Scolaris Content Display Scolaris Content Display

Base de datos Cochrane de Revisiones Sistemáticas Protocolo - Intervención

Strategies to improve the implementation of workplace‐based policies or practices targeting tobacco, alcohol, diet, physical activity and obesity

Declaraciones de intereses de los autores

Versión publicada: 19 diciembre 2016 Historial de versiones

https://doi.org/10.1002/14651858.CD012439

Esta versión no es la más reciente

ver la versión actual 14 noviembre 2018
Contraer todo Desplegar todo

Abstract

This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:

The primary aim of this review is to determine the effectiveness of implementation strategies for policies, practices or programmes that aim to improve health behaviours or reduce unhealthy behaviours commonly associated with risk factors for chronic disease in the workplace. Specifically, this review will target interventions that address diet, physical inactivity, obesity, risky alcohol use and tobacco use.

In addition, this review will determine:

  • the effectiveness of implementation strategies on health behaviour outcomes (nutrition, physical activity, obesity, alcohol use and smoking);

  • the cost‐effectiveness of these strategies;

  • the existence of adverse outcomes resulting from the implementation of these strategies.

Background

Description of the condition

The most prevalent modifiable risk factors for chronic disease are poor diet, physical inactivity, tobacco use, risky alcohol consumption and obesity (Lim 2012). For example, the World Health Organization (WHO) estimates that 205 million males and 297 million females are obese (a body mass index (BMI) of 30 kg/m2 or more) as of 2008 (WHO 2011). In 2005, the WHO also estimated that 36% of men and 8% of women smoke tobacco, although the prevalence is lower among selected Organisation for Economic Co‐operation and Development (OECD) countries, at 21.3% and 12.7% for men and women respectively (OECD 2015; WHO 2008). Moreover, 20% of males and 27% of females are estimated to be physically inactive, with this proportion almost doubling in high‐income countries (Kohl 2012). Cumulatively, the Institute for Health Metrics and Evaluation has reported that in 2013 these risk factors accounted for over 25% of the total global disease burden, over 26 million deaths and over 650 million disability‐adjusted life years (IHME 2013).

These risk factors permeate the workforce to the extent that poor worker health has a significant impact on business economic potential. Estimates from Australia in 2009 suggest that the cost of poor health in workplaces equates to over AUD 60 billion for individuals, employers and the community alike (including medical costs, compensation costs, lost income and loss of productivity) (SWA 2015). For individuals, there is evidence to indicate that as many as one in four Australian employees will report having an alcohol use disorder (Teesson 2010). There is also evidence that between 12% and 30% of the Australian workforce smokes cigarettes and that this is higher among blue collar workers (Scollo 2015). Moreover, 57% to 66% of employed males and approximately 45% of employed females are overweight (a BMI of 25 kg/m2 or more), and 60% of males and 50% of females report not meeting recommended physical activity guidelines (Mummery 2005). All in all, the major modifiable risk factors for chronic disease place a high burden on workforces.

Description of the intervention

In 1979, the WHO proposed their 'Global Strategy for Health for All By 2000', wherein workplaces were identified as a settings to target and improve a range of health outcomes and health behaviours (WHO 1981). Indeed, in 2014 alone, adults from OECD countries spent an average of 36.8 hours per week in paid employment (OECD 2015a). This suggests that workplaces provide a unique opportunity to implement strategies to address multiple health behaviours and risk factors that are associated with chronic disease given 1) the potential for workplace‐based interventions to reach a large number of adults for prolonged periods each working day, and 2) the potential capacity for worksites to structure individual‐ and organisational‐level interventions (Pelletier 2011). Workplace‐based strategies could therefore make a significant contribution to the improvement of a range of health behaviour issues among adults.

