Developing acceptable contraceptive methods: Mixed-method findings on preferred method characteristics from Burkina Faso and Uganda

Background: Unmet need remains high in developing regions. New contraceptive technologies may improve uptake and use. This study examines desirable product characteristics. Methods: We added a module to the female questionnaire of the PMA2020 surveys in Burkina Faso and Uganda and conducted 50 focus group discussions (FGDs) with women, 10 FGDs with men, and 37 in-depth interviews (IDIs) with providers across the two countries. FGDs with women and IDIs with providers included a semi-structured ranking exercise on pre-selected product characteristics. Results: Effectiveness, duration, few side effects, cost, and access were the characteristics most commonly reported as important in choosing a method by survey respondents across both countries. Half or more of women surveyed in each country would like a method that lasts at least one year, while 65% in Burkina Faso and 40% in Uganda said they would use a method causing amenorrhea. Qualitative findings show that women want methods with minimal and predictable side effects. Reactions to increased bleeding were negative, especially in Burkina Faso, but perspectives on reduced bleeding were more mixed. Women and providers preferred methods that are discreet and not user-dependent, and associate duration with convenience of use. Some women in Uganda expressed concerns about the invasive nature of long-acting methods, and cost was an important consideration in both countries. In the ranking exercise, discreet use and few side effects often ranked high, while causing amenorrhea and not requiring a pelvic exam often ranked low. Conclusion: Product development should consider user preferences for success in these settings.


Abstract
: Unmet need remains high in developing regions. New Background contraceptive technologies may improve uptake and use. This study examines desirable product characteristics.
: We added a module to the female questionnaire of the PMA2020 Methods surveys in Burkina Faso and Uganda and conducted 50 focus group discussions (FGDs) with women, 10 FGDs with men, and 37 in-depth interviews (IDIs) with providers across the two countries. FGDs with women and IDIs with providers included a semi-structured ranking exercise on pre-selected product characteristics.
: Effectiveness, duration, few side effects, cost, and access were Results the characteristics most commonly reported as important in choosing a method by survey respondents across both countries. Half or more of women surveyed in each country would like a method that lasts at least one year, while 65% in Burkina Faso and 40% in Uganda said they would use a method causing amenorrhea. Qualitative findings show that women want methods with minimal and predictable side effects. Reactions to increased bleeding were negative, especially in Burkina Faso, but perspectives on reduced bleeding were more mixed. Women and providers preferred methods that are discreet and not user-dependent, and associate duration with convenience of use. Some women in Uganda expressed concerns about the invasive nature of long-acting methods, and cost was an important consideration in both countries. In the ranking exercise, discreet use and few side effects often ranked high, while causing amenorrhea and not requiring a pelvic exam often ranked low.
: Product development should consider user preferences for Conclusion success in these settings.

Introduction
Of the 885 million women of reproductive age who want to avoid a pregnancy in developing regions, 214 million have an unmet need for modern contraception 1 . This figure includes many women who have ever used a modern contraceptive but discontinued its use. Primary reasons for unmet need are concerns about side effects and health risks, infrequent sex, opposition from partners, and misperceptions about pregnancy risk, all of which could potentially be addressed by adapting current methods or developing new technologies that more adequately meet women's needs 2 . If these reasons could be overcome, unintended pregnancies could reduce by as much as 59% 3 .
There is growing interest by donors in accelerating the development of new contraceptive technologies for women. Calliope, the Contraceptive Pipeline Database, currently lists 109 technologies that are in active development 4 . However, research and development is slow and expensive, and the contraceptive field is littered with failed or abandoned efforts. For example, though this list may not be exhaustive, Calliope identifies an additional 51 products whose development is currently stalled 4 . Moreover, successful product development hinges not only on bringing products to the market but also on their acceptability among intended users. To this end, the development of methods that women (and men) want to use requires incorporating users' needs and preferences into the design of new products early in the process. Evidence that can help establish target product profiles is essential to focus research and development resources on those products that are most likely to fit into users' lives and be successful. Although they differ in the specific approaches used and span different stages of the development process, several examples of efforts aiming to include user perspectives in the development of contraceptives such as SILCS diaphragms, pericoital contraception, vaginal rings and microarray patches or to elucidate preferred contraceptive features are available in the literature 5-12 .
The research presented in this article was conducted in Burkina Faso and Uganda to gauge the perspectives of potential contraceptive users, providers, and decision-makers on six long-acting methods at various stages of development and to gain broader insight into acceptable and desired product characteristics. Findings related to the six methods (a new copper intra-uterine device (IUD), a levonorgestrel intra-uterine system, a new single-rod implant, a biodegradable implant, a longer-acting injectable, and a method of non-surgical permanent contraception) are reported elsewhere 13 . The findings presented here seek to inform product development more broadly. The two sets of findings are reported separately because they were produced using different methodological approaches to acceptability research (i.e. reaction to specific products vs. perspectives on contraceptive method characteristics) and thereby produce information that is relevant at different stages of the product development process. Specific objectives addressed in this paper are to 1) describe desirable product characteristics, and 2) explore why these characteristics matter to potential users.

