A spatial panel analysis of prostate cancer detection across the United States has found that the age-adjusted incidence of the disease fell at the state level over the 2010s, while screening activity, specialist availability and demographic composition showed distinct ecological associations with observed case counts. The study, published in BMC Health Services Research, tracked 48 contiguous states plus the District of Columbia from 2010 to 2020, assembling 539 state-year observations to examine how reported incidence moves alongside healthcare resources and population characteristics.
The unweighted mean state-level incidence declined from 139.27 cases per 100,000 males in 2010 to 108.23 per 100,000 in 2020. The authors caution that this downward trend reflects both changes in underlying disease occurrence and shifts in case detection, since observed incidence is a product of true biology and diagnostic intensity, and prostate cancer incidence is particularly sensitive to prostate-specific antigen (PSA) testing practices.
To capture geography, the researchers evaluated spatial clustering with Global Moran’s I and local statistics. Annual values ranged from 0.052 to 0.115, indicating modest positive spatial autocorrelation with permutation-based pseudo-P values below 0.05. In plain terms, states with similar reported incidence rates were not scattered randomly across the map but tended to sit near one another.
The centerpiece of the analysis was a two-way fixed-effects spatial Durbin model using inverse-distance weights, which allowed the team to separate direct, within-state associations from model-defined spatial spillover effects. Direct positive associations emerged for unemployment (an estimate of 1.902, with a 95 percent confidence interval of 0.871 to 2.933), the proportion of Black males (4.927, 3.399 to 6.455), the proportion of males aged 65 and older (2.051, 0.109 to 3.993), and urologist density (4.073, 0.704 to 7.442). Each estimate corresponds to a one-percentage-point increase in the exposure, except urologist density, measured as one additional urologist per 100,000 population.
PSA testing told a more nuanced story. Its direct within-state association with incidence was not statistically significant (0.278, −0.042 to 0.599), but the model identified a positive spatial spillover association (2.123, 0.324 to 3.923) and a positive total association (2.402, 0.610 to 4.194). This pattern means that, in this modeling framework, PSA testing levels in nearby states were associated with higher observed incidence within a given state, even after accounting for within-state testing levels.
Not every association survived scrutiny under alternative specifications. The proportion of males aged 65 and older showed a negative spatial spillover association (−14.342, −27.871 to −0.813), while the uninsured proportion showed no significant association with incidence. When the researchers swapped inverse-distance weights for Queen contiguity weights or applied a two-year lag to PSA testing, the spillover signal became nonsignificant, though it remained significant when analyses were restricted to survey years in which PSA testing was directly observed. The urologist-density direct association also lost significance in lagged models, underscoring how sensitive these ecological estimates are to modeling choices.
The authors are explicit about the limits of what such a design can show. As a longitudinal ecological study, it analyzes states rather than individuals, and no causal relationships or actual interstate healthcare pathways can be established from these data. The spillover estimates are model-defined quantities that depend on the assumed spatial weights and temporal structure, not measured flows of patients or services across borders. Incidence, moreover, mixes genuine disease occurrence with case detection, so an association with testing or specialist supply may reflect diagnostic intensity as readily as disease burden.
Within those constraints, the findings describe prostate cancer detection in the United States as a spatially structured, resource-related phenomenon in this dataset. The positive association with the proportion of Black males and the urologist-density association stood out among the contextual characteristics examined, while the uninsured proportion showed no significant association with observed incidence in the primary model.
The authors conclude that PSA testing, specialist resources and contextual characteristics each showed distinct ecological associations with observed incidence, and that all such findings should be read through the twin lenses of case detection and underlying disease occurrence — a distinction that remains central to interpreting prostate cancer statistics.
Subject of Research: Spatial associations of PSA testing, urologist density, insurance coverage, and demographic factors with state-level prostate cancer incidence in the United States, 2010–2020
Article Title: PSA testing, specialist healthcare resources, insurance coverage, and spatial spillover associations with state-level prostate cancer incidence in the United States: a spatial panel analysis, 2010–2020
Article References: Liu, Y., Wei, J., Ma, Z., Xu, L., & Zhang, P. (2026). PSA testing, specialist healthcare resources, insurance coverage, and spatial spillover associations with state-level prostate cancer incidence in the United States: a spatial panel analysis, 2010–2020. BMC Health Services Research. https://doi.org/10.1186/s12913-026-15925-w
Image Credits: AI Generated
DOI: 10.1186/s12913-026-15925-w
Keywords: prostate cancer, PSA testing, spatial analysis, urologist density, health insurance, incidence, United States, ecological study, spatial Durbin model, health services research, spatial spillover, testing
News Source: Nathaniel Bowman. (October 23, 2026). PSA Testing Shows Cross-State Links to Prostate Cancer Detection in US Spatial Analysis. Scienmag.
