Thread by @tomtom_m

Quite a few people have been sharing a copy of part of a South African death certificate #onhere, (including some medical practitioners) claiming that this shows that Covid deaths are being exaggerated, misidentified, and misreported. A thread on Cause of Death Certification. /1

2/ What was being shared? Let’s look at one of the images.

Other tweets have made the opposite claim: (e.g.:

https://twitter.com/NadAhm3/status/1352340191323840514

)

3/ The Notification of Death form in South Africa, DHA1663, is in two parts. A sample can be found on pp9-12 of this document: samrc.ac.za/sites/default/…

4/ Part A, pages 9-11, of the referenced document, covers the basic information on the decedent. When a death is reported to Home Affairs, some – not all, by any means! – of the information in Part A is captured in real time, in order to produce the death certificate.

5/ The death certificate, which is produced in real time, allows the executors of the deceased’s estate to begin the long process of winding it up; for insurance claims to be made etc

6/ What goes on to the certificate is very limited: ID number, Full Name; dates of birth and death; office where the death was notified; and – simply – whether the death was due to natural or unnatural causes.

7/ Note that in Part A, other than the determination of unnatural death (which would require a forensic report or autopsy), there is no indication given as to cause of death.

8/ (This is of relevance to those producing the weekly SAMRC excess deaths report; that information is simply not available. We are asked why we do not report on death certificates with ‘covid’ – because that’s not captured in real time.

9/ Information on causes of death is collected in Part B (copy on page 12 of the linked document) and has to be completed by a doctor, professional nurse or forensic pathologist. On completion, Part B is sealed, and is only processed much later, by Stats SA. More on this later.

10/ Field 77 deals with causes of death. This section is itself in two parts (Parts 1 and 2), and follows the WHO recommendations on how causes of death should be recorded.

11/ Part 1 offers UP TO FOUR sequential causes of death – starting from immediate, and working back to the last entry, the underlying cause.

12/ In the screenshot shown in the 2nd tweet, these are metabolic acidosis (which immediately caused death), caused by septicaemia, caused by pancreatitis. Part 2 asks about ‘other significant conditions contributing to death but not resulting in underlying cause given in Part 1’

13/ On the right hand side, we can see (‘for office use only’), blocks for recording the ICD-10 codes of each causal contributor.

14/ The International Classification of Diseases (ICD, version 10) is the current standard for coding disease and cause of death. (It is also in the process of being upgraded to version 11).

15/ A description of ICD-10 can be found here: who.int/classification… and there are numerous online search engines that allow looking up of ICD codes.

16/ The WHO guidance is clear: for tabulations of cause of death, it is the last mentioned item *in part 1* that should be recorded as the underlying cause of death. In the example, then, that would be pancreatitis.

17/ The allocation of ICD-10 codes is highly specialised (and increasingly automated). Stats SA receives the (still sealed) Part B, along with the Part A that had been partially captured at the time of notification.

18/ Stats SA then set about coding-up the causes of death, the process of which is described here: samrc.ac.za/sites/default/…. It takes some time. In March 2020, the coded causes of deaths in 2017 were released.

19/ Anonymised unit record data are in the public domain, and deaths from 1998-2017 are available on Stats SA’s website (warning: the file is LARGE).

20/ Again, Stats SA follows WHO recommended practice, determining the underlying cause of death from the last-mentioned cause in Part 1 of Section 77. Information from Part 2 is captured and coded, but this does not go into the causal chain.

21/ That said, a large (c25%) proportion of deaths in South Africa have an ‘unknown’ cause (coded R99). This means one cannot use these data uncritically.

22/ In the example given, COVID will be recorded as a contributory factor, but will not be recorded as the underlying cause of death, or even an element of the causal chain. And researchers will, typically be most interested in the underlying cause.

23/ However, there is scope for editing / correcting the causes where the sequence is illogical, or deemed to be incorrect. How much this happened historically, or will happen post-Covid, is unclear.

24/ My understanding is that this is quite uncommon (historically). But this is something to keep an eye on…

25/ If not changed, the tabulations which we will eventually be able to draw on will allow us to see the death in the screenshot as one that COVID may have contributed to the pancreatitis being fatal, but COVID was not the underlying cause of death.

26/ The second example may well have Covid on the causal pathway. (Curious however that the first showed a picture of a sealed, confidential, portion though)

27/ Doctors in South Africa are trained in how to complete death notification forms, so it is somewhat alarming that some medical practitioners have shared the original image, uncritically, and without alerting their followers to the misinformation. /fin