We performed a secondary analysis on the HER Salt Lake initiative, a prospective cohort study conducted by the University of Utah Division of Family Planning in partnership with Planned Parenthood of Utah (NCT02734199). ^,, The study addressed barriers to contraception by providing free access and person-centered counseling between 2015 and 2017, with survey follow-up through 2020. ^,, Contraceptive methods available to participants included: the arm implant (68 mg etonogestrel [ENG]), the LNG IUD (52 mg or 13.5 mg), the injection (150 mg/mL depot medroxyprogesterone acetate), POPs (variable formulations), combined oral contraceptives (COCs) (variable formulations), the vaginal ring (0.015 mg ethinyl estradiol and 0.120 mg ENG), the copper intrauterine device (CuIUD), condoms, or a subscription to Natural Cycles (NC). ^,, For analysis, methods were grouped as follows: combined methods included COC and ring users; progestin-only methods included implant, injection, LNG IUD, and POP users; and nonhormonal methods included CuIUD, condom, and NC users. We further grouped condom and NC users into nonhormonal behavioral methods (NHBMs). Our analysis included the 4425 participants who enrolled in the survey arm of the study. ^,, Participants completed the consent process with trained study staff prior to enrollment and could discontinue surveys at any time (IRB 0065794). Participants were followed for 3 years and could change or discontinue their method choice at any point without cost. ^,, We excluded participants who enrolled with only emergency contraception or no method, and those missing self-reported baseline mood-related symptoms.

Surveys occurred at enrollment and months 1, 3, and 6, with participants completing them online. ^,, Surveys collected detailed information on participant demographics (age, race, sexual identity, religion, education, and parity) and contraceptive continuation (“Are you still using the same method of contraception that you received at the beginning of this study?”). Additionally, participants who reported that they were no longer using the method they initiated at enrollment could identify reasons for discontinuation at each survey time point (“What are the reasons that you are no longer using the method you received at the start of the study [check all that apply]?”). Reasons included: moodiness or depression, bleeding issues (spoting, irregular, etc.), excessive bleeding, cramping, pain, breast symptoms, weight gain, weight loss, bloating, skin problems, pain during intercourse, partner complaint, intrauterine device fell out/expelled, wanted to get pregnant, had a positive pregnancy test, just wanted a different method, and a free text option for “other”. ^,,

Prospective surveys also asked about contraceptive side effects and side effect experiences. Self-reported frequency of mood-related symptoms was measured at baseline, month 1, month 3, and month 6 by asking whether patients had experienced mood-related symptoms in the past 30 days. Participants were asked about moodiness (M) and depression (D), separately, and could answer: (A) Have not had in the past 30 days; (B) Once a month; (C) A couple of days a month; (D) Once a week; (E) A couple of days a week; or (F) Every day. We created a composite variable for M/D categorized as “Never” (if a participant marked “Have not had in the past 30 days”), “Occasional” (if a participant marked “Once a month” or “A couple of days a month”), “Frequently” (if a participant marked “Once a week” or “A couple of days a week”), or “Daily” (if a participant marked “Every day”), for either variable. Additionally, we cross-tabulated the composite M/D variable between self-reported moodiness and depressive symptoms to assess symptom overlap and concordance at both enrollment and 6 months.

We used R v4.5.1 for analyses. We used descriptive statistics and chi-square tests to assess differences between grouped method types for various demographics, including self-reported experiences of moodiness and depressive symptoms before HER Salt Lake, age, sexual identity, race, religion, education, and parity.

One outcome of interest was self-reported reasons for discontinuation. Fisher’s exact tests assessed relationships between discontinuation reasons among participants who stopped their original method within the first 6 months and contraception type. We also used the same methods to assess relationships between reasons for discontinuation and individual contraceptive method types.

Another outcome of interest was discontinuation. Discontinuation was measured by asking whether participants were still using the method they received at enrollment (Yes/No) and what methods participants had used within the prior 4 weeks. Variables of interest included participant self-reported frequency of mood-related symptoms at each time point and method choice. We used mixed-effects models to assess the odds of discontinuation while controlling for the experience of M/D at each survey wave and the length of time in the study. Random intercepts were used for each participant to account for intraindividual correlations. Models were stratified by method type to assess whether self-reported M/D affected discontinuation differently among variable hormone types. We did not control for demographic variables in these models, as some contraceptive types had low counts of discontinuation at certain time points, and the inclusion of many covariates disrupted the models.