Cultural adaptation of an evidence-based lifestyle intervention for diabetes prevention in Chinese women at risk for diabetes: results of a randomized trial

Abstract

Background

The prevalence of both obesity and type 2 diabetes has been on the rise in China. This randomized controlled trial was conducted to test the feasibility and effectiveness of an evidence-based diabetes prevention program in Yuci, Shanxi Province, China from 2012 to 2014.

Methods

Women with pre-diabetes, ages 25–65 y, were assigned randomly to a comparison (n=75) or 6-mo lifestyle intervention condition (n=109). Weight, fasting glucose, hemoglobin A1c and self-reported diet and physical activity were measured at baseline, 6 mo and 12 mo.

Results

All measures except fasting glucose improved favorably in both comparison and intervention participants at the 6- and 12-mo follow-ups. Participants in the intervention group lost more weight (−0.91 kg, p<0.05) and had a lower body mass index (−0.39 kg/m², p<0.05) than the comparison group at follow-up. A total of 31.6% (31/98) and 16.2% (11/68) of the participants in the intervention and comparison groups, respectively, achieved the weight loss goal of 5% at follow-up. There was no significant group difference in outcome measures at the 12-mo follow-up. Participants in the intervention group also showed favorable changes in self-reported diet and physical activity measures.

Conclusions

A lifestyle intervention to prevent diabetes in at-risk women in community health centers in China is feasible and acceptable but effect sizes were small.

Chinese women, Diabetes prevention, Diabetes risk, Evidence-based intervention, Lifestyle intervention

Introduction

There has been a significant global increase in type 2 diabetes, especially in developing countries. A recent study estimated that the prevalence of diabetes and pre-diabetes were 11.6% (12.1% for men and 11% for women) and 50.1% (52.1% for men and 48.1% for women), respectively, in China in 2010.¹ Health and economic burdens associated with diabetes have posed substantial challenges to the healthcare systems in low- and middle-income countries. There is an urgent need to develop evidence-based prevention programs to address these challenges in China and other developing countries. However, previous efforts in translating evidence-based interventions to racial/ethnic minority and non-Western populations have produced mixed results.^2,3

Randomized controlled trials (RCTs) such as the Chinese Da Qing Impaired Glucose Tolerance and Diabetes Study (CDQDPOS),⁴ the US Diabetes Prevention Program (DPP)⁵ and the Finnish Diabetes Prevention Study (DPS)⁶ have demonstrated the efficacy of intensive lifestyle intervention in the reduction of risks and the incidence of diabetes in pre-diabetic populations. Translation of DPP and DPS as evidence-based interventions has produced promising outcomes in healthcare and community settings in high-income countries.⁷

Recently the Chinese government has restructured its healthcare system to address the increase in lifestyle-related chronic diseases. In 2006 the Ministry of Health officially endorsed the Community Health Service system and began to promote the urban primary healthcare model through Community Health Centers (CHCs) and Community Health Stations (CHSs), using a six-in-one model: clinical care, health education, family planning, rehabilitation, health maintenance and prevention.⁸ The cities are divided into communities of 10 000–25 000 individuals with one CHC or CHS per community, with the health workers being responsible for screening and managing chronic disease for all members of their community. The newly initiated Community Health Service system has opened a new venue for conducting community-based health promotion programs that can aggressively attack diabetes in China using the latest advancements in diabetes prevention and translation research.

We conducted a translation study to evaluate a community-based lifestyle intervention program, Pathway to Health (PATH), to reduce the risk of diabetes in a sample of Chinese women at risk for diabetes. The purposes of this RCT were to test the feasibility and acceptability of adapting the DPP curriculum for implementation in CHCs by trained community health educators and to test the effects of a 6-mo intervention on body weight, hemoglobin A1c (HbA1c) and cardiorespiratory fitness at 6- and 12-mo follow-ups. Clinical indicators such as the percentage of participants achieving a 5% weight loss and a 10% reduction in HbA1c and the number needed to treat were examined. We also examined the effect of program attendance on the outcome measures.

