Development of an Intervention to Support Reproductive Health of Garment Factory Workers in Cambodia

Background: Garment factory workers in Cambodia are potentially a vulnerable population in terms of support for reproductive health issues and access to services, as the majority are young women aged under 30 years who have migrated from rural areas away from their family and community support. The aim of this paper was to describe the development of an intervention to support the reproductive health of garment factory workers in Cambodia. Methods: The research was conducted by a multidisciplinary team with backgrounds in public health, linguistics, digital cultures and service delivery in a suburb of Phnom Penh where many garment factories cluster. Informed by intervention mapping approaches, we conducted a needs assessment with general and participant observation and semi-structured interviews, followed by intervention development activities including specifying possible behaviour change, designing the intervention and producing and rening intervention content. Results: Our research identied some challenges that Cambodian garment factory workers experience regarding contraception and abortion. Concerns or experience of side-effects were identied as an important determinant leading to non-use of effective contraception and subsequent unintended pregnancy. Financial constraints or a desire to space pregnancies were the main reported reasons to seek an abortion. Information about medical abortion given to women by private providers was often verbal, with packaging and the drug information leaet withheld. Given the observed widespread use of social media among factory workers, we developed three short ‘edutainment’ videos about contraception which were evaluated after one month. In addition we adapted three informative videos made by Marie Stopes International (MSI) from English to the Khmer language, and also adapted the MSI medical abortion ‘Mariprist’ instruction leaet to a simple video format. Conclusions: We describe the development of an intervention to support reproductive health among garment factory workers in Cambodia. These videos could be further improved and additional videos could be developed. More work is required to develop appropriate and effective interventions to support reproductive health of garment factory workers in Cambodia. Trial registration: N/A

we adapted three informative videos made by Marie Stopes International (MSI) from English to the Khmer language, and also adapted the MSI medical abortion 'Mariprist' instruction lea et to a simple video format.

Background
Cambodia has made considerable improvements in economic and health indicators in the last two decades, with much of Cambodia's economic growth attributed to garment exports. (1) (2) In 2015, the estimated number of garment factory workers in Cambodia was estimated to be around 650,000, of whom 85% were female.(3)(4) Garment factory workers are potentially a vulnerable population in terms of support for reproductive health issues and access to services. A study of female factory workers conducted in 2014 reported that 80% were younger than 30 years, half had only primary school education, and most had migrated from rural areas away from their family and community support and did not have adequate access to affordable, high quality reproductive health information and services to meet their diverse needs. (5) In Cambodia, reproductive health services are available through the public and private sectors, including nongovernmental organisations (NGOs). A wide variety of contraception methods are available. Abortion was legalised in 1997 and medical abortion was approved by the Cambodian Ministry of Health in 2009. (6) Previous studies have examined medical abortion provision in clinics and pharmacies. A study in 2015 reported good knowledge of medical abortion among health care providers in clinics and pharmacies in Cambodia. (6) A study of clinical outcomes of medical abortion in 2019 reported no increased need for additional treatment among women seeking medical abortion from pharmacies compared with clinics, but that women were less likely to feel prepared for what happened after taking the pills or to be offered contraceptive methods when getting medical abortion pills. Users of the different services varied with 26% of women in the pharmacy group being garment factory workers compared with 9% in the clinic group. (7) This current study was conceived based on the prior observation that many independent private healthcare providers are located in the vicinity of garment factories, but not much is known about what kind of reproductive health services they provide, particularly medical abortion and contraception, and what is the experience of clients who visit them. The project is a collaboration between Marie Stopes International Cambodia (MSIC) and three universities in the United Kingdom. The study protocol and description of family planning practices are described elsewhere. (8) (9) The aim of this paper is to describe the development of an intervention to support the reproductive health of garment factory workers in Cambodia.

