Old Topics‎ > ‎


Mortality Forum Dec1996-March1997

17-XII-1996 Q1 Gangrene as underlying cause

(Rule A /UC Chapter XVIII)

[I have received a reply from Patricia Wood, Canada, and I have updated the summary with her remarks. The new text is marked with "*" in the left margin, I hope I do not violate any other newsgroup convention I do not know about! /Lars Age Johansson]

The certificate was:

1 (a) Septicemia
1 (b) Amputation of foot
1 (c) Gangrene

                 Underlying cause:

Brazil          Y83.5

*Canada     Y83.5

France        Y83.5

Ireland         R02

Sweden       Y83.5

USA           Y83.5

[So, there is fairly good agreement between the countries. Some comments from the countries on how they arrived at the underlying cause:]

Brazil /Augusto Hasiak Santo/: The identified underlying cause (UC) was Y83.5, since gangrene is ill-defined and the code for post-surgical septicemia is not a valid underlying cause.

*Canada /Patricia Wood/: For this particular certificate I am in agreement that the underlying cause of death would be ICD-10 code Y83.5. The tentative underlying cause, ICD-10 code R02, selected by the General Principle, would be reselected by an application of Rule A.

I think that this case is a very good one for querying the certifier for information about the cause of the gangrene. Unfortunately, in Canada, we find it difficult to institute any sort of consistent querying policy, particularly since our vital statistics system is decentralized.

France /Gerard Pavillon/: Gangrene is discarded as ill-defined condition. A query is sent to the physician asking for the cause of the gangrene. If the physician does not answer, amputation of foot (Y83.5) is coded as the underlying cause.

USA /Julia E Raynor /: Select gangrene, R02 by General Principle. When surgery is performed for an ill-defined condition (Chapter XVIII), reselect using modification Rule A. Code uc to E code for surgery since the complication, septicemia T814 is indexed to Chapter XIX. Refer to Volume 2, page 88, 4.2.12 External causes.

Sweden /Lars Age Johansson/: We would have coded this certificate to 443.9 (peripheral vascular disease NOS) in ICD-9, which, of course, was not according to the rules. The reason was that when we asked the certifiers for the cause of the gangrene, the answer was very often "peripheral vascular disease". In ICD-10 we will do as France and try to find out the cause of the gangrene, and otherwise select Y83.5.


17-XII-1996 Q2 – Q3

17-XII-1996 Q2 Sudden death as underlying cause (Rule A /UC Chapter XVIII)

17-XII-1996 Q3 Cardiac arrest as underlying cause (Rule A /UC Chapter IX)

[A new comment from Sweden:]

[The certificates:]


1 (a) Sudden cardiac death
1 (b) -
1 (c) -
2 Renal failure, diabetes, inguinal hernia



1 (a) Cardiac arrest
1 (b) -
1 (c) -
2 Renal failure, diabetes, inguinal hernia

[Suggested coding:]

                           #2.                  #3.
Australia                                      I461 (E142?)

Brazil                  N19                  N19

Canada               I461                  I469

England              I461                   I461 [Lin, I suppose you mean I46.9?]

Ireland               I461                   E102 (1st choice) or I469

France               I461 (N19)         I469 (N19)

Sweden              N19                  N19

USA                    I461                I469


*Sweden /Lars Age Johansson/: Of course, "renal failure" is not a very satisfying underlying cause. The reason why we in Sweden do not count it as a mode of dying or ill-defined condition is that our medical queries have shown that a renal failure NOS is often a longstanding condition, and quite often due to some disease that has not been reported, like chronic nephritis. We would not assume that a renal failure reported with diabetes is always due to the diabetes, since our queries have shown that the renal disease quite often predates the diabetes, especially in older people. I do not know if this would be true for other countries as well. Sweden has an old population, and perhaps we have more type II diabetics than other countries.

This is one point where the Swedish modification of the ACME decision tables differs from the original NCHS tables. We have removed many of the links with diabetes, since so many old people in our country have indeed developed diabetes, but in a comparatively mild form.