A number of systematic reviews and meta‐analyses have been published in the last 10 years regarding the effectiveness of workplace interventions in influencing health behaviours (Anderson 2009; Barr‐Anderson 2011; Benedict 2008; Cahill 2014; Fichtenberg 2002; Fishwick 2013; Freak‐Poli 2013; Geaney 2013; Kahn‐Marshall 2012; Maes 2012; Malik 2014; Mhurchu 2010; Rongen 2013; To 2013; Vuillemin 2011; Wong 2012). Reviews of workplace interventions targeting dietary behaviour (Anderson 2009; Geaney 2013; Maes 2012; Mhurchu 2010), and tobacco use (Cahill 2014; Fichtenberg 2002; Fishwick 2013; Freak‐Poli 2013), have typically reported that such interventions yield modest improvements in these behaviours. Reviews of interventions targeting physical inactivity (Barr‐Anderson 2011; Malik 2014; To 2013; Vuillemin 2011; Wong 2012), obesity (Benedict 2008; Vuillemin 2011), or risky alcohol use (Ames 2011; Kolar 2015; Lee 2014), however, have reported mixed results, although within such reviews effective programmes have been identified.

Nonetheless, implementation of effective workplace interventions is required if they are to benefit public health (Bero 1998). 'Implementation' is defined as the use of strategies to adopt and integrate evidence‐based health interventions and to change practice patterns within specific settings (Glasgow 2012). Specifically, implementation research is the study of strategies designed to integrate health policies, practices or programmes within specific settings (for example, workplaces) (Schillinger 2010). The National Institutes of Health recognises implementation research as a component of the third stage (‘T3’) of the research translation process and as being essential if health innovations are to generate health improvements in the community (Glasgow 2012).

There are a range of potential strategies that can improve the likelihood of implementation of interventions to address healthy eating, physical activity, obesity, tobacco or harmful alcohol use. In health services research, for example, the Cochrane Effective Practice and Organisation of Care (EPOC) Group taxonomy has been developed to characterise educational, behavioural, financial, regulatory and organisational strategies that can improve professional practice and health care (EPOC 2015). Specific implementation strategies included in the taxonomy include continuous quality improvement, educational materials, performance monitoring, local consensus processes and educational outreach visits (EPOC 2015). Such strategies have also been utilised in settings such as schools (Nathan 2012), childcare services (Finch 2012; Jones 2015), and sporting clubs (Kingsland 2015) to improve implementation of evidence‐based health interventions and they could similarly be applied to workplaces to improve implementation of chronic disease prevention policies and practices.

How the intervention might work

Strategies that improve the implementation of workplace‐based health‐related policies may be effective if they address the determinants impeding implementation. However, the determinants of policy and practice implementation are complex. That is, a number of factors have been reported to impede implementation of health promotion initiatives in the workplace settings (Cherniack 2010). For example, a workshop convened by the National Institutes of Health and the Centers for Disease Control and Prevention to advance knowledge and implementation of effective strategies to reduce chronic disease risks in the workplace identified many barriers to worksite programme implementation (Sorensen 2011). These barriers included lack of employee interest, limited staff resources, cost, misalignment of incentives and insufficient support from management, while others have identified workplace financial, structural and cultural issues (Cherniack 2010). Moreover, implementation theoretical frameworks, including Damschroder's Consolidated Framework for Implementation Research and the behaviour change wheel, also suggest that barriers to implementation are complex, operate at multiple levels and include individual, organisational, cultural, social, political and other macro‐levels factors (Damschroder 2009; Michie 2011). Similarly, such frameworks suggest that a sound understanding of implementation context and barriers is required in order to correctly apply implementation frameworks and select strategies that best address the determinants of implementation (Michie 2008; Michie 2011).

Why it is important to do this review

The lack of evidence regarding effective strategies to improve the implementation of health behaviour policies in workplaces represents a significant gap in the health promotion and implementation science literature. Future workplace interventions will benefit significantly from a comprehensive review of strategies to improve the implementation of interventions targeting diet, physical inactivity, risky alcohol use, tobacco use and obesity. This review will also provide a firm evidence base for health promotion practitioners, as well as other end‐users including employers or insurers, regarding the design and implementation of interventions to promote healthy behaviours within workplaces.

