Does MMPI assessed at medical school admission predict psychological problems in later years?

Objective Psychological distress among medical students is related to personality. The Minnesota Multiphasic Personality Inventory (MMPI) is a common instrument used to assess personality and psychological problems during the medical school admission process in Thailand. The purpose of this study was to examine how the MMPI can predict medical students’ psychological problems including perceived stress, anxiety, depression, interpersonal difficulties as well as self-esteem in later years. Results Anxiety and depressive symptoms were predicted by the psychopathic deviation, psychasthenia, and schizophrenia scales of the MMPI, while perceived stress was predicted by schizophrenia scale of MMPI. Social introversion predicted interpersonal difficulties. No MMPI scale was found to predict self-esteem. Electronic supplementary material The online version of this article (10.1186/s13104-019-4524-5) contains supplementary material, which is available to authorized users.


Introduction
Medical education is challenging and demanding. Apart from intellectual ability, personality traits are believed to be important factors that affect medical school performance, psychological adaptation, motivation in medical education and psychological problems, which may interfere with study [1][2][3].
Assessing psychological problems and personality traits is a routine practice as a part of the medical students' admission process. The Minnesota Multiphasic Personality Inventory (MMPI) is an instrument that has been used globally to assess personality and psychological issues. Poor academic performance was related to the MMPI scales of hysteria, psychopathic deviate, and schizophrenia [4].
Currently, medical schools in Thailand use the MMPI as a tool to assess psychopathology during the student selection process. Although MMPI results typically are not used for admission decision, they help direct the attention of the interview committee to applicants who score high on certain MMPI scales.
Common psychological problems among medical students include anxiety, depression, and suicidal ideation, all of which relate to personality [2][3][4][5]. Evidence has shown that Thai medical students tended to have higher scores on most subscales of the MMPI by the third year of medical school [6]. This suggests that psychopathology can also relate to personality traits.
On the other hand, positive attributes such as selfesteem play important roles not only in medical performance [7], but also in influencing a student's sense of well-being [8]. Self-esteem has been shown to be related to extraversion and neuroticism [9], and some items of MMPI were drawn to form a self-esteem content scale in the MMPI-2 [10]. However, to the best of our knowledge, the relationship between MMPI clinical scales and selfesteem has not yet been explored.
This study examined Thai medical student's MMPI personality profiles, assessed during the admission process at a Chiang Mai medical school, in predicting changes regarding psychological problems and self-esteem during subsequent years. We hypothesized that some MMPI clinical scales would predict these changes. If so, employing the MMPI during the admission process would help to identify those students at risk of incurring psychological problems or diminished levels of self-esteem.

Methods Design
This study employed an observational and prospective design to explore using the MMPI to predict both positive and negative psychological outcomes.

Participants
Participants were medical students who were administered the MMPI as part of the medical school admission process. Two hundred and fifty five students, who passed the written entrance examination for the medical school, completed the MMPI before the admission interview. The MMPI results were then provided to the interview committee as additional data to aid in the committee's decision-making. Of the 250 students who was ultimately admitted into the medical school, 203 students participated in this study; however 201 students completed various psychological measures that assessed self-esteem, perceived stress, perceived social support, interpersonal problems, and psychological symptoms (Time 1). Participants were followed-up for changes in previously noted psychological variables and asked to complete the same measures when they were in their third year of medical school (Time 2). One hundred and ninety-six students completed the psychological measures during their third year of medical study (Additional file 1: Figure S1). Mean age of the participants was 18.64 years old (SD = .59), and 60.7% were female.
The Thai adaptation team concluded that the American norms were acceptable to use with Thai clients because prior work indicated that normal samples in Thailand scored in a similar range as the American subjects on the MMPI scales [6].

Perceived stress scale (PSS)
This scale measures how an individual perceives stress. It is a 10-item self-report instrument that uses a 5-scale Likert format 0 (never) to 4 (very often), with total score ranges from 0 to 40 [13]. Higher scores indicate greater perceived stress. The Thai version of the PSS-10 demonstrated good reliability and validity [14] and the present study yielded a Cronbach's alpha of .85.

Outcome inventory (OI-21)
The Outcome Inventory is a self-rating questionnaire that measures four common mental health problems: anxiety, depression, interpersonal difficulties and somatic complaints. It includes 21 questions assessed using Likert scales that range from 1 (not at all) to 5 (very much). A higher score indicates a higher level of psychopathology [15]. In the present study, only anxiety, depression and interpersonal difficulties were used and yielded a Cronbach's alpha of .89.

Rosenberg self-esteem scale (RSES)
The RSES is a self-rating questionnaire that measures self-worth or self-esteem. It includes 10 questions assessed using Likert scales that range from 1 (strongly disagree) to 4 (strongly agree). The Thai RSES demonstrated good validity and reliability, and the present study yielded a Cronbach's alpha of .88 [16].

