Contraception xx (2014) xxx – xxx
Original research article
Does switching contraceptive from oral to a patch or vaginal ring change the likelihood of timely prescription refill?☆,☆☆ Amy Law a , Yi-Chien Lee b , Magdaliz Gorritz b , Leo Plouffe a,⁎ a
Bayer HealthCare, Whippany, NJ, USA LASER Analytica, New York, NY, USA Received 28 January 2014; revised 2 April 2014; accepted 13 April 2014 b
Abstract Objective: This study evaluated contraceptive refill patterns of women insured commercially in the US who switched from oral contraceptives (OCs) to the patch or vaginal ring and assessed if switching contraceptive methods changes refill patterns. Study Design: Women aged 15–44 with ≥ 2 patch or ring prescriptions and ≥ 2 OC prescriptions before the first patch/ring prescription were identified from the MarketScan® Commercial database (1/1/2002-6/30/2011). Refill patterns 1-year pre- and postindex date (first patch/ ring prescription) were evaluated, and women were categorized as timely or delayed refillers on OCs and patch/ring. Regression modeling was used to investigate the association between refill patterns and contraceptive methods and switching effects on refill patterns. Results: Of 17,814 women identified, 7901 switched to the patch, and 9913 switched to the ring. Among timely OC refillers, the percentage of timely refills decreased (patch: 95.6% to 79.4%, pb.001; ring: 96.5% to 74.3%, pb.001). However, among delayed OC refillers, the percentage of timely refills improved (patch: 47.9% to 72.2%, pb.001; ring: 50.4% to 64.0%, pb.001) during patch/ring use. Nonetheless, compared to timely OC refillers, women who were delayed OC refillers had 1.68-fold [95% confidence interval (CI): 1.52–1.84, pb.001] and 1.85-fold greater odds (CI: 1.69–2.02, pb.001) of being a delayed refiller while on the patch and ring, respectively. Conclusion: Switching to the patch or ring may improve refill behavior for women who have problems refilling OCs timely; however, the magnitude of the improvement may fail to improve ultimate contraceptive efficacy by simply switching to the patch or ring. Implications: The impact on timely refills of switching from OCs to either the patch or ring is complex and varies depending on the pattern of timely refills on OCs. © 2014 Elsevier Inc. All rights reserved. Keywords: Contraception; Contraceptive patch; Contraceptive ring; Oral contraceptives; Refill patterns
1. Introduction Approximately 50% of all pregnancies are unintended in the US, and it is estimated that approximately half are among women who become pregnant during contraceptive use [1,2]. When used consistently and correctly, oral contraceptives (OCs) are more than 99% effective, with the first-year pregnancy rate being 0.3% to 0.5% among users who
☆
Funding: This research was supported by Bayer HealthCare. Conflict of Interest: A Law and L Plouffe are employees of Bayer HealthCare Pharmaceuticals Inc. Y Lee and M Gorritz are employees of LASER Analytica and served as paid consultants to Bayer HealthCare Pharmaceuticals Inc. for conducting this study. ⁎ Corresponding author. Bayer HealthCare Pharmaceuticals. Tel.: +1 862 404 5596. E-mail address: [email protected] (L. Plouffe). ☆☆
http://dx.doi.org/10.1016/j.contraception.2014.04.005 0010-7824/© 2014 Elsevier Inc. All rights reserved.
manage near-perfect compliance to OC therapy [3,4]. However, 19% to 47% of OC users miss at least one pill per cycle [5]. Trussell (2011) reported that among women in the US, typical use of OCs is associated with 9% of women experiencing an unintended pregnancy within the first year of use compared to 0.3% of women who have perfect use [3]. Minimizing dosing frequency is one suggested strategy of improving compliance to contraceptives [6]. The contraceptive patch (hereinafter referred to as “patch”) is a transdermal system applied once weekly for 3 weeks followed by a 1-week pause, while the contraceptive ring (hereinafter referred to as “ring”) is applied once monthly for 21 days followed by a 1-week pause. As these methods of contraception have less frequent dosing than OCs, they may potentially improve compliance and, thereby, efficacy, especially among women who do poorly on OCs. However, several trials reporting pregnancy outcomes showed no difference in contraceptive
2
A. Law et al. / Contraception xx (2014) xxx–xxx
efficacy between the patch or ring and OCs, increasing the uncertainty around the actual contribution of the various dosing regimens of short-acting contraception to improved clinical efficacy [7]. Results of clinical trials, wherein treatment interventions are highly controlled, may not be generalizable to that of women in the real-world setting. Although actual medication use is not possible to assess using administrative claims databases, other studies have shown that studying medication acquisition (e.g., refill patterns) assists in understanding medication adherence through extrapolation [8,9]. This retrospective study aimed to evaluate real-world patterns of contraceptive refills among women insured commercially in the US who switched from using OCs to the patch or the ring and determine whether refill patterns change after switching contraceptive methods.
