WHO Medical Certification of Cause of Death

Hi @akanter,

At the ICRC we are making a first iteration at implementing a form to produce the WHO certification of cause of death:

Starting with section I, the form captures two sets of data:

  1. The disease/condition directly leading to death, (a).
  2. Two possible antecedent causes, (b) and (c).

The existing DHIS2 tracker (see reference 1) goes up to four possible antecedent causes numbered (b), (c), (d) and (e).

In OpenMRS terms I would tend to see this as five obs constructs for (a), (b), (c), (d) and (e) that share the same questions:

  1. Cause of death (Question/Coded), eg. ICD10-K720.
  2. Interval between onset and death (Question/Numeric), eg. 1.
  3. Units of interval between onset and death (Question/Coded), eg. Hours.
  4. Underlying cause of death (Question/Boolean), eg. True.

Hopefully there is a more elegant way to that but I would basically end up with five convenience sets to record those five “same” obs constructs (same in the sense that they are made of the same questions):

(a) Disease or condition directly leading to death
(b) First antecedent disease or condition leading to death
(c) Second antecedent disease or condition leading to death
(d) Third antecedent disease or condition leading to death
(e) Fourth antecedent disease or condition leading to death

How do we go about recording this?

Cc @ball @burke @jesplana @icrc.thonorio @ifernandes


  1. (DHIS2) WHO Cause of Death - Tracker and Event System Design
  2. (WHO) Medical certification of cause of death : instructions for physicians on use of international form of medical certificate of cause of death, 4th ed

PIH implemented a death certificate form. Unfortunately it’s never been used in production. (Long complicated story but no reflection on the functionality)

Thanks @ball! Actually your custom tag <causeOfDeathList/> pointed me (here) to CIEL:1816.

And I guess that’s what I was maybe looking for…

@akanter is CIEL:1815 (Diagnosis sequence number) the numbering found on the death certificate going from (a) to (e)? I think so as per its description:

The number/order of a diagnosis on a list or death certificate.

If yes then that’s it, we will just need to add those questions to the CIEL:1816 ConvSet:

  • Interval between onset and death (Question/Numeric)
  • Units of interval between onset and death (Question/Coded)
  • Underlying cause of death (Question/Boolean)

162569 is the current cause of death list which can be used to capture Causes of death from a death certificate. I am not sure the need to include the “underlying cause of death” as opposed to contributor to cause of death as the ordering on the death certificate is usually sufficient. Can someone clarify if that is truly required?

I agree that the interval and units would need to be added.

Was what the MVP verbal autopsy used. I would propose using the duration of illness and duration units in the 162569 concept set.

This article about Death Certificate was useful to a non-clinician (me).

Source of information = 162566 Death certificate

@akanter you mean that the last one in the list is de facto the underlying cause of death?

I think people expect that the ordering of the top three are the causes of death and >3 are just associated. I don’t know if specifying one “underlying cause of death” has any meaning and why we would need to capture that element on each diagnosis.

In short this comes from DHIS2, and I don’t know why DHIS2 does that.

@jesplana would it be ok to just assume that the last entry is de facto the underlying cause of death? So the one reported as ‘Results’ with the encompassing SMoL code?

@mksd I think so but this also assumes that the person filling in the death certificate enters cause(s) of death in the correct order (which is how it’s done on paper anyway…).

If the cause of death is ‘unknown’ then the underlying cause of death is also ‘unknown’.

I think DHIS2 modeled it that way to allow ‘easy’ data analysis. As it was designed in a tracker program, it has a program rule that takes the data from a data element that has the check next to the ‘underlying cause of death’ and pushes the data value to the question: Underlying cause of death. Which allows for generating a list of ‘underlying causes of death’.

Here’s a short guide from the WHO:

Thanks @jesplana.

Ok that’s what I was suspecting, they added one more question to facilitate analysing the data. Let’s try to not do that in OpenMRS as it would require to add one more boolean question to CIEL:162569.

@akanter what about those ‘Results’ above, made in DHIS2 of:

  1. The underlying cause of death (as text).
  2. The SMoL code.
  3. The ICD-10 code.


Looks like 1. is unnecessary anyway, since it’s the FSN of 2. So the ‘Results’ are in the end made of two coded concepts:

  1. The SMoL code.
  2. The “ultimate” cause of death ICD-10 code.

How would you go about modelling this?

Hi @akanter . Any thoughts on @mksd comments about ‘Results’ on DHIS2?

Thanks for your help on this.

Cc: @mksd @jesplana

1 Like

Wow, I had no idea about all of this. We hadn’t needed to include SMoL or the one underlying cause. Since there is no way to know what the ultimate cause is from just the list, I do think we need a flag concept which would be applied to the last diagnosis on the list within Section I. I will look into whether CIEL should have SMoL codes as maps.

1 Like

@akanter thanks for checking if CIEL should have SMoL codes as maps. I’ll wait for your feedback.

I don’t know what the actual user demand is for these codes. Since CIEL has more specific codes, those are what are supposed to be used. The SMoL codes can be calculated from the ICD-10 codes so capturing them in CIEL is a large overhead that probably is not warranted (unless I hear otherwise). As for underlying cause, I think we do need to add a new boolean flag which would be set for the primordial cause in section I.

1 Like

1 Like

@akanter it sound good to me to have the boolean concept under CIEL:162569 members. It would actually solve our current situation for the underlying cause of death. Thanks!