Piecewise regression equations for estimating stature: an anthropometric study in Spanish females

Saco-Ledo G., Porta J., Monson T. A., Brasil M. F., Atamturk D., DUYAR İ.

INTERNATIONAL JOURNAL OF LEGAL MEDICINE, vol.134, no.6, pp.2289-2296, 2020 (SCI-Expanded) identifier identifier identifier

  • Publication Type: Article / Article
  • Volume: 134 Issue: 6
  • Publication Date: 2020
  • Doi Number: 10.1007/s00414-020-02350-9
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, BIOSIS, Criminal Justice Abstracts, EMBASE, MEDLINE, DIALNET
  • Page Numbers: pp.2289-2296
  • Keywords: Anthropometry, Forensic anthropology, Human identification, Stature estimation, Tibia
  • Istanbul University Affiliated: Yes


In forensic anthropology, generic equations are generally preferred for estimation of stature. However, recent studies have demonstrated that regression equations specific to stature groups yield more accurate predictions. Almost all previous studies have been conducted on male subjects, and it is not currently known how well such equations work for females. Therefore, this study aims to test whether regression equations specific to stature groups work for females as well. To this end, a cross-sectional study was conducted to estimate stature on a sample of 351 Spanish adult females. The participants were randomized into a calibration group (n = 185) and a validation group (n = 166). Equations for stature estimation based on tibial length were developed in the calibration group, which was categorized according to stature (short, medium, and tall) using the 15th and 85th percentiles as cut-off points. The standard errors of the estimations (SEEs) for the group-specific regression equations (SEE = 2.35-2.66 cm) were lower than for the general formula derived for all participants of the calibration group (SEE = 3.46 cm). The specific equations resulted in smaller differences between estimated and recorded statures than the generic equation when we tested the equations with the validation group. Additionally, the SEE values of the stature-specific equations are lower compared to generic equations applied to other human populations. In conclusion, the group-specific equations from tibial length have high accuracy compared with previously derived equations for Spanish females and other populations. This procedure for estimating stature thereby improves the tools available to forensic scientists.