Methods
We conducted a cross-sectional, mixed-method study in Burkina Faso and Uganda. In each country, we added a module of 12 questions to the female questionnaire in Round 4 of PMA2020 surveys and conducted focus group discussions (FGDs) with women and with men and in-depth interviews (IDIs) with providers, as described below.

Study settings
When this study was conducted in 2016-17, modern contraceptive prevalence among women in union in Burkina Faso was 25% and unmet need 29% 14 . Implants dominated the method mix (48% of the method mix), followed by injectables (33%) 14 . In Uganda, 32% of married women were using a modern method, while unmet need was 30% 15 . Injectables (56% of the method mix) and implants (15% implants) were the most popular methods 15 . The quantitative and qualitative components of the study were conducted separately, as described below.

Amendments from Version 1
We wish to thank the reviewers for their constructive comments. We made clarifications in the text and address some comments below.
Our data were part of a larger study that also solicited opinions on specific products. The two parts of the study utilize different approaches to acceptability research and produce information that can be relevant at different stages of the product development process.
While we agree that reflecting on the comparative advantages and disadvantages of various methodological approaches is valuable, it warrants a much longer discussion than what can be incorporated in this paper. In comparison with proportional ranking or discrete choice experiments, the simple ranking exercise we used is limited in its ability to support analysis of trade-offs because it does not include a measure of difference between items. We added some text on these two points under limitations.
We included some text addressing the value of acceptability research in the broader context of product financing in the introduction and conclusion. We added text on UHC to the discussion.
Qualitative and quantitative data collection solicit information in different ways. We believe that this provides for a combined, stronger understanding but may, in some cases, result in conflicting findings. We added a reference to a secondary analysis we conducted to further explore the two sets of results on amenorrhea.
The other component of the study (around the six methods) included information on the inclusion of hormones into the description of the products. We found that "hormones" did not translate directly into the local languages, which led to conflation with other things such as experiencing side effects. We agree, however, that the notion of including hormones is qualitatively different and warrants further attention. Our analyses report by contraceptive use.
Qualitative transcripts are not made available for ethical reasons.
Any further responses from the reviewers can be found at the end of the article REVISED Quantitative component PMA2020 surveys collect data from a nationally representative sample of households and service delivery points to monitor key health and development indicators, as fully explained elsewhere 16 . A survey of all women of reproductive age living at sampled households is embedded in the household survey 16 . Data are collected by a network of trained female resident enumerators using mobile phones. The added module was implemented at the end of the regular female questionnaire. In Uganda, questions were asked of all women who were not using a permanent contraceptive method and said they would be interested in using a new contraceptive product that may become available in the future. In Burkina, questions were asked of all current users of any non-permanent method and of non-users who said they thought they would use contraception in the future. Written consent was obtained prior to implementing the female questionnaire, with additional oral consent collected prior to proceeding with the added module due to its research purpose. Data were collected in April-May 2016 in Uganda and between November 2016 and January 2017 in Burkina Faso.
In the module, women were asked about the characteristics that were most important to them in choosing a method, their preferred method duration, and whether they would choose a method that stopped their period during use (see Table 3 for exact question wording). Data were analyzed descriptively in Stata 14, adjusting for the complex sampling design and unit nonresponse to the female questionnaire with sampling variables and weights provided with the dataset, and methods appropriate for subpopulation analysis. We performed design-adjusted Rao-Scott tests (p<0.05) to compare differences in responses between contraceptive user groups (current users, past users, never users).