Anyway, many other countries handle the cardiac arrests differently than we do, and perhaps we should follow Cleo Rooney's suggestion and query these certificates. However, before we change our coding, we would like to make a study (100 certificates with cardiac arrest reported as the underlying cause of death, collect the medical records, check how often 'our' underlying cause is correct). I'm afraid we'll have to wait until ICD-10 runs more smoothly.

Australia /John Donovan/: We have been following this correspondence with great interest. If a certificate shows

I (a) Cardiac arrest
I (b)
I (c)
II Renal failure, diabetes, inguinal hernia

and we are asked to apply the rules, we have to remember that the rules are really there only to help us when a certificate is unsatisfactory. We might recall that any document is usually taken at face value, and in this case the certificate is part of a document which tells us that cardiac arrest directly led to the death, and that renal failure, diabetes, and inguinal hernia all contributed to the death but were not directly related to the cardiac arrest. As a doctor, I cannot make sense of this combination of statements, and the best approach is to query, as already suggested by the ONS, whose wording of a query for this certificate is good. But then what do you do if there is no reply, which happens in 10% of our queries? I can envisage cardiac arrest being due to renal failure but that is no less a mode of dying. So if forced to choose from Part II without regard to any rules I would assume the diabetes had caused the renal failure. If following the rules I would choose the cardiac failure. Perhaps we need to agree on a list of 'modes of dying' for the purposes of the rules.

Brazil /Augusto Hasiak Santo/: For us, sudden cardiac death and cardiac arrest were considered as ill-defined and latter the selected underlying cause was the renal failure (N19). No linkage was done with diabetes.

Canada /Patricia Wood/: For the first of these two death certificates I would select, as underlying cause, ICD-10 code I46.1 and for the second ICD-10 code I46.9. There are no direct consequence relationships nor linkage relationships with any of the conditions recorded in Part II.

I agree that sudden cardiac death and cardiac arrest are ill-defined conditions, or perhaps modes of dying rather than causes of death. I think that it would be useful to have an international consensus on a list of conditions outside Chapter XVIII that would be considered to be ill-defined or conditions that would be considered to be modes of dying.

England /Lin Shane/:

2. 1a Sudden cardiac death I46.1
     2 Renal failure, diabetes, inguinal hernia N19, E14.9, K40.9

Underlying cause=I46.1

3. 1a Cardiac arrest I46.1
     2 Renal failure, diabetes, inguinal hernia N19, E14.9, K40.9

Underlying cause=I46.1

Comments from Cleo:
We all hated having to code to the cardiac arrest, but could not think of any genuine ICD-10 rule which allowed us to use linkage, vague or unspecified condition, or a direct sequel to select any of the conditions in part 2. We could recommend that all such certificates are referred back to the certifier with the specific question 'was the terminal event given in part 1 of this certificate likely to have been due to any of the diseases you have mentioned in part 2? If so, which do you think was the cause of this death?

What does Andre think?

Sweden /Lars Age Johansson/: I agree that the definition of "ill-defined condition" in the text of Rule A mentions only conditions coded to Chapter XVIII. But according to the definition of causes of death (Vol. 2, p 30), "modes of dying" are not to be reported as causes of death. The examples given are heart failure (I50.9, Chapter IX) and respiratory failure (J96.9, Chapter X), so apparently there is no restriction to Chapter XVIII there. On the other hand, the ICD gives no instructions on what to do if a "mode of dying" is reported as the underlying cause. Would it be unreasonable to assume that something that should not have been reported in the first place should also not be accepted as the underlying cause of death, and treat the "modes of dying" in the same way as the ill-defined causes?

This discussion has developed in a direction that was a bit surprising to us at Statistics Sweden. We believed that the case with the cardiac arrest was crystal clear, and that all countries would select the renal failure. What we hoped to find out with this question was if other countries would handle the cardiac death in the same way as the cardiac arrest. Well, you did, but not quite as we had expected.

France /Gerard Pavillon/:

#2.: Underlying cause : Sudden cardiac death (I46.1).

In fact, in France, most of the coders would select Renal failure as the underlying cause, considering Sudden cardiac death as an ill-defined cause. It is more interesting to select Renal failure, but this would need to set precisely the conditions other than chapter XVIII that could be considered as ill-defined.

