I would like to conduct a study to look at the effect COVID-19 pandemic lockdown has on the following:

  • number of cases at the hospital emergency department (broken down into the different triages and different complaints/diagnoses
  • waiting time to consultation
  • returns to hospital emergency department within 24 hours (percentage data)
  • subsequent admission to general ward or ICU

The pre-period (i.e. before lockdown) and the post-period (i.e. after implementation of lockdown) are not consecutive. We are looking at the same months both in pre- and post-periods, each having a total of 60 days.

Would it be okay if I use t-test to compare pre- and post-period data? But that would mean aggregating the data into 1 data point pre and 1 data point for post. Would we lose resolution here?

Or should I do an interrupted time series, aggregating by day?

Appreciate any advice!

  • $\begingroup$ If you were to use a t test, what would $n$ be? Are you looking at various locations? Countries, cities, US states? $\endgroup$ – BruceET Oct 9 '20 at 5:22
  • $\begingroup$ I think for the case of number of cases at the hospital emergency department; i could use the different triage level as the n, but that would just be 4. Or I could do a pairwise t-test and make day-to-day comparison. So 1 march 2019 vs 1 march 2020, I don't know if this make sense $\endgroup$ – HNSKD Oct 9 '20 at 6:41
  • $\begingroup$ If you have only one location, then doing a t test is probably not the best way to analyze the data. In a time series approach, it may be best to try to match days of the week SMTWThFSa rather than days of the month. $\endgroup$ – BruceET Oct 9 '20 at 6:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.