Qualitative component
The qualitative component was conducted in five PMA2020 enumeration areas per country, with one urban enumeration area from the region surrounding the capital and four rural enumeration areas each selected from a different region (Boucle du Mouhoun, East, North, and South-West in Burkina Faso and Central, Eastern, Northern, and Western in Uganda). The study included FGDs with women who were aged 15-17 and married or aged 18-49 regardless of marital status, FGDs with adult men (≥18 years), and in-depth interviews (IDI) with a convenience sample of facility-based family planning providers. For women, separate FGDs were conducted with current contraceptive users (separating long-acting and other methods in Uganda but combining them in Burkina Faso) and women who were not currently using a modern method.
In Burkina Faso, women and men were mobilized through local health centers. In Uganda, women were mobilized by Village Health Teams (VHTs), although nurses at health centers and women themselves helped identify potential participants in a few cases. Men were primarily identified through female FGD participants who agreed to have their partner contacted, with some men also mobilized through VHTs and other male participants. Reflecting the service delivery context in each country, providers in Burkina Faso represented the public sector while those in Uganda covered public and private service delivery channels. We aimed to complete two FGDs with each category of women, one FGD with men, and three (Burkina Faso) or five (Uganda) IDIs with providers per region.
Masters-level research assistants (RAs) conducted all interviews in the local languages (FGDs) or in English or French (providers) in April-May 2016 in Burkina Faso and February 2016 in Uganda. Three topic guides were developed (women, men, providers): each explored participants' perspectives on and experiences with long-acting reversible contraceptive (LARC) methods, with RAs being instructed to emphasize probing on contraceptive decision-making and bleeding changes in FGDs with women. In FGDs with women and IDIs with providers, these questions were followed by a semi-structured ranking exercise on important product characteristics (see Figure 1 and Figure 2 for full list). Using 15 illustrated cards with women and 16 text-only cards with providers that represented the same concepts (the additional concept for providers was "the method does not contain hormones"), participants were asked to rank the cards in order of importance for women in choosing a method. In FGDs, participants were asked to agree on one final ranking for the group.
FGDs were held in private rooms at pre-arranged locations in the communities, or in a few cases, in a quiet and discrete location outside, while providers were interviewed at their place of work. Written consent was obtained from all participants. On average, FGDs with women lasted 118 minutes, FGDs with men 108 minutes and IDIs with providers 89 minutes. All FGD participants were offered a refreshment and received soap in Burkina Faso and 20,000 shillings (USD 5.40) in Uganda. Providers received a refreshment in Burkina Faso and 12,000 shillings (USD 3.25) in Uganda. All interviews were audio-recorded and translated and transcribed into French or English. Transcripts were uploaded to NVivo 11 for coding and thematic analysis. (All qualitative interviews also included a section exploring perspectives on six products in development; corresponding transcript excerpts were not included in the analysis presented in this paper). Detailed memos described the main dimensions of each main code, and matrices in Excel were used to summarize variations in key themes by country and participant type. Ranking data were analyzed separately for women and providers by computing the frequency at which each card was ranked in the top or bottom three cards.

