The association between skeletal lesions and tuberculosis diagnosis using a probabilistic approach

Dorthe Dangvard Pedersen, George R. Milner, Hans Jørn Kolmos, Jesper Lier Boldsen

Research output: Contribution to journalArticlepeer-review

29 Scopus citations


Sensitivity and specificity estimates for 18 skeletal lesions were generated from modern skeletons for future paleoepidemiological analyses of tuberculosis prevalence in archaeological samples. A case-control study was conducted using 480 skeletons from 20th century American skeletal collections. One-half of the skeletons were documented tuberculosis cases (Terry Collection). The remaining age and sex-matched skeletons were controls (Bass Collection). The association between 18 candidate skeletal lesions and tuberculosis was established by comparing lesion distributions in case and control groups. Lesion indicators at six locations – visceral surface of ribs, ventral vertebral bodies, lateral part of ilium, acetabular fossa, iliac auricular surface, and ulna olecranon process - occurred significantly more often among cases than in controls, and were associated with one another. The most useful indicator proved to be a bony reaction on ventral thoracic and lumbar vertebral bodies. Its presence means a 53.3% probability of a true tuberculosis diagnosis. Because of the nature of the reference sample – 20th century American cases – sensitivity and specificity estimates will better estimate disease prevalence in archaeological samples from cultural settings where pulmonary tuberculosis predominated. The general approach of this proof-of-concept study is applicable to other diseases that occur commonly and affect bone.

Original languageEnglish (US)
Pages (from-to)88-100
Number of pages13
JournalInternational Journal of Paleopathology
StatePublished - Dec 2019

All Science Journal Classification (ASJC) codes

  • Pathology and Forensic Medicine
  • Archaeology


Dive into the research topics of 'The association between skeletal lesions and tuberculosis diagnosis using a probabilistic approach'. Together they form a unique fingerprint.

Cite this