Objective: The Multidimensional Pain Inventory (MPI) has been used widely to assess the psychosocial adaptation in chronic pain patients. The MPI can also be used to classify patients into one of three primary subgroups. However, studies have reported that anywhere from 3% to 30% of patients are unclassifiable, mostly attributable to an excessive number of missing responses to questions. The purpose of this study was to examine the reasons for the large number of missing responses and subsequently to reduce the number of patients who cannot be classified within one of the three primary MPI subgroups.
Design: Two versions of the MPI were evaluated on two samples of patients (N = 147; and N = 143) with persistent pain being evaluated by the Workers Compensation Board of British Columbia, Canada. One version used the published MPI instructions and the second modified the instructions to clarify the meaning of the term "significant other." In addition, patients were required to designate a significant other who would serve as the anchor for all responses related to a significant other.
Results: Approximately 18% of patients could not be classified within one of the three primary profiles and were determined to be unanalyzable using the MPI standard classification procedure. Most patients who could not be classified had excessive missing data to those questions asking about support received from "significant other." Almost all of these patients with missing data were unmarried and living alone. Modifying the instructions of the MPI to specify the meaning of the term "significant other" resulted in an almost threefold (6.3% vs. 18%) reduction in the percentage of patients whose responses could not be analyzed and consequently classified into one of the three primary MPI subgroups. The revised instructions did not lead to any difference in the actual distribution of patients assigned to one of the three profiles. Thus, using the modified instruction did not alter the proportion of patients assigned to one of the three primary groups.
Conclusion: Modification of the MPI instructions clarifying the definition of a significant other can greatly reduce the number of patients who are classifiable within one of the three primary MPI profiles. Thus, the modified instructions of the MPI (hence the MPI-M) have a significant advantage over the original version. The mean scores on the individual scales of the MPI and the MPI-M are comparable. Moreover, the results demonstrate that the proportion of the previously unanalyzable patients classified within one of the three primary subgroups is comparable to the original proportion of patients assigned to each of the primary subgroups. Thus, it may not be necessary to reestablish the norm scores based on the modified instructions. Future studies and clinicians who wish to use the MPI for the purpose of classifying patients on the basis of psychosocial and behavioral factors should make use of the MPI-M that incorporates the clarification of the meaning of "significant other."