We found 171 randomised controlled trials meeting our criteria for examining online learning for healthcare professionals’ knowledge and skills acquisition. Overall, online learning demonstrated large positive effects when compared to static controls (e.g., waitlists, self-directed textbook reading, electronic or print materials) and small but significant positive effects compared to face-to-face instruction. Results seemed consistent regardless of the learning design, intervention characteristics, professional group, clinician experience, or clinical setting. Overall, the findings suggest that online programmes may be more effective than traditional methods for improving healthcare professionals’ knowledge and skills, with the added advantages of scalability and flexibility for healthcare professional development^20–23.

Prior reviews of online learning were small ^e.g.,14 or focused on a single skill or context^25,26,30. In contrast, our findings show that online learning is effective across diverse professional groups, experience levels, practice settings, and income classifications (Figures 3 and 5). These consistent findings across different contexts establish online learning as a reliable approach for healthcare professional development. Our broad inclusion criteria also gave us the power to test a large number of potential moderators. Few of these moderators demonstrated consistent effects. Given the heterogeneity we observed, this means differences between effect sizes are driven by differences we did not or could not assess from the study’s report (e.g., the institutional support for online learning⁴⁹, or the degree of end-user co-design⁵⁰).

A few significant moderators emerged. Benefits of online learning were particularly pronounced in lower-middle income countries when compared against static controls, likely because there were fewer high-quality alternatives to the training provided by the trials in these settings^18,19. In these contexts, geography and limited infrastructure are key barriers to professional development^22,26, and online learning’s digital delivery, flexible access, and low cost^21,23 directly address these constraints. In high-income countries, where professionals have greater resources and access to professional learning, these advantages are less distinctive and the marginal benefit of any single programme is smaller by comparison. Even so, online learning improved knowledge and skill acquisition compared to static control resources and face-to-face delivery across all income settings, suggesting its value is universal but its potential to reduce global health inequalities is greatest where access to professional development is most limited²².

Another moderator of note was the type of outcome assessed, observed only when comparing online learning to face-to-face instruction. Online learning produced larger effects for knowledge acquisition than for skills acquisition, which aligns with evidence that structured digital content is well suited to transferring declarative knowledge. More notably, online learning still outperformed face-to-face instruction for skills acquisition, challenging the assumption that physical presence is the primary driver of procedural development. Other systematic reviews have shown online video leads to bigger improvements in skill acquisition than knowledge development⁵¹, arguably because video can easily show authentic skill demonstrations. Benefits of online learning may reflect the inherent variability of face-to-face clinical training, where the learning experience depends on available cases, supervisors, and setting, meaning practitioners may develop inconsistent levels of competence. Online learning, by contrast, delivers standardised instruction consistently across learners regardless of context, which can mitigate the inconsistencies of traditional professional learning.

Interactivity level did not moderate online learning effectiveness in either comparison, contrary to the Interactive, Constructive, Active, and Passive (ICAP) framework’s prediction that constructive and interactive engagement produce greater learning gains than passive modes²⁷. Two explanations are plausible. Conceptually, the lack of a moderating effect may stem from the fact that all engagement levels (i.e., active, constructive, interactive, and passive) produced significant positive effects with tightly clustered effect sizes (Figure 3 & 5). When every delivery format consistently outperforms control conditions, specific differences between categories become difficult to isolate, suggesting that the primary driver of benefit is the online learning itself rather than the specific level of interactivity. Methodologically, interactivity was categorised from published intervention descriptions rather than direct measurement of learner experience, which may have reduced sensitivity to detect meaningful differences. Overall, the findings suggest that even less resource-intensive, passive online learning formats provide a robust and scalable alternative to traditional professional development for advancing professional knowledge and skills.

It remains unclear whether any learning design principles such as feedback, scaffolding, simulation, and group collaboration, enhance the effects of online learning on knowledge and skill acquisition. This may be an artefact of our design’s sensitivity to detect between-study moderators. Studies in our review covered different content for different audiences, measuring different outcomes. It might therefore be difficult to detect effects of studies using specific learning design principles. Robust assessments of these learning design principles exist, for example, where the same training is conducted with and without online videos⁵¹. Learning designers should use design principles supported by existing systematic reviews, such as letting learners collaborate^28,29, use quizzes^52,53, design clear multimedia⁵⁴, scaffold the training⁵⁵ or provide authentic simulations⁵⁶.