Table 15 Summary and interpretation of main results
Outcome | Analysis | Meta-analysis results | Conclusions–interpretations |
|---|---|---|---|
COVID-19 impacts | |||
Perceived training disruption of learners | Overall | 71.1% (67.9–74.2), I2 = 98.7%, N = 66 870 | A considerable rate of learners likely perceived some extent of disruption of training amidst the pandemic |
Invasive vs non-invasive experience | Invasive: 75.8% (71.4–79.9), I2 = 98.2%, N = 23 047; non-invasive: 69.7% (64.4–74.8), I2 = 98.7%, N = 25 463 | Learner perceived disruption of training was high in terms of both invasive procedures and non-invasive clinical experience, though the former was more prominent | |
By WHO region | AMR: 67.1% (61.3–72.8), I2 = 97.9%, N = 13 430 vs EUR: 71.1% (65.9–76.0), I2 = 97.8%, N = 15 249 vs EMR: 71.6% (60.7–81.3), I2 = 99.3%, N = 12 019 vs SEAR: 84.5% (80.3–88.4), I2 = 95.3%, N = 7 809 vs WPR: 69.9% (60.2–78.8), I2 = 97.0%, N = 3 964; psubgroup < 0.001 | The highest learner rate perceiving training disruption was recorded in the SEAR. These rates may be examined in combination with the satisfaction and preference rates for online learning methods. However, the disruption should be considered multifactorial (e.g., redeployment, decrease of case numbers, etc.) and dissatisfaction with virtual delivery of education may just be one of the contributing factors | |
Learner redeployment | Overall | 29.2% (25.3–33.2), I2 = 95.3%, N = 11 527 | Approximately 3 out of 10 learners might have been redeployed due to the pandemic |
By WHO region | AMR: 24.7% (19.5–30.3), I2 = 94.4%, N = 4 838 vs EUR: 35.2% (28.8–41.8), I2 = 94.6%, N = 4 156 vs AFR: 40.7% (10.2–75.8), I2 = 97.0%, N = 276 vs EMR: 25.9% (9.5–46.6), I2 = 96.5%, N = 648 vs SEAR: 13.7% (0.1–43.8), I2 = 97.7%, N = 420; psubgroup = 0.092 | When compared with their colleagues in the AMR, learners in the EUR likely exhibited higher redeployment rates due to the pandemic | |
Learners rethinking career plans | Overall (and sensitivity analysis) | 21.5% (17.1–26.3), I2 = 99.5%, N = 134 623; [21.8% (17.2–26.8), I2 = 99.1%, N = 35 955 after exclusion of studies with N > 25 000 to minimize risk of duplicate population] | A considerable rate of learners reconsidered their career plans (residency/practice/expertise) due to the COVID-19 pandemic |
At least moderate scaled learner anxiety | Overall | 32.3% (28.5–36.2), I2 = 99.4%, N = 95 927 | Amidst the COVID-19 pandemic, approximately one-third of learners might have screened positive for anxiety of at least moderate severity |
GAD-7 only (and sensitivity analysis) | 32.1% (26.6–37.9), I2 = 99.5%, N = 53 658 (low risk of bias studies only: 32.2% (26.0–38.7), I2 = 99.5%, N = 45 382) | Learner rates of at least moderate anxiety did not materially change when only studies that used the GAD-7 screening tool (and their low-risk of bias-subset) were analyzed | |
By ISCO-08 HW group | Medical doctors: 30.4% (25.6–35.3), I2 = 99.5%, N = 76 730 vs nursing professionals: 33.0% (20.1–47.4), I2 = 98.5%, N = 3 196 vs dentists: 32.4% (25.4–39.7), I2 = 96.3%, N = 4 812 vs Pharmacists: 50.0% (45.6–54.5), I2 = 19.1%, N = 643; psubgroup < 0.001 | Pharmacy learners might have screened positive for at least moderate anxiety at significantly higher rates than the other occupational groups. Anxiety is likely multifactorial and, therefore, reasons leading to higher anxiety in this occupational group might have not been investigated in this paper | |
By training level | Undergraduate: 34.9% (30.2–39.9), I2 = 99.4%, N = 63 736 vs postgraduate: 28.4% (23.2–34.0), I2 = 98.4%, N = 19 343; psubgroup = 0.079 | Although anxiety is multifactorial, higher anxiety observed in undergraduate learners could be partially attributed to their lower satisfaction with online learning compared to their postgraduate counterparts | |