2. Materials and methods 2.1. Study population Women aged 15–44 with ≥ 2 patch prescriptions (ORTHO EVRA®, Janssen Pharmaceuticals, Titusville, NJ, USA) or ring prescriptions (NUVARING®, Merck & Co., Whitehouse Station, NJ, USA) and ≥ 2 OC prescriptions before the first patch or ring prescription were identified from the Truven Health Analytics MarketScan® Commercial Claims and Encounters (CCE) database between 1/1/2002 and 6/30/2011. National Drug Codes (NDC) and Healthcare Common Procedure Coding System (HCPCS) codes used to identify patch/ring claims are listed in the Appendix. The CCE database includes inpatient and outpatient medical claims and outpatient pharmacy claims for beneficiaries covered by employer-sponsored private health insurance. The medical claims data facilitate longitudinal studies by providing standardized, patient- and provider-level de-identified data spanning extensive time periods, and the database has been used in many other studies [10,11]. OC claims were identified by NDC of known OCs and other OCs with therapeutic class 168. The index date was defined as the date of the first patch/ring prescription filled after an OC prescription. Women were required to have ≤60-day gap between the last day of supply of an OC and a patch or ring prescription, ≥ 2 prescriptions during pre- and postindex periods to avoid having women who were just trying out the contraceptive method (e.g., for a single cycle or in combination with prescription samples) and ≥ 12 continuous months of health plan enrollment prior to and after the index date. Women were excluded if they had claims for incomplete contraceptive cycles or claims for more than 3 months supply of drugs since such claims may represent miscoding in the claim records. Women with claims for extended-cycle OCs and progestin-only pills were also excluded. 2.2. Measurements Refill patterns were evaluated over 365 days prior to the index date for OC use and over 365 days postindex date for
patch/ring use until the end of healthcare coverage, the end of the study period, discontinuation from OC/patch/ring or a pregnancy diagnosis, whichever occurred first. Discontinuation was defined as a gap of N 42 days between the end of previous contraceptive day of supply and the next contraceptive claim date. When unequal follow-up time was observed between contraceptive methods, the minimum follow-up time was applied to both methods. The total number of refills and the proportion that were timely, delayed or late were determined. A timely refill was defined as having no gap (0 days) between contraceptive claims after taking days of supply into account, while a delayed refill was defined as having a gap of 1–14 days between contraceptive claims, as similarly used by Nelson et al. (2008) [12]. Late refill was defined as having a gap of 15– 42 days between contraceptive claims. After evaluating the distribution of refill patterns, women were categorized as timely or delayed refillers. Women were considered timely refillers if at least 80% of their claims were characterized as timely refills. The cutoff of 80% was chosen as it was close to the median percentage of timely refills across the different treatment groups. In addition, the threshold for good medication adherence is commonly set at 80% [13]. 2.3. Statistical analysis Refill patterns while on OCs versus patch/ring were statistically compared using chi-square tests. A critical value of 0.05 was used to determine statistical significance. All statistical analyses were carried out using statistical analysis system (SAS) 9.3.
2.4. Multivariable regression analysis Logistic regression with stepwise selection was conducted to model the relationship between refill patterns on OCs and on the patch/ring while adjusting for other independent variables. The outcome for these models was whether a woman was a timely refiller (yes vs. no) while on the patch/ ring, and the main effect was whether a woman was a timely refiller while on OCs. The following covariates were initially included: age, US region, index year, health plan type, other disorders of the female genital tract, number of different OCs used, year of first OC claim, the gap between termination of OCs and initiation of the patch/ring, duration of follow-up period and monthly copayment difference. Only variables significant at the 0.2 level remained in the final models. Generalized estimating equation (GEE) regression models with a logit link and a binomial distribution assumption were used to investigate whether switching contraceptive methods had a significant effect on the percentage of timely refills. The final GEE models included covariates selected for the final logistic models. An interaction term between the switch effect and OC refill pattern was included to test for a differential switch effect among timely and delayed OC refillers.
A. Law et al. / Contraception xx (2014) xxx–xxx
The overall study design and analysis approach were reviewed and approved by the author team and an internal review committee prior to the execution of the study.
3
3. Results
prescription were identified between 1/1/2002 and 12/31/2011 (Fig. 1). After applying the selection criteria, the final study population included 17,814 women, of which 7901 switched to the patch and 9913 switched to the ring. Table 1 summarizes the demographics and clinical characteristics of study groups.
3.1. Study population
3.2. Refill patterns
A total of 102,128 women with ≥2 patch or ring prescriptions and ≥ 2 OC prescriptions before the first patch or ring
For both study groups, the proportion of timely refills decreased after switching to either the patch or ring versus
Fig. 1. Flow of selection of study population.
4
A. Law et al. / Contraception xx (2014) xxx–xxx
Table 1 Demographics and clinical characteristics of women who switched from OCs to the patch or ring (n= 17,814) Characteristic
Switch from OC Switch from OC to patch n=7901 to ring n= 9913
Age, years, mean (SD) Median Age, years, n (%) 15–18 19–25 26–30 31 + Region, n (%) Northeast North central South West Unknown Index year, n (%) 2002–2004 2005–2007 2008–2011 Health plan type, n (%) Comprehensive EPO HMO POS PPO POS w/capitation CDHP HDHP Comorbidities*, n (%) Inflammatory disease of pelvic organs Other disorders of genital tract** Neoplasms of reproductive organs or breast Psychological disorders Other
26.8 (7.8) 26.0
26.0 (6.8) 25.0
1415 (17.9%) 2329 (29.5%) 1522 (19.3%) 2635 (33.4%)
1107 (11.2%) 4078 (41.1%) 2270 (22.9%) 2458 (24.8%)
689 (8.7%) 2281 (28.9%) 3402 (43.1%) 1503 (19.0%) 26 (0.3%)
1245 (12.6%) 2807 (28.3%) 3921 (39.6%) 1928 (19.5%) 12 (0.1%)
4901 (62.0%) 2018 (25.5%) 982 (12.4%)
926 (9.3%) 2788 (28.1%) 6199 (62.5%)
649 (8.3%) 40 (0.5%) 1673 (21.4%) 1287 (16.4%) 3850 (49.2%) 250 (3.2%) 70 (0.9%) 13 (0.2%)
332 (3.4%) 103 (1.1%) 2005 (20.7%) 1093 (11.3%) 5642 (58.2%) 84 (0.9%) 349 (3.6%) 89 (0.9%)
962 (12.2%)
1261 (12.7%)
2512 (31.8%) 153 (1.9%)
3076 (31.0%) 154 (1.6%)
964 (12.2%) 1766 (22.4%)
1332 (13.4%) 2754 (27.8%)
*Within 1-year preindex; ** disorders include endometriosis, genital prolapse, fistula involving female genital tract, noninflammatory disorders of ovary fallopian tube and broad ligament, uterus, cervix, vagina, vulva, perineum, or other genital organs, pain and other symptoms associated with female genital organs, disorders of menstruation and other abnormal bleeding from female genital tract, menopausal and postmenopausal disorders, and female infertility. SD: standard deviation; EPO: exclusive provider organization; HMO: health maintenance organization; POS: point-of-service; PPO: preferred provider organization; CDHP: consumer-driven health plan; HDHP: high deductible health plan.
before (Fig. 2A and B). In general, regardless of switch method, there was a higher proportion of timely refillers while on OCs compared to when using the patch (63.6% on OC vs. 58.2% on patch; pb.001) or ring (67.8% on OC vs. 48.6% on ring; pb.001) (Table 2). Results from the multivariate logistic models indicated that, compared to timely OC refillers, women who were delayed refillers during OC use had 1.68-fold greater odds [95% confidence interval (CI): 1.52–1.84, pb.001] of being a delayed refiller while on the patch and 1.85-fold greater odds (CI: 1.69–2.02, pb.001) of being a delayed refiller while on the ring.
