Surgery for Obesity and Related Diseases ] (2014) 00–00

Original article

Validity of Minnesota Multiphasic Personality Inventory – 2 – Restructured Form (MMPI-2-RF) scores as a function of gender, ethnicity, and age of bariatric surgery candidates Ryan J. Marek, M.A.a,*, Yossef S. Ben-Porath, Ph.D.a, Martin Sellbom, Ph.D.b, John L. McNulty, Ph.D.c, Leslie J. Heinberg, Ph.D.d a Department of Psychological Sciences, Kent State University, Kent, Ohio Research School of Psychology, Australian National University, Canberra, Australia c Department of Psychology, University of Tulsa, Tulsa, Oklahoma d Cleveland Clinic Lerner College of Medicine, Department of Medicine, Cleveland, Ohio Received May 21, 2014; accepted October 3, 2014 b

Abstract

Background: Presurgical psychological screening is used to identify factors that may impact postoperative adherence and surgical outcomes in bariatric surgery candidates. Minnesota Multiphasic Personality Inventory – 2 Restructured Form (MMPI-2-RF) findings have demonstrated utility for this task. Objectives: To explore whether there are clinically meaningful gender, ethnicity, or age differences in presurgical MMPI-2-RF scores and the validity of these scores in bariatric surgery candidates. Methods: The sample was composed of 872 men and 2337 women. Ethnicity/race groups included 2,204 Caucasian, 744 African American, and 96 Hispanic individuals. A sample of 165 were not included in the ethnicity/race analyses because they were of another descent. Ages groups included 18–35 year olds (n ¼ 454), 36–49 year olds (n ¼ 1154), 50–64 year olds, (n ¼ 1246), and 65 years old or older (n ¼ 355). Validity data, obtained via a retrospective chart review, were available for a subset patients (n ¼ 1,268) who were similarly distributed. Step-down hierarchical regression analyses were conducted to assess for differential validity. Results: Bariatric surgery candidates produced comparable MMPI-2-RF scores in all subsamples, indicating that the test norms generalize across demographic groups. Validity findings were also generally comparable, indicating that MMPI-2-RF scores have the same interpretive implications in demographically diverse subgroups of bariatric surgery candidates. Conclusions: The MMPI-2-RF can assist in presurgical psychological screening of demographically diverse bariatric surgery candidates. (Surg Obes Relat Dis 2014;]:00–00.) r 2014 American Society for Metabolic and Bariatric Surgery. All rights reserved.

Keywords:

Obesity; MMPI-2-RF; Bariatric surgery candidates; Demographic characteristics; Diversity; Gender; Ethnicity; Age; Presurgical Psychological Screening; Assessment; Evaluation

Bariatric surgery has demonstrated good weight loss and quality of life outcomes for many surgical weight loss * Correspondence: R. J. Marek, M.A., Kent State University, Department of Psychological Sciences, 600 Hilltop Drive Kent, Ohio 44242, United States. E-mail: [email protected]

seeking patients; however, a subset of patients experience significant psychological difficulties both before and following the procedure [1–6]. Research indicates that psychopathology may be related to bariatric surgery outcomes [3–6], and the American Society for Bariatric and Metabolic Surgery recommends use of objective psychological assessment measures as part of a presurgical psychological

http://dx.doi.org/10.1016/j.soard.2014.10.005 1550-7289/r 2014 American Society for Metabolic and Bariatric Surgery. All rights reserved.