Objectives

The primary aim of this review is to determine the effectiveness of implementation strategies for policies, practices or programmes that aim to improve health behaviours or reduce unhealthy behaviours commonly associated with risk factors for chronic disease in the workplace. Specifically, this review will target interventions that address diet, physical inactivity, obesity, risky alcohol use and tobacco use.

In addition, this review will determine:

  • the effectiveness of implementation strategies on health behaviour outcomes (nutrition, physical activity, obesity, alcohol use and smoking);

  • the cost‐effectiveness of these strategies;

  • the existence of adverse outcomes resulting from the implementation of these strategies.

Methods

Criteria for considering studies for this review

Types of studies

Given the often complex nature of evaluation studies of implementation trials, we will include a broad range of study types in this review. Although randomised controlled trials (RCTs) are considered the most reliable and robust design for establishing the effectiveness of an intervention, there are often practical issues regarding their implementation that preclude them from being a feasible design type in the context of workplace interventions. As such, we will include any trial with a parallel control group including:

  • RCTs and cluster‐RCTs;

  • quasi‐RCTS and cluster quasi‐RCTs;

  • controlled before and after studies (CBAs) and cluster‐CBAs.

We will only include studies that 1) compare a strategy to improve implementation of a healthy eating, physical activity, alcohol or smoking cessation policy with no intervention or ‘usual practice', or 2) compare two or more strategies to improve implementation of healthy eating, physical activity, alcohol or smoking cessation policies.

There will be no restriction on the length of the study follow‐up period, language of publication or country of origin. However, we will exclude studies that do not include implementation of a workplace policy or practices as a specific aim. We will not include studies that do not report baseline measures of the primary outcome. We will exclude studies with only one intervention or control site in line with the Cochrane EPOC Group recommendations (EPOC 2015).

Types of participants

We will include studies undertaken in any ‘workplaces’ or ‘worksites’, in any location in any country, which are staffed by paid employees. We will include workplaces of any industry sector including health, education, finance, retail manufacturing, information technology, agriculture, construction or mining. Participants could include paid employees at any level of the workplace organisation, or other officials or organisations who could influence the implementation of workplace health‐promoting programmes, practices or policies.

Types of interventions

We will include any intervention with the primary intent of improving implementation of a healthy eating, physical activity, alcohol cessation or smoking cessation policy, practice or programme in a workplace, to improve the health of employees (e.g. healthy cafeteria options or smoke‐free policies). Interventions could be based on quality improvement initiatives, education and training, performance feedback, prompts and reminders, implementation resources, financial incentives, penalties, communication and social marketing strategies, professional networking, the use of opinion leaders or implementation consensus processes, as well as other strategies included in the EPOC taxonomy (EPOC 2015). Interventions may be singular or multicomponent. We will also include interventions to support the implementation of strategies in the workplace to enhance the use of external services to encourage health behaviour change of workers, such as incentive schemes to encourage gym membership. We will exclude interventions focused outside of the workplace.

Types of outcome measures

Primary outcomes

The primary outcomes will be any measure of the implementation of a workplace policy, procedure or practice to improve the diet, physical inactivity, obesity, risky alcohol use and tobacco use of its employees. For example, the percentage of workplaces implementing a food service with menu labelling, or the mean number of health‐promoting practices implemented by workplaces to promote physical activity. Data on these outcomes might be obtained from self–report measures (e.g. completed by workplace staff or health promotion officers), direct observation by researchers, audits of workplace records or audits of data collected by external organisations (e.g. parent company, government). Such outcome data could be extracted from the primary outcomes of implementation initiatives or process evaluation data from trials.

