Acne is one of the most common diseases worldwide, affecting 85% of adolescents.1 In addition, acne frequently persists into adulthood, with more than 50% of women reporting acne in their 20s and 35% of women reporting acne in their 30s.2 The pathogenesis of acne is multifactorial, with production of androgens and sebogenic hormones playing an important role.3 Given that several forms of contraception can influence these hormonal signaling pathways, choice of contraception may influence the incidence and severity of acne in women.
Several placebo-controlled trials have demonstrated that combined oral contraceptives (OCs) are an effective treatment for acne and patients with acne are frequently prescribed combined OCs, with three combined OC products U.S. Food and Drug Administration–approved for acne.4–9 Although there are multiple combined OC formulations, and combined OCs containing progestins with antiandrogenic properties such as drospirenone may have greater effectiveness, a recent Cochrane Review concluded that there were no important and consistent differences among combined OCs’ influence on acne.8,10–14
However, acne can also be a reason for contraceptive discontinuation and different contraception classes may have differing effects on acne.15–19 Two small randomized trials have found that acne may be more likely to be reported among users of progestin-only methods than combined OCs.20,21 In addition, a survey study of patient-reported effects of contraception on acne suggests that progestin-only methods, including depot medroxyprogesterone acetate (DMPA) injections, the etonogestrel subdermal implant, and levonorgestrel-releasing intrauterine devices (IUDs) worsened acne, whereas combined OCs improved acne.14
Much of the existing literature is limited by its focus on a comparison of only a few contraception classes and reliance on patient report for acne outcomes. In addition, because combined OCs are an effective treatment for acne, it is possible that some of the incident acne reported with other contraceptives could be due to the discontinuation of prior combined OCs rather than from the contraceptive method itself. There is also a lack of data on the effects of progesterone-only contraception and long-acting reversible contraception methods on acne. The purpose of this study was to evaluate the association of contraceptive method on the incidence and severity of acne among new contraception users and among women who switched contraceptive methods. Based on their mechanism of action, we hypothesized that combined OCs would have a beneficial effect, the copper IUD would have a neutral effect, and progesterone-only OCs, levonorgestrel IUDs, the etonogestrel implant, and DMPA injection would have a negative effect on acne.
This study was a retrospective cohort study of female patients prescribed contraception in the Optum de-identified Clinformatics Data Mart Database, a de-identified commercial claims database for approximately 12–14 million individuals annually in the United States. These data include both medical and pharmacy claims, as well as patient demographic information such as age and gender. The patient population available in the database is similar to the demographics of the United States population with respect to gender, age, and geographic distribution.22
Study inclusion criteria were: 1) females aged 12–40 years; 2) at least 1 year of continuous enrollment before and after the index date, which was defined as the date of the first use of contraception; 3) at least 1 year of contraception use after the index date. Participants were excluded if they had a history of pregnancy within 1 year before or 1 year after the index date.
For the primary analysis of the incidence of clinical encounters for acne in the year after initiation of contraception, additional exclusion criteria were: 1) any pharmacy claim for an acne topical (ie, benzoyl peroxide, topical retinoids, topical antibiotics) before the index date and 2) any encounters with an International Classification of Diseases, Ninth (ICD-9) or Tenth Revision (ICD-10) code for acne (706.1, L70.0, L70.1, L70.8, L70.9) before the index date.
Contraception classes were defined as combined OCs, progestin-only OCs, DMPA injection, etonogestrel implant, levonorgestrel IUDs, and the nonhormonal copper IUD. Combined OCs were identified by their National Drug Codes. Prescriptions were consolidated into courses of therapy, with prescriptions separated by fewer than 30 days considered to be part of a single course of therapy. Injections for DMPA were identified by their National Drug Codes and Current Procedural Terminology codes and injections separated by fewer than 120 days (to allow for potential small gaps of up to 30 days between injections owing to logistical issues) were considered part of a single course of therapy. Placement and removal of IUDs and the etonogestrel implant were identified by their associated Current Procedural Terminology and ICD codes. If the same type of IUD was replaced at the time of IUD removal, these IUDs were considered part of a single course of therapy.