Materials and methods

Study design and sample

This 1-y RCT was conducted in a non-metropolitan community in collaboration with the local health department, municipal hospital and two CHCs in Yuci, Shanxi Province, China from May 2013 to November 2014. Study outcomes were assessed at baseline and the 6- and 12-mo follow-ups. The majority of the study participants were recruited from patients enrolled in two CHCs. We also recruited participants from a hospital-based health screening center that offers annual health screenings to community residents for a fee. The study eligibility criteria were female residents aged 25–65 y, pre-diabetes (based on a previous health exam), overweight or obese (body mass index [BMI]≥24), not physically active (based on a short physical activity [PA] screening)⁹ and expressed interest in lifestyle changes. Individuals who had diagnosed diabetes or reported the use of medication for diabetes were excluded from the study. All participants completed a preliminary eligibility screening and provided informed consent. After completing the baseline assessment, the participants in two cohorts were assigned randomly to either a comparison group (40%) or a 6-mo lifestyle intervention group (60%), in blocks of 10 participants, using a randomization table by the study statistician. Enrollment and randomization were performed by trained research staff. There was no contact with the participants during the last 6 mo of the study. We planned to recruit 280 female participants, which would provide an 80% power (α=0.05 for a two-tailed test) to detect a moderate intervention effect (δ=0.4–0.6) on weight loss, assuming a retention rate of 90% over three measurement points. The study protocol was approved by the Institutional Review Board at the University of Texas at San Antonio and North Dakota State University. Because it was not stipulated by the funding agencies, this study was not registered as a clinical trial.

Study outcome and process measurements

Trained research staff measured the participant’s weight, height and waist circumference with light clothes twice and the average was used. Venous blood plasma was collected for HbA1c and fasting glucose after an 8-h fast. The blood samples were assayed at the city’s municipal hospital. Heart rate (beats per minute [bpm]) at the completion of a submaximal 3-min step test was used to evaluate cardiorespiratory fitness.¹⁰ Information on clinical indicators (percentage of participants losing ≥5% of body weight and lowering HbA1c ≥10%) was also assessed.

Lifestyle factors, including physical activity and diet, were assessed. The participants completed the International Physical Activity Questionnaire – Short Form (IPAQ-SF), which provides estimates of habitual levels (min/d) of vigorous PA (VPA), moderate PA (MPA) and walking for exercise in the last 7 d.¹¹ Because VPA was not reported by the participants, we only analysed the minutes for MPA and walking. The participants also completed a dietary survey to assess the eating habits and quality of diet based on the Chinese Dietary Guidelines.¹² The 40-question survey assessed meal and snack patterns; the kind and amount of staple and beverage consumption; the frequency of dairy, meat, organ meat, fish, legume, fruit and vegetable consumption and cooking methods (e.g., fried, stir fried, steamed, boiled). The scores ranged from 50 to 200, with higher scores indicating a higher-quality diet. Attendance in group health education classes was recorded. Demographics (age, marital status, work status, childcare responsibilities), family history of diabetes, menopause status, medication use and new diagnoses of diabetes were also collected.

Description of the intervention treatments

PATH intervention

The PATH was a group-based lifestyle intervention built on the DPP Group Lifestyle Balance program¹³ and a feasibility study of a community-based lifestyle modification program for diabetes risk reduction in Mexican American women at risk for diabetes in San Antonio, TX, USA.¹⁴ PATH was a goal-based behavioral intervention with three goals: 5% weight loss, ≥30 min of MPA (3000–4000 moderate-intensity steps/d or an equivalent amount of other forms of PA) on most days of the week and a reduction in weekly caloric intake of 1000–1400 calories (200 calories/d) by adopting healthy eating practices (limiting foods high in fat and starch and increasing vegetables and fruits). PATH participants received training to individualize the three goals into short- and long-term goals and re-evaluate the goals with the support of Community Health Educators (CHEs). For example, each participant identified the 5% weight goal based on her weight at baseline and developed short-term weight loss goals. Intervention activities were adapted to address differences in lifestyle and culture, traditional values associated with foods (e.g., preference for high starch foods), cooking practices, resources available and healthcare practices in China. We also modified the nutritional messages and recommendations based on the Chinese Food Guide Pagoda.¹⁵ PATH program materials are available for public use at http://www.pathdpp.com.

