By Andy Skean
National estimates of arthritis prevalence typically rely on a single survey question about doctor‐diagnosed arthritis, without using survey information on joint symptoms, despite some patients with only the latter having been shown to have arthritis.
The sensitivities of the current surveillance definition are only 53% and 69% in patients aged 45 to64 and 65 or older, respectively, resulting in misclassification of nearly one‐half and one‐third of people in these age groups.
To better assess how this patient population is treated, Reza Jafarzadeh, DVM, MPVM, PhD, and colleagues sought to estimate arthritis prevalence based on an expansive surveillance definition that is adjusted for measurement errors in the current definition. Their results were published in Arthritis Care & Research.
“The current arthritis definition and the methodology used by the CDC since 2002 was known to be inaccurate for prevalence estimation,” says Dr. Jafarzadeh. “An underestimate resulted in the under-investment of arthritis research, awareness, and prevention, which is going to cost us significant healthcare expenditure and loss of productivity as a result of disability down the road.”
Using the 2015 National Health Interview Survey, the researchers developed a Bayesian multinomial latent class model for arthritis surveillance based on doctor‐diagnosed arthritis, joint symptoms, and whether symptom duration exceeded 3 months.
Of 33,672 participants, 19.3% of men and 16.7% of women aged 18 to 64 and 15.7% of men and 13.5% of women aged 65 or older affirmed joint symptoms without doctor‐diagnosed arthritis. The measurement error–adjusted prevalence of arthritis was 29.9% in men aged 18 to64, 31.2% in women aged 18 to64, 55.8% in men aged 65 and older, and 68.7% in women aged 65 and older.
“Estimating disease prevalence on a national level from surveys is a practical method, but a challenging task,” says Dr. Jafarzadeh. “While the survey data areuseful in understanding the national burden of disease, the methodology to make inferences on prevalence at federal agencies needs to be revised and improved substantially.”