Search strategy

The search strategy located published studies. One researcher (ZB) searched MEDLINE, PsycInfo, CINAHL, Web of Science (S2 File: search strategies) and Google Scholar databases on the 21^st February 2023. The reference lists of all included sources of evidence were screened for additional studies, alongside relevant citation searches.

Eligibility criteria

The inclusion criteria for eligible intervention studies (based on behaviour change wheel (BCW) definitions; see explanation in ‘synthesis of results’ below [23]) were: (a) directly or indirectly related to improving the health of contact centre employees; (b) published in English; (c) published since 2002. Studies published prior to 2002 were excluded as a previous review [26] identified no relevant research before this.

Evidence selection

Identified citations were collated and uploaded into Endnote (Version X9) with duplicates removed using Endnote’s duplicate identification strategy and then manually. References were uploaded to the screening tool Rayyan [32] for independent assessment by two reviewers (ZB, CS) against inclusion criteria. The same two reviewers independently screened all titles and abstracts, followed by full-text assessments for eligible citations. Any disagreements between reviewers were resolved through discussion with an additional author (LG).

Charting the data

Two reviewers (ZB, CS) developed, tested and calibrated a data-charting tool in Excel by extracting data from four randomly selected documents. Discussions of the results informed tool adaptations. For the full data-charting process, each source was charted independently by two reviewers (ZB, CS). Data was collated with any disagreements resolved through discussion.

Critical appraisal of individual sources of evidence

We critically appraised the quality of included interventions by assessing the risk of bias that each study displays. This appraisal did not impact the inclusion decisions, as guided by a scoping review framework [28]. We used the Cochrane RoB2 tool [33] to appraise randomised controlled trials, the ROBINS-I tool [34] to appraise quasi-experimental trials and the NHLBI quality assessment tool for pre-post studies [35]. One pre-post study was not appraised, as the main focus of the study was to assess the acceptability and feasibility of the pre-post trial (see S4 File for included studies reference list:15).

Synthesis of results using the Behaviour Change Wheel (BCW)

Sources identified were mapped to the nine intervention functions of the BCW (education, enablement, training, coercion, restriction, environmental restructuring, incentivisation, persuasion, modelling) [23] to systematically describe each intervention, and the behaviour change techniques (BCT) used [36]. A detailed account of the BCW is available [37]. This approach will support researchers and stakeholders to interpret the evidence-base, informing future research and practice. To address research question one, extracted characteristics summarise the extent, range and nature of the evidence. Within this, two reviewers (ZB, CH) systematically coded intervention components within included studies to a) the nine BCW intervention functions, and b) 93 BCT [36] using detailed intervention descriptions (S1 Table: intervention description table). One reviewer (ZB) had completed BCT taxonomy training. Results were synthesised using relational analysis to present the interventions by their main intended outcomes; this method allows for a rich ‘joined-up description’ within the analysis [38]. Accordingly, we present findings for research question 2 (effectiveness) and three (acceptability and feasibility) interchangeably within the results. Evidence gaps are discussed throughout to address research question four.

Selection of sources of evidence

A PRISMA study flow diagram [39] (Fig 1) details the screening process and reasons for exclusion at full text. Database searches and reference checking returned 328 records. After removing duplicates, 231 titles and abstracts were screened, and the full text of 40 records were screened. Fourteen records were excluded resulting in 26 eligible records for research question one. Two articles (S4 File:10,22) reported two separate and eligible intervention studies. Accordingly, 28 intervention studies from 26 intervention articles were eligible for research question two (intervention effectiveness). Five intervention studies were eligible for research question three (intervention acceptability and feasibility). A detailed description of each intervention is available (S1 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. PRISMA scoping review flow diagram.