The 6-mo intervention included 12 large-group health education sessions on nutrition, PA, behavioral monitoring/goal setting; 10 small-group sessions for goal setting and evaluation, counselling and social support; and a final celebration party. The intervention was delivered by CHEs who were recruited from the chronic disease management team in the two CHCs. The CHEs received 50 h of training in lifestyle interventions, PA, nutrition and community-based health promotion. At the beginning of the intervention the participants established their personalized behavioral goals and identified strategies to achieve the goals under the guidance of the CHEs in a PATH Participant’s Handbook, which included a PATH description, individual goal worksheets, weekly logs and class handouts. The participants received training on walking with a pedometer and strengthening exercises with a resistance band, an exercise DVD for exercise at home, an oil and salt measuring cup to monitor the amount of oil and salt used in cooking and information on healthy cooking with recipes for reducing starch and fat intake. The class topics and schedules are shown in Table 1. Participant’s weight was recorded at each meeting. CHEs also made weekly phone calls to the participants for problem solving and support if they missed a session. The participants used the PATH Handbook to record their goal progress and monitor their weekly dietary and physical activities.

Table 1.

PATH health education class and small-group meeting schedule

Week no.	Meeting type and topic
Week 1	Class 1: Starting on the PATH to health—are you ready? (Importance of physical activity and diet)
Week 2	Class 2: Starting on the PATH to health: changing your lifestyle
Week 3	Class 3: Knowing how much you move
Week 4	Small group: Activity form, goals, food diary
Week 5	Class 4: Nutrition (Healthy eating—the Chinese food pyramid)
Week 6	Small group: Review game, food diary, self-evaluation and goals
Week 7	Class 5: Being active: a way of life
Week 8	Small group: Review, activity form, resistance band (DVD)
Week 9	Class 6: Nutrition: carbohydrates
Week 10	Small group: Review game (Jeopardy), food diary, self-evaluation, resistance band
Week 11	Class 7: Nutrition: protein
Week 12	Small group: Basic cooking skills, food diary
Week 13	Class 8: Nutrition: tip the calorie balance
Week 14	Small group: Review game (Jeopardy), self-evaluation
Week 15	Class 9: Problem solving
Week 16	Small group: Self-evaluation, problem solving
Week 17	Class 10: Managing your environment
Week 18	Small group: Food labels
Week 19	Class 11: The slippery slope of lifestyle change
Week 20	Small group: Menopause discussion
Week 21	Class 12: Ways to stay motivated
Week 22	Small group: Wheel of fortune game
Week 23	Celebration

All intervention sessions lasted 60 min and were conducted at two CHCs and a health screening center. The health education session began with a weigh-in and review of the step count for the previous week, followed by a review of the previous class content, a 30-min lecture (using participatory teaching methods) and finally a discussion. Small-group meetings were attended by three to five participants and included a review of previous material, discussion of issues related to topics of previous lectures, sharing of problem-solving strategies and evaluation of individual goals. Voluntary resistance band exercise sessions were offered 30 min before class time from week 8. Participants in the intervention group were allowed to make up missed health classes at a later time.

Comparison group: general healthy lifestyle education

Participants in the comparison group received a counselling session with a nurse to review the results of the baseline testing, a package with information on nutrition and PA, a pedometer with instructions, an oil and salt measuring cup and a notebook for taking notes at health classes and for tracking weight and PA, but they were not given specific counselling or reinforcement to do so. The participants were invited to attend six general health education classes on PA, nutrition, chronic diseases (obesity, diabetes, heart diseases) and menopause at the same venue as the intervention group. The sessions were 3–4 wk apart and lasted about 1 h. The participants did not have small-group meetings or group exercise sessions.

Data analysis

Demographic, baseline outcome and process measures were tested with a one-way F test for continuous variables and χ² test for categorical variables to examine the difference between the comparison and intervention groups. Generalized estimation equations (GEEs) for repeated measures were used to test the time effect as well as the time×treatment interaction on differences in the outcome measures, with planned contrasts between the comparison and intervention groups at the 6- and 12-mo follow-ups. One random effect was included in the model to account for the autocorrelations from repeated measurements. Cohort, demographic (age, education, marital status, and childcare) and health history variables (menopause status and medication use), as fixed effects, were included as covariates in the model. Similarly, we tested the difference in self-reported PA and dietary quality between the comparison and intervention groups. We also explored the effect of program attendance (<70% vs ≥70% session attendance) on the outcome measures and self-reported behavioral measures. The significance level was set at p<0.05 (two-sided test). SPSS Statistics 22 (IBM, Armonk, NY, USA) was used for all data analysis.

