Proposed section structure
This topic is best organized around clinical context, trial design, the specific post hoc adherence question, quantitative findings, interpretation for practice, and study limitations. The article therefore follows this sequence: highlights, background, study design and analytic approach, key results, clinical interpretation, strengths and limitations, implications for practice and policy, and citation details.
Highlight
In a post hoc analysis of 11,908 pregnancies from a large randomized trial, low-dose aspirin effects on preterm delivery were not materially altered by when treatment began between 6 and 13 weeks’ gestation.
Measured adherence was high overall, with 85.5% of participants taking more than 90% of assigned tablets, and higher adherence did not appear to strengthen aspirin’s effect on preterm delivery, very preterm delivery, or perinatal mortality.
No meaningful effect modification was seen across regions in Africa, Asia, and Latin America, supporting consistency of the parent trial findings across diverse low-resource settings.
The analysis reinforces the main trial message: in this dataset, the observed aspirin effects were not dependent on modest differences in total exposure within the studied initiation window.
Background
Preterm delivery remains one of the leading causes of neonatal morbidity and mortality worldwide, with the greatest burden concentrated in low- and middle-income countries. Preventive strategies that are inexpensive, scalable, and safe in early pregnancy are therefore of major public health interest. Low-dose aspirin has long been studied in obstetrics, primarily for prevention of preeclampsia and related placental complications, but its potential role in reducing preterm birth has also drawn increasing attention.
Biologically, aspirin may influence pregnancy outcomes through inhibition of platelet thromboxane production, improvement in placental perfusion, and modulation of inflammatory pathways implicated in abnormal placentation. These mechanisms provide a plausible rationale for starting treatment early in pregnancy, during the period of trophoblastic invasion and placental vascular remodeling. This mechanistic framework has also led clinicians to ask whether the magnitude of aspirin benefit depends on how early therapy starts and how consistently patients take it.
That question has practical importance. In real-world antenatal care, especially in resource-constrained settings, women may present at different gestational ages and adherence can vary because of supply limitations, travel barriers, competing demands, side effects, or misunderstanding of preventive therapy. If aspirin’s effectiveness were strongly dependent on near-perfect adherence or very early initiation, implementation strategies would need to prioritize these factors aggressively. Conversely, if benefit is robust across a reasonable range of initiation timing and exposure, programs could focus more broadly on access and uptake.
The present study addresses that issue through a post hoc analysis of the multinational ASPIRIN trial dataset, examining whether total exposure, approximated through gestational age at treatment initiation and pill-count-based adherence, modified aspirin’s effects on preterm delivery and related outcomes.
Study design and analytic approach
This was a post hoc analysis of a randomized controlled trial conducted in hospitals in low-resource settings across Africa, Asia, and Latin America. The parent trial randomized women between 6 0/7 and 13 6/7 weeks’ gestational age to low-dose aspirin or placebo. Outcomes were available for 11,908 of 11,943 randomized women, a remarkably complete follow-up rate for a multicentre trial spanning multiple regions.
Medication was distributed in prepackaged two-week allotments, and adherence was assessed every two weeks using pill counts. Although pill counts do not guarantee ingestion, they remain a pragmatic and widely used adherence metric in large trials, particularly where electronic monitoring is not feasible.
The primary outcome for this post hoc analysis was preterm delivery. Secondary outcomes included preterm delivery before 34 weeks and perinatal mortality. The central analytic question was whether aspirin’s effect differed according to two exposure-related variables: gestational age at treatment initiation and overall adherence. Investigators estimated relative risks and 95% confidence intervals using Poisson models adjusted for site, treatment arm, and gestational age at treatment initiation. They also evaluated whether results varied by geographic region.
The median gestational age at treatment initiation was 10.1 weeks, with an interquartile range of 8.6 to 12.0 weeks, placing most participants squarely within the end of the embryonic and early placental development period that many clinicians consider mechanistically relevant for aspirin response.
Key findings
Overall adherence was high
The first notable result is simply that adherence in the trial was excellent. Fully 85.5% of mothers had greater than 90% adherence to treatment. This matters because it indicates a relatively narrow distribution of exposure, which is reassuring from a trial-conduct standpoint but also important when interpreting the interaction analyses. When most participants are highly adherent, the study is best positioned to detect large effect modification, but more subtle adherence-response gradients may be harder to identify.