2

R. J. Marek et al. / Surgery for Obesity and Related Diseases ] (2014) 00–00

screening (PPS) [7]. The long-documented validity advantages offered by objective psychometric testing in conjunction with a clinical interview support more precise, empirically-validated information about the patient than a clinical interview alone [8]. Objective psychological assessment can aid or challenge a diagnostic consideration, provide clinical information on risk factors that the patient may be sensitive to disclosing during an interview, and some instruments can assess the extent to which a candidate is over- or underreporting symptoms [9]. Approximately 45% to 85% of clinics assessing bariatric surgery candidates use objective psychometric testing as part of a PPS [10]. The 2 most widely used instruments are the Minnesota Multiphasic Personality Inventory – 2 (MMPI-2) [11] and the Beck Depression Inventory – II [12]. A newer instrument, the MMPI-2 Restructured Form (MMPI-2-RF) [13,14], was developed in an effort to psychometrically improve the MMPI-2, provide less subject burden, revise the validity scales, and provide a straightforward interpretative strategy of the substantive scales for psychologists/psychiatrists. Restructuring the MMPI-2 involved identification and subsequent removal of general factor variance that has long been recognized as impairing the discriminant validity of the Clinical Scales of the instrument [15–16]. This general factor variance, termed demoralization, is conceptualized as a general distress construct (e.g., feeling sad, unhappy, and feeling overwhelmed) that limits the discriminant validity of many other psychological self-report instruments [17]. Removal of demoralization from the Clinical Scales of the MMPI-2 led to the development of the more homogenous Restructured Clinical (RC) Scales [18]. Following the construction of the RC Scales, a similar process using the MMPI2 item pool led to the development of MMPI-2-RF. The 338-item MMPI-2-RF provides less subject burden for test takers (25–35 minute computer administration as evidenced in normative samples) than the 567-item MMPI-2. The validity scales of the inventory allow the clinician to assess for protocol validity via 2 noncontent based responding scales (random and fixed responding), 5 overreporting scales, and 2 underreporting scales. Protocol validity reflects the extent to which the MMPI-2-RF substantive scale scores obtained by a specific test-taker are credible and interpretable. For example, if a patient scores 80 T-score points or higher on the Variable Response Inconsistency scale, the MMPI-2-RF substantive scale scores are considered invalid owing to excessive random responding, and, therefore, should not be interpreted [13]. Similarly, a score of 70 T-Score Points or higher on Adjustment Validity indicates that the patient underreported by presenting him or herself as remarkably well-adjusted [13]. Consequently, the absence of clinically elevated scores on the substantive scales cannot be interpreted as indicating the absence of problems assessed by those scales [13]. The more homogenous substantive scales of the MMPI-2RF assess broadband psychological constructs associated with many psychological disorders that are prevalent among

bariatric surgery candidates [1–6]. Specifically, the MMPI-2RF substantive scales assess constructs associated with emotional, behavioral, and thought dysfunction as well as various somatic/cognitive complaints and interpersonal functioning. The instrument also contains the Personality Psychopathology – 5 Scales, which are conceptually and empirically linked to the hybrid-trait model of personality disorders outlined in Section III of the Diagnostic and Statistical Manual of Mental Disorders, 5th Edition (DSM-5) [20]. Moreover, the straightforward interpretive strategy is ideal for prescreening bariatric surgery candidates. For example, an elevation on the Low Positive Emotions scale of the MMPI-2-RF indicates that the patient reported significant anhedonia and that the patient likely displays vegetative symptoms of depression that warrant further evaluation. Additionally, an elevation on the Low Positive Emotions scale points to possible treatment indications, such as the need to evaluate for antidepressant medication [13]. The MMPI-2-RF has demonstrated good reliability, validity, and generalizability of scale scores across bariatric surgery candidate samples [1,21–24]. For example, the RC Scales of the MMPI-2-RF demonstrate better reliability and discriminant validity coefficients than the original Clinical Scales of the MMPI-2 in a sample of bariatric surgery candidates [24]. Studies by Marek et al., [1] and Tarescavage et al. [23] present similar normative data and validity coefficients across 2 different bariatric surgery seeking samples. Moreover, associations have been reported between presurgical MMPI-2-RF scores and 1-and 3-month postoperative somatic concerns, psychological distress, and maladaptive eating behaviors [22]. Specifically, higher presurgical MMPI-2-RF scores on scales measuring emotional dysfunction were associated with greater psychological distress at follow-ups. Thus, patients with T-scores greater than 64 on MMPI-2-RF scales such as Demoralization, Self-Doubt, and Dysfunctional Negative Emotions should be viewed as being at increased risk for postoperative adjustment-related difficulties and feeling overwhelmed by the changes that follow bariatric surgery. Presurgical MMPI-2-RF scales assessing somatization were predictive of greater somatic concerns at both follow-ups, suggesting that patients who score above 64 on scales such as Somatic Complaints and Neurologic Complaints are likely to report a higher frequency of difficulties associated with nausea, pain, or other common medical complications following surgery. Lastly, higher presurgical scores on MMPI-2-RF scales measuring constructs associated with binge eating disorder (e.g., Dysfunctional Negative Emotions, Cognitive Complaints, Antisocial Behaviors, etc.) identify patients who have poor impulse control and have a low tolerance frustration before surgery and are at higher risk to be nonadherent to dietary restraints [22]. Such findings suggest the clinical utility of the instrument for treatment planing which, in turn, will lead to treatment interventions that may improve postoperative outcomes. An important step in evaluating the validity of an instrument is to determine whether it is demographically biased.