Results

Baseline

The flow of study participants is displayed in Figure 1. After completing baseline assessment, 184 participants were randomized to an intervention or comparison group. Participant retention rates were 95% (n=174) at the 6-mo follow-up and 86% (n=159) at the 12-mo follow-up.

Figure 1.

PATH participants flow diagram.

Participant characteristics are shown in Table 2. The average age of the participants was 51.96 y (SD 7.22), with two-thirds being postmenopausal. Half of the participants had less than a high school education (≤9 y of school). Almost all participants were married, and nearly half had the responsibility of taking care of a grandchild. The health history of the participants demonstrated a high-risk profile for diabetes. Close to one-third had a family history of diabetes (n=57), 26% (n=48) were taking medication for hypertension and 69.9% (n=124) had metabolic syndrome.¹⁶ There was no difference in the outcome measures, self-reported MPA and walking between the comparison and intervention groups at baseline, except heart rate response to the step test, with the participants in the comparison group being less fit.

Table 2.

Comparisons of study participant characteristics and outcome measures at baseline^a

Study variable name		Comparison group (n=75)	Intervention group (n=109)	All (n=184)
Age (y), mean (SD)**		53.27 (7.17)	51.06 (7.15)	51.96 (7.22)
Study cohort, n (%)	1	33 (44.0)	51 (46.8)	84 (45.7)
	2	42 (56.0)	58 (53.2)	100 (54.3)
Education, n (%)**	1–6 y	10 (13.3)	13 (11.9)	23 (12.5)
	7–9 y	36 (48.0)	33 (30.3)	69 (37.5)
	10–12 y	23 (30.7)	33 (30.3)	56 (30.4)
	≥13 y	6 (8.0)	30 (27.5)	36 (19.6)
Marriage status, n (%)	Married	72 (96.0)	104 (95.4)	176 (95.7)
	Not married	3 (4.0)	5 (4.6)	8 (4.3)
Childcare responsibility, n (%)	No	44 (58.7)	58 (53.2)	102 (55.4)
	Yes	31 (41.3)	51 (46.8)	82 (44.6)
Family diabetes history, n (%)	No	57 (76.0)	70 (64.2)	127 (69.0)
	Yes	18 (24.0)	39 (35.8)	57 (31.0)
Menopause status, n (%)	No	22 (29.3)	38 (34.9)	60 (32.6)
	Yes	53 (70.7)	71 (65.1)	124 (67.4)
Taking hypertension medication, n (%)	No	57 (76.0)	79 (72.5)	136 (73.9)
	Yes	18 (24.0)	30 (27.5)	48 (26.1)
Taking lipids medication, n (%)	No	71 (94.7)	105 (96.3)	176 (95.7)
	Yes	4 (5.3)	4 (3.7)	8 (4.3)
Metabolic syndrome, n (%)*	No	17 (22.7)	39 (35.8)	56 (30.4)
	Yes	58 (77.3)	70 (64.2)	128 (69.6)
Weight (kg), mean (SD)		68.59 (8.42)	67.82 (9.10)	68.13 (8.81)
Waist circumference (cm), mean (SD)		96.95 (7.27)	95.41 (7.41)	96.04 (7.37)
BMI (kg/m2), mean (SD)		27.43 (2.75)	27.42 (2.91)	27.42 (2.84)
HbA1c (%), mean (SD)		6.04 (0.45)	6.04 (0.51)	6.04 (0.49)
Fasting glucose, mean (SD)		5.38 (0.53)	5.38 (0.59)	5.38 (0.57)
Heart rate in the step test (bpm), mean (SD)**		139.32 (13.61)	134.66 (15.06)	136.52 (14.64)
Diet quality score, mean (SD)*		105.43(11.37)	102.16 (12.25)	103.50 (11.97)
Daily self-reported moderate physical activity, n (%)*	<30 min/d	67 (89.3)	99 (90.8)	166 (90.2)
	30–59 min/d	2 (2.7)	8 (7.3)	10 (5.4)
	≥60 min/d	6 (8.0)	2 (1.8)	8 (4.3)
Daily self-reported walking, n (%)	<30 min/d	51 (68.0)	76 (72.4)	127 (70.6)
	30–59 min/d	13 (17.3)	12 (11.4)	25 (13.9)
	≥60 min/d	11 (14.7)	17 (16.2)	28 (15.6)

*p<0.10, **p<0.05.

