Acceptability, side-effects and discontinuation of second and third-generation oral contraceptive pills in Bangladesh: A quasi-experimental study

Background The third-generation oral contraceptive pill (3G-OCP) has improved side-effect prole compared to the second-generation (2G-OCP). In Bangladesh, comparative data on these two generations of OCPs is nonexistent. This study aimed to compare acceptability, side-effects and discontinuation rates between 3G- and 2G-OCPs. Methods This quasi-experimental study was conducted from October 2017 to June 2018 in two unions of one sub-district in Bangladesh. From each intervention and control unions, 1400 women were enrolled after screening for selection criteria. All the women in the intervention union received 3G-OCP while those in the control union had 2G-OCP. Women from both the groups were provided six cycles of OCPs, two at enrollment, and two each at two subsequent home-visits at 2-months apart by the health workers. Data was collected thrice: immediately after enrolment, and two subsequent follow-up visits at 2-months interval by the study interviewers. Life table analysis was done to compare cumulative discontinuation rates and Hazard Ratio (HR) was estimated for likelihood of discontinuation of 3G-OCP as compared to 2G-OCP due to side-effects. Results After 24-weeks of enrollment, 69.1% (n=967) of the women from the intervention and 58.0% (n=812) from the control group continued with the method. The major reason for discontinuation was pill use-related side-effects (3G-OCP: 20.4% vs. 2G-OCP: 19.5%). Initially, the reported side-effects for 3G-OCP were higher than those for 2G-OCP (47.3% vs. 33.2%). However, after 24-weeks of use, the corresponding gures became similar (9.5% vs. 8.7%). The cumulative discontinuation rate due to side-effects was 22.8% and 25.2% for 3G- and 2G-OCP respectively which was not statistically signicant (p=0.14). After adjusting for potential covariates, the likelihood of discontinuation of 3G-OCP

OCP though the difference was not statistically signi cant. After adjusting for other potential sociodemographic characteristics, the likelihood of discontinuation of 3G-OCP was 14%, lower than 2G-OCP which was statistically signi cant at 10% level. The study observed three deaths, one in intervention and two in control groups of which one death occurring due to stroke in the control group was possibly related to the use of 2G-OCP. In conclusion, the study found relatively improved acceptance of the 3G-OCP than the 2G-OCP. Though the reported side-effects of the two types of OCPs was similar after 24-weeks of use, one mortality due to stroke, related to use of 2G-OCP, demands further studies in large populations.

Background
The rst oral contraceptive pill (OCP) introduced in the 1960s marked a step into great progress for women's health and autonomy worldwide. The pill allowed women to be able to decide on the spacing of and the number of children they wished to have using a noninvasive method. Although this contraceptive method initiated several controversies at rst, it was later strongly adopted and became the principal contraceptive method in the United States and other developed nations, used by approximately 8 to 10 million women in the US by 1970 [1].
While the OCP was very effective at controlling birth, its users reported several side effects. Research associated the high estrogen levels in the pill with increased risk of blood clots, strokes and myocardial infarction [2]. By 1962, about 132 reports of thromboembolism had been reported in OCP users and had resulted in 11 deaths [3]. In an effort to reduce these risks, second-generation OCP (2G-OCP) was developed. These pills contained a reduced dose of hormones -estrogens and progestin compared to the rst-generation pills, and were meant to be safer for the women, cause fewer side effects and be overall more e cient than the former pill. Unfortunately, though successful at protecting against certain risky cancers [4], the 2G-OCP was still associated with risk of hypertension and presented daunting side effects such as dizziness, nausea, migraines and weakness. This discouraged women from staying on the pill, making its continuation rates low [5].
The third-generation OCP (3G-OCP) was hence developed to address the continuous low adherence of 2G-OCP. The 3G-OCP is an association of low-dose ethinylestradiol and potent testosterone-derived progestin, low in androgenic activity, making it less harmful for metabolism, weight gain, acne and other notable side effects [6] including a reduced risk of myocardial infarction [7]. When compared to 2G-OCP, 3G-OCP is associated with 29-45% lower rates of discontinuation due to weight gain [8]. The 3G-OCP also has shown potential improvements for a reduced risk of myocardial infarction and are more tolerated by the users. Findings from a transnational study conducted in 1997 showed that 3G-OCP users had approximately one-third of the risk of myocardial infarction when compared to 2G-OCP, the rst pill to be associated with no risk of myocardial infarction [7].
These aforementioned, evidence-based characteristics of the 3G-OCP motivated the idea of its introduction in the National Family Planning Programme (NFPP) in Bangladesh. About 62% of the country's currently married women are using a contraceptive method and 25% of the women use OCP as a primary method of contraception [9]. Oral contraception though widely accepted, also has a high stoppage rate which therefore affects its success in the country. In 2017 while the overall discontinuation rate of contraception was 37%, the OCP discontinuation rate was even higher (42%) [9].
As of 2014, among the women in Bangladesh who were using OCPs, 50% of them were using 2G-OCP, making it the most commonly used modern method of contraception in the country [10]. Yet, a study conducted in rural Bangladesh illustrated that approximately 50% of the women who were on the 2G-OCP, experienced side effects such as dizziness, nausea, blurring of vision and weakness [5]. The purpose of this study was to obtain evidence-based data on acceptability, side-effects and discontinuation rates of 3G-OCP adherence, as they compare to those of the 2G-OCP, as well as to identify reasons for discontinuation and compliance.

