Millon Behavioral Medicine Diagnostic

Chronic Pain: (Fishbain et al., 2001; n = 185, male = 95; Chronic pain patients with myofascial pain syndrome)

• Base rate of 75 with scores of 75 or greater indicating presence of the characteristic that the scale is measuring.

Chronic Pain: (Labbé et al., 1989;  n = 247 (132 male, 115 female); Mean Age = 47.0 (males) and 50.8 (females)—age difference was not significant) 

• Scores for chronic pain patients and the normative population (controls) were not significantly different on most MBHI scales. However, some differences did exist for the male and female groups:
  • Female pain patients were less likely to exhibit a cooperative coping style than female controls (p < 0.05)
  • Female pain patients were more likely to indicate the psychogenic attitudes of future despair and somatic anxiety than female controls (p < 0.05)
  • Male pain patients were more likely to exhibit an introversive coping style (p < 0.05) and less likely to demonstrate cooperative (p < 0.05) or sociable (p < 0.01) coping styles
  • Male pain patients were more likely to indicate the psychogenic attitudes of premorbid pessimism and somatic anxiety than male controls (p < 0.05), as well as future despair (p < 0.01)

Concurrent validity:

Lower Back Pain: (Lee-Riordan & Sweet, 1994; n = 60, male = 44, mean age = 47.2 (13.3), subjects sought treatment at multidisciplinary pain clinic, low-back pain of at least 6 months duration as primary complaint, mean duration of pain = 75.9 (108.2) months, pain rated as it usually occurs from 1 = mild to 5 = unbearable)

• Excellent negative correlations between MBHI scales and the Minnesota Multiphasic Personality Inventory (MMPI) K scale (p < 0.001):
  • Premorbid Pessimism (r = -0.63)
  • Future Despair (r = -0.64)
  • Social Alienation (r  = -0.61)
  • Allergic Inclination (r = -0.64)
  • Cardiovascular Tendency (r = -0.70)
  • Life Threatening Reactivity (r = -0.64)
  • Emotional Vulnerability (r = -0.68)
• Adequate negative correlations between MBHI scales and the MMPI K scale (p < 0.001):
  • Chronic Tension (r = -0.50)
  • Somatic Anxiety (r = -0.40)
  • Gastrointestinal Susceptibility (r = -0.59)
  • Pain Treatment Responsivity (r = -0.56)
• Excellent positive correlations between MBHI scales and the Minnesota Multiphasic Personality Inventory (MMPI) F scale (p < 0.001):
  • Future Despair (r = 0.61)
  • Pain Treatment Responsivity (r = 0.67)
  • Life Threatening Reactivity (r = 0.65)
• Adequate positive correlations between MBHI scales and the MMPI F scale (p < 0.001):
  • Premorbid Pessimism (r = 0.59)
  • Social Alienation (r = 0.58)
  • Somatic Anxiety (r = 0.32)
  • Allergic Inclination (r = 0.59)
  • Gastrointestinal Susceptibility (r = 0.49)
  • Cardiovascular Tendency (r = 0.59)
  • Emotional Vulnerability (r = 0.56)
• Excellent correlations between MBHI Pain Treatment Responsivity Scale (PP) and MBHI Coping Styles (p < 0.001)
  • Positively correlated with Inhibited Style (r = 0.72) and Sensitive Style (r = 0.79)
  • Negatively correlated with Sociable Style (r = -0.67) and Confident Style (r = -0.62)
• Adequate to Excellent  positive correlations between MBHI Pain Treatment Responsivity Scale (PP) and other MBHI scales (p < 0.001):
  • Premorbid Pessimism (r = 0.72)
  • Future Despair (r = 0.80)
  • Social Alienation (r = 0.69)
  • Allergic Inclination (r = 0.62)
  • Gastrointestinal Susceptibility (r = 0.55) 
  • Cardiovascular Tendency (r = 0.71)
  • Life Threatening Reactivity (r = 0.75)
  • Emotional Vulnerability (r = 0.76)

Discriminant validity:

Chronic Pain: (Gatchel et al., 1985; n = 23 (headache patient group), n = 24 (other chronic pain group), n = 21 (normal controls)