^aGroup differences were tested with one-way F test for continuous variables and χ² test for categorical variables.

Primary study outcomes: weight and cardiometabolic indicators

All outcome measures except fasting glucose improved from baseline to the 6-mo follow-up and from baseline to the 12-mo follow-up in participants in both groups. Fasting glucose increased across time in both the comparison and intervention groups. The results of GEE analysis (see Table 3) revealed a significant time effect on the outcome measures with adjustment for multiple comparisons, suggesting participants in both groups had favorable changes over the study period. There was a significant time×treatment interaction on the changes in body weight and BMI at the 6-mo follow-up that favored the intervention participants, after adjustment for covariates. No other group difference was found at the 6- and 12-mo follow-ups.

Table 3.

Estimated marginal means (M) and standard errors (SE) for study outcome measures at baseline and 6- and 12-mo follow-ups

Outcome measures		All participants†			Comparison group			Intervention group			Group difference at 6 mo‡	Group difference 12 mo‡
		Baseline	6 mo	12 mo	Baseline	6 mo	12 mo	Baseline	6 mo	12 mo
Weight (kg)	M	67.32a,b	66.08a	66.45b	67.51b	66.72b	66.82b	67.13b	65.44b	66.07b	0.91*	0.38
	SE	0.74	0.77	0.77	1.13	1.18	1.17	0.88	0.92	0.92	0.45	0.45
BMI (kg/m2)	M	27.20a,b	26.71a	26.87b	27.19	26.90	26.94	27.20	26.52	26.80	0.39*	0.16
	SE	0.22	0.24	0.24	0.34	0.36	0.37	0.28	0.30	0.31	0.18	0.18
Waist circumference (cm)	M	94.86a,b	93.39a	93.96b	95.58	94.45	95.08	94.14	92.33	92.84	0.68	0.80
	SE	0.60	0.63	0.60	0.87	0.94	0.89	0.74	0.79	0.75	0.69	0.60
HbA1c (%)	M	5.99a,b	5.78a	5.66b	5.96	5.76	5.6	6.02	5.8	5.71	0.01	–0.05
	SE	0.04	0.04	0.04	0.06	0.06	0.06	0.05	0.05	0.05	0.08	0.09
Fasting glucose (mmol/l)	M	5.3a,b	5.49a	5.63b	5.26	5.44	5.62	5.34	5.55	5.63	–0.04	0.07
	SE	0.05	0.04	0.05	0.07	0.07	0.07	0.06	0.05	0.05	0.08	0.08
Heart rate (bpm)	M	138.05a,b	131.54a	133.89b	140.33	134.60	137.56	135.77	128.49	130.22	1.54	2.77
	SE	1.26	1.29	1.22	1.88	1.93	1.81	1.54	1.57	1.48	1.94	1.97

Same superscripted letters denote group difference (all p<0.05 with Bonferroni adjustment for multiple comparisons).

*p<0.05.

†One-way post hoc test for time effect between baseline and 6-mo follow-up and baseline and 12-mo follow-up.

‡Planned contrast for group difference at 6- and 12-mo follow-ups.

Goal achievement

Table 4 shows comparisons of the percentage of the participants achieving 5% weight loss goal between the comparison and intervention groups. More participants in the intervention group (31.6% [n=31]) than in the comparison group (16.2% [n=11]) achieved the weight loss goal at the 6-mo follow-up. There was no significant difference at the 12-mo follow-up. Even though there was no difference in percentage of participants with a 10% reduction in HbA1c between the comparison and intervention groups, 18.9% (n=31) and 26.2% (n=43) of all participants lowered their HbA1c more than 10% at the 6- and 12-mo follow-ups.

Table 4.