Study design and setting
This quasi-experimental study was conducted in two unions (one intervention and one control) of the Sarankhola sub-district of the Bagerhat district of Bangladesh where OCP discontinuation rate was high (30% during January-December, 2016) [11]. The study was conducted from October 2017 to June 2018.

Study population
All the married women of reproductive age (MWRA) in the two selected unions of Sarankhola sub-district (4096 in the intervention union and 3182 in the control union) represented the study population.
Speci cation of the 3G-and 2G-OCP For this study, 3G-OCP was procured from a renowned pharmaceutical company, named, Renata Limited, Bangladesh, which was available in the market with the brand name 'Desolon'. Each Desolon pill contains Desogestrel BP (0.15mg) and Ethinylestradiol BP (0.03mg). Each cycle of 'Desolon' comes in a strip with only 21 pills and does not contain iron tablets. To make the strip similar to the 2G-OCP, the brand name of which is 'Sukhi' and is available in the NFPP, customization of the strips of 'Desolon' was done by Renata Limited by adding seven additional iron tablets (each containing Ferrus Fumerates BP 75.0 mg) for the current study. For supplying six months' cycle among the 1400 clients of OCP, a total of 8,400 cycles of 3G-OCPs were procured from Renata Limited for this study purpose.
Sample selection and enrolment of women in the study For enrollment in the study, a total of 2001 women from the intervention union and 2020 women from the control union were screened for selection criteria. The inclusion criteria were i) married women within 15 and 39 years of age; ii) not pregnant at the time of enrollment; iii) no desire for having a baby in next 1 year, and iv) willing to use OCP as a contraceptive method for the next 6 months. Exclusion criteria were: i) lactating mothers with a ≤6-months-old baby; ii) known case of hypertension; iii) BMI ≥ 30; iv) migraine; v) blurring of vision; vi) varicose vein, and vii) known case of diabetes mellitus. The selection criteria were adapted in the local context using the WHO guideline [12]. Family Welfare Assistants (FWAs), a cadre of frontline government health workers, who were trained for 5 days to work in this study, screened and enrolled the subjects. From each intervention and control unions, 1400 women who passed the selection criteria and gave consent to participate in the study were enrolled. The study protocol was approved by the Institutional Review Board of International Centre for Diarrhoeal Disease Research, Bangladesh (icddr,b).

Distribution of OCPs
After enrollment, the FWAs provided two cycles of 3G-OCP to each woman in the intervention unions, and those in the control unions were provided with two cycles of 2G-OCP. The FWAs also oriented the women in both the groups about the process of using the OCPs and their tentative side-effects. In case of any major side-effect, the women were advised to communicate with the concerned FWAs or visit the satellite clinics or the nearest family planning clinic or health complex for necessary advice.
All the participants were visited twice by the FWAs, rstly at the 2 nd month, and secondly at the 4 th month as part of their regular home visit schedules. During each visit, in the intervention and control unions, the FWAs provided the women with 2 additional cycles of OCPs of the respective types after their health checking and necessary counselling.