Fig. 2. Refill patterns of women before (preindex) and after (postindex) switching from OCs to the patch (A) and before and after switching to the ring (B).
3.3. Switch effects on refill pattern After adjusting for other covariates, the odds of having a timely refill after switching to the patch decreased compared to while on OCs (Odds ratio [OR]=0.78; CI: 0.74–0.82, pb.001) (Table 3). Among the study group who switched from OCs to the patch, the percentage of timely refills before and after switching were 81.7% and 76.0%, respectively. Differential switch effects were observed between timely and delayed OC refillers. After adjusting for these effects, the mean percentage of timely refills among women who were timely refillers decreased from 95.6% while on OCs to 79.4% while on the patch (pb.001). Among delayed refillers, the mean
Table 2 Contraceptive compliance of women while on OCs and while on the patch or ring (n= 17,814) Switch from OC to patch n=7901
Timely refillers, n (%) Delayed refillers, n (%)
Switch from OC to ring n=9913
OC
Patch
p Value
OC
Ring
p Value
5027 (63.6) 2874 (36.4)
4601 (58.2) 3300 (41.8)
b.001
6725 (67.8) 3188 (32.2)
4822 (48.6) 5091 (51.4)
b .001
A. Law et al. / Contraception xx (2014) xxx–xxx Table 3 Odds of having a timely refill after switching to the patch or ring versus before switching and estimated percent of timely refills before and after switching to the patch or ring
After switch to patch versus before
Odds ratio
CI lower
CI upper
p Value
0.78
0.74
0.82
b.001
Estimated % of CI lower CI upper p Value timely refills Entire OC switch to patch population Before switch (OC) After switch (patch) Stratified by refill pattern Timely OC refiller Before switch (OC) After switch (patch) Delayed OC refiller Before switch (OC) After switch (patch)
81.7% 76.0%
80.2% 74.1%
83.1% 77.7%
b.001
95.6% 79.4%
95.1% 77.6%
96.0% 81.0%
b.001
47.9% 72.2%
45.4% 70.0%
50.4% 74.3%
b.001
Odds ratio
CI lower
CI upper
p Value
0.42
0.46
b.001
Estimated % of CI timely refills lower
CI upper
p Value
84.1% 69.4%
83.6% 68.7%
84.5% 70.1%
b.001
96.5% 74.3%
96.3% 73.6%
96.7% 75.0%
b.001
50.4% 64.0%
49.6% 56.3%
51.3% 62.8%
b.001
After switch to ring versus 0.44 before
Entire OC switch to ring population Before switch (OC) After switch (ring) Stratified by refill pattern Timely OC refiller Before switch (OC) After switch (ring) Delayed OC refiller Before switch (OC) After switch (ring)
percentage of timely refills increased from 47.9% while on OCs to 72.2% while on the patch (pb.001); however, the percentage of timely refills while on the patch remained below the 80% threshold required to be categorized as a timely refiller. Fig. 3 presents the estimated percentage of timely refills based on the GEE model further broken down by refill pattern while on OCs into five categories (0–20%, 20–40%, 40–60% and 60–80% timely refills). The timely OC refiller group (80–100% of timely refills) was the only group with a decreasing trend (96% to 80%) of timely refills while on the patch. For the remaining subgroups, the percentage of timely refills increased to levels between 66 and 76%, but no group reached the 80% threshold to be classified as timely refillers. Results were similar for women who switched to the ring from OCs; however, the switch effect was more pronounced (Table 3, Fig. 4).
5
4. Discussion In this study of nearly 18,000 women who had commercial insurance in the US, approximately one third did not refill their OC prescriptions in a timely manner. Overall, switching to the patch or ring did not improve refill patterns as greater proportions of women on the patch and ring were delayed refillers (42% and 51%, respectively). A recent study conducted by Pittman et al. (2011), in which contraception compliance was also extrapolated through refill pattern, reported that over two thirds of women who used OCs, the patch or the ring did not have timely refill compliance [9]. However, Pittman et al. (2011) defined timely refill compliance differently [9]. In addition, the latter study was part of the CHOICE Project and was conducted on a population of women who had just chosen to use OCs, the patch or ring [9]. It included a higher percentage of younger women (66% ≤24years of age) and noncommercially insured women (37%) than ours, which may have contributed to the higher frequency of noncompliance observed while on OCs, the patch or ring in comparison to the results of our study [9]. In contrast to the idea that switching to alternative contraceptive methods, which require less frequent administration than OCs may improve compliance and thus contraceptive efficacy [14], our study, based on an extrapolated measure of compliance, suggests that the dynamics of compliance to various contraceptive methods are complex. After adjusting for covariates, among women who switched to the patch, the percentage of timely refills declined approximately 6% (81.7% to 76.0%), and among women who switched to the ring, the decline was nearly 15% (84.1% to 69.4%). These overall results, however, fail to highlight the full dynamics of refill patterns. After adjusting for differential switch effects among timely and delayed OC refillers, we did observe that switching to the patch or ring had a positive effect on delayed OC refillers but a negative effect on timely OC refillers. In regard to the overall study group that switched to the patch, our data contrast with the results from randomized clinical trials in which compliance was compared among women treated with different contraceptive methods. Audet et al. (2001) and Archer et al. (2004) both reported that women in the US and/or Canada who use the patch are more likely to comply with the regimen than OC users (Audet: patch=88.2% vs. OC=77.7%; Archer: patch=88.7% vs. OC= 79.2%) [15,16]. Clinical trials have shown conflicting results regarding compliance between OC users and ring users with one reporting greater compliance among ring users and another reporting much less compliance among ring users than OC users [7]. Clinical trials include actual validated measures for compliance; our study can only extrapolate compliance based on refill patterns. In addition, clinical trials compared two independent populations, while the current study focused on a single population that switched from OCs to the patch/ring, with each woman serving as her own control. In summary, the seemingly conflicting results may be due, in part, to different target populations and the approach to measure or extrapolate compliance.
6
A. Law et al. / Contraception xx (2014) xxx–xxx
Fig. 3. Estimated percentage of timely refills before and after switching from OCs to the contraceptive patch by percentage of timely refills while on OCs (in 20% intervals) after adjusting for other covariates.
