Secondary Logo

Journal Logo

Original Research

Treatment of Unscheduled Bleeding in Continuous Oral Contraceptive Users With Doxycycline

A Randomized Controlled Trial

Kaneshiro, Bliss MD, MPH; Edelman, Alison MD, MPH; Carlson, Nichole PhD; Morgan, Kristin; Nichols, Mark MD; Jensen, Jeffrey MD, MPH

Author Information
doi: 10.1097/AOG.0b013e3181e0119c
  • Free

Continuous administration of combined oral contraceptives initially gained legitimacy as a treatment for menstrual-related disorders. In recent years, continuous administration of combined oral contraceptives has gained popularity as an effective and safe contraceptive method that allows a woman to control when and if she has a menstrual period. Compared with the traditional cyclic regimen, continuous administration decreases the overall number of scheduled bleeding days.1 However, it is associated with unscheduled bleeding and spotting.2 Rates of unscheduled bleeding are highest in early cycles and gradually decrease over time. In evaluation of bleeding patterns in women using continuously dosed oral contraceptive pills, Miller and Hughes1 reported that the complete absence of bleeding and spotting was achieved in only 16% of continuous-use participants during the first 3 months of treatment. This represents a major nuisance to women and may lead to method discontinuation as well as unplanned pregnancy.

Although the molecular mechanisms that regulate endometrial bleeding are not completely understood, matrix metalloproteinases (MMPs), a family of zinc-dependent proteases that degrade the extracellular matrix, are known to play a role.3 They are found locally in the endometrium and their levels of expression have been studied in menstruation,3 dysfunctional uterine bleeding,4 and the unscheduled bleeding associated with progestin-only contraceptives.5,6 Progesterone has been shown to regulate MMP activity with increased expression of MMP-3 and MMP-9 associated with the levonorgestrel intrauterine system, subdermal levonorgestrel, and depo-medroxyprogesterone acetate exposure.6–9 Up-regulation of MMPs has been associated with unscheduled bleeding, and inhibition is believed to result in decreased endometrial bleeding.4,9

In the 1980s, it was discovered that the common antibiotic doxycycline is also a potent inhibitor of MMPs.10 Matrix metalloproteinases inhibition with doxycycline has been previously studied in the treatment of unscheduled bleeding in women using an etonorgestrel contraceptive implant and successfully reduced the duration of bleeding.11 To estimate whether MMP inhibition could reduce unscheduled bleeding in women initiating continuous oral contraceptive pills, we designed a study to treat women with doxycycline or placebo at the onset of bleeding episodes. The primary objective of this study was to estimate whether doxycycline would decrease unscheduled bleeding and spotting in women using continuous oral contraceptive pills through MMP inhibition. Secondary objectives were to describe bleeding patterns, safety and side effects, and satisfaction with doxycycline treatment.


A randomized, double-blind study was conducted at Oregon Health & Science University between June 2007 and March 2009. The protocol was approved by the Institutional Review Board at Oregon Health & Science University. Recruitment was done through radio, television, and print advertisements. Women were invited to participate if they were between the ages of 18 and 45 years, were in good health, and had no medical contraindications to continuous oral contraceptive pill therapy. To avoid the common transition bleeding seen with combined hormonal contraceptive initiation, all participants were required to have been on cyclic hormonal contraceptives (combined oral contraceptive pills, the contraceptive patch, or the contraceptive ring) without unscheduled vaginal bleeding before enrollment. Potential participants were excluded if they had an intrauterine device or contraceptive implant in place, had used depot medroxyprogesterone acetate within 9 months, were smokers, or had a hypersensitivity to any of the tetracyclines. All participants underwent written informed consent.

The study flow is outlined in Figure 1. All women enrolled in the study took a 20-microgram ethinyl/90-microgram levonorgestrel pill (Lybrel; Wyeth Pharmaceuticals, Madison, NJ) dosed in a continuous fashion over four 28-day cycles (112 days of hormonally active pills). Participants were randomly assinged to either doxycycline 100 mg or placebo taken orally twice a day for 5 days starting on the first day of unscheduled bleeding. If another bleeding or spotting episode occurred more than 7 days after completion of treatment with the study medication (doxycycline or placebo), the 5-day regimen was repeated. To determine whether unscheduled bleeding was eliminated or just postponed, the study medication was discontinued after the first 84 days, and for the remaining 28 days, bleeding patterns were observed on the continuous oral contraceptive pill alone.

