Interventions for the prevention of risk factors and incidence of type 2 diabetes in the work environment: a systematic review

ABSTRACT OBJECTIVE To evaluate the effectiveness of interventions aimed at the prevention of risk factors and incidence of type 2 diabetes in the workers population. METHODS Systematic review of interventions aimed at adult workers at risk of type 2 diabetes published in Medline, Embase, Web of Science, Central Cochrane Registry of Controlled Trials, and Lilacs. Randomized trials, quasi-experimental research and cohort studies were selected; in English, Spanish and Portuguese; published from 2000 to 2017. Intervention effectiveness was evaluated concerning the incidence of type 2 diabetes and a significant reduction in body weight, or another anthropometric or metabolic parameter. RESULTS 3,024 articles were generated, of which 2,825 that did not answer the research question were eliminated, as well as 130 that did not evaluate original interventions, 57 carried out outside the workplace and two reviews; so that 10 selected items remained. Interventions based on structured programs previously evaluated and integrated into the workplace had a favorable impact on the reduction of body weight and other risk factors. CONCLUSIONS The effectiveness of lifestyle interventions for the prevention of type 2 diabetes should be based on structured programs with proven effectiveness and adapted to the workplace, with employer participation in the provision of schedules and work environments.


INTRODUCTION
Type 2 diabetes (DT2) is estimated to affect 425 million people, which represent 8.8% of the world's adult population 1 . The numbers are expected to increase to 693 million by 2045 2 . The prevention of DM2 through structured programs to promote physical activity and healthy diet, also known as lifestyle interventions (LSI), have proved to be effective in reducing the risk of DM2 3,4 . The Finnish Diabetes Prevention Study (FDPS) 5 and the Diabetes Prevention Program (DPP) 6 propose as a success indicator the reduction of 5.0 and 7.0% of the initial body weight, respectively. However, the success and sustainability of these interventions also depends on the context in which they are developed 7 . Work environment represents a challenge for LSIs, considering its known risk factors, such as inadequate diet 8,9 and sedentary lifestyle 10,11 , added to other yet uunevaluated factors, such as work stress 12 . The prevention of DM2 and its complications in labor contexts aims to maintain workers' health and well-being, in addition to avoiding absenteeism 13 , low productivity 14 and significant expenses that affect the health system 15,16 . This study aimed to evaluate the effectiveness of interventions for the prevention of DM2 risk factors in the worker population.

METHODS
A systematic review of interventions aimed at an adult worker population at risk of DM2 was performed. Original articles in English, Spanish and Portuguese, designed as randomized trial, quasi-experimental or cohort studies, and published between 2000 and 2017 were considered eligible. Such studies included interventions for the prevention of DM2 risk factors, based or not on a previously evaluated program. The study population included adult workers with risk factors for DM2, such as overweight, obesity, pre-diabetes and/or metabolic syndrome. Those with some type of diabetes at the beginning of intervention were excluded.
A highly sensitive search algorithm was used for the PubMed, Embase, Web of Science, and Lilacs databases and in the Central Cochrane Register of Controlled Trials (CENTRAL) (Chart).
A list of titles and abstracts was created, based on results of the database search with a bibliographic reference manager (EndNote TM x8, USA). The selection strategy included a first screening of titles and/or abstracts and a second screening based on full-text reading by two of the authors (JI and NH), resolving the discrepancies of both steps with a third reviewer (CU-G).
Intervention effectiveness considered the incidence of DM2 or a significant reduction of initial body weight, as this parameter is recognized as the main DM2 risk predictor 17,18 , as well as of any other anthropometric or metabolic parameter recognized as a DM2 risk predictor (Figure 1).

RESULTS
The search generated 3,083 articles, of which 59 duplicate articles were eliminated, leaving 3,024. In the first screening, 2,825 articles that did not answer the research question were excluded. In the second screening, 189 articles were excluded because they had another study design (n = 67), did not consider the work environment (n = 57) and did not exclude participants with DM2 (n = 39). Therefore, 10 articles were selected, of which six were randomized trials and four were quasi-experimental studies testing the before and after of a single group (Figure 2). Results show heterogeneity in the number, duration, and content of intervention sessions evaluated, as well as diverse age groups, educational levels and work environments of the participants (Table 1). • Web of Science: #1: (TS=("Occupational Health" OR "Primary Prevention" OR "Occupational Health Services" OR "Preventive Health Services" OR "Complementary Therapies" OR "Integrative Medicine" OR "Models, Organizational" OR "Behavior Therapy" OR "Tai Ji" OR "Yoga" OR "Qigong")); #2: (TS=("diabetes" OR "Metabolic Syndrome" OR "Prediabetic State")); #3: (TS=(("adult population" OR "Adult" OR "Occupational Groups" OR "Occupational Health Nursing" OR "Workplace" OR ("Occupational Groups" AND "Population") OR "Work-related" OR "Worksite" OR "Working population" OR "Economically active population"))) AND LANGUAGE: (  Interventions were classified as: conventional LSI (seven), virtual counseling LSI (two), and exclusively nutritional (one). Of these, six were based on structured programs for the prevention of DM2, such as the DPP 19-21 , FDPS 22,23 and the Life Style Modification Program for Physical Activity and Nutrition program (LiSM10!) 24 . Two interventions were ground on DM2 prevention recommendations based on guidelines from The National Institute for Health and Care Excellence (NICE) 25 and the Japan Diabetes Society and the American Diabetes Association (JDS/ADA) 26 . In addition, one intervention was based on a DM2 management program called Healthy Living with Diabetes Program 27 and one did not report any background 28 (Table 2). Bias risk assessment was described and assessed as recommended by the Cochrane Collaboration for randomized studies 29 ( Figure 3).
According to the effectiveness assessment, six studies showed statistically significant reduction in bodyweight [20][21][22]24,25,30 , all of which were based on structured programs for the prevention of DM2, such as DPP [19][20][21] , FDPS 22 and LiSM10! 24 , as well as on the recommendations given by NICE 25 . On the other hand, a nutritional intervention showed a reduction in caloric intake with subsequent reduction of two-hour postprandial glucose levels, based on the JDS/ADA recommendations 26 . In contrast, two non-effective interventions focused on the management of DM2 27,28 , and one had limited sessions and prolonged follow-up 23 . No studies showed adverse effects in the study population.
None of the seven effective studies showed results in terms of reducing the risk of DM2 incidence, since their design did not allow it. However, two of them showed results  regarding the percentage of initial body weight reduction 19,20 . Kramer et al. 19 determined that 5.0% reduction of average body weight was achieved by 45.0% and 7.0% of the participants involved in the intervention and the control group, respectively 19 . In contrast, Weinhold et al. 20 determined that 32.4% of the intervention group managed to lose at least 7.0% of the initial body weight, a percentage which was significantly greater than for the control group (2.9%; p <0.01).  On the other hand, they also evaluated economic and non-economic incentives, as well as the role played by the employer's participation in the intervention results. Faghri et al. 21 was the only study to offer a cumulative economic incentive for lost weight. Despite being a low intensity intervention, it was successful in reducing weight, decreasing the risk of DM2 and leading to a healthy diet. Employer participation generally did not go further than supporting the enrollment of participants. The facilities in enrollment, arrangement of environments, and work schedules in the studies of Kramer et al. 19 and Weinhold et al. 20 were linked to the success of the intervention.

