Abstract
Spinal cord stimulation (SCS) has variable effectiveness in controlling chronic pain. Previous research has demonstrated that psychosocial factors are associated with diminished results of SCS. The objective of this investigation is to examine associations between pre-implant psychological functioning as measured by the Minnesota Multiphasic Personality Inventory–2–Restructured Form (MMPI-2-RF) and SCS outcomes. SCS candidates at two sites (total N = 319) completed the MMPI-2-RF and measures of pain, emotional distress, and functional ability as part of a pre-implant psychological evaluation. At an average of 5 months post-implant, patients completed the measures of pain and emotional distress a second time. Poorer SCS outcomes and poorer patient satisfaction were associated with higher pre-implant MMPI-2-RF scores on scales used to assess emotional dysfunction, somatic/cognitive complaints, and interpersonal problems. Ways through which pre-implant psychological evaluations of spinal cord stimulator candidates can be informed by MMPI-2-RF findings are discussed.
Since the 1960s, patients with chronic, intractable pain have had the possibility of achieving some relief of their symptoms through the use of an implantable spinal cord stimulator (SCS). Based on Gate Control Theory (Melzack & Wall, 1965), SCS involves implantation of a device that delivers electrical stimulation along ascending nerve pathways. SCS is theorized to override the transmission of pain signals, thereby reducing pain intensity (Campbell, Jamison, & Edwards, 2013). SCS implantations have become more common and are now used in many chronic pain conditions, including failed back surgery syndrome (FBSS), complex regional pain syndrome, postherpetic neuralgia, refractory angina pectoris, neuropathic pain, and refractory migraine (Cameron, 2004).
About 10,000 to 20,000 permanent SCS implants are performed annually (Simpson, Cholewicki, & Grauer, 2006; Sparkes et al., 2010). However, the long-term effectiveness of stimulation is variable. In a review of the literature that defined success as greater than 50% reduction in pain, success rates ranged from 62% for phantom limb pain and FBSS to 84% for complex regional pain syndrome (Cameron, 2004). In a recent demonstration, Eldabe, Kumar, Buchser, and Taylor (2010) randomized 100 FBSS patients to either SCS or conventional medical management. At 6 months post-implant, 64% of patients had achieved the 50% reduction in leg pain criterion (vs. 18% of conventional medical management patients; see also North, Kidd, Zahurak, James, & Long, 1993). It is also noteworthy that the costs of reducing pain via SCS can be quite high. The average cost of SCS over the first 24 months post-implant was $52,091 (Hollingworth, Turner, Welton, Comstock, & Deyo, 2011). Furthermore, the complication rate of SCS is also significant, 20% to 25%, with the most frequent complications being infections and migration of the electrodes (Kumar et al., 2008; Turner, Loeser, Deyo, & Sanders, 2004).
Given that SCS is a procedure with high costs, significant medical risks, and variable effectiveness, it is imperative that factors be identified that can assist in patient selection, pointing to those for whom SCS is unlikely to be effective. There is a growing recognition that psycho-social factors can influence the outcome of many surgical procedures (see Block & Sarwer, 2013). As reviewed by Sparkes et al. (2010), several studies have found diminished SCS effectiveness is associated with depression as assessed by the Minnesota Multiphasic Personality Inventory (MMPI; Hathaway & McKinley, 1943), the MMPI-2 (Butcher, Graham, Ben-Porath, Tellegen, & Dahlstrom, 2001), and other psychological assessment instruments (Burchiel et al., 1995; Jamison & Edwards, 2013; North, Kidd, Wimberly, & Edwin, 1996; Olson, Bedder, Anderson, Burchiel, & Villanueva, 1998; Ruchinskas & O’Grady, 2000). In other reviews, pre-implant somatization, anxiety, and poor pain coping were related to diminished SCS results (Celestin, Edwards, & Jamison, 2009; Jamison & Edwards, 2013). Recent research confirms the adverse impact on SCS results of one specific negative form pain coping, catastrophizing (Sumner & Lofland, 2014). Other significant risk factors appear to be any type of workers’ compensation claim (Hollingworth et al., 2011), longer pain duration, and age (North et al., 1996).
