Combined Oral Contraceptive Side Effects: Incidence Rates Across Clinical Trials

Why Incidence Numbers Matter More Than Labels

Package inserts list side effects in broad categories. They do not tell you that nausea peaks at week two of the first pack and then drops substantially for most women. They do not tell you that your risk of a blood clot depends heavily on which progestin is in your pill.

Numbers matter. A clinician can tell you "this pill carries a very low absolute VTE risk" and be technically correct while leaving you without the context to weigh that against your personal history. This article pulls trial-level data so you have actual percentages, not adjectives.

How COC Trials Are Designed and Why That Limits the Data

Most key trials for individual COC formulations enroll 1,000 to 3,000 women and run for 13 menstrual cycles. They are designed primarily to establish contraceptive efficacy (Pearl Index), not to characterize rare adverse events. The FDA's combined hormonal contraceptive guidance reflects this reality: post-market surveillance and pooled analyses are the main source of safety signal data for low-frequency events.

Women have been historically under-represented in cardiovascular safety substudies of these trials, and women of color remain under-represented in key efficacy studies. When you see incidence percentages below, note whether the source is a manufacturer-sponsored Phase III trial (mostly young, healthy, screened women), a pharmacoepidemiological database (broader population, more confounders), or a meta-analysis.

Common Side Effects: Incidence by System

Nausea and Gastrointestinal Effects

Nausea is the side effect women most commonly cite in the first one to two months. Across pooled data from key trials of 20 to 35 mcg ethinyl estradiol (EE) formulations, nausea incidence ranges from approximately 5 to 20 percent in cycle 1, with most women experiencing resolution by cycle 3 without stopping the pill.

Lower EE doses (10 to 20 mcg) are associated with less nausea but higher rates of unscheduled bleeding. Taking the pill with food or at bedtime reduces symptomatic nausea for most users.

Vomiting is less common, reported in roughly 1 to 5 percent of users in short-term trials. Persistent nausea beyond cycle 3 warrants a formulation review.

Unscheduled Bleeding and Spotting

Breakthrough bleeding (BTB) is the most common reason women discontinue COCs in the first three months. A pooled analysis of 18 Phase III trials found unscheduled bleeding or spotting in 10 to 30 percent of women during cycle 1, falling to under 10 percent by cycle 3 for most formulations.

Continuous or extended-cycle regimens (e.g., 84/7 or 365-day dosing) carry higher BTB rates in the first three months, typically 20 to 40 percent, before stabilizing. A trial of the 90 mcg levonorgestrel/20 mcg EE extended-cycle pill reported 40 percent of women experiencing spotting in the first 91-day cycle, dropping to approximately 15 percent by the third cycle.

Breast Tenderness

Cyclic breast tenderness is reported in approximately 5 to 11 percent of women across key trials. It is more common with higher EE doses (35 mcg) and tends to resolve within two to three cycles. Women in the late perimenopause who start or restart COCs for cycle regulation may notice more prominent breast changes because baseline breast density is already in flux.

Headache

Headache is listed as common in virtually every COC label, with trial incidence ranging from 6 to 18 percent depending on the formulation and the baseline rate in the placebo or comparison arm. The distinction that matters clinically is between tension-type headache (no contraindication to COC use) and migraine with aura (an absolute contraindication per WHO Medical Eligibility Criteria, Category 4).

Women who experience new-onset visual aura after starting a COC should stop the pill and seek evaluation the same day.

Mood Changes and Depression

This is an area where sex-specific data exists and the conversation in trials has historically been inadequate. A large Danish cohort study published in JAMA Psychiatry (2016) by Skovlund et al. followed more than one million women and found that COC users had a relative risk of first antidepressant use of 1.23 (95% CI 1.22 to 1.25) compared to never-users, with the highest association in adolescents (RR 1.80 for ages 15 to 19). This is an observational finding with confounders including the bidirectional relationship between hormonal contraception initiation and social/relationship stress, but it is a real signal that should not be dismissed.

Manufacturer trial data for mood outcomes is inconsistent. Most key trials used general well-being questionnaires rather than validated depression instruments. Women with a personal history of depression or premenstrual dysphoric disorder (PMDD) deserve a specific conversation before starting a COC because mood sensitivity to exogenous progestins varies considerably.

The WomanRx Side-Effect Timing Framework for COCs:

| Time Point | What to Expect | Action Threshold | |---|---|---| | Weeks 1 to 4 (cycle 1) | Nausea, breast tenderness, BTB in up to 30% | Watchful waiting unless severe | | Cycles 2 to 3 | Most GI and breast symptoms resolve | If unchanged, consider lower EE dose | | Cycle 4 and beyond | Baseline tolerated side-effect profile | Persistent BTB or mood change warrants formulation switch | | Any time | New visual aura, leg swelling, chest pain | Stop pill, seek care the same day |

Venous Thromboembolism: The Risk That Needs Real Numbers

VTE is the most discussed serious adverse event with COC use. The absolute risk is low but real, and it varies significantly by progestin generation. Every woman deserves to hear the numbers, not just "rare."

