Details on the design, protocol, and statistical analysis plan of the ATAMIS trial have been previously published [7,9]. In brief, the ATAMIS trial was a multicenter, open-label, blinded-endpoint, randomized clinical trial enrolling 3,000 patients between December 20, 2016, and August 9, 2022, which aimed to assess the effect of clopidogrel plus aspirin in patients with acute mild-to-moderate ischemic stroke. The eligibility criteria included an age of 18 years or older, functioning independently before stroke (modified Rankin Scale [mRS] scores, 0-1; range, 0 [no symptoms] to 6 [death]), and diagnosis with acute mild-to-moderate ischemic stroke (baseline National Institute of Health Stroke Scale [NIHSS] scores, 4-10; range, 0-42, with higher scores indicating greater stroke severity) within 48 hours after stroke onset. The exclusion criteria were patients who met the eligibility criteria for intravenous thrombolysis or endovascular therapy and had a clear indication for anticoagulation or history of intracerebral hemorrhage. All study procedures were reviewed and approved by the Ethics Committee of the General Hospital of Northern Theater Command (Approval Number: k [2016] 6-1). Written informed consents were obtained from either the patients or their legally authorized representatives. This study was registered at ClinicalTrials.gov (NCT02869009). All patients from the modified intention-to-treat analysis set were included in the post hoc analysis. If patients were diagnosed with atrial fibrillation following randomization, they received anticoagulation therapy after antiplatelet therapy, in accordance with the national guidelines, and were included in the full analysis set based on the intention-to-treat principle. The detailed screening of patients for the analysis has been previously reported [7]. The current analysis followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline (Appendix 1).

In the ATAMIS trial, patients were randomly assigned to receive either dual antiplatelet therapy with clopidogrel plus aspirin or monotherapy with aspirin alone. Further details of antiplatelet therapy have been described in a previous report [7]. In this post hoc analysis, according to OTT using 24 hours as a cutoff in which dual antiplatelet therapy was recommended for minor stroke by guideline [1], eligible patients were divided into two subgroups based on treatment time: 0-24 hours group (OTT <24 h) and 24-48 hours group (OTT ≥24 h). Neurological status, measured with NIHSS score, was evaluated at admission, and at 7 and 14 days after randomization. Follow-up data, including the assessment of clinical prognosis, recurrent events, and safety events, were collected 90 days after randomization.

In this post hoc analysis, all the outcomes paralleled the ATAMIS trial [7]. The primary outcome was occurrence of END at 7 days, defined as an increase in NIHSS of greater than 2 points compared with baseline (excluding cerebral hemorrhage) [10]. The secondary outcomes were excellent functional outcome at 90 days, defined as an mRS score of 0 to 1; a shift in the distribution of the ordinal mRS score at 90 days; a change in the NIHSS score compared with baseline at 14 days; time from randomization to occurrence of new stroke [11] within 90 days; and time from randomization to the occurrence of other vascular events (pulmonary embolism, peripheral vessel or cardiovascular event) or all-cause death within 90 days. Safety outcomes included the occurrence of any bleeding events or intracranial hemorrhage during the trial, given that these were directly associated with antiplatelet therapy. Detailed information for the outcome assessment is shown in the Supplementary Methods.

Considering the potentially imbalanced characteristics between the groups in the post hoc analysis, the primary analysis was adjusted. Continuous variables are summarized as median (interquartile range [IQR]) and categorical variables as frequencies (percentages). The baseline characteristics between groups were compared using the Mann-Whitney U test or chi-square test. For the treatment effect of outcomes, we estimated the absolute number of events, as well as the risk difference (RD), odds ratio (OR), geometric mean ratio (GMR), or hazard ratio (HR) with their 95% confidence intervals (CIs).

First, as a continuous variable, restricted cubic spline analysis with three knots at the 10th, 50th, and 90th percentile [12] was applied to explore the nonlinear association between OTT, as well as the probability of the primary outcome compared with antiplatelet therapies. The probability was calculated using binary logistic regression analysis, and the best-fit line was drawn according to probability and OTT. As a categorical variable, the primary outcome was analyzed for each subgroup.

Second, both unadjusted and adjusted generalized linear models and Cox regression analyses were performed to evaluate the effect of antiplatelet therapies on outcomes according to OTT, and in each time subgroup. The primary adjusted model included any imbalanced baseline variables (P<0.1), as well as pre-specified covariates (age, sex, history of diabetes, history of hypertension, NIHSS score at randomization, and presumed stroke cause based on the Trial of Org 10172 in Acute Stroke Treatment [TOAST] classification [13]) in the ATAMIS trial. Adjusted interactions between the OTT and the effect of antiplatelet therapy on outcomes were assessed by including imbalanced variables between time subgroups.

Third, a subgroup analysis of primary outcome was conducted according to six pre-specified factors (age [<65 yrs or ≥65 yrs], sex [female or male], history of diabetes [yes or no], history of hypertension [yes or no], NIHSS score at randomization [<7 or ≥7], and presumed stroke cause), in order to investigate the characteristics of patients benefiting from dual antiplatelet therapy in each time subgroup.

