This ecological study examined 491,769 COVID-19 cases across 1,831 census block groups in Miami-Dade County, March 2020 to May 2021. Getis-Ord Gi* hot spot analysis and negative binomial regression with spatial eigenvector filtering were run for five three-month periods. Early hot spots had median household incomes of $52,022 to $55,676, but from September 2020 the median was $77,069. Cold spots then gathered in historically Black block groups along the I-95 corridor, with 19% uninsured against 13% countywide. From then, a higher Black population share predicted lower reported rates (IRR 0.991 to 0.995). Testing rates were unavailable, so detection gaps could not be measured directly.
Why it is interesting: Reported case rates here appear to reflect access to testing as much as transmission. Known risk factors appeared protective, and block group analysis exposed a gradient ZIP code averages would hide.
Abstract — the authors' own words, CC BY
Passive disease surveillance systems may systematically undercount burden in low-resource minority neighborhoods, yet fine-scale evidence of this phenomenon remains limited. We examined spatiotemporal patterns of reported COVID-19 cases across Miami-Dade County's diverse neighborhoods and investigated which socioeconomic and demographic characteristics predicted areas of elevated and suppressed case rates. In an ecological, cross-sectional analysis of 491,769 geocoded cases across 1,831 census block groups from March 2020 to May 2021, we applied negative binomial regression with spatial eigenvector filtering and Getis-Ord Gi* hot spot analysis across five pandemic periods aligned with major policy shifts. The analysis revealed a detection paradox. Early hot spots concentrated in lower-income census block groups (median household income: $52,022-$55,676), but by September 2020, hot spots had shifted to more affluent areas (median: $77,069). Historically Black neighborhoods along the I-95 corridor emerged as persistent cold spots despite multiple structural risk factors: lower incomes, reduced insurance coverage, and limited healthcare access. From Period 3 onward, predominantly Black census block groups showed significantly lower reported rates (incidence rate ratio [IRR]: 0.991-0.995, p < 0.001). Household overcrowding was tested but retained in no final model. These paradoxical patterns suggest that as testing expanded, surveillance increasingly tracked healthcare utilization rather than transmission. What appeared as neighborhood "protection" likely reflected barriers to testing access shaped by transportation constraints, work schedule inflexibility, and historical relationships with medical institutions. Census block group analysis revealed these dynamics obscured at broader geographic scales. The findings challenge the assumption that reported case rates approximate true disease burden and have direct implications for emergency response planning: surveillance systems require complementary detection streams, risk-adjusted monitoring for underdetection, and targeted outreach where poverty and limited access to testing are concentrated but reported rates are low.
Cite this item
BibTeX
Free sign-up. A digest of the new papers twice a week, grouped by domain.
Other drivers in Data, models and systems
-
Forecasts and their record
How the forecasts are made, how well they explained the past, and how well mortality can be forecast — and by whom.
4 this week -
What linked data reveals
What linked registries make it possible to know about whole populations
8 this week -
AI in medicine
Whether machine learning predicts, detects and decides better than the tools we have, and whether that has yet changed an outcome for a population.
3 this week -
Health systems and access
What the health system can provide, who gets it, and whether the gaps in health between groups are widening.
4 this week -
Value in health
What a treatment, health programme or policy is worth against what it costs, and how far the models that decide it can be trusted.
5 this week -
Revisions and new sources
What the figures are built from, what is newly available, and how much the numbers move when they are revised.
6 this week