Timing of initiation did not modify the aspirin effect
For each 1-week increase in gestational age at treatment initiation, the treatment risk ratio for preterm delivery was 0.97 with a 95% confidence interval of 0.93 to 1.02. This estimate is close to 1.0, and the confidence interval crosses the null comfortably, indicating no persuasive evidence that starting aspirin a week later within the 6- to 13-week window materially changed its effect on preterm delivery.
The same pattern held for more severe outcomes. For preterm delivery before 34 weeks, the corresponding relative risk was approximately 0.98 with a 95% confidence interval of 0.89 to 1.07. For perinatal mortality, it was 1.04 with a 95% confidence interval of 0.96 to 1.12. In practical terms, these estimates argue against a meaningful interaction between treatment effect and start week within the time frame studied.
This is clinically useful. Many aspirin discussions in obstetrics hinge on whether treatment must begin very early to be effective. These data suggest that, at least between 6 and 13 weeks’ gestation, modest differences in start timing did not measurably alter the direction or magnitude of aspirin’s effect on the study’s principal outcomes.
Higher measured adherence did not amplify benefit
The adherence analysis yielded similarly neutral findings. For each 5% increase in adherence, the treatment risk ratio for preterm delivery was 1.01 with a 95% confidence interval of 0.99 to 1.04. For preterm delivery before 34 weeks, it was approximately 0.99 with a 95% confidence interval of 0.95 to 1.03. For perinatal mortality, it was 1.02 with a 95% confidence interval of 0.98 to 1.06.
These estimates again cluster very near 1.0 and provide no evidence that higher adherence meaningfully strengthened aspirin’s effect. The interpretation is not that adherence is clinically irrelevant, but rather that within this trial population, where adherence was already generally high, there was no detectable exposure-response interaction for the outcomes evaluated.
No effect modification by region
The absence of effect modification by region further supports internal consistency. The trial was conducted across Africa, Asia, and Latin America, where baseline risks, health-system infrastructure, nutritional factors, infectious burden, and obstetric care patterns can differ substantially. Despite this heterogeneity, the interaction analyses did not suggest that the relation between aspirin, timing, adherence, and outcomes depended on region.
For clinicians and policy planners, this regional consistency is encouraging because it implies that the trial’s main conclusions are not being driven by a single geography or care environment.
Clinical interpretation
The major clinical message is narrow but important: in this large post hoc analysis, aspirin’s observed effects on preterm delivery, very preterm delivery, and perinatal mortality were not modified by when therapy began within the studied 6- to 13-week interval or by measured adherence as assessed by pill counts.
That conclusion should be interpreted in the context of the parent trial rather than in isolation. Post hoc analyses do not create new causal proof in the same way as a prespecified primary analysis. Instead, they test whether the original treatment effect appears to vary across patient or treatment-exposure subgroups. Here, the answer appears to be no.
For front-line antenatal care, especially in low-resource settings, the results are somewhat reassuring. They suggest that clinicians do not need to assume that women who start aspirin at 11 or 12 weeks rather than 7 or 8 weeks will necessarily lose benefit, at least regarding the outcomes examined here. Likewise, while adherence counseling remains essential, these data do not indicate that small incremental differences in already-high adherence produced substantial differences in treatment effect.
At the same time, the study should not be over-read as proving that timing and adherence never matter. The initiation window was relatively early overall, and adherence was unusually strong. The findings therefore apply best to women who start between 6 and 13 weeks and whose adherence is generally high, not to those initiating much later or taking medication sporadically.
Strengths and limitations
Strengths
The analysis has several strengths. First is scale: nearly 12,000 women contributed outcome data. Second is diversity: the multicentre design across low-resource settings enhances relevance to populations that bear a disproportionate share of global preterm birth burden. Third is frequent adherence assessment: pill counts every two weeks provide richer longitudinal exposure data than a single self-reported adherence measure late in pregnancy. Fourth is the use of relative risk modeling with adjustment for site and gestational age at treatment initiation, which is appropriate for these outcomes and settings.
Limitations
The main limitation is that this was a post hoc analysis. Interaction testing often has less statistical power than main-effect analyses, so failure to detect effect modification does not completely exclude modest heterogeneity. Second, adherence measured by pill count can misclassify true ingestion; tablets removed from packaging are not necessarily consumed. Third, the high adherence rate reduced contrast between exposure levels, making it harder to detect a dose-response relation if one exists. Fourth, the findings address initiation only within a relatively early window. They do not answer whether starting after 14 or 16 weeks changes efficacy, a question that remains relevant to real-world antenatal booking patterns.