Comparisons of 5% weight loss and 10% reduction of HbA1c at 6- and 12-mo follow-ups

	≥5% weight loss		≥10% HbA1c reduction
	6 mo	12 mo		6 mo	12 mo
Comparison (n=68), n (%)	11 (16.2)a	15 (22.1)	Comparison (n=68), n (%)	17 (25.0)	21 (30.9)
Intervention (n=98) , n (%)	31 (31.6)a	28 (28.6)	Intervention (n=96), n (%)	14 (14.6)	22 (22.9)
All (n=166), n (%)	42 (25.3)	43 (25.9)	All (n=164), n (%)	31 (18.9)	43 (26.2)

Same superscripted letters denote group difference (χ² test, p<0.05).

Diet and PA changes

There was a significant difference in dietary quality scores between the intervention and comparison groups at the 6- and 12-mo follow-ups, suggesting that the intervention participants had more improvement in diet quality than the comparison participants (see Table 5). Table 6 compares the differences in the levels of self-reported MPA and walking for exercise between the treatment groups. While all participants in both treatment conditions had similar levels of participation in MPA and walking at baseline, the levels of MPA and walking increased at the 6- and 12-mo follow-ups in both groups (p<0.05). The participants in the intervention group also engaged in higher levels of MPA at the 12-mo follow-up (31.3% [n=30] vs 14.9% [n=10]) and walking at the 6-mo follow-up (47.3% [n=26] vs 27.5% [n=11]) compared with the comparison group (p<0.05).

Table 5.

Changes in dietary quality scores at 6- and 12-mo follow-ups†

	Change at 6 mo*	Change at 12 mo*
Comparison	6.31 (9.63)a	8.90 (10.77)b
Intervention	12.61 (12.18)a	13.89 (13.36)b
All	10.04 (11.60)	11.83 (12.57)

Same superscripted letters denote group difference (all p<0.05 with Bonferroni adjustment for multiple comparisons).

*p<0.05 with Bonferroni adjustment for multiple comparisons.

†One-way post hoc test for comparison of difference between baseline and month 6 and baseline and month 12.

Table 6.

Self-reported PA at baseline and 6- and 12-mo follow-ups

		Self-reported MPAa			Self-reported walkingb
		<30 min/d	30–59 min/d	≥60 min/d	<30 min/d	30–59 min/d	≥60 min/d
Baseline (n=184)	Control (n=79)	67 (89.3)	2 (2.7)	6 (8.0)	51 (68.0)	13 (17.3)	11 (14.7)
	Intervention (n=105)	99 (90.8)	8 (7.3)	2 (1.8)	76 (72.4)	12 (11.4)	17 (16.2)
6-mo follow-up (n=95)	Control (n=40)	23 (56.1)	8 (19.5)	10 (24.4)	11 (27.5)	18 (45.0)	11 (27.5)
	Intervention (n=55)	37 (67.3)	7 (12.7)	11 (20.0)	11 (20.0)	18 (32.7)	26 (47.3)
12-mo follow-up (n=163)	Control (n=67)	49 (73.1)	8 (11.9)	10 (14.9)	18 (26.5)	24 (35.3)	26 (38.2)
	Intervention (n=96)	50 (52.1)	16 (16.7)	30 (31.3)	21 (22.6)	29 (31.2)	43 (46.2)

Data presented as n (%).

^aLinear trend for self-reported moderate physical activity at the 12-mo follow-up (Somer’s d=0.21, p<0.003).

^bLinear trend for self-reported walking at the 6-mo follow-up (Somer’s d=0.17, p<0.07).

Trial participation

The average attendance of the 23 sessions was 67.5%, with 66 intervention participants attending 70% or more of the sessions. However, attending 70% of the sessions was not associated with improvement in the outcome measures at the 6- and 12-mo follow-ups (data not shown). There was no unintended effect reported by the study participants (Table 7).

Table 7.

Health education class attendance*

Treatment group	Sessions attended (n), mean (SD)†	Sessions attended (%), mean (SD)†	≥70% of sessions attended, %‡
Comparison	3.46 (1.64)a	57.7 (27.4)a	31.0a
Intervention	15.53 (7.60)a	67.5 (33.0)a	66.0a

Same superscripted letters denote group difference (p<0.05).