MMPI-2-RF Validity in Bariatric Surgery Candidates / Surgery for Obesity and Related Diseases ] (2014) 00–00

Differential predictive validity occurs when test scores systematically over or underpredict relevant criteria (intercept bias) as a function of group membership (i.e., gender, ethnicity, age, etc.) or when they are differentially associated with relevant extratest criteria (slope bias). If present, differential predictive validity could impact the utility of a measure for use in a population. For example, a test would be problematic if it was able to successfully assess for anxiety-related symptoms in women but unable to do so for men. Past studies have identified intercept bias in a small number of analyses when using the MMPI-2, with underprediction of psychopathology for African American men and women. However, the effect sizes reported were small to modest across all analyses [25,26]. To date, the MMPI-2-RF has not been evaluated for differential validity among bariatric surgery candidates. We sought to determine whether there are clinically meaningful differences between: 1) gender, 2) ethnicities (i.e., Caucasian, African American, and Hispanic patients) and 3) and age groups (i.e., 18–35 year olds, 36–49 year olds, 50–64 year olds, and Z65 year olds) seeking bariatric surgery on the validity and substantive scales of the MMPI-2-RF. We hypothesized that male and female candidates would produce comparable MMPI-2-RF scores, as has been reported in other settings [1,23,24]. We also hypothesized that there would be no meaningful differences (typically defined as 5 T score points or greater [27]) between MMPI-2-RF scale scores among ethnic groups, based on the inconsistent results reported across other samples using MMPI instruments [25,26]. Based on prior research findings [28], we hypothesized that older candidates would produce similar scores to those of younger patients. A second goal was to test for differential predictive validity of MMPI-2-RF scores as a function of gender, race/ethnicity, or age when predicting risk factors identified from a semistructured interview. Based on prior research [25,26] we hypothesized that there would be no evidence of slope bias and that any intercept bias would be of minimal magnitude.

Materials & Methods Participants Our initial sample included 3,341 consecutive patients who completed the MMPI-2-RF as part of a PPS at the Cleveland Clinic. Of these individuals, we excluded 132 patients ( 4.00 %) with invalid test results according to published MMPI-2-RF guidelines [14]: Cannot Say Z18, Variable Response Inconsistency – Revised Z80, True Response Inconsistency – Revised Z80, Infrequent Responses Z120, & Infrequent Psychopathology Responses Z100. Caucasians were more likely to produce an invalid MMPI-2-RF protocol than minorities [χ2 (1, N ¼ 3341) ¼ 4.48, P ¼ .03, φ ¼ .04], but this constitutes a small effect. No other statistically significant differences were found. The remaining sample (n ¼ 3,209) included 72.83% women and 27.17% men, 68.68% Caucasians, 22.84%

3

African Americans, 2.62% Hispanics, and 5.86% another ethnicity. 14.15% of individuals were young adults between the ages of 18–35 years old, 35.96% were mid-life adults between the ages of 36–49 years old, 38.83% were older adults between the ages of 50–64 years old, and 11.06% were 65 years of age or older. Extratest criterion data were available for 1,268 consecutively assessed patients. The remaining candidates were tested after the chart data used in this study were collected. This subsample was composed primarily of women (72.40%), and included 27.60% men: 68.45% Caucasian, 24.76% African Americans, 2.44% Hispanic, and 4.35% were of another ethnicity. The mean age of the sample was 49.81 years old (SD ¼ 11.59; Range 22–81). Patients’ mean presurgical body mass index was 49.45 kg/m2 (SD ¼ 10.93 kg/m2) and the sample had an average of 13.90 years of education (SD ¼ 2.52). The study was approved by the hospital’s Institutional Review Board. Measures Minnesota Multiphasic Personality Inventory – 2 Restructured Form [13,14]. The MMPI-2-RF is composed of 338-items, scored on 9 validity and 42 substantive scales. The test is theoretically grounded and organized in a hierarchical manner, consistent with contemporary psychopathology models. [19] Ben-Porath (2012) [29] provides a comprehensive review of the constructs assessed by the instrument, as well as an expansion on the method and rationale used to develop the MMPI-2-RF. The test has been validated in inpatient and outpatient mental health, forensic, medical, and nonclinical samples [14]. Binge Eating Scale (BES) [30]. The BES is a 16-item, self-report measure of binge eating severity with good internal consistency (Cronbach’s α ¼ .90) [31], high sensitivity, and adequate specificity in bariatric surgery candidate samples [32]. Semi-Structured Psycho-Diagnostic Interview. For a subset of patients included in this study (n ¼ 1,268), a semi-sturuced interview that is mandated by the Cleveland Clinic was administered by a doctoral level psychologist or a postdoctoral fellow under their supervision. Information obtained during the interview included: an evaluation of DSM-IV-TR diagnoses, including binge eating disorder, and a history and current mental health treatment including outpatient therapy, inpatient hospitalizations, and medication use. Furthermore, the interview assessed for a history of physical or sexual abuse, a history of past suicide attempts, and a history or current use of alcohol or other illicit substances. Procedure All patients underwent a standard PPS. Candidates were administered the MMPI-2-RF, BES, and a preoperative, semi-structured psycho-diagnostic interview. Criterion data

