A key issue with confirmed cases is that reported results have far more to say about testing practices than underlying ground truth. A confirmed case requires an infected individual, a test, and a positive result.<p>(False positives are rare, false negatives fairly common.)<p>At county-level resolution, I'd argue that direct comparisons are at best difficult, and may be effectively meaningless. Rapid rises in values may indicate outbreaks ... or more intense monitoring. Low results, especially in populated counties, are a very likely <i>warning</i> sign -- the outbreak can now be presumed nearly universal. As a rule of thumb, you might rank US state public health effectiveness by dates of first detection of community spread of COVID-19, earlier being better.<p>Death rates are far more reliable true measures than monitoring, though they both lag infections (by about three weeks, though confirmed cases by less, possibly only one week), and are far thinner -- about 1% of all infections 3 weeks ago given typical course of illness.<p>The reliability comes from the fact that bodies are far less convenient to hide, and a far smaller and targeted testing target.<p>Test results, coverage, and negative:positive ratios (higher is better, suggesting better coverage) are all highly useful, as is case severity mix -- more mild cases suggests more complete testing.