4.1. Limitations and strengths As administrative claims data are recorded for billing purposes rather than health research, data may be inconsistently reported across sites or insurance plans. The data quality for certain key variables was inconsistent (e.g., unusual and extreme values for days of supply were observed). However, since the sample size was large, it was possible to exclude women with unusual claims, thus minimizing the possible bias that may result from using inconsistent data. In the final regression models, index year, health plan type and duration of follow-up period influenced whether a woman was a timely refiller while on the patch, and age, health plan type and duration of follow-up period influenced whether a woman was a timely refiller while on the ring. These factors were controlled for accordingly. Women with only one claim for an OC, patch or ring were excluded from the study population. As an exploratory evaluation, we determined the number of women excluded due to having only one OC claim and found that the percentage of such women was not large. Thus, this exclusion was not likely to have influenced the main study findings. The evaluation of women with only one OC prescription may be interesting in a
separate future research study interested in examining women who discontinue contraceptive methods after only one try. Actual use of contraceptive methods could not be determined in this study based on analysis of prescription database claims. However, many other studies have used prescription fill data from pharmacy claims as a proxy for drug use [8–12]. By using claims data, we were able to assess the real-world refill pattern, thus avoiding bias resulting from patient awareness of monitoring. Other observational studies have also used refill pattern as a surrogate for medication compliance [9,17]. In this study, women were considered timely refillers if at least 80% of their claims were characterized as timely refills. As a sensitivity analysis, we examined the influence of different threshold criteria (75%, 85%) on the odds of having a timely refill after switching to the patch or ring versus before switching and observed similar results as in the default analysis. Finally, the results from this study may not be generalizable to women who do not have commercial insurance. 5. Conclusions The results from this study suggest that switching from OCs to either the patch or ring does not necessarily improve compliance, as estimated by refill pattern, as suggested. Switching to the patch or ring may prove beneficial for women who have problems refilling their OCs; however, the magnitude of the improvement may fail to improve ultimate contraceptive efficacy by simply switching to the patch or ring. Acknowledgments Editorial assistance was provided by Melissa Lingohr-Smith of Novosys Health and was funded by Bayer HealthCare. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.contraception.2014.04.005. References
Fig. 4. Estimated percentage of timely refills before and after switching from OCs to the contraceptive ring by percentage of timely refills while on OCs (in 20% intervals) after adjusting for other covariates.
[1] Finer LB, Zolna MR. Unintended pregnancy in the United States: incidence and disparities, 2006. Contraception 2011;84:478–85 [PubMed: 22018121]. [2] Finer LB, Henshaw SK. Disparities in rates of unintended pregnancy in the United States, 1994 and 2001. Perspect Sex Reprod Health 2006;38:90–6 [PubMed: 16772190]. [3] Trussell J. Contraceptive failure in the United States. Contraception 2011;83:397–404 [PubMed: 21477680]. [4] Potter L, Oakley D, de Leon-Wong E, Canamar R. Measuring compliance among oral contraceptive users. Fam Plann Perspect 1996;28:154–8 [PubMed: 8853280]. [5] Pierson RA, Archer DF, Moreau M, Shangold GA, Fisher AC, Creasy GW. Ortho Evra(tm)/Evra(tm) versus oral contraceptives: follicular development and ovulation in normal cycles and after an intentional dosing error. Fertil Steril 2003;80:34–42 [PubMed: 12849799]. [6] U.S. selected practice recommendations for contraceptive use, 2013: adapted from the World Health Organization Selected Practice
A. Law et al. / Contraception xx (2014) xxx–xxx
[7]
[8]
[9]
[10]
[11]
Recommendations for Contraceptive Use, 2nd Edition. Morb Mortal Wkly Rep 2013;62(RR05):1–6 [Available at: http://www.cdc.gov/ mmwr/preview/mmwrhtml/rr6205a1.htm?s_cid=rr6205a1_w]. Lopez LM, Grimes DA, Gallo MF, Stockton LL, Shulz KF. Skin patch and vaginal ring versus combined oral contraceptives for contraception. Cochrane Database Syst Rev 2013(4):CD003552 [PubMed:23633314]. Steiner JF, Prochazka AV. The assessment of refill compliance using pharmacy records: methods, validity, and applications. J Clin Epidemiol 1997;50:105–16 [PubMed: 9048695]. Pittman ME, Secura GM, Allsworth JE, Homco JB, Madden T, Peipert JF. Understanding prescription adherence: pharmacy claims data from the contraceptive CHOICE Project. Contraception 2011;83:340–5 [PubMed:21397092]. Xu X, Macaluso M, Ouyang L, Kulczycki A, Grosse SD. Revival of the intrauterine device: increased insertions among US women with employer-sponsored insurance, 2002-2008. Contraception 2012;85:155–9 [PubMed:22067778]. Bonafede MM, Johnson BH, Wenten M, Watson C. Treatment patterns in disease-modifying therapy for patients with multiple sclerosis in the United States. Clin Ther 2013;35:1501–15 [PubMed:24139422].
7
[12] Nelson AL, Westhoff C, Schnare SM. Real-world patterns of prescription refills for branded hormonal contraceptives. Obstet Gynecol 2008;112:782–7 [PubMed:18827120]. [13] Kozma CM, Dickson M, Phillips AL, Meletiche DM. Medication possession ratio: implications of using fixed and variable observation periods in assessing adherence with disease-modifying drugs in patients with multiple sclerosis. Patient Prefer Adherence 2013;7:509–16 [PubMed:23807840]. [14] Ingersoll KS, Cohen J. The impact of medication regimen factors on adherence to chronic treatment: a review of the literature. J Behav Med 2008;31:213–24 [PubMed: 2868342]. [15] Audet MC, Moreau M, Koltun WD, Waldbaum AS, Shagold G, Fisher AC, et al. Evaluation of contraceptive efficacy and cycle control of a transdermal contraceptive patch vs an oral contraceptive: a randomized controlled trial. JAMA 2001;285:2347–54 [PubMed: 11343482]. [16] Archer DF, Cullins V, Creasy GW, Fisher AC. The impact of improved compliance with a weekly contraceptive transdermal system (Ortho Evra) on contraceptive efficacy. Contraception 2004;69:189–95 [PubMed: 14969665]. [17] Huybrechts KF, Ishak KJ, Caro JJ. Assessment of compliance with osteoporosis treatment and its consequences in a managed care population. Bone 2006;38:922–8 [PubMed:16330270].