4

R. J. Marek et al. / Surgery for Obesity and Related Diseases ] (2014) 00–00

from the semi-structured interview were coded by trained research assistants using a retrospective chart review. Interrater reliability (Kappa and intraclass correlation coefficients) between coders was good [n ¼ 50; .96 (range ¼ 0.81–1.00)]. For the current investigation, we chose criteria that were deemed a priori to be relevant to the criteria based on previously reported findings [1]. These criteria and the MMPI-2-RF scales analyzed are included in Supplemental Tables 4-9. Analytic Plan Demographic Comparisons on MMPI-2-RF Scale Scores (n ¼ 3,209). MMPI-2-RF scale scores were compared for gender, ethnic group, and age group (Supplemental Tables 1–3). Because only 20 Hispanic men were available, they were not included in our analyses. Following convention, we considered a difference clinically meaningful if it was 5 T-score points or greater (corresponding to a medium effect size) [27]. If a clinically significant difference was found, we then compared, for the relevant groups, the percent of individuals who scored at or above designated interpretive cutoffs in the MMPI-2-RF manual [13]. Differential Validity Analyses (n ¼ 1,268). We tested for 2 aspects of differential predictive validity; slope and intercept bias, using step-down hierarchical multiple regression [33,34]. For each criterion and relevant MMPI2-RF scale, we compared a regression (prediction) model that included the MMPI-2-RF scale with one that included that scale, the potential moderator variable, and the interaction of the 2 (full model). All continuous predictors were centered around their means and centered values were also used in the computation of the moderating variable [35]. Ordinary least square (OLS) regression models were used for continuous variable, and R2 was examined to compare models (.010 ¼ Small Effect Size; .090 ¼ Medium Effect Size; .250 ¼ Large Effect Size). A significant increment in R2 obtained by the full model compared with the prediction model indicated bias. Logistic regression models were compared for dichotomous criteria. Because Pseudo-R2 s (e.g., Cox & Snell’s R2, McFadden’s R2) are unreliable estimates of variance accounted for in nonnormal regression models, [36] log-likelihood ratio tests were used instead. A chi-square value was computed for each model. To assess the model fit using the chi-square value, effect size ω’ (.10 ¼ Small Effect; .30 ¼ Medium Effect; .50 ¼ Large Effect) [37] was used to capture variance accounted in the logistic models. As with the OLS regression models, if there was a significant increment in ω’ obtained by the full model compared with the prediction model, this was an indication of bias. Across both types of models, if the test just described yielded an indication of bias, then follow-up tests for intercept and slope bias were conducted. In all analyses, at least a medium effect size was required to achieve clinical significance.

Results Means and standard deviations for the 51 MMPI-2-RF scales are reported by gender in Supplemental Table 1. Overall, men and women produced very similar scores. However, there were a few small gender differences that were consistent with what is reported in the MMPI-2-RF Technical Manual [14]. Women scored higher on the Symptom Validity – Revised and Multiple Specific Fears scales than men, and men scored higher than women on the Behavior/Externalizing Dysfunction and Mechanical/Physical Interests scales. In comparisons of the percent of individuals who scored at or above designated interpretive cutoffs in the MMPI-2RF manual [13] on the Symptom Validity–Revised scale there were no clinically meaningful differences. On the Behavior/Externalizing Dysfunction scale, a greater proportion of women than men produced a low score, but this difference was not clinically significant. On Behavior/ Externalizing Dysfunction, a larger percentage of men than women produced clinically elevated scores – a difference that is typically observed in other samples [15]. On the other hand, a greater proportion of women than men produced clinically elevated scores on multiple specific fears, which has also been reported in other settings [15]. Finally, a higher percentage of men than women produced clinically elevated scores on Disconstraint–Revised, but this difference was not clinically meaningful. Displayed in Supplemental Table 2 are means and standard deviations on the 51 MMPI-2-RF scales broken down by ethnicity within gender. The various ethnic groups produced very similar scores with a few exceptions. African American men scored higher than Caucasian men on Uncommon Virtues, whereas Caucasian men scored higher on Introversion/Low Positive Emotionality than did African American men. Among women, African Americans scored higher on Multiple Specific Fears and AggressivenessRevised than Caucasian women. The only clinically significant finding for Hispanic women was that they scored higher than Caucasian women on Uncommon Virtues. On scales on which significant ethnic or racial group differences were found, a greater proportion of African American men scored at or above the designated interpretive cutoff on Uncommon Virtues than did Caucasian men, but this difference was not clinically significant. Although negligible, a greater percentage of Caucasian men reached the threshold on Introversion/Low Positive Emotionality than African American men. On Multiple Specific Fears and Aggressiveness-Revised, a greater proportion of African American women scored at or above the designated cutoffs compared with Caucasian women; however, none of the differences were clinically meaningful. Listed in Supplemental Table 3 are age breakdowns within gender. The only clinically significant difference occurred between 18–35 year olds and 65þ year olds on

MMPI-2-RF Validity in Bariatric Surgery Candidates / Surgery for Obesity and Related Diseases ] (2014) 00–00 Table 1 Logistic regression analyses of bariatric surgery candidate criteria on Minnesota Multiphasic Personality Inventory - 2 - Restructured Form Scale Scores Effect sizes

