In 1978, Goodlin et al published a case report in The Lancet , describing a 31-year-old woman presenting with thrombocytopenia at 15 weeks’ gestation. Her obstetric history had been tumultuous: 2 preterm births with neonatal demise, both arising from “recurrent toxemia.” Perhaps with some alarm, the treating team found her platelet count was already just 59,000/μL, with further tests confirming peripheral consumption. To boost her platelet count, they first tried heparin for 4 weeks without avail.

They then turned to aspirin. At the rather hefty dose of 600 mg, 3 times a day, her platelet count commenced a slow but inexorable upward trajectory, eventually doubling after 10 weeks. But there was more. In contrast to her previous ill-fated pregnancies, she never developed hypertension for the 22 weeks while she was on aspirin. At 32 weeks’ gestation, the team ceased the aspirin due to concern it could close the fetal ductus arteriosus. Almost immediately she deteriorated, developing severe hypertension and her platelets acutely plummeted. Aspirin was promptly recommenced and her platelet count rose once more. She was eventually delivered at 34 weeks’ gestation, precipitated by fetal heart rate decelerations on a nonstress test where a live-born (albeit growth-restricted) fetus was delivered. This was the first case report of aspirin use in pregnancy and it set in train remarkable momentum, culminating in 3 decades of earnest clinical trials to determine whether aspirin can prevent preeclampsia. Some 30,000 women have now been randomized in clinical trials making aspirin easily one of the most intensely studied drugs in obstetrics.

Drawing on many of these clinical trials, 2 important meta-analyses have been published on aspirin use in this issue of the American Journal of Obstetrics and Gynecology . They have set out to address very specific important clinical questions that resonate with everyday clinicians: whether the dose and timing of aspirin commencement matters.

Bujold et al previously published a meta-analysis of placebo-controlled randomized clinical trials that suggested starting aspirin ≤16 weeks’ gestation is impressively effective in preventing preeclampsia, while there was no statistically significant effect when aspirin was commenced >16 weeks’ gestation. Furthermore, they reported starting aspirin ≤16 weeks’ gestation also significantly reduced the risk of related obstetric complications, such as perinatal death and fetal growth restriction. The impact of this meta-analysis, and a subsequent publication by the group in 2013, has been considerable. There are clinical guidelines now recommending that aspirin is commenced ≤16 weeks’ gestation for maximum preventative impact. This 16-week threshold has seeped into the practicing minds of many clinicians across the world.

In this issue of the American Journal of Obstetrics and Gynecology , the same team has published a new meta-analysis examining whether the dose of aspirin matters. They also revisited the 16 weeks’ gestation threshold and updated their meta-analysis with 3 new trials. Once more, the team found commencing aspirin ≤16 weeks’ gestation was strongly effective in reducing the risk of preeclampsia (relative risk [RR] 0.57), while the RR of 0.81 when starting it >16 weeks remained statistically not significant.

Importantly, they found a dose-dependent effect where doses of 100 mg of aspirin seemed more effective than lower doses. Table 2 of their report suggests a stronger effect with increasing aspirin dose, an effect that is more profound for severe preeclampsia. In studies that randomized women >16 weeks’ gestation, this dose-dependent relationship disappeared.

Life for the obstetrician-gynecologist at the coalface would be clean and simple if we could translate this to a crisp, incontrovertible recommendation to commence 100 mg of aspirin <16 weeks. However, also published in this edition of the American Journal of Obstetrics and Gynecology are powerful conflicting data. Meher et al performed an individual participant data (IPD) meta-analysis using data from the PARIS (Perinatal Antiplatelet Review of International Studies) study published almost a decade ago. This analysis showed a far more modest 10% reduction effect of aspirin on the risk of preeclampsia, and there was no difference in this risk reduction if aspirin was commenced either <16 or >16 weeks’ gestation. They arrived at the same conclusion for the risk of fetal demise and delivering a small-for-gestational-age infant: starting aspirin <16 or >16 weeks’ gestation made no difference.