Participant characteristics
A total of 2,743 women in Burkina Faso and 2,403 women in Uganda completed the module added to the PMA2020 surveys; this represented 86% and 63% of all women who completed the main female questionnaire, respectively, and 99.8% and 100% of those eligible for the module. The mean age was close to 27 years old in both countries, and approximately three-quarters of participants were married ( Table 1). The proportions of both current and past contraceptive users were higher in Uganda than in Burkina Faso where 56% of women had never used a modern method.
We completed 50 FGDs with women, 10 FGDs with men, and 37 IDIs with providers across the two countries (Table 2). For FGDs with women, the mean age and parity were similar across participant groups in Burkina Faso (29 years old, 4 children) and Uganda (31 years old, 4 children). Men were 39 years old on average in Burkina Faso and 38 in Uganda; the mean number of children for men was 5 in both countries. In both countries, the mean provider age was 39 and, on average, participants had nine years of experience in their current designation. The sample spanned a range of providers, from physicians to midwives.
Results from PMA2020 survey module Table 3 shows results from questions in the PMA2020 survey module. In both countries, the five characteristics that were most commonly reported as being important in choosing a method were effectiveness, duration of protection against pregnancy, few side effects (besides bleeding changes), cost, and access. Not changing menstrual bleeding was reported as important by 18.3% of women in Uganda and 10.1% of women in Burkina Faso.
Other characteristics that were cited by more than 10% of women in Burkina Faso were that the method can be used discreetly, and in Uganda that the method is not painful to receive and does not affect sex. In addition, there were statistically significant differences between user groups. For example,  the current and past user groups included higher proportions of women who reported few side effects and no changes in menstrual bleeding as important compared to the never user group in both countries.
Almost three-fifths of women in Burkina Faso and half in Uganda said they would prefer a method that they would take every year or every few years; around 30% of women in each country favored a method that would be taken every month or few months. Among those who preferred methods lasting one or more years, the mean preferred duration was 3.8 years in Burkina Faso and 3.6 years in Uganda (data not shown). The preferred duration for women who wanted methods lasting one or a few months was 3.5 months in Burkina Faso and 3.9 months in Uganda on average (data not shown). In Uganda, 11.6% of women favored permanent methods compared to 3.2% in Burkina Faso.
Overall, 64.5% of women in Burkina Faso and 39.8% of women in Uganda said they would choose a method that stopped their period during use. We found no statistically significant differences between user groups in either country for either preferred method duration or willingness to use a method causing amenorrhea.
Results from ranking of important product characteristics Figure 1 and Figure 2 show the proportion of data collection events (summed across FGDs with women and IDIs with providers) within each country in which each characteristic was ranked among the top three and the bottom three cards in the ranking. In both countries, the ability to use the method discreetly and the method not causing side effects like headache or abdominal pain ranked high in over half of data collection events, while causing a woman's period to stop during use and not requiring a pelvic exam ranked low.
Other characteristics that often ranked high included quick return to fertility, the partner liking the method, the method being less expensive than other methods and friends or relatives recommending the method in Burkina Faso, and the ability to use the method while breastfeeding, having a regular period and the provider recommending the method in Uganda. Other characteristics often ranking low were lasting more than six % FGDs item was ranked among: % provider IDIs item was ranked among:   Qualitative results from FGDs with women and with men and IDIs with providers We collated codes from the analysis of FGD and IDI transcripts into three broader emerging themes to explain data: 1) perspectives on bleeding side effects, 2) perspectives on other product characteristics, 3) characteristics influencing method choice. Similarly, many providers in both countries reported observing a range of attitudes related to reduced bleeding among their female clients, but a strong dislike of heavy bleeding. Some women in Uganda and two-thirds of providers in Burkina Faso also mentioned irregular bleeding as an issue for contraceptive users.

Perspectives on bleeding side effects: All bleeding changes
A few women in both countries suggested that lighter bleeding was acceptable, but amenorrhea was not. A common concern among women about amenorrhea was that the blood might accumulate in the body. Several women in both countries believed amenorrhea was connected to other side effects, including abdominal pain and swelling in the belly, but also backache in Uganda. In several FGDs in both countries, women suggested the accumulated blood may lead to illness. Some women in Burkina Faso worried about dirt not being flushed out of the body in the absence of menses.  In Uganda, a number of participants had concerns about the release of a large quantity of accumulated blood once menses resumed following amenorrhea.
[Whenyoudonotbleedatall]iswhenyougetworriedthat whenitcomes,youmightbleedtodeath.Thisiswhatwefear. (25-year-oldIUDuserinUganda,1562-5) Several women in Uganda and a few providers in both countries noted that amenorrhea could cause women to believe they were pregnant. On the other hand, a few users in Burkina Faso saw the absence of menses as indicative of contraceptive effectiveness, while a few other Burkinabe women agreed that the resumption of menses after use was more important than menstruating during use.
At times, women linked heavy bleeding to a combination of heavier and prolonged periods, but they often did not explicitly differentiate between the two (or even with irregular bleeding). Women in both countries associated a range of health concerns with heavy bleeding, from weakness or fatigue to dying from massive blood loss. In many cases, women and providers saw heavy bleeding as a condition requiring medical attention. Several women said that more nutritious foods may be needed to counteract blood loss, but often saw improved nutrition as unattainable. A couple of providers in Uganda and a handful in Burkina Faso were specifically concerned about anemia.