*Intervention group had 23 health education classes and small-group meetings; comparison group had 6 health education sessions.

†One-way F test.

‡χ² test.

Discussion

The findings of this RCT demonstrate that Chinese women at risk for diabetes significantly lowered their body weight and BMI after participation in a 6-mo lifestyle intervention program at the 6-mo follow-up compared with participants who received general healthy lifestyle education. However, the group difference had disappeared at the 12-mo follow-up. There was no other significant difference in the outcome measures. However, all of the outcome measures except fasting glucose improved from baseline to the 6- and 12-mo follow-ups in the intervention as well as comparison participants.

Translation studies of DPP with an RCT design are rare in China. Although the CDQDPOS was among the first studies that demonstrated the efficacy of lifestyle intervention to prevent diabetes in adults with pre-diabetes,⁴ the finding has not been tested using a translation research paradigm to examine its effectiveness and scalability in community settings.¹⁷ The PATH intervention closely mimicked the key concepts and components of the DPP lifestyle intervention⁵ that has been successfully adapted for different population groups and settings in the USA and other countries.^7,18 In addition, we adapted the DPP’s intervention strategies and activities to accommodate both cultural and lifestyle preferences of Chinese women. Intervention impacts on body weight, dietary quality and PA confirmed the PATH intervention has produced desirable behavioral changes, as anticipated.² High levels of program compliance measured by attendance suggest that the program was feasible and acceptable regarding delivery format, behavioral intervention strategies and program activities. The dropout rate in the study was consistent with those reported in other translation studies.¹⁹ Together, the evidence suggests that we have successfully adapted an evidence-based intervention for implementation in Chinese CHCs, even though we were unable to demonstrate its effectiveness in some important outcome measures.²⁰ Future studies should compare the effectiveness of the CDQDPOS and DPP using the translation research paradigm.

Weight loss (5–7%) is the primary indicator of success for risk reduction in diabetes prevention trials as well as translation studies.¹⁹ However, studies in addition to ours have reported weight loss of less than 5% in ‘real-world’ settings in non-Western countries.^18,21 The PATH participants showed a consistent trend of reduced weight, BMI and waist circumference over time as well. The magnitude of reduction, however, was considerably smaller compared with Western studies. This is consistent with diabetes prevention studies conducted in Asian populations. No significant weight loss was reported in the study participants at the 6-y evaluation in the CDQDPOS,⁴ and weight loss was also small in Japanese adults with pre-diabetes in two Japanese DPP translation trials.^22,23 While the average BMI was ≥35 in Western studies, the average BMI in our study and the Japanese studies was less than 28. Therefore, as important as it is to lose weight, the expected amount would be less than the 5% in populations with a lower overall body mass compared with DPP translation studies in Western populations. It is also interesting to note that PATH program attendance (attending ≥70% of sessions) did not affect the outcomes, contrary to findings in other studies.^7,18 The Chinese women in the study tended to be older, have a low level of education, to not work outside the home and to have childcare responsibilities. It is unclear how these participant characteristics might have influenced their responsiveness to the intervention.²⁴ We also want to note that the PATH curriculum started with PA followed by nutrition education, whereas studies that started with nutrition education reported improved long-term weight loss in Western populations. Finally, lack of a maintenance program may contribute to the diminished program effects at the 12-mo follow-up. Continued participant support was shown to be critical in achieving long-term effects among community-based lifestyle intervention participants, whereas in our study we offered no maintenance program from month 6 to month 12.²⁵

In general, changes in clinical measures such as HbA1c and fasting glucose were minimal in DPP translation studies.²⁶ A recent meta-analysis found that the use of lifestyle interventions in routine clinical practice positively influenced body weight–related measures but not clinical measures in patients at high risk for diabetes.²⁷ In our study there is a non-significant favorable trend of a reduction in HbA1c in the intervention participants, while fasting glucose increased over the study period. Others also reported an increase in fasting glucose in intervention participants that can be attributed to regression to the mean over time.²⁸ Among the two clinical indicators, more intervention participants reached the 5% weight loss goal at the 6-mo follow-up, which is consistent with previous studies.¹⁹ No other significant difference was found, although both groups showed improvement in the clinical indicators.

