Article Text

Download PDFPDF

Effects of a worksite tobacco control intervention in India: the Mumbai worksite tobacco control study, a cluster-randomised trial
  1. Glorian Sorensen1,2,
  2. Mangesh Pednekar2,3,
  3. Laura Shulman Cordeira1,
  4. Pratibha Pawar3,
  5. Eve M Nagler1,2,
  6. Anne M Stoddard4,
  7. Hae-Young Kim5,
  8. Prakash C Gupta2,3
  1. 1Center for Community-Based Research, Dana-Farber Cancer Institute, Boston, Massachusetts, USA
  2. 2Harvard T.H. Chan School of Public Health, Boston, Massachusetts, USA
  3. 3Healis-Sekhsaria Institute for Public Health, Navi Mumbai, India
  4. 4Statistical Consultant, Pelham, Massachusetts, USA
  5. 5New England Research Institutes, Watertown, Massachusetts, USA
  1. Correspondence to Dr Glorian Sorensen, Center for Community-Based Research, Dana-Farber Cancer Institute, 450 Brookline Avenue, Boston, MA 02215, USA; Glorian_Sorensen{at}


Objectives We assessed a worksite intervention designed to promote tobacco control among workers in the manufacturing sector in Greater Mumbai, India.

Methods We used a cluster-randomised design to test an integrated health promotion/health protection intervention, the Healthy, Safe, and Tobacco-free Worksites programme. Between July 2012 and July 2013, we recruited 20 worksites on a rolling basis and randomly assigned them to intervention or delayed-intervention control conditions. The follow-up survey was conducted between December 2013 and November 2014.

Results The difference in 30-day quit rates between intervention and control conditions was statistically significant for production workers (OR=2.25, p=0.03), although not for the overall sample (OR=1.70; p=0.12). The intervention resulted in a doubling of the 6-month cessation rates among workers in the intervention worksites compared to those in the control, for production workers (OR=2.29; p=0.07) and for the overall sample (OR=1.81; p=0.13), but the difference did not reach statistical significance.

Conclusions These findings demonstrate the potential impact of a tobacco control intervention that combined tobacco control and health protection programming within Indian manufacturing worksites.

Trial registration number NCT01841879.

  • Global health
  • Cessation
  • Disparities
  • Low/Middle income country
  • Socioeconomic status

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


There is an urgent need for effective tobacco control initiatives in response to the rising global burden of tobacco, particularly in the developing world.1 ,2 WHO predicts that of the more than 8 million people globally who are expected to die from tobacco-related causes by 2030, 80% will be from low and middle income countries (LMICs).3 Tobacco use in multiple forms is widespread among India's population of 1.2 billion. Tobacco use prevalence rates are 48% among men and 20% among women.4–6 In 2010, over 1 million Indians were expected to die from tobacco-related causes.7 ,8 Although India was an early signatory to the Framework Convention on Tobacco Control, and smoking is banned in public places such as enclosed worksites,9 enforcement mechanisms are weak, few resources are available to support tobacco use cessation, the prevalence of quitting remains low, and social norms rarely support quitting.4

Growing attention is being paid to these mounting global disparities in tobacco use.10 ,11 The National Cancer Institute recently formed its Center for Global Health aimed at reducing the global burden of cancer.12 In the USA and other high-income countries, employer initiatives have contributed to tobacco control. Such efforts are uncommon within India, although policy leaders and Indian employers alike are increasingly examining the implications of non-communicable diseases (NCD) and risk-related behaviours such as tobacco use for productivity and economic growth.13–17

WHO,18 the US National Institute for Occupational Safety and Health (NIOSH),19 and others20 ,21 have recommended new models for workplace interventions that aim to reduce hazards in the physical and psychosocial work environments, while also providing support for individual behavioural changes such as quitting tobacco use. This approach has been shown to be efficacious in promoting smoking cessation, particularly for blue-collar workers who face dual health risks because of their high rates of risk-related behaviours, such as tobacco use, and exposures to occupational hazards.22–24 For example, in the USA, we found that an intervention that integrated health promotion, including tobacco control, with occupational safety and health, resulted in a doubling of quit rates among production workers, compared to no significant changes among non-production workers.23 These findings suggest that production workers, who are more likely to be exposed to hazards on their jobs, may be especially receptive to these integrated approaches.