Dependent variable MMPI-2-RF Scale n

* 0

ω

Prediction bias Δω0

Gender analyses Past suicide attempt(s) RCd 1268 .23 .13 MLS 1268 .20 .14 SUI 1268 .24 .15 Past psychotropic medication use RCd 1234 .29 .17 RC2 1234 .28 .18 RC7 1234 .24 .17 MLS 1234 .31 .18 Current psychotropic medication use RC2 1262 .21 .11 MLS 1262 .28 .12 History of outpatient Therapy NEGE-r 1218 .22 .13 Major Depressive Disorder RC2 1222 .27 .11 History of being physically abused FML 1233 .25 .12 Ethnicity analyses Past psychotropic medication use RCd 1149 .31 .21 RC2 1149 .29 .19 RC7 1149 .27 .21 MLS 1149 .32 .20 Current psychotropic medication use RC2 1176 .24 .17 MLS 1176 .31 .17 History of outpatient therapy RCd 1134 .25 .11 NEGE-r 1134 .21 .13 Age analyses History of outpatient therapy NEGE-r 1218 .21 .11 No current Exercise program MLS 1172 .21 .11

Slope bias Δω0

Threshold bias Δω0

.03 .03 .04

.13 .14 .15

.04 .02 .03 .01

.17 .18 .17 .18

.01 .02

.11 .12

.04

.09

.08

.08

.00

.12

.07 .06 .04 .05

.20 .18 .20 .19

.05 .05

.16 .16

.07 .04

.08 .09

.06

.01

.01

.11

FML ¼ Family Problems; MLS ¼ Malaise; NEGE-r ¼ Negative Emotionality/Neuroticism – Revised; RC2 ¼ Low Positive Emotions; RC7 ¼ Dysfunctional Negative Emotions; RCd ¼ Demoralization; SUI ¼ Suicidal/Death Ideation. * 10 ¼ Small Effect Size; .30 ¼ Medium Effect Size; .50 Large Effect Size.

5

conducted, including the resulting beta weights, log-likelihood ratio statistics, and P–values, are reported in Supplemental Tables 4-9. 12 of the 40 gender bias analyses reached statistical significance; however, effect sizes were insignificant to small. There was no evidence of slope bias. Small effect sizes for intercept bias were detected. In all cases, results indicated an overprediction of the criteria for men. Specifically, Demoralization, Malaise, and Suicide/Death Ideation demonstrated a small degree of overprediction of past suicide attempts. Demoralization, Low Positive Emotions, Dysfunctional Negative Emotions, and Malaise overpredicted past psychotropic medication use and a Major Depressive Disorder diagnosis and Low Positive Emotions and Malaise overpredicted current psychotropic medication use. Inefficacy overpredicted a Major Depressive Disorder diagnosis as well. Demoralization and Negative Emotionality/Neuroticism – Revised overpredicted a history of outpatient therapy. Family Problems overpredicted a history of being physically abused. To reiterate, none of these statistically significant findings would result in clinically meaningful prediction errors. Of the 40 analyses for ethnicity, 10 indicated statistical significance, but produced a small effect size at most. There was evidence of slope bias for Demoralization predicting past psychotropic medication use – indicating better prediction for Caucasians. However, the effect size was small and clinically insignificant. Statistical significance for intercept bias was found in the 10 analyses, with evidence of small magnitude under prediction of criteria in African Americans. Specifically, models containing Demoralization, Low Positive Emotions, Dysfunctional Negative Emotions, and Malaise indicated an underprediction of past psychotropic medication use, whereas Low Positive Emotions and Dysfunctional Negative Emotions underpredicted current Table 2 OLS regression analyses of bariatric surgery candidate criteria on MMPI-2RF substantive scales Effect sizes

Dependent variable MMPI-2-RF Scale n

*

R2

Prediction bias ΔR2

Slope bias ΔR2

Intercept bias ΔR2

o .001 .001

.010 .013

.001

.015

Ethnicity analyses Binge Eating Scale score RC7 FML

1095 .099 1095 .036

.010 .013

Age analyses

Multiple Specific Fears. Specifically, women 65 or older scored higher than 18–35 year old women on Multiple Specific Fears. When comparing the proportion of individuals who scored at or above the recommended interpretative cutoffs, women in the 65 or older group scored meaningfully higher than women 18–35 years old. Statistically significant findings for differential predictive validity analyses are reported in Tables 1 and 2. All analyses

BMI at intake MLS

1221 .049

.014

BMI ¼ Body Mass Index; FML ¼ Family Problems; MLS ¼ Malaise; MMPI-2-RF ¼ Minnesota Multiphasic Personality Inventory–2 Restructured Form; OLS ¼ Ordinary Least Square; R2 ¼ Coefficient of Determination Adjusted for Sample Size; RC7 ¼ Dysfunctional Negative Emotions. * 010 ¼ Small Effect Size; .090 ¼ Medium Effect Size; .250 ¼ Large Effect Size.

