and their waist circumference will be measured using SECA 201 ergonomic retractable tape with using a standard protocol 26 . Those who are overweight or obese (waist circumference ≥80 cm) 22 and with a random capillary blood glucose of >110 mg/dl 27 will be invited to undergo a 2-hr OGTT in clinics organized in their neighbourhoods. These individuals will be considered to be at high risk for having prediabetes or diabetes.
Stage 2: Clinics. The high-risk individuals will attend clinics organized in their neighbourhoods, where they will undergo a 2-hr OGTT, body composition, anthropometric, and blood pressure measurements, and interviews to complete the questionnaires. Those diagnosed with T2DM on the OGTT will be referred to the nearby health facility for treatment and care. Body fat composition will be assessed using TANITA DC 360 (pole). Blood pressure will be measured using an OMRON HEM-7120 blood pressure monitor using the standard protocol 26 .
The weight and height will be measured using SECA 813 flat scale and SECA 213 stadiometer, respectively, with a standard protocol 26 . Blood samples will be collected in a fasting state (8-10 hours of fasting) and will be centrifuged within 30 mins after collection and transported to a lab accredited by the National Accreditation Board for Testing and Calibration Laboratories (NABL) 28 . Questionnaires will be administered by trained staff to collect information on demographics, diet 29 , physical activity 26 , tobacco use 26 , alcohol use 26 , diabetes knowledge 30 , prediabetes knowledge 31 , health-related quality of life (HRQoL) 32 ; self-efficacy for managing chronic diseases 33 , risk perception for diabetes 34 , social support 35 , and sleep hygiene 36 using standardized questionnaires (Table 1). Those who are diagnosed with diabetes at 12 th and 24 th month assessment will be referred to the nearby health facility for treatment and care, and will continue to participate in the trial. All assessments done at the baseline will be repeated at 12 months and 24 months.
Staff conducting physical measurements and administering questionnaires at baseline and follow-up visits will be blinded to participants' group allocation. Intervention staff will not be engaged in the data collection process and the statistician engaged in data analysis will be blinded to the group allocation.
All study materials can be found in the Extended data 37 .

Outcomes
The primary outcome for the effectiveness study will be normoglycemia, defined as FPG<5.6 mmol/l and 2-hr PG<7.6 mmol/l, at 24 months 3 . Secondary outcomes will include incidence of T2DM (fasting plasma glucose ≥ 7.0 mmol/l or 2hr PG ≥ 11.1 mmol/l or HbA1c ≥6.5%), insulin sensitivity, beta cell function, weight, body mass index, waist circumference, FPG, 2-hr PG, blood pressure, body composition measures (fat percent and muscle mass) and psychosocial variables (health-related quality of life, self-efficacy/self-empowerment, risk-perception for diabetes, social support and sleep hygiene). Implementation outcomes include reach, effectiveness, adoption, implementation and maintenance of the intervention 16 .
Study groups Control group. Participants in the control group will receive a health information leaflet at baseline in local language (Malayalam) on strategies for diabetes prevention. No further engagement will be there in the control group apart from follow-up assessments at 12 and 24 months.
Intervention group. The study participants in the intervention group will receive an intensive lifestyle modification program for a period of 12 months through group and individually mentored sessions. The intervention will include behavioural determinants such as self-efficacy, risk perception, social support, and behavioural change techniques, including knowledge enhancement, self-monitoring, goal setting and review, and peer support 38 (Table 2). The intervention will be intense as the engagement is not just limited to group sessions, but also involves individualized support through peer mentors and individualized instructions through a mobile application, specifically made for this trial.

Theory of intervention.
The intervention program objectives, theory-based methods, and strategies for intervention engagement and implementation are given in Table 3. Briefly, the intervention was adapted from a successfully conducted lifestyle-based diabetes prevention trial in Kerala, the K-DPP 38 Program objectives will be achieved through "personal learning" and "environmental change" using evidence-based behaviour change techniques with strategies targeting participants at individual, interpersonal and community levels. Figure 3 shows a thematic representation of the program goals with the targets and the strategies adopted at individual and group level using the behaviour techniques. The intervention will focus primarily on key behavioural risk factors, including unhealthy diet, physical inactivity, tobacco use, alcohol use, and sleep. Based on the dietary recommendations for the prevention of T2DM 39 and pertinent research findings among    individuals with IFG, the dietary intervention will include consumption of a low-calorie diet (~1500 calories per day i.e., 500 calories lower than the daily requirement for women) 39,[42][43][44] and consuming food with low glycaemic index. Other dietary recommendations include changing the quality of dietary fat from using saturated to unsaturated, increasing the intake of whole grains and foods rich in fibre and decreasing the intake of sugar-sweetened beverages, sweets, and highly processed products [45][46][47] . The dietary goals are: <30% of total energy intake from fat, 400-600 grams of fruit and vegetable intake a day, 5 cups (400 gms of cooked rice intake per day), <25 gms of free sugar intake a day, and <5 gms of salt intake per day 45 .

