This systematic review used the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [3]. The PRISMA guidelines are commonly used for multiple forms of research methodologies [4]. The authors also worked with a biomedical librarian on all search processes.

Inclusion criteria required the following items: first, a study that explicitly examined emotional intelligence as a predictor for success during matriculation in medical school; second, markers of success that were defined by written examination scores, continuous assessments, final examinations, weighted grade point average (wGPA), and association with success defined through traditional medical school admission criteria and failures; third, details provided about the sample including demographics, etc.; and fourth, published in English. Excluded papers were those that were examined a single construct of emotional intelligence (e.g., empathy).

Three databases were used during the search strategy: PubMed, Cumulative Index to Nursing and Allied Health Literature (CINAHL), and Education Resources Information Center (ERIC). All three were searched from their establishment to July 2016.

Search terms used in the systematic review were (MH “school admissions+”) AND (MH “emotional intelligence”) in CINAHL, “School admission criteria” [MeSH terms] OR “medical school” [all fields] AND “emotional intelligence” [MeSH terms] in PubMed, and Admission criteria and emotional intelligence in ERIC. Key terms used in each search included: school admissions, medical school, emotional intelligence, and admission criteria. The search was designed to be sensitive, and not specific, since emotional intelligence is recent terminology within the medical search engines.

One author (CJC) performed a title screen and abstract review in PubMed. Two authors (CJC and CEC) then completed full-text reads to determine appropriateness for inclusion of each article, with a 100% consensus. The same screening process was repeated in CINAHL and ERIC, with duplicate articles removed when found. One author (CEC) also performed a hand search of references from the included papers.

One author (CJC) read each of the studies for associations between emotional intelligence scores and the measured success subjects experienced in medical school, along with associations between emotional intelligence and traditional selection measures. This author extracted the relevant data, noting the statistical relationships to emotional intelligence.

The study included any papers with explicit mechanisms to evaluate emotional intelligence in self-report format. We considered the measures of Mayer-Salovey-Caruso Emotional intelligence Test (MSCEIT) [5], Situational Test of Emotional Understanding (STEU) [6], Situational Test of Emotion Management (STEM) [6], and the Emotional Intelligence Inventory (EQ-I) [7] as qualifying emotional intelligence measures. The MSCEIT is an ability-based test created to measure the four branches of the emotional intelligence model that was developed by Mayer et al. [5] and consists of 141 items and takes 30–45 minutes to complete, providing 15 main scores: total emotional intelligence score, two area scores, four branch scores, and eight task scores [5]. The STEU is a test developed by MacCann and Roberts [6] that examines emotional understanding, a key component of emotional intelligence. The test-taker is required to choose which five emotions is most likely to result from an emotional situation in each item. STEM, a test also developed by MacCann and Roberts [6], assesses emotion management, the capacity of individuals to regulate their emotions by reducing negative feelings and enhancing positive ones. Subjects are given a situation then must choose from four strategies to apply to improve emotions and manage the problem. EQ-I [7] was developed to assess the Bar-On model of emotional-social intelligence. This test is a self-report measure designed to examine a number of constructs related to emotional intelligence, consisting of 133 items and taking approximately 30 minutes to complete.

We accepted medical school success measures such as written examination scores, continuous program-based assessments, final examinations, wGPA, and failures such as failing grades and dropouts. With respect to medical school admission assessments, we accepted scores on interviews, Undergraduate Medicine and Health Sciences Admission (UMAT) scores, Tertiary Entrance Rank (TER), wGPA, and autobiographical test scores. UMAT is an admissions test that was developed by the Australian Council for Educational Research and used to assist with selection of students into the medicine, dentistry and health science degree programs at undergraduate level. The TER was a tertiary entrance score used as a tool for selection to universities in Australia that was replaced by the Australian Tertiary Admission Rank in 2010, in all territories except Queensland. The wGPA is a grade point average that takes into account the difficulty of classes and increases accordingly. Autobiographical scores could include materials such as personal statements, essays, and letters.

Methodological quality was assessed by one author (CJC) using the Newcastle-Ottawa Scale [8] for assessing the quality of nonrandomized studies in meta-analyses. A ‘star system’ is used through which a study is judged based on three broad perspectives: the selection of the study groups; the comparability of the groups; and the method of achieving the outcome of interest for cohort studies (Appendix 1).