6

R. J. Marek et al. / Surgery for Obesity and Related Diseases ] (2014) 00–00

psychotropic medication use. Demoralization and Negative Emotionality/Neuroticism-Revised underpredicted a history of outpatient psychotherapy and Dysfunctional Negative Emotions and Family Problems underpredicted Binge Eating Scale scores. Of the 40 analyses for age, 3 indicated statistical significance with small effect sizes indicating prediction bias. There was evidence of slope bias for Negative Emotionality/Neuroticism-Revised predicting history of outpatient therapy – indicating better prediction for younger individuals. Evidence of intercept bias was found in 3 analyses. Specifically, Negative Emotionality/Neuroticism-Revised overpredicted a history of outpatient therapy for older adults, and Malaise overpredicted older adults not exercising and having a higher presenting body mass index.

Discussion The 3,209 bariatric candidates included in our demographic analyses produced very similar results on the MMPI2-RF across all comparison groups. Their scores were also very similar to those reported in the MMPI-2-RF Technical Manual [14] for a different sample of bariatric surgery candidates, reflecting the generalizability of MMPI-2-RF results for this population. Mean scores were within normal limits for all groups, with the exception of the Malaise scale, a measure reflecting the extent to which individuals present as being physically debilitated. This finding is similar to those reported in other medical samples, such as spine surgery and spinal cord stimulator candidates [38,39]. In the gender comparisons, women scored higher than men on Symptom Validity-Revised, however virtually none of the men or the women in this sample scored above the cutoff recommended for use with individuals with significant medical problems (T score 100). With regard to ethnic and racial group comparisons, here too, very few meaningful scale score differences were found, and those that did emerge did not generalize across gender. In the age group comparisons, the patients included in our sample produced remarkably similar MMPI-2-RF scores across all age ranges. Predictive validity analyses yielded mostly nonsignificant findings with a sample size that provided ample power. For the relatively small number of criteria where significant results were obtained, effect sizes were generally in the small to clinically insignificant range. In most of these cases, intercept, but not slope bias, was indicated; that is MMPI-2-RF scores were comparably valid, with only inconsequential levels of over or underprediction of criteria detected. In the gender analyses, negligible to smallmagnitude intercept differences indicated overprediction of a number of criteria for men. For race/ethnicity, smallmagnitude underprediction of the criteria was found for African Americans. For age, small-magnitude overprediction of criteria for older adults was found in 3 analyses.

In contrast with the demographic variables, MMPI-2-RF scale scores accounted for most of the predicted variance in the study criteria. Moreover, the bias effect sizes were uniformly negligible to small. These results, coupled with the considerable similarity in mean scores across groups, indicate that MMPI-2-RF findings can be interpreted similarly for women and men, African Americans, Hispanics, and Caucasians, and for bariatric surgery candidates of all age groups. The primary limitation of the present investigation is that our sample came from a single setting. However, the sample was broad and included large numbers of older adults and minorities. Nonetheless, replication with other bariatric surgery candidate samples can further bolster our conclusions. Another limitation is that our criteria included only crosssectional risk factors. Prospective surgical outcome data were not available for this investigation. It should be noted that rather than attributing the predictive bias to the particular MMPI-2 RF scale alone, an equally plausible explanation is that the relevant predictive criteria used to identify the scale's slope or intercept bias are themselves biased based on demographic or ethnic factors. That is, interviewers evaluating bariatric surgery patients may themselves be under-recording the presence of criteria based on demographic or ethnic characteristics of the patient. Nonetheless, the absence of meaningful bias in these cross-sectional analyses bodes well for the prospect that similar results will be obtained in prospective studies of surgical outcome prediction, which should be the focus of future research. Conclusions Overall, these results contribute to an accumulating body of literature available to guide use of the MMPI-2-RF in presurgical psychological screenings of bariatric surgery candidates [1,22–25]. Consistent with American Society for Bariatric and Metabolic Surgery guidelines, the MMPI-2-RF assesses a broad range of psychological risk factors that may not always be detected in a clinical interview (for example, due to limited face-to-face time or reluctance of candidates to disclose information they may not feel comfortable discussing in a face-to-face interview). The results of this study indicate that MMPI-2-RF test results can guide, and provide support for treatment recommendations, and other important clinical decisions pertaining to the presurgical psychological evaluation of demographically diverse bariatric surgery candidates. Disclosures Yossef Ben-Porath is a paid consultant to the MMPI-2RF publisher, the University of Minnesota and Distributor, Pearson. As co-author of the MMPI-2-RF, he receives royalties on sales of the test. Portions of this manuscript were presented at the Society for Personality Assessment Research Symposium in