Research documenting the impact of psychosocial factors on SCS outcome has been influential. Currently, many insurers and third-party payers either require or recommend a psychological evaluation as a precondition for the SCS trial. However, the standards for such an evaluation are unsettled. The newest iteration of the MMPI, the MMPI-2 Restructured Form (MMPI-2-RF; Ben-Porath & Tellegen, 2008/2011; Tellegen & Ben-Porath, 2008/2011), may provide an effective means for assessing psychosocial factors associated with SCS outcomes. A previous study by Block, Ben-Porath, and Marek (2013) established normative values on the MMPI-2-RF for both SCS and spine surgery candidates, finding no clinically significant differences in the mean profiles for these two patient groups. In separate studies, they examined the associations between MMPI-2-RF scale elevations, as well as other psychosocial factors, with spine surgery outcome at 6 months postoperative (Block, Marek, Ben-Porath, & Ohnmeiss, 2014; Marek, Block, & Ben-Porath, 2015). Several outcome measures were used, including reported pain level, the Oswestry Disability Index (ODI; a measure of functional ability; Fairbank, Couper, Davies, & O’Brien, 1980), medication status (whether the patient was using narcotic, nonnarcotic, or no medication), work status, negative affect, as well as an overall outcome measure that combined these measures. Poorer spine surgery outcome was significantly associated with presurgical MMPI-2-RF scale scores drawn from three major domains. In the Emotional Dysfunction domain, poorer surgical outcomes were associated with patients’ overall level of emotional distress (scale Emotional/Internalizing Dysfunction); feelings of being overwhelmed, dissatisfied, and sad (scale Demoralization); lack of ability to experience positive emotions (scale Low Positive Emotions); dysfunctional negative emotions including anxiety, anger, and fear (scale Dysfunctional Negative Emotions); as well as specific problem areas including feelings of helplessness (scale Helplessness/Hopelessness), and scales assessing self-doubt, inefficacy, stress/worry, anxiety, anger, and behavior-restricting fears. The second group of MMPI-2-RF scales associated with lessened spine surgery outcomes were those reflecting somatic/cognitive complaints, including complaints of diffuse somatic symptoms (scale Somatic Complaints); a sense of poor health, weakness, and incapacity (scale Malaise); and specific complaints of a gastrointestinal (scale Gastrointestinal Complaints), head pain (scale Head Pain Complaints), neurological (scale Neurological Complaints), and self-reported problems with memory (scale Cognitive Complaints). The final set of MMPI-2-RF scales associated with poorer spine surgery results reflected difficulties in interpersonal functioning, including specific problems of conflictual relationships with family members (scale Family Problems), social avoidance (scale Social Avoidance), and a preference for being alone (scale Disaffiliativeness). In addition, levels of persecutory ideation (scale Ideas of Persecution) were found to be correlated with poorer spine surgery results.
Noting that MMPI-2-RF profiles are not meaningfully different between spine surgery and SCS patients (Block et al., 2013), we hypothesized that MMPI-2-RF correlates of reduced SCS outcome will be similar to those found in spine surgery candidates. That is, we expected that higher presurgical MMPI-2-RF scores on measures of emotional dysfunction, somatic complaints, and interpersonal problems will be associated with reduced SCS results.