By WHO region | AMR: 32.4% (25.9–39.4), I2 = 98.5%, N = 13 977 vs EUR: 38.5% (30.8–46.4), I2 = 99.3%, N = 28 246 vs AFR: 33.1% (15.8–53.1), I2 = 94.0%, N = 862 vs EMR: 40.4% (34.1–46.8), I2 = 98.7%, N = 17 824 vs SEAR: 26.6% (20.2–33.6), I2 = 97.4%, N = 6 759 vs WPR: 15.3% (9.7–21.8), I2 = 99.4%, N = 26 196; psubgroup < 0.001 | Learners in the WPR might have screened positive for anxiety of at least moderate severity at significantly lower rates compared to their counterparts in the other regions. Learner anxiety rates may have also been lower in the SEAR compared to the EMR and EUR. The continent analysis further showed significantly higher anxiety rates in South compared to North America (and Asia). This difference could not have been revealed by the WHO regional analysis. Combined interpretation of these analyses is therefore essential (Table 5) | |
By gender | Female: 39.8% (29.5–50.4), I2 = 99.5%, N = 18 384 vs male: 25.4% (17.6–34.2), I2 = 98.4%, N = 7 913; psubgroup = 0.038 | In line with the relevant literature, female gender may have been associated with increased anxiety rates | |
By year of study end date (2020 vs 2021) | 2020: 28.7% (24.8–32.8), I2 = 99.1%, N = 55 368 vs 2021: 41.9% (35.0–48.9), I2 = 98.8%, N = 22 016; psubgroup = 0.001 | Learner rates of at least moderate anxiety may have been higher in 2021 compared to 2020, reflecting potential accumulation as pandemic continued to evolve. This finding could indicate that policies for prevention of learners’ anxiety should have been implemented early during the pandemic | |
At least moderate scaled learner depression | Overall | 32.0% (27.9–36.2), I2 = 99.4%, N = 84 067 | Amidst the COVID-19 pandemic, approximately one-third of learners screened positive for depression of at least moderate severity |
PHQ-9 only (and sensitivity analysis) | 32.8% (25.3–40.7), I2 = 99.6%, N = 39 876 (low risk of bias studies only: 31.0% (23.0–40.0), I2 = 99.6%, N = 32 803) | Learner rates of at least moderate depression did not materially change when only studies that used the PHQ-9 screening tool (and their low-risk of bias-subset) were analyzed | |
By training level | Undergraduate: 35.0% (29.9–40.3), I2 = 99.4%, N = 55 559 vs Postgraduate: 25.7% (17.7–34.5), I2 = 99.4%, N = 18 269 vs continuing: 21.6% (8.3–39.0), I2 = 94.5%, N = 911; psubgroup = 0.098 | As with anxiety, undergraduate learners may have screened positive for depression of at least moderate severity at higher rates than their postgraduate counterparts | |
By WHO region | AMR: 32.7% (23.1–43.0), I2 = 99.2%, N = 11 937 vs EUR: 35.9% (26.5–45.9), I2 = 99.5%, N = 25 235 vs EMR: 43.6% (36.2–51.2), I2 = 99.0%, N = 17 011 vs SEAR: 26.4% (15.6–38.9), I2 = 99.1%, N = 5 885 vs WPR: 14.9% (12.0–18.1), I2 = 97.4%, N = 22 606; psubgroup < 0.001 | Learners in the WPR might have screened positive for depression of at least moderate severity at significantly lower rates compared to their counterparts in the other regions (especially AMR, EUR, EMR). Regional differences in anxiety and depression rates of at least moderate severity might follow a similar pattern, with the highest rates being observed in the EMR, followed by the EUR, AMR, SEAR and WPR. However, some of these differences may be due to chance alone. As with anxiety, significantly higher depression rates were found by studies in South America when compared with studies conducted in the other continents (Table 5) | |
Learner scaled burnout | Overall | 38.8% (33.4–44.3), I2 = 99.0%, N = 35 808 | Almost 4 out of 10 learners might have screened positive for burnout syndrome amidst the pandemic |
MBI and variants only (and sensitivity analysis) | 46.8% (38.6–55.1), I2 = 98.4%, N = 17 134 (low risk of bias studies only: 43.5% (35.3–51.9), I2 = 98.4%, N = 16 964) | Studies using the MBI and its variants revealed higher learner burnout rates. This may be a more accurate estimation of learner burnout rates or an overestimation due to potentially higher false-positive rates observed when using certain MBI variants | |