Methods
This research project was conducted during 2018-19 in a suburb of Phnom Penh where many garment factories cluster, with the co-operation of three factories and a number of private healthcare providers. We were a multidisciplinary team with backgrounds in public health, linguistics, digital cultures and service delivery. The research was conducted by different members of the research team at different times. Each member focused on their area of expertise and methods, while participating in data gathering done by other members. Findings from the different phases and approaches informed subsequent data gathering, in an iterative process aimed at understanding the wider context of medical abortion and contraception rst (needs assessment), and develop possible interventions later. Research methods included general and participant observation, semi-structured interviews and periodic meetings in the UK and Cambodia to discuss ndings. (8) The methods were informed by intervention mapping methods,(10) but in a more holistic manner, which took into consideration the different disciplines involved. We aimed to conduct a needs assessment followed by intervention development activities including specifying possible behaviour change targets, designing the intervention and producing and re ning intervention content. Coding of the data was done in both a deductive way, utilizing codes derived from the structure of the intervention mapping needs assessment process, and inductively, using codes that emerged from the more open-ended methods, including observations and semistructured interviews with factory workers. The end result was not a pure application of intervention mapping but rather a modi ed approach to it.
In order to learn about the wider context of factory workers' choices regarding reproductive health, we conducted a short ethnographic study of their lives, living conditions, work places, and digital technology use. We joined several women during their lunch breaks in factory grounds, chatting about their experiences and getting a sense of who they were friends with, what they discussed during breaks, how they found their current jobs and what their future plans were. We also shared meals with them in nearby restaurants after working hours and at weekends, and were able to visit some of their homes. After an initial period of "hanging out" we conducted semi-structured interviews that focused primarily on their mobile phone use, to understand what sites/apps they used, what content they consumed online and off, and whether they searched for speci c information. Some of the interviews covered healthcare and health-related topics. We also did semi-structured interviews with owners of mobile phone shops and service providers near the factories, to triangulate what we learnt from the interviews about mobile phone use.
In order to learn about garment factory workers' use of reproductive health services we observed the practice of and interviewed garment factory in rmary workers and a variety of private healthcare providers ranging from simple drug stores to larger clinics. Interviews explored practices around reproductive health, especially with regards to contraception and abortion provision. With the co-operation of selected private providers we conducted interviews with women seeking abortion services. Interviews explored reasons for seeking abortion services and their experiences. Further details regarding recruitment, data collection and analysis are reported elsewhere. (9) During the course of the project we organized several meetings to share and discuss behaviours observed in research that could feasibly be targeted and potential interventions resulting from the needs assessment activities, and then design, produce and re ne a pilot intervention. In order to address limitations in the provision of accessible contraceptive information and information materials for factory workers in general and speci cally at the time of seeking medical abortion we decided to make three 'edutainment' videos about family planning and one instructional video about how to take MA correctly. We collaborated with a Cambodian media company (Phare) and during the process sought input from garment factory workers in a pre-testing workshop. (11) The videos were designed to target barriers to contraception use that were identi ed in the needs assessment, ethnographic and interviews studies. (9) The videos were released through MSIC's communication channels and their impact was evaluated by an independent digital media company (Havas Champagne media).(12) Ethical approval was obtained from the Cambodia Human Research Ethics committee (number: 094NECHR), the London School of Hygiene and Tropical Medicine (LSHTM) (number: 14646), and Marie Stopes International (MSI) ethics committee (number: 003-18). This research project was funded by the Arts & Humanities Research Council (AHRC) (number: AH/R006091/1) and a linked UK Research and Innovation Global Impact Accelerator Award (GIAA). Informed written consent was obtained from all interview and focus group participants in this study.