Fig. 1.
Fig. 1.:
Study flow.Kaneshiro. Doxycycline and Unscheduled Bleeding. Obstet Gynecol 2010.

Participants who met eligibility criteria based on a telephone screen underwent an in-person visit that consisted of informed consent, demographic and medical history taking, and physical examination including a blood pressure check, pelvic examination, cervical cytologic evaluation, and testing for gonorrhea and chlamydia. If participants had documented cervical cytology in the previous 6 months, cervical cytology was not repeated. Participants were asked to continue on their current cyclic contraceptive method until they returned for their second study visit.

The second visit was scheduled between days 1 and 5 of their next menstrual period provided they had a negative screen for gonorrhea and chlamydia and cervical cytology that did not require colposcopy or an excisional procedure according to the American Society for Colposcopy and Cervical Pathology guidelines.11 After confirmation of a negative urine pregnancy test result, participants were block randomized using a computer-generated randomization scheme. To conceal treatment groups to investigators and participants, the Oregon Health & Science University Research Pharmacy packaged indistinguishable gel capsules containing either doxycycline or placebo into identical bottles labeled only with instructions and the participant identifier. Individuals received a 2-month supply of contraceptive pills, study medication, and a study calendar to document daily bleeding events, pill compliance, and adverse effects on a daily basis. Bleeding was recorded daily as bleeding, spotting, or no bleeding or spotting based on criteria recommended by Mishell et al.12 Spotting was defined as minimal blood loss that did not require new use of any type of sanitary protection, including panty liners. Bleeding was defined as evidence of blood loss that required the use of sanitary protection with a tampon, pad, or panty liner. Bleeding episodes were defined as days of bleeding or spotting that were bounded by 2 days of no bleeding or spotting. In accordance with the criteria described by Mishell et al,12 all bleeding in this study was defined as unscheduled.

Participants recorded headache, nausea, acne, mood/depression, menstrual pain, vaginal candidiasis (yes/no), and any other notable event on a daily basis. They were also asked to make a daily notation on their calendars (yes/no) of whether they took their contraceptive pill and study medication. If an oral contraceptive pill was missed, they were instructed to take this pill as soon as they remembered and to note the date and time on their calendars.

Although the only scheduled pregnancy test took place at the second study visit, if a patient had a suspicion of pregnancy during the study period, a urine pregnancy test was repeated. A follow-up telephone call was made 2 weeks after randomization, and follow-up study visits took place every month for the duration of the study. At the end of 2 months, participants received another 2-month supply of contraceptive pills and a 1-month supply of study medication. At the end of 3 months, when they stopped the study medication, participants completed a satisfaction questionnaire. At the end of 4 months, at the time of study completion, participants completed a second satisfaction questionnaire. Satisfaction with bleeding pattern and treatment was assessed using a 100-mm visual analog scale.

The primary outcome was the number of bleeding and spotting days for the first 84 days of the study. A two-sample t test was used to compare the mean number of bleeding and spotting days during the first 84 days between the two groups. To understand whether the bleeding was just postponed, additional secondary analyses comparing the mean number of bleeding and spotting days for the last 28 days and the total study period (112 days) were also compared using two-sample t tests and repeated measures analysis of variance. To address distributional assumptions, an appropriate transformation was applied to bleeding data so that parametric tests could be used, and normality was verified with P-P plots.

Analyses were performed on the basis of intention to treat. In the event of incomplete data resulting from study discontinuation, the proportion of days with each bleeding pattern (absence of bleeding/spotting, spotting, bleeding) was calculated by dividing the total number of bleeding, spotting, or absence of bleeding/spotting days by the number of women-days of participation. The proportions were then multiplied by 28 to obtain the expected number of days of each bleeding pattern during the 28-day interval in which they discontinued the study. Demographics, satisfaction, compliance with pills, and medication-associated adverse effects were compared between groups using χ2 or Fisher exact tests for categorical variables and t tests for continuous factors. Statistical analyses were performed using the Statistical Program for Social Sciences (SPSS 16.0 for Windows; SPSS Inc, Chicago, IL).

