Open AccessCATEGORICAL VERSUS CONTINUOUS MEASURES OF PAIN: IMPLICATIONS FOR COMMUNITY‐BASED RESEARCH
Author(s) -
Jana Mossey,
Rollin M. Gallagher,
Mcp Hahnemann
Publication year - 2002
Publication title -
pain medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.893
H-Index - 97
eISSN - 1526-4637
pISSN - 1526-2375
DOI - 10.1046/j.1526-4637.2002.20248.x
Subject(s) - categorical variable , mcgill pain questionnaire , medicine , physical therapy , ambiguity , adaptation (eye) , physical medicine and rehabilitation , psychology , visual analogue scale , statistics , computer science , mathematics , neuroscience , programming language
Jana Mossey, PhD, MSN, and Rollin M. Gallagher, MD, MPH, MCP Hahnemann University Introduction: The Parmelee adaptation of the McGill Pain Questionnaire (PAMPQ), widely used in studies of elderly, creates values ranging between 5 (‘no pain’) and 25 (‘severe pain’). The scale, when treated as a continuous measure, is statistically efficient but difficult to interpret for values between 6 and 24. In longitudinal studies, change scores are conveniently calculated as the difference between two consecutive pain assessments. The range of change scores for the PAMPQ, −20 to +20, denotes direction of change but clinical ambiguity. Categorical classification provides an alternative. Using two PAMPQ items tapping current pain and pain‐related activity limitations, individuals can be classified as: 1) No pain; 2) pain without activity limitations; and 3) pain with activity limitations. The clarity of this three‐level measure is intuitively appealing. Change in pain can be identified by cross‐classifying individuals at consecutive assessments, although these measures may mask small changes and restrict analytic options. Method: Longitudinal data on 353 community dwelling elders, based on the PAMPQ, permitted us to compare the ability of categorical and continuous pain measures to characterize cross‐sectional pain levels and change over two years. Results: Categorical cross‐sectional analyses classified individuals as: (1) no pain (41%); (2) pain without limitations (23%); and (3) pain with activity limitations (36%). Additional analyses showed statistically significant mean differences between groups 1, 2 & 3 on the continuous form of PAMPQ. Substantial score overlaps occurred between pain groups (2 & 3): each had 51% with scores between 10 and 16, making interpretation of individual scores difficult. Longitudinal comparisons revealed similar difficulty. Continuous difference scores ranged from −19 to 16 (mean .849; sd 4.639). Most experiencing a 2 level change on the pain classification variable had difference scores>(+/−4). Comparable change occurred in 45% changing 1 pain level and 25 % with no change. Discussion : These findings highlight differences between continuous and categorical pain measures, with implications for pain research.