In addition, per-protocol analysis was restricted to patients who received the completed intervention, as specified in the protocol [9], while propensity score matching analysis for all outcomes was performed as a sensitivity analysis. Propensity score matching was performed to generate a new cohort with a balanced sample size between the time subgroups to control for sample-size bias. Furthermore, patients were analyzed by grouping with different OTT cutoff values, such as 12 hours or 18 hours.

Detailed information regarding the adjusted models, propensity score matching, and statistical analysis is presented in the Supplementary Methods. All analyses were exploratory, and all P values were nominal. Two-sided P values <0.05 were considered significant. All statistical analyses were performed using the SPSS software (version 26.0; IBM Corp., Armonk, NY, USA) and R software (version 4.1.0; R Foundation for Statistical Computing, Vienna, Austria).

A total of 2,915 patients from the ATAMIS trial were enrolled in this post hoc analysis, including 1,757 patients in the 0-24 hours group (OTT: median [IQR], 9.0 h [5.4 to 14.5 h]) and 1,158 patients in the 24-48 hours group (OTT: median [IQR], 28.6 h [25.1 to 41.2 h]). The median (IQR) age was 66 (59-73) years, and 1,020 (35.0%) patients were women. In total, 902 patients were assigned to receive clopidogrel plus aspirin, and 855 patients were assigned to receive aspirin alone in the 0-24 hours group, whereas 600 and 558 patients were assigned to receive these treatments in the 24-48 hours group, respectively (Figure 1).

Table 1 shows the baseline characteristics of patients assigned to different antiplatelet therapies in each time subgroup. In the 0-24 hours group, a higher proportion of patients with diabetes was assigned to the clopidogrel plus aspirin group. In the 24-48 hours group, a higher proportion of patients with a history of transient ischemic attack were assigned to receive aspirin alone. We identified differences in age, sex, current smoking, current drinking, NIHSS score at admission, OTT, presumed stroke cause, and location of the responsible vessel between the 0-24 hours and 24-48 hours groups.

As a natural continuous characteristic of OTT, the probability of END at 7 days was compared between the clopidogrel plus aspirin and aspirin alone groups according to OTT, which revealed that the probability decreased with a longer OTT (Figure 2A and B). Compared with aspirin alone, the analysis revealed a significantly decreased risk of END after receiving clopidogrel plus aspirin therapy (4.8% vs. 6.7%; adjusted RD, -1.9%; 95% CI, -3.6% to -0.2%; P=0.03) (Table 2). However, no significant interactions were observed (adjusted P=0.13). As such, we explored the nonlinear association between continuous OTT and the primary outcome using restricted cubic spline analysis. Although no significant nonlinear association was found (P for nonlinear=0.06) (Figure 2C and D), the odds ratio for END at 7 days, as well as that for antiplatelet therapy, decreased with longer OTT and plateaued at 48 hours.

The distribution of mRS scores at 90 days between the clopidogrel plus aspirin and aspirin alone groups is presented in Figure 3A. None of the secondary or safety outcomes were significantly different between clopidogrel plus aspirin and aspirin alone across the continuous OTT, and no significant interactions were found (Table 2).

Compared with aspirin alone, the analysis revealed significantly decreased odds in END following clopidogrel plus aspirin treatment in the 0-24 hours group (5.7% vs. 9.2%; adjusted RD, -3.7%; 95% CI, -5.5% to -2.0%; P<0.01) (Table 3), while no significant difference was found in the 24-48 hours group (3.5% vs. 2.9%; adjusted RD, 0.6%; 95% CI, -0.8% to 2.0%; P=0.40) (Table 3). Significant interactions between the time subgroups and the effect of antiplatelet therapy on the primary outcome were observed (adjusted P=0.03).

The distribution of mRS scores at 90 days between the clopidogrel plus aspirin and aspirin alone groups in each time subgroup is presented in Figure 3B and C. None of the secondary or safety outcomes showed significant differences between the clopidogrel plus aspirin and aspirin alone in each time subgroup, and no significant interactions were found (Table 3).

To explore the characteristics of patients who benefited from early dual antiplatelet therapy, we further investigated the association between OTT, as well as the effect of antiplatelet therapy on the primary outcome in relation to age, sex, history of hypertension, history of diabetes, stroke severity, and presumed stroke cause (Figure 4). In each characteristic subgroup, clopidogrel plus aspirin showed a lower risk of END at 7 days than aspirin alone in the 0-24 hours group compared with the 24-48 hours group, which was similar to the result observed in the primary analysis. Notably, significant interactions between the OTT subgroup and the effect of antiplatelet therapy on the primary outcome were identified in patients with NIHSS scores of 4 to 7 at admission (P=0.046) and an undetermined stroke etiology (P=0.01).

All outcomes analyzed in patients with propensity score matching and from the per-protocol analysis set were similar to those in patients from the modified intention-to-treat analysis set. As the number of OTT groups increased, the results also indicated that patients benefited more from earlier dual antiplatelet therapy. The results of the sensitivity analysis are presented in the Supplementary Results.