Data-collection
Data was collected by 4 female Field Research Assistants (FRAs) who were provided with a 15-day training on the data collection tools, including interview techniques. A full-time eld manager was engaged in the study area to monitor the entire eld implementation process.
Within 2 days of enrolment, an FRA visited the participants at their homes and interviewed them for sociodemographic and reproductive characteristics including their previous contraceptive use history. Each participant was further followed-up at the 3 rd and 6 th months after enrollment for detailed information on the OCP use, side-effects and compliance. Women, who discontinued the method use, were asked about the time of, and reasons for discontinuation. A semi-structured pre-tested questionnaire was used for these follow-up interviews. All the completed questionnaires were regularly reviewed at the eld for completeness and accuracy of the intervention under the supervision of the eld manager. For internal constancy, necessary corrections were made after checking entered data with the questionnaire.

Statistical analysis
Descriptive as well as analytical statistics were performed to report ndings on compliance, side-effects and discontinuation by type of OCPs. To summarize various reported side-effects, ndings are presented under four broad categories: i) direct pill use related problems; ii) probable pill use related problems; iii) possible pill use-related problems, and iv) other health problems in relation to the likelihood of sideeffects of OCP use. Life table technique was used to compute rates of method discontinuation due to side-effects by taking into account the varying length of time for which the subjects remained in the study with the Log-Rank test (Mantel-Cox) to compare the intervention to control. The Cox's Proportional Hazard model was used to estimate the Hazard Ratio (HR) for assessing the likelihood of discontinuation of 3G-OCP as compared to 2G-OCP after adjusting for selected socio-demographic variables. All the statistical analyses were performed using Statistical Package for the Social Sciences (SPSS) for Windows version 20.0 (Chicago, IL, USA).

Results
The similarities and differences in socio-demographic characteristics of the OCP acceptors in the two study groups are shown in Table 1. Mean age of each of the 3G-OCP and 2G-OCP acceptors was 29 years. When comparing the educational status, about 60% of the OCP acceptors in both the groups had at least incomplete secondary or higher education. Only 4% of the acceptors in the intervention group and 6% in the control group had no education. About 47% of the husbands from each group of OCP acceptors had incomplete secondary or higher education. About 10% of the husbands of 3G-OCP users as compared to 14% of those accepting 2G-OCP had no education. For the employment status of the OCP acceptors, only about 5% in each intervention and control group were involved with any incomegenerating activities. The OCP acceptors in the control group had relatively higher monthly family expenditure as compared to that in the intervention area (Taka 11,644 vs. Taka 11,575 equivalent to USD The incidence of different reported side-effects between the 3G-OCP and 2G-OCP acceptors at different time points during the 6-month follow-up period of the study is shown in Table 2. Initially, the 3G-OCP acceptors reported a higher incidence of side-effects (47.3%) as compared to the 2G-OCP acceptors (33.2%). However, at the end of 6 months of use, the incidence of reported side-effects of the two types of OCPs in the study convened to about 9%.
Direct pill use related problems The analysis noted a signi cantly higher incidence of any direct pill use related problem among the 3G-OCP users as compared to 2G-OCP users in the rst four months of follow-up. However, in the 5 th and 6 th months this variation deteriorated. Nevertheless, in the 6th-month follow-up, the 2G-OCP users reported a higher incidence of vertigo and nausea as compared to the corresponding gures by 3G-OCP users. In the 6th-month, any direct pill use related problem by the 3G-OCP users was even lower (4.7%) than the reported gure by the 2G-OCP users (6.0%), though the difference was not statistically signi cant ( Table  2). Probable pill use related problems Comparison of reported probable pill use related problems showed an overall decline in reported sideeffects in both groups of OCP users (reduced from 11.1-3.9% in 3G-OCP and 7.4-1.5% in 2G-OCP users). However, overall, there was a persistent higher incidence of related side-effects among the 3G-OCP users than the 2G-OCP users. When analyzed by types of side-effects, white discharge, excessive bleeding and blurred vision were signi cantly higher in the 3G-OCP users than those using 2G-OCP (Table 2).
Possible pill use related problems Our analysis demonstrates a higher incidence of any possible pill use related problem in 3G-OCP users than those using the 2G-OCP in each of the rst four-month time points consecutively. However, at the 5th and 6th month time point this variation was eliminated ( Table 2).