Secondary outcomes

  1. Any measure of diet, physical activity (including sedentary behaviours), tobacco or alcohol use, or weight status. Such measures could be derived from any data source including direct observation, questionnaire, or anthropometric or biochemical assessments. We will exclude studies focusing on malnutrition/malnourishment.

  2. Estimates of absolute costs or any assessment of the cost‐effectiveness of strategies to improve the implementation of policies, practices or programmes in workplaces.

  3. Any reported adverse consequences of a strategy to improve the implementation of policies, practices or programmes in workplaces. This could include impacts on worker health (for example, an increase in injury following the implementation of physical activity‐promoting practices), workplace operation or staff attitudes (for example, impacts on staff motivation or cohesion).

Search methods for identification of studies

We will perform searches for peer‐reviewed and grey literature in electronic databases, handsearch relevant journals and screen the reference lists of included trials. We will also search on the internet.

Electronic searches

We will search the following electronic databases:

  • Cochrane Central Register of Controlled trials (CENTRAL, current issue);

  • MEDLINE (up to 2016);

  • MEDLINE In‐Process (up to 2016);

  • The Campbell Library;

  • PsycINFO (up to 2016);

  • Education Resource Information Center (ERIC) (up to 2016);

  • CINAHL (up to 2016);

  • SCOPUS (up to 2016).

The MEDLINE search strategy is described in Appendix 1. The search strategy will use search filters published in other systematic reviews for research design (Waters 2011), setting (Cahill 2014; Freak‐Poli 2013), physical activity and healthy eating (Dobbins 2013; Guerra 2014; Jaime 2009), obesity (Waters 2011), tobacco use prevention (Thomas 2013), and alcohol misuse (Foxcroft 2011). We will also use a search filter for the intervention (implementation strategies) that has been employed in previous Cochrane Reviews (Williams 2015; Wolfenden 2015), and was originally developed based on common terms in implementation and dissemination research (Rabin 2008; Rabin 2010).

Searching other resources

We will search the reference lists of all included trials for citations of other potentially eligible studies. We will conduct handsearches of all publications for the past five years in the journal Implementation Science and the Journal of Translational Behavioural Medicine. Furthermore, we will conduct searches of the WHO International Clinical Trials Registry Platform (www.who.int/ictrp/) and ClinicalTrials.gov (www.clinicaltrials.gov) to identify studies in progress or completed that may be eligible. We will include studies that have not yet been published in the 'Characteristics of ongoing studies' table of the review. We will also make contact with the authors of included trials, experts in the field of implementation science and key organisations to identify any relevant ongoing or unpublished trials or grey literature publications (e.g. HMIC, OpenGrey, ProQuest Dissertations and Theses).

Data collection and analysis

Selection of studies

Two review authors will independently screen abstracts and titles for potentially eligible studies. Review authors will not be blind to author or journal information. We will perform screening using a standardised screening tool developed based on the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a). We will adapt the tool, which has been previously used by the author team in other systematic reviews (Williams 2015; Wolfenden 2015), for the content and setting of this review and we will pilot it before use. We will obtain the full texts of potentially eligible trials for further examination. For all ineligible manuscripts, we will document the primary reason for exclusion in the 'Characteristics of excluded studies' table. Discrepancies between review authors regarding study eligibility will be resolved by consensus or, when required, by a third review author.

Data extraction and management

Two review authors (from pool of five authors: LW, TR, SY, CM, JW) will independently extract information from the included trials. Those extracting data will not be blind to author or journal information. We will extract data using a form developed based on the recommendations in the Cochrane Public Health Group Guide for Developing a Cochrane Protocol (CPHG 2011). We will adapt the form, which has previously been used by the author team in other systematic reviews (Williams 2015; Wolfenden 2015), for use in this review and we will pilot it before use. We will resolve discrepancies between review authors regarding data extraction by consensus and, where required, via a third review author.

Specifically, we will extract the following information:

  • Study eligibility as well as the study design, date of publication, type of workplace, country, participant/service demographic/socioeconomic characteristics and number of experimental conditions, as well as information to allow assessment of study risk of bias.

