Monday, 23 March 2020

Data insufficiency in understanding Covid

Because most Covid victims are with pre-existing problems, to really understand what it does, to whom and how much (which should be essential for any reaction from Govt to individual level), data set we analyse should at least have,
  • Total deaths expected (for season),
  • Total deaths with covid,
  • Normal deaths by causes
  • Deaths by causes with covid
  • Normal comorbidities
  • Covid deaths comorbidities
Plus like there are attempts to capture skewness in Covid victims for parameters like Male/Female, Blood Group, comorbidities, similar data should be captured for Unaffected/mild/sever/victims for parameters like below and others to make data readable.
  • Food habit (Veg/NonVeg, Traditional/Western, fresh/processed)
  • Alcohol consumption
  • Smoking,
  • Treatment and outcome (Allopathy/Ayurved/other TM/combination + change in post survival health)
Lets take few hypothetical examples for a hypothetical city/state/country with population of 4 cr. for the active period (say three months) of outbreak to demonstrate how a plain 'death' number doesn't tell you much. (Examples are deliberately showing different sides extreme points for easier understanding)
Here, it appears that,
Example 1: Only those who would have anyway died have died, i.e. no extra deaths and yet Covid death number stands at huge 50k.
Example 2: More total deaths with same covid deaths, i.e. most died like 1 above but few who would have lived for next few months if not infected, died few months earlier due to covid. i.e. extra deaths over longer period is zero.
Example 3: Compared to 1 & 2, covid deaths are less by 10k at 40k but here, 10k extra have died who would not have died in ultra short term and 5k in short term which makes this a worse scenario compared to 1 & 2.
Example 4: deaths attributed are drastically lower at 10k and yet its similar to 3.
Example 5: death count similar to 4, but all are extra deaths and huge no. comes from with no pre-existing conditions.
Example 6: Something similar to what we think when we see numbers now.

My guess is, reality would be somewhere between 3 to 6, but unless we know what it is (together with skewness data), there is no way to tell how big and what kind of problem is this and what kind of short/medium and long term (sustainable) changes will help.

I dont know if/how/how much data can be collected, but our Vaidyas could also profile unaffected/mild/sever/dead cases based on Vata, Pitta, Kaph and Satv, Rajas, Tamas, dhatus etc balance/imbalance and other relevant classifications.

No comments:

Post a Comment