Learner scaled insomnia | Overall | 30.9% (20.8–41.9), I2 = 99.2%, N = 9 906 | Almost one-third of learners might have screened positive for insomnia amidst the pandemic. Combining the findings on anxiety, depression, burnout, and insomnia (all as per measurements with validated scales) it appears that HW learners may be considered as a vulnerable group for “mental health disruption”, as they are simultaneously faced with two distinct and equally challenging tasks, namely education and patient care |
By year of study end date (2020 vs 2021) | 2020: 24.6% (14.5–36.3), I2 = 99.2%, N = 7 941 vs 2021: 50.5% (31.4–69.5), I2 = 98.0%, N = 1 512; psubgroup = 0.023 | As with anxiety, learner rates of insomnia may have been higher in 2021 compared to 2020, reflecting potential accumulation as pandemic continued to evolve. This finding could indicate that policies for prevention of learners’ insomnia should have been implemented early during the pandemic | |
Outcomes of policies | |||
Satisfaction with online | Learner (and sensitivity analysis) vs faculty | Learner: 75.9% (74.2–77.7), I2 = 99.3%, N = 425 466 [76.2% (74.0–78.3), I2 = 99.2%, N = 226 348 after exclusion of studies with N > 25 000 to minimize risk of duplicate population]; faculty: 71.8% (66.7–76.7), I2 = 93.9%, N = 6 525 | HW learners and faculty might have been generally satisfied with online learning methods during the pandemic, with faculty appearing somewhat less satisfied than learners. A potential explanation could be that faculty may have encountered the extra challenge of attempting to engage their audiences |
Learner satisfaction with online learning | Theoretical vs practical vs clinical experience (and sensitivity analyses) | Theoretical: 67.5% (64.7–70.3), I2 = 99.5%, N = 252 931 (67.6% (64.4–70.7), I2 = 99.4%, N = 153 372 after exclusion of studies with N > 25 000 to minimize risk of duplicate population); Practical: 85.4%, (82.3–88.2), I2 = 99.2%, N = 153 445 [85.5% (82.5–88.2), I2 = 98.6%, N = 53 886 after exclusion of studies with N > 25 000 to minimize risk of duplicate population]; clinical experience: 86.9% (79.5–93.1), I2 = 98.5%, N = 8 640 | During the pandemic, HW learners might have been more satisfied with online practical courses and online true clinical experience involving patients than with predominantly theoretical online courses. When lack of interaction/practice was addressed to the possible extent, satisfaction seemed to increase |
By training level | Undergraduate: 71.9% (69.8–74.0), I2 = 99.4%, N = 361 819 vs postgraduate: 79.1% (75.4–82.5), I2 = 96.0%, N = 14 611 vs continuing: 86.8% (82.0–91.0), I2 = 95.3%, N = 6 173; psubgroup < 0.001 | Satisfaction with online learning seemed to significantly increase as training level increased. Accessibility and flexibility of this format may have better suited the likely busier schedules of learners at higher training stage | |
By WHO region | AMR: 84.0% (80.9–87.0), I2 = 97.7%, N = 31 019 vs EUR: 78.8% (74.4–82.9), I2 = 99.3%, N = 61 616 vs AFR: 86.1% (70.4–96.7), I2 = 98.5%, N = 2 680 vs EMR: 59.6% (53.9–65.1), I2 = 99.3%, N = 48 152 vs SEAR: 60.9% (53.8–67.8), I2 = 99.2%, N = 23 949 vs WPR: 78.5% (74.2–82.4), I2 = 99.7%, N = 238 209; psubgroup < 0.001 | Learner satisfaction with virtual learning methods might have been lower in the EMR and SEAR when compared to that in the AMR, EUR, AFR and WPR. Lower satisfaction might be attributed to lower availability of resources, potential connectivity issues or difficulty in accessing necessary equipment in these regions. Learners in the AFR might have experienced accessibility or other issues with the in-person format even before the onset of the pandemic. The need to bypass such issues may have reinforced their satisfaction with online options | |
Learner preference for learning method | Online vs face-to-face vs blended | Online: 32.0% (29.3–34.8), I2 = 98.7%, N = 94 452; face-to-face: 48.8% (45.4–52.1), I2 = 99.0%, N = 97 903; blended: 56.0% (51.2–60.7), I2 = 96.9%, N = 14 992 | Learners seemed to prefer the existence of an in-person component in their curriculum. The virtual component was potentially preferred as part of a blended educational system rather than a purely distant format |