Research context
A summary of sources of information to gather information for the needs assessment is shown in Table 1. Through observation, discussion and the limited available literature we were able to learn about the context and lives of garment factory workers in Phnom Penh.(13) There were several factories in the area each employing several thousand workers, typically working six days per week. Workers sat at sewing machines or did other tasks at individual work stations. Between the noise and the factory oors lay-out, there were limited opportunities to talk among colleagues during working hours. Meal times were when workers gathered in small groups, with different people (different ages and different place of origins) coming together rather than a xed group of friends. The breaks tended to be short, and people either chatted or looked at their phones while having their meals. In terms of living conditions, workers who came from far away areas rented rooms in apartment blocks nears the factories, sharing with family or other co-workers, often from their same area. Apartments were carved up in different rooms typically rented separately, so space was very limited and privacy of any kind hard to come by. While nearby guest houses provided the opportunity for discreet romantic relationships, factory workers tended to be surrounded by other people at all times. Shops around factories catered to the needs of the mostly young, hard-working residents: food and daily necessities shops, mobile phone sellers and service providers, and licensed and unlicensed healthcare providers, from simple drugstores to clinics with consultation rooms and beds. There were no government or NGO clinics in the immediate vicinity of our research area, although some private providers had previously received trainings from NGOs through social marketing or social franchising activities. The factories we visited all had in rmaries for rst-aid and minor ailments. Mobile phone shops sold new and used mobile phones, ranging from USD 15 for an old style feature phone to USD 200 for a smartphone. Specialized shops created Facebook accounts on behalf of their clients, did small repairs, downloaded apps and downloaded content such as lm and music, cheaper than using contract data to download online content. To complete the general picture of the life of factory workers, it is worth noting that as of January 2019, the minimum monthly wage was USD182 per month, (14) which could increase with overtime or for skilled workers, or be lower if workers did fewer hours or worked for workshops instead of established factories. Our interviewees reported paying an average rent of USD 35 per month (for a shared room) plus about USD5 in utilities. Finally, garment factory workers tended to be more in debt than the average Cambodian (40% of factory worker households are in debt, versus a national rate of 37%). (15)  Observations in factories (working space, canteen, in rmary, public spaces), in shops and markets near factories and in 1 worker's home. This provided a useful perspective in the spaces where workers live, what kind of goods and services they have access to, how far they need to go to access speci c services.
Formal interviews with 33 female factory workers. "Hanging out" at break times, over meals and at their homes for informal conversations. This allowed us to build rapport, and to contextualize, triangulate and clarify information received from a variety of sources.

Sources of information for family planning and abortion
Literature and grey literature on contraception and abortion needs and services. No research exists on use of web sources for health/sexual health information in Cambodia.
-Direct observation of online activities did not yield any nding regarding searching for family planning information.
-Direct search on YouTube for 'family planning' 'abortion' 'medical abortion' 'contraceptive pill' showed several videos on the topics, some instructional some editorial. All videos had comments, many from 2019 (regardless of when the video was rst published), which indicate increasing engagement with online sources to look for family planning information.
Interviews with factory workers indicated a strong reliance on family and friends for information related to contraception and abortion. Medical practitioners were also cited as a source of information, but less in uential. Interviewees who were asked directly denied looking for family planning information online, but some said they looked for other health information on Facebook or Youtube.

Reproductive health
We conducted 16 interviews with women seeking abortion from private providers located near the factories and 13 interviews with private providers working in different facilities such as in rmaries, pharmacies and bigger clinics. Findings from these interviews are reported elsewhere. (9) In brief, for health issues during worktime, workers were able to use the factory in rmary, and outside of work the option to use public or private providers. The main role of the in rmaries was to treat minor ailments. Only the largest factory provided contraception. In rmary staff reported that women would seek consultation elsewhere in the event of an unintended pregnancy.
Factory workers' opportunities for using contraception to prevent pregnancy were constrained by their long working hours, limiting their ability to attend distant clinics with restricted opening hours e.g. for long acting contraception. Pharmacies close to their home and workplace were therefore the most practical option for healthcare seeking. There was sensitivity and stigma related to obtaining contraception or MA. Women seeking supplies from drug stores or pharmacies would need to wait until no one else was at the counter or whisper to the provider to achieve any privacy. For young unmarried women privacy in taking visible contraception such as the pill was often essential due to negative social attitudes to sexual activity for unmarried women, yet very di cult to achieve in cramped shared living conditions. Young women either shared a room with other young women or sometimes lived with an older female relative. There was also the suggestion that young women in relationships with older, wealthier men (meeting in nearby guest houses) might be unwilling or unable to insist on condom use. For married women long periods living away from partners meant that contraception was only needed intermittently and unpredictably.
Regarding contraception use in general, women reported receiving information from a variety of sources such as family, friends, healthcare providers, or the media. Women had concerns about contraceptives. Some women reported discontinuing contraception due to experience of side effects such as 'body heat' sensation, weight loss due to reduced appetite, vomiting, menstruation changes, skin changes, dizziness and fatigue. Some women had never used contraception due to fear of side-effects including fear of infertility.
Financial constraints or a desire to space pregnancies were the main reported reasons to seek an abortion. In most cases, women obtained information about abortion from family and friends. In a few cases, women went directly to a clinic. None reported seeking information via the internet. Although most private providers were owned by a medical professional such as a doctor or pharmacist, the day-to-day running was often done by a nurse or pharmacist who was a family member or hired staff. Most had undergone formal training but some had learnt on the job or through short training courses. There was limited use of treatment guidelines or protocols. Several providers expressed a wish for more training, including Comprehensive Abortion Care (CAC) training. All of the providers offered short-acting contraceptives and medical abortion. Some larger clinics offered ultrasound, surgical abortion (vacuum aspiration) and had beds for patients. A variety of medical abortion products were available in the private providers around the factories. In some cases someone would request the drugs on behalf of someone else. Information given to women was often verbal, with packaging and the drug information lea et withheld, sometimes upon request of the women to protect their privacy in their homes, but also to prevent clients from knowing the brand name and looking for the drug elsewhere. In the cases where a drug information lea et was given it was not always in the Khmer language nor had pictorial instructions. Hence women often had to remember the sequence of taking the drugs, side-effects and potential warning signs. Information on postabortion family planning was variable and follow up was generally in the case of problems rather than routine.