Baseline Population Risk

Non-pill-using women of reproductive age have a VTE incidence of approximately 2 per 10,000 woman-years. Pregnancy raises that to 29 per 10,000 woman-years, and the postpartum period to 300 to 400 per 10,000 woman-years in the first six weeks.

COC VTE Risk by Progestin Type

A 2012 BMJ meta-analysis by Stegeman et al. and updated pharmacoepidemiological data from the EMA's 2013 CHMP review consistently show:

• Levonorgestrel (2nd generation), norethindrone: approximately 5 to 7 per 10,000 woman-years
• Desogestrel, gestodene (3rd generation): approximately 9 to 12 per 10,000 woman-years
• Drospirenone (4th generation): approximately 9 to 12 per 10,000 woman-years
• Norgestimate: approximately 5 to 7 per 10,000 woman-years

The FDA updated its labeling guidance in 2012 to reflect that some progestins carry approximately double the VTE risk of levonorgestrel. The absolute increase is still small, roughly 4 to 5 extra events per 10,000 women per year, but for a woman with factor V Leiden or a family history of unprovoked DVT, that difference is clinically meaningful.

Who Has Higher Absolute Risk

Thrombophilia (Factor V Leiden, prothrombin gene mutation, protein C/S deficiency) multiplies the baseline COC VTE risk by 10 to 35 times. ACOG Practice Bulletin 206 on thrombophilia recommends against routine thrombophilia screening before COC initiation in asymptomatic women without a personal or strong family history, but targeted testing is appropriate when clinical suspicion is present.

Obesity (BMI >30) roughly doubles VTE risk independent of COC use, and the two risks are additive. Smoking over age 35 is a Category 4 contraindication per WHO MEC.

Arterial Events: Stroke and Myocardial Infarction

Ischemic Stroke

Women with migraine with aura have an approximately threefold elevated baseline ischemic stroke risk. COC use in the presence of aura migraine increases that risk further. The WHO Category 4 restriction (benefits do not outweigh risks) applies to any COC in women with current migraine with aura regardless of age.

Population-level absolute risk of ischemic stroke attributable to COC use in healthy, non-smoking women under 35 without migraine remains very low: approximately 4.4 attributable cases per 100,000 woman-years in a large Danish study by Lidegaard et al. Published in the BMJ.

Myocardial Infarction

COC-associated MI risk is extremely low in young non-smoking normotensive women. Risk increases substantially with smoking, hypertension, and diabetes. A Cochrane review by Baillargeon et al. found that the absolute excess MI risk attributable to low-dose COC use in healthy women is fewer than 1 case per 100,000 woman-years.

Effects on Blood Pressure

COCs raise systolic blood pressure by an average of 1 to 5 mmHg in most users, a modest effect. In approximately 1 to 5 percent of users, a clinically significant rise (>140/90 mmHg) occurs. Blood pressure should be checked before initiation and at the three-month visit.

WHO MEC Category 3 (risks generally outweigh benefits) applies to women with adequately controlled hypertension; Category 4 applies to women with systolic ≥160 mmHg or diastolic ≥100 mmHg.

Effects on Weight

Weight gain is one of the most frequently cited fears about the pill. The evidence does not consistently support a meaningful causal effect. A Cochrane review by Gallo et al. (2014) found no evidence from randomized trials that COC use causes significant weight gain compared to non-hormonal methods. Mean weight change across trials was <1 kg.

That does not mean no individual woman gains weight on the pill. Fluid retention from the estrogen component and appetite changes from the progestin component are real. Women who notice sustained weight gain beyond the first two cycles deserve a formulation review rather than dismissal.

Effects on Libido and Sexual Function

Female sexual dysfunction is under-studied in COC trials. The pill reduces free testosterone by increasing sex hormone-binding globulin (SHBG), and SHBG levels may remain elevated for months after stopping the pill. A study in the Journal of Sexual Medicine found that women taking COCs had significantly higher SHBG and lower free androgen index than controls, with some data suggesting SHBG elevation persists six months after discontinuation.

Decreased libido and vaginal dryness are not consistently captured in key trials because validated sexual function instruments were not routinely administered. The honest answer is that the evidence gap here is significant: we know the mechanism by which COCs could reduce libido, but trial-level frequency data is sparse.

Cancer Risk: Cervical, Breast, and Colorectal

Cervical Cancer

Long-term COC use (>5 years) is associated with a modestly increased risk of cervical cancer in women positive for HPV. A 2007 Lancet meta-analysis by Moreno et al. found the relative risk of cervical cancer with 5 or more years of COC use was approximately 1.90 (95% CI 1.69 to 2.13). The absolute excess risk is small and decreases after discontinuation. This does not change the recommendation to use COCs for eligible women, but it reinforces the importance of HPV vaccination and regular cervical screening.