Our sample size of 66 (33 in each group) was based on expected differences in the primary outcome of total bleeding and spotting days. Previous studies documented a mean of 19.2 days (standard deviation, 12.01) of bleeding and spotting over the first 90 days when a 20-microgram levonorgestrel/100-microgram ethinyl pill was administered in a continuous fashion.13 The sample size was calculated to detect a 50% reduction in bleeding (β=0.80, α=0.05) and accounted for a 30% dropout rate.


Of the 70 participants who met inclusion criteria after a telephone screen, 66 were randomly assinged and 60 completed all four cycles of the study. Reasons for study discontinuation and when it occurred are illustrated in Figure 2. Demographic characteristics were similar between the study groups (Table 1). Most participants were white, in their late 20s, and nulliparous and were using cyclic oral contraceptive pills before randomization.

Fig. 2.
Fig. 2.:
Reasons for study discontinuation.Kaneshiro. Doxycycline and Unscheduled Bleeding. Obstet Gynecol 2010.
Table 1
Table 1:
Demographic and Menstrual Characteristics of the Study Population at Randomization

The mean number of bleeding and spotting days for the first 84 days was 23.06 (standard error, 3.47) days for the doxycycline group and 19.24 (3.42) for the placebo group (P =.32; Table 2). The mean number of bleeding-only days was 8.74 (2.21) for the doxycycline group and 6.48 (1.64) for the placebo group. The days of bleeding or spotting or both for each of the 28-day cycles are depicted graphically in Figure 3. There were no significant differences for each of the 28-day cycles, for the 84-day treatment interval, or over the entire 112-day study period. A repeated measures analysis of variance showed that the mean number of days of bleeding did not differ between treatment groups (P=.44) nor did bleeding patterns differ between treatment groups over time (P=.83). However, the analysis did indicate that the number of days of bleeding increased between the first and second months (P=.037). Bleeding patterns between groups were identical in the posttreatment cycle.

Table 2
Table 2:
Mean and Median Days of Bleeding and Bleeding or Spotting or Both for the First 84 Days, Last 28-Day Cycles, and All 112 Days of the Study Period
Fig. 3.
Fig. 3.:
Average number of days of bleeding or spotting or both during each 28-day cycle. Participants received either doxycycline (100 mg twice daily) or placebo for 5 days during bleeding/spotting episodes during cycles 1–3. There was no treatment during cycle 4. There were no statistically significant differences in bleeding/spotting days between groups.Kaneshiro. Doxycycline and Unscheduled Bleeding. Obstet Gynecol 2010.

Both groups used a similar number of courses of study medication (doxycycline 2.64 [1.88], placebo 2.55 [2.08], P=.87). There were no statistically significant differences in the reported number of missed oral contraceptive pills during the first 84 days of the study period between the doxycycline and placebo groups (P=.92; Table 3). In terms of compliance with the study medication, participants did not always treat bleeding/spotting episodes with the study medication, as is demonstrated in the discrepancy between the number of bleeding and spotting episodes and the number of courses of study medication.

Table 3
Table 3:
Bleeding and Spotting Episodes, Study Medication Administration, and Missed Oral Contraceptive Pills for the First 84 Days of the Study Period

There was no difference in the duration of the longest bleeding/spotting episode between the two study groups in the first 84 days of the study period (doxycycline 9.42 [2.41], placebo 8.90 [1.96], P=.70) and the last 28 days of the study period (doxycycline 5.35 [1.60], placebo 5.71 [1.70], P=.95; Table 3). Although most women in both groups experienced between one and three bleeding episodes during the study period, women in the placebo group were more likely to experience no bleeding (doxycycline 3.0% compared with placebo 18.2%) and were also more likely to experience more than four bleeding episodes (doxycycline 24.2% compared with placebo 36.4%; P=.04).

