Commentary: Effects of diagnostic thresholds and research vs service and administrative diagnosis on autism prevalence
Bibliographic record
Abstract
King and Bearman are to be congratulated on their sophisticated analysis of the Californian Department of Developmental Services (DDS) database. In contrast with previous attempts to examine diagnostic substitution and diagnostic accretion (both in the same data source and in national administrative data sources), which allowed time trends but not individual child-level diagnostic substitutions and accretions to be examined, they demonstrated that children previously classified with ‘mental retardation’ account for one-quarter of the measured increase in autism prevalence in the DDS. However, King and Bearman highlight the fact that this leaves nearly three-quarters of the increase to be explained by other factors. The information available in administrative databases such as the DDS do not allow for any test of what these ‘other factors’ might be. Thus, their analysis does not answer the ‘great questions’ that have engaged both the scientific community and the general public: has there been a real increase in incidence and, if so, why? What has been the impact of changes in diagnostic practice, public and professional awareness of autism and other methodological factors (e.g. broadening of our concept of autism, different methods of ascertainment, inclusion of individuals with average IQ and those with other neuropsychiatric and medical disorders) that likely account for much of these dramatic time trends? Service administration databases are not prevalence studies and changes in recorded need might reflect changes in entitlement or availability of particular services, rather than true changes in prevalence. However, over the past 10 years epidemiological studies have also found much higher rates of autism than previously; so the administrative increase identified in the DDS data has also been reflected in the findings from prevalence studies. For several decades following Lotter’s seminal study, the autism prevalence figure that he reported of 4.5 cases per 10 000 was broadly accepted, although 20 per 10 000 children was noted as showing the broader ‘triad of impairments’ characterized by Wing and Gould. Over the past decade, prevalence rates for autism in the range of between 20 and 40 per 10 000 have been reported. The rates for the broader autism spectrum have ranged from 60 per 10 000 to close to and even over 100 per 10 000. However, quantifying the effects of the various methodological factors that might help explain the higher figures found in more recent studies is difficult. Williams et al. conducted a systematic review of autism prevalence studies and found that diagnostic criteria [higher prevalence in studies using International Classification of Diseases (ICD-10) or Diagnostic and Statistical Manual of Mental Disorders (DSM-IV)], age of the sample (higher prevalence in younger samples) and study location (higher prevalence in Japanese vs US studies), are systematically related to the prevalence figures reported across 37 studies. A significant proportion (61%) in the variance in prevalence rates across studies could be explained by these factors. However, Williams et al. concluded that the three factors identified might be acting as a proxy for other study characteristics that affect prevalence rate. In a univariate analysis, they found that decade of study had the largest
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.020 | 0.172 |
| Meta-epidemiology (narrow) | 0.002 | 0.002 |
| Meta-epidemiology (broad) | 0.003 | 0.003 |
| Bibliometrics | 0.003 | 0.003 |
| Science and technology studies | 0.004 | 0.005 |
| Scholarly communication | 0.004 | 0.006 |
| Open science | 0.012 | 0.003 |
| Research integrity | 0.060 | 0.043 |
| Insufficient payload (model declined to judge) | 0.018 | 0.011 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".