Outcomes were evaluated during 1 year of follow-up after the index date. For our primary analysis of the incidence of clinical encounters for acne after starting contraception, the outcome was defined as the first encounter with an ICD code for acne after the index date.23
In addition, to evaluate the association of contraception class on acne severity, a subgroup analysis was performed among a cohort of patients with a history of clinical encounters for acne examining the incidence of treatment escalation from topical acne medications to an oral tetracycline-class antibiotic in the year after initiation of contraception. This outcome was chosen because tetracycline-class antibiotics are the most common systemic therapy prescribed for acne.9,24
To evaluate the association of discontinuing a combined OC on acne, a subgroup analysis was performed evaluating incident clinical encounters for acne among patients who had been using a combined OC for at least 90 days and then discontinued this combined OC and either started another form of contraception within 90 days of discontinuation (which could be a different combined OC) or did not receive any prescription for contraception for the next year. To reduce risk of loss to follow-up for those on no contraception, at least one clinical encounter of any type after the index date was required in this analysis.
Age, history of polycystic ovarian syndrome, and calendar year in which the contraception was started were evaluated as factors that may influence the incidence of clinical encounters with an ICD code for acne and type of contraceptive prescribed. In addition, the total number of encounters (independent of those for acne) in the year before and the year after the index date were evaluated as factors that may influence the likelihood of having a clinical encounter with an ICD code for acne owing to increased interaction with the health care system.
Multivariable Cox proportional hazard models, adjusting for each of the covariates above, were used to evaluate the association between contraception class and the development of each of the outcomes of interest.25 Results are presented as adjusted hazard ratios (HRs), adjusted odds ratios, and 95% CIs. The assumption of proportionality was confirmed using log-log plots.
To evaluate for the potential influence of unmeasured confounding in our primary analysis, a sensitivity analysis was performed using E-values.26 Because ICD-9 codes for acne have a sensitivity and positive predictive value of approximately 85%, there is the possibility that some patients with acne may not receive an ICD code for acne before the index date.23 If these patients with acne who did not receive an ICD code before index data are also more likely to receive one form of contraception (for example, combined OCs, because these are U.S. Food and Drug Administration–approved for acne) and more likely to receive an ICD code for acne after the index date, it could result in confounding by indication. The E-value represents the minimum strength of association that an unmeasured confounder would need to have with both the treatment and outcome to fully explain away a specific treatment–outcome association. To help contextualize the results, we also evaluated the odds of having a history of acne for each type of contraception, controlling for age, history of polycystic ovarian syndrome, and calendar year in which the contraception was started. Statistical analyses were performed in Stata 15. This study was deemed exempt by Institutional Review Board of the University of Pennsylvania because it involved de-identified data. This study was conducted in adherence with the STROBE guidelines.27
A total of 336,738 females that met the study criteria for the primary analysis (Table 1). The most common contraceptives prescribed were combined OCs (82.9%). Other methods included levonorgestrel IUDs (10.6%), copper IUD (2.1%), etonogestrel implant (1.9%), progestin-only OC (1.3%), and DMPA injection (1.2%).
Incident acne was recorded among 2–8% of participants in the study, with higher rates among younger participants, as shown in Table 1. Compared with combined OCs, the copper IUD (adjusted HR 1.14; 95% CI 1.01–1.29) and levonorgestrel IUDs (adjusted HR 1.09; 95% CI 1.03–1.16) were associated with increased risk of clinical encounters with an ICD code for acne in the first year after the index date. The etonogestrel implant (adjusted HR 0.83; 95% CI 0.72–0.94) and the DMPA injection (adjusted HR 0.70; 95% CI 0.59–0.84) were associated with decreased risk of incident clinical encounters for acne (Table 2). The unadjusted absolute difference in the proportion of participants with acne between contraception classes was less than 1% (Table 1). The results were similar in a sensitivity analysis after excluding all participants with a history of polycystic ovarian syndrome (data not shown).