This is all very perplexing. How can we possibly reconcile these opposing findings? The methodology of the IPD meta-analysis ensures that clinical characteristics of all individual patients from different studies are available for analysis. The use of IPD meta-analysis has several advantages over the use of aggregate data: more up-to-date information can be included than was available at the time of the original trial’s publication; analyses across studies are standardized; results of previously missing or poorly reported outcomes are incorporated; and clinical decisions can be personalized by assessing different treatment effects for subgroups, ensuring the optimal dose, timing, and delivery method of the interventions can be studied. Meher et al also reached more statistical power–they examined data from 9241 women to derive the RR of preeclampsia if aspirin was commenced <16 weeks and data from 21,429 women to calculate the RR of preeclampsia if aspirin was started >16 weeks. In contrast, Roberge et al included 5130 participants who started aspirin ≤16 weeks and 15,779 participants who started aspirin >16 weeks.

Apart from this increase in statistical power, the study of Meher et al is more protected against publication bias that may be present in the study by Roberge et al (see their funnel plot [Figure 4, A] and the accompanying statistical analysis ).

Also, it should be noted that there have been no clinical trials that have randomized women to starting aspirin <16 weeks’ gestation vs afterwards. Thus, there is no evidence addressing this question in direct comparisons: both meta-analyses have examined data generated from clinical trials, drawing their conclusions from indirect comparisons.

Turning to the possible biological mechanism of action does not really help us make our minds up whether aspirin will work better, if started earlier in pregnancy. The enthusiasts believing aspirin works better if commenced early might refer to the long-standing belief that aspirin somehow facilitates early placental embedding, a process that is in fact poorly understood but is likely to be complete by 16 weeks’ gestation. However, aspirin also increases prostacyclin (vasodilator) and may also decrease endothelial (blood vessel) dysfunction, actions that could make it effective in preventing clinical preeclampsia well >20 weeks’ gestation.

So here we are in 2017, >23 trials and 30,000 women later, still debating what to do. With the available evidence, clinicians on the ground need pragmatic advice that is useful for day-to-day care. We suggest the following. Firstly, the dose-response data in the report of Roberge et al for the outcome of severe preeclampsia look pretty convincing and suggest that the aspirin dose to prevent preeclampsia should not be <100 mg. Until this is assessed in an IPD or head-to-head comparative studies, clinicians should prescribe aspirin at dose of at least 100 mg. Secondly, while both reports concur that commencing aspirin <16 weeks’ gestation has preventative effects, the IPD of Meher et al representing >21,000 women suggest it is still beneficial to start aspirin even if commenced >16 weeks’ gestation. As to the magnitude of the risk reduction, the risk of preeclampsia is reduced by about 10% if aspirin is commenced >16 weeks’ gestation. It is tricky to decide what we should say is the risk reduction if it is started <16 weeks’ gestation. The IPD approach of Meher et al with a significantly larger sample size may be arguably more robust than the meta-analysis of Roberge et al. Therefore, advising patients that aspirin reduces the risk of preeclampsia by 10% is probably closer to the truth based on the current evidence.

Our responsibility does not end here. As researchers, we must design studies that address the knowledge gaps. The PARIS IPD meta-analysis has published a reanalysis of its data 10 years after its conception, but the IPD has not been updated. The hypothesis that an earlier start is beneficial and a high dose is better should be tested in trials directly comparing an early vs later start. We note with anticipation that the Aspirin for Evidence-based Preeclampsia Prevention Trial is due to report its findings soon. It is a study where nearly 30,000 women are being screened in the first trimester for their risk of developing preeclampsia. Around 1700 identified as high risk are then randomized to aspirin or placebo. The quality and independence of systematic reviews can be improved by complying with guidelines such as those recommended by the Institutes of Medicine, and by following transparent protocols. Finally, we appeal to clinicians on the ground to contribute to studies as part of their routine practice. Currently, the costs of trial are in the millions and one of the main drivers is difficult recruitment, a responsibility that rests with all of us.

In the past decade and a half, the pace of discoveries in the field of preeclampsia have accelerated and this has led to some new drug possibilities that could prove effective to treat or prevent preeclampsia. The line of proposed new drugs are an eclectic mix–sildenafil, esomeprazole, metformin, and even pravastatin. There may be others. Thus, it is possible that in coming years a drug will be proven in clinical trials to be clearly better than aspirin at preventing preeclampsia and other major complications. Let’s hope so–that could finally draw to a close the circular debates over the timing, dose, and the effectiveness of aspirin.