https://doi.org/10.1371/journal.pone.0298150.g001

Characteristics of sources of evidence

Related to research question one, 14 studies were published between 2003–2011 and 14 between 2012–2022. Most of the 28 intervention studies were conducted in high-income countries (S2 Table: characteristics of included intervention studies): USA (6/28, 21.4%), UK (5/28, 17.9%), Australia (4/28, 14.3%), Germany (2/28, 7.1%) and one each (3.6%) in Finland, Austria, Denmark, Singapore and Taiwan China. Five interventions were conducted in upper middle-income countries (South Africa, 3/28, 10.7%; Turkey, 1/28, 3.6%, Iran, 1/28, 3.6%) and one intervention in a lower middle-income country (India, 1/28, 3.6%). No studies were conducted in low-income countries. The number of participants totalled 2,774 with samples ranging from 14 (S4 File:11,12) to 646 (S4 File:14). Most studies included contact centre advisors only (23/28, 82.1%). One study each (3.6%) recruited advisors with a disability (S4 File:4), voice problems (S4 File:5), employees who smoke (including advisors, managers, admin staff, researchers/analysts) (S4 File:14), advisors and team leaders (S4 File:16), and all employees (including advisors, admin staff, support staff) (S4 File:7). From studies reporting participant age (19/28, 67.9%), the mean was 32.5 years (mean range 23.1 (S4 File:14) to 40.0 years (S4 File:1,18)). From studies reporting participant gender (25/28, 89.3%), the mean proportion of females was 65.7% (range 19.7% (S4 File:14) to 100% (S4 File:21,26)) and males was 34.3% (range 0% (S4 File:21,26) to 80.3% (S4 File:14)). From studies reporting participant ethnicity (6/28, 22.2%), Caucasian was most represented (mean 77.7%, range 47.8% (S4 File:18) to 100% (S4 File:12)).

Ten of the 28 studies (35.7%) were randomised controlled trials (RCTs) (including two clustered RCTs), eight (28.6%) were quasi-experimental trials (controlled before and after), and ten (35.7%) were pre-post studies (within-subjects design). Five interventions were single component (5/28, 17.9%) (S4 File:4,21,22,23; note, S4 File:22 reported two separate and eligible intervention studies within one article). The remainder were multicomponent (23/28, 82.1%).

In relation to the BCW, environmental restructuring was used in 24/28 (85.7%) intervention studies, followed by training (19/28, 67.9%), education (12/28, 42.9%), enablement (10/28, 35.7%), persuasion (6/28, 21.4%), incentivisation (2/28, 7.1%), and modelling (1/28, 3.6%). No study used coercion or restrictions. The three most used BCT were instruction on how to perform the behaviour (training function), adding objects to the environment (environmental restructuring function) and behavioural practice and rehearsal (training function). See ‘Synthesis of evidence by intervention outcome’ section for full BCT details.

Twelve of the 28 (42.9%) studies were underpinned by theory, including stress/mindfulness theory (5/28, 17.9%), job redesign theory (5/28, 17.9%) and behaviour change theory/the socioecological model (2/28, 7.1%). Nine interventions lasted <3 months (32.1%), ten lasted 3–6 months (35.7%) and five >6–12 months (17.9%). Intervention length was unclear for four studies (14.3%). Most interventions occurred in an office setting and one of these interventions included a home-based component (S4 File:1). The intervention delivery/implementation location was unclear in two studies (S4 File:14,26). Over half the interventions involved researchers delivering all or part of the intervention (15/28, 53.6%). This was followed by interventions partly delivered by individuals working within the organisations (participatory research participants, team leaders, management; 5/28, 17.9%). One study each (3.6%) had all, or part of the intervention delivered by either group facilitators with previous experience of receiving the intervention, a clinical councilor/social worker, an occupational health and safety officer, a speech teacher/language therapist, an expert tobacco counsellor, or an external consultant in organisational development. It was unclear who delivered the intervention in eight studies (S4 File:4,6,9,14,21,23,24,26).

Many outcomes were measured, including health outcomes in 19/28 intervention studies (67.9%; stress-related indicators, visual fatigue, musculoskeletal discomfort, job related wellbeing, vocal health), behavioural outcomes in 6/28 studies (21.4%; sitting time, physical activity, tobacco use), indirect measures of health in 3/28 studies (10.7%; job control, job satisfaction), and intervention acceptability and/or feasibility in 5/28 studies (17.9%).

Source quality

For the RCTs, four studies had low bias for all sections, five had some concerns for the measurement of the outcome, and two of these also had high bias for adherence to the intervention (S3 Table: ROB assessment tables). One study had some concerns for assignment to the intervention and the selection of reported results, and another had some concerns with the randomization process. Risk of bias was generally higher for the quasi-experimental studies than the RCTs, typically due to confounding in five of the eight studies (S3 Table). None of these studies received low bias for all categories. Some concerns arose for deviations from the intervention due to poor adherence and for measurement of the outcome due to self-report measurements. One article (S4 File:10) lacked sufficient information to reliably judge the quality of each section. For pre-post studies, six of the included studies were judged to be ‘fair’ and three were ‘poor’ in relation to their risk of bias (S3 Table). One study (S4 File:26) did not report receiving ethical approval.