By training level | Undergraduate: 29.5% (26.5–32.6), I2 = 98.5%, N = 62 459 vs postgraduate: 39.7% (33.2–46.4), I2 = 98.2, N = 16 911 vs continuing: 39.9% (27.7–52.7), I2 = 97.4%, N = 3 369; psubgroup = 0.007 | Postgraduate learners likely preferred the virtual format significantly more than their undergraduate counterparts. This is in accordance with findings on satisfaction. Accessibility and flexibility of this format may have better suited their likely busier schedules | |
By WHO region | AMR: 38.3% (31.5–45.2), I2 = 98.2%, N = 16 146 vs EUR: 37.3% (32.7–42.1), I2 = 98.2%, N = 30 492 vs AFR: 29.7% (11.5–51.9), I2 = 98.3%, N = 1 102 vs EMR: 33.1% (26.2–40.4), I2 = 98.7%, N = 13 421 vs SEAR: 22.7% (18.4–27.3), I2 = 97.9%, N = 17 276 vs WPR: 29.7% (15.9–45.6), I2 = 99.4%, N = 8 282; psubgroup < 0.001 | Preference for the purely virtual format appeared to be lower for learners in the SEAR when compared to their counterparts in the AMR, EUR and EMR. Focusing on the comparison of the SEAR and the EMR, and combining the results with those of satisfaction per WHO region, it is likely that the lower satisfaction with the virtual courses in the EMR region may have resulted more from issues emerging during their delivery rather than the virtual format itself. The same might not apply for countries of the SEAR, in which learners may have perceived the virtual-only format as less feasible, regardless of how well the courses were actually delivered | |
Learner preference for face-to-face learning | By training level | Undergraduate: 50.9% (46.9–54.9), I2 = 99.1%, N = 70 146 vs Postgraduate: 47.6% (39.9–55.4), I2 = 97.8%, N = 8 217 vs continuing: 30.7% (21.1–41.2), I2 = 95.3%, N = 3 066; psubgroup = 0.003 | In accordance with preference for the virtual format, preference for the in-person educational format might have been significantly higher for undergraduate learners than their counterparts at senior training stage. However, preference rates for in-person learning were likely higher than those for virtual training for learners of all levels |
Learners wanting to keep learning method post-pandemic | Online-only vs blended | Online: 34.7% (30.7–38.8), I2 = 99.0%, N = 59 765; blended: 68.1% (64.6–71.5), I2 = 98.4%, N = 49 585 | Learners were likely in favor of maintaining the virtual format post-pandemic along with their in-person curricular activities rather than maintaining it on its own |
Learners wanting to keep blended learning post-pandemic | By WHO region | AMR: 75.7% (64.8–85.2), I2 = 98.5%, N = 5 195 vs EUR: 74.8% (68.6–80.6), I2 = 97.0%, N = 8 182 vs AFR: 76.5% (52.4–94.1), I2 = 94.6%, N = 813 vs EMR: 55.8% (46.2–65.2), I2 = 98.8%, N = 9 489 vs SEAR: 56.7% (48.9–64.2), I2 = 97.6%, N = 7 037 vs WPR: 62.2% (55.6–68.6), I2 = 97.2%, N = 13 507; psubgroup < 0.001 | As more learners have likely expressed the desire to maintain the virtual format as part of a blended system, rates of learners in favor of a future blended system were generally in accordance with the rates of satisfaction with online methods, except for the WPR. The lower-than-expected rates in the WPR might be attributed to saturation with the virtual format (even as part of a blended system and despite the potentially high quality of its delivery), considering that the pandemic struck this region first and transition to the virtual format might have occurred there first |
Effectiveness of learning methods | Comparator vs intervention | SMD = − 1.09 (− 1.21 to -0.96), I2 = 98.2%, N = 49 911 [SMD = − 1.11 (− 1.25 to − 0.96), I2 = 97.9%, N = 24 432 after exclusion of studies with N > 25 000 to minimize risk of duplicate population] Pre vs Post-intervention (phase 2): SMD = − 1.31 (− 1.46 to − 1.16), I2 = 98.1%, N = 42 060 Comparator (previous method) vs intervention SMD = − 0.28 (− 0.48 to − 0.09), I2 = 94.3%, N = 4 489 | Learning methods applied during the pandemic seemed overall effective as they likely managed to significantly improve learners’ mean knowledge or acquired overall skills’ scores compared to pre-training status or the respective pre-pandemic methods. A main limitation of these studies is that they are based on evaluations right after the intervention without long-term follow-up. That often leads to overvalued effectiveness of the interventions. That is more evident in the studies comparing knowledge/skills’ scores before and after the intervention |
