Skip to main content

Interpretation and limitations

What the data support​

The current aggregate data support descriptive small-area comparison, resource-quality accounting, within-area heterogeneity analysis, rank and classification comparison, HCS hypertension triangulation, prespecified sensitivities, and supplementary ecological models.

The results can identify places where a coarse label hides clinically recorded tract variation and where additional public-health assessment may be useful.

What the data do not support​

The analysis does not:

  • estimate prevalence among all Chicago residents;
  • establish that one source is correct;
  • infer unmet need or service need;
  • locate or evaluate health services;
  • estimate individual risk;
  • identify causal intervention targets; or
  • prove literal information loss from mathematical aggregation.

The tract and community-area values were direct outputs from the same source at different geographic levels. Their differences are cross-frame differences.

Principal limitations​

  1. Inclusion depends on care seeking, participating systems, coding, and geocoding.
  2. Aggregate data cannot independently verify the person-level eligibility and cross-system deduplication sequence.
  3. Unresolved zero or suppressed values and the denominator threshold change the analytic population by condition.
  4. The complete 50%-area alternative frame requires 2 tracts absent from the current analytic file.
  5. Direct tract and community-area exports may differ in construction beyond geographic scale.
  6. Rank categories depend on the eligible distribution and tie rules.
  7. PLACES and HCS differ from Chicago Health Map in population, phenotype, period, and estimation method.
  8. Cluster bootstrap intervals do not fully represent spatial dependence or source-estimate uncertainty.
  9. The supplementary ecological model has at most 77 areas and residual spatial structure.
  10. Combined diabetes was excluded because its source semantics remained unresolved.
  11. Temporal and geographic transportability were not tested because a frozen 2025 extract and compatible external-network extract were unavailable.

Data or analyses that would strengthen the paper​

  • an aggregate annual flow of eligible adults, duplicates removed, geocoding exclusions, and contributing systems;
  • source-owner confirmation of eligibility, deduplication, diagnosis, denominator, zero, suppression, and reliability rules;
  • compatible uncertainty for Chicago Health Map and PLACES;
  • a complete 50%-area alternative tract frame;
  • a frozen 2025 out-of-time replication; and
  • a second metropolitan multi-system EHR dataset with direct tract and direct coarser-area estimates.