Original research article
Does switching contraceptive from oral to a patch or vaginal ring change the likelihood of timely prescription refill?☆,☆☆ Amy Law a , Yi-Chien Lee b , Magdaliz Gorritz b , Leo Plouffe a,⁎ a
Bayer HealthCare, Whippany, NJ, USA LASER Analytica, New York, NY, USA Received 28 January 2014; revised 2 April 2014; accepted 13 April 2014 b
Abstract Objective: This study evaluated contraceptive refill patterns of women insured commercially in the US who switched from oral contraceptives (OCs) to the patch or vaginal ring and assessed if switching contraceptive methods changes refill patterns. Study Design: Women aged 15–44 with ≥ 2 patch or ring prescriptions and ≥ 2 OC prescriptions before the first patch/ring prescription were identified from the MarketScan® Commercial database (1/1/2002-6/30/2011). Refill patterns 1-year pre- and postindex date (first patch/ ring prescription) were evaluated, and women were categorized as timely or delayed refillers on OCs and patch/ring. Regression modeling was used to investigate the association between refill patterns and contraceptive methods and switching effects on refill patterns. Results: Of 17,814 women identified, 7901 switched to the patch, and 9913 switched to the ring. Among timely OC refillers, the percentage of timely refills decreased (patch: 95.6% to 79.4%, pb.001; ring: 96.5% to 74.3%, pb.001). However, among delayed OC refillers, the percentage of timely refills improved (patch: 47.9% to 72.2%, pb.001; ring: 50.4% to 64.0%, pb.001) during patch/ring use. Nonetheless, compared to timely OC refillers, women who were delayed OC refillers had 1.68-fold [95% confidence interval (CI): 1.52–1.84, pb.001] and 1.85-fold greater odds (CI: 1.69–2.02, pb.001) of being a delayed refiller while on the patch and ring, respectively. Conclusion: Switching to the patch or ring may improve refill behavior for women who have problems refilling OCs timely; however, the magnitude of the improvement may fail to improve ultimate contraceptive efficacy by simply switching to the patch or ring. Implications: The impact on timely refills of switching from OCs to either the patch or ring is complex and varies depending on the pattern of timely refills on OCs. © 2014 Elsevier Inc. All rights reserved. Keywords: Contraception; Contraceptive patch; Contraceptive ring; Oral contraceptives; Refill patterns
1. Introduction Approximately 50% of all pregnancies are unintended in the US, and it is estimated that approximately half are among women who become pregnant during contraceptive use [1,2]. When used consistently and correctly, oral contraceptives (OCs) are more than 99% effective, with the first-year pregnancy rate being 0.3% to 0.5% among users who
☆
Funding: This research was supported by Bayer HealthCare. Conflict of Interest: A Law and L Plouffe are employees of Bayer HealthCare Pharmaceuticals Inc. Y Lee and M Gorritz are employees of LASER Analytica and served as paid consultants to Bayer HealthCare Pharmaceuticals Inc. for conducting this study. ⁎ Corresponding author. Bayer HealthCare Pharmaceuticals. Tel.: +1 862 404 5596. E-mail address: [email protected] (L. Plouffe). ☆☆
http://dx.doi.org/10.1016/j.contraception.2014.04.005 0010-7824/© 2014 Elsevier Inc. All rights reserved.
manage near-perfect compliance to OC therapy [3,4]. However, 19% to 47% of OC users miss at least one pill per cycle [5]. Trussell (2011) reported that among women in the US, typical use of OCs is associated with 9% of women experiencing an unintended pregnancy within the first year of use compared to 0.3% of women who have perfect use [3]. Minimizing dosing frequency is one suggested strategy of improving compliance to contraceptives [6]. The contraceptive patch (hereinafter referred to as “patch”) is a transdermal system applied once weekly for 3 weeks followed by a 1-week pause, while the contraceptive ring (hereinafter referred to as “ring”) is applied once monthly for 21 days followed by a 1-week pause. As these methods of contraception have less frequent dosing than OCs, they may potentially improve compliance and, thereby, efficacy, especially among women who do poorly on OCs. However, several trials reporting pregnancy outcomes showed no difference in contraceptive
2
A. Law et al. / Contraception xx (2014) xxx–xxx
efficacy between the patch or ring and OCs, increasing the uncertainty around the actual contribution of the various dosing regimens of short-acting contraception to improved clinical efficacy [7]. Results of clinical trials, wherein treatment interventions are highly controlled, may not be generalizable to that of women in the real-world setting. Although actual medication use is not possible to assess using administrative claims databases, other studies have shown that studying medication acquisition (e.g., refill patterns) assists in understanding medication adherence through extrapolation [8,9]. This retrospective study aimed to evaluate real-world patterns of contraceptive refills among women insured commercially in the US who switched from using OCs to the patch or the ring and determine whether refill patterns change after switching contraceptive methods.
