Is it appropriate to compare rates of (for instance) tea consumers among patients admitted to a hospital with some disease, with rates of tea consumers in general population in the same country? (Given that in terms of age and sex representation, the group of patients and general population are reasonably similar).
Would difference between (for instance) 10 % of tea consumers (with upper limit of 95% confidence interval 12%) and 20% of tea consumers in general population be convincing as evidence for a protecting effect of tea consumption against the disease?
I do understand that randomized case-control study would be much more convincing. Nevertheless, can the above mentioned be considered as evidence too?
Provided that there was not something associated with the particular hospital that biased the patient referral for or against tea drinkers, this could be conceived as a case-base study and the relative exposed/non exposed (tea /non tea) compared between the cases (patients with specific disease) and the base (general population) giving an odds ratio.
eg patients (100 say) general pop (100000 say)
tea 10 20000
no tea 90 80000
Odds ratio = (10 x 80000) / 90 x 20000 = 0.44 (preventive effect)
You ask: "Is it appropriate to compare..." I guess it depends on what you want to find out, i.e. what is the source population that you want to make conclusions about?
Example:
for instance you want to find risk factors for severe (need for hospitalisation) symptoms of COVID-19.
Among all males aged 70-79 patients submitted to a hospital with severe COVID-19 you find 40% are smoking.
In your general population you known that only 20% of males aged 70-79 are smoking.
You conclude that smoking poses and atributional risk for severe COVID-19 symptoms.
This conclusion neglects that you are only at risk of developing COVID-19 if you are infected with SARS-COV-2. The populations that you should actually compare the hospitalised patients with is not all males aged 70-79, but infact only those at risk of developing either no, mild or severe symptoms, namely those who have been infected.
To conclude my example: if, smoking is associated with a higher risk of getting infected (e.g. smokers could be from another socioeconomical class, have other movement-patterns, adhere less public recommendations on transmission prevention, more often live in urban areas...) their could actually be e.g. 40% smokers among all males aged 70-79 that are infected with SARS-COV-2, and the exposure would be equally present in the group that develops mild, and those who develop severe disease.
And just to be clear. This was an example. I do believe smoking increases the risk of developing severe COVID-19 if infected, based on what we currently know about the disease.
I think, that a conclusion to consider all possible explanations without a prejudice, is a a reasonable one. (Keeping an eye on trend is another one)
- Badges
- Science topic
- Similar topics
- Medicine
- Epidemiology