In the analysis to evaluate for the potential influence of unmeasured confounding of contraceptive method on incident acne, the E-Value point estimates ranged from 1.40 to 2.21 (Table 3). These signify the strength an unmeasured confounder would need to have with both the treatment and outcome to fully explain away the results. For example, the E-Value estimate for the DMPA injection was 2.21, with a lower 1.67 E-Value estimate for the 95% CI. If there were an unmeasured confounder that is associated with both being prescribed a DMPA injection and the development of incident acne with a strength of association greater than 1.67, the CI could be moved to include the null, and we could no longer conclude that DMPA injection is associated with a lower risk of incident acne than combined OCs among new contraceptive users.
Compared with combined OCs, the copper IUD (adjusted HR 1.44; 95% CI 1.00–2.06) and levonorgestrel IUDs (adjusted HR 1.34; 95% CI 1.10–1.64) were associated with increased risk of treatment escalation from topical acne medications to an oral tetracycline-class antibiotic in the first year after the index date (Table 2, n=21,178). None of the other comparisons were statistically significant.
Among those who switched from a combined OC to another form of contraception, compared with combined OCs, progestin-only OCs (adjusted HR 1.70; 1.23–2.35), levonorgestrel IUDs (adjusted HR 1.93; 95% CI 1.69–2.22), the etonogestrel implant (adjusted HR 1.45; 95% CI 1.08–1.95), and the copper IUD (adjusted HR 1.70; 95% CI 1.23–2.35) were each associated with increased risk of clinical encounters for acne (Table 4, n=113,708). Switching to DMPA injections or no contraception were not associated with statistically significant differences in the risk of incident clinical encounters for acne.
We evaluated the association of contraception method on the development of incident clinical encounters for acne and acne severity among two groups: new contraceptive users, and individuals switching between contraceptive methods. In each of our analyses, participants using a copper IUD or a levonorgestrel IUD were slightly more likely than those prescribed combined OCs to have incident or worsening acne. Participants using the etonogestrel implant or the DMPA injection were slightly less likely to have incident clinical encounters for acne, although these contraceptives were not statistically significantly associated with treatment escalation.
It is important to emphasize that although we found some differences in the relative risk of acne between contraceptive methods, they were small. Differences in the absolute risk of acne are small as well. In our study, the frequency of patients seeking care for incident acne varied by contraception type and age as shown in Table 1, but was low overall (less than 5%), and the absolute difference in incident clinical encounters for acne between contraception classes was less than 1%. The risk of acne should be weighed alongside other important factors including individual patient preferences and medical history, the effectiveness of the method, bleeding profile, and potential side-effects. In addition, given the potential effects of multiple acne treatments on a developing fetus (ie, topical retinoids, oral antibiotics, spironolactone, and isotretinoin),24 choosing an effective and patient-centered contraceptive option can help reduce the risk of fetal complications from treatments for acne.
In our analysis among individuals switching from combined OCs to other contraceptive methods, the HRs for incident acne were generally higher for each contraceptive than they were in our analysis of new contraceptive users. These findings suggest that some of the observed associations in other studies between noncombined OC contraception classes and acne may be due to prior discontinuation of combined OCs before starting the noncombined OC contraception method, and that overall, large differences in the risk of acne between contraceptive classes are unlikely.14
Consistent with a prior randomized trial comparing a combined OC with a levonorgestrel IUD, we found that acne was more likely to be reported among users of the IUD.21 In contrast, our finding that the etonogestrel implant and DMPA injections were associated with a decreased risk of incident acne among new contraceptive users conflicts with several prior studies.14,20 In addition, these differences were not present in the treatment escalation or treatment switching analyses. Although it is possible that these contraception methods may be protective against acne, it is important to consider the possibility of unmeasured confounding. Notably, participants who receive the etonogestrel implant and DMPA injections are substantially less likely to have a history of acne, with odds ratios that are greater in magnitude than the associated E-values for the 95% CIs (Table 3). We therefore urge caution in interpretation of these results because, if an unmeasured confounder, such as acne that is not coded, was also associated with decreased risk of subsequent acne codes detected in our study, these findings could be biased.