2. Materials and methods 2.1. Study population Women aged 15–44 with ≥ 2 patch prescriptions (ORTHO EVRA®, Janssen Pharmaceuticals, Titusville, NJ, USA) or ring prescriptions (NUVARING®, Merck & Co., Whitehouse Station, NJ, USA) and ≥ 2 OC prescriptions before the first patch or ring prescription were identified from the Truven Health Analytics MarketScan® Commercial Claims and Encounters (CCE) database between 1/1/2002 and 6/30/2011. National Drug Codes (NDC) and Healthcare Common Procedure Coding System (HCPCS) codes used to identify patch/ring claims are listed in the Appendix. The CCE database includes inpatient and outpatient medical claims and outpatient pharmacy claims for beneficiaries covered by employer-sponsored private health insurance. The medical claims data facilitate longitudinal studies by providing standardized, patient- and provider-level de-identified data spanning extensive time periods, and the database has been used in many other studies [10,11]. OC claims were identified by NDC of known OCs and other OCs with therapeutic class 168. The index date was defined as the date of the first patch/ring prescription filled after an OC prescription. Women were required to have ≤60-day gap between the last day of supply of an OC and a patch or ring prescription, ≥ 2 prescriptions during pre- and postindex periods to avoid having women who were just trying out the contraceptive method (e.g., for a single cycle or in combination with prescription samples) and ≥ 12 continuous months of health plan enrollment prior to and after the index date. Women were excluded if they had claims for incomplete contraceptive cycles or claims for more than 3 months supply of drugs since such claims may represent miscoding in the claim records. Women with claims for extended-cycle OCs and progestin-only pills were also excluded. 2.2. Measurements Refill patterns were evaluated over 365 days prior to the index date for OC use and over 365 days postindex date for
patch/ring use until the end of healthcare coverage, the end of the study period, discontinuation from OC/patch/ring or a pregnancy diagnosis, whichever occurred first. Discontinuation was defined as a gap of N 42 days between the end of previous contraceptive day of supply and the next contraceptive claim date. When unequal follow-up time was observed between contraceptive methods, the minimum follow-up time was applied to both methods. The total number of refills and the proportion that were timely, delayed or late were determined. A timely refill was defined as having no gap (0 days) between contraceptive claims after taking days of supply into account, while a delayed refill was defined as having a gap of 1–14 days between contraceptive claims, as similarly used by Nelson et al. (2008) [12]. Late refill was defined as having a gap of 15– 42 days between contraceptive claims. After evaluating the distribution of refill patterns, women were categorized as timely or delayed refillers. Women were considered timely refillers if at least 80% of their claims were characterized as timely refills. The cutoff of 80% was chosen as it was close to the median percentage of timely refills across the different treatment groups. In addition, the threshold for good medication adherence is commonly set at 80% [13]. 2.3. Statistical analysis Refill patterns while on OCs versus patch/ring were statistically compared using chi-square tests. A critical value of 0.05 was used to determine statistical significance. All statistical analyses were carried out using statistical analysis system (SAS) 9.3.
2.4. Multivariable regression analysis Logistic regression with stepwise selection was conducted to model the relationship between refill patterns on OCs and on the patch/ring while adjusting for other independent variables. The outcome for these models was whether a woman was a timely refiller (yes vs. no) while on the patch/ ring, and the main effect was whether a woman was a timely refiller while on OCs. The following covariates were initially included: age, US region, index year, health plan type, other disorders of the female genital tract, number of different OCs used, year of first OC claim, the gap between termination of OCs and initiation of the patch/ring, duration of follow-up period and monthly copayment difference. Only variables significant at the 0.2 level remained in the final models. Generalized estimating equation (GEE) regression models with a logit link and a binomial distribution assumption were used to investigate whether switching contraceptive methods had a significant effect on the percentage of timely refills. The final GEE models included covariates selected for the final logistic models. An interaction term between the switch effect and OC refill pattern was included to test for a differential switch effect among timely and delayed OC refillers.
A. Law et al. / Contraception xx (2014) xxx–xxx
The overall study design and analysis approach were reviewed and approved by the author team and an internal review committee prior to the execution of the study.
3
3. Results
prescription were identified between 1/1/2002 and 12/31/2011 (Fig. 1). After applying the selection criteria, the final study population included 17,814 women, of which 7901 switched to the patch and 9913 switched to the ring. Table 1 summarizes the demographics and clinical characteristics of study groups.
3.1. Study population
3.2. Refill patterns
A total of 102,128 women with ≥2 patch or ring prescriptions and ≥ 2 OC prescriptions before the first patch or ring
For both study groups, the proportion of timely refills decreased after switching to either the patch or ring versus
Fig. 1. Flow of selection of study population.
4
A. Law et al. / Contraception xx (2014) xxx–xxx
Table 1 Demographics and clinical characteristics of women who switched from OCs to the patch or ring (n= 17,814) Characteristic
Switch from OC Switch from OC to patch n=7901 to ring n= 9913
Age, years, mean (SD) Median Age, years, n (%) 15–18 19–25 26–30 31 + Region, n (%) Northeast North central South West Unknown Index year, n (%) 2002–2004 2005–2007 2008–2011 Health plan type, n (%) Comprehensive EPO HMO POS PPO POS w/capitation CDHP HDHP Comorbidities*, n (%) Inflammatory disease of pelvic organs Other disorders of genital tract** Neoplasms of reproductive organs or breast Psychological disorders Other
26.8 (7.8) 26.0
26.0 (6.8) 25.0
1415 (17.9%) 2329 (29.5%) 1522 (19.3%) 2635 (33.4%)
1107 (11.2%) 4078 (41.1%) 2270 (22.9%) 2458 (24.8%)
689 (8.7%) 2281 (28.9%) 3402 (43.1%) 1503 (19.0%) 26 (0.3%)
1245 (12.6%) 2807 (28.3%) 3921 (39.6%) 1928 (19.5%) 12 (0.1%)
4901 (62.0%) 2018 (25.5%) 982 (12.4%)
926 (9.3%) 2788 (28.1%) 6199 (62.5%)
649 (8.3%) 40 (0.5%) 1673 (21.4%) 1287 (16.4%) 3850 (49.2%) 250 (3.2%) 70 (0.9%) 13 (0.2%)
332 (3.4%) 103 (1.1%) 2005 (20.7%) 1093 (11.3%) 5642 (58.2%) 84 (0.9%) 349 (3.6%) 89 (0.9%)
962 (12.2%)
1261 (12.7%)
2512 (31.8%) 153 (1.9%)
3076 (31.0%) 154 (1.6%)
964 (12.2%) 1766 (22.4%)
1332 (13.4%) 2754 (27.8%)
*Within 1-year preindex; ** disorders include endometriosis, genital prolapse, fistula involving female genital tract, noninflammatory disorders of ovary fallopian tube and broad ligament, uterus, cervix, vagina, vulva, perineum, or other genital organs, pain and other symptoms associated with female genital organs, disorders of menstruation and other abnormal bleeding from female genital tract, menopausal and postmenopausal disorders, and female infertility. SD: standard deviation; EPO: exclusive provider organization; HMO: health maintenance organization; POS: point-of-service; PPO: preferred provider organization; CDHP: consumer-driven health plan; HDHP: high deductible health plan.
before (Fig. 2A and B). In general, regardless of switch method, there was a higher proportion of timely refillers while on OCs compared to when using the patch (63.6% on OC vs. 58.2% on patch; pb.001) or ring (67.8% on OC vs. 48.6% on ring; pb.001) (Table 2). Results from the multivariate logistic models indicated that, compared to timely OC refillers, women who were delayed refillers during OC use had 1.68-fold greater odds [95% confidence interval (CI): 1.52–1.84, pb.001] of being a delayed refiller while on the patch and 1.85-fold greater odds (CI: 1.69–2.02, pb.001) of being a delayed refiller while on the ring.
