To determine prescription contraceptive use in the UK.
To determine prescription contraceptive use in the UK.
Observational study using a primary care database.
The Health Improvement Network (THIN).
Women in THIN aged 12–49 years in 2008, registered with their primary care doctor for at least 5 years, and with a prescription history of at least 1 year were included.
THIN was searched using the Read and MULTILEX codes for the following methods: combined oral contraceptives (COCs), progestogen-only pills (POPs), copper intrauterine devices (Cu-IUDs), the levonorgestrel-releasing intrauterine system (LNG-IUS), progestogen-only implants, progestogen-only injections, and contraceptive patches.
Prevalence, switching, and duration of prescriptions.
A cohort of 194 054 women was identified. The prevalence of contraceptive use was: COCs, 16.2% (95% confidence interval, 95% CI 16.1–16.3%); POPs, 5.6% (95% CI 5.5–5.6%); Cu-IUD, 4.5% (95% CI 4.4–4.5%); LNG-IUS, 4.2% (95% CI 4.1–4.2%); progestogen-only implants, 1.5% (95% CI 1.5–1.6%); progestogen-only injections, 2.4% (95% CI 2.3–2.4%); and contraceptive patches, 0.1% (95% CI 0.1–0.2%). Within 1 year, 9.8% of new COC users switched to alternative COCs, and 9.0% changed to a different method. Among new COC users who did not switch method, 34.8% did not continue use beyond 3 months, and were no longer using a prescription contraceptive.
Among users of oral contraceptives who did not switch method, over one-third did not continue use beyond 3 months. This supports current UK guidelines recommending a follow-up consultation with a healthcare professional 3 months after the first prescription of COCs.
Women have a considerable choice of effective contraceptive methods, with each method requiring different levels of user involvement. Barrier methods, such as condoms, must be employed at every act of intercourse, and oral contraceptives require daily administration. Long-acting reversible contraceptives (LARCs), which comprise intrauterine devices (IUDs), progestogen-only implants, and progestogen-only injections, require the least level of user involvement. The most effective contraceptive methods do not depend on regular user action; hence, LARCs have the lowest failure rates of reversible contraceptive methods. In a recent prospective study in the USA, the contraceptive failure rate among women using LARCs was 0.27 per 100 participant-years, compared with 4.55 per 100 participant-years among women using other methods.[2, 3] Continuation rates also influence the effectiveness of contraception, because women may use no contraception while deciding which method to use next.
In the UK, contraceptives have been provided free of prescription charges since 1974, and continuation rates and switching patterns are unlikely to be influenced by the costs of the individual contraceptives and the required logistics for refilling prescriptions. This is in contrast to other countries, where studies have shown that access is a major determinant of continuation rates of combined oral contraceptives (COCs). In the USA, up to 61% of women have been found to switch contraceptive methods within 2 years of starting them. IUDs and progestogen-only implants have been found to have the lowest rates of discontinuation at 1 year. In another US study, more than 80% of women who discontinued use of a new oral contraceptive within 6 months of prescription either failed to adopt another method or changed to a less effective one. Factors that predicted the early discontinuation of oral contraceptives included side effects such as nausea and mood changes. Women may also associate the use of oral contraceptives with weight gain or an increased risk of developing cancer. Data regarding switching patterns and discontinuation for the UK are lacking. In countries such as the USA, where contraceptives can be charged to the user, cost may play a much bigger part in uptake and continuation; therefore, data should be extrapolated to the UK with caution.
The purpose of this study is to evaluate contraceptive use and switching patterns among women of childbearing potential in the UK in 2008 using data from The Health Improvement Network (THIN).
The Health Improvement Network (THIN) is a medical research database of anonymised patient records, and includes data from 3.7 million active patients. The information is recorded by primary care doctors (PCDs) in 562 participating general practices across the UK, covering 6.2% of the UK population. The data in THIN is representative of the UK population, and has been validated for pharmacoepidemiological research. Participating PCDs systematically record data in THIN as part of their routine care, including information on: patient demographics, consultations, diagnoses from specialist referrals and hospital admissions, results of laboratory tests, and prescriptions written. Information is anonymised before being sent to the database for use in research projects. Read codes are used to record patient information such as diagnoses, demographic factors, and therapeutic or surgical procedures. Multifunctional Standardised Lexicon for European Community Languages (MULTILEX) codes are used to record drug prescriptions.
