Ask a family living with dementia with Lewy bodies how common it is and, until this year, the honest answer was a shrug dressed up as a statistic. Textbooks called it the second most common cause of degenerative dementia after Alzheimer disease, then cited ranges so wide they were useless for planning a clinic rota, let alone a life. That gap is what Daniele Urso, Stefano Giannoni-Luza and colleagues at the University of Bari 'Aldo Moro' in Tricase, with co-authors at Cambridge and Geneva, set out to close in JAMA Neurology, published online on 11 May 2026. Their systematic review and meta-analysis screened 2,520 records, read 88 full papers, kept 16 population-based studies and pooled 12 of them.

The headline figure is 4.79 new cases per 100,000 person-years across all ages, with a 95 percent confidence interval of 3.90 to 5.88. That is a small number until you split it by age. Among people aged 65 and over, pooled incidence was 46.85 per 100,000 person-years (95 percent CI 23.78 to 92.30); among those under 65 it was 0.34 (0.14 to 0.83). Prevalence, the count of people living with the condition at a given moment, is thinner still: only one included study reported an all-age figure, 19.13 per 100,000 population, and the pooled figure for people aged 65 and over, 352.26 per 100,000, came from just two studies of populations older than 70 and carried a confidence interval running from 112.25 to 1,099.79. Incidence was numerically higher in men than women, 5.45 against 4.32 per 100,000 person-years, a difference that was not statistically significant. For scale, the authors note their pooled DLB incidence sits above previously reported population figures for frontotemporal dementia (2.28), amyotrophic lateral sclerosis (1.59), progressive supranuclear palsy (0.81) and corticobasal syndrome (0.28) per 100,000 person-years.

The second finding is the disagreement itself, and it is not a footnote. Between-study heterogeneity was high, at an I2 of 85 percent or more, meaning the studies were not measuring quite the same thing in quite the same way. The included work used the McKeith consensus criteria, DSM classifications and ICD coding, and came from registries, cohorts and multiphase surveys; heterogeneity in reconstructed cohort studies was lower than in claims-based ones. The authors flag that diagnostic criteria changed over the period covered, that clinical and pathological overlap with Alzheimer disease and Parkinson disease dementia creates a real risk of misclassification, that most studies came from high-income countries, and that they did not formally test for publication bias because there were too few studies per outcome. A pooled estimate built on that foundation is a floor, not a census.

What sharpens the point is what happens when one population is followed prospectively with the most recent criteria. In Neurology, the same Bari group reported a two-year, population-based incidence study in Salento, southern Italy, covering 767,356 residents from 1 March 2023 to 28 February 2025 and using the 2017 DLB Consortium criteria with standardised assessment of core features. Sixty-two incident cases emerged over 1,534,712 person-years: crude incidence 4.04 per 100,000 person-years, age-standardised 3.77, higher in men (4.87 crude) than women (3.27), peaking at ages 80 to 84 at 30.83 per 100,000 person-years. Onset before 65 accounted for 6.5 percent of cases. Alzheimer biomarkers, by spinal fluid or amyloid PET, were available in only 45.2 percent of those patients, and three quarters of those tested were positive. In other words, even a well-run modern study can only biomarker-check about half its cases, and most of the studies feeding the global pool predate that possibility altogether.

For a household, the practical value of these figures is not personal risk. It is leverage. When a support charity argues for a specialist nurse, when a health board decides whether DLB gets its own care pathway or is folded into a generic dementia service, when a trial sponsor works out how many sites it needs to recruit, the answer has always turned on numbers nobody could defend. There is now a citable set, from population data rather than memory clinics, with its uncertainty printed alongside. The investigators write that the results offer "a robust reference for future epidemiologic research and public health planning", while pressing for standardised diagnosis and for populations that current studies barely touch.

The sex pattern deserves the same caution. Higher incidence in men appeared in both the pooled analysis and the Salento cohort, but in the pooled data the difference did not reach statistical significance, and in Salento the sexes differed in how the condition looked rather than only in how often it occurred: men scored higher on motor severity (MDS-UPDRS part III 23.7 against 14.4, p = 0.002), women higher on neuropsychiatric burden (Neuropsychiatric Inventory total 24.0 against 13.0, p = 0.041). Those are group averages from one region of Italy, not a rule about any individual, and they say nothing about what any one person's course will be.