Suppose perfect lockdowns, so no new infections after lockdown, not even family members. Suppose incubation is 1w, then 1w of worsening syntoms, and then death

Suppose perfect lockdowns, so no new infections after lockdown, not even family members. Suppose incubation is 1w, then 1w of worsening syntoms, and then death at 3w, consistent with Italy and other countries (sp. in places with enough beds, deaths seemed to have occurred later). Then, isn't the whole premise wrong? You want to study lockdowns but only look at deaths from infections BEFORE the lockdown? Am I missing something?