All females in THIN of childbearing age (12–49 years), from 1 January to 31 December 2008, inclusive (Figure 1), registered with their PCD for at least 5 years, and who had a computerised prescription history of at least 1 year were identified. The contraceptive methods examined during the study period were: oral contraceptives, comprising COCs and progestogen-only pills (POPs); LARCs, comprising copper IUDs (Cu-IUDs), the levonorgestrel-releasing intrauterine system (LNG-IUS), progestogen-only implants, and progestogen-only injections; and contraceptive patches.
To ascertain the number of women prescribed each contraceptive method, independent automatic searches were performed by means of computer algorithms using the specific Read and MULTILEX codes for each contraceptive method (see Supporting information). The use of the vaginal ring was also investigated; however, no use was reported. The start date (date of first use) was assigned as the earliest date of the relevant Read or MULTILEX code entry for each patient. The treatment pattern of the methods prescribed during the first year was analysed for each contraceptive user.
Each user cohort was stratified according to previous use of prescription contraceptives. New users had no record of using each respective contraceptive method any time before the start date, whereas past users had employed this contraceptive method before the start date. New users of each contraceptive method were further stratified according to previous use of any of the other contraceptive methods described above into first-time users and prior contraceptive users. First-time users had no record of previous use of any contraceptive method before the start date. By contrast, prior contraceptive users had previously employed a contraceptive method other than the new method prescribed (Figure 1).
The 1-year incidence and prevalence was calculated for each contraceptive method in 2008. In the literature, COCs have often been categorised according to the progestogen component as first generation, second generation, and third generation. In this study, users of COC were classified according to these subtypes. COCs that contain the progestogens drospirenone or cyproterone were classified separately. To identify established LARC users, the typical life cycles for Cu-IUDs, the LNG-IUS, and progestogen-only implants were considered. Searches were conducted in the 10 years before the start date (date of insertion) for Cu-IUDs, in the 5 years before the start date for the LNG-IUS, and in the previous 3 years for progestogen-only implants. It was also ensured that there was no record of removal before the start date in 2008 among these established LARC users. Progestogen-only injections have an 8–12-week life cycle; hence, the equivalent analysis was not undertaken for users of this method. It should be noted that in calculating the prevalence of each individual contraceptive method, a woman could contribute to more than one contraceptive method group during 2008.
Several Read codes are non-specific and do not distinguish between Cu–IUDs and the LNG-IUS. In such cases, if there were no accompanying specific MULTILEX code, they were assigned as new Cu-IUD users, provided there was no recorded use of a Cu-IUD in the 10 years before the start date, no recorded use of the LNG-IUS in the 5 years before the start date, or there was no record of LNG-IUS removal in the 5 years before or 90 days after the recording of the non-specific read code. New LNG-IUS users were identified based on the specific MULTILEX or read codes, and were only required to be free of using an LNG-IUS (read or MULTILEX code) within the previous 5 years.
To avoid misclassification of established LARC users, the most recent record indicating the presence of the existing Cu-IUD, LNG-IUS, or progestogen-only implant was identified that was within the time frame specified previously for the life cycle of each LARC method (Cu-IUD, 10 years; LNG-IUS, 5 years; progestogen-only implant, 3 years). From the most recent record up to the start date, women with an existing Cu-IUD, LNG-IUS, or progestogen-only implant meeting any of the following criteria were excluded.
Demographics and switching patterns were investigated for new users of contraceptives; however, because of the small size of the contraceptive patch cohort (n = 434), this group was not studied further.
Data on age, smoking status, and body mass index were collected for any time before the start date, prioritising the most recent entry.
For each contraceptive method, the number of prescriptions for that method and the number of prescriptions for other methods were evaluated among new users in the first year of use. Individuals with two or more concurrent prescriptions at the start date were excluded from this analysis. Users of COCs and POPs who did not switch method were categorised into three mutually exclusive groups according to the duration of prescription during the first year of use: up to 3 months; from 3 months to 6 months, and longer than 6 months. Women whose duration of prescription was <6 months were considered to have discontinued their use of prescription contraceptives. A small proportion of women, approximately 1% of each cohort, switched to more than one contraceptive method and was excluded.