Method
Participants
Participants were referred to the Behavioral Medicine Division of Texas Back Institute and the Mercy Hospital Multidisciplinary Pain Management for a Presurgical Psychological Screening (PPS) prior to undergoing an SCS. Patients were referred from physicians both inside and outside of the medical centers. Four-hundred and fourteen consecutively referred individuals who underwent an SCS were invited to participate in an SCS outcome study. Specifically, 245 patients were asked to participate from the Texas Back Institute and 169 patients were invited to participate from the Mercy Hospital. Of those, 345 consented to participate (n = 193 and n = 152, respectively). Statistical analyses found no demographic differences between those who consented and those who did not. Of those who consented, 26 participants produced invalid MMPI-2-RF protocols and were excluded from further analyses based on the recommended guidelines for identifying invalid MMPI-2-RF results (Tellegen & Ben-Porath, 2008/2011). Specifically, a protocol was considered invalid if a patient scored at or above the cutoffs recommended in the test manual on the following Validity Scales: Cannot Say ≥18, Variable Response Inconsistency–Revised ≥80, True Response Inconsistency–Revised ≥80, Infrequent Responses = 120, and/or Infrequent Psychopathology Responses ≥100. Statistical analyses indicated that there was a significant gender differences between those who produced valid or invalid MMPI-2-RF protocols, such that men were more likely to produce invalid results than women (χ2[1] = 6.10, p = .014, φ = .13). Of the remaining 319 participants, 118 were men and 201 were women. The mean age was 53.36 years (SD = 13.89), and the mean years of education was 13.09 (SD = 3.06). Of the patients who responded at follow-up (n = 161), the average follow-up time for self-report packets to be returned was 146.71 days (SD = 31.73; median = 144.50; range = 42-239). There were no significant demographic differences between those who responded and those who did not respond at follow-up; however, patients who scored higher on the ODI administered prior to the implant procedure (M = 54.48, SD = 20.05) were more likely to not respond at follow-up (t[170.75] = 1.99, p = .05, d = .26) than patients who did respond at follow-up (M = 49.63, SD = 15.95). There were also no significant demographic differences between patients recruited at the two sites. The study was approved by the Kent State University Institutional Review Board.
Measures
Minnesota Multiphasic Personality Inventory–2–Restructured Form
The MMPI-2-RF (Ben-Porath & Tellegen, 2008/2011; Tellegen & Ben-Porath, 2008/2011) was congruent with contemporary models of psychopathology (Kotov et al., 2011; Krueger & Markon, 2006; Sellbom, Ben-Porath, & Bagby, 2008). Specifically, the substantive scales of the MMPI-2-RF measure emotional, thought, behavioral, somatoform, and interpersonal dysfunction and two interest scales. In addition, the instrument contains the PSY-5 scales, which are conceptually similar to personality disorder constructs outlined in Section III of the DSM-5 (Anderson et al., 2013). The scores from the instrument are reliable and have been validated across a number settings, including medical (such as bariatric surgery and spine surgery clinics), forensic, mental health (inpatient and outpatient), personnel screening, and nonclinical settings (Tellegen & Ben-Porath, 2008/2011).
Oswestry Disability Index
The ODI (Fairbank et al., 1980) is a 10-item, self-report questionnaire that measures the impact of pain on patients’ functional ability. The scores represent the percentage of functional disability. The index score has demonstrated good reliability (Cronbach’s alpha [α] ranging from .76 to .91) in previous studies (Fairbank & Pynsent, 2000; Kopec et al., 1996; Marek, Block, et al., 2015), good mean interitem coefficients (Marek, Block, et al., 2015), and good validity when comparing scores to actual performance, such as lifting, walking, and so on (Fisher & Johnston, 1997).
Patient Self-Reported Survey Data
This survey (published in Online Appendix A of a prior study in Marek, Block, et al., 2015) is a self-report measure of patients’ pain levels, pain-related interference with lifestyle, implant outcome expectations, and current emotional states. Patients were also asked to rate the extent to which pain interferes with their function on a 0 to 10 scale (higher value indicating more interference). Similarly, patients were asked on a 10-point Likert-type scale if they were satisfied with their results (10 indicated perfectly satisfied) and if their outcome met their expectations (10 indicated outcomes perfectly met their expectations). We reversed the satisfaction with implant results and the meeting patients’ expectation scores so that a higher score indicated less satisfaction and not meeting patient expectations. Participants rated the intensity of emotions they experienced on Likert-type scales (Not at all to Extremely), and pain on a 0 to 10 Likert-type scale (higher value indicating more pain). Exploratory factor analyses indicated that the current emotional state items can be aggregated into two scales: Negative Affect and Positive Affect (Block et al., 2014; Marek, Block, et al., 2015). We used the Negative Affect scale in the current investigation because it was marked by items such as depressed mood, worry, anger, and fear. The overall score reflects the average rating of items that loaded onto Negative Affect.