We applied this approach in the Mumbai Worksite Tobacco Control Study, which tested the effectiveness of a worksite intervention, the Healthy, Safe, and Tobacco-free Worksites programme, in increasing tobacco use cessation among workers in the manufacturing sector. The purpose of this manuscript is to present findings related to the study's main outcome, tobacco use cessation, focusing on the following hypotheses: (1) tobacco use cessation rates, defined as 6-month continuous abstinence of all tobacco use (primary outcome), and 30-day abstinence (secondary outcome), will be significantly greater in intervention worksites compared to control worksites; and (2) improvements in tobacco use cessation rates in the intervention group relative to the control group will be greater among production workers compared against managers and office workers.


This cluster randomised controlled trial (RCT) tested the effectiveness of the Healthy, Safe, and Tobacco-free Worksites programme intervention in increasing tobacco use cessation among workers. The worksite was the unit of intervention and randomisation. This study was a collaboration among the Harvard T.H. Chan School of Public Health, the Dana-Farber Cancer Institute, the New England Research Institutes in Boston, Massachusetts, and the Healis-Sekhsaria Institute for Public Health in Mumbai. This study was approved by the Healis and Harvard Chan School Institutional Review Boards, as well as by the Indian Council of Medical Research, and has been registered with and the Clinical Trial Registry of India.

Study population and sample

We recruited 22 worksites to this study, 2 of which were included in a pilot test of the intervention. Twenty worksites were included in the full-scale RCT; 10 were randomly assigned to intervention and 10 to control. Worksite eligibility criteria included: (1) employ at least 200 production workers; this definition was expanded to include at least 60% of the workforce to be comprised of production workers, with the goal of recruiting medium-sized to large-sized companies with a large population of production workers; (2) be located in the Greater Mumbai area, including the Mumbai, Thane or Raigad Districts; (3) be involved in some type of manufacturing and (4) be willing to provide a current employee roster as part of survey planning. Participating worksites agreed to be randomly assigned to the intervention or the delayed intervention control condition; participate in the intervention if assigned to the intervention condition; and participate in the planned data collection activities, regardless of assigned condition. Written consent was obtained from each worksite when the worksite was recruited.

Companies recruited to the study manufactured a range of products. The recruitment process included: (1) publicising the study to key stakeholders and prospective participants; (2) identifying potential worksites using industry lists, referrals and in-person visits to industrial areas; (3) communicating with company representatives to validate worksite information and introduce the study; (4) meeting with key stakeholders at the company to determine each company's interest and eligibility and (5) confirming company participation by obtaining a formal letter of participation.25 Of the 20 companies participating in the study at baseline, 17 participated in the final survey. One intervention and two control worksites dropped out of the follow-up survey due to personnel changes (intervention site); competing priorities, including labour negotiations; and loss of interest related to dissatisfaction with being assigned to the control condition. The trial was otherwise completed as planned.

Intervention conditions

Worksites were randomly assigned to the experimental intervention or delayed intervention control condition on a rolling basis in blocks of four worksites, after each block of sites completed the baseline survey. Random assignments were made using a random number generator by a biostatistician in Boston to eliminate any possibility that investigators who conducted the baseline survey could influence the assignments. At the follow-up survey, the survey assistants were masked as to the assignment of each worksite; intervention and evaluation staff were independent of each other.

Experimental intervention group—‘The Healthy, Safe, and Tobacco-free Worksites programme’: The 9-month intervention addressed changes at the management and worker levels, and aimed to capitalise on and supplement existing efforts in occupational safety and health (OSH), including coordination with the companies’ medical and safety officers (table 1). The intervention was designed based on careful formative research, including pilot testing of a part of the intervention in two pilot worksites. All study health educators participated in a 5-day training prior to delivering the intervention.