DISCUSSION
Interventions based on structured programs for the prevention of DM2 showed wide effectiveness, which was linked to the degree of similarity with the base program. On the other hand, interventions based on recommendations from the NICE guidelines were statistically effective, but present aspects to be considered. The study by Limaye et al. 25 was based on the NICE guideline and shows significant weight reduction, but this is achieved due to an increase in weight in the control group. In this study, the average bodyweight reduction in the intervention group does not reach less than 5.0% post-intervention.
Uneffective interventions 23,27,28 did not have a structured session plan 23 , and two of them were focused on the management of DM2 27,28 . Viitasalo et al. 23 , in a pre-and post-test study carried out on employees of an airline company, showed a slight increase in anthropometric parameters after the intervention. Such intervention had an average of 1.6 sessions, which is less than in any other existing program; although based on the FDPS, the program differed from the original. Regarding the two interventions aimed at DM2 management, they included a pre-diabetes population in a context of workers with DM2. These types of interventions were based on work welfare programs and, despite having resources, the intervention approach is important.  Overall, the selected articles make a brief description of the labor factors that would favor the development of DM2. Limaye et al. 25 evaluated an intervention in IT employees in India who reported a sedentary lifestyle, inadequate eating habits, and work stress. Maruyama et al. 24 evaluated Japanese office employees with prolonged periods of work, which would favor a high caloric intake. In contrast, Watanabe et al. 26 , evaluated employees with working conditions related to high caloric intake at night and short periods of sleep.
Employees who carried out these activities are within the framework of "white-collar workers". This population has a higher risk of being overweight and obese due to the sedentary lifestyle they experience at work, favoring the increase in body mass index 10 .
On the other hand, work with night, rotating shifts, or non-regular, called "shift work," are associated with the development of DM2 due to insufficient hours of sleep and poor eating habits 31 . Nonetheless, these factors were not evaluated in the intervention, but they had to be considered in the evaluation of its impact.
Other studies also developed communication plans to encourage changes in lifestyle 19,20,22,27,28 and even provided environments and schedules for the development of work sessions 19,20,27 . Some interventions provide basic incentives such as glucometers 27,28 , discount cards when initiating and completing the intervention 19,27 , and economic incentives for lost weight 21 . However, the long-term results regarding the maintenance of the weight lost by this type of interventions seem to be inconsistent 32 .
Those employers who offered greater support in the intervention, such as facilities in enrollment, provision of environments (topical, rest environments, coffee shops, etc.) and schedules at work favored the implementation and success of the intervention 19,20 . Such benefits can be integrated to the intervention, to achieve and maintain the proposed objectives. Companies with food service can implement healthy menu options for people at risk of DM2, and even for diabetics 33 . In our review, interventions that do not report the presence of a program with such benefits are those aimed primarily at "white-collar workers" [24][25][26] .
The limitations of our review are linked to the heterogeneity of the interventions, which makes meta-analysis impossible. On the other hand, there are no results in terms of DM2 incidence risk due to the short intervention periods of most studies evaluated. Thus, the evaluation of intervention effectiveness is flexible, as they are based on the statistically significant reduction of body weight (p <0.05), but not on the reduction of body weight percentage or on clinically-significant weight reduction (reduction of at least 5.0% post-intervention). In addition, the risk of bias in randomized trials is high. The studies of Weinhold et al. 20 and Watanabe et al. 26 show a higher proportion of unclear risk, but with a better rating, while the article by Kramer et al. 19 has the lowest rating.
Interventions with a favorable impact on the reduction of body weight and other risk factors of DM2 in the workplace are mainly those based on a program previously evaluated and carried out in the workplace with employer participation. Longitudinal studies are required to evaluate these interventions as potential programs to reduce the incidence of DM2 in the workplace.