This study has several limitations. The use of electronic claims data can be a powerful tool for understanding subgroup effects such as contraceptive class, but it can be limited by loss to follow-up, indication bias (eg, patients with existing acne are more likely to receive combined OCs), and ascertainment bias. Although ICD codes for acne have been validated in a dermatology population, the accuracy of these codes has not been extensively studied in nondermatology populations.23 If dermatologists prescribe combined OCs more frequently than other contraceptives and subsequently code for incident acne more frequently than other providers, this could bias our results towards the null (ie, underestimation of the relative benefit of combined OCs over other contraceptives). Although we controlled for the number of visits before and after the index date to address ascertainment bias, it is possible that unmeasured confounders such as acne that is not coded could influence our observed results. To explore this potential issue, we performed a sensitivity analysis using E-values to quantify the potential magnitude of unmeasured confounding that would be required to influence our results.
Prior research has shown that combined OCs can reduce acne, and this study highlights that combined OCs are associated with a beneficial effect with respect to acne compared with other options and that the loss of a beneficial effect from combined OCs when switching to another method may partially explain why other contraception options have traditionally been felt to worsen acne. However, absolute differences between forms of contraception were small. Overall, patients should be reassured that combined OCs can reduce acne, and that there are unlikely to be major differences among other contraception classes on acne.
1. Stathakis V, Kilkenny M, Marks R. Descriptive epidemiology of acne vulgaris in the community. Australas J Dermatol 1997;38:115–23.
2. Collier CN, Harper JC, Cafardi JA, Cantrell WC, Wang W, Foster KW, et al. The prevalence of acne in adults 20 years and older. J Am Acad Dermatol 2008;58:56–9.
3. Layton AM, Eady EA, Whitehouse H, Del Rosso JQ, Fedorowicz Z, van Zuuren EJ. Oral spironolactone for acne vulgaris in adult females: a hybrid systematic review. Am J Clin Dermatol 2017;18:169–91.
4. George R, Clarke S, Thiboutot D. Hormonal therapy for acne. Semin Cutan Med Surg 2008;27:188–96.
5. Redmond GP, Olson WH, Lippman JS, Kafrissen ME, Jones TM, Jorizzo JL. Norgestimate and ethinyl estradiol in the treatment of acne vulgaris: a randomized, placebo-controlled trial. Obstet Gynecol 1997;89:615–22.
6. Maloney JM, Dietze P, Watson D, Niknian M, Lee-Rugh S, Sampson-Landers C, et al. Treatment of acne using a 3-milligram drospirenone/20-microgram ethinyl estradiol oral contraceptive administered in a 24/4 regimen: a randomized controlled trial. Obstet Gynecol 2008;112:773–81.
7. Koltun W, Maloney JM, Marr J, Kunz M. Treatment of moderate acne vulgaris using a combined oral contraceptive containing ethinylestradiol 20 μg plus drospirenone 3mg administered in a 24/4 regimen: a pooled analysis. Eur J Obstet Gynecol Reprod Biol 2011;155:171–5.
8. Arowojolu AO, Gallo MF, Lopez LM, Grimes DA. Combined oral contraceptive pills for treatment of acne. The Cochrane Database of Systematic Reviews 2012, Issue 7. Art. No.: CD004425. DOI: 10.1002/14651858.CD004425.pub6.
9. Barbieri JS, James WD, Margolis DJ. Trends in prescribing behavior of systemic agents used in the treatment of acne among dermatologists and nondermatologists: a retrospective analysis, 2004-2013. J Am Acad Dermatol 2017;77:456–63.e4.
10. van Vloten WA, van Haselen CW, van Zuuren EJ, Gerlinger C, Heithecker R. The effect of 2 combined oral Contraceptives containing either drospirenone or cyproterone acetate on acne and seborrhea. Cutis 2002;69:2–15.
