Attributes associated with training effectiveness based on >1 study |
Location of training activities: where HCPs routinely work (on-site) versus all training off-site | Direct evidence*: none. No head-to-head study examined this attribute. Indirect evidence*: having some or all training on-site was more effective than all training off-site by a mean of 6.0–10.4 %-points. Details in online supplemental appendix 1, table E, row 1.
|
Use of clinical practice as a training method | Direct evidence*: none. No head-to-head study examined this attribute. Indirect evidence*: training with clinical practice was more effective than training without clinical practice by a mean of 6.9–7.4 %-points. Details in online supplemental appendix 1, table E, row 2.
|
Time since training | Direct evidence*: change over time in the marginal effect of supervision given training was 0.3 %-points per month (p=0.58) for 0.5–5.5 months after training. Indirect evidence*: mean effect of training only (without supervision) decreased by 0.8–1.0 %-points per month after training, with the effect predicted to reach zero after 19.8–22.5 months, on average. Mean effect of training plus supervision did not decrease over time (there was a trend of increasing effect of 0.2–0.3 %-points per month, which was not statistically significant). The latter result was sensitive to outliers. Details in online supplemental appendix 1, table E, row 3, and figure C.
|
Baseline performance level | Direct evidence*: none. No head-to-head study examined this attribute. Indirect evidence*: mean effect of training decreased by 0.11–0.15 %-points for every 1 %-point increase in baseline performance level. Details in online supplemental appendix 1, table E, row 4.
|
Attributes associated with training effectiveness based on only 1 study (ie, interpret with caution) |
Tailoring in-service training to HCPs’ stage of readiness to change | Direct evidence*: training tailored to HCPs’ stage of readiness to change was more effective than non-tailored training by a median of 23.3 %-points. Indirect evidence*: none. The HCPPR database did not include this attribute. Details in online supplemental appendix 1, table E, row 5.
|
Preservice training with group feedback about pretraining evaluation results | Direct evidence*: preservice training with group feedback about pretraining evaluation results was more effective than with individual feedback by 19.0 %-points. Indirect evidence*: none. Modelling was not performed for preservice because there were too few studies. Details in online supplemental appendix 1, table F.
|
Training on a protocol versus training on clinical acumen | Direct evidence*: training on a protocol-based model (HCPs applied screening results to an algorithm), combined with supervision and integration of services, was more effective than training on clinical acumen (what HCPs did with screening results was left to their discretion), combined with supervision and integration of services, by 8.4 %-points. Indirect evidence*: none. The HCPPR database did not include this attribute. Details in online supplemental appendix 1, table E, row 6.
|
Attributes with an unclear association with training effectiveness because direct and indirect evidence was contradictory |
Trainee group size | Direct evidence*: small group training (ie, 2–14 participants) was somewhat more effective than large group training (ie, >14 participants), by a median of 5.3 %-points. Indirect evidence*: large group training was somewhat more effective than small group training by a mean of 5.8–6.1 %-points. Details in online supplemental appendix 1, table E, row 11.
|
Trainers with content expertise | Direct evidence*: training by trainers with content expertise (doctors) was slightly more effective than training by trainers without content expertise (paramedics), by 2.5 %-points. Indirect evidence*: training when ‘all trainers were content experts’ was less effective than when not all trainers were content experts, by a mean of 16.1 %-points. Details in online supplemental appendix 1, table E, row 12.
|
Use of non-interactive lectures as a training method | Direct evidence*: training with a non-interactive lecture or session (as a sole training method) was more effective than interactive training (as a sole training method) by a median of 5.0 %-points. Indirect evidence*: no significant association. However, all univariable regression coefficient β values were less than 5.0 %-points (range: −2.5 to 3.8 %-points; all non-significant). Details in online supplemental appendix 1, table E, row 13.
|
Interaction between in-service training duration and the topic complexity of the training | Direct evidence*: for training on single topics, training effectiveness might have increased with course duration; and for training on multiple topics, training effectiveness seemed unrelated to course duration. Indirect evidence*: for training on single topics, training effectiveness was unrelated to course duration; but for training on multiple topics, effectiveness increased with longer course duration. Details in online supplemental appendix 1, table E, row 14.
|