Who, when, where, and why: A systematic review of “late diagnosis” in autism

Abstract

An autism diagnosis can be a critical milestone toward effective and affirming support. Despite the sharp increase in the number of studies focused on late diagnosis over the last 15 years, there remains no consensus as to what constitutes a late diagnosis of autism, with cutoffs ranging from infancy to middle adulthood. This preregistered systematic review evaluated (a) the field's current quantification of late diagnosis in autism, (b) how the threshold for late diagnosis varies as a function of demographic and population factors, and (c) trends over time. Of the 11,697 records retrieved, N = 420 articles met inclusion criteria and were extracted. Articles spanned 35 years (1989–2024) and included participants from every continent except Antarctica. Only 34.7% of included studies provided a clear threshold for “late diagnosis” (n = 146/420). Late diagnosis cutoffs averaged 11.53 years (range = 2–55 years; median = 6.5 years) with a bimodal distribution (3 and 18 years). The threshold for late diagnosis varied by participant location, F(5,140) = 10.4, p < 0.0001, and sample age, F(5,140) = 20.1, p < 0.0001. Several key rationales for age determinations emerged, including access to services, considerations for adult diagnoses, and data driven approaches. What authors consider to be a “late” diagnosis of autism varies greatly according to research context. Justifications for a specific late-diagnosis age cutoff varied, underscoring the need for authors to contextualize their conceptualizations.