Procedure
Across both sites, patients underwent a standard PPS that included a semistructured interview with a psychologist, a review of medical records, psychometric testing, and completion of an algorithm designed for use as part of PPS in spine surgery and SCS candidates (Block, Gatchel, Deardorff, & Guyer, 2003; Block, Ohnmeiss, Ben-Porath, & Burchett, 2011; Block & Sarwer, 2013). Prior to the PPS, patients were invited to consent to participate in the follow-up study. Participants were asked to consent to have data collected during three follow-up periods after their implant: 3 to 6 months, 12 months, and 24 months. The study is ongoing, and only data collected at the time of the patients’ initial follow-up are used in the current investigation. Participants were mailed packets of outcome measures around their first post-implant date. The packet of outcome measures included the ODI and the Patient Self-Report Survey Data. The implant date was recorded at the time of the psychological evaluation. However, because the implant date was variable (e.g., rescheduling or cancellations), it was not uncommon for the patient to receive their packets outside of the 3-month period. Descriptive statistics (reported in the Participants section) showed that, with the exception of a few outliers, a majority of patients responded approximately 5-month post-surgery.
Statistical Analyses
Descriptive statistics (means and standard deviations) on MMPI-2-RF scale scores for the combined sample are reported in Table 1. Preliminary analyses indicated that there were no meaningful differences between pre-implant MMPI-2-RF scale scores and the two implant sites. Because not all 319 participants who responded at baseline replied to the initial follow-up period (n = 161 responders) and no statistically significant differences between demographic and MMPI-2-RF substantive scale scores and responders versus nonresponders at the follow-up period were found, pairwise deletion was used to handle missing data across time. Pre-implant and post-implant scores and change in pain levels, interference with lifestyle, ODI scores, and Negative Affect scores between baseline and follow-up are reported in Table 2. The magnitude of changes was interpreted in terms of effect sizes (Cohen’s d: .20 = small change, .50 = moderate change, and .80 large change; Cohen, 1988).
Means and Standard Deviations for MMPI-2-RF Scales: SCS Candidates.
Note. MMPI-2-RF = Minnesota Multiphasic Personality Inventory–2–Restructured Form; SCS = spinal cord stimulation; M = mean; SD = standard deviation.
Descriptive and Inferential Statistics Between Pre- and Post-Implant Outcome Measures.
Note. M = mean; SD = standard deviation; ODI = Oswestry Disability Index. Cutoffs for interpretation of Cohen’s d are as follows: .20 indicates a small effect size, .50 indicates a medium effect size, and .80 or larger indicate a large effect size.
Pearson product–moment correlations were computed between the pre-implant MMPI-2-RF substantive scale scores and pre-implant ODI and Patient Self-Report Survey Data results (Table S1; all Supplemental Materials are available at http://asm.sagepub.com/supplemental) as well as postoperative Patient Self-Report Survey Data and ODI scores (Table S2). Because a number of statistical tests were conducted using data reported in Tables S1 and S2, a Bonferroni-corrected alpha was computed (0.05/42 = 0.0012) in order to lower the risk of making a Type I error. In addition, correlation coefficients were considered clinically meaningful if they reach an effect size of r ≥ 0.20. The magnitudes of the correlation coefficients were interpreted in accordance to empirical guidelines for psychological–medical research (<.20 = Small effect; .20 to .30 = Modest effect; >.30 = Large effect) outlined by Hemphill (2003).
Hierarchical regression analyses were calculated to examine the incremental contribution of the MMPI-2-RF substantive scales after controlling for pre-implant pain levels, pain-related interference with lifestyle, ODI scores, and Negative Affect scores (Table S3). Pre-implant measures were entered in a first block of the analysis followed by MMPI-2-RF Substantive Scales entered in a second block by scale set (e.g., Higher-Order, RC Scales, PSY-5 Scales). MMPI-2-RF scales were selected for inclusion in the test of incremental prediction if the scale was meaningfully associated with an outcome measure. An MMPI-2-RF Substantive Scale was determined to be incrementally predicative of outcome if the associated change in R2 and standardized beta weight were statistically significant.