Other health problems
Overall reporting of other health problems was low (ranging from 1.2-1.7%) in the rst follow-up that gradually reduced to 0% at 6 months of use (Table 2). No variation in other reported health problems between 3G-OCP and 2G-OCP users was observed over time.

Status of continuation with 3G-OCP and 2G-OCPs
Among the enrolled women in both intervention and control groups, 967 (69.1%) and 812 (58.0%) continued with taking the OCPs respectively at the end of 6-month follow-up (Fig. 1). The crude discontinuation rates due to side-effects were 20.4% (n = 285) and 19.5% (n = 273) for 3G-OCP and 2G-OCPs respectively. A substantial proportion of subjects in each intervention (9.0%, n = 126) and control (18.1%, n = 254) group, discontinued the method due to various social reasons (husband's disapproval, husband away, wanted child were the top three reasons for method discontinuation) ( Table 3). During the 6-month follow-up period, 4 women in the intervention group and 7 in the control group reported getting pregnant. When these women were asked about the reasons for becoming pregnant, the majority (75.0% in the intervention and 57.1% in the control group) said about stopping the pill due to side-effects. About one-fourth of the women in each intervention and control group mentioned forgetting taking the pill as a reason for becoming pregnant (Table 3). 0.5 Death 0.0 Lost to follow-up 0.9  (Fig. 2). The same analysis also showed that the major share of these discontinuations occurred within the rst 4 weeks of OCP adherence (14.6% for 3G-OCP and 19.6% for 2G-OCP).
Reasons for discontinuation of OCP due to side-effects Table 4 demonstrates the proportion of women who discontinued OCP due to various side-effects. Most of the side-effect related discontinuations occurred within the rst follow-up visit for both the groups of women. At the end of the rst follow-up visit, among the women who discontinued OCP, over 90% in both intervention and control groups discontinued the method use due to direct pill use related problem in which vertigo, nausea, excessive vomiting and migraine were the major reasons for discontinuation. At the end of the rst follow-up visit, about a quarter of the participants (25.0% in the intervention and 27.7% in the control group), reported discontinuing the OCP due to a probable pill use related problem in which blurred vision, excessive bleeding during menstruation and white discharge were major reported reasons. At the end of the rst follow-up visit, 13.1% and 14.5% of the discontinuations for the 3G-OCP and 2G-OCP respectively were due to any possible pill use related problem in which lower abdominal pain, acidity/gastric, weakness, and frequent urination were the main problems. Other health problems such as loss of appetite, breathing di culties, back pain etc. were reported by around 4% of the women from in each intervention and control group as reasons for discontinuing the method use.  The study also observed 3 deaths, 1 in the intervention area and 2 in the control group. The death in the intervention group was due to diarrhoea and therefore not likely to be related to the use of 3G-OCP under this study. Of the two death cases in the control group, one was due to a stroke that was likely to be related to the use of 2G-OCP. The other death case in the control group had pre-existing kidney and heart disease and a history of taking 2G-OCP. Despite being enrolled in the study, the latter woman did not use the 2G-OCP in the current study thus this death was not related to the study intervention.  Table 5 demonstrates that after adjusting for socio-demographic covariates, the 3G-OCP users were 14% less likely to discontinue the method as compared to those using the 2G-OCP (HR = 0.86, p = 0.075) which was statistically signi cant at 10% level of error. We also observed that women with increasing age, low education level, previous use of OCP, and low family income were more likely to discontinue OCP (Table 5). Hazard model of this study revealed that after 6 months of the OCP adherence, the 3G-OCP was 14% less likely to be discontinued when compared to the 2G-OCP in Bangladesh (HR = 0.86, p = 0.075). This gure also supports the above premise.
Although overall results are advantageous, the measured discontinuation rates of 25.2% and 22.8% for 2G-OCP and 3G-OCP respectively are still relatively high. In addition to displaying relatively high rates of discontinuation, the study also expressed similarities with other previous studies [13]. Side effects were one of the main reasons for OCP discontinuation causing approximately 20% of discontinuation amongst participants in both groups. The second reason for discontinuation was social reasons. In a setting like Bangladesh, social factors can frequently play direct roles in health outcomes as the community is very homogenous and traditional; therefore, immediate change is not easy. Social factors contributed to 9% of discontinuation among the intervention group and 18% in the control group by the end of the study.
The major reason for OCP discontinuation in this study was various side effects, causing 20% of discontinuation in the intervention group and 19% in the control group. These results re ect similar ndings from previous studies conducted in 2006 in Bangladesh, that listed side effects as one of the three major reasons for method discontinuation [5]. Unlike some past studies, this study asked participants what speci c side effects they experienced while on the pills, to identify the side effects that were most connected to discontinuation. The relevant side effects in the research were categorized into 4 different groups: direct pill related problems, probable pill use-related problems, possible pill use related problem and other health problems. The most reported side effects leading to discontinuation were direct pill use-related problems. These problems included, most frequently identi ed vertigo, nausea, blurred vision, excessive vomiting, migraine and others. In 2G-OCP and 3G-OCP groups, vertigo and nausea caused more than 30% of direct pill related discontinuation. In 2G-OCP acceptors, vertigo and nausea caused about 35% of discontinuation, which was 33% in 3G-OCP acceptors.