Synthesis of evidence by intervention outcome

The intervention studies were mapped to the BCW intervention categories and BCT, and synthesised to display the total number of functions used across all interventions (Table 1). The interventions were then categorised into the following sections based on the reported primary outcome or intended primary aim: i) health behaviours (sedentary behaviour, physical activity, smoking); ii) physical health outcomes (musculoskeletal health, vocal health, visual health, sick building syndrome); iii) mental health outcomes (stress, job control, job satisfaction, wellbeing). While we acknowledge that most studies measured multiple outcomes (see S2 Table for all the study outcomes i.e., S4 File:15’s primary outcome related to sitting time [health behaviour] but they also measured musculoskeletal outcomes [physical health]), this categorisation approach brings order to the synthesis and allows discussion of research question two and three within the following sections.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Summary of studies mapped to the behaviour change wheel (BCW) intervention functions and behaviour change techniques (BCT).

https://doi.org/10.1371/journal.pone.0298150.t001

Health behaviours

Physical health outcomes

Mental health outcomes

Research question one—What is the extent, range, nature, and quality of the intervention evidence?

This scoping review identified a low number of peer-reviewed, health-promoting intervention studies for contact centre advisors (28 studies since 2002). Comparatively, another review [40] identified 34 studies (2009–2017) for interventions involving sit-stand desks within a traditional office workplace. Given contact centre advisors are at high risk of poor health due to their working conditions [3, 6–8] and social determinants of health [11, 12, 41], there is an urgent need for more health interventions research in this setting.

Globally, the US holds the largest proportion of contact centres, followed by the Philippines and India [42]. Our review highlighted that interventions were mainly conducted in high-income countries (e.g., US), with few conducted in middle- (e.g., Philippines, India) and low-income countries. Contact centre advisors in low-to-middle income countries likely face even greater risks to health (lower pay, lower levels of education, poor housing, poor working conditions [43]) compared to those in higher-income countries. Accordingly, while more intervention research is needed globally, there is a particular need for health intervention research in low-to-middle income countries that employ a large proportion of global contact centre workforce.

Most participants within the intervention studies were relatively young contact centre advisors (mean of 32.5 years) working day shift hours. Only one study focused on disabled advisors and one on advisors with voice problems. Therefore, contact centre advisors underrepresented in the current evidence include older adults, night workers, and disabled workers. This is problematic as night workers are likely to suffer from additional negative effects on sleep quality, food habits, addictions, social and mental health [44], poor working conditions are likely to have a more severe impact on disabled workers, and, amidst an aging population, the highest incidence of mental health short-term disability claims within the work environment are among those aged 40–49 years [45]. Future intervention research that examines the needs of, and develops interventions for, these especially vulnerable contact centre advisor sub-groups, is warranted.

Few studies adopted an RCT design (35.7%, including two clustered RCTs). This number is low compared to 55.9% of RCT’s identified within a similar review assessing interventions for reducing sitting at work [40]. Fewer RCT’s indicates lower quality evidence to inform intervention guidance. Despite this, it is acknowledged that RCTs pose a high risk of contamination between groups, meaning future research should consider clustered RCT’s as a more feasible design within the contact centre setting [11].

The most common intervention functions examined in contact centres were environmental restructuring (adding objects to the environment) and training (instruction on how to perform the behaviour). Environmental restructuring may be common due to the need to tackle health problems associated with working for prolonged periods on a computer in a static, seated posture [46]. Training may also be common due to established, existing training structures operating within contact centres for employees. In contrast, modelling and incentivisation were seldom used. The modelling function was only used in one intervention study (S4 File:15) with stand-up champions encouraging advisors to sit less and move more at work. This was perceived ineffective, as advisors were often unsure who the champions were. Future interventions using modelling in contact centres should promote awareness of the champions, and may find the effective use of movement champions in non-contact centre office environments informative [47]. Regarding incentivisation, only one job redesign intervention (across two studies) aimed to change behaviours through measurement and reward structures (bonuses, raises, management approval). This may be because job redesign interventions require organisational commitment to adjust structural components, or the financial cost of incentives is too high for centres. Health interventions within non-contact centre office environments have effectively used financial incentives to increase employee health [48], which may be informative for future interventions using this method in contact centres. Finally, no interventions featured coercion or restrictions, which have previously been perceived as unacceptable strategies within a workplace environment [49].

Less than half of the interventions identified were underpinned by theory and those without an underpinning theory were mostly ergonomic interventions to improve vocal, visual or musculoskeletal health. This is consistent with previous reviews describing a ‘strikingly small’ proportion of ergonomic intervention studies with underpinning theory [50], despite researchers identifying relevant theories [51]. Theory may help to explain the mechanisms behind the effect of an intervention, however, research has indicated that theory-based versus no-stated theory interventions do not differ in effectiveness [52]. Theory can be a valuable resource, but it does not always ensure the effectiveness of interventions; theory may be inconsistently operationalised (put into practice), inappropriate for specific contexts or flawed [53, 54].

