Knee osteoarthritis has doubled in prevalence since the mid-20th century

Abstract

Knee osteoarthritis (OA) is believed to be highly prevalent today because of recent increases in life expectancy and body mass index (BMI), but this assumption has not been tested using long-term historical or evolutionary data. We analyzed long-term trends in knee OA prevalence in the United States using cadaver-derived skeletons of people aged ≥50 y whose BMI at death was documented and who lived during the early industrial era (1800s to early 1900s; n = 1,581) and the modern postindustrial era (late 1900s to early 2000s; n = 819). Knee OA among individuals estimated to be ≥50 y old was also assessed in archeologically derived skeletons of prehistoric hunter-gatherers and early farmers (6000-300 B.P.; n = 176). OA was diagnosed based on the presence of eburnation (polish from bone-on-bone contact). Overall, knee OA prevalence was found to be 16% among the postindustrial sample but only 6% and 8% among the early industrial and prehistoric samples, respectively. After controlling for age, BMI, and other variables, knee OA prevalence was 2.1-fold higher (95% confidence interval, 1.5-3.1) in the postindustrial sample than in the early industrial sample. Our results indicate that increases in longevity and BMI are insufficient to explain the approximate doubling of knee OA prevalence that has occurred in the United States since the mid-20th century. Knee OA is thus more preventable than is commonly assumed, but prevention will require research on additional independent risk factors that either arose or have become amplified in the postindustrial era.

Keywords: aging; arthritis; evolutionary medicine; mismatch disease; obesity.

Conflict of interest statement

The authors declare no conflict of interest.

Figures

Fig. 1.

Knee OA prevalence during different time periods. (A and B) Knee OA prevalence from regression models controlling for sex (A) as well as age, BMI, sex, and ethnicity (B). Dark and light gray bars are from unmatched and matched analyses, respectively (B). (C and D) Knee OA prevalence ratios from regression models including sex (C) as well as age, BMI, sex, and ethnicity (D) as predictor variables. Black and light gray dots are from unmatched and matched analyses, respectively (D). Age and BMI were entered into models as continuous variables, but effects are reported for 10-y and 5-U intervals, respectively (D). Whiskers represent 95% CIs. Ethnicity effects are reported in Table S3.

Fig. 2.

Age-related change in knee OA prevalence controlling for BMI, sex, and ethnicity. Shading represents 95% CIs.