Side effects being a primary reason that led to OCP discontinuation motivated the review of the variances between reported side effects amongst the intervention and control groups. Regarding the timing of reported side effects, as shown in the results, within the rst four months of follow-up, more participants from the intervention group ended OCP adherence. However, a signi cant shift happened in the fth month. The number of women, from the intervention group, who were discontinuing the pill, started reducing more rapidly than of the women in the control group. By the end of the study, at 6 months, more OCP acceptors from the control group were discontinuing the pill. The fact that approximately 97% of the women in this study had been on a contraceptive pill prior to this study may serve as a potential explanation to the dynamic timing of the discontinuation due to side effects trend. The Rosenberg et al.
1998 study articulated in their study that, recent pill use was the only signi cant variable in predicting discontinuation, meaning that prior exposure to an OCP made a difference on if the women stayed on the pill or not [14]. In 2017, the NFPP released their 2018-2020 prioritized actions and listed as commitment 7 to provide free and adequate contraceptives to NGOs, private clinics and hospitals and garment factory clinics with trained family planning personal [15]. This effort was meant to increase the accessibility of contraceptives to women all over the country; hence more women were taking the 2G-OCP. The 3G-OCP, however, was yet to be introduced in the NFPP in Bangladesh, therefore, most of these women were probably using the 2G-OCP as their contraception method prior to participating in this study. This means that most women in the control group had already been exposed to the 2G-OCP and had a chance to adapt to its potential side effects the rst time they took the method. Most of the OCP acceptors from the intervention group, however, were getting exposed to the 3G-OCP for the rst time and therefore experiencing these side effects for the rst time leading to more initial discontinuation.
Inversely, while former studies report clear favourable trends between higher educational status and OCP discontinuation rates, this study's results challenge those previous ndings. Women in the intervention group who had no educational background or incomplete primary education, in this study, reported lower discontinuation rates than women with higher education. Also, women in the intervention group, from families with high monthly expenditure (20,000 + BDT, equivalent to USD 235+) also showed signi cantly higher discontinuation rates in comparison to women from other expenditure brackets in both groups. Both of these ndings are different from the results shown in previous studies [16,17]. Muhindo et al., in 2015, reported that lower education level was an indicator for poor contraceptive adherence in Uganda [16] and Mahumud et al., in 2015 reported that signi cantly higher rates of discontinuation were pronounced among married Bangladeshi women who were less educated [17]. These new results could consequently be due to the presence of the health workers who provided thorough and personal counselling to each participant during the study implementation period by making all the women, especially the less privileged ones, well aware that some side effects would be apparent as a result of being on the pill. A 1998 study on oral contraception continuation by Rosenberg et al., expressed that pill discontinuation is especially likely if the side effects come as a shocker to the pill acceptors [14]. Hence, understanding the origin and possibility of side effects onset prior to taking the pill could have been one of the reasons that kept the less educated OCP acceptors on the pill even after the inception of side effects. A 2006 study on OCP discontinuation in rural Bangladesh found that women who had consulted a trained worker prior to getting on the pill had a minimum use period of twelve months while women who did not have proper counselling from a trained worker had a minimum use time of one month [5]. A more recent study aimed at identifying the practices that led to unintended pregnancies in Bangladesh, recommended after its ndings, to increase trained eld workers who can ensure distribution of OCP supplies and provide counselling and support for the families, through the family planning program [18].
In addition, the discrepancy between women's educational background and discontinuation rates could be explained by the notion of self-authorization and exposure to more contraceptive options. Women with higher education levels are likely to be employed which makes them more nancially capable, enabling them to seek different contraceptive methods if side effects become too discomforting. It could also be inferred that these women are more empowered to make independent decisions on their bodies and OCP continuation, leading them to stop if they are displeased with the side effects.
From the results, we also observed three death cases during the implementation of the OCP intervention.
One of the deaths was of a participant in the intervention group and the 2 others from participants in the control group. The death case in the intervention group had previously been on the 2G-OCP for over 15 months and even after starting the 3G-OCP, did not report any major health changes. This individual suffered and died from diarrhoea therefore her death was not associated with the 3G-OCP. One of the 2 deaths in the control group had no previous use of OCP. She, however, was hypertensive, was not taking her antihypertensive drug regularly and subsequently died of a stroke. Although the 2G-OCP can protect against some cancers [4], previous research has also linked it with increased risks of hypertension, migraines and other side effects [5]. Subsequently, this death could be identi ed as likely to be related to 2G-OCP use. The other death in the control group was of a woman who had been on the NFPP 2G-OCP for 3 years prior to the start of the study. Immediately after her enrollment in this study, this woman was diagnosed as having kidney and heart complication and spent 2 months in the hospital, during which she was not taking the 2G-OCP. Although she made it out of the hospital, her complications caused her death which therefore cannot be connected to the 2G-OCP use in this study. Yet, it is important to mention that her prior 3 years on the 2G-OCP might have played a role in the status of her overall health.

