Debunking myths about contraceptive safety among women in Kingston, Jamaica: Pilot randomized controlled trial

Abstract

Objectives:

1) to create a short motion graphic video to debias women, using evidence from cognitive psychology, of two common myths about safety of intrauterine devices (IUDs) and implants in Jamaica and 2) to conduct a pilot study to evaluate video effectiveness.

Study Design:

We conducted a series of three focus group discussions (FGDs) among target users to inform the development process of the script, story, character and look of the intervention video. We randomized young, female non-users of long-acting contraception at risk of pregnancy at a public clinic in Kingston in 2018-2019 to watch either the intervention (n=113) or control video (n=112). We used logistical regression to evaluate perceptions of method safety, naturalness, and uptake after three months of follow up.

Results:

Almost all (n=220; 97.8%) participants completed the 3-month interview. More women in the intervention arm perceived IUDs to be safe (59.1%) compared to the control arm (43.6%; p=0.02). Perceived implant safety increased from enrollment to follow up in the intervention and control arms (10.9 and 2.7 percentage-point increases, respectively); however, the difference between arms at follow up was not statistically significant (p=0.57). This appeared to be due to arm imbalances at enrollment. Study arms did not differ at follow up in perceived IUD naturalness (p=0.36) or implant naturalness (p=0.68).

Conclusions:

Findings from a pilot study of a video intervention suggest that using debiasing strategies from cognitive psychology has the potential to address misconceptions about contraceptive safety. A larger trial with adequate power is warranted.

1. Introduction

Nearly half (47%) of pregnancies among reproductive-age women in Jamaica were unintended according to the last national reproductive-health survey conducted in 2008 [1]. Among reproductive-age women in union, 33% were using a modem method of contraception, most often condoms (19%), oral contraception (13%) and injectable contraception (10%). In contrast, use of long-acting reversible contraception (LARC), consisting of the intrauterine device (IUD) and implant was infrequent in Jamaica (0.9 and 0.5, respectively) and did not differ between the health regions of the country. Government data show that the copper intrauterine device (IUD; the only type available in the public sector) and the implant accounted for low, and declining, proportions of the overall method mix from 2008 to 2015 [1,2]. Concern about side effects and health risks is a primary barrier to contraception use in Latin America and the Caribbean [3]. In Jamaica, anecdotal evidence from talking with young women attending for healthcare suggested that two myths about the safety of LARC are prominent: the copper in the copper IUD “poisons” women, and implants harm women by causing “bad blood to build up inside” (Dr. Natalie Medley-Singh, obstetrician-gynecologist, written communication, January 31, 2017).

Computer-based interventions for promoting contraception have demonstrated varying degrees of effectiveness for a range of outcomes [4]; however, none have focused on correcting misinformation about contraception safety using debiasing methods grounded in theory. Attempts to correct misinformation may be ineffective in updating people’s memory and can even backfire. Correcting misinformation is more difficult than intervening with people who simply lack relevant knowledge. When people are ignorant on a topic, they often can still make good decisions using simple heuristics [5] and, importantly, they tend to hold low levels of confidence in their decision-making, remaining more open to intervention [6,7]. In contrast, people who believe in a myth may have strong convictions for maintaining it, and reject scientific evidence presented by others [8].

Cognitive psychology recently has identified effective strategies for debiasing [8]. These strategies for debiasing were shown in a recent randomized controlled trial to be effective in debunking myths about vaccine safety among skeptics [9]. We used these strategies to develop a short (<5 minutes) motion graphic video to debunk the two common myths among young women in Jamaica. We evaluated the effect of exposure to the intervention video on young women’s beliefs about the safety and naturalness of the two LARC methods, and adoption of a LARC method, in a pilot randomized controlled trial in Kingston, Jamaica.