MMPI-2-RF Validity in Bariatric Surgery Candidates / Surgery for Obesity and Related Diseases ] (2014) 00–00

Arlington, VA (March, 2014) and the 48 th Annual MMPI Symposium on Recent Research in Scottsdale, AZ (April, 2014). Appendix Supplementary data Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j. soard.2014.10.005. References [1] Marek RJ, Ben-Porath YS, Windover A, et al. Assessing Psychosocial Functioning of Bariatric Surgery Candidates with the Minnesota Multiphasic Personality Inventory-2 Restructured Form (MMPI-2RF). Obes Surg 2013;23:1864–73. [2] Mitchell JE, Selzer F, Kalarchian MA, et al. Psychopathology before surgery in the longitudinal assessment of bariatric surgery-3 (LABS3) psychosocial study. Surg for Obes Relat Dis 2012;8:533–41. [3] Semanscin-Doerr DA, Windover A, Ashton K, Heinberg LJ. Mood disorders in laparoscopic sleeve gastrectomy patients: does it affect early weight loss? Surg for Obes Relat Dis 2010;6:191–6. [4] de Zwaan M, Enderle J, Wagner S, et al. Anxiety and depression in bariatric surgery patients: a prospective, follow-up study using structured clinical interviews. J Affect Disord 2011;133:6–8. [5] Colles SL, Dixon JB, O'Brien PE. Grazing and loss of control related to eating: two high-risk factors following bariatric surgery. Obesity 2008;16:615–22. [6] White MA, Kalarchian MA, Masheb RM, Marcus MD, Grilo CM. Loss of control over eating predicts outcomes in bariatric surgery patients: a prospective, 24-month follow-up study. J Clin Psychiatry 2010;71:175–84. [7] Block AR, Sarwer DB. Presurgical Psychological Screening: Understanding Patients, Improving Outcomes. In: Block AR, Sarwer DB, eds. 1st ed. Washington, D.C.: American Psychological Association; 2013. [8] Grove WM, Meehl PE. Comparative efficiency of informal (subjective, impressionistic) and formal (mechanical, algorithmic) prediction procedures: The clinical-statistical controversy. Psychol. Public Policy Law 1996;2(2):293–323. [9] LeMont D, Moorehead M, Parish M, Reto CS, Ritz SJ. Suggestions for the pre-surgical psychological assessment of bariatric surgery candidates. American Society for Bariatric Surgery 2004:1–29. [10] Heinberg LJ. The role of psychological testing for bariatric/metabolic surgery candidates. Bariatric times: Clinical developments and metabolic insights in total bariatric patient care. 2013. Available online at http://bariatrictimes.com/the-role-of-psychological-testingfor-bariatricmetabolic-surgery-candidates/. [11] Butcher JN, Graham JR, Ben-Porath YS, Tellegen A, Dahlstrom WG. Minnesota Multiphasic Personality Inventory–2 (MMPI-2): Manual for administration and scoring. Rev. ed. Minneapolis, MN: University of Minnesota Press; 2001. [12] Beck AT, Steer RA, Brown GK. Manual for the Beck Depression Inventory-II. San Antonio, TX: Psychological Corporation; 1996. [13] Ben-Porath YS, Tellegen A. The Minnesota Multiphasic Personality Inventory-2 Restructured Form (MMPI-2-RF): Manual for administration, scoring, and interpretation. Minneapolis, MN: University of Minnesota Press; 2008/2011. [14] Tellegen A, Ben-Porath YS. The Minnesota Multiphasic Personality Inventory-2 Restructured Form (MMPI-2-RF): Technical Manual. Minneapolis, MN: University of Minnesota Press, 2008/2011.