Table 1

Intervention components by condition

Organizational-level interventions for management: Study staff provided consultation on and recommendations for the adoption and implementation of a worksite tobacco control policy. All worksites were encouraged to implement a tobacco control policy; the consultation with management focused on compliance with legislation banning smoking in public places, and additionally encouraged expansion of policies to include prohibiting use of smokeless tobacco. As part of efforts to support and build on the company's existing OSH structure, a safety expert conducted an industrial hygiene walk-through to identify potential workplace hazards, and provided a brief consultation based on the report, recommending changes to reduce exposures.

Individual-level interventions for workers: All workers were eligible to participate in the intervention. The programme targeted tobacco users and non-users, with the aim of providing support for quitting and building social norms around tobacco control, and was specifically designed for the production workers. The intervention centred around six health education events for workers conducted by trained study health educators (table 1). Each event was offered onsite during multiple 15–20 min sessions delivered during a single day at every intervention company. These health education events and corresponding materials were designed to increase workers’ understanding of the risks associated with tobacco use, their motivations to quit or help someone quit, and the skills and social support needed for cessation. The information was provided through discussions, audiovisuals and participatory activities. The messages and materials, additionally, linked tobacco control with the work environment to enhance their salience and communicate messages about OSH. Worksite OSH improvements were highlighted as ‘success stories’, and used health education events to communicate changes made to the work environment and worksite policies to workers.

Control group: Worksites randomised to this condition were offered two health education events as incentives for participation. One health education event—either on HIV/AIDS or stress management—was provided between the baseline and follow-up surveys; this event was unrelated to tobacco use so as not to influence the cessation rates in the control group. After the follow-up survey, a tobacco-related health education event was provided (table 1).

Data collection

At each survey time point, we surveyed a census of permanent workers employed by the worksite, based on a roster provided by the company. To ensure that we included all employees, new workers not included on the roster were also surveyed.

At baseline, prior to randomisation, we surveyed workers in the 20 worksites on a rolling basis as worksites were recruited to the study (July 2012–July 2013). Survey administration, conducted onsite in English, Hindi and Marathi, included options for interviewer and self-administered formats; we used highest education level attained as a proxy for literacy and offered those who completed grade 12 or higher the self-administered version, and those with less education were interviewed by a trained interviewer. All workers provided verbal consent, as is culturally appropriate, prior to the beginning of each survey. The follow-up survey was conducted using the same methods in 17 of the 20 sites, between 14 and 18 months postcompletion of the baseline assessment (December 2013 to November 2014).

In addition, we tracked implementation of the intervention protocol. For the management intervention, senior study staff responsible for this intervention completed process tracking forms for each meeting, including meeting attendees, total meeting time, and topics addressed. Health educators tracked the worker-level intervention; for each health education event, process tracking included the total number of participating workers, average time spent on each session with workers, whether or not specific topics or activities were implemented as per the protocol, and the number of materials displayed at each event. These data were used to track fidelity to the intervention protocol and dose of intervention delivered.


Tobacco use and cessation were measured by self-report using standard items.26–28 Tobacco use status was assessed at baseline and final, and included assessment of current users of any tobacco, using separate items to assess smoking and use of smokeless tobacco. Tobacco use cessation was defined as cessation of using any tobacco product, measured among respondents to the follow-up survey who reported using tobacco in the last 18 months, to capture those using tobacco at the beginning of the intervention. Among these users, we defined 6-month continuous cessation (primary outcome) as having quit using any tobacco in the past 18 months and not using any tobacco in the past 6 months. In the same sample, 30-day tobacco quit status (secondary outcome) was defined as having quit using tobacco in the past 18 months and not using any tobacco in the past 30 days. No changes were made to trial outcomes.