2. Methods

2.1. Video theory and development

The study team included the Jamaican medical director of the study site (TH-K), a Jamaican expert in health promotion and behavior change communication (AB), a professor of design in the U.S. (MP) and two epidemiologists in the U.S. who work on trials related to communication about contraception (MJS and MFG). We developed the intervention video in an iterative process in which we solicited feedback from the target population via a series of three focus group discussions (FGDs) using structured interview guides. Each FGD included eight women (not necessarily the same women) from the target population. We recorded, and then transcribed, the FGDs. We used the findings to inform the video script and story development process to ensure that the video content was culturally-appropriate, eye-catching and engaging, and contributed to the target population’s understanding of the key messages.

The intervention video included an overview of the IUD and implant, their relative effectiveness in preventing pregnancy compared to other methods, and information to debunk the two prevalent myths (i.e., copper poisoning from IUDs and build-up of blood from implants). The content was designed to account for four cognitive factors that make myths “sticky” [8]. First, the continued influence effect holds that in neutral situations (i.e., those in which people lack motivation for believing one piece of information over another) simply providing retractions will fail to eliminate the influence of misinformation [9]. The video used two techniques that have been shown to enhance retraction effectiveness: repetition of the retraction and providing an alternative explanation to fill the coherence gap. Second, the video did not articulate the myths to be debunked to avoid the familiarity backfire effect [9], in which people retain familiarity with the myth but lose the memory “tag” that labels the information as a myth [10,11]. Third, the overkill backfire effect holds that simple myths cognitively are more attractive than more complicated retractions [13,14]. To reduce the possibility of this effect, we employed few refutations, which were conveyed in clear language using culturally-appropriate drawings. For example, we used graphics to explain that Jamaican diets typically include trace amounts of copper. Finally, the worldview backfire effect occurs when evidence that threatens the recipien’s worldview causes the person to strengthen their belief in the myth [9]. To address this effect, we used a self-affirmation exercise on personal values before showing the intervention video [15].

Finally, even superficial characteristics of presenting information have been shown to influence people’s processing of the message and their acceptance of the information as true. Thus, our video used easy-to-read fonts [16] with high instead of low color contrast [17]. Characters in the video were co-designed with our focus group discussants to reflect culturally relevant representations of Jamaican clinic doctors and clients including attention to fashion and hair styles. The video script used rhyming (e.g., “no need to bleed”) or short statements to deliver the key messages and was narrated by a native Jamaican speaker, which can increase the credibility of the message [18]. Specifically, the video was narrated by a health promoter who had hosted a prime time television program for over a decade in Jamaica, which could be expected to contribute to the target population’s trust in the message.

2.2. Study participants

Female, Jamaican interviewers enrolled young women, who were attending a large public clinic in Kingston during November 2018 to March 2019. Women had to meet the following eligibility criteria for the pilot randomized controlled trial: be 18-25 years of age; not using LARC or sterilization; sexually active (defined as ≥1 penile-vaginal act in past month); not known to be pregnant; not breastfeeding; and not desiring pregnancy in the next 12 months. We limited the study to women 18-25 years of age because young women might have different beliefs about LARC, and motivations to use LARC, compared to older women. We did not enroll women under the age of 18 years as the legal status of providing contraception, in general, to minors in Jamaica is unclear [19]. Furthermore, health care providers in Jamaica are unlikely to be willing to insert LARC in young women under 18 years of age because of the interpretation of the Child Care and Protection Act 2004 that providing contraceptive advice and treatment to minors is illegal [20]. To participate, women also had to agree to provide a telephone number for follow-up contact and provide written consent. Ethical review committees at the Jamaica Ministry of Health and OSU approved the study.

2.3. Recruitment, screening and enrollment visit

Interviewers recruited and screened eligible women waiting at the clinic to determine their eligibility. Using a tablet with the web-based version of REDCap [21], interviewers administered the enrollment questionnaire to record participant data on demographics and contraceptive-related knowledge, attitudes and behaviors. We measured women’s baseline assessment of IUD safety with the question, “Do you think the IUD is very safe, mostly safe, mostly unsafe, very unsafe or decline?” We measured women’s baseline assessment of IUD naturalness with the question: “Do you think the IUD is very natural, mostly natural, mostly unnatural, very unnatural or decline?” Parallel baseline questions were asked about implant safety and naturalness.