Whenever I over bleed my blood volumes reduce and I also physically lose strength and also you may find that I do not have sufficient foods which are healthy because whenever youbleedyouhavetoeatagooddietsothatyoucanregain your weight and strength. (19-year-old non-user in Uganda, 1242-11)
Many women and a few providers in both countries described how prolonged bleeding and irregular bleeding interfered with daily life, making them difficult to tolerate. A common concern was the disruption of sex and marital relations, which a few Burkinabe and Ugandan women said could lead men to cheat or even leave their wife.

If your husband tries to have sex with you, once, twice, three times and that you say it is the time of your period, he will end up saying that you are lying. He will think it is suspicious. Some husbands can even ask you this question: 'What period flows and never ends?'…you risk packing your bags (your husband can kick you out). (45-year-old Burkinabe non-user who had previously used an implant, 2442-2)
Other areas of interference somewhat varied based on geography and other participant characteristics. For example, several women in the South-West region of Burkina Faso talked about taboos around cooking while menstruating, while a few Muslim participants said women were not able to pray during their period because they were considered impure. Another concern that was mentioned by several women from different regions in both countries was restricted mobility and fear of getting soiled in public. A 33-year non-user in Burkina Faso who previously used the implant and the injectable said: In addition, acceptability is associated with mode of delivery. Participants in half of FGDs with short-acting method users in Uganda and many providers in both countries identified discreet use as an important advantage of injectable contraception.

Mostmotherscomeforfamilyplanningwithnoconsentwith theirhusbands.So,whentheyaregiventheinjectionnobody canknowthattheyhavehadaninjectionjustforonevisit;they getaninjectionandtheygoback.(ProviderinUganda,1223)
When it comes to devices being inserted, myths around implants or IUDs migrating in the body and concerns that IUDs may negatively affect sex persist among a few women. In Uganda, several FGDs with women not currently using long-acting methods in Uganda also identified concerns with implants and IUDs, including fear of implant insertion and concerns that IUDs could cause cervical cancer. Many women and a few men believed that side effects were based on the compatibility of a method with the chemistry of each person's body (their "blood"). Many FGDs, especially in Uganda, included a discussion of compatibility testing as an element of service provision to determine what method would suit a woman best, and some believed that such "blood tests" already existed.

Whenthehealthworkerscameonanoutreacharrangement, they inserted implants in several women; some found them compatibleandyetothersfailedsotheywentforremoval,so we need health workers to come, provide us with contraceptivecounseling,testourbloodtorealizeifourblood[bodies] suits such methods then they do the insertion. (38-year old injectableuserinUganda,1251-5)
A number of users described a process of trial-and-error in finding a method with acceptable side effects, and several expressed reluctance to try a different method once they find one that fits their bodies.

