Reproducibility indices applied to cervical pressure pain threshold measurements in healthy subjects

Objectives: To apply various statistical indices for reproducibility analysis of pressure pain threshold measurements and to derive a preferred pressure pain threshold measurement protocol based on these indices. Methods: The pressure pain threshold of 3 pairs of right and left homologous cervical region sites were measured in 20 healthy subjects (10 women, 10 men) using a hand-held pressure algometer. Measurements took place on 2 occasions (test 1 and test 2) separated by a mean interval of 1 week. On each testing session, the site-related pressure pain thresholds were measured 3 times each according to 2 different protocols. Protocol A consisted of a repetitive order, namely 3 consecutive measurements at each site before proceeding to the next, whereas protocol B consisted of an alternate order in which 3 consecutive rounds of all individually tested sites took place. For test 1, protocol A was followed by protocol B with an hour interval. For test 2, the reverse order took place. Results: The findings revealed no significant differences between the two protocols and indicated a significant rise (P < 0.0001) in the absolute scores from test 1 to test 2 in both protocols. Absolute values (mean +/-SD) derived from the entire sample of pressure pain threshold sites ranged from 140 +/- 60 to 198.7 +/- 95 kPa (1.60 +/- 0.6 to 1.99 +/- 0.95 kg/cm(2) respectively). No significant gender or side differences were noted. Pearson r as well as the intraclass correlation coefficient revealed good to excellent reproducibility for both protocols and for all sites measured: r = 0.79-0.94 and intraclass correlation coefficient((3.3)) 0.85-0.96, respectively. To define site-specific cutoff values indicating change at the 95% confidence level, 1.96*SEM was calculated, and its values ranged from 31.6 to 58.2 kPa, which correspond to 16.8% to 32.8% of the absolute mean values. In addition, the limits of agreement, which depict the individual test-retest differences relative to their mean, indicated a heteroscedastic trend. Discussion: The two protocols yielded very similar results. However. on the grounds of patient's comfort and compliance as well as facility of application, protocol B stands out as the more preferred between the two.