Videos about family planning in Cambodia
Given the observed widespread use of Facebook and Youtube among factory workers, and the preference for receiving information via images, video and voice rather than text that emerged from interviews and observations, we searched Youtube for information on contraception, abortion and medical abortion, using a variety of search terms that re ected both medical and more popular terms. There were few results, and none of the videos we found had been posted by NGOs or government sources, possibly because such videos, although they exist, are not optimized for Khmer language and thus not ndable when searching in Khmer. The videos that existed were not professionally produced, often consisting of little more than a slide show or a static image with a voice-over, but they were very descriptive and factual, and were beginning to show views in the thousands and tens of thousands. A detailed analysis of this phase of the research will be reported elsewhere, but our working hypothesis, based on YouTube searches, interviews with factory workers and mobile providers, and the increasing number of mobile internet users in the country, was that the use of the internet to seek health information is just at the beginning for the large number of factory workers who are just starting to go online, and that it will increase signi cally in the next few years.

Intervention development
We held a mid-point workshop in Cambodia in order to discuss the research ndings and their implications. Findings were also discussed in separate meetings with stakeholders in the Cambodian Ministry of Health and other NGOs. Consistent with previous research in Cambodia, we identi ed concerns or experience of side-effects as an important determinant leading to non-use of effective contraception and subsequent unintended pregnancy.
(16)(17) These experiences or concerns were shared by potential users of contraception and healthcare providers. Some of the reported side-effects such as 'body heat', or 'weight loss' (as opposed to weight gain) did not appear to be reported in standard contraception provider guidance or materials developed for potential users. We discussed the possibility that Cambodian women could experience more pronounced hormonal side-effects compared with western women due to differences in size and metabolism. This hormonal 'mismatch' theory has been described elsewhere,(18) but it was beyond the scope of our study to explore this systematically. We also discussed similarities and differences between Cambodian womens reports and those of women in the UK talking about hormonal contraception being 'unnatural' making them feel 'out of balance', which again is not widely raised in materials developed for potential users. (19) An attempt was made to construct a logic model of the problem, as per Intervention Mapping techniques (Fig. 1). In this model we considered the health problem to be unintended pregnancy (clearly for women who did not want the pregnancy), and the 'at-risk' population to be factory workers with unmet need for contraception. We considered behaviors of the at-risk group and environmental agents, and determinants leading to those behaviors.
A summary of potential interventions is shown in Table 2 using some terminology from the World Health Organization classi cation of digital health interventions. (20) We felt interventions aimed primarily at providers would be challenging due to the number of different providers, their competing nancial and business concerns and the need for sustained effort and follow up, and was beyond the scope of the project. It was apparent that some providers had received training on reproductive health topics such as contraception and abortion from NGOs through social marketing or other programmes. Most providers were interested in more training. We facilitated additional training for those that requested it but it was beyond the scope of this project to provide a comprehensive training package. There was also interest from providers in peer-to-peer support for questions and advise leveraging social and informal ties,(21) such as through social media or instant messaging, potentially moderated by a reproductive health expert from an NGO or department health. There did not seem to be any capacity to do this at scale at the time of the study as it would have required a medium-term investment in human resources and a joint approach to be effective. Marie Stopes Ladies, an MSI global service delivery channel that has recently been started in Cambodia, is utilizing such a strategy, by setting up a Facebook group to share information among the service providers. (22) In terms of interventions aimed at factory workers, we considered ways of direct provider-to-client communication, such as enrolling potential users of contraception through private providers or in rmaries, but opted against this due to concerns about maintaining communication, due to the high frequency of phone and/or SIM switching, or potential harm arising from other people listening to messaging, since phones were often accessible to several people aside from their owner (23)(24). -Can be easier to access in private than written lea ets.
-Needs to be found online, against existing competing videos that might be less accurate but are ranked higher in search results