A total of 574 185 women were eligible for study, of which 194 054 had used at least one of the designated contraceptive methods during the study period. COCs were used by the greatest proportion of women (16.2%, 95% confidence interval, 16.1–16.3%), whereas POPs were used by 5.6% of women (95% CI 5.5–5.6%) (Figure 2). Among LARC methods, the use of the intrauterine devices Cu-IUD and LNG-IUS had the highest prevalence (4.5%, 95% CI 4.4–4.5%; 4.2%, 95% CI 4.1–4.2%). The use of the contraceptive patch had the lowest prevalence (0.1%, 95% CI 0.1–0.2%).
The majority of users of each contraceptive method had used the method previously, and ranged from 52.7% of users of the contraceptive patch to 92.1% of Cu-IUD users (Table 1). Among new users of COCs, the vast majority were first-time users (90.2%). In contrast, the majority of new users of all other methods had used contraceptives previously.
|Past users (n = 160 996)||New users (n = 33 058)|
|Total||First-time users||Prior contraceptive users|
|Combined oral contraceptive (n = 92 980)||81 437||87.6||11 543||12.4||10 409||90.2||1134||9.8|
|Progestogen-only pill (n = 31 885)||22 570||70.8||9315||29.2||1677||18.0||7638||82.0|
|Copper intrauterine device (n = 23 180)||21 347||92.1||1833||7.9||453||24.7||1380||75.3|
|Levonorgestrel-releasing intrauterine system (n = 22 934)||18 572||81.0||4362||19.0||788||18.1||3574||81.9|
|Progestogen-only implant (n = 8604)||5660||65.8||2944||34.2||504||17.1||2440||82.9|
|Progestogen-only injection (n = 13 648)||10 976||80.4||2672||19.6||577||21.6||2095||78.4|
|Contraceptive patch (n = 823)||434||52.7||389||47.3||53||13.6||336||86.4|
New users of COCs were typically younger than new users of other forms of contraception (mean age 19.2 years, 95% CI 19.1–19.3 years; Table 2). By contrast, new users of IUDs had the highest mean age (Cu-IUD, 35.2 years, 95% CI 34.8–35.6 years; LNG-IUS, 38.4 years, 95% CI 38.2–38.6 years).
|Total (n = 32 669)a||Combined oral contraceptive (n = 11 543)||Progestogen-only pill (n = 9315)||Copper intrauterine device (n = 1833)||Levonorgestrel-releasing intrauterine system (n = 4362)||Progestogen-only implant (n = 2944)||Progestogen-only injection (n = 2672)|
|Age in years, mean (95% CI)||19.2 (19.1–19.3)||30.7 (30.5–30.9)||35.2 (34.8–35.6)||38.4 (38.2–38.6)||25.2 (24.9–25.5)||25.0 (24.6–25.3)|
|BMI, n (%) a|
|15–19||1230 (9.2)||805 (8.8)||116 (6.8)||227 (5.6)||236 (7.3)||278 (8.9)|
|20–24||2521 (33.3)||3030 (32.0)||660 (33.8)||1553 (30.7)||813 (29.2)||779 (31.3)|
|25–29||1159 (19.0)||1986 (20.7)||472 (23.4)||1170 (23.5)||545 (22.4)||482 (22.0)|
|≥30||626 (11.2)||1987 (21.2)||375 (17.4)||1000 (21.4)||517 (21.4)||414 (19.9)|
|Unknown||6007 (27.3)||1507 (17.5)||210 (18.8)||412 (18.9)||833 (19.9)||719 (18.0)|
|Smoking, n (%) a|
|Non-smoking||7698 (62.2)||4789 (51.9)||930 (46.8)||2238 (46.8)||1315 (43.1)||1267 (46.9)|
|Current||1731 (16.0)||2504 (26.7)||442 (28.0)||972 (29.7)||1015 (33.4)||920 (33.6)|
|Former||801 (17.9)||1879 (19.7)||442 (22.5)||1137 (22.7)||518 (21.6)||408 (17.9)|
|Unknown||1313 (4.0)||143 (1.7)||19 (2.9)||15 (0.9)||96 (1.9)||77 (1.7)|
Overall, 9.8% of COC users switched to an alternative brand, and 9.0% switched to a method other than a COC (Table S3). For each subtype of COC, first-time users were more likely to switch to a different COC than switch to a different contraceptive method. In contrast, prior users of contraceptives were more likely to switch to an alternative method than to a different brand of COC. Of those who did not switch method, approximately one-third (34.8%) did not continue the prescription beyond 3 months, and approximately one-half (50.1%) had discontinued use by 6 months. A similar pattern was observed for women using POPs. Among those who did not switch to another method, 38.2% discontinued use within 3 months, rising to approximately one-half (49.7%) by 6 months. Overall, 26.9% of POP users switched to an alternative method.