  • Characteristics of the implementation strategy, including the duration, number of contacts and approaches to implementation, the theoretical underpinning of the strategy (if noted in the study), information to allow classification against the EPOC taxonomy to enable an assessment of the overall quality of evidence using the Grades of Recommendation, Assessment, Development and Evaluation (GRADE) approach, as well as data describing the consistency of the execution of the intervention with a planned delivery protocol.

  • Trial primary and secondary outcomes, including the data collection method, validity of measures used, name of tool used, scale of measure (range), number of participants analysed in each comparison group, effect size (its value, 95% confidence interval (CI) and P value) and measures of outcome variability.

  • Source(s) of research funding and potential conflicts of interest.

Assessment of risk of bias in included studies

Two review authors will assess risk of bias independently using the Cochrane EPOC Group 'Risk of bias' criteria (EPOC 2015a). We will assign a risk of bias classification ('high', 'low' or 'unclear') to each of the following assessment criteria: sequence generation, allocation concealment, protection against contamination, blinding of outcome assessment, baseline outcome, baseline characteristics, selective outcome reporting, missing outcome data and other risks of bias (EPOC 2015a). Additionally, we will include a criterion for 'potential confounding' for the assessment of the risk of bias in non‐randomised trial designs (Higgins 2011). We will also include additional criteria for cluster‐randomised controlled trials including 'recruitment to cluster', 'baseline imbalance', 'loss of clusters', 'incorrect analysis' and 'compatibility with individually randomised controlled trials' (Higgins 2011). We will document the risk of bias of the included studies in 'Risk of bias' tables. We will also assign an on overall risk of bias to each study, giving consideration to all such study characteristics. Where required, a third review author will adjudicate discrepancies regarding the risk of bias that cannot be resolved via consensus.

Measures of treatment effect

We anticipate differences in the measures of primary and secondary outcomes reported in the included studies, which may preclude the use of summary statistics to describe treatment effects and necessitate a narrative synthesis. Nonetheless, we will make attempts to conduct meta‐analysis using data from the included trials for all outcomes where it is appropriate to do so. In such cases, for binary outcomes, we will calculate the standard estimation of the risk ratio (RR) and a 95% confidence interval (CI). For continuous data, we will calculate the mean difference (MD), where a consistent measure of outcome is used in the included trials. Where different measures are used to examine the primary outcome, we will calculate the standardised mean difference (SMD).

Unit of analysis issues

Clustered studies

We will examine all clustered trials for unit of analysis errors. Where they occur, we will document unit of analysis errors in the 'Risk of bias' tables. For cluster‐randomised trials that have performed analyses at a different level to that of allocation, without appropriate statistical adjustment for clustering, we will calculate the trial’s effective sample size for use in meta‐analysis. We will also utilise the intracluster correlation co‐efficient derived from the trial (if available), or from another source (for example, using the intracluster correlations derived from other, similar trials). We will calculate the design effect using the formula provided in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a).

Studies with more than two treatment groups

We will follow the procedures described in the Cochrane Handbook for Systematic Reviews of Interventions for trials with more than two intervention or comparison arms to avoid double‐counting study participants in meta‐analysis (Higgins 2011a). Specifically, to avoid unit of analysis errors for a study that could contribute multiple, correlated comparisons, we will first seek to combine all relevant experimental groups into a single study group to create a single pair‐wise comparison, when possible. Otherwise we will split the shared group into two or more groups with smaller sample sizes before pooling data.

Dealing with missing data

We will contact the authors of included trials to provide additional information if any outcome data are unclear or missing. We will not impute outcome data if unavailable. We will record any instances of potential selective or incomplete reporting of outcome data in the 'Risk of bias' tables. We will examine the potential impact of missing data on the pooled estimates of intervention effects as part of our sensitivity analyses.