We offered six general healthy lifestyle education classes to the comparison group, which were well attended. This enhanced treatment over the traditional control treatment led the comparison participants to engage in behavioral changes. Anecdotal information revealed that the provision of health screenings and providing a free pedometer motivated the comparison participants to change their lifestyle, resulting in an improvement in the outcome measures and process measures in the comparison participants. Future research should explore this minimalistic approach in resource-poor communities in developing countries.²⁴

Improvement in diet (reduced calorie and fat intake) and PA (150 min of MVPA/wk) and frequent self-monitoring were closely associated with weight loss and a reduced incidence of diabetes in DPP²⁹ and DPP translation studies.^30,31 Less is known about the effectiveness of lifestyle interventions that are based on Western lifestyles and healthcare systems in non-Western low- and middle-income countries.³² We were unable to quantify the changes and achievement of goals in diet and PA accurately with the use of self-reported measures in our study. Nevertheless, we found consistent improvement in dietary quality and PA in the intervention participants, suggesting that lifestyle interventions based on behavior change theories can be applied to the Chinese population.

There were several limitations in our study that might have affected the generalizability and validity of the study. First, the study was underpowered due to not recruiting the number of participants (57% of targeted sample) based on the power estimation. We encountered many challenges in recruiting study participants, which could be improved by better recruitment outreach and participant education and using multiple recruitment strategies. Some eligible patients refused to participate because of concern about the legitimacy of the study or time conflicts with their family or work obligations. Another unanticipated problem was that many of the eligible patients could not be reached with the contact information in the database at the CHCs, which was the main source of the participant pool. Although the poor recruitment and participation rate did appear to be systematic, the smaller sample size failed to provide adequate power to test the research questions and therefore limited the generalizability of the study findings. Second, PATH intervention only lasted 6 mo, compared with an intervention duration of 6–12 mo in other translation studies.^18,32 Future studies should test interventions with a longer duration and a maintenance program in Chinese populations. Third, there were concerns about the accuracy of blood assays that were performed in the regional hospital laboratory in three separate batches without a calibration protocol. As a result, there were large variations and standard deviations in the measurements. Finally, there were technical difficulties and inconsistencies in implementing the fitness test protocol that led to underestimation of heart rates in some participants.

Conclusion

With the global increase in diabetes rates, cultural adaptation and translation of evidence-based interventions are urgently needed. Findings from our study demonstrate that the PATH intervention is feasible and acceptable for implementation in Chinese CHCs and holds great promise to prevent diabetes in Chinese residents at risk for diabetes. But it should be noted that the effect sizes are small. Future implementation studies are needed to test PATH effectiveness in a large RCT with refinement of the intervention based on feedback from this study.

Authors’ contributions: ZY and MS (project co-directors) were responsible for conceptualizing the study and securing the funding for the study. ZY, MH, YF and JP were responsible for development of the intervention program and data collection protocol and staff training. JP supervised the study implementation and data collection. ZY, JP and RJ were responsible for database management and data analysis. All authors participated in the interpretation of results and preparation of the manuscript. The study protocol is available upon request to ZY. ZY and MS are guarantors of the paper.

Acknowledgements: The authors acknowledge the support of Shanxi Yuci Public Health Bureau and the Center for Disease Prevention in the development and implementation of the study. Heartfelt appreciation also goes to the Yuci residents who participated in the study. Finally, it would have been impossible to complete the study without the hard work and dedication of the Community Health Educators and the Evergreen staff at the Yuci office. The PATH intervention program was based on a translation planning study conducted in San Antonio, TX, USA that was supported by the National Institutes of Health (NIH R34DK084203 to ZY).

Funding: This project is supported by a BRIDGES Grant (ST10-019) from the International Diabetes Federation. BRIDGES, an International Diabetes Federation project, is supported by an educational grant from Lilly Diabetes. The authors also received a gift from the Coca Cola Company (Atlanta, GA, USA) to support the project.

Competing interests: None declared.

Ethical approval: The study protocol was approved by the Institutional Review Boards of the University of Texas at San Antonio and North Dakota State University.

Author notes

The authors wish it to be known that in their opinion the first and last authors should be regarded as joint first authors.