Among users of LARCs, first-time users had higher continuation rates (i.e. were more likely to use only that method during the first year), compared with prior users of contraceptives. The women most likely to use only that method during the first year were users of the LNG-IUS (95.7 and 89.2% for first-time users and prior users of contraceptives, respectively; Table 3). Conversely, those using progestogen-only injections were least likely to use only that method during the first year (77.1 and 70.2% for first-time users and prior contraceptive users, respectively).
|Contraceptive methoda||User type, n (%)b||Use of only that method during the first year, n (%)|
|Copper intrauterine device (n = 1783)||First-time 451 (25.3)||410 (90.9)|
|95% CI 23.3–27.4||95% CI: 87.9–93.2|
|Prior 1332 (74.7)||1126 (84.5)|
|95% CI 72.6–76.7||95% CI 82.5–86.4|
|Levonorgestrel-releasing intrauterine system (n = 4170)||First-time 768 (18.4)||735 (95.7)|
|95% CI 17.3–19.6||95% CI 94.0–96.9|
|Prior 3402 (81.6)||3034 (89.2)|
|95% CI 80.4–82.7||95% CI 88.1–90.2|
|Progestogen-only implant (n = 2700)||First-time 478 (17.7)||379 (79.3)|
|95% CI 16.3–19.2||95% CI 75.4–82.7|
|Prior 2222 (82.3)||1619 (72.9)|
|95% CI 80.8–83.7||95% CI 71.0–74.7|
|Progestogen-only injection (n = 2601)||First-time 573 (22.0)||445 (77.7)|
|95% CI 20.5–23.7||95% CI 74.1–80.9|
|Prior 2028 (78.0)||1424 (70.2)|
|95% CI 76.3–79.5||95% CI 68.2–72.2|
The majority of users of oral contraceptives who switched method changed to an alternative oral contraceptive method rather than to an LARC (Table S1). The majority of LNG-IUS users who switched contraceptive method changed to oral contraception (COC and POP: 87.9% for first-time users and 75.4% for prior contraceptive users, respectively) rather than to a Cu-IUD (6.1% for first-time users and 5.1% for prior contraceptive users, respectively). Likewise, the majority of first-time and prior contraceptive users of progestogen-only implants (95.6 and 80.8%, respectively) and progestogen-only injections (76.3 and 78.3%, respectively) who switched method changed to oral contraception (COCs and POPs). The only exception to this pattern of switching was first-time users of Cu-IUDs, 47.4% of whom switched to the LNG-IUS.
Looking at COC subtypes, when switching from second-generation COCs, the largest group of first-time users (41.7%) switched to drospirenone, whereas the largest group of prior contraceptive users (48.6%) switched to first-generation COCs. The largest groups of users of all other COC subtypes switched to second-generation COCs (Table S2).
In our study, oral contraceptives were used by 21.8% of UK women aged 12–49 years in 2008, with approximately three times more women choosing a COC over a POP. Among LARC methods, IUDs had the highest prevalence (4.5 and 4.2% for Cu-IUD and the LNG-IUS, respectively). At the end of 1 year, most LARC users (70–95%) were still maintaining their use of the method. Although COC use had the highest prevalence of all the methods examined, among women who did not switch method in the first year of use more than one-third did not continue the prescription beyond 3 months, and were no longer using a prescription contraceptive. Among women using COCs who switched method, a similar proportion of women switched to an alternative COC (9.8%) as switched to an alternative contraceptive method (9.0%).