7

[15] Jackson DN. The dynamics of structured personality tests. Psychological Review 1971;78(3):229–48. [16] Meehl PE. Reactions, reflections, projections. In: Butcher JN, editor. Objective personality assessment: Changing perspectives. Oxford, UK: Academic Press; 1972. p. 131–89. [17] Tellegen A. Structures of mood and personality and their relevance to assessing anxiety, with an emphasis on self-report. In: Tuma AH, Mason J, eds. Anxiety and the Anxiety Disorders. Hillsdale, NJ: Routledge; 1985. p. 681–706. [18] Tellegen A, Ben-Porath YS, McNulty JL, Arbisi PA, Graham JR, Kaemmer B. The MMPI-2 Restructured Clinical (RC) scales: Development, validation, and interpretation. Minneapolis, MN: University of Minnesota Press; 2003. [19 ] Krueger RF, Markon KE. Reinterpreting comorbidity: a model-based approach to understanding and classifying psychopathology. Annu Rev Clin Psycho 2006;2:111–33. [20] Anderson JL, Sellbom M, Bagby RM, Quilty LC, Veltri CO, Markon KE, Krueger RF. On the convergence between PSY-5 domains and PID-5 domains and facets: implications for assessment of DSM-5 personality traits. Assessment 2013;20:286–94. [21] Marek RJ, Ben-Porath YS, Ashton K, Heinberg LJ. Minnesota multiphasic personality inventory-2 restructured form (MMPI-2-RF) scale score differences in bariatric surgery candidates diagnosed with binge eating disorder versus BMI-matched controls. Int J Eat Disord 2014;47:315–9. [22] Marek RJ, Ben-Porath YS, Merrell J, Ashton K, Heinberg LJ. Predicting One and Three Month Postoperative Somatic Concerns, Psychological Distress, and Maladaptive Eating Behaviors in Bariatric Surgery Candidates with the Minnesota Multiphasic Personality Inventory-2 Restructured Form (MMPI-2-RF). Obes Surg in press. [23] Tarescavage AM, Wygant DB, Boutacoff LI, Ben-Porath YS. Reliability, Validity, and Utility of the Minnesota Multiphasic Personality Inventory–2–Restructured Form (MMPI–2–RF) in Assessments of Bariatric Surgery Candidates. Psychological Assessment 2013;25:1179–94. [24] Wygant DB, Boutacoff LI, Arbisi PA, Ben-Porath YS, Kelly PH, Rupp WM. Examination of the MMPI-2 Restructured Clinical (RC) Scales in a Sample of Bariatric Surgery Candidates. Journal of Clinical Psychology in Medical Settings 2007;14(3):197–205. [25] Arbisi PA, Ben-Porath YS, McNulty J. A comparison of MMPI-2 validity in African American and Caucasian psychiatric inpatients. Psych Assess 2002;14:3–15. [26] Monnot MJ, Quirk SW, Hoerger M, Brewer L. Racial bias in personality assessment: Using the MMPI-2 to predict psychiatric diagnoses of African American and Caucasian chemical dependency inpatients. Psych Assess 2009;21:137. [27] Graham JR. MMPI-2: Assessing Personality and Psychopathology, 5th ed. New York, NY: Oxford University Press; 2011. [28] Heinberg LJ, Ashton K, Windover A, Merrell J. Older bariatric surgery candidates: is there greater psychological risk than for young and midlife candidates? Surg Obes Relat Dis 2012;8:616–22 [29] Ben-Porath YS. Interpreting the MMPI-2-RF. Minneapolis, MN: University of Minnesota Press; 2012. [30] Gormally J, Black S, Daston S, Rardin D. The Assessment of Binge Eating Severity among Obese Persons. Addict Behav 1982;7: 47–55. [31] Ashton K, Drerup M, Windover A, Heinberg L, Brief four-session group. CBT reduces binge eating behaviors among bariatric surgery candidates. Surg Obes Relat Dis 2009;5:257–62. [32] Grupski AE, Hood MM, Hall BJ, Azarbad L, Fitzpatrick SL, Corsica JA. Examining the binge eating scale in screening for binge eating disorder in bariatric surgery candidates. Obes Surg 2013;23:1–6. [33] Nunnally J, Bernstein I. Psychometric Theory, 3rd ed. New York, NY: McGraw-Hill, 1994.

8

R. J. Marek et al. / Surgery for Obesity and Related Diseases ] (2014) 00–00

[34] Lautenschlager GJ, Mendoza JL. A Step-Down Hierarchical Multiple Regression Analysis for Examining Hypotheses About Test Bias in Prediction. Appl Psych Meas 1986;10:133–9. [35] Tabachnick BG, Fidell LS. Using Multivariate Statistics, 6th ed. Pearson Education Inc.; 2012. [36] Long J.S. Regression models for categorical and limited dependent variables: Sage; 1997. [37] Cohen J. Statistical power analysis for the behavioral sciences. 2nd ed. New York, NY: Psychology Press; 1988.

[38] Block AR, Ben-Porath YS, Marek RJ. Psychological risk factors for poor outcome of spine surgery and spinal cord stimulator implant: a review of the literature and their assessment with the MMPI-2-RF. Clin Neuropsychol 2013;27:81–107. [39] Marek RJ, Block AR, Ben-Porath YS. The Minnesota Multiphasic Personality Inventory–2–Restructured Form (MMPI-2-RF): Incremental Validity in Predicting Early Postoperative Outcomes in Spine Surgery Candidates. Psychological Assessment 2014. http://dx.doi. org/10.1037/pas0000035.

Validity of Minnesota Multiphasic Personality Inventory – 2 – Restructured Form (MMPI-2-RF) scores as a function of gender, ethnicity, and age of bariatric surgery candidates.

Presurgical psychological screening is used to identify factors that may impact postoperative adherence and surgical outcomes in bariatric surgery can...
289KB Sizes 6 Downloads 4 Views

Recommend Documents