Hospitalisation Rate For Adult Respiratory Conditions (part 4)

Lessons from an undisclosed NHS Trust

In my previous post in this series we discovered that the weekly hospitalisation rate for adult respiratory admissions to the emergency department of a sizeable UK NHS Trust during weeks 11 to 40 of 2020 were no different to those observed for the same weeks in 2017 and 2019, with 2018 being an exceptionally quiet year. It was only during weeks 41 to 53 of 2020 could we detect signs of elevated rates over and above that of previous years. We also discovered that rates for weeks 1 to 6 of 2021 were no different to those observed for the same weeks in 2017, with weeks 7 to 38 of 2021 mirroring those for 2017 and 2019. In sum we didn’t find much that would suggest a pandemic had gripped the nation aside from weeks 41 to 53 of 2020, when hospitalisation rates were exceptionally high compared to historic values. This is bizarre because it suggests that the first and second waves were non-existent in terms of hospitalisation of adult respiratory admissions; only in the third wave do we see evidence of something going on!

To clarify the situation I ran the analysis again but including only those adult respiratory admissions that required emergency treatment for respiratory conditions. The attached slide is not at all what I was expecting! I would have put good money on the three waves of 2020 distinguishing themselves in elevated hospitalisation rates of adult respiratory admissions requiring emergency treatment but we find the opposite: all five years merge into a tangle of coloured spaghetti in which we are less able to distinguish anything tasting of a novel viral outbreak, and a deadly one at that.

The only peculiar features my eyeballs note is zero intake of these cases during weeks 40 to 43 of 2017 (this may be a localised bed management issue or a sudden surge of very mild respiratory disease). The second wave pandemic of 2020 rears its head for weeks 40 to 44 and that’s about it – there’s nothing showing for the first or third waves. Weeks 1 to 38 of 2021 with all the new variants try to make a showing but nothing really sticks out. None of this makes any sense.

In terms of rigorous statistical testing I once again adopted a paired samples T-Test using a Bonferroni correction for five repeated tests (the nominal alpha level of p=0.05 becomes p=0.01). In the geeky table below we find three statistically insignificant comparisons and four highly statistically significant comparisons. These us that weekly hospitalisation rates for the 37 matching weeks of 2021 and 2020 are not significantly different (p=0.083). They also tell us that hospitalisation rates for the 37 matching weeks of 2021 and 2019 are not significantly different (p=0.653). They do, however, tell us that hospitalisation rates for the 37 matching weeks of 2021 and 2018 are highly significantly different (p<0.001), as are the 37 matching weeks of 2021 and 2017 (p<0.001). In terms of full year comparisons they tell us that the 52 matching weeks of 2020 and 2019 are not significantly different (p=0.592), whereas the 52 matching weeks of 2020 and 2018 are highly significantly different (p<0.001), as are the 52 matching weeks of 2020 and 2017 (p<0.001).

This is interesting because if we pick the bones out of this then 2020 and 2019, when compared week-by-week, are pretty much identical in terms of hospitalisation rates adult respiratory admissions that required emergency treatment, with overall annualised means fetching up at 0.682 and 0.671 respectively. What kind of pandemic mimics the previous year?