Limitations
A major limitation of this study was the short duration of study implementation period (6 months only). This prevented observation of long-term effects of each type of OCP used in this study. Because the rst adherence months of OCP have been proven to be a more sensitive period for pill initiators than pill switchers [14], a longer follow up time was necessary to allow the 3G-OCP acceptors to get comfortable with the pill.
Also, the short timing of study implementation did not allow for a 'wash out-period' for those study participants who had previously been using other hormonal contraceptive methods, prior to being enrolled in this study. The fact that some participants might have had a different hormonal balance than their natural balance while other were at natural balance, might have affected the ndings especially when the study had measured a variable closely related to hormonal con guration. Alongside, only two unions were used for patient recruitment. Both the unions were in the same sub-district and under the same district. Participants for the 3G-OCP were recruited from one union and participants for the 2G-OCP were from the other union, resulting in small and similar sample groups. Nonetheless, during the analysis, discontinuation rates of OCP types were adjusted for previous contraceptive use history and other sociodemographic characteristics of the participants.

Conclusion
Like previous research suggested, this study also found that the 3G-OCP in compared to the 2G-OCP would be a better method of contraception to administer to women of Bangladesh. Although a study with a larger sample size and longer implementation period might show the results more clearly, this study serves some evidence-based ndings in recommending the addition of 3G-OCP in the National Family Planning Programme (NFPP) in Bangladesh.
Efforts should be ensured through quality counselling so that women stay on the pill for the rst four to six months, that they will be less likely to discontinue the method. It will also be essential for the NFPP to have targeted strategies for the different strata of women that it caters to. Although Bangladesh can be described as a country that holds culture and tradition very highly, it is also important to note that advancements in health, education and several other sectors are encouraging people and more signi cantly women to nd ways to protect and aid their personal health, sometimes despite cultural norms. For example, targeting those women who have a history of using OCP and who have a low socioeconomic and or educational background may help improve the continuation rate of the 3G-OCP, as the study showed that this population is less likely to discontinue this pill. Nonetheless, the study observed three death cases, of which at least one was likely to be associated with the use of 2G-OCP, no inference could be drawn on the association of the use of OCP type with mortality under this study, for which further studies are recommended. Enrollment and follow-up status of the subjects in the study