Interviewers then randomized women (using the REDCap randomization function, which serves to conceal the allocation process) in a balanced design to either watch a short motion graphic video on 1) debiasing about LARC safety (intervention arm) or 2) methods to control mosquitos (control arm). The interviewer asked participants assigned to the intervention arm to complete a self-affirmation exercise; specifically, they asked participants to report the degree to which they agreed with a series of statements on personal values. An example statement read: “I must stand up for what I believe in, even in the face of strong opposition.” The interviewer then asked them to reflect briefly about one of the values that they identified as important. The self-affirmation exercise, which did not involve data collection and did not affect the analysis, was designed to combat the worldview backfire effect [15]. Participants in both arms watched their assigned video on a low-cost tablet in a quiet location. Study staff asked participants whether they wanted to receive routine contraceptive counseling; to mimic real-life conditions, we did not influence the clinic’s standard of care related to contraception counseling or provision.

2.4. Follow-up questionnaire

Interviewers telephoned participants three months post-enrollment to administer a short follow-up questionnaire. The follow-up questionnaire collected data on beliefs about the safety and naturalness of the IUD and implant, receipt of contraceptive counseling, and current method use. The follow-up interview included the same four questions used at baseline to measure IUD safety, IUD naturalness, implant safety and implant naturalness. Based on interviews with key informants from family planning clinics in Kingston conducted for earlier research [22], the three-month interval was adequate for ensuring that participants who decided at the enrollment visit to initiate a LARC method had sufficient time to obtain it.

2.5. Analysis

To evaluate the effectiveness of the intervention video compared to the control, we present p-values from logistical regression models for the differences between study arms in the primary outcomes measured after three months of follow up. Primary outcomes included perceiving the IUD and implant as safe or natural (i.e., four separate outcomes) and adopting the use of either LARC method. We also evaluated whether the groups differed at follow up in terms of other contraception use and having discussed the IUD or implant with a health care provider since study enrollment. We used the intent-to-treat analysis population, which consisted of participants with non-missing data for the primary outcomes at follow up.

3. Results

Of the 320 women who screened for the study, 95 did not meet the eligibility criteria. A few women who were recruited did not agree to be screened; however, the exact number is unknown as staff did not document this infrequent occurrence. Overall, 225 women were eligible and consented to participate in the study and were randomized to the intervention (n=113) or control arm (n=112) (Figure). Five participants were lost to follow up. Thus, the intent-to-treat analysis population consisted of 220 women. Participants at enrollment had a mean age of 21.3 years (standard deviation, 2.4; range, 18-25). Most participants had completed an educational level of high school or less (56.4%-59.6%), were not employed full-time (74.6%-81.8%), and were married, in a common-law union or cohabiting (62.4%-65.7%) (Table 1). At enrollment, 39.1% and 52.7% of participants in the intervention arm perceived the IUD and implant, respectively, as very or mostly safe. Among those in the control arm, 38.2% and 64.6% perceived the IUD and implant, respectively, as very or mostly safe.

Fig. 1.

Disposition of potential participants.

Table 1.

Characteristics and behaviors at enrollment by study arm for intent-to-treat population, among women (N=220) 18-25 years of age attending public clinic in Kingston in 2018-2019