The existing information regarding contraceptive use in the UK is based on patient surveys. The most recent survey, conducted by the Office for National Statistics, was based on a sample of 1093 women. THIN is one of the largest national collections of primary care data, and this study comprised a cohort of almost 200 000 users of contraceptives. The patients included in this database are representative of the entire UK population with respect to age, sex, and geographical region. Therefore, THIN is a rich source of information for examining contraceptive use in the UK. Although this current study examines only prescription contraceptive methods, the data in THIN have been recorded by healthcare professionals and are likely to be more reliable than patient surveys.
In our study, women using progestogen-only injections showed the lowest level of adherence among those using LARCs. It is worth noting that norethisterone enantate (Noristerat®) is only licensed for a maximum of two injections, so in this case short-term use does not necessarily represent discontinuation. However, users of norethisterone enantate comprised only 0.1% of all women using progestogen-only injections, and therefore this is unlikely to have distorted the results. A further potential source of error is that some LNG-IUS users may have been misclassified as Cu-IUD users; new IUD users identified by a non-specific Read code were assigned as Cu-IUD users in the absence of an accompanying specific MULTILEX code.
To be eligible for inclusion in our study, women were required to have an extensive medical record history, a computerised prescription history of more than 1 year, and to have been registered with their PCD for at least 5 years. These eligibility criteria may have resulted in geographically mobile women being excluded, and had these criteria not been imposed the sample size would have increased by 10–15%. Also, in this study, contraceptive use was based on prescriptions issued rather than information on dispensing or actual use. Furthermore, some hormonal contraceptives are licensed for indications other than contraception; thus, some minor overestimation is possible. Perhaps more significantly, information regarding contraceptives provided outside general practice may not be captured in THIN, although, in the UK, general practice is cited as the most common provider of contraceptives for women.
The pattern of contraceptive use varies throughout the world. Oral contraception has been found to be the most widely prescribed method in Europe and the USA.[16, 17] In contrast, in China, oral contraceptives were used by fewer than 1% of married women aged 20–49 years, whereas nearly half used Cu-IUDs. According to UK guidelines, LARCs are more cost-effective than COCs; however, LARCs were only used by approximately one-third of the women evaluated in this study. Misconceptions about the risks and benefits of IUDs have been found to contribute to the low prevalence of use. Contrary to the evidence, PCDs assessed pelvic inflammatory disease and ectopic pregnancy to be major complications associated with IUD use, and in a US study, more than one-third of doctors would not recommend the use of Cu-IUDs, primarily because of concern over medical liability. This is in contrast to POPs, which, despite drawbacks such as some requiring a strict 3-hour window of daily administration, were the second most frequently used method in this study.
The intentional short-term use of oral contraceptives is a contributing factor to discontinuation: a prospective cohort study showed that almost one-quarter of women who discontinued after 6 months did so because of a lack of continuing need. Discontinuation may also result from undesirable side effects, which would usually decrease over time. Women using COCs experience varying tolerability profiles, which may mean that the combination or regimen needs to be adjusted to find a COC more suited to the needs of the patient.
An alternative explanation for the high level of discontinuation of oral contraceptive use at 3 months is that some women may not have returned to their PCD to obtain further supplies after completion of the initial prescription. In fact, the World Health Organization's Department of Reproductive Health and Research concluded that restricting the number of cycles of pills can result in the unwanted discontinuation of the method, and recommends the provision of up to 1 year's supply of oral contraceptives at a woman's initial (and return) visit.
In the US-based CHOICE project, women failed to adhere to the correct use of oral contraceptives after a mean of 6.8 months, despite the provision of contraceptives free of charge. In this case, adherence was calculated using the time that participants took to obtain re-fills every month, and users with a lapse of >90 days between re-fills were considered to have discontinued the method and were excluded from the study. The differences in study design mean that caution is needed in making direct comparisons with the present study; however, both studies highlight the difficulty in adhering to long-term medication regimens.
Combined oral contraceptives (COCs) were the most frequently prescribed contraceptive method in the UK in 2008; however, among women who did not switch method in the first year of use, more than one-third did not continue the prescription beyond 3 months, and may have been at risk of unintended pregnancy. UK guidelines recommend a follow-up consultation with a healthcare professional 3 months after the first prescription of COCs.