Assessment of heterogeneity

We will assess heterogeneity via a number of methods for each outcome pooled in quantitative synthesis. First, we will visually inspect forest plots for the extent to which CIs overlap. Second, we will conduct Chi2 tests, with a P value of < 0.05 as evidence of statistical heterogeneity. Finally, we will calculate the I2statistic (Higgins 2011). We will consider an I2 statistic of > 50% as representing substantial heterogeneity and we will seek consensus between review authors regarding the appropriateness of meta‐analysis. We will not perform meta‐analysis if the I2 statistic is > 90%.

Assessment of reporting biases

The comprehensive search strategy for this review will reduce the risk of reporting bias. Nonetheless, we will generate funnel plots for each outcome and compare published reports with trial protocols and trial registers, where such reports are available, to identify instances of potential reporting bias. We will document any instances of potential reporting bias in the 'Risk of bias' tables.

Data synthesis

We will consider any implementation strategies in our review, therefore we anticipate that the pooled effect size estimation will be affected by the potential heterogeneity for each outcome. Therefore, we will use a random‐effects model to estimate the pooled effect size and its 95% CI where two or more trials with suitable data are available for the outcomes. We will not pool data from randomised and non‐randomised trial designs. Similarly, we will not pool data from non‐randomised studies of different study designs.

Given the variety of possible intervention strategies and outcome measures, we anticipate that a meta‐analysis of included trials will not be possible for the primary review outcome. In this instance, we will group implementation strategies using the Cochrane EPOC Group taxonomy (EPOC 2015), and we will describe the median effect size (with range) of the primary implementation outcomes for trials reporting dichotomous or continuous outcomes as has been done in previous Cochrane Reviews (Ivers 2012).

We will include a 'Summary of findings' table to present the key findings of the review. We will generate the table based on the recommendations included in the Cochrane Handbook for Systematic Reviews of Interventions. The 'Summary of findings' table will include: 1) a list of all primary outcomes of the review; 2) a measure of absolute or relative magnitude of intervention effect, or both (if meta‐analysis is performed); 3) the number of participants and studies addressing each outcome; 4) a grade for the overall quality of the body of evidence for each outcome; and 5) any pertinent comments to assist interpretation (Higgins 2011a). In particular, the table will provide key information concerning the quality of evidence, the magnitude of effect of the interventions examined and the sum of available data on the main outcomes.

Two review authors will rate the overall quality of the evidence for each outcome using the GRADE system (Guyatt 2011), with any disagreements resolved via consensus or, if required, by a third review author. The GRADE system defines the quality of the body of evidence for each review outcome regarding the extent to which one can be confident in the review findings. The GRADE system requires an assessment of methodological quality, directness of evidence, heterogeneity, precision of effect estimates and risk of publication bias. We will assess all GRADE domains to make judgements on the quality of the evidence. We will use the GRADE quality ratings (from 'very low' to 'high') to describe the quality of the body of evidence for each review outcome and we will include these in the 'Summary of findings' table.

Subgroup analysis and investigation of heterogeneity

We will describe the characteristics of included trials according to population, intervention, comparison, outcome and study design to establish clinical and methodological heterogeneity across included studies. We will perform subgroup analyses to explore heterogeneity where the I2 value is > 50%. Specifically, to explore heterogeneity we will form subgroups based on the intervention population, intervention, targeted health behaviour and study design characteristics. Additionally, we will perform subgroup analyses based on the scale of implementation. Specifically, we will perform subgroup analyses for interventions targeting implementation in 50 or more organisational units (workplace sites or departments).

Sensitivity analysis

We will perform sensitivity analysis for the primary trial outcomes by removing studies with a high risk of bias and by removing outliers contributing to statistical heterogeneity. If visual inspection of the forest plots identifies outliers (i.e. where the CI of a trial does not overlap with other included studies), we will contact the authors of the trial to confirm the trial data and we will perform the sensitivity analyses with the trial removed to assess any impact on the pooled estimates of effect.