	Intervention arm (n=110)	Control arm (n=110)
	n (%)	n (%)
Age in years
18-19	34 (30.9)	34 (30.9)
20-23	27 (24.6)	44 (40.0)
24-25	49 (44.6)	32 (29.1)
Highest educational grade achieved
High school or less	62 (56.4)	65 (59.6)
Vocational / skills training	32 (29.1)	26 (23.9)
Some college or university	16 (14.6)	18 (16.5)
Fulltime employment
Yes	28 (25.5)	20 (18.2)
No	82 (74.6)	90 (81.8)
Union status
Married, common-law, cohabiting	68 (62.4)	71 (65.7)
Other	41 (37.6)	37 (34.3)
Have live children
Yes	57 (52.3)	43 (39.8)
No	52 (47.7)	65 (60.2)
Current contraceptive method*
Injectable	18 (16.4)	18 (16.4)
Oral contraception	5 (4.6)	5 (4.6)
Male condoms	43 (39.1)	51 (46.4)
Female condoms	0 (0.0)	2 (1.8)
Fertility-based awareness or withdrawal	7 (6.4)	7 (6.4)
None	49 (44.6)	51 (46.4)
Perception of IUD safety
Very or mostly safe	43 (39.1)	42 (38.2)
Very or mostly unsafe	44 (40.0)	38 (34.6)
Decline	23 (20.9)	30 (27.3)
Perception of implant safety
Very or mostly safe	58 (52.7)	71 (64.6)
Very or mostly unsafe	39 (35.5)	32 (29.1)
Decline	13 (11.8)	7 (6.4)

^*Could provide multiple responses

At follow up, 59.1% of women in the intervention arm perceived the IUD to be very or mostly safe. This represented a 20 percentage-point increase from enrollment. Higher proportions of women in the intervention arm (59.1%) perceived the IUD to be very or mostly safe at follow-up relative to the proportion in the control arm (43.6%; p=0.02; Table 2). At follow up, 63.6% of women in the intervention arm perceived the implant to be very or mostly safe, which was a 10.9% percentage-point increase from enrollment. However, this perception did not differ between the two study arms at follow up (p=0.57). Furthermore, the two study arms did not differ at follow up in terms of perceptions of IUD naturalness (p=0.36) or implant naturalness (p=0.68).

Table 2.

Contraception perceptions and use at 3-month follow up by study arm for intent-to-treat population, among women^a 18-25 years of age attending public clinic in Kingston in 2018-2019

	n (%)	p-valueb
Perceives IUD to be very or mostly safe
Intervention arm	65 (59.1)	0.02
Control arm	48 (43.6)
Perceives IUD to be very or mostly natural
Intervention arm	37 (33.9)	0.36
Control arm	31 (28.2)
Perceives implant to be very or mostly safe
Intervention arm	70 (63.6)	0.57
Control arm	74 (67.3)
Perceives implant to be very or mostly natural
Intervention arm	41 (37.3)	0.68
Control arm	44 (40.0)
Discussed IUD or implant with provider since enrollment
Intervention arm	79 (71.8)	0.20
Control arm	70 (63.6)
Any current contraception use
Intervention arm	70 (63.6)	0.54
Control arm	65 (59.6)
Current IUD or implant use
Intervention arm	0 (0.0)	--
Control arm	0 (0.0)
Current injectable contraception use
Intervention arm	27 (24.6)	0.06
Control arm	16 (14.6)

^aRandomly assigned to intervention (n=110) and control arm (n=110)

^bFrom logistical regression model for the difference between study arms

The frequency of reporting having discussed the IUD or implant with a health care provider since enrollment appeared to be higher in the intervention arm (71.8%) relative to the control arm (63.6%); this difference, though, was not statistically significant (p=0.20; Table 2). The two arms did not differ in any current contraception use at follow up (63.6% and 59.6% in the intervention and control arms, respectively; p=0.54), and no one in either arm reported current IUD or implant use. The frequency of reporting current injectable contraception use was higher in the intervention arm (24.6%) compared to the control arm (14.6%), but this difference was not statistically significant (0.06).

4. Discussion

Young women at risk of pregnancy in Kingston, Jamaica who were exposed to a short video on debunking culturally-specific myths about LARC had more positive perceptions about IUD safety at the three-month follow up compared to those exposed to the control video. The two study arms did not differ in perceptions about implant safety at follow up. However, the arms were imbalanced at enrollment: almost 12 percentage-points fewer women in the intervention arm perceived the implants as very or mostly safe compared to control women at enrollment. Thus, an improvement in perception of implant safety in the intervention arm appears to have been masked by this arm imbalance, which likely was due to inadequate study power in the pilot study.