When relatively large numbers of independent variables are included in regression models, it is important to assess the model for stability. In other words, the regression models should be able to be applied to a similar sample drawn from the same population without losing accuracy in the prediction of the criterion. This can be ascertained through cross-validation using two independent samples. A lack reduction in R2s between models implies that the regression model is stable. Although there is no agreed on “rule-of-thumb” as to how large a difference between R2s there has to be for a regression model to be considered unstable, Kleinbaum, Kupper, and Muller (1988) suggest a difference of <.10 in R2 between models as evidence of model stability. Because of the number of predictors, the modest correlation between variables, and the small sample size associated with outcome in the current investigation, R2s in the regression models may be overestimated. Because the current sample is not large enough to be divided for the purpose of cross-validation and a second sample is not available, we tested the stability of the regression models using the Predicted Residual Sums of Squares (PRESS) method (Holiday, Ballard, & McKeown, 1995; Myers, Montgomery, & Anderson-Cook, 1990; Palmer & O’Connell, 2009), which does not require data splitting or a second sample. The PRESS statistic is the sum of squares of deleted residuals derived from a regression equation. Therefore, the PRESS statistic should be similar to the sum of squares for the error (SSE) used in ANOVA calculations of our models if our regressions are stable. Large deviations between the PRESS statistic and the SSE of our regressions would indicate model instability. R2s can be calculated based on the PRESS statistic (Holiday et al., 1995; Myers et al., 1990; Palmer & O’Connell, 2009). The calculation for PRESS R2s is as follows: 1 − [Predicted Residual Sums of Squares/Total Sum of Squares]. Small deviations (<.10) between the R2s derived from the hierarchal regressions and PRESS R2s indicate that the analyses will likely replicate in similar samples drawn from the same population (Holiday et al., 1995).
To document the practical implications of our results, relative risk ratios (RRRs) were computed for statistically significant and meaningful findings reported in Table 3 for various MMPI-2-RF scale score cutoffs (e.g., 55T, 60T, 65T, 70T, 75T). RRRs were computed by dividing the risk of achieving suboptimal implant results for patients who score at or above a pre-implant MMPI-2-RF scale score cutoff by the risk among those who score below a MMPI-2-RF scale score cutoff. A RRR was deemed statistically significant if 1 was not in the 95% confidence interval. The variables of current pain levels, current pain-related interference with lifestyle, if the patients were satisfied with their implant results, and if their outcome met their expectations were dichotomized such that 5 to 10 indicated suboptimal implant results whereas a score of <5 indicated good outcome. For the ODI, a score at or greater than 40 indicated poor functional ability as this is the recommended cut-point (Little & MacDonald, 1994). The Negative Affect score was not dichotomized because this is not a dimensional construct. RRRs reported in Table 3 reflect analyses that were statistically significant and yielded the highest RRR for the criteria at an MMPI-2-RF cut-score.
Statistically Significant Minnesota Multiphasic Personality Inventory–2–Restructured Form Relative Risk Ratios for Post-Implant Criteria.
Note. MMPI-2-RF = Minnesota Multiphasic Personality Inventory–2–Restructured Form; BR = base rate; SR = selection ratio; RRR = relative risk ratio; CI = confidence interval; ODI = Oswestry Disability Index; RCd = Demoralization; EID = Emotional/Internalizing Dysfunction; RC2 = Low Positive Emotionality (RC2); MLS = Malaise; COG = Cognitive Complaints; STW = Stress/Worry; ANP = Anger Proneness; NEGE-r = Negative Emotionality/Neuroticism–Revised; SFD = Self-Doubt; SUB = Substance Abuse.
Results
Presented in Table 1 are descriptive statistics for all 51 MMPI-2-RF scale scores for SCS patients combined from both sites after removing invalid MMPI-2-RFs. Descriptive and inferential statistics for the measures assessed at baseline and postoperative follow-up are reported in Table 2. Patients reported statistically significant and meaningful decreases in pain levels, pain-related interference with lifestyle, and functional disability as measured by the ODI at follow-up. On the other hand, there was a statistically significant increase in ratings of Negative Affect following SCS, though this effect was marginal.
