Cause-of-death-attribution data is intrinsically unreliable
Assignment of cause of death, with infectious diseases and comorbidity, is not only technically difficult (e.g., Simonsen et al., 1997; Marti-Soler et al., 2014) but also contaminated by physician-bias, politics and news media.
This has been known since modern epidemiology was first practiced. Here is Langmuir (1976) quoting the renowned pioneer William Farr, regarding the influenza epidemic of 1847:
Farr uses this epidemic to chide physicians mildly on their narrow views pointing out that sharp increases were observed not only in influenza itself but in bronchitis, pneumonia and asthma and many other non-respiratory causes, he states:
'… there is a strong disposition among some English practitioners not only to localize disease but to see nothing but the local disease. Hence, although it is certain that the high mortality on record was the immediate result of the epidemic of influenza, the deaths referred to that cause are only 1,157.'
And, such bias is generally recognized by leading epidemiologists (Lui and Kendal, 1987):
… the decision to classify deaths into "pneumonia and influenza" is subjective and potentially inconsistent. On one hand, the effect of influenza or influenza-related pneumonia may be underestimated because underlying chronic diseases, particularly in the elderly, are usually noted as the cause of death on the death certificate. On the other hand, after influenza activity has been publicly reported there may be an increased tendency to classify deaths as due to "pneumonia and influenza," thereby amplifying the rate of increase in P&I deaths or, when a decline in influenza activity is reported, a bias toward decreasing the classification of deaths related to "pneumonia and influenza" may result. Surveys to evaluate these possibilities have not been done.
3
One can reasonably expect that in the current world of social media, with a World-Health-Organization-declared (WHO-declared) “pandemic”, such bias will only be greater compared to its presence in past viral respiratory disease epidemics.