#3.: Underlying cause : Cardiac arrest (I46.9). Same remark as for case 2.

USA /Julia E Raynor/:
example 2

uc 461

Select sudden cardiac death, I461 by General Principle. There is no linkage with any condition in Pt 2 in Volumes 1, 2, or 3. We would not code I461 as a direct sequel of either condition in Pt 2.

example 3

uc I469

Select cardiac arrest, I469 by General Principle. Explanation is the same as example 2.

Sweden /Lars Age Johansson/: We would think along the Brazilian lines and regard both the sudden cardiac death and the cardiac arrest as ill-defined or modes of dying, and not accept them as underlying cause.

[Of course, how you decide to handle cases like these might influence your statistics (and the comparability). Any comments from the other countries? Would it be possible to agree on a list of ill-defined conditions?
Lars Age]

23-I-1997 Q1 – Q2

23-I-1997 Q1 Linkage acute bronchiolitis - chronic obstructive pulmonary disease  (Rule C / UC Chapter X)

23-I-1997 Q2 Linkage pneumonia - chronic obstructive pulmonary disease (Rule C / UC Chapter X)

[New comments from Sweden and USA. Apparently we have a fairly good agreement on the first question, and we would all select something in J44 as the underlying cause, but different fourth characters. For the second question, it would be easier if we had some certificates to discuss.]

Questions from France, dated 23.1.1997:

1. In volume 2, page 58, J20 (acute bronchitis) is linked with J44 (other chronic pulmonary disease), when in Volume 1 page 525, the whole block J20-J22 excludes J44.1 and J44.0. Should not J21 (acute bronchiolitis) and J22 (unspecified acute lower respiratory infection) follow the same behaviour than J20?

2. J44.1 (Chronic obstructive pulmonary disease with acute exacerbation) is excluded from J00-J06 and J20-J22. There is no indication for J12-J18 when combined with J44.1. In these cases however, it would seem sensible to apply the same rule and to prefer the chronic disease to the acute episode.


*Sweden /Lars Age Johansson/: The first one: Is it correct to say that acute bronchiolitits (J21) is an acute exacerbation of chronic obstructive pulmonary disease? If so, the first exclusion note on Vol 1:525 would apply to that category as well, and the resulting underlying cause would be J44.1. The second exclusion on the same page ("lower respiratory infection") should cover J22 ("Unspecified acute lower respiratory infection"), and we would have J44.0 as the underlying cause of death. I do not understand why just J20 is mentioned in Vol 2, and not J21-J2 as well.

The second one: In this case we would apply Rule 3 and consider the pneumonia a direct consequence of the chronic condition. This is Swedish coding practice and not an ICD instruction, but at the beginning of ICD-9 we made a decision to consider all conditions of a "complication" nature in a specific organ a direct sequel of any serious disease of that same organ. That principle would apply to this case, but I have a feeling that some of you may have comments on this "complication principle".

*USA /Julia E Raynor/: Decisions for this group of questions are:


We interpret the " EXCLUDES " note on page 525, Volume 1 to mean only that
        1. chronic obstructive pulmonary disease (J449) links with mention of acute exacerbation (J441) for a code of J441 and
        2. that chronic obstructive pulmonary disease (J449) links with mention of acute lower respiratory infection (J22) for a code of J440.

The U.S. follows the recommended linkages in ICD-10 Volumes and does not add any additional ones. We will DS J21 (acute bronchiolitis) to J449 (chronic obstructive pulmonary disease) and code J449. We will DSC (direct sequel with combination code) J22 to J449 with combination code of J440.

England /Lin Shane/:
1. We feel that coders would not have a problem with this. Using Volume 3 and then Volume 1, we would arrive at J44 and accept the definitions as expressed in the ICD-10 . We regard the exclusion note on page 525 of Volume 1 as a general reference to the block and not just to J20. (See page 19 Volume 2).

2. Similar to #1, we do not think that coders would have a problem. As Gerard infers, it would be easier if we had example records.