Reported in Table S1 are Pearson product–moment correlations between the 42 MMPI-2-RF substantive scale scores and the pre-implant measures from the Patient Self-Reported Survey Data and functional ability measured on the ODI. The patterns of associations were similar to those reported in other studies using these baseline measures in similar samples (Block et al., 2013; Marek, Block, et al., 2015). For example, the Restructured Clinical (RC) Scales Demoralization (RCd), Somatic Complaints (RC1), Low Positive Emotions (RC2), and Dysfunctional Negative Emotions (RC7) were all modestly to substantially correlated with ODI and Negative Affect scores. Moreover, Demoralization (RCd), Somatic Complaints (RC1), and Low Positive Emotions (RC2) were moderately associated with pain-related interference with lifestyle. Ideas of Persecution (RC6) and Aberrant Experiences (RC8) were also modestly associated with Negative Affect scores. Moving to the Specific Problems Scales, all the Somatic/Cognitive Specific Problems Scales were associated with Negative Affect scores. Malaise (MLS) yielded a substantial correlation with pain-related interference with lifestyle Behavior Restricting Fears (BRF) was modestly associated with functional disability as measured on the ODI. None of the Externalizing Specific Problems scales were associated with the pre-implant measures. The Family Problems (FML) scale had modest associations with ODI scores and Negative Affect.
Table S2 contains Pearson product–moment correlations between the 42 MMPI-2-RF substantive scales assessed prior to implant and the post-implant measures from the Patient Self-Reported Survey Data and functional ability measured on the ODI. On the three Higher-Order scales, pre-implant Emotional/Internalizing Dysfunction (EID) scores were modestly to substantially correlated with all outcome measures. Pre-implant scores on Thought Dysfunction (THD) were associated with higher post-implant Negative Affect scores. Among the Restructured Clinical (RC) Scales, the strongest pre-implant predictors of negative outcome after implant were scores on the emotional dysfunction measures: Demoralization (RCd) and Dysfunctional Negative Emotions (RC7). Additionally, Somatic Complaints (RC1), Low Positive Emotions (RC2), and Aberrant Experiences (RC8) were associated with higher Negative Affect scores post-implant. Among the Somatic/Cognitive Specific Problems scales, pre-implant Malaise (MLS), Neurological Complaints (NUC), and Cognitive Complaints (COG) scores were modestly correlated with post-implant Negative Affect scores. In addition, post-implant ODI scores were associated with pre-implant Malaise (MLS) scores. Of the Internalizing Specific Problems Scales, all scales except Multiple Specific Fears (MSF) were associated with post-implant Negative Affect scores. Stress/Worry (STW) was associated with all post-implant measures. Inefficacy (NFC) was associated with higher post-implant pain levels whereas Anger Proneness (ANP) was associated with higher ODI scores post-implant. Self-Doubt (SFD) was associated with implant not meeting patients’ expectations and less satisfaction with implant results. In terms of the Externalizing Specific Problems Scales, Substance Abuse (SUB) was correlated with implant not meeting patients’ expectations. Of the Interpersonal Specific Problems Scales, pre-implant scores on Family Problems (FML) and Social Avoidance (SAV) were associated with post-implant Negative Affect scores. Among the PSY-5 scales, modest to large associations were observed between pre-implant Negative Emotionality/Neuroticism Revised (NEGE-r) and all post-implant measures (except for post-implant pain levels). Additionally, Introversion/Low Positive Emotionality Revised (INTR-r) was moderately associated with post-implant ODI and Negative Affect scores.
Table S3 lists the MMPI-2-RF Substantive Scales that predicted outcome significantly beyond scores on baseline measures. Overall, the MMPI-2-RF accounted for 2% to 11% of additional variability in SCS outcomes. Moreover, PRESS R2s (discussed earlier) indicated that our models would accurately predict these criteria in other samples, as evidence by small differences (<.10) in the R2s derived from our models and the PRESS R2s. Emotional/Internalizing Dysfunction (EID), Demoralization (RCd), Stress/Worry (STW), and Negative Emotionality/Neuroticism–Revised (NEGE-r) incrementally predicted all outcome measures after controlling for scores on associated baseline measures. Family Problems (FML) predicted postoperative pain levels, pain-related interference with lifestyle, and Negative Affect above and beyond preoperative variables. Last, Cognitive Complaints (COG) incrementally predicted higher Negative Affect scores after controlling for baseline scores.