Fig. 2. Refill patterns of women before (preindex) and after (postindex) switching from OCs to the patch (A) and before and after switching to the ring (B).
3.3. Switch effects on refill pattern After adjusting for other covariates, the odds of having a timely refill after switching to the patch decreased compared to while on OCs (Odds ratio [OR]=0.78; CI: 0.74–0.82, pb.001) (Table 3). Among the study group who switched from OCs to the patch, the percentage of timely refills before and after switching were 81.7% and 76.0%, respectively. Differential switch effects were observed between timely and delayed OC refillers. After adjusting for these effects, the mean percentage of timely refills among women who were timely refillers decreased from 95.6% while on OCs to 79.4% while on the patch (pb.001). Among delayed refillers, the mean
Table 2 Contraceptive compliance of women while on OCs and while on the patch or ring (n= 17,814) Switch from OC to patch n=7901
Timely refillers, n (%) Delayed refillers, n (%)
Switch from OC to ring n=9913
OC
Patch
p Value
OC
Ring
p Value
5027 (63.6) 2874 (36.4)
4601 (58.2) 3300 (41.8)
b.001
6725 (67.8) 3188 (32.2)
4822 (48.6) 5091 (51.4)
b .001
A. Law et al. / Contraception xx (2014) xxx–xxx Table 3 Odds of having a timely refill after switching to the patch or ring versus before switching and estimated percent of timely refills before and after switching to the patch or ring
After switch to patch versus before
Odds ratio
CI lower
CI upper
p Value
0.78
0.74
0.82
b.001
Estimated % of CI lower CI upper p Value timely refills Entire OC switch to patch population Before switch (OC) After switch (patch) Stratified by refill pattern Timely OC refiller Before switch (OC) After switch (patch) Delayed OC refiller Before switch (OC) After switch (patch)
81.7% 76.0%
80.2% 74.1%
83.1% 77.7%
b.001
95.6% 79.4%
95.1% 77.6%
96.0% 81.0%
b.001
47.9% 72.2%
45.4% 70.0%
50.4% 74.3%
b.001
Odds ratio
CI lower
CI upper
p Value
0.42
0.46
b.001
Estimated % of CI timely refills lower
CI upper
p Value
84.1% 69.4%
83.6% 68.7%
84.5% 70.1%
b.001
96.5% 74.3%
96.3% 73.6%
96.7% 75.0%
b.001
50.4% 64.0%
49.6% 56.3%
51.3% 62.8%
b.001
After switch to ring versus 0.44 before
Entire OC switch to ring population Before switch (OC) After switch (ring) Stratified by refill pattern Timely OC refiller Before switch (OC) After switch (ring) Delayed OC refiller Before switch (OC) After switch (ring)
percentage of timely refills increased from 47.9% while on OCs to 72.2% while on the patch (pb.001); however, the percentage of timely refills while on the patch remained below the 80% threshold required to be categorized as a timely refiller. Fig. 3 presents the estimated percentage of timely refills based on the GEE model further broken down by refill pattern while on OCs into five categories (0–20%, 20–40%, 40–60% and 60–80% timely refills). The timely OC refiller group (80–100% of timely refills) was the only group with a decreasing trend (96% to 80%) of timely refills while on the patch. For the remaining subgroups, the percentage of timely refills increased to levels between 66 and 76%, but no group reached the 80% threshold to be classified as timely refillers. Results were similar for women who switched to the ring from OCs; however, the switch effect was more pronounced (Table 3, Fig. 4).
5
4. Discussion In this study of nearly 18,000 women who had commercial insurance in the US, approximately one third did not refill their OC prescriptions in a timely manner. Overall, switching to the patch or ring did not improve refill patterns as greater proportions of women on the patch and ring were delayed refillers (42% and 51%, respectively). A recent study conducted by Pittman et al. (2011), in which contraception compliance was also extrapolated through refill pattern, reported that over two thirds of women who used OCs, the patch or the ring did not have timely refill compliance [9]. However, Pittman et al. (2011) defined timely refill compliance differently [9]. In addition, the latter study was part of the CHOICE Project and was conducted on a population of women who had just chosen to use OCs, the patch or ring [9]. It included a higher percentage of younger women (66% ≤24years of age) and noncommercially insured women (37%) than ours, which may have contributed to the higher frequency of noncompliance observed while on OCs, the patch or ring in comparison to the results of our study [9]. In contrast to the idea that switching to alternative contraceptive methods, which require less frequent administration than OCs may improve compliance and thus contraceptive efficacy [14], our study, based on an extrapolated measure of compliance, suggests that the dynamics of compliance to various contraceptive methods are complex. After adjusting for covariates, among women who switched to the patch, the percentage of timely refills declined approximately 6% (81.7% to 76.0%), and among women who switched to the ring, the decline was nearly 15% (84.1% to 69.4%). These overall results, however, fail to highlight the full dynamics of refill patterns. After adjusting for differential switch effects among timely and delayed OC refillers, we did observe that switching to the patch or ring had a positive effect on delayed OC refillers but a negative effect on timely OC refillers. In regard to the overall study group that switched to the patch, our data contrast with the results from randomized clinical trials in which compliance was compared among women treated with different contraceptive methods. Audet et al. (2001) and Archer et al. (2004) both reported that women in the US and/or Canada who use the patch are more likely to comply with the regimen than OC users (Audet: patch=88.2% vs. OC=77.7%; Archer: patch=88.7% vs. OC= 79.2%) [15,16]. Clinical trials have shown conflicting results regarding compliance between OC users and ring users with one reporting greater compliance among ring users and another reporting much less compliance among ring users than OC users [7]. Clinical trials include actual validated measures for compliance; our study can only extrapolate compliance based on refill patterns. In addition, clinical trials compared two independent populations, while the current study focused on a single population that switched from OCs to the patch/ring, with each woman serving as her own control. In summary, the seemingly conflicting results may be due, in part, to different target populations and the approach to measure or extrapolate compliance.
6
A. Law et al. / Contraception xx (2014) xxx–xxx
Fig. 3. Estimated percentage of timely refills before and after switching from OCs to the contraceptive patch by percentage of timely refills while on OCs (in 20% intervals) after adjusting for other covariates.
