EVALUATION OF STATISTICAL ILLITERACY IN LATIN AMERICAN 1 CLINICIANS AND OF THE EFFICACY OF A 10-HOUR COURSE

Statistical Illiteracy is highly prevalent among Latin American clinicians. Short-term 47 educational interventions are effective but, their benefits quickly fade away if they are not 48 periodically reinforced. Medical boards and Medical schools need to periodically teach and 49 evaluate statistical proficiency to ameliorate these issues. 50


Introduction 24
All medics require statistical interpretation skills to keep up to date with the scientific 25 advances and evidence-based recommendations of their specific field. However, statistical 26 illiteracy among clinicians is a highly prevalent problem with far-reaching consequences. The 27 few available studies that report statistical literacy improvements after educational 28 interventions do not report for how long these benefits last. We measured for the first-time 29 statistical proficiency among Latin-American clinicians with different levels of training and 30 evaluated the efficacy of a 10-hour course at multiple timepoints. 31 Methods 32 Using an online questionnaire, we evaluated self-perceived statistical proficiency, scientific 33 literature reading habits and statistical literacy (using an adaptation of the Quick Risk Test) 34 across multiple levels of medical training. Separately, we evaluated statistical proficiency 35 among Internal Medicine residents at a tertiary centre in Mexico City immediately before, 36 immediately after and one and two months after a 10-hour statistics course using the same 37 adaptation (allowing for "I don't know" answers) of the Quick Risk Test. Scores across 38 multiple time points were compared using Friedman's Test. 39

Results 40
Data from 392 clinicians from 9 Latin American countries were analyzed. Most clinicians 41 (85%) failed our adaptation of the Quick Risk Test (mean score = 2.6/10, IQR:1.4). The 10-42 hour course significantly improved the scores of the Internal Medicine Residents (n=16) from 43 3.8/10, IQR:1.8 to 8.3/10, IQR:1.4 (p<0.01). However, scores dropped after one and two 44 months to 7.7/10, IQR:1.6 and 6.1 / 10, IQR:2.2, respectively. 45 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) Most medical schools and Medical board recognise the importance of Statistical Skills for 55 practising clinicians 1 . However, evidence shows that even experienced clinicians struggle 56 with assimilating the differences and implications of fundamental statistical concepts such as 57 odds ratio versus absolute risk and sensitivity versus positive post-test probability 2 . 58 Moreover, essential concepts such as absolute risk changes, number needed to treat/screen, 59 intention-to-treat analysis and Bayesian probability are often overlooked when making 60 clinical decisions and when explaining the implications of tests and treatments to patients 3,4 . 61

62
The implications of Statistical Illiteracy among clinicians are frequent and range from 63 generating individual ethical problems 5-7 to health-policy misinformed decisions 8 . Moreover, 64 improving health statistics among medical doctors has been put forward as one of the seven 65 goals for improving health during this century 9 . 66 67 Importantly, evidence also suggests that cheap, easy-to-implement and short-term 68 interventions can improve statistical skills among clinicians 10 . In their 2018 study, Jenny,69 Keller and Gigerenzer 11 demonstrated that a 90-minute training session in medical statistical 70 literacy improved the performance (from 50% to 90%) in 82% of the participants using a 71 multiple-choice Statistics test. However, it was not evaluated how quickly these 72 improvements fade away after the educational intervention. 73

74
In this study, we estimated Statistical Literacy among Latin American clinicians and 75 evaluated the efficacy of a 10-hour Statistics course across multiple timepoints. 76 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 14, 2021. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Since scores were not normally distributed, we compared them using Friedman's Test using 119 the R function "friedman.test" from the R package "stats" version 3.6.2. Normality was 120 evaluated using Shapiro-Wilk tests using the function "shapiro.test" in the same R package. 121 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Survey responses and Statistical Literacy results among Latin American clinicians. 123
A total of 403 responses were collected, however, 11 were discarded due to having 124 incomplete data. In total 392 from 9 different countries and 53 different medical schools were 125 included in the analysis. Most respondents (82%) were Mexican. Scores were not 126 significantly different across different levels of medical training. Table 1 summarizes their 127 answers in the survey and the overall performance in the test. Table 2  CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

FUNDING, DATA SHARING AND CONFLICT OF INTEREST DISCLOSURES 195
This study did not receive funding. The authors declare they do not have conflicts of interest 196 to disclose. Data for research purposes will be shared upon request to the corresponding 197 author. Dr Adrian Soto-Mota is the guarantor of the integrity of this work. 198 199 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 14, 2021. ; https://doi.org/10.1101/2021.05.08.21256882 doi: medRxiv preprint