During the first 84 days, there were no differences in the side effect profiles between the doxycycline and the placebo groups including headache (doxycycline 63.6% compared with placebo 69.7%, P=.83), nausea (doxycycline 63.6% compared with placebo 51.5%, P=.54), acne (doxycycline 48.8 compared with placebo 38.7%, P=.54), mood complaints (doxycycline 57.6% compared with placebo 40.7%, P=.29), pain with menses (doxycycline 72.7% compared with placebo 54.8%, P=.12), or vaginal candidiasis (doxycycline 15.2% compared with placebo 9.7%, P=.58). No major adverse outcomes were reported in either group during the treatment and posttreatment cycles. Of the two participants who discontinued the study secondary to medication side effects, one participant in the placebo group discontinued because of nausea and one participant in the doxycycline group discontinued because of moodiness and acne.

After 84 days of using a continuous combined oral contraceptive with the study medication, participants in the doxycycline and placebo groups reported similar satisfaction with bleeding patterns using a 100-point visual analog scale (Table 4). Although it appeared that more participants in the placebo group (54.8%) preferred to take continuous contraceptives than in the doxycycline group (37.9%), these results were not statistically different. After 112 days of taking a continuous combined oral contraceptive pill, a majority of participants in both groups reported that they planned to continue taking oral contraceptive pills (doxycycline 69.0% compared with placebo 87.1%, P=.08), with most participants reporting that they planned to use continuous administration (doxycycline 37.9% compared with placebo 54.8%, P=.25). At the end of the study period, participants were asked whether they preferred to take a medication every day to prevent bleeding or preferred to take a medication to treat bleeding if it started. Notably, most participants, regardless of allocation, reported that they would have preferred to take a medication to prevent bleeding (doxycycline 69.0% compared with placebo 87.1%, P=.13).

Table 4
Table 4:
Midstudy and Final Participant Satisfaction Results Using a 100-Point Visual Analog Scale


Continuous administration of combined oral contraceptive pills is becoming increasingly popular as a contraceptive method. For most women, amenorrhea is a goal of therapy and unscheduled bleeding is the primary reason for dissatisfaction and discontinuation of the approach. Because MMPs are known to play a role in endometrial degradation,3–6 we tested the hypothesis that an MMP inhibitor could reduce this bleeding. Results from our randomized, blinded, placebo-controlled study demonstrate that doxycycline, initiated at the onset of unscheduled bleeding or spotting, is not an effective treatment and does not decrease the duration of unscheduled bleeding/spotting or the number of bleeding or spotting episodes.

It is notable that the mean number of days of bleeding for the placebo group was actually lower than that of the doxycycline group and the lower bound (−5.95 days) of the 95% confidence interval (CI) did not include our ad hoc definition (9-day reduction) of a clinically meaningful difference. In fact, these data rule out the possibility of the placebo group having more than 6 additional days of bleeding compared with the doxycycline group based on the bounds of the CI. The fact that the CI does not include any clinically meaningful differences indicates that, although this study was insignificant, it was not underpowered. The amount of unscheduled bleeding reported in this study was higher than has been reported in other studies of continuous oral contraceptives1,13 but similar to the amount reported in a study that used the same oral contraceptive formulation we used,14 suggesting external validity to our findings.

There were small changes in the bleeding patterns of participants in the two groups that might be explained by treatment effects. Participants in the placebo group were more likely to have more than four episodes of bleeding, but the doxycycline group tended to experience longer episodes. There was a nonsignificant trend toward more bleeding in the doxycycline group women during each treatment cycle, but not in the posttreatment cycle. Taken together, these results suggest that inhibition of MMPs, once bleeding has started, may have a destabilizing effect on the endometrium in women using continuous oral contraceptives. However, the overall treatment effect of doxycycline on unscheduled bleeding, the number of bleeding and spotting days, and the length of the longest bleeding episodes were not significantly different from placebo. Any alterations in bleeding in the treatment group are unlikely to be related to the absorption of oral contraceptive steroids, as pharmacokinetic studies have demonstrated that concomitant use of doxycycline and other tetracyclines do not alter oral contraceptive steroid levels.15,16