Covariates: Sociodemographic characteristics were measured by self-report using standard items, including age, gender, tenure at worksite (less than 1 vs 1 year or more), economic well-being (ownership of refrigerator, vehicle or neither), education and marital status. We also assessed employee's job title using information provided by the worksite; we coded job title as production or non-production based on employees’ department and job title provided by the employer.

Statistical methods

Initial power calculations were based on a planned recruitment of 20 worksites randomised into the two groups with an average of 72 tobacco users per worksite for a total sample size of 720 workers per group. Based on prior findings regarding quit rates among tobacco users in Mumbai,29 we assumed that 2% of users in the control group would quit using tobacco; also assuming a within-worksite correlation of 0.05, this sample size provided 80% power to detect a quit rate of 9% in the intervention group as statistically significantly different from the control group rate at the 5% significance level.

We controlled for the clustering of workers in worksites in all analyses. Preliminary analysis assessed balance in the two conditions with respect to age, gender, occupation/position, worksite tenure, economic well-being, education, marital status and tobacco use using mixed-effect linear modelling methods, with intervention as a fixed effect and worksite as a random effect. For analysis of the intervention effect, we used mixed-effect logistic regression analysis of quit status on intervention group (fixed effect), controlling for worksite (random effect). We did not control for randomisation block.


Twenty manufacturing worksites that employed 7633 workers were recruited to the study on a rolling basis. A total of 6880 workers (90.1%) responded to the baseline survey. Three of the 20 worksites did not complete the study (1 intervention and 2 control). As per the protocol, all workers present on the day of the follow-up survey were invited to participate in the survey. Based on the number of workers included on the worksite rosters at baseline, the estimated response rate in these nine intervention worksites was 89.6% (3362/3753), and in the eight control worksites, 91.2% (2776/3043). For the follow-up survey, the estimated response rate was 78.3% (3117/3980) in the nine intervention worksites and 88.9% (2976/3346) in the eight control worksites.

Among those completing the follow-up survey, 765 (24.5%) from the intervention worksites and 704 (23.7%) from control worksites reported they were tobacco users at the beginning of the intervention, and were included in the analysis of cessation (figure 1)

Sample characteristics: Sociodemographic characteristics of workers in the two groups were comparable at baseline, with no statistically significant differences (data not shown). At baseline in the 20 worksites, the tobacco use rates were 24.8% in intervention worksites and 20.0% in control worksites (p=0.44). We compared the baseline sample characteristics and tobacco use prevalence between worksites that completed the study and those that dropped out (data not shown); there were no significant differences in tobacco use prevalence (23.1% in worksites completing the study, 21.2% in the three that dropped out) and in most other covariates; where differences were significant, the magnitude of the difference was small (eg, workers in worksites completing the study were slightly older, somewhat more likely to be employed for more than a year, and somewhat more likely to be married compared with those in sites that dropped out). There were no differences in the sociodemographic characteristics of respondents to the follow-up survey between intervention and control worksites (table 2).

Table 2

Participants’ characteristics by intervention group among participants in the follow-up survey in the nine intervention and eight control worksites

Intervention delivery: Based on the analysis of the process tracking data, 7 of the 10 intervention worksites implemented all 6 planned health education events; 2 worksites implemented two events, and 1 worksite did not conduct any event and dropped out of the study. For the intervention for managers, the tobacco policy consultation was implemented in all 10 intervention worksites; 1 of the 10 worksites refused the OSH consultation.

Tobacco use cessation: As shown in table 3, the intervention resulted in a doubling of the 6-month cessation rates among workers in the intervention worksites compared to those in the control; this difference in rates was not statistically significant for the overall sample (OR=1.81; p=0.13) or for production workers only (OR=2.29; p=0.07). The within-worksite correlation was 0.10. The magnitude of the difference in 30-day quit rates between the two groups was comparable to the 6-month rates and was statistically significant for production workers (OR=2.25, p=0.03), although not for the sample overall (OR=1.70; p=0.12).