Presented in Table 3 are RRRs for each criterion at the most optimal MMPI-2-RF scale score cutoff. As can be seen, Demoralization (RCd) significantly raised the risk for poor results on all outcome and satisfaction measures. Other scales from the Emotional Dysfunction dimension significantly increased the risk of reporting poorer functional ability, negative affect and dissatisfaction, with RRRs ranging from 1.42 to 1.80. Elevated scores on Somatic/Cognitive scales Malaise (MLS) and Cognitive Complaints (COG) increased the risk of reporting more functional impairment (RRRs = 1.61). Elevated Substance Abuse (SUB) scores increased the risk of patients’ negative evaluations of spine surgery by up to 1.71.
Discussion
The results of the current study demonstrate that pre-surgical MMPI-2-RF scale elevations are significantly correlated with various outcome criteria at an average of about 5 months post-implant. In general, MMPI-2-RF scales that are associated with poorer outcomes of spine surgery are also associated with lesser outcomes of SCS and with reduced patient satisfaction. Many scales reflecting emotional dysfunction were found to be associated with poorer outcomes. Multiple scales reflecting somatic/cognitive dysfunction were also associated with poorer SCS results and satisfaction. The only interpersonal functioning scale associated with poorer outcomes were Family Problems (FML) and Social Avoidance (SAV).
Associations between pre-surgical emotional dysfunction SCS outcome warrants discussion. MMPI-2-RF scales in this domain that were found to be associated with poorer outcome assess feelings of being overwhelmed, highly distressed, and dissatisfied with life, in addition to an inability to experience positive emotions, higher levels of anxiety and anger, and lack of self-confidence. Higher scores on these scales have also been to be associated with adverse outcomes of spine surgery (Block et al., 2014; Marek, Block, et al., 2015), as well as with higher levels of negative affect and maladaptive eating behaviors at 3 months following bariatric surgery (Marek, Ben-Porath, Merrell, Ashton, & Heinberg, 2014). Patients with these symptoms are likely to have problems with motivation and with recognition of treatment gains, whereas patients who do not experience such issues tend to have better SCS results. For example, in one study of spine surgery results (Skolasky, Mackenzie, Wegener, & Riley, 2008), “patient activation”—defined as a patient’s propensity to engage in adaptive health behaviors—was correlated with a sense of greater self-efficacy and lower levels of depression. In addition, greater patient activation was also associated with better adherence to postoperative physical therapy. Patients at the highest levels of activation also showed greater decreases in pain and in disability as a result of surgery (Skolasky, Mackenzie, Wegener, & Riley, 2011). At the other end of the emotional dysfunction spectrum, depression among chronic pain patients is associated with underreporting of gains in functional ability during multidisciplinary treatment (Kremer, Block, & Atkinson, 1983), and with the tendency to overinterpret sensations as painful (Geisser, Haig, Colwell, Wiggert, & Shirey, 1999). Thus, it appears that emotional dysfunction can affect cognition, motivation, compliance, and pain perception in ways that bode poorly for the outcome of SCS.
Somatic/cognitive dysfunction scales, including those that reflect preoccupation with health concerns, multiple somatic complaints, and feelings of being weak and incapacitated, were also correlated with poorer SCS outcomes. This finding is consistent with previous literature examining associations between scores on earlier versions of MMPI instruments and both spine surgery as well as SCS outcomes. As reviewed by Block et al. (2003), the MMPI and MMPI-2 scales assessing heightened somatic sensitivity, including Hypochondriasis (Scale 1) and Hysteria (Scale 3), are related to diminished outcomes in these settings. The excessive sensitivity to somatic complaints assessed by these scales also bodes poorly for the outcome of comprehensive pain management programs (Vendrig, Derksen, & de Mey, 1999). In previous research (Block, Vanharanta, Ohnmeiss, & Guyer, 1996), we found elevations on these two scales to be associated with reporting of excessive pain during discography, a medical diagnostic test designed to identify disrupted intervertebral discs and to link such aberrant disc structure with the patient’s pain complaints.