Discussion
This research provides insight into important method characteristics for product development from the perspectives of end users in Burkina Faso and Uganda. In both countries, desirable product characteristics include effectiveness, a long duration of action, and few side effects. Avoiding side effects or bleeding changes was comparatively more important for women with prior contraceptive experience than for never users. Qualitative findings from FGDs and IDIs generally support survey results but give additional emphasis to aversion to bleeding changes and discreet use. Both sets of results also highlight cost and access as important considerations. While the ranking exercise did not highlight duration as important (and even suggests it may not be a priority relative to other considerations, especially in Burkina Faso), this apparent discrepancy may be explained by the definition of long-acting that was used in the cards (i.e. lasting more than six months) given that survey results indicate many women would prefer longer durations. Overall, these results are broadly aligned with key concerns linked to unmet need, including side effects, opposition from partners and misperceptions about pregnancy risk. They also highlight the fact in addition to designing product that people will like and want to use, ensuring that these are accessible and affordable will be critical aspects on the continuum from product development to successful introduction.
Half or more of women in each country would like a method that lasts at least a year, with an average preferred duration of more than three years, and close to a third would like a method lasting at least one month with an average preferred duration that would be slightly longer than the current three-month injectable. Approximately 10% of Ugandan women favored permanent methods. Qualitative results show that these preferences are at least partly driven by increased convenience associated with reduced user-dependence and number of clinic visits. Further research is needed to better understand the relative importance of fertility intentions and convenience in preferences for method duration. This is particularly important for new product development as methods allowing for self-use, such as Sayana Press, also have potential to eliminate the challenge of having to visit providers.
Mode of delivery has design implications for contraceptives that directly impact other product characteristics such as duration or side effects. From the perspectives of potential users, however, mode of delivery also has important implications in its own right. The ranking exercise highlighted the ability to use a method without others knowing as important. The opportunities injectables offer for discreet contraceptive use were especially noted in both countries. While longer-acting methods are appreciated, women in Uganda in particular expressed concerns about their invasive nature.
While having few side effects is desirable, participants recognize that side effects are unavoidable with most methods but also are aware that how they manifest varies across individuals.
Beyond the importance of counseling on the range of possible side effects associated with a method, our findings draw attention to a desire for more predictability that could be gained through compatibility testing for each potential user. Given that experimentation with different methods and anticipation of costs of treatment and removal both factor into contraceptive decision-making, more effective side effect treatment or reversibility (in the sense of the ability to interrupt use at any time with ensuing and immediate disappearance of side effects) also warrant attention as part of product development and introduction strategies.
Not all side effects were viewed equally, especially when it came to bleeding changes, which can be a cause of contraceptive dissatisfaction, discontinuation, and nonuse 17 . Our findings also link bleeding changes to method choice. Qualitative results indicate negative attitudes towards heavier bleeding, particularly in Burkina Faso, due to health concerns, anxiety over potential treatment costs, and negative lifestyle implications. On the other hand, perspectives on amenorrhea were mixed. In the PMA2020 survey module, over half of Burkinabe women and two-fourths of Ugandan women said they would use a method causing amenorrhea. However, in the ranking exercise, pausing periods was ranked consistently low in both countries, suggesting that while amenorrhea may be tolerable, it may not be desirable. A secondary analysis further examines women's perspectives on contraceptive-induced amenorrhea, reporting on the factors associated with acceptability of amenorrhea and on the reasons underlying women's attitudes 18 . In addition, the research identified persisting misperceptions, notably linked to health concerns associated with blood accumulating in the body, that should be addressed for existing and future methods.
Finally, our findings reveal that cost and access are pervasive themes. While affordability and availability directly influence economic and geographic access to contraceptive options at uptake, this research is a reminder that they factor more broadly into contraceptive decision-making through financial and practical considerations related to resupply, side effect management, and removal. As advances are being made towards including family planning into insurance schemes to promote universal health coverage schemes, attention should be paid to cover a range of family planning services to alleviate concerns and enhance product acceptability.

Strengths and limitations
The mixed-method approach and the triangulation of views from providers and clients on important method characteristics with insight gained from actual contraceptive decisions and experiences strengthen confidence in our ability to identify desirable product attributes. Our ability to add a module to the PMA2020 surveys in the two countries allows us to capture the views of nationally representative samples of women. However, the limited number of questions and survey context restrict the breadth and depth of the information that could be collected. Insufficient probing and lack of clarity in qualitative transcripts prevents us from clearly capturing nuances in perspectives on some aspects of bleeding changes such as intensity vs. duration of period or monthly bleeding vs. spotting. The simple ranking exercise offers an additional data point; however, it lacks a measure of difference between items that would allow an exploration of trade-offs. While we were able to gain insight into desirable product characteristics, more research is needed to examine how characteristics can be combined optimally (taking into account both feasibility and acceptability) and how the optimal combination might vary across countries and populations of users.