Comments from Cleo:

I think both the linkage and the exclusion are trying to achieve the same thing, viz. if both acute and chronic bronchitis are mentioned, or chronic bronchitis with any acute respiratory condition in J20-J22 are mentioned, then code to J44.-, i.e. that the pre-existing chronic condition is emphasised.

In #2, the codes which do have the exclusion are mostly trivial conditions or unlikely to be fatal on their own. This is not true of pneumonia, even though it too may be a complication of chronic bronchitis. I do agree it would be clearer with examples.


23-I-1997 Q3 Linkage HIV - complications in Part II

(Rule C / UC Chapter I)

[A new comment from the US. So far everyone has been of the same opinion as to the underlying cause, but the principles for multiple cause coding differ.]

1(a) Pneumonia (pneumocystis carinii)
1(b) AIDS
2 Kaposi's sarcoma


                                UC               MC

Brazil                      B22.7

Canada                   B22.7

England                  B22.7             B206/B24*B210

France                    B22.7

*USA                    B22.7             B59/B54*C469


*USA /Julia E Raynor/:

Underlying cause is B227. We will interpret B227 to include "with mention of " complications of AIDS that are classified to two or more 3 digit categories in B200-B238. We will also use , as Augusto indicated, infectious conditions outside Chapter 1 such as pneumonias and meningitis.

Brazil /Augusto Hasiak Santo/: The final underlying cause is B22.7, after linkages of AIDS with both Pneumocystis carinii pneumonia and Kaposi's sarcoma.

Regarding AIDS coding, Celso Escobar Pinheiro and I have been preparing decision tables based on real Brazilian death certificates and on what can be assumed theoretically from ICD. Firstly, considering that physicians include very often more than four affections in these death certificates, linkages of AIDS and these affections are considered as "with mention of" and not as AIDS considered as their "originating antecedent cause" (due to linkage). Besides this, we have decided to consider infectious affections outside Chapter I also as infectious ones; i.e., pneumonias, meningitis, etc.

[Augusto and Celso, would you mind sharing your AIDS decision tables with us?]

Canada /Patricia Wood/: For this particular certificate I selected as underlying cause of death ICD-10 code B22.7. I based my decision on the note for use in underlying cause mortality coding, ICD-10, Volume II, page 51, B22.7. Rather than ignoring the Kaposi's sarcoma reported in Part II though I considered it to be included in B22.7 as one of the "multiple diseases classified elsewhere" resulting from AIDS.

In the index Kaposi's sarcoma specified as resulting from HIV disease is coded to B21.0. Does that imply that on a certificate like this one we can not code the Kaposi's sarcoma to B21.0 because it is recorded in Part II and not specified as "due to" the AIDS? I would be inclined to assume that Kaposi's mentioned anywhere on a certificate with AIDS or HIV disease would be coded to B21.0.

England /Lin Shane/:

3. 1a Pneumonia (pneumocystis carinii) B20.6
      b AIDS B24
    2 Kaposis Sarcoma B21.0

Underlying cause=B22.7

In this instance (when coding manually) we would deviate from our general practise for multicause coding in order to be able to reconcile the underlying cause.

France /Gerard Pavillon/: About the Swedish death certificate mentioning AIDS, Jean would code B22.7 on the basis of the example 25 (Vol II, p40). The fact that Kaposi is mentioned in part II is ignored.

Sweden /Lars Age Johansson/: I am certainly no HIV specialist, but I believe that some complications to HIV are so typical of the disease that it would be safe to code them as complications regardless of where on the certificate they have been mentioned. Again, an international list would be useful, and perhaps we could have a discussion of the tables Augusto and Celso have prepared.


3-II-1997 Q1 Interpretation of the new wording?

(Rule B)

[A new comment from France:]

[The certificate - change in "trivial" rule:]

1a) renal failure
  b) septicemia
  c) dermatitis

[Suggested coding:]

                      UC              MC

Australia        Rule B not applicable

Brazil             L309            N19/A419/L309

England         A419           N19/A419/L309

*France         L309            N19/A419/L309

Sweden         A419            N19/A419/L309

USA             L309            N19/A419/L309


*France /Gerard Pavillon/: We do not think that dermatitis is a trivial condition and we accept the sequence mentioned by the certifier.