Few interventions were implemented long-term, with the longest being 1-year. No interventions had follow-up data collection points beyond 1-year, which is similarly reported in another workplace health intervention review [40]. Most interventions were office-based, with only one containing a home-based component (S4 File:1). This is problematic, as the COVID-19 pandemic sparked a shift to hybrid working, with 64% of contact centre advisors working remotely in 2021 and this predicted to continue in the long-term [55]. Accordingly, there is an urgent need for contact centres and researchers to understand the needs of hybrid/remote workers when developing, adopting and implementing health-promoting interventions. More long-term follow-up intervention studies are also needed.

The multiple outcomes evaluated across the identified interventions may be a result of the many behavioural and health issues that contact centre advisors face. However, despite being linked to work-related stress [56] and social determinants of health [13–16], only five intervention studies targeted physical activity/sedentary behaviour, and only one study targeted smoking. Further, no intervention targeted alcohol consumption or diet. This demonstrates a gap in the evidence compared to workplace health interventions targeting diet (17 identified) [57] and alcohol consumption (18 identified) in traditional office environments [58]. Future research may explore whether behavioural interventions reported as effective in more traditional office environments, are equally effective for contact centre employees.

Research question two—What is the current evidence regarding intervention effectiveness?

Most interventions reported positive effectiveness results for the primary intended outcome. Only four interventions failed to report effective results, including an ergonomic checklist (S4 File:10[study 2]), a screen filter to reduce visual fatigue (S4 File:4) and two studies putting plants into the workplace to improve wellbeing (S4 File:22[study 1 and 2]). These studies can be interpreted as being amongst the most simplistic interventions, based on the BCW intervention function mapping, with the latter three being single component interventions. This is in-line with a systematic review assessing workplace health promoting interventions which stated that multi-component interventions were more effective than the single-component interventions [59].

Four (14.3%) interventions identified in this review are cited within health strategy and guidance documents for contact centres, as produced by trade (labour) unions and private sector organisations [21, 22, 60]. These interventions focused on air quality and ergonomic training solutions. In contrast, to the authors’ knowledge, the remaining 24 intervention studies identified in this review are not cited in any health strategy or guidance document for contact centres. This highlights a lack of translation of published scientific evidence into practice, and the need for better collaboration between researchers and stakeholders concerned with health promotion in contact centres. Further, there is a need for evaluation of the ‘good practice’ recommendations within existing documents to understand their effectiveness, acceptability, and feasibility. In combination, these actions can help produce evidence-informed health strategy and guidance documents, and promotion of those documents at scale across the industry could improve the health of hundreds of thousands of contact centre advisors.

Research question three—What is the current evidence regarding intervention acceptability and feasibility?

Overall, there was a low proportion of studies reporting acceptability and/or feasibility (5/28 studies). All studies appeared acceptable to participants (S4 File:1,13,15,16,17). Regarding feasibility, one study stated that stand-capable hot desks were not feasible (S4 File:16) and one study highlighted lack of time as a potential barrier as participants needed more time to practice a mindfulness programme (S4 File:1). This is likely to be a common challenge for contact centre interventions, as advisors have little autonomy and flexibility surrounding break times [61]. More acceptability and feasible research is needed within this setting due to its unique working conditions.

Strengths and limitations

This is the first systematic scoping review on this topic to be submitted for peer-review and provides a needed update on a non-peer reviewed publication in 2010. This review utilised a comprehensive search strategy across four databases and google scholar to identify health-promoting interventions for contact centre advisors. To ensure all relevant studies were captured, the search strategy and inclusion criteria remained broad, ensuring a physical, mental and social health focus. The coding framework was based on the established BCW and BCT to systematically describe the range and nature of the evidence, providing structure to the findings. The risk of bias assessment for applicable studies provides the reader with an overview of the quality of the evidence-base, highlighting common biases such as confounding within quasi-experimental designs. This resulted in a recommendation for future research to consider clustered RCTs as a preferable study design to reduce bias within contact centre research.

This review’s restriction to behavioural and health outcomes could be a limitation. Business and productivity-related outcomes could prove informative for contact centre stakeholders and should be considered for future reviews. This review is also limited in its capacity to make recommendations for the effectiveness of individual interventions, instead this scoping review provides a descriptive account of the available evidence [28]. Excluding studies that were not published in English was also a potential limitation, however, this did not affect the findings of the review as only three studies were not available in English, none were interventions and would not have been eligible for inclusion.