-Privacy issues, as it remains in search history
Given the widespread consumption of online videos, we considered that this medium could be the most effective in terms of reach and engagement, so we developed three short 'edutainment' videos about contraception. In addition we adapted three informative videos made by MSI from English to the Khmer language, and also adapted the MSI medical abortion 'Mariprist' instruction lea et to a simple video format. These videos were not formally evaluated. The three short 'edutainment' videos were aimed at users or potential users of contraception aimed to address personal determinants, particularly knowledge and attitudes, towards contraception. Our objective was to frame contraception in a positive way, whilst acknowledging these frequently experienced and discussed context speci c side-effects such as body heat but being careful not to perpetuate them or dismiss them as imagined or unimportant. We felt that videos aimed at the general public that could be watched on mobile phones could be an effective way to deliver an intervention whilst limiting potential for harm. We decided to make three videos because female factory workers are not homogenous. Working from the data we gathered from eldwork, we constructed two "personas," that is ctional characters that represent characteristics associated with distinct groups of women. (25) The rst persona represents married women, possibly with children, interested in spacing out births, and able to discuss contraception and abortion with their husbands. The second represents single women who may or may not be in a stable relationship, and who are interested in information about contraception and medical abortion, but don't necessarily discuss it with their partners.
We wrote a terms of reference and invited local media agencies to pitch ideas. We then worked in collaboration with Phare agency to produce the videos. (11) This involved providing the agency with information on target audience and a visit to garment factory area for a pre-testing workshop with garment factory workers to seek their feedback on some prototype videos and opinions on how to further improve the content. During this workshop it was discussed that some of the factory workers would act in one of the videos. There was a two-way process by which the scripts were developed. English language versions of the scripts are shown in Appendix 1. The videos were lmed and Khmer subtitles added and the videos were released sequentially through MSIC's social media communication channels. (26)(27) The rst video ("Mother") was aimed at married parous women, showing a conversation between a mother and her children whereby the woman was experiencing side-effects from using contraception with the children providing reassurance in a light hearted way. The key message aimed to be that there is a contraceptive method that will t with you. This was lmed by the agency in Battambang. The second video ("Love") was aimed at unmarried nulliparous women with the concept of a romance story being watched on a smartphone by a goup of factory workers. The key message aimed to be being in a romantic relationship and still having other life opportunities because of using contraception. The romance scenes were lmed in Battambang and the factory worker scenes were lmed in Phnom Penh. The third video ("Baby") was aimed at married or unmarried women with the concept of a Khmer karaoke-style comedy dance/song with a key message of nding the right method that ts with you and your family.
The videos were posted at one monthly intervals on Facebook and evaluated by Havas Champagne, a Cambodian media company after one month.(12) The videos were boosted on the MSIC Facebook page aimed to engage the target audience, female factory workers and reach the wider public. Geo-targeted ads were used to identify and show content where most factories were located. A click to message campaign was also used to encourage people to send messages to MSIC Facebook page. The videos were placed on the News Feed and Facebook suggested videos feed, and Facebook Watch feed. A shortened version rather than the original baby video was released. Overall, unique reach (number of people who saw a post a least once) was 3,462,176 with 2,839,255 engagements (number of people interacting with the content e.g. like, share, comment, reaction), 7,876,734 video plays, 16,000,000 impressions (number of people who saw a post, but may include multiple views by the same people), and 25,637 click to messages (number of people who click the "Send to Message" button that will lead to MSIC Facebook messenger). An automated response was set for those who sent a Facebook message and MSIC counselors provided more comprehensive information regarding the topic they asked about and referred to MSIC clinics or services if required. Key metrics per each video are shown in Table 3. In terms of Click and engagement, the 'Love' video achieved the highest click amount to MSI. Figure 2 shows trends in total calls to the MSIC contact centre and shows the number of Facebook messages sent to MSIC during the project period. Both gures show increases through sept and oct with a decrease afterward corresponding to the timing of release of the 'love' and 'song' videos. Data was not collected from the MSIC contact centre or clinics regarding whether referrals were as a result of any speci c content. De nitions: Reach is the number of people who saw a post at least once. Engagement is the number of people interacting with the content. Example is like, share, comment, reactions. Click to Action measures the number of people who click the "Send message" button that will lead to MSIC Facebook messenger.

