Initial searches returned a total of 220 records. Of these, 29 fulfilled eligibility criteria and were included in the systematic review, with others excluded as indicated in the flow diagram provided in Fig. 1. Included studies were from a diverse array of countries, including 14 from India [10,11,12,13,14,15,16,17,18,19,20,21,22,23], 4 from the United States of America [24,25,26,27], 2 from Malaysia [28, 29], 2 from Saudi Arabia [30, 31], 2 from Oman [32, 33], 2 from Iran [34, 35], and one from each of Canada [36], the United Arab Emirates [37] and Palestine [38]. The sample size of the included studies ranged from 38 [10] to 425 [21]. The structure and delivery of crosswords employed varied substantially. Methodological quality varied with MERSQI scores ranging from 5.5 to 15.5 (mean score 10.1). The most common methodological limitations were studies being conducted at a single institution [10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38] and utilising only cross-sectional or post-test-only methodologies [10, 12,13,14,15,16,17,18,19, 21,22,23,24,25,26,27,28,29,30,31,32,33, 36, 37], and a number of studies also had unclear response rates [13, 15, 16, 19, 22, 24, 28, 30, 31, 34, 36, 37] (see Table 3, Supplementary Information D).

Flow diagram demonstrating study selection and exclusion rationale

There were seven studies that report of the conduct of randomised trials [11, 12, 15, 20, 33, 35, 38]. Six examined student performance on knowledge-based assessments in the crossword groups compared to another group [11, 12, 15, 20, 35, 38]. In the study of speech therapy students by Zamani et al., both groups had similar performance prior to the intervention, and one month after the educational intervention the group that received crosswords had a significantly higher test score than the traditional teaching group (18.26 vs 16.10, P = 0.001) [35]. Similarly, in Gaikwad & Tankhiwale, the interventional group had an absolute learning gain of 33.9% as compared to the control group having an absolute learning gain of 18.55% (statistical significance not presented) [11]. These improvements in knowledge as evaluated through test scores were supported by other studies that examined trainees’ perceptions of their knowledge gain. For example, in Sannathimmappa et al. the proportion of students who "strongly agreed" with a statement that "Solving crossword puzzles improved my examination scores" was 69.3% [33]. Conversely, Shenoy & Rao compared crosswords to student led tutorials, and demonstrated improved test scores in the tutorial group, compared with that of the crossword group [20], though there was no control group who received traditional teaching methods to allow for comparison with standard practice. In the five studies that presented results regarding student experience, the responses were consistently positive with students reporting they enjoyed crosswords and would like to undertake further crosswords in future [11, 15, 20, 33, 35].

Eight studies fulfilled inclusion criteria that employed either single-group pre- and post-test analyses [16, 17, 19, 23, 24, 27, 37] or non-randomised two group studies [34]. Four studies presented data on the educational impact on test scores. Three studies demonstrated an increase in test scores following the application of crossword puzzles [23, 24,