Conclusion
Soliciting end-user input early in the product development process is critical to optimize product design and the use of resources and maximize the chances of success. This research provides insight on product characteristics that should be prioritized in product development efforts. While many findings were consistent across Burkina Faso and Uganda, a few were more nuanced and different across the two countries. This reinforces the fact that preferences are part of a larger social and service-delivery context that will ultimately affect demand for new products and deserves careful attention as part of introduction strategies.

Data availability
Underlying data Quantitative survey data are available with PMA2020 household/female datasets for Burkina Faso (Round 4/2016b) and Uganda (Round 4/2016) at the PMA2020 website (https://www. pma2020.org) or IPUMS PMA website (https://pma.ipums. org). On both sites, datasets are free to download, but users are required to register and provide a description of the proposed research or analysis. Full qualitative transcripts are not available for ethical reasons because even after removing directly identifiable information such as names and addresses, participant identity may be difficult to fully conceal, and research locations may remain potentially identifiable, presenting a risk of deductive disclosure. However, topic guides, ranking exercise cards and relevant transcript excerpts are available from the authors on reasonable request.

Acknowledgments
The study team would like to thank the PMA2020 global team, including Linnea Zimmerman, Selamawit Desta and Shani Turke, and the PMA2020 country principal investigators, including Georges Guiella and Frederick Makumbi for the opportunity to add a question module to the PMA2020 surveys, and for their constructive input in designing the module. We also wish to acknowledge Claire Brennan, Dominika Fiolna, Saskia Guerrier, Anna Lawton, and Seth Zissette for support to the qualitative analysis process and Mario Chen for review of quantitative data analyses. In terms of methodological innovation, the addition of a module to the PMA2020 to elicit women's In terms of methodological innovation, the addition of a module to the PMA2020 to elicit women's perspectives on contraceptive characteristics that were important to them was a good one. The authors complement the PMA2020 survey data with qualitative information. Having said this, I also wanted the paper to compare and contrast this type of methodology and the findings with those used by market research firms. For example, what sorts of biases were avoided or included? Some market research efforts have shown study participants a range of new products and solicited opinions on each. In contrast, in this paper, study participants were not shown products but asked to comment on attributes. It will be important for the authors to elucidate how the combination of PMA2020 and qualitative data shed new or different light. Relatedly, it appears that the ranking of product characteristics was the first step of the qualitative research component. It seemed to have some elements of conjoint analysis; it will be helpful for the authors to flesh this out further -for example, what were the trade-offs that the research participants were willing to make on specific characteristics?

Saumya Ramarao
Were questions asked of the HIV/STI prevention capability of the contraceptive? This will be important to report given imminent release of the ECHO trial results.
It will be helpful if the authors could address the financing of product development and introduction since the purpose of eliciting user input early is to improve the chances of success and better use of resources. Relatedly, since research participants mention the importance of cost considerations in their choice-making, the authors should comment on the policy context of UHC where advances are being made for the inclusion of FP into insurance packages, and other financing mechanisms. Minor comments follow: Introduction, last paragraph: How do the two papers -the current one and the other -differ? The differences are not clear. Also, please include which six methods are included so that the reader is aware.
Results section, first paragraph: 37% of women in Uganda did not complete the product characteristic module. This is a fairly high proportion of drop outs. It will be useful to see if the women who completed the module were different from the women who didn't to examine self-selection. Table 3: "...as being important in choosing a method were effectiveness, duration of protection against a pregnancy, few side-effects, costs and access." The authors should comment on how these align/or not with reasons for unmet need which include side-effects, infrequent sex, opposition from partners, and being postpartum/breastfeeding.