Risk of bias across studies was evaluated by examining grey literature, conference proceeding, and other environments to identify research on emotional intelligence and medical admissions that had not been formally published.

Strength of association measures included any acceptable correlation coefficient (measure used depended on the data used), which represents the strength of the presumed linear association between the variables in question, and ranges from −1 to +1 [9]. A correlation coefficient of zero indicates that no linear relationship exists between two tested variables, whereas a value closer to +1 suggests a positive correlation and a value of -1 suggests a negative correlation.

This study was exempt from the local institutional human ethics review board of both institutions.

In total, 1,332 articles were initially identified through database search of PubMed, CINAHL, and ERIC. A hand search also resulted in seven articles. Once duplicates were removed, a total of 1,336 articles remained. After title screening and abstract review, 26 articles were pulled and reviewed in full. The final search yielded six manuscripts [10,11,12,13,14,15], analyzing emotional intelligence as a predictor for success in medical school or as a predictor for medical school admissions [16] (Fig. 1).

Total population across all studies was found to be 1,859 (range, 51 to 659). Three of the studies [10,11,12] were longitudinal and three were cross-sectional [13,14,15]. The medical students who participated in the studies included undergraduate medical students (i.e., Australia), graduate medical students, and those applying to and then matriculating into medical school. Success in medical school was defined in these studies by clinical competency assessments, bio-medical knowledge based assessments, scores on written examinations, objective structures clinical examinations (OSCEs), the number of nominations for excellence in clinical performance, number of failures, interpersonal academic performance, intellectual academic performance, tests of cognitive ability, continuous assessments, and final examinations (Table 1).

There were four tools used to measure emotional intelligence. These tools included: the MSCEIT [11,12,13,14,15]; EQ-I [11]; the STEU and the STEM [10]. Emotional intelligence was compared as a predictor to success to the UMAT [13], interview, TER scores [13], and other traditional measures such as wGPA and autobiographical sketch scores.

The second question of the Newcastle-Ottawa Scale was excluded as it was not applicable to the design of the studies. Because of the way in which emotional intelligence is measured, the ascertainment of emotional intelligence scores among all studies were considered written self-reports. Most studies lacked controls, with the exception of one in which nonresponse bias was controlled for by having applicants respond as part of their medical school curriculum. Further trends included a lack of adequate follow-up in all but one study; however, minimal follow-up was still long enough for outcomes to occur. All but one study assessed the outcomes through record linkage. Outcomes of interest (success in medical school or a correlation between emotional intelligence and traditional measures of acceptance) appeared not to be present at the start of any studies. All studies were similar in their results of risk of bias assessment (Table 2).

Four studies [10,11,12,14] examined emotional intelligence as a predictor for success in medical school. One study was cross-sectional [14], the remaining were longitudinal [10,11,12]. Chew et al. [14] found that students who were more emotionally intelligent tended to perform better in both the continuous assessments (r=0.24, P=0.003) and the final professional examination (r=0.21, P=0.01). Libbrecht et al. [10] found that emotional intelligence correlated positively with interpersonal academic performance (r=0.21, P<0.001). Humphrey-Murto et al. [12] found no significant correlations when evaluating associations between scores on the MSCEIT and written examination scores for interview and matriculated studies in two cohorts but did find a statistically significant positive correlation between the year 4-total OSCE score and the MSCEIT score (r=0.35, P<0.001). A final study [11] found that at first assessment EQ-I scores were negatively associated with performance on the clinical competency modules (r=-0.38, P<0.01). At a second assessment, an association between EQ-I scores and poor performance on the clinical competency module were found once again [11] (Table 1).

Two studies [13,15] examined traditional medical school admissions criteria and how it related to emotional intelligence; both studies were cross-sectional. One study found no significant correlations between emotional intelligence total and any of the selection scores for the first cohort [13]. The second study found a lack of substantial relationships between emotional intelligence scores and traditional medical school admission measures [15].

There were no instances in which grey literature was found, suggesting no publication bias. There were also no instances of selective reporting among accepted studies.