This study found that contraceptive continuation among hormonal contraceptive users was low among women obtaining contraceptive methods at public facilities in Cape Town. The findings underscore the prevalence of low levels of continuation among users of SAM (oral pills, injectables), as well as the early removal of LARC (implant, IUD) methods, without switching to alternative methods.
Overall, all-method contraceptive continuation was 39.5%, which is generally low compared to other studies in the same setting (85.8%, 66%) as well as to the national estimate of 71% cited in the 2016 SADHS [5, 16, 25]. Despite this, all-method continuation was generally higher compared to method-specific continuation, as it incorporates method switching [14]. The low proportion of all-method continuation in this population could be attributed to high SAM use (95.6%) which is negatively impacted by method non-adherence [43], as well as stockouts of most contraceptive methods at high volume facilities (additional file 4). This is important within the context of how continuation was measured within this study. The study made use of a novel measurement technique suited to the dichotomous and spread nature of the data. SAM were measured by a count of adherence to the method itself through repeat appointment attendance and dispensing of methods. While LARC methods were measured through whether the method was removed, replaced, or changed through-out the period. The low adherence observed was due to method breaks or temporary discontinuation which often occurs among injectable users, and studies conducted elsewhere reported method breaks in up to 43.6% of injectable users [7]. Local research has found that breaks may occur intentionally with users wanting to give their bodies a ‘break’ or because healthcare providers did not want to provide a re-injection due to the individual being late for their scheduled appointment (and potentially pregnant) [7, 10, 44]. In addition, there was a high discontinuation rate of the implant due to early removal, a LARC method which elsewhere has been found to have higher rates of continuation across studies and countries [5, 27, 45–47]. This has been attributed predominantly to method side-effects [27]. In addition, as found in the sub-study, there were a deficit of injectables, implants and IUDs in high volume facilities (see additional file 4).
The main factors associated with continuation in the study were higher age and dual method use. These findings are consistent with other studies in the region [7, 16, 25]. The findings are also consistent with other studies in SA that show that older women are a more motivated contraceptive user group [25].
TB and HIV infections were selected as covariates as these affect the metabolism of certain contraceptive methods and may influence method selection for contraception in an HIV and TB endemic setting such as SA [48]. In the study, being on TB treatment was negatively associated with continuation. This result could be due to national clinical guidelines stipulating cessation of use of combined hormonal contraceptives such as the oral pill and injectables, when using TB drugs [49]. This may influence method choice, the method mix available, provider prescribing patterns and method switching. Interestingly, HIV status, ART use and STI treatment were not significantly associated with continuation and the prevalence of HIV in this population was low at 2%. In contrast, some studies have found that HIV positive women are a more motivated contraceptive user group, with high rates of continuation and acceptability of contraceptives in study settings [50]. This may be in part attributable to FP becoming an important part of HIV standards of care [50].
Contraceptive usage trends observed are comparable to other studies exploring contraceptive use patterns. While injectables and oral pills had the highest usage, it also had the lowest proportion of first method (baseline) continuation (NET-EN-8.2%, oral pills-11.3%, DMPA-IM-17.6%). This is consistent with findings from a 60-country analysis of discontinuation across Africa, Asia, Latin America, and Eastern Europe, which showed that SAM have the highest discontinuation probabilities compared to other contraceptives (40–50%) [40].
We found poorer levels of first-method continuation than those found in other SA studies [7, 11, 14, 22, 51]. These higher rates of discontinuation in comparison to LARC methods can be explained by the ease with which SAM can be discontinued (without the intervention of a healthcare provider) In addition, it requires greater adherence (daily adherence for oral pills) [23, 33]. Yet, within the SA context, high rates of discontinuation may often be due to transient relationships. If a woman is not in a relationship and is not sexually active, she does not have unmet need and may therefore cease use. Furthermore, marriage and cohabitation rates are very low [17], in contrast to other countries in SSA [21]. Women may not see partners for long periods of time and therefore, may discontinue due to side effects, due to sexual inactivity or may use condoms provided by their partner.
The high level of first method (baseline) discontinuation may be further due to method dissatisfaction because of side effects, the most common reason for method switching and discontinuation [27]. Side effects, leading to method switching, are commonly due to disturbances in menstrual patterns [23]. But this switching generally occurs in motivated contraceptive users wanting to avoid a pregnancy and are influenced by factors such as the method mix available [23]. Although the method mix was not measured directly in the study, a limited method mix, the preponderance of injectable usage and drug stockouts at high volume facilities may explain the lower proportions of switching in the study. Stockouts of some contraceptives may deter women from switching contraceptive methods as no other preferable method is available.
In addition to the high usage of SAM and low continuation of methods, there was also low switching of methods in the population. The study found that < 10% of women switched methods overall (7.9%), which is similar to rates found in a 1995 SA study (11.9%) [25] with the national estimate from the SADHS 2016 being much lower (4%) [5]. Conversely, a 2019 HIV-1 prevention study conducted in Cape Town, found higher levels of switching, particularly for SAM users, with 56% switching to a LARC method (60% DMPA, 58% NET-EN, 43% of OCP users), as women were offered multiple methods at each monthly study visit [16]. The low level of switching combined with high discontinuation found in our study could be due to dissatisfaction with SAM as there was high injectable usage in the population, with 68% of women using injectable methods at least once, combined with a limited method mix at several facilities and pharmacy stockouts. Baseline SAM users had particularly low levels of switching (DMPA-IM: 7.6%; NET-EN: 15.3%; and OCP: 11.9%). Switching to a more suitable contraceptive tailored to the user’s needs should be encouraged for these method users if they are dissatisfied with their current method, are experiencing side effects or have unmet need.