Results from
The quantitative data indicate that 64.5% of women in Burkina Faso and 39.8% of Ugandan women would choose a method that stopped their periods. However, the qualitative data do not support this. Were there differences in how the questions were asked/broached in the quantitative and qualitative phases? If possible, it will be good for the authors to include the FGD and IDI discussion guides.
Were the questions in the quantitative and qualitative components similar in terms of the questions, sequencing, and how they were raised? It seems that this could explain why some of the data do not triangulate well, especially on amenorrhea.
Perspectives on product characteristics -finding that methods not user-dependent were perceived to be convenient. This will be a function of the knowledge that women have of the current methods that are in the program, and not necessarily of new products.
that are in the program, and not necessarily of new products.
It will be useful to include references from other efforts to include user perspectives -notably the SILCS diaphragm, pericoital contraception, and vaginal rings.

Is the work clearly and accurately presented and does it cite the current literature? Partly
Is the study design appropriate and is the work technically sound? Yes

If applicable, is the statistical analysis and its interpretation appropriate? Partly
Are all the source data underlying the results available to ensure full reproducibility? Yes

Are the conclusions drawn adequately supported by the results? Yes
No competing interests were disclosed.

Competing Interests:
Reviewer Expertise: Reproductive health research, product introduction, quality of care I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
Author Response 10 Sep 2019 , FHI 360, Washington, USA Aurélie Brunie Thank you for taking the time to review and for these comments. Edits were made to the text in response to this feedback and some direct responses are also provided below.
The data presented in this paper were part of a larger study that also solicited opinions on six specific products (comments 1 & 4). The two parts of the study (perspectives on the six products and insight into desired product characteristics) utilize different approaches to acceptability research and produce information that can be relevant at different stages of the product development process. We made some edits to clarify this in the introduction. We also agree that reflecting on the comparative advantages and disadvantages of various methodological approaches to soliciting input for informing product development is valuable; however, it warrants a much longer discussion than what can be incorporated in this paper. We provide some information under limitations. The authors are currently working on compiling our experiences using different approaches across three studies, which we hope will address this comment in more depth.
We used a simple ranking exercise (comment 1). In comparison with proportional ranking or discrete choice experiments, this method is limited in its ability to support analysis of trade-offs 1.
We used a simple ranking exercise (comment 1). In comparison with proportional ranking or discrete choice experiments, this method is limited in its ability to support analysis of trade-offs because it does not include a measure of difference between items. We added a note on this under limitations.
We tried addressing the value of acceptability research in the broader context of product financing in the introduction and added some text in the conclusion to reinforce this point (comment 3). Some text was also added to the discussion regarding UHC. We also commented on how our results relate to reasons for unmet need (comment 6).
Thank you for catching the error in wording regarding the proportion of women who responded to the module (comment 5). These were the proportions of all women who responded to the PMA 2020 female questionnaire. When looking at the proportion of those eligible to answer our module, the proportions who agreed to respond were much higher: 99.8% and 100% respectively.
Qualitative and quantitative data collection solicit information in different ways (comments 7 & 8). We believe that this makes for a combined, stronger understanding of the topics being addressed but may indeed, in some cases, result in conflicting findings. Related to amenorrhea specifically, we conducted a secondary analysis further exploring the two sets of results. We added a reference to this work in the paper.
No questions were asked that were related to dual protection (comment 2).
No competing interests were disclosed. , which permits unrestricted use, distribution, and reproduction in any medium, provided the original Attribution License work is properly cited.

Cynthia Woodsong
Ipas, Chapel Hill, NC, USA I very much appreciate the work described in this paper, and look forward to citing it in the future, once it is indexed. The issues the authors present are persistent and persistently overlooked by those seeking to develop new contraceptive methods, as well as increase uptake and continuation of existing methods. The study design, particularly choice of countries and study populations, is excellent. I could fill a page of my commendations and agreements with many points made in this paper. The design, data, insights and conclusions are a significant contribution to the field. I hope this paper gets into the hands of a wide range of people. Some of the conclusions are quite unique, and offer explanations to some deep-seated challenges with contraceptive uptake and continuation.
I have a few comments and suggestions below.
With regard to my responses to the peer review form: -The information is "clearly and accurately presented," but it does not provide sufficient citations to