11. Kelly S, Davies E, Fearns S, McKinnon C, Carter R, Gerlinger C, et al. Effects of oral contraceptives containing ethinylestradiol with either drospirenone or levonorgestrel on various parameters associated with well-being in healthy women: a randomized, single-blind, parallel-group, multicentre study. Clin Drug Investig 2010;30:325–36.
12. Thorneycroft IH, Gollnick H, Schellschmidt I. Superiority of a combined contraceptive containing drospirenone to a triphasic preparation containing norgestimate in acne treatment. Cutis 2004;74:123–30.
13. Mansour D, Verhoeven C, Sommer W, Weisberg E, Taneepanichskul S, Melis GB, et al. Efficacy and tolerability of a monophasic combined oral contraceptive containing nomegestrol acetate and 17β-oestradiol in a 24/4 regimen, in comparison to an oral contraceptive containing ethinylestradiol and drospirenone in a 21/7 regimen. Eur J Contracept Reprod Health Care 2011;16:430–43.
14. Lortscher D, Admani S, Satur N, Eichenfield LF. Hormonal contraceptives and acne: a retrospective analysis of 2147 patients. J Drugs Dermatol 2016;15:670–4.
15. Darney P, Patel A, Rosen K, Shapiro LS, Kaunitz AM. Safety and efficacy of a single-rod etonogestrel implant (Implanon): results from 11 international clinical trials. Fertil Steril 2009;91:1646–53.
16. Grunloh DS, Casner T, Secura GM, Peipert JF, Madden T. Characteristics associated with discontinuation of long-acting reversible contraception within the first 6 months of use. Obstet Gynecol 2013;122:1214–21.
17. Ayadi AE, Diamond-Smith N, Harper C, Kohn JE, Rocca C. Young women's experience of side effects and contraceptive discontinuation. Contraception 2017;96:287–8.
18. Fan G, Kang S, Ren M, Weisberg E, Lukkari-Lax E, Roth K, et al. A single-arm phase III study exploring the efficacy and safety of LNG-IUS 8, a low-dose levonorgestrel intrauterine contraceptive system (total content 13.5 mg), in an Asia-Pacific population. Contraception 2017;95:371–7.
19. Teal SB, Turok DK, Chen BA, Kimble T, Olariu AI, Creinin MD. Five-year contraceptive efficacy and safety of a levonorgestrel 52-mg intrauterine system. Obstet Gynecol 2019;133:63–70.
20. Bahamondes L, Brache V, Meirik O, Ali M, Habib N, Landoulsi S, et al. A 3-year multicentre randomized controlled trial of etonogestrel- and levonorgestrel-releasing contraceptive implants, with non-randomized matched copper-intrauterine device controls. Hum Reprod Oxf Engl 2015;30:2527–38.
21. Suhonen S, Haukkamaa M, Jakobsson T, Rauramo I. Clinical performance of a levonorgestrel-releasing intrauterine system and oral contraceptives in young nulliparous women: a comparative study. Contraception 2004;69:407–12.
23. Ejaz A, Malaiyandi V, Kim WB, Rogalska T, Alhusayen R. Validating the diagnostic code for acne in a tertiary care dermatology centre. Eur J Dermatol EJD 2015;25:469–71.
24. Zaenglein AL, Pathy AL, Schlosser BJ, Alikhan A, Baldwin HE, Berson DS, et al. Guidelines of care for the management of acne vulgaris. J Am Acad Dermatol 2016;74:945–73.e33.
25. Schmidt TH, Khanijow K, Cedars MI, Huddleston H, Pasch L, Wang ET, et al. Cutaneous findings and systemic associations in women with polycystic ovary syndrome. JAMA Dermatol 2016;152:391–8.
26. VanderWeele TJ, Ding P. Sensitivity analysis in observational research: introducing the E-value. Ann Intern Med 2017;167:268–74.
27. von Elm E, Altman DG, Egger M. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet Lond Engl 2007;370:1453–7.