Table 4

Graded severity of localised demyelination and axonal degeneration as judged by electroneurography (ENG) of the ulnar nerve (four grades, 0–3) in relation to forceful work (force-score, dichotomised)

Graded severity of ulnar nerve affection according to ENG
Localised demyelinationAxonal degeneration
Cases fromUnadjusted OR (95% CI)Adjusted OR* (95% CI)Cases fromUnadjusted OR (95% CI)Adjusted OR* (95% CI)
Study 1Study 2Study 1Study 2
n=323n=396n=322n=394
Force-score >01.18 (0.68 to 2.07)1.07 (0.60 to 1.90)1.60 (0.89 to 2.86)1.41 (0.76 to 2.59)
Severity×force†
 091051.001.0021831.001.00
 1221971.36 (0.68 to 2.73)1.31 (0.64 to 2.69)781981.00 (0.50 to 2.00)0.93 (0.45 to 1.92)
 273862.18 (0.99 to 4.80)2.22 (0.97 to 5.06)971071.73 (0.80 to 3.70)1.88 (0.85 to 4.19)
 321982.28 (1.09 to 4.76)2.22 (1.03 to 4.78)12661.26 (0.54 to 2.91)1.30 (0.54 to 3.13)
p‡0.100.16
  • Results of conditional logistic regression analyses comprising the total study population (n=2513 original cases with matched referents). Numbers of original cases in study 1 (ulnar neuropathy) and study 2 (ulnar neuropathy-like symptoms) are shown.

  • *Adjusted for body mass index (continuous), pack-years of smoking (continuous), alcohol consumption (continuous), side-specific fractures (never/ever), full anaesthesia within a 5-year period up to the index year (no/yes), predisposing disorders (no/yes), use of crutches within a 5-year period up to the index year (no/yes), hand–arm intensive sports (0, 1, 2) and weight loss ≥10 kg within half a year during a 5-year period up to the index year (no/yes).

  • †The method used interaction terms between force-score (0, >0) and four levels of severity (0–3). Matched controls followed their index case to one of the four levels of severity (to obtain this, they were technically given the severity score of their index case). For each level of severity, the OR for the interaction term represents the effect of a force-score >0 as compared to a force-score of 0. This corresponds to analyses stratified by severity, which explains why no OR is shown for severity.

  • ‡Likelihood-ratio test of the presented model as compared with the corresponding adjusted model without the interaction term.