Breast Cancer

The relationship between COCs and breast cancer remains one of the most debated in women's health. A 2017 New England Journal of Medicine study by Morch et al. followed 1.8 million Danish women and found a relative risk of breast cancer of 1.20 (95% CI 1.14 to 1.26) for current or recent COC users compared to never-users. The absolute excess was approximately 13 extra breast cancers per 100,000 women per year.

For context, at age 20 to 29 the baseline annual breast cancer incidence is extremely low, so a relative risk of 1.20 translates to a very small absolute number. For a 45-year-old perimenopausal woman using a low-dose COC for cycle regulation, the absolute risk is somewhat larger and warrants individualized discussion.

Colorectal Cancer

COC use is associated with a modest reduction in colorectal cancer risk. A meta-analysis in the American Journal of Gastroenterology found an approximately 18 percent reduction in risk with ever-use. This protective effect does not offset the importance of age-appropriate colorectal cancer screening.

Pregnancy, Postpartum, and Lactation

This section is mandatory reading if you are pregnant, recently postpartum, or breastfeeding.

During Pregnancy

COCs are contraindicated in confirmed pregnancy. Ethinyl estradiol is not classified under the old FDA letter system for drugs approved before 2001, but post-marketing data have not confirmed teratogenicity from first-trimester exposure. The ACOG position is that inadvertent COC exposure in early pregnancy does not require pregnancy termination based on drug safety alone, but the pill should be stopped immediately upon pregnancy confirmation.

Postpartum and Lactation

Estrogen-containing contraceptives suppress milk production by inhibiting prolactin activity. WHO MEC assigns Category 4 (do not use) to COCs in the first 21 days postpartum for all women regardless of breastfeeding status, due to the markedly elevated postpartum VTE risk. From 21 to 42 days postpartum, Category 3 applies if additional VTE risk factors are present.

For breastfeeding women, progestin-only methods (POP, implant, hormonal IUD) are preferred because they do not affect milk supply and carry Category 1 or 2 ratings throughout lactation per WHO MEC.

If a breastfeeding woman chooses to use a COC after 42 days postpartum, she should be counseled that milk volume may decrease and infant exposure to EE via breast milk, while low, is not zero. A study in Contraception found EE transfer into breast milk at approximately 0.02 percent of the maternal dose, well below levels expected to cause harm, but current guidance still favors non-estrogen methods during lactation.

Perimenopause

Women in perimenopause using low-dose COCs for cycle regulation, contraception, or vasomotor symptom management carry a higher absolute cardiovascular baseline risk than women in their 20s. The same relative VTE risk from the progestin translates to a higher absolute number at age 48 than at age 25. ACOG Practice Bulletin 110 on noncontraceptive uses of hormonal contraceptives recommends that perimenopausal women without contraindications may use COCs through menopause, but blood pressure and lipids should be rechecked annually.

Who This Is Right For and Who Should Avoid It

Women who are generally good candidates:

• Reproductive-age women without contraindications seeking reliable contraception (typical-use failure rate of approximately 9 percent per year)
• Women with PCOS seeking cycle regulation, androgen suppression for acne or hirsutism, or endometrial protection
• Women with endometriosis using continuous COC to suppress lesions
• Women with heavy menstrual bleeding or dysmenorrhea (COCs reduce menstrual blood loss by approximately 40 to 50 percent)
• Perimenopausal women without cardiovascular contraindications who need contraception and cycle control

Women who should use an alternative or avoid COCs entirely (WHO MEC Category 3 or 4):

• Migraine with aura at any age
• Current or personal history of VTE or PE
• Known thrombophilia
• Hypertension ≥160/100 mmHg
• Smokers aged 35 or older
• <21 days postpartum
• Active or recent breast cancer
• Breastfeeding women <6 weeks postpartum
• Women with diabetes with vascular complications

FAERS and Post-Market Signals

The FDA Adverse Event Reporting System (FAERS) reflects real-world use, not controlled trial conditions, and suffers from under-reporting and selection bias. The top adverse event categories reported for COC products in FAERS align with trial data: nausea, headache, mood changes, and breakthrough bleeding dominate. Post-market pharmacovigilance added the drospirenone VTE signal in the mid-2000s, leading to the 2012 FDA label update requiring progestin-specific VTE risk language.

Post-market data has also generated signals around depression, particularly in adolescents, that were not captured in pre-approval trials. The Skovlund 2016 Danish cohort remains the most cited evidence for this association, with a relative risk of first antidepressant prescription of 1.80 in adolescents aged 15 to 19.

Evidence Gaps: What We Do Not Know

Women have been under-represented in COC safety substudies in three consistent ways. First, most key trials explicitly exclude women with relevant comorbidities (hypertension, diabetes, BMI over 35), so the data do not reflect the real-world population. Second, adolescent-specific safety data beyond the Skovlund cohort is sparse for most formulations. Third, women of color have participated in roughly 10 to 20 percent of the enrollment in key US-based Phase III trials despite being disproportionately affected by some comorbidities that influence COC risk.

Where data in women of color, adolescents, or women with managed chronic conditions is extrapolated from predominantly white, healthy trial populations, that extrapolation should be disclosed. We have done so throughout this article.