• Four MBHI scales significantly differentiated between the headache pain group and the other chronic pain group:
  • Inhibited Style: (F(1,67) = 4.49, p < 0.05)
  • Sensitive Style: (F(1,67) = 4.76, p < 0.05)
  • Chronic Tension: (F(1,67) = 10.01, p < 0.01)
  • Allergic Inclination: (F(1,67) = 4.22 p < 0.05)

• Based on 2,113 medical patients (Millon et al.,1982) 
• The Reliability Check of the MBHI suggested a valid profile (Sweet et al., 1985). This measure detects whether or not respondents may be responding in random fashion. If a respondent indicates “Yes” to any of the Reliability Check items their profile is labeled “questionable” and considered unreliable. If their score is two or more, then their assessment is marked “invalid” (Lee-Riordan & Sweet, 1994).
• It is of note that scales have considerable content overlap and may measure highly similar dimensions (Lee-Riordan & Sweet, 1994). 

Few patients have base rates over 75 (denotes presence of construct) on any of the MBHI scales. This could indicate a “floor effect” (Murphy, Sperr, & Sperr, 1986). 

Medical Patients and Non-Clinical Subjects: (Millon et al., 1982, n = 2,565 (2,113 patients with a variety of medical conditions and 452 subjects from non-clinical settings, MBHI)

• Acceptable test-retest reliability for all MBHI subscales (ICCs ranged from 0.76 to 0.90), with the exception of the Emotional Vulnerability scale (ICC = 0.59).

Medical Patients: (Millon et al., 2006; n = 720, female = 52%, patient conditions included: heart problems, cancer, diabetes, gynecological, problems, chronic pain, accident/injury, back pain, headaches, neurological problems, gastrointestinal problems, organ transplants, and HIV/AIDS, MBMD)

• Acceptable test-retest reliability for all MBMD subscales (ICCs ranged from 0.71 to 0.92)

Medical Patients and Non-Clinical Subjects: (Millon et al., 1982, n = 2,565 (2,113 patients with a variety of medical conditions and 452 subjects from non-clinical settings, MBHI)

• Adequate to Excellent internal consistency for nearly all MBHI subscales (Kuder-Richardson-20 (KR-20) coefficients ranged from 0.72 to 0.90), with the exception of the Cooperative (KR-20 = 0.68) and Confident (KR-20 = 0.66) coping styles.

Medical Patients: (Millon et al., 2006; n = 720, MBMD)

• Adequate to Excellent internal consistency for most MBMD subscales (Cronbach’s alpha ranged from 0.74 to 0.89).
• Poor internal consistency for Coping Styles (Cronbach’s alpha):
  • Cooperative (0.68)
  • Sociable (0.54)
  • Confident (0.61)
  • Non-Conforming (0.67)
  • Forceful (0.67)
  • Respectful (0.62)
• Poor internal consistency for Treatment Prognostics (Cronbach’s alpha):
  • Medication Abuse (0.65)
  • Information Discomfort (0.47)
  • Problematic Compliance (0.62)

Bariatric Surgical Candidates: (Walfish, S. et al., 2008; examination of reliability data published in Miller Behavioral Medicine Diagnostic (MBMD) manual)

• Poor: Cronbach’s alpha < 0.70 for 16 of the 32 scales: disclosure (0.54), desirability (0.47), debasement (0.69), introversive (0.65), cooperative (0.63), sociable (0.59), confident (0.61), nonconforming (0.58), forceful (0.58), respectful (0.56), medication abuse (0.59), information discomfort (0.22), utilization access (0.68), problematic compliance (0.55), adjustment difficulties (0.69), and psych referral (0.64)
  • Recommended use of alternative measures to assess the 16 constructs with questionable reliability

Bariatric Surgical Patients: (Hoyt & Walter, 2022; n = 224, mean age = 40.8 (12.5), female = 198 (88.4%), BMI was retrospectively recorded from medical visits over a 60-month time period at ten different time points: pre-surgical baseline, 3, 6, 9, 12, 18, 24, 36, 48, and 60 months.)