While myths about contraceptive safety appear to be prevalent in many populations [23], the degree to which these misconceptions prevent contraception use – or lead to use of a less effective method – is unknown. For example, women might use contraception despite safety concerns if they preferentially value other factors. A range of other considerations could be more influential for women’s decision-making about contraceptive method use including the ability to be protected against unwanted pregnancy, non-contraceptive benefits of specific methods, or method attributes related to convenience or ability to use discretely.²² Women’s understanding of the relative effectiveness of specific methods also might play an important role in the decision to use a highly-effective method [24–26]. Finally, other structural factors, such as access, cost, provider attitudes [27,28], could prevent method use even if women are successfully debiased of their misconceptions.

We selected the study sites, in part, because LARC are provided without user fees and, thus, financial barriers to LARC use should not be present. Our primary outcomes also included measures of contraceptive method initiation. However, no women in the study in either arm initiated IUD or implant use during the study. During study planning, stocks of IUDs and implants were available at nearby family planning clinics. Study procedures, though, did not include replenishing stocks at the study site or nearby clinics during the study; our intent was to introduce the study video without otherwise modifying the setting to enable us to measure the effects of a simple intervention that would be easy and inexpensive to scale up. Given the low availability of the IUDs and implants in Jamaica [29], the devices may have been unavailable at the study site to participants who wanted to initiate their use. Future research should ensure the availability of adequate supplies of contraceptive devices during data collection. Use of injectable contraception at three months was higher in the intervention arm (24.6% vs. 14.6%); however, this difference was not statistically significant (p=0.06). The intervention video graphically showed the typical effectiveness of the condom, injectable, implant and IUD, which could have motivated women to seek out the most effective contraceptive method from the types available to them.

A primary weakness is that the project was undertaken as a pilot study to develop the intervention video and collect preliminary data on its effectiveness; thus, we did not perform as a power calculation as the sample size was restricted by the limited pilot study funding. Consequently, insufficient power could have prevented us from detecting differences between study arms. Furthermore, a single viewing of the video by female patients might be inadequate intensity of exposure to the study intervention to debias women of their misconceptions about safety or change their behavior related to contraception use. Also, many myths about contraceptive safety are possible; we might have failed to identify the most critical myths to debunk for the study population. Interventions that target health care providers, who could hold misconceptions about contraception safety, also might be needed to increase LARC uptake. Finally, we measured perceptions of method safety and naturalness three months after exposure to the video. Given the timing of the assessment, we might have failed to detect immediate effects on women’s perceptions after video exposure as well as longer term declines in any effects.

Given the high follow up (97.8% of those enrolled), the study findings are unlikely to have been influenced by attrition bias. Another study strength was the use of the low-cost tablet for showing the video and the use of opt-in design for receiving contraceptive counseling that followed standard care at the study clinic. We used this design to ensure that the study intervention could be easily scaled up in non-research settings using only minimal additional resources. Finally, the intervention video employed evidenced-based strategies for debiasing [8,9] and was developed through an iterative and collaborative process with the target population in an attempt to ensure the use of messages that were persuasive both in terms of content and style for combatting the key myths.

In summary, findings from this pilot study suggest that the intervention video, which was designed using debiasing strategies from cognitive psychology, might be effective in overcoming culturally-specific misconceptions about LARC safety among young women at risk of pregnancy. It is critical to safeguard against reproductive coercion when designing and implementing LARC-related interventions. For example, others have warned against LARCs being pushed onto young women as a means of controlling their fertility to reduce the public costs of pregnancy [30]. Achieving true reproductive autonomy also requires that decision-making be free of misinformation about contraceptive safety. Future research could test whether exposing women to a video employing debiasing strategies can lead to a behavioral change related to contraception uptake and continuation. This research also could include qualitative component to improve our understanding of how misconceptions about contraceptives might affect decision-making regarding their use.

Implications:

Evidence from a pilot randomized controlled trial suggested that use of debiasing strategies from cognitive psychology could be effective in correcting women’s misconceptions about contraception safety and thus show promise for the design of future contraceptive promotion videos to increase uptake.