Overview
Page A is the curated page at dizziness-and-balance.com/disorders/dizzy_epi.html. It is authored by Timothy C. Hain, MD, and was last modified on March 31, 2023. It contains 3844 words.
Page B is the AI page at dizziness-and-balance.com/ai/disorders/dizzy_epi.htm. Its byline states that it was written by Gemini AI on 07-09-2026. It contains 805 words.
Content quality
Page A is a detailed, evidence-based synthesis. It breaks dizziness into categories (otologic, medical, neurological, psychiatric, and undiagnosed), explains why definitions and study populations distort the literature, and offers a specific bottom-line estimate: about 15% of the population has dizziness, of whom 40% have otologic dizziness, 10% have central dizziness, 25% have medical dizziness, and 25% have undiagnosed dizziness. It also presents multiple tables of measurements, including one-year prevalence figures from individual studies (ranging from 3.13% in Taiwan insurance claims to 36.2% in an Australian cross-section), table-level data for Meniere's disease, BPPV, migraine vertigo, orthostatic dizziness, and unlocalized dizziness, plus a table of diagnoses in dizzy patients across different clinical settings. The page critiques specific studies and discusses methodological problems at length.
Page B is a general narrative overview. It reports that dizziness affects roughly 15% to 30% of the general adult population, that prevalence in people over 65 can reach over 30-50%, and that the annual incidence is approximately 5-10%. It lists risk factors (age, gender, comorbidities, medications, lifestyle), describes four subtypes (vertigo, disequilibrium, presyncope, non-specific dizziness), and discusses falls and healthcare burden. It contains no tables, no specific study-by-study data, no breakdown of diagnostic categories, and no discussion of how dizziness is defined or why definitions matter.
Page A is heavily grounded in specifics: it names studies, settings, populations, and percentages throughout. Page B stays at a general level and does not engage with the diagnostic ambiguity that Page A treats as a central issue. Page A also provides the only quantitative statements about the US impact of dizziness, such as dizziness being the primary complaint in 2.5% of all primary care visits, or about 8 million visits per year, and an estimated economic impact of roughly 2.25 billion dollars per year.
References to journal articles
Page A contains a dedicated reference list with 45 entries to journal articles and other sources, and its text and tables cite many additional studies inline (for example, Lai et al 2011, Humphriss and Hall 2011, Mendel et al 2010, Wiltink et al 2009, Gopinath et al 2009, Neuhauser et al 2008, and Kroenke et al 2000).
Page B contains zero references to journal articles. It names no studies, no authors, and gives no citations for any of its figures.
Authoritativeness
Page A is the more authoritative of the two. It is attributed to a named physician, carries a modification date, relies on a long list of cited studies, and is candid about uncertainty, stating for example that "relatively little is known about the precise incidence and prevalence of dizziness" and that its own estimates "may change" with more data. It also evaluates its sources critically: it calls the inference that all balance dysfunction is vestibular dysfunction a "leap of logic," and it says that ICD-10 codes "are not to be relied upon for anything important."
Page B is far less authoritative. It is attributed solely to Gemini AI, has no named human author, contains no citations, and makes broad claims without any supporting evidence. It does not state the limits of the data, does not compare the results of different studies, and does not mention any of the specific research literature that Page A discusses. Its confident presentation is not backed by any verifiable source within the page.
Hallucinations in the AI page
Judged only against the text of Page A, the following statements in Page B are unsupported or inconsistent with the supplied text:
- Page B claims that BPPV "is the most common cause of recurrent vertigo, with a lifetime prevalence around 2.4%." Page A contains no lifetime prevalence of 2.4% for BPPV. The BPPV figures in Page A are a 1.5% one-year prevalence in 4869 adult Germans, 9% in elderly patients, 0.46% in German insurance records, and 17.1-41% of dizzy patients in clinical settings.
- Page B claims that for individuals over 65 years, the prevalence of dizziness "can soar to over 30-50%." The highest older-age figures in Page A are 36% (2011 persons over age 70 in Sweden) and 30% (elderly metanalysis). No source in Page A supports a figure of 50%.
- Page B states that longitudinal studies suggest an annual incidence of approximately 5-10% in the adult population. Page A provides no annual incidence figure for adults at all.
- Page B closes with a line reading "All rights reserved © 2023." This conflicts with the page's own byline, which dates the article as written by Gemini AI on 07-09-2026, and the year 2023 is not corroborated anywhere in the text of Page A or Page B.
Summary table
| Aspect | Page A (curated) | Page B (AI) |
|---|---|---|
| Word count | 3844 | 805 |
| Named author | Timothy C. Hain, MD | Gemini AI |
| References to journal articles | 45 in the reference list, plus further inline citations | None |
| Data tables | Several | None |
| Specific prevalence estimates | Many, with individual studies cited | A few broad ranges, uncited |
| Methodological critique | Extensive | Absent |
| Hallucinated or unsupported figures | None identified | At least three unsupported figures, plus an inconsistent copyright year |
Overall, Page A serves as a dense, critical, and referenced source on the epidemiology of dizziness, while Page B is a short, unreferenced overview that introduces unsupported numerical claims. For any reader seeking reliable figures, Page A is clearly the more trustworthy of the two.