• When all MBMD scales were included simultaneously, Significantly worse fit was observed for the model using MBMD scales (-2LL = 10176.14) than for the model based on demographic factors (-2LL = 9074.84) (Chi-square = 1101.3, p < 0.001)
• With each MBMD scale entered separately, the Bayesian information criterion (BIC) values for each model showed that the addition of each individual MBMD scale resulted in significantly worse fit than the model based on demographic factors for all scales, with BIC values increasing by an average of 38.9 (SD = 2.2) across all models (range = 31.8-42.0)
• Overall results indicated that pre-surgical MBMD scales do not predict BMI outcomes over a 5-year period in bariatric surgery candidates.  

Predictive Validity:

Cardiac transplant candidates and recipients: (Brandwin, M. et al., 2000; 103 patients referred for heart transplant evaluation, mean age = 48.93, male = 84 (82%), 37 (36%) receiving transplant, all patients tested with MBHI as part of comprehensive initial work-up for transplant candidacy.

• Main effect for cluster based on MBHI profiles in which all scales across the three domains in which higher scores indicate increased psychological stress/higher stress-related illness vulnerability were higher for one cluster (high distress) than the other cluster (low distress), p < 0.001.
  • Significant differences in mortality by cluster membership at 1 and 5 year follow-up, Chi-square = 8.93, p < 0.005, and 8.16, p < 0.005, respectively, with high distress cluster patients having a greater proportion of deaths in each group.
  • Cluster membership was a significant predictor of mortality at 5-year follow-up, Chi-square = 10.59, p < 0.002.

Cardiac transplant recipients: (Harper et al., 1998; n = 90, male = 86%, mean age = 53.16 (11.19), all received transplants with average waiting time = 147 (141) days, survival status following 56-month observation period: 61 patients still living and 29 deceased)

• Survival curves for low and high risk groups defined by the median split of composite MBHI risk factor scores (Wald = 7.24, p = 0.07) showed the low risk leveled off at about 0.8 at 1000+ days, while the curve for the high risk group indicated the cumulative probability of survival to be slightly better than 0.5 for the same time period.
• An analysis of the postsurgical care required as an index of the cost of the transplant associated with the low and high risk groups revealed that the two groups were significantly different (Levene’s t corrected for unequal variances = 2.09, p = 0.041), with the high risk group requiring more than twice the amount of care as the low risk group.
• For the Care Rate variable, the MBHI scales Pain Threat Responsivity (β = 0.45, t = 4.420, p = 0.0000) and Cooperative coping style (β = 0.21, t = 2.11, p = 0.037) were the best predictors of post-transplant care required.
• For the Infection Rate variable, the Future Despair (β = 0.65, t = 3.39, p = 0.001) and Life Threat Reactivity (β = -0.44, t = 12.32, p = 0.02) in combination were the best predictors of post-transplant care required.

The MBHI coping scales significantly discriminated between good and poor pre-transplant compliance as well as interview judgments of good and poor coping and support resources, with modest accuracy. The MBHI also was superior to interview judgments in predicting post-transplant survival time and medical care utilized by patients. Certain scales were also positively associated with physical parameters of pre-transplant and post-transplant status. (Harper et al., 1998, p. 563).   

HIV-Positive Men and Women: (Cruess et al., 2007; n = 117, mean age = 42.75 (8.27), male = 63%; inclusion criteria: a) HIV-positive diagnosis, b) age 18 to 65, c) currently on stable highly active antiretroviral therapy (HAART) medication for at least one month; exclusion criteria: a) currently prescribed other medication affecting the immune system, b) history of chemotherapy or whole body irradiation, and c) history of chronic illness (other than HIV) associated with persistent generalized lymphadenopathy or permanent changes in the immune system.)

Baseline Characteristics on the Millon Behavioral Medicine Diagnostic Among HIV-Positive Men and Women

Predictive validity:

Human Immunodeficiency Virus (HIV): (Pereira et al., 2004; n = 28, female = 100%, African American = 87%, HIV+ women at risk for cervical cancer)

• Greater inhibited coping style on the MBHI was determined by regression analysis to be a significant predictor of greater non-adherence to clinic visits during both 1- (β = 0.45, p = 0.05) and 2-year (β = 0.54, p = 0.02) follow-up, even after controlling for depressed mood.

Human Immunodeficiency Virus (HIV): (Goodkin et al., 1992; n = 62, male = 100%, mean age = 33.8 (7.7); 66.1% exclusively homosexual, asymptomatic HIV-1 seropositive at CDC stages II and III)

• Significant positive association of MBHI active coping style with natural killer cell cytotoxicity (NKCC) that was independent of all other variables, t  = 2.48, p = 0.02.

