Serosurvey suggests ~21% of NYC was exposed to SARS-COV-2 despite 1 month lockdown, while less impacted states are now lifting lockdown. Paradoxically, the latter may be at much higher risk than the former, and mildly impacted states may suffer badly for their mistake. A thread.

If lower # of cases in some states are due to lower contact rates than in NYC, epidemics there will be lower and relaxing lockdowns will increase cases but won't lead to NYC-intensity epidemics. But another possibility is more likely: they are simply earlier in epidemic curve.

If so, then they will suffer doubly due to "epidemic momentum". Nice 2011 paper by @DeirdreHoll
doi:10.1371/journal.pcbi.1001076 Fig 3
showed how temporary interventions can not only greatly reduce peak but also total fraction infected.

Key detail: reductions depend on the efficacy of control and time implemented. Temp. interventions started before much transmission occurs shift time of epidemic later but do little to reduce the height b/c no immunity has built up, as has occurred in many less impacted states.

In contrast, control measures implemented during peak transmission both "flatten the curve" and lead to much smaller epidemics when interventions are lifted b/c of buildup of herd immunity. Thus, current mildly impacted states may suffer huge epidemics if lockdowns are lifted.

To be clear, best strategy is to lockdown early and shift to test-contact trace-quarantine (T-CT-Q) when T-CT_Q capacity set up (hopefully soon NYC!). But worst time to lift control measures without T-CT-Q is when few cases have occurred. I worry for many states.

@CT_Bergstrom wrote a nice thread illustrating this idea that mirrors the paper by @DeirdreHoll cited above: https://twitter.com/CT_Bergstrom/status/1252075528711860224