OA risk over 40 years predictable at age 18 years
Individual and population risk for knee osteoarthritis over a 40-year period is predictable among individuals aged 18, using just a few easily measured covariates with moderate discrimination, according to findings published in Arthritis Care & Research.
“We are aware of only one study predicting knee OA using easily obtainable candidate predictors: the Nottingham 12-year risk prediction models,” Karin Magnusson, PhD, of Lund University, Sweden, and colleagues wrote. “The study sample in which it was developed comprised middle-aged persons and the model included the predictors age, female sex, BMI, physically demanding work, family history of OA and knee injury. ... The extent to which a simple model including variables available in adolescents can predict the long-term risk of knee OA is currently unknown.”
To simplify the 12-year Nottingham model for knee OA and determine whether it could predict long-term risk in young men, the researchers studied data from 40,118 men aged 18 years who had been conscripted into the Swedish military between September 1969 and May 1970. Examiners at that time measured height and weight, as well as data on knee injuries. The researchers linked that data to diagnoses of knee OA entered between 1987 and 2010 in the Swedish National Patient Register.
The original Nottingham model included age, sex, BMI, knee injury, occupational risk and family history of OA as predictors; for the study, Magnusson and colleagues used age, BMI and knee injury, as they are available during adolescence. In addition, they examined the discrimination of the simplified model using area under receiver operating characteristics curve (AUC) in their sample.
According to the researchers, the original Nottingham 12-year model had an AUC of 0.7 (95% CI, 0.61-0.79) in the model development sample, and 0.6 (95% CI, 0.58-0.63) in an external validation sample. In their study, the researchers calculated the AUC for the modified model predicting 40-year risk to be 0.6 (95% CI, 0.590.61). Under this model, a patient aged 18 years with a BMI of 30 and a knee injury would have 22% risk for knee OA over a 40-year period, three times greater than that of a peer with a BMI of 25 and no knee injury.
“We have simplified and externally validated a risk prediction model to be used for the prediction of knee OA with moderate discrimination in young men,” Magnusson and colleagues wrote. “Further studies are required for finding a predictive model that effectively discriminates between persons with and without high risk of knee OA. Yet, so far, the Nottingham risk prediction model has been externally validated in at least three different samples including our study. Risk prediction tools may become increasingly important for encouraging risk reduction at the individual and population level.” – by Jason Laday
Disclosure: Magnusson reports no relevant financial disclosures. Please see the study for all other authors’ relevant financial disclosures.