Conversely, method switching among baseline implant users was high (36%). Issues have been noted with the implant at the health system, provider, and user level leading to early removals and a decline in its use [52–55]. A local study found that among discontinuers, up to 90% discontinued due to side effects, such as bleeding pattern changes and headaches [54]. After initial high levels of uptake following its introduction in SA in 2014, there has been an almost 50% decline in insertions year upon year [52]. This is mirrored in our study which found that less than half of users (40%) continued this method, leading to low baseline method continuation. Although switching for baseline implant users were higher than other methods, 63% of baseline implant discontinuers ceased contraceptive use altogether, indicating issues with continuation at the user level. This high level of discontinuation normally does not occur with LARC methods which generally have elevated continuation rates at ≥ 84% in multiple regions, globally [45, 54, 56]. This indicates that the implant may be a method with high user dissatisfaction in this population, most likely linked to side effects, inadequate pre-insertion counselling or is being discouraged by providers. A local qualitative study conducted in Cape Town revealed that providers typically favoured removing implants in HIV-infected women prescribed the widely used ART, efavirenz, instead, promoting injectables, due to contraindications [57]. In the same study, providers also expressed contraceptive preferences which discouraged the use of implants for younger women [57]. In contrast, the IUD had the lowest baseline usage of < 0.5% but had the highest first method continuation at 87%. The method also requires a trained health-care professional to insert and remove it [33], but, compared to the implant, displayed higher continuation and less switching. This may indicate that it is initiated and used by a more motivated group of contraceptive users. It may also point to a lack of trained providers widely available for this service [57].
Method switching is also considered a proxy measure of increasing quality of care, with increased switching being an indication of better service quality [30]. Therefore, switching needs to be encouraged within this population when methods are no longer suitable or preferred by women. In addition, comprehensive patient centred counselling is important and has been linked to both increased continuation [58] and higher levels of switching, globally [28]. This counselling should be underpinned by women’s rights, providing the salient informational messaging (informing women about a methods side effects, how they can manage side effects and what other methods are available for use) [5].
Another factor which contributes to discontinuation of baseline methods is poor method adherence. This is particularly pronounced for injectable users amongst whom poor method adherence and high discontinuation rates have been noted [11, 22, 27]. The study shows high levels of injectable use coupled with low contraceptive continuation of injectables, leaving women vulnerable to unintended pregnancy, if they have unmet need. This is concerning and needs attention especially as large proportions of young women use injectables. A 2001 SA study found a continuation rate of 42% at one year for DMPA-IM and 41% for NET-EN users. By two years, this dropped to 21% [22]. First method (baseline) continuation was even lower in our study. NET-EN displayed the lowest first-method continuation at 8.24%. Nearly 60% of all baseline NET-EN users are young women aged 15–24. This key sub-group are most at risk of an unintended pregnancy and simultaneously at higher risk of HIV-1 acquisition [40].
The evidence suggests that if contraceptive counselling is not engaged in a shared decision- making, patient-centred framework, a woman is less likely to understand why she is using the method she was given and therefore less likely to use the method correctly [59, 60]. High rates of discontinuation may indicate that stockouts were common in 2017 and that women did not want to switch to another available method.
Limitations
Participants ‘entered’ and ‘exited’ the study multiple times, potentially taking method breaks, making accounting for time in the study difficult. Due to this, discontinuation proportions in the study may be inflated (Table 3). Several participants may have entered the study period toward the end of the year and therefore only reflected as having one or two method repeats (and therefore seen as discontinuers). It is unknown whether they followed through with this method in the following year, creating a form of information bias as they are assumed discontinuers. SAM continuation was measured by a count of method repeats for each visit and therefore, is less comparable to other continuation studies in SA (which measure continuation rates or probabilities through self-report or survey data). This is part of the method measurement, as this was routine data and therefore did not have continuous participant data monthly to facilitate a ‘contraceptive calendar method’. We only calculated continuation for one calendar year and consequently cannot prognosticate moving forward. It is also unknown whether the baseline method used in the study were users preferred method (which could account for their discontinuation). Another potential limitation is that women may have decided to obtain their method through the private sector or the provincial health service in cases when there were stockouts in city facilities, which our study was also unable to measure. Furthermore, within the study, we do not know the reasons for discontinuation, some of which could be due to cessation of sexual activity, meaning they discontinued but do not have unmet need. It is assumed that once contraception is initiated, there is an ongoing need for contraception. Without data which could be supplied by another study design, this is impossible to ascertain. Therefore, the unmet need is probably overstated in the study. Nonetheless, the results of this study have practical applications for training and service delivery across city facilities, identifying opportunities for improvement and the novel study measures provide an opportunity to use routine service data in a meaningful way at low cost.