Table 3

Tobacco use cessation by intervention group in the nine intervention worksites and eight control worksites completing the follow-up survey

We computed a post hoc analysis of the power to detect a between-group difference in this study, using the observed 6-month cessation data. The within-worksite correlation for the resulting sample was 0.10, and the quit rate in the control group was 4.8%. Given these findings, the actual power of the study to detect the observed quit rate (8.4%) in the intervention group as significantly different from that in the control group was 21%. The final sample size provided 80% power to detect a difference of 10 percentage points between the two groups.


We found a doubling of quit rates between intervention and control worksites among production workers; this difference was statistically significant for 30-day quit rates but not for 6-month sustained cessation. To our knowledge, this is the first worksite tobacco control study in India using a rigorous randomised controlled design that has demonstrated the efficacy of a worksite intervention for workers’ tobacco use cessation.

There is an emerging literature exploring the potential of worksite approaches to reducing risks associated with NCDs. For example, Prabhakaran et al16 reported that a worksite intervention programme to reduce cardiovascular disease (CVD) risk factors among employees of six industrial worksites was successful in reducing overall CVD risk, and reported significant reductions in mean body weight, waist circumference, blood pressure, serum cholesterol and plasma glucose levels in the intervention group relative to the control. Thakur and colleagues described the development and feasibility of a 12-month intervention in three industries to address the physical work environment, psychosocial work environment and promoting healthy behaviour, with a particular focus on NCD risk factors. Others have used oral cancer screening for individual workers to promote cessation.30–33 This manuscript contributes to this emerging literature on the potential efficacy of workplace interventions designed to improve behaviours associated with NCD risk, such as tobacco control. Building on the existing OSH infrastructure provided a potentially sustainable approach for engaging companies in tobacco control efforts.34

These findings may be limited because they rely on self-reported tobacco use. We took numerous precautions to minimise reporting bias, including use of anonymous surveys, use of separate intervention and survey staff, and informing workers that the results would only be communicated in the aggregate, and that no personal information would be provided to their employer. Biochemical validation of cessation was not feasible in this population-based study conducted in a worksite setting in which the appearance of drug testing would be unwelcome; in addition, other reports in India have found that biochemical assessments may have low sensitivity in detecting use of smokeless tobacco, especially prevalent in this population.35 ,36 Also, because of the influx of new workers, the cessation rates include those workers who were only partially exposed to intervention, possibly diluting the final results. Also, we estimated our response rate using rosters provided by the worksites, which may have been incomplete or somewhat out of date. The lower response rate in the intervention sites at follow-up may have led to some non-response bias in our estimated effects.

This study was underpowered to detect the difference in 6-month quit rates observed here. Our power was compromised by the loss of three sites, including one intervention site and two control sites. In addition, although we estimated the 6-month quit rate in the intervention group accurately for our power calculations, we underestimated both the quit rate in the control group and the within-worksite correlation. With increasing awareness of the harms of tobacco use, tobacco use cessation is slowly progressing in India. Future studies in India will need to be powered to appropriately take into consideration an increasing rate of quitting likely to be observed as part of the secular trend. Also, although these results may have relevance beyond the Mumbai area, we acknowledge that generalisability may be restricted to similar worksites in the manufacturing sector in this region.

This study used a rigorous randomised design, with worksites as the unit of intervention and randomisation; the analyses controlled for the clustering of respondents within worksites. This study was implemented with manufacturing workers who face dual risks related to occupational hazards and are likely to have higher rates of tobacco use than white-collar workers. The intervention was designed to address tobacco control in a comprehensive manner, with the additional focus on OSH based on recent recommendations from WHO, NIOSH and others.18–21 This study demonstrates the feasibility and potential efficacy of implementing a tobacco control intervention in a challenging setting, requiring support from management as well as a committed team of health educators who sometimes travelled long distances to worksites and offered intervention events across shifts to ensure broad access. Despite the challenges to implementation, the intervention was delivered as planned in a majority of sites.

