The main finding of our study is that current HC use was not significantly associated with an increased risk of attempted suicide in women of fertile age, even when considering both current and former use, and even in a high-risk population of previous suicide attempters. In addition, the current use of combined HC, in particular that of EE-containing preparations, was associated with a lower risk of attempted suicide compared to non-use of HC even after controlling for covariates in women without psychiatric disorders.

Our finding that the current use of HC is not significantly associated with an increased risk of AS in fertile-aged women contrasts with results reported by recent studies in the Nordic countries. A Danish study found a higher risk of the first AS in women aged 15–33 years using HC both currently (relative risk, RR 1.97, 95% CI 1.85–2.10) and previously (RR 3.40, 95% CI 3.11–3.71) compared with never-users [5]. Similarly, a large Swedish register-based study of women aged 15–22 years found an increased risk of suicidal behaviour (attempted or completed suicide) in users of combined oral contraceptives (HR 1.36, 95% CI 1.18–1.56) and progestin-only pills (HR 1.75, 95% CI 1.44–2.12) [6]. Further, two Swedish studies found a higher use of antidepressants among young (16–31 years, and especially in the youngest age-groups) users of progestin-only contraceptives, in particular of non-oral methods such as intrauterine systems, implants, injections, and transdermal patches, compared with users of combined hormonal contraceptives [22, 23]. A Korean study also reported positive associations between use of oral contraceptives and suicidality (ideation and attempt) in women over 20 years of age; however, the latter compared, with a retrospective design, lifetime prevalence of oral contraceptive use and period prevalence of suicidality, and did not provide any information on the type of oral contraceptives or on other types of HC [24]. One of the plausible reasons underlying these different findings is the inclusion of older women in our sample, where the prevalence of AS in general is low and who usually have a distinct profile of HC use. In this regard, it must be acknowledged that older women are more likely to use long-active reversible contraceptive methods, such as the levonorgestrel-releasing intrauterine system, which were not all included in our study, given that most of these methods are provided free-of-charge by some municipalities in Finland and can be used for up to five years. However, our results did not change in age-stratified analyses. Another possible explanation to our findings is that women with psychiatric disorders may be less likely to use HC [16], but more likely to attempt suicide. However, no associations between HC use and a higher risk of AS emerged in analyses stratified by psychiatric history, nor in the high-risk population of previous suicide attempters. While these results have clinical implications as they suggest that HC does not represent a risky prescription in this population, it remains however possible that inadequate power partly accounts for lack of a significant association, especially concerning progestin-only contraceptives, which are known to have the worst effects on mood and were in fact related to a higher proportion of suicide attempts in the current study.

The observation of a negative association between current use of EE-containing preparations and AS risk is a novel one, although driven substantially by the group of women without any psychiatric disorders. Overall, these results are in line with those of a recent Swedish register-based study conducted on a selected population of over 20,000 women with premenstrual disorder, which suggested that use of HC, and in particular of combined HC, was associated with no or a lower risk of AS in women with as well as without psychiatric disorders [14]. On the other hand, Skovlund et al. [5] found an increased risk associated with the use of all EE-containing combined oral contraceptives, but the highest risk was seen in users of patch, vaginal ring, and progestin-only products. Differently from the Danish study, we controlled our results for the current use of any psychotropic medications, including antipsychotics and anxiolytics in addition to antidepressants. This may be relevant, as we have previously found positive associations between HC use and the use of psychotropic drugs of any class [25]. The opposite findings may also reflect the effect of unknown confounding factors that cannot be easily uncovered in register-based studies. Similarly as Skovlund et al. [5] we also found a tendency for a higher AS risk with the use of progestin-only products, such as norethisterone and desogestrel containing oral contraceptives. However, in contrast with the Danish results, we did not find associations between the risk of AS and the use of vaginal ring or patch. Even though the negative impact of norethisterone on mood has been observed previously [26, 27], our results need caution, given that the use of norethisterone-only pills in Finland is rather limited (Table S1). In general, the use of progestin-only methods is not contraindicated during lactation; however, the tendency did not change after controlling for a recent delivery.

Another key finding of our study is a tendency for a lower AS risk in women using drospirenone and EE. Drospirenone has progestogenic, antimineralcorticoid and antiandrogenic activities. These characteristics are thought to contribute to its favourable influence on mood also in women suffering from premenstrual dysphoric disorder [28,29,30,31]. This finding is of relevance, as oral contraceptives containing drospirenone (and third-generation progestins such as desogestrel, etonogestrel, and gestodene) are used more commonly in Finland than in the other Nordic countries, and oral contraceptive containing drospirenone is the most used combined oral contraceptive in Finland (Table S1).

Our study has some limitations. By using registry data derived mainly from specialist healthcare, we were not able to detect less severe cases of AS that remained unknown to the healthcare system, which likely account for almost half of the self-reported AS [32]. Similarly, we were not able to distinguish between attempted suicide and other types of suicidal and self-harm behaviours (superficial self-harm, stress-relieving self-harm, cry for help, etc.), since we did not have access to individual medical records. Additionally, as HC use was defined as redeemed prescriptions rather than on its actual use monitored in clinical practice, misclassification cannot be ruled out. However, because the Social Insurance Institution of Finland does not reimburse HC, it is likely that most who purchased the drug did in fact use it. On the one hand, by using at least two redeemed HC prescriptions during the 180-day period as the criterion for current HC use, we cannot exclude a null finding driven by the inclusion of HC users in the non-user group. On the other hand, it cannot be excluded that women currently not using HC at the time of the AS (during the past 6 months) were in fact former users, who discontinued their contraceptive due to side effects. For example, Skovlund et al. [5] found a higher risk of AS in former users than in current users of HC. However, a recent study found only marginal increase in the risk estimates for depression when using never-users compared to non-users of HC as the reference category [33]. Importantly, our results did not change when splitting the HC non-user category into former and never-users. Because records of HC use are included in the Finnish register only starting from 2017, a longer follow-up period was not possible. Another limitation arises from the lack of information on the precise contents of the contraceptive preparations used, which precluded any analyses on the effect of different doses of EE. Additionally, we lacked information on the use of non-hormonal methods (e.g., copper intrauterine device, and barrier methods) as well as contraceptives obtained free-of-charge as part of municipal programs, concerning especially long-active reversible contraceptive methods (in particular the hormonal intrauterine system). Further, we managed to capture a nearly complete proportion of those women having a history of psychiatric disorders with a diagnosis from specialised healthcare, reimbursement rights and redeemed psychotropic medications, although some patients may have contacted only primary healthcare and not claimed their special reimbursement rights. Finally, we cannot exclude that the detected associations are confounded by external unaccounted factors.

Among the strengths of our study is the use of Finnish register data of proven high quality [34], and the identification of AS cases based on the diagnostic ICD-10 codes from specialised healthcare from 2018 to 2019. The nested case–control design we used produces unbiased estimates and is free from weaknesses of the ordinary case–control design. It uses correct sampling of controls that accounts the follow-up time [35, 36], and appears more straightforward than a cohort design with respect to recent, non-cumulative exposures. In addition, the control women were matched by age.

Taken together, our results convey the reassuring message to fertile-aged women seeking contraception that current HC use was not significantly associated with an increased risk of attempted suicide. At the same time, they once more stress the importance of a personalized choice of the best and safest contraceptive option, which should include the assessment of mental health status and suicide risk.