Statistical analysis
Sociodemographic data such as sex and age were described by percentage and mean (SD). Pearson's or Spearman's rank correlation coefficients were used to examine the linear relationship between MMPI scales and outcome measures, i.e., anxiety, depression, perceived stress and self-esteem. Two-sample t-tests, ANOVA, Kruskal-Wallis or Mann-Whitney U tests were used to compare outcome measures between time 1 and time 2, as appropriate.
Because time serves as an important predictor in this case, multilevel linear models were applied to assess the relationship between the independent variables of interest including time and MMPI scales, while dependent (outcome) variables included anxiety, depression, interpersonality difficulties, perceived stress, and self-esteem. Fixed effects models were fitted first and then random intercepts and slopes were introduced using maximum likelihood methods. Intercepts and slopes for each subject were allowed to vary for the MMPI scales. All data were analyzed using IBM SPSS 22.0. The data were normally distributed (skewness and kurtosis between − 2 and + 2), and the final best fitting models are presented below. Table 1 shows the different outcomes between time 1 (on admission) and time 2 (year 3 of the study). No differences were found among psychological symptoms. However, self-esteem and perceived stress scale scores were significantly higher in year 3 than at admission (p < .05 and p < .001, respectively).

Results
Additional file 2: Table S1 shows the mean and standard deviation of the MMPI scales, which were mostly in an acceptable range (40 to 60), except for defensiveness, which was slightly, but predictable, higher. In terms of sex, no significant differences were found on outcome measures except for anxiety. On the MMPI scales, significant differences were found for hypochondriasis, depression, hysteria, masculinity/femininity and psychasthenia. Table 3 shows that after sex and age were accounted for, four MMPI scales related to the change in outcome scores: psychopathic deviation, psychasthenia, schizophrenia, and social introversion. Changes in the selfesteem scales were not predicted by any MMPI scales.

Discussion
The aim of the present study was to examine the predictive validity of the MMPI at admission against subsequent psychological problems and self-esteem. Even  though no significant differences were found concerning anxiety and depression between time 1 (year 1) and time 2 (year 3), we found the schizophrenia, psychopathic deviation, and psychasthenia scales predicted individual changes among these psychological distress scores. Overall, the schizophrenia scale was the best predictor of negative mental health. As expected, changes in interpersonal difficulties were predicted by social introversion MMPI scale. Interestingly, depression was not predicted by the MMPI depression scale. We suspect that the participants were aware of the role of the MMPI in the medical school admission process and hence tended to present themselves in a favorable light and underreport their symptoms for fear that they might not be granted entry into the medical school or be labeled as mentally unhealthy. This assumption is supported by the high correlation between the defensive scale and psychological problems. Notably, even though no difference was found regarding the means for anxiety and depression for the whole group, schizophrenia, psychopathic deviation and psychasthenia were shown to predict changes for participants. In addition, the schizophrenia, psychopathic deviation and psychasthenia scales have been shown to predict behavioral problems such as drug abuse. Related research showed that perceived stress was significantly higher in the medical students in the third year comparing to the first year [17]. However, to the best of our knowledge, no study has focused on associations between MMPI scales and changes in perceived stress. We found the schizophrenia scale predicted changes in perceived stress. Low scores on reality testing of the schizophrenia scale has been shown to relate to low scores on measures of emotional intelligence [18] and medical students with higher emotional intelligence experienced lower levels of stress [17].
High scores on the MMPI schizophrenia scale relate to unusual beliefs and eccentric behaviors, and do not necessarily mean that participants met the criteria of schizophrenia; instead, high scores (e.g., t-score of 65 or above) suggest personality processes associated with increased liability to developing schizophrenia-related illness [19]. High scores on the schizophrenia scale most often result from poor social skills, limited skill in judgment, and impaired logical thinking [20], making those with high scores more likely to experience psychological problems.
In medical settings, where the MMPI is most appropriately used, patients with high scores on schizophrenia, psychopathic deviation and psychasthenia scales should be monitored to allow for early intervention to prevent further psychological problems. Medical students in higher clinical years (years 4-6) tend to have more psychological problems due to clinical training encounter [21]. Hence, further investigation that includes higher clinical years may provide a more comprehensive picture of the value of the MMPI in predicting psychological problems encountered by medical students.

Conclusion
The MMPI predicted higher scores on measures of anxiety, depression, interpersonal difficulties and perceived stress. Schizophrenia, psychopathic deviation and psychasthenia were the MMPI scales found to be most predictive, as found in other studies. It may be useful to use the MMPI to identify students at risk of developing psychopathology so that appropriate interventions could be provided as early as possible.

Limitations
The MMPI, administered during the medical school admissions process, was likely to be biased because of the demand characteristics of the situation. Even though biases are somewhat corrected using regression analysis, it is rather difficult to assess the influence of situational variables. In addition, the results may not be generalized to other settings, because MMPI scores obtained on admission may not reflect the true personality and psychopathology of the respondents. Finally, an older version of the MMPI was used for this study, and further exploration using a more recent version of the test is warranted.

Additional files
Additional file 1: Figure S1. Flow chart of the study.
Additional file 2: Table S1. MMPI scale scores on admission.