There is also a conceptual issue common to adherence analyses in randomized trials: adherence may correlate with other health behaviors and social determinants. Although randomization balances treatment assignment, it does not randomize adherence itself. The authors appropriately analyzed effect modification rather than claiming a causal adherence effect, but clinicians should still recognize the potential for residual complexity in these relationships.
How this fits with the broader aspirin-in-pregnancy literature
Aspirin has established importance in selected pregnancies at increased risk for preeclampsia, and several professional societies recommend low-dose aspirin prophylaxis for appropriate risk groups. Earlier work has suggested that aspirin may be more effective for preeclampsia prevention when initiated before 16 weeks, particularly in high-risk patients, although results have not always been uniform across meta-analyses and trials. The present analysis addresses a different but related question: not whether aspirin works in general, but whether the effect seen in this specific trial varied according to start time within an already early initiation range and according to measured adherence.
Its neutral interaction findings do not negate the broader mechanistic argument for early therapy. Rather, they suggest that within the early first-trimester to early second-trimester window studied here, the effect was stable. This distinction is important. A stable effect between 6 and 13 weeks is fully compatible with the idea that much later initiation could be less effective, a possibility this analysis was not designed to resolve.
Implications for practice and policy
For clinicians, the practical takeaway is to continue offering indicated low-dose aspirin promptly in early pregnancy without creating unnecessary concern over small differences in start week within the 6- to 13-week interval. Programs should still aim for early antenatal engagement, but these data provide some reassurance when initiation occurs later within that window.
For health systems in low-resource settings, the results support implementation models that prioritize broad access, simple dispensing, and regular follow-up. The very high adherence achieved in the trial may itself offer implementation lessons: frequent contact, prepackaged medication allotments, and structured pill counts may have contributed to excellent treatment exposure. Even if higher adherence did not modify effect in this analysis, maintaining strong adherence remains important for fidelity to any preventive intervention.
For researchers, the study highlights the need for future work that more directly models cumulative dose, treatment discontinuation patterns, and later initiation windows. Analyses incorporating pharmacodynamic markers or digital adherence tools could also clarify whether true biological exposure behaves differently from pill-count estimates.
Conclusion
This post hoc analysis of the ASPIRIN trial provides a clinically useful negative finding: among women who started low-dose aspirin between 6 and 13 weeks’ gestation in low-resource settings, the treatment’s effects on preterm delivery, preterm delivery before 34 weeks, and perinatal mortality were not meaningfully modified by gestational age at initiation or by measured adherence. The data strengthen confidence that the parent trial findings are not explained by differential exposure across regions or by small differences in start timing within the studied window. The results are most applicable to early pregnancy initiation and generally high adherence, and they should be interpreted as complementary to, not replacing, the broader evidence base on aspirin prophylaxis in pregnancy.
Funding and ClinicalTrials.gov
Trial registration: ClinicalTrials.gov identifier: NCT02409680.
The abstract provided does not specify the funding source for this post hoc analysis. Readers should consult the full BJOG publication and the parent trial report for complete funding and sponsor details.
References
1. Mwenechanya M, Tembo AM, Chomba E, Hoffman MK, Goudar SS, Moore JL, Metgud MC, Somannavar MS, Okitawutshu J, Lokangaka AL, Tshefu AK, Bose CL, Bauserman MS, Figueroa L, Garces A, Krebs NF, Jessani S, Saleem S, Goldenberg RL, Kurhe K, Das P, Patel AB, Hibberd PL, Nyongesa P, Esamai F, Bucher SL, Goco N, Hemingway-Foday JJ, Babineau DC, McClure EM, Silver RM, Derman RJ, Carlo WA, ASPIRIN Study Group Collaborators. Does Adherence Modify Low-Dose Aspirin Effects in Pregnancy? Post Hoc Analysis From a Multi-Centre Trial. BJOG. 2026 Apr 19. PMID: 42002305.
2. Hoffman MK, Goudar SS, Kodkany BS, Metgud M, Somannavar MS, Okitawutshu J, et al. Low-Dose Aspirin for the Prevention of Preterm Delivery in Nulliparous Women with a Singleton Pregnancy (ASPIRIN): a randomised, double-blind, placebo-controlled trial. Lancet. 2020;395(10220):285-293.
3. American College of Obstetricians and Gynecologists; Society for Maternal-Fetal Medicine. Low-Dose Aspirin Use During Pregnancy. ACOG Committee Opinion No. 743. Reaffirmed 2023.