4.1. Limitations and strengths As administrative claims data are recorded for billing purposes rather than health research, data may be inconsistently reported across sites or insurance plans. The data quality for certain key variables was inconsistent (e.g., unusual and extreme values for days of supply were observed). However, since the sample size was large, it was possible to exclude women with unusual claims, thus minimizing the possible bias that may result from using inconsistent data. In the final regression models, index year, health plan type and duration of follow-up period influenced whether a woman was a timely refiller while on the patch, and age, health plan type and duration of follow-up period influenced whether a woman was a timely refiller while on the ring. These factors were controlled for accordingly. Women with only one claim for an OC, patch or ring were excluded from the study population. As an exploratory evaluation, we determined the number of women excluded due to having only one OC claim and found that the percentage of such women was not large. Thus, this exclusion was not likely to have influenced the main study findings. The evaluation of women with only one OC prescription may be interesting in a
separate future research study interested in examining women who discontinue contraceptive methods after only one try. Actual use of contraceptive methods could not be determined in this study based on analysis of prescription database claims. However, many other studies have used prescription fill data from pharmacy claims as a proxy for drug use [8–12]. By using claims data, we were able to assess the real-world refill pattern, thus avoiding bias resulting from patient awareness of monitoring. Other observational studies have also used refill pattern as a surrogate for medication compliance [9,17]. In this study, women were considered timely refillers if at least 80% of their claims were characterized as timely refills. As a sensitivity analysis, we examined the influence of different threshold criteria (75%, 85%) on the odds of having a timely refill after switching to the patch or ring versus before switching and observed similar results as in the default analysis. Finally, the results from this study may not be generalizable to women who do not have commercial insurance. 5. Conclusions The results from this study suggest that switching from OCs to either the patch or ring does not necessarily improve compliance, as estimated by refill pattern, as suggested. Switching to the patch or ring may prove beneficial for women who have problems refilling their OCs; however, the magnitude of the improvement may fail to improve ultimate contraceptive efficacy by simply switching to the patch or ring. Acknowledgments Editorial assistance was provided by Melissa Lingohr-Smith of Novosys Health and was funded by Bayer HealthCare. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.contraception.2014.04.005. References
Fig. 4. Estimated percentage of timely refills before and after switching from OCs to the contraceptive ring by percentage of timely refills while on OCs (in 20% intervals) after adjusting for other covariates.
[1] Finer LB, Zolna MR. Unintended pregnancy in the United States: incidence and disparities, 2006. Contraception 2011;84:478–85 [PubMed: 22018121]. [2] Finer LB, Henshaw SK. Disparities in rates of unintended pregnancy in the United States, 1994 and 2001. Perspect Sex Reprod Health 2006;38:90–6 [PubMed: 16772190]. [3] Trussell J. Contraceptive failure in the United States. Contraception 2011;83:397–404 [PubMed: 21477680]. [4] Potter L, Oakley D, de Leon-Wong E, Canamar R. Measuring compliance among oral contraceptive users. Fam Plann Perspect 1996;28:154–8 [PubMed: 8853280]. [5] Pierson RA, Archer DF, Moreau M, Shangold GA, Fisher AC, Creasy GW. Ortho Evra(tm)/Evra(tm) versus oral contraceptives: follicular development and ovulation in normal cycles and after an intentional dosing error. Fertil Steril 2003;80:34–42 [PubMed: 12849799]. [6] U.S. selected practice recommendations for contraceptive use, 2013: adapted from the World Health Organization Selected Practice
A. Law et al. / Contraception xx (2014) xxx–xxx
[7]
[8]
[9]
[10]
[11]
Recommendations for Contraceptive Use, 2nd Edition. Morb Mortal Wkly Rep 2013;62(RR05):1–6 [Available at: http://www.cdc.gov/ mmwr/preview/mmwrhtml/rr6205a1.htm?s_cid=rr6205a1_w]. Lopez LM, Grimes DA, Gallo MF, Stockton LL, Shulz KF. Skin patch and vaginal ring versus combined oral contraceptives for contraception. Cochrane Database Syst Rev 2013(4):CD003552 [PubMed:23633314]. Steiner JF, Prochazka AV. The assessment of refill compliance using pharmacy records: methods, validity, and applications. J Clin Epidemiol 1997;50:105–16 [PubMed: 9048695]. Pittman ME, Secura GM, Allsworth JE, Homco JB, Madden T, Peipert JF. Understanding prescription adherence: pharmacy claims data from the contraceptive CHOICE Project. Contraception 2011;83:340–5 [PubMed:21397092]. Xu X, Macaluso M, Ouyang L, Kulczycki A, Grosse SD. Revival of the intrauterine device: increased insertions among US women with employer-sponsored insurance, 2002-2008. Contraception 2012;85:155–9 [PubMed:22067778]. Bonafede MM, Johnson BH, Wenten M, Watson C. Treatment patterns in disease-modifying therapy for patients with multiple sclerosis in the United States. Clin Ther 2013;35:1501–15 [PubMed:24139422].
7
[12] Nelson AL, Westhoff C, Schnare SM. Real-world patterns of prescription refills for branded hormonal contraceptives. Obstet Gynecol 2008;112:782–7 [PubMed:18827120]. [13] Kozma CM, Dickson M, Phillips AL, Meletiche DM. Medication possession ratio: implications of using fixed and variable observation periods in assessing adherence with disease-modifying drugs in patients with multiple sclerosis. Patient Prefer Adherence 2013;7:509–16 [PubMed:23807840]. [14] Ingersoll KS, Cohen J. The impact of medication regimen factors on adherence to chronic treatment: a review of the literature. J Behav Med 2008;31:213–24 [PubMed: 2868342]. [15] Audet MC, Moreau M, Koltun WD, Waldbaum AS, Shagold G, Fisher AC, et al. Evaluation of contraceptive efficacy and cycle control of a transdermal contraceptive patch vs an oral contraceptive: a randomized controlled trial. JAMA 2001;285:2347–54 [PubMed: 11343482]. [16] Archer DF, Cullins V, Creasy GW, Fisher AC. The impact of improved compliance with a weekly contraceptive transdermal system (Ortho Evra) on contraceptive efficacy. Contraception 2004;69:189–95 [PubMed: 14969665]. [17] Huybrechts KF, Ishak KJ, Caro JJ. Assessment of compliance with osteoporosis treatment and its consequences in a managed care population. Bone 2006;38:922–8 [PubMed:16330270].