L.C.S. and L.A.G.R. are employed by CEIFE, which has received a research grant from Bayer Pharma AG for this work, A.M. is an employee of Bayer HealthCare Pharmaceuticals Inc. (USA), and M.A.W. was an employee of AB Bayer (Sweden). L.A.G.R. has been paid as a consultant for Bayer Pharma AG (Germany).
L.C.S., L.A.G.R., A.M., and M.A.W. contributed to the study design and the interpretation of the data. L.C.S. and L.A.G.R. collected and analysed the data. All authors participated in the development and revision of the article.
The THIN data collection scheme is approved by the UK Multicentre Research Ethics Committee (ref. no. 07H1102103). In accordance with this approval, the study protocol was reviewed and approved by an independent scientific review committee (ref. no. 11–041).
L.C.S. and L.A.G.R. are employed by CEIFE, which has received a research grant from Bayer Pharma AG for the submitted work, A.M. is an employee of Bayer HealthCare Pharmaceuticals Inc. (USA), and M.A.W. was an employee of AB Bayer (Sweden). L.A.G.R. has been paid as a consultant for Bayer Pharma AG (Germany).
We thank Dr Charlotte Cookson and Dr William Gattrell of Oxford PharmaGenesis™ Ltd for medical writing support, funded by Bayer Pharma AG. The authors would like to thank CSD Medical Research Limited for providing access to the THIN database.
This retrospective cohort study by Cea-Soriano et al. offers a glimpse into the contraceptive prescribing practices and continuation rates in the UK in 2008. It also highlights potential barriers that prevent women from continuing oral contraceptives (OCs) and from receiving one of the most effective contraceptives, intrauterine contraception (IUC). In this study, 194 054 female contraceptive users registered in The Health Improvement Network (THIN) were followed for 1 year to assess contraceptive use and switching patterns. Twenty-two percent of women used OCs, whereas only 9% used either the copper intrauterine device (Cu-IUD) or levonorgestrel intrauterine system (LNG-IUS). Of note, the mean age of women using OCs was 19.2 years, whereas the mean age of Cu-IUD and LNG-IUS users was 35.2 and 38.4 years, respectively. It is unclear whether this finding that OC users tended to be younger than IUC users is a reflection of provider or patient preference. Other recent studies from the UK (Middleton et al. Eur J Contracept Reprod Health Care 2011;16:311–318; Bharadwaj et al. Eur J Contracept Reprod Health Care 2011;16:201–209) suggest that both young women and their providers harbour misconceptions about the eligibility criteria and safety profile for IUC use among this population.
Another important finding was that among the new users of combined OCs (COCs), 9% switched to another prescription method and 41% discontinued use within 6 months. In fact, over a quarter of new COC users (28%) discontinued use within 3 months of initiation, whereas over 90% of new IUC users were still using IUC at 1 year. As continuation in this study was based upon the duration and number of prescriptions for the method, the high COC discontinuation rate may have been associated with a high rate of prescriptions that were written for only 3 months of use, which is the recommended prescription length for new COC users in the UK (FFPRHC Guidance: First prescription of combined oral contraception, 2006). However, these short-term prescriptions may create a barrier for COC continuation. A systematic review suggested that providing a greater number of pill packs was associated with increased COC continuation and that it was associated with fewer pregnancy tests, fewer pregnancies, and a lower cost per client (Steenland et al. Contraception 2013;87:605–610). Therefore, the World Health Organization recommends that women be provided with up to a 1-year supply of COCs at their initial visit and concluded that restricting the number of cycles of pills can result in unwanted discontinuation of the method and an increased risk of pregnancy (WHO Selected Practice Recommendations for Contraceptive Use, 2004). No evidence exists regarding whether a routine follow-up visit after initiating COCs improves correct or continued use, and a systematic review that examined the incidence of hypertension among women who began using a COC found that few women developed hypertension after initiating it (Steenland et al. Contraception 2013;87:611–624). However, for women who have an elevated risk for developing hypertension, it may be prudent to continue the recommendation that these patients return for a blood pressure check 3 months after COC initiation.
The author has no conflicts of interest to declare.
University of North Carolina, Department of Obstetrics & Gynecology, 3031 Old Clinic Building, CB#7570, Chapel Hill, NC 27599-7570, USA