An interpersonal functioning scale that was correlated with poorer SCS outcome was Family Problems (FML)—a measure of conflictual family relationships, dislike of family members, feelings of being unappreciated, and not supported by family members. Feelings of being supported (loved and cared for, and willingly assisted by family) have been found to be critical to recovery in many medical conditions. For example, Schade, Semmer, Main, Hora, and Boos (1999) found that support from the spouse was significantly associated with greater relief after lumbar laminectomy/discectomy. Similarly, when examining recovery from hip fracture surgery, patients with low levels of social support achieved less improvement in walking ability at 2 months following surgery than did patients with higher support levels (Mutran, Reitzes, Mossey, & Fernandez, 1995). Lack of support and relationship dissatisfaction have been found to have further deleterious effects. We have studied marital satisfaction among spouses of chronic pain patients, finding that dissatisfied spouses tend to attribute the patient’s pain to psychological rather than physical causes (Block & Boyer, 1984) and to have more negative outcome expectations for patients (Block, Boyer, & Silbert, 1985). Interactions with spouses and family members can influence the course of pain in other ways. Following behavioral concepts of chronic pain (Fordyce, 1976), it is now firmly established that family members, by responding to the patient in a solicitous fashion can reinforce both pain complaints (Block, Kremer, & Gaylor, 1980) and activity limitations (Flor, Kerns, & Turk, 1987), negatively affecting treatment outcomes. The current findings linking reduced SCS outcome with high scores on the FML scale, combined with previous research on social support and on pain reinforcement by family members, point to the importance of considering the social environment of the SCS candidate.
The current results have a number of limitations. First, despite multiple attempts at obtaining follow-up packets from participants, only about one-half of the subjects completed the follow-up survey. Although no significant demographic or MMPI-2-RF scale score differences between responders and nonresponders were found, it is possible that the strength of associations between the MMPI-2-RF scale scores and outcome measures would be higher if there were a greater rate of follow-up survey completion. An important limitation that potentially attenuates prediction coefficients in both the current sample and in similar studies is that patients with no psychopathology or well-controlled psychopathology are more likely to be cleared for procedures they are seeking. Patients who are identified as having a psychosocial risk factor that may impede surgical/implant results are likely to receive pre- or postoperative psychological treatment. Moreover, patients evidencing higher functional disability scores on the ODI in the current investigation were also more likely not to follow-up. Understanding how these methodological considerations affect prediction coefficients has been discussed in greater detail with bariatric surgery patients (Marek, Tarescavage, et al., 2015). Last, Butcher and Tellegen (1978) warned against collapsing samples across important demographic characteristics in studies using MMPI instruments without first exploring possible systematic differences. We lacked statistical power to conduct such analyses in the current investigation.
Another limitation of the current study is reliance on self-report to assess outcomes. Because pain is a subjective phenomenon, self-report is important to assess for individual differences of perceived pain levels. However, assessing functional ability through objective measurement such as a functional capacity examination may lead to different conclusions. Finally, these data measure only outcome at an average of 5 months post-implant. Longer follow-up intervals may significantly alter the relationship on MMPI-2-RF to outcome and satisfaction.
In summary, the current study provides support for the inclusion of a standardized psychological assessment instrument such as the MMPI-2-RF in the pre-implant psychological evaluation of candidates for SCS. A number of MMPI-2-RF scales assessing emotional dysfunction, somatic/cognitive dysfunction, and interpersonal relations were strongly associated with poorer outcome of SCS. Elevations on these scales indicated that patients had up to a 1.856 times greater risk of achieving poor outcome. These findings are generally consistent with previously reported data from our group on the association between pre-surgical MMPI-2-RF scores and spine surgery results, with the exception that, unlike spine surgery patients, employment status at 6 months postprocedure was not associated with any MMPI-2-RF scales for the SCS patients. Future studies will allow researchers and clinicians greater ability to assess psychosocial risk factors in SCS candidates and enhance the outcome of pain control procedures.
Footnotes
Declaration of Conflicting Interests
The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Yossef Ben-Porath is a paid consultant to the MMPI-2-RF publisher, the University of Minnesota Press, and Distributor, Pearson. As coauthor of the MMPI-2-RF, he receives royalties on sales of the test.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was partially funded through a grant from the University of Minnesota Press to the first author, Andrew Block.