Not all women who initiate therapy with continuous oral contraceptives experience problematic bleeding, and our clinical experience suggests that there is a susceptible phenotype that will never achieve amenorrhea. Our current understanding of the mechanism of endometrial changes induced by contraceptive steroids does not allow us to predict how an individual woman will respond to continuous or cyclic use. Although we randomized participants to treatment group and there were no significant differences in baseline characteristics, differences reflecting endometrial response appear to have existed between the groups. There were significantly more participants in the placebo group who experienced no bleeding or spotting during the study period. This pattern is likely reflective of an underlying endometrial bleeding propensity rather than a treatment effect of the study medication, as participants who did not have unscheduled bleeding did not begin treatment with the study medication. However, when participants who never experienced unscheduled bleeding during the study are excluded from the analysis, a treatment benefit from doxycycline is still not observed.

The dosage and administration pattern of doxycycline used in this study was based on a study by Weisberg et al in which doxycycline 100 mg administered twice a day for 5 days modestly decreased unscheduled bleeding in etonorgestrel implant users.17 An endometrial biopsy study demonstrated that MMP-2 and MMP-9 were inhibited with doses of doxycycline as low as 20 mg administered twice a day in etonorgestrel users.18 It is possible that a different dose or dosing frequency is required for MMP inhibition in the endometrium of users of combined contraceptive products that contain estrogen, as this hormonal milieu may have a different effect on MMP expression in the endometrium. It is also possible that MMP inhibition needs to occur before the onset of unscheduled bleeding (eg, when MMP levels are low) and that once endometrial degradation begins, it is not possible to halt the cascade. In an animal study exploring MMP activity and inhibition, doxycycline was given at the time of progesterone withdrawal, and although MMP activity was inhibited in the endometrium of mice, this did not correlate with less endometrial degradation.19 Although rodent studies are not directly translational to humans, this may explain why we did not see a decrease in unscheduled bleeding in our study.

The World Health Organization has recommended using 90-day reference intervals to analyze bleeding patterns that occur in women using noncyclic contraceptive methods.20 Although the 90-day reference interval makes sense for progestin-only methods, we chose to use an 84-day reference period because oral contraceptive pills, including extended and continuous formulations, are uniformly packaged in 28-day packs. Had our findings been significant, the results of this study could easily have been applied to the typical use of continuous oral contraceptive pills.

Some important limitations of this study should be noted. It has been recommended that studies exploring bleeding patterns with contraceptive use should use daily real-time electronic collection devices.12 However, as these systems are extremely expensive and subject to other problems, we used traditional paper diaries to collect bleeding data. Because we randomized women to treatment, inconsistencies in reporting and pill use should have been distributed evenly between the groups. Any bias would likely lead to a rejection of the hypothesis.

Our study participants frequently had questions about the study medication regimen including when to start and stop treatment. This was addressed by frequent telephone calls and study visits with research personnel and can be seen in our results in the discrepancy between the number of episodes of bleeding and spotting and the number of treatment courses of study medication. The difficulty that our research participants had with correct use of treatment for all bleeding episodes suggests that the approach may not have been clinically useful even if a beneficial effect was seen.

Despite a majority of women reporting some unscheduled bleeding and spotting during the study period, most women reported that they planned to continue taking combined oral contraceptive pills continuously rather than returning to cyclic dosing. This observation underscores the high acceptability of continuous administration of oral contraceptive pills. As most participants in this study reported that they would rather take a medication to prevent unscheduled bleeding than start a medication once unscheduled bleeding occurred, it is important that future studies address preventive measures rather than treatment methods for unscheduled bleeding.


