Investigators assessed long-term functional recovery using the modified Rankin Scale (mRS) at five years. They compared patients discharged on amlodipine with those discharged on nifedipine and modeled the full distribution of mRS using ordinal logistic regression. The adjusted analysis produced an odds ratio (OR) of 0.76 for worse outcomes in the amlodipine group versus nifedipine (95% CI 0.65–0.89; p < 0.001), indicating lower mRS scores among amlodipine users at the five-year mark. Short declarative facts matter: the primary outcome was function, not mortality. The registry data also showed that 51.76% of weighted amlodipine users had an mRS > 0 at 5 years compared with 59.74% of weighted nifedipine users, yielding an OR of 0.71 (95% CI 0.61–0.83; p < 0.001). The study team reported these differences consistently across multiple analytic approaches.
The dataset derived from the Third China National Stroke Registry (CNSR-III), which enrolled consecutive adult ischemic stroke and TIA patients between 2015 and 2018. From that registry the authors selected patients discharged with either amlodipine or nifedipine documented in prescribing records. To address baseline imbalance, investigators implemented inverse probability of treatment weighting (IPTW). They explicitly adjusted models for patient age, sex, body mass index, systolic blood pressure at admission, and onset date. After IPTW, the effective sample used in comparisons included 1,847 patients in the amlodipine arm and 1,058 in the nifedipine arm. The source did not report additional randomized allocation, since this was an observational registry analysis rather than a controlled trial.
Data indicated a clinically measurable shift in the entire mRS distribution favoring amlodipine at five years. The ordinal model — adjusted and weighted — returned an OR of 0.76 for worse mRS with amlodipine versus nifedipine. A secondary dichotomous outcome (mRS > 0) also favored amlodipine, with roughly eight percentage points fewer patients affected in the amlodipine group after weighting. The study authors presented these effects as statistically significant and clinically relevant within the constraints of the dataset. Short sentence: the effect persisted in adjusted analyses. Longer sentence: the investigators noted that findings were robust across the primary analytic approach and that observed differences in function were not explained solely by the covariates included in their models, although residual confounding cannot be excluded.
The research team performed sensitivity analyses to test whether the primary result was dependent on the analytic technique. The source reported that sensitivity analyses yielded consistent results, with effect estimates aligning with the primary IPTW-based models. The article did not list every sensitivity scenario or subgroup boundary in the abstract; detailed tables and figures were provided in the manuscript for readers seeking the full set of diagnostics. What the authors documented clearly: multiple analytic checks did not reverse the direction of benefit seen with amlodipine, and the magnitude of association remained similar across the sensitivity procedures that they described.
This investigation carried the limitations inherent to observational registry research. The authors acknowledged that causal inference is limited by nonrandomized treatment allocation and potential unmeasured confounding. The analysis relied solely on registry data, and the manuscript reported a relatively high loss-to-follow-up at five years, which could bias long-term estimates. The source also stated that individual-level patient data are not publicly available because of Chinese health data protections and institutional ethical restrictions; access requires formal application to the data access committee. The authors concluded that randomized controlled trials are warranted to confirm whether the association between discharge amlodipine and improved five-year functional outcomes is causal and to guide optimal CCB selection after ischemic stroke or TIA.
Limitations left open specific operational questions that the registry could not answer: dosing regimens, adherence over time, exact switching between CCB agents during follow-up, and adjudicated causes of functional decline were not reported in the abstract and therefore cannot be assumed. The manuscript ends with concrete next steps recommended by the authors — prospective randomized evaluation — rather than broad, untestable claims.