Discussion
In this paper we have described the process by which we developed interventions to support reproductive health of garment factory workers in Cambodia, primarily three short 'edutainment' videos about contraceptive methods. Our research identi ed some challenges that Cambodian garment factory workers experience regarding contraception and abortion. Our ndings of unmet need for contraception due social and structural constraints limiting womens opportunity to use contraception and their experience or concerns about side-effects is in keeping with previous literature.(17) Some side-effects reported, such as the experience of 'body heat' with the contraceptive pill appear to be more common in Cambodia compared with elsewhere and possible reasons for this could be further explored. Future interventions should be sensitive to the nding that contraceptive side effects may vary by context, and those critical to non-use may not be recognised in World Health Organization or other global literature and guidelines. Information about family planning often came from friends or family, and mass media campaigns in some cases. Use of private providers would in general result in a bias towards information about short-acting methods as few had the training to provide long-acting methods. Furthermore, providers may not have done formal family planning training and would reinforce myths or concerns about methods. Hence we heard many reports of contraception use and subsequent discontinuation which could result in an unintended pregnancy. We considered that videos portraying contraception in a positive way whilst addressing context speci c issues could be an effective way to change attitudes and behaviour regarding contraception use.
Our three edutainment videos aimed to contain different key messages aimed at both married and unmarried women. By working with a media company, it was possible to incorporate fresh perspectives, and to involve factory workers in the production of the videos. One of the main challenges was maintaining the key messages in the script when working in two languages. The videos were promoted through MSIC's communication channels and key metrics evaluated by an independent company. The videos had a greater reach and number of views if compared other videos on the MSIC Facebook page, but it is di cult to know how much of this was due to the 'boosting' effect. The 'Reach' of the videos was similar, however there were differences in the number of 'Clicks' and 'Engagement'. According to Havas/Champagne this could be explained by the creative and message of the 'Love' video which was more relatable to younger audiences. MSIC saw an increase in calls and messages especially in October, which was the month where they found the highest number of inbound calls and Facebook messages with messages doubled compared to previous month. The videos may have in uenced people to directly reach out to MSIC contact centre for more information and led to subsequent use of services. However data was not collected from the MSIC contact centre or clinics regarding whether referrals were as a result of the videos and therefore we cannot assume a causal relationship between our videos and the increased contact centre activity.
In general it can be observed that the videos were less polished or professional in production if compared with other reproductive health videos such as produced by Population Services International Khmer or BBC Media Action.(28)(29) However, they are unique in how they involved and address our targeted population in re ecting a narrative language that is familiar and appreciated and could serve as examples of videos that could be developed and evaluated further. Themes of love and romance in particular could be persued further and we recommend collaboration with creative media companies to contribute fresh ideas. The feedback from Champagne media was to use videos with shorter length and less texts on the thumbnail could improve performance in terms of reach, watch time and higher chance of better audieance retention. It is recommended to focus on mobile phones only, since there was virtually no tra c coming from other devices.
With regards to abortion, we observed that a variety of medical abortion products were widely available from the private providers, with variation in practice as observed elsewhere.(6) Witholding the lea et when selling medical abortion was sometimes done to protect the client, according to the provider, but suggests 'information asymmetry' whereby the client is less informed and able to shop around. In any case, women purchasing medical abortion often had to rely on verbal information and remember how to take the product, what to expect and if they should return for follow up. As we were not able to identify any reputable existing videos, in addition to the videos described in this article, we adapted a medical abortion lea et into a video format which can be viewed on the MSIC Facebook page. (26) This was not formally evaluated in the way the contraception videos were. The impact of providing support for medical abortion in different formats could be evaluated in a further study. In the meantime, our ndings on lea et retention and comprehension by pharmacy clients have contributed to an evidence brief published by Marie Stopes International on supporting self-management of medication abortion from pharmacies. (30) In this project we did not develop any interventions directly aimed at providers. This was partly because it was di cult to establish the exact status of the providers in terms of current licence, approval to provide certain types of services and quality of medications. Whilst we found varied adherence to best practice, it was encouraging that some providers expressed interest in further training or peer-support groups. Interventions to support or systems to inspect private providers should be explored further.
Strengths of this study include the interdisciplinary approach, collaborations between different academic institutions, an NGO and a media company, and the independent evaluation of the videos. This study has several limitations. Being a qualitative study, the number of participants was relatively small and may not be representative of neither garment factory workers nor providers. Our goal was to reach a good variety of participants, but it is possible that the factories and providers who agreed to participate in the study had better conditions and services compared with other ones. Regarding those seeking abortion services, single women were under-represented in our interviews, and remain a very hard group to reach. Negative experiences of contraception may have been over-reported as many of the women we interviewed had experienced an unintended pregnancy. It is also possible that women's experiences of contraception, particular side-effects were di cult to translate from Khmer to English, although this was mitigated by having a Cambodian midwife and a linguist on the research team. Intervention mapping provides a rigorous detailed step by step approach to developing interventions.(31) In our experience, although we found elements of the approach useful, particularly in conceptualising the problem, other aspects were more challenging to apply. The range of local partners involved were limited to MSI, pharmacists, drug store owners and the factories which combined with the limited time frame and funding for the project constrained our options for the types of intervention we could consider. It was challenging to explain and translate details of intervention mapping across languages, cultures and disciplines working in different paradigms, especially as most of the team were unfamiliar with this approach. A strength of our work was the wide range of disciplines, cultures and perspectives we came from. As a result, our discussions identi ed similar or overlapping constructs which different people labelled in different ways. In the process of allowing the range of perspectives to contribute in ways that participants felt comfortable we abandoned some of the IM mapping steps which were experienced as too rigid, formulaic and as coming from and using labels from one particular paradigm. We sought to employ participatory approaches in our needs assessment, factory workers views and local MSI staff were involved in developing the concepts for the videos. We obtained feedback from factory workers regarding the concept and content of videos allowing them to inform the process. Some factory workers volunteered to take part in a video. The process resulted in clearer documentation of interventions developed, but alternate interventions might have been developed using the approach applied more strictly or in a different way.