England /Lin Shane/: We agree with Patricia. In ICD-9 the underlying cause would be assigned to Dermatitis. Following the rule in ICD-10 volume 2, we would have to select the septicemia as the underlying cause. Although we can appreciate the reasoning, we are concerned to read Julia's reply since we shall be using MICAR and ACME. Should we assume that the software will be programmed to reflect the ICD-9 rather than the ICD-10 rule?

USA /Julia E Raynor/: I have already commented on the trivial rule but would like to offer an explanation for NCHS coding of John's Case 3.

  a) left lower lobe pneumonia 486
Part II left leg ulcer, dermatitis, urinary tract infection 7071 6929 5990

UC 486 We assign pneumonia, lobe to 486 based on instruction from Official WHO Amendments and Corrections to IC D-9, Volume 2. Instructions are---
page 408 volume 2- add parenthetical term Pneumonia...(lobe)
page 410 volume 2 - delete subterm

-lobe- see Pneumonia, lobar

This list in included in Instruction Manual, Part 2e.

Australia /John Donovan/: We have also been following this correspondence with great interest. Skin conditions are not necessarily trivial in the elderly! Here are three certificates from our trial of 13000 certificates coded both manually using our conventions and automatically by MICAR and ACME. The certifiers were not sent queries concerning any of these certificates. All were written on forms which had a blank line, not labelled I (d), below line I (c).

Case 1 male 89

I (a) Septicaemia
I (b) Infected ulcer left foot
I (c)
II Alzheimer's dementia

This was coded manually to ICD-9 707.0 and automatically to 707.1. The latter is in accordance with ICD-9 rules as the ulcer is not specified as decubitus ulcer.

Case 2 male 86
I (a) Septicaemia
I (b) Foot ulceration (chronic)
I (c) Renal failure and debility
Alzheimer's dementia
II Ulcerating oesophagitis, poor appetite

This was coded manually to ICD-9 707.1 and automatically to 290.1. The former is in accordance with ICD-9 rules, Rule 1.

Case 3 male 94

I (a) Left lower lobe pneumonia
I (b)
I (c)
II Left leg ulcer, dermatitis, urinary tract infection

This was coded manually to ICD-9 707.0 following an Australian rule used until 1996 that over age 75 when pneumonia is all that is in Part I and there are non-trivial conditions in Part II, the first non-trivial condition from Part II is selected as underlying cause; it was coded automatically to 486. I would not have used either of these codes: 481 is correct, in my view. There is no indication that the ulcer was a decubitus ulcer, lobar pneumonia is not a likely direct consequence of a leg ulcer, and the automatic coding does not recognise pneumonia of a named lobe as lobar (this has been pointed out to NCHS).

From 1997 we are using the rules of the automatic coding systems MICAR ACME etc unmodified.

Sweden /Lars Age Johansson/: I must confess that I have never quite understood this rule. But the comments to both Example 40 and 41 in ICD-9 -

"Dental caries, selected by the General rule, is ignored since this condition was not reported as the cause of a more serious complication" ...

"Dermatitis selected by rule 2, is ignored since it was not reported as the cause of a more serious complication"

- seem to indicate that these two trivial conditions should have been selected IF they had caused a serious complication. In ICD-10, these examples have been removed, and my conclusion is that a trivial condition should not be coded as UC, even if it is the cause of a serious complication. I have read the phrase "a trivial condition unlikely to cause death" as meaning "unlikely to cause death on its own", and so reselected the underlying cause. But, as Augusto says, it seems that the rule does not really say anything about this case. Does anyone know something about the intentions behind the change?

I agree with John that it is far from clear what conditions we should regard as trivial, and that some distinction must be made between young and old people. That was also the main reason why we in Sweden removed most of the trivial causes from the ACME decision tables when we modified them - we felt that you had to know the age of the deceased before you could decide whether to use Rule 6 or not.