In conclusion, this worksite intervention resulted in a doubling of tobacco use cessation among production workers, providing an innovative and effective response to the rising need for evidence-based tobacco control interventions in India. Few evidence-based approaches are currently available to inform the creation of an infrastructure to support tobacco use cessation. This randomised trial is of particular importance because it was effective in promoting tobacco use cessation among production workers, where prevalence rates are especially high. This intervention aimed to build on the OSH infrastructure within Indian worksites, which may be engaged to provide a platform for broad-based initiatives to support workers’ safety and health, including through tobacco control. This study points to important next steps for future worksite interventions, with important implications for policymakers within India as well as more broadly in LMICs.

What this paper adds

  • Blue-collar workers are more likely to use tobacco than white-collar workers, a disparity that is consistently observed in both high-income countries, such as the USA, and low and middle income countries (LMIC), such as India.

  • Workplace interventions that integrate health promotion with occupational safety and health have been shown to increase tobacco use cessation rates for blue-collar workers in the USA.

  • With its large population and high prevalence of multiple forms of tobacco use, India represents an important case example of the need for interventions to redress disparities in tobacco use by occupation.

  • There is a significant need to address the accelerating use of tobacco in LMICs, and to attend to disparities in tobacco use in these areas.

  • Prevalence of quitting is low in India, and few supports are available for those wanting to quit.

  • There is little evidence to inform worksite-based tobacco control interventions in LMICs, including in India.

  • This worksite intervention resulted in a doubling of tobacco use cessation among production workers.

  • As a result of this study, we now know that a worksite tobacco control intervention that builds on a company's occupational safety and health infrastructure can successfully improve tobacco use cessation rates among production workers in Indian manufacturing worksites.


Dr Sorensen expresses appreciation for the support of the Rockefeller Foundation and its Bellagio Center Academic Writing Residency, which provided time and space for preparing the first draft of this manuscript. The authors wish to thank the numerous investigators and staff members in India and the USA, who contributed to this study, including from the USA, Linnea Benson-Whelan, Ellen Connorton, Diane Chen, Josh Gagne, Tarub Mabud, Taylor Schwartz, Clara Setiawan, Melanie Silverman, Shamaila Usmani, K Viswanath, Lorraine Wallace and the Health Communication Core, Dana-Farber Cancer Institute (; and from India, Sameer Narake, Mayuri Sawant, Juhi Gautam, Nirmal Ahuja, Bharti Prabhakar, Ravi Govande, VN Prabhu and Ashish Mishra. In addition, this work could not have been completed without the participation of the 22 worksites in Mumbai, participants in the formative research conducted to plan this study, the study's scientific advisory board, and Healis-Sekhsaria Health Educators and staff.



  • Contributors GS ensured that the study maintained rigorous scientific standards, conceptualised the manuscript and was responsible for drafting the manuscript, incorporating the co-authors’ input, and finalising the draft; she is responsible for the overall content as guarantor. MP was the lead person responsible for study implementation, and contributed to writing the manuscript. LSC contributed to conceptualising and planning the intervention, and contributed to writing the paper. PP was responsible for planning and implementation of the process tracking and analysed these data for the paper; she also contributed to management of the field operations in India, and contributed to writing the manuscript. EN was responsible for planning the intervention and training of interventionists, contributed to quality control for study implementation, and contributed to writing the paper. AMS contributed to overall decisions about study design and implementation, and contributed to writing the paper. H-YK conducted the analyses for the paper and contributed to writing the paper. PCG contributed to overall study design and planning, and in conceptualising and writing the paper.

  • Funding This study was funded by the National Cancer Institute, grant number R01 CA140304 and K05CA10866.

  • Competing interests None declared.

  • Ethics approval Healis Institutional Review Board, Harvard T.H. Chan School of Public Health Institutional Review Board, Indian Council of Medical Research.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement Unpublished data from this study are currently being analysed by the research team; persons interested in accessing the data should contact the lead author.