Intervention content
The adherence to diet will be assessed every month using a 24-hour dietary recall 48 . Other measures of intervention goal include increasing physical activity to the recommended levels of at least 150 minutes of moderate-vigorous physical activity per week 40 . 7-9 hours of sleep at night 41,49 , and no use of alcohol and tobacco.

Intervention delivery
The intervention will be delivered through 12 sessions (individual and group based), one session per month, over a period of 1 year (Table 3). Intervention delivery will be supported using pretested and piloted educational materials such as flip charts for group-based sessions and a mobile based application for individual sessions. The mobile application will also serve as an interactive platform for a personalized diet planning and reporting.
The group sessions (45-60 minutes duration) will be organised in the participants' neighbourhood, mostly in homes or local health centres, delivered by the research team (postgraduates in public health/ social work degree) in the initial phase, followed by trained volunteers as "peer mentors". Peer mentor, a group nominated volunteer, will undergo a five-day capacity building training program to guide and assist the participants in making realistic goals for lifestyle modification with the support of the extended community stakeholders.

Evaluation framework
In addition to the baseline, 12th month and 24 th month assessment for the primary and secondary outcomes, process evaluation of core interventions at community, peer mentor and participant levels will be undertaken ( Figure 4). Evaluation process will be facilitated through regular monitoring via participant feedback report, peer mentor feedback report, feedback on quality of training, pre-and post-training knowledge evaluation and other interactions (mobile app/telephonic contact).

Data analysis
Data will be collected and entered using a data entry template in an ODK platform by the data collectors and managed in a cloud-based server which will only be accessed by the Principal Investigator (EM). Subsequent to the data collection, the data will be used only using a participant identity number and all the personal identifiers will be masked. Only deidentified data will be shared with other investigators, if required. Data quality will be ensured during data collection and data analysis. During the data collection, the research team will verify the data for missing values and if present, will be rectified then and there. Furthermore, randomly identified 5% of the participants data will be verified with the participants through telephone for quality check. The data from the ODK platform will be exported to SPSS and data cleaning will be done manually. Any outliers or implausible values will be identified and will be checked with participants over phone, if necessary.
The analysis will follow the intention-to-treat (ITT) principle. Characteristics of the participants at baseline will be compared between study groups using descriptive statistics. The primary outcome will be analysed using the generalized estimating equations (GEE) with an appropriate working correlation structure and a binomial family with 'log' link function to estimate the relative risk (and 95% confidence interval [CI], p value). Standard errors will be based on Huber-White sandwich estimator, which will provide valid CIs even in case of misspecification of the correlation structure 50 . For secondary outcomes, continuous variables will be analysed using mixed-effects linear regression models, which will include all available data at baseline, 12 and 24 months. Study group (intervention vs. control), timepoint (follow-up vs. baseline) and a study group-by-time point interaction will be specified as fixed effects. Random effects will be specified for wards, to account for the clustered study design, and for participants, to account for correlation between the repeated measurements on the same individual. Categorical variables will be analyzed using log-binomial models. All p values reported will be 2 tailed, and a p of 0.05 will be considered statistically significant. Analyses will be performed with Stata version 14.2 (StataCorp LP, College Station, Texas, USA).

Ethics approval
The study was approved by the Institutional Human Ethics Committee of the Central University of Kerala (CUK/ IHEC/2019/034_A, 21 st November 2019). Written informed consent will be obtained from all study participants. The risk from the intervention to the participants is anticipated to be negligible as the intervention involves only lifestyle modification and no pharmacological drugs. Data safety monitoring will be done by the research team.

Discussion
This paper describes the protocol for an intense lifestyle intervention program among women with i-IFG to facilitate regression to normoglycemia at 24 months. This study will provide the first evidence on the effects of lifestyle intervention in regressing i-IFG to normoglycemia among Indians. The findings of the study will be disseminated through public engagement, reports and publications.

Strengths and limitations
This is one of the first studies globally to evaluate the effects of a novel lifestyle intervention among women with i-IFG. Further, the community-based nature of the intervention would facilitate future sustainability and scalability in India and other similar settings. However, since the study population comprises only women, the findings cannot be generalized to men.

Trial status
The trial is currently in the screening phase. have strong face validity and low risk, but there is less evidence for effectiveness in a given context. I may be misreading, but the trial seems to be primarily motivated by the need for evidence on the effectiveness of lifestyle interventions among people with isolated FPG. The protocol has much less detail about the implementation outcomes than the effectiveness outcomes. I did not see a co-primary implementation outcome specified. The individually randomized nature of the trial (rather than cluster RCT) also limits the RE-AIM assessment at the provider/clinic/health system levels. If the authors deem this more in line with a type 2 design, they may wish to add more support for their thinking.

Not applicable
Competing Interests: No competing interests were disclosed.
Reviewer Expertise: Diabetes, global health, implementation research, health systems research I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.