Learner satisfaction with pandemic face-to-face learning | Overall | 93.0% (89.1–96.2), I2 = 95.4%, N = 6 263 | Learner satisfaction with the in-person learning activities that were employed during the pandemic, was likely high (probably even higher than that with online activities). Learners might have been that satisfied either due to the in-person format inside a curriculum full of virtual activities or because of the COVID-19-related character of many of these activities, with the latter potentially providing them with essential knowledge/skills to deal with this pandemic |
Learner satisfaction with online assessment | Overall | 68.8% (60.7–76.3), I2 = 98.6%, N = 11 072 | Learner satisfaction with virtual evaluation methods was likely moderate to high, probably reflecting a balance between convenience or better scores and potential cheating or perception of unfairness |
By training level | Undergraduate: 62.5% (52.4–72.1), I2 = 98.9%, N = 9 221 vs postgraduate: 86.6% (78.1–93.3), I2 = 86.5%, N = 726; psubgroup < 0.001 | Satisfaction with online evaluation might have been significantly higher for postgraduate learners compared to undergraduates. Postgraduate learners may have perceived the distant format as more flexible or even easier to prepare for, which are essential advantages, especially in the context of a likely busier schedule | |
By WHO region | AMR: 82.3% (70.3–91.8), I2 = 96.1%, N = 1 589 vs EUR: 87.3% (82.1–91.8), I2 = 65.9%, N = 632 vs EMR: 61.4% (41.0–79.9), I2 = 99.5%, N = 5 355 vs SEAR: 52.7% (37.5–67.5), I2 = 98.2%, N = 2 449 vs WPR: 55.0% (31.3–77.5), N = 882; psubgroup < 0.001 | Exactly as with training methods, learner satisfaction rates with virtual assessment might have been lower in the EMR and SEAR when compared to those in the AMR and EUR (data on WPR are limited and less credible). This reinforces the robustness of this review’s findings on regional differences in satisfaction rates and indicates that satisfaction may represent more the learners’ views on the distant format of the innovations rather than their primary aim (i.e., training or assessment). However, data on virtual innovations for assessment are far more limited than that focusing on virtual responses for education | |
Learner online vs face-to-face assessment scores | Previous/in-person vs virtual/new method | SMD = − 0.68 (− 0.96 to − 0.40), I2 = 98.1%, N = 12 513 | Learners likely achieved significantly higher scores when undertaking online assessment compared to pre-pandemic in-person evaluation methods. This finding may be attributed to easier examination formats, lower examination demands, given the circumstances, or inadequate supervision of participants |
Learners’ actual participation in volunteering activities | Overall | 27.7% (19.1–37.3), I2 = 99.7%, N = 39 046 | An encouraging rate of learners might have volunteered during the pandemic |
By training level | Undergraduate: 32.4% (20.6–45.4), I2 = 99.8%, N = 32 541 vs postgraduate: 9.1% (0.4–26.2), I2 = 99.0%, N = 2 059; psubgroup = 0.029 | Undergraduate learners might have volunteered at higher rates than their graduate counterparts. This finding may be attributed more to the availability of time of undergraduates rather than differences in willingness to volunteer | |
Learners’ intention to volunteer | Overall | 62.2% (49.2–74.4), I2 = 99.8%, N = 28 728 | A considerable rate of learners might have intended to volunteer during the pandemic, consisting of a valuable pool of available volunteers willing to assist, if needed |