HIV-Positive Men and Women: (Cruess et al., 2007; n = 117)

• The MBMD Depression*, Cognitive Dysfunction*, Emotional Lability*, Dejected**, Denigrated**, and Medication Abuse** were all significantly associated with the categorical variable of adherence (>=95%) vs. non-adherence (<95%) on the Medical Adherence Training Interview (MATI) at baseline (*p < 0.05, **p < 0.01)
  • An overall logistic regression model incorporating these significant scales and controlling for alcohol use and substance use accounted for 26.9% of the variance and correctly assigned 76.9% of participants as either adherent or non-adherent on the MATI at baseline, Chi-square ( 6, n  = 117) = 14.193, p = 0.030.
  • Scores on the Medication Abuse scale alone were uniquely associated with the categorization of adherence or non-adherence and increased the odds of correct categorization by 1.26 times in the overall model.
• Higher scores on the Anxiety-Tension**, Cognitive Dysfunction***, Inhibited*, Dejected**, Cooperative**, Functional Deficits*, Pain Sensitivity**, Interventional Fragility**, and Medication Abuse** scales were significantly associated with taking more drugs than prescribed over the past month at baseline (*p < 0.05, **p < 0.01, ***p <0.001)
  • Inclusion of these scales in an overall linear regression model controlling for education level was significant (R² = 0.17, F(10,114) = 2.19, p = 0.024)
• The MBMD Medication Abuse scale was significantly associated with the categorical variable of adherence (>=95%) vs. non-adherence (<95%) on the MATI at 3-month follow-up (following medical adherence training) (p < 0.01)
• Higher scores on the Medication Abuse Scale were significantly associated with higher mean percentage of weekdays in which medications were skipped over the past month. 
  • An overall logistic regression model incorporating the Medication Abuse scale and controlling for age and baseline MATI scores was significant at T2 (R² = 0.34, F(3, 79) = 13.04, p < 0.001).
  • Higher scores on the Medication Abuse scale alone were uniquely associated with higher percentage of weekdays in which medications were skipped (p = 0.006)

Heart Failure: (Farrell et al., 2011; n = 105, mean age = 57.38 (10.75), male = 62%, volunteer sample from larger parent study (n = 177) recruited from outpatient clinics at private and county hospitals, inclusion criteria = primary diagnosis of heart failure and Spanish or English fluency, exclusion criteria = current drug or alcohol abuse, current diagnosis of HIV or other immune disorder, current treatment for cancer, or any psychiatric or cognitive disorder that would impede ability to complete questionnaires)

• Mean MBMD medication abuse scale score = 11.54 (1.65)
• Mean MBMD problematic compliance score = 28.72 (3.65)

Predictive validity:

Heart Failure: (Farrell et al., 2011; n = 105)

• Adequate correlation between MDMB medication abuse scores and medication adherence (r = 0.308, p <0.001)
• Significant predictive ability of the medication abuse scale to predict medication adherence (with depression and medical abuse scores included in the model), β = 0.236, t[102] = 2.113, p = 0.037
• Significant predictive ability of the medication abuse scale to predict medication adherence (with hostility and medical abuse scores included in the model), β = 0.244, t[102] = 2.506, p = 0.014
• The Sobel test statistic for each of the mediation models mirrored the regression results by confirming that Medication Abuse partially mediates the relationship between adherence and hostility (Sobel z-score = 1.97, p = 0.061) and completely mediates the relationship between adherence and depression (Sobel z-score = 1.88, p = 0.049)  

Localized Prostate Cancer: (Cruess et al., 2013; n = 66, males aged >= 45, mean age = 68.18 (7.07), cancer stage T1a-T2b, mean time since prostate cancer diagnosis = 18.03 (6.49) months, undergone radical prostatectomy or radiotherapy within past 18 months; mean time since treatment = 12.18 (4.11) months; participants reassessed at 3-, 6-, and 12-month follow-ups)

Baseline Characteristics on the Millon Behavioral Medicine Diagnostic (MBMD) Among Male Patients Diagnosed With Localized Prostate Cancer