Conclusions
In conclusion, in this paper we describe the development of an intervention to support reproductive health among garment factory workers in Cambodia. These videos could be further improved and additional videos could be developed. More work is required to develop appropriate and effective interventions to support reproductive health of garment factory workers in Cambodia. Availability of data and materials

List Of Abbreviations
The data that support the ndings of this study are available on request from the corresponding author [CS]. The data are not publicly available because they contain information that could compromise research participants' privacy and consent.

SISTER:
If you want a long term family planning Implant is a good choice. It is effective for 3-5 years. It is a small soft plastic pipe. It takes the provider only 5 minutes to insert Implant to put it under arm skin or remove it. And when you want to have another child you just go to see provider to remove it. When it is take out you can have the baby again as normal.

SISTER:
The Intra Uterine Device or IUD is effective for 10 years. It is small and does not release hormones. It stops sperm from meeting the eggs to fertilise. No surgery is required to insert and remove IUD. It makes woman have regular periods.

MUM:
Thank you kids for all your advises! BABY: Don't worry Mum! Be con dent with your choice for our future. We are here supporting you. Now you know there are other contraception methods for you to choose.

VOICE OVER:
For more information about nding the right method that ts with for your health, or if contraception goes wrong and you need emergency contraception, call Marie Stopes to speak to an adviser 012 999 002/098 999102. Or nd us on FB athttps://www.facebook.com/MarieStopesKh Figure 1 Logic model of the problem