1. Miller L, Hughes JP. Continuous combination oral contraceptive pills to eliminate withdrawal bleeding: a randomized trial. Obstet Gynecol 2003;101:653–61.
2. Anderson FD, Hait H. A multicenter, randomized study of an extended cycle oral contraceptive. Contraception 2003;68:89–96.
3. Salamonsen LA, Woolley DE. Matrix metalloproteinases in normal menstruation. Hum Reprod 1996;11(suppl 2):124–33.
4. Galant C, Berliere M, Dubois D, Verougstraete JC, Charles A, Lemoine P, et al. Focal expression and final activity of matrix metalloproteinases may explain irregular dysfunctional endometrial bleeding. Am J Pathol 2004;165:83–94.
5. Chegini N, Rhoton-Vlasak A, Williams RS. Expression of matrix metalloproteinase-26 and tissue inhibitor of matrix metalloproteinase-3 and -4 in endometrium throughout the normal menstrual cycle and alteration in users of levonorgestrel implants who experience irregular uterine bleeding. Fertil Steril 2003;80:564–70.
6. Vincent AJ, Zhang J, Ostor A, Rogers PA, Affandi B, Kovacs G, et al. Matrix metalloproteinase-1 and -3 and mast cells are present in the endometrium of women using progestin-only contraceptives. Hum Reprod 2000;15:123–30.
7. Vincent AJ, Malakooti N, Zhang J, Rogers PA, Affandi B, Salamonsen LA. Endometrial breakdown in women using Norplant is associated with migratory cells expressing matrix metalloproteinase-9 (gelatinase B). Hum Reprod 1999;14:807–15.
8. Salamonsen LA, Zhang J, Hampton A, Lathbury L. Regulation of matrix metalloproteinases in human endometrium. Hum Reprod 2000;15(suppl 3):112–9.
9. Marbaix E, Kokorine I, Henriet P, Donnez J, Courtoy PJ, Eeckhout Y. The expression of interstitial collagenase in human endometrium is controlled by progesterone and by oestradiol and is related to menstruation. Biochem J 1995;305(pt 3):1027–30.
10. Burns FR, Stack MS, Gray RD, Paterson CA. Inhibition of purified collagenase from alkali-burned rabbit corneas. Invest Ophthalmol Vis Sci 1989;30:1569–75.
11. Wright TC Jr, Massad LS, Dunton CJ, Spitzer M, Wilkinson EJ, Solomon D. 2006 consensus guidelines for the management of women with abnormal cervical screening tests. J Low Genit Tract Dis 2007;11:201–22.
12. Mishell DR Jr, Guillebaud J, Westhoff C, Nelson AL, Kaunitz AM, Trussell J, et al. Recommendations for standardization of data collection and analysis of bleeding in combined hormone contraceptive trials. Contraception 2007;75:11–5.
13. Edelman AB, Koontz SL, Nichols MD, Jensen JT. Continuous oral contraceptives: are bleeding patterns dependent on the hormones given? Obstet Gynecol 2006;107:657–65.
14. Archer DF, Jensen JT, Johnson JV, Borisute H, Grubb GS, Constantine GD. Evaluation of a continuous regimen of levonorgestrel/ethinyl estradiol: phase 3 study results. Contraception 2006;74:439–45.
15. Murphy AA, Zacur HA, Charache P, Burkman RT. The effect of tetracycline on levels of oral contraceptives. Am J Obstet Gynecol 1991;164(1 pt 1):28–33.
16. Neely JL, Abate M, Swinker M, D'Angio R. The effect of doxycycline on serum levels of ethinyl estradiol, norethindrone, and endogenous progesterone. Obstet Gynecol 1991;77:416–20.
17. Weisberg E, Hickey M, Palmer D, O'Connor V, Salamonsen LA, Findlay JK, et al. A pilot study to assess the effect of three short-term treatments on frequent and/or prolonged bleeding compared to placebo in women using Implanon. Hum Reprod 2006;21:295–302.
18. Zhao S, Choksuchat C, Zhao Y, Ballagh SA, Kovalevsky GA, Archer DF. Effects of doxycycline on serum and endometrial levels of MMP-2, MMP-9 and TIMP-1 in women using a levonorgestrel-releasing subcutaneous implant. Contraception 2009;79:469–78.
19. Kaitu'u TJ, Shen J, Zhang J, Morison NB, Salamonsen LA. Matrix metalloproteinases in endometrial breakdown and repair: functional significance in a mouse model. Biol Reprod 2005;73:672–80.
20. Belsey EM, Machin D, d'Arcangues C. The analysis of vaginal bleeding patterns induced by fertility regulating methods. World Health Organization Special Programme of Research, Development and Research Training in Human Reproduction. Contraception 1986;34:253–60.
© 2010 by The American College of Obstetricians and Gynecologists.