USA /Julia E Raynor/ Concerning the trivial rule--We have been aware that the wording for the trivial rule is different between the 9th and 10th revision. Our decision is to treat the trivial rule for the 10th the same as for the 9th. This means that if the trivial condition caused another condition, we will not consider it trivial and reselect. We consider the dermatitis "likely " caused the death.

Brazil /Augusto Hasiak Santo/: Regarding the question submitted by Patricia, our comments are the following ones:

The assumption is that the phrase "unlikely to cause death" has the same value as the former "part b" of Modification Rule 6 of ICD-9. In this death certificate, "dermatitis" "likely" caused the death.

The Report of the Working Party on the Review of Selection and Modification Rules for Underlying Cause of Death, Budapest, 19-23 April 1983, regarding former Rule 6, says: "The Group agreed to delete rule 6 (b); it was rarely applied and there seemed no justification for preferring a completely unrelated condition. There was considerable discussion as to whether rule 6(a) should be extended to all conditions but it was decided to keep the restriction to trivial conditions. Its application was widened to include untoward complications as well as adverse reactions. The Group did not accede to requests for a list of trivial conditions, since it would be an exceedingly long list, but did define the term more precisely. Rule 6 would now read: "Where the selected underlying cause was a trivial condition, i.e., one which would not lead to death on its own, and death was the result of an adverse reaction to or untoward complication of its treatment, select the adverse reaction or untoward complication." If ICD-10 provided a mechanism to identify both the condition treated and the adverse reaction or untoward complication, rule 6 might be dropped."

Again, the Report of the Consultation on Rules and Definitions in Death Certification in Relation to ICD-10, Titchfield, 6-10 April 1987, regarding Rule 6, reads:

"This rule relating to "trivial conditions" was considered to be inappropriate as it made an unreasonable distinction between the therapeutic misadventures related to trivial conditions and those associated with serious conditions (see Rule 12). It was agreed that further definition of trivial conditions was required on a "such as" basis and that they should not be accepted as the underlying cause of death except under exceptional circunstances. The part of ICD-9 rule preferring a more serious unrelated condition was recommended for deletion so that this rule would read:

"Where the selected underlying cause is a trivial condition unlikely to cause death and the death was the result of an adverse reaction to treatment of the trivial condition, select the adverse reaction.""

It seems that the possibility of a trivial condition giving rise to a more serious condition was not dealt very clearly in both meetings. Even on page 45 of the Manual of Instructions for ICD-10 there is not an example of a this pathological sequence.


 5-II-1997 Q1 Coding of surgery without stated reason for the surgery

[We have an answer from France that somewhat deviates from the beautiful consensus we have had so far... May I ask Gerard to develop their reasoning a bit: in cases with recent surgery and a subsequent complication, would you always code the surgery itself as the underlying cause of death, and not to the condition that was the cause of the operation?]

[The certificate:]

1a) Toxemia
1b) Gangrene of right inferior limb
1c) Infection of arterial prosthesis

[Suggested coding:]

                  UC            MC

Brazil           I779        T828/T828/Y831 T827

Canada        I779        R688/R02/T827 Y831

England        I779         R688/R02/Y831 T827

*France      Y831         R688/T828/T827 Y831 (if recent prothesis)

                    I779         R688/T828/T827/I779 (if no answer)

Sweden         I779        R688/R02/I779 Y831 T827

USA              I779       R688/R02/T827 Y831 I779


*France /Gerard Pavillon/: Query to the certifier to know if the prosthesis is recent or not.

England /Lin Shane/: Similar to Lars and Julia, we would assign the underlying cause to the disease of the site indicated by the surgery, in this case Unspecified disorder of the arteries.

Canada /Patricia Wood/: I would apply the following ICD-10 codes . . . R688/R02/T827 Y831

Then, accepting a causal relationship between the components of the sequence of events leading to death I would select, as the tentative underlying cause of death, Y83.1. Next, considering the note in Volume 2, page 71, 4.2.6, I would reselect, as the underlying cause of death, "the code for the residual category for the organ or site indicated by the name of the operation," I77.9.

Sweden /Lars Age Johansson/: We would, like Augusto, apply the instruction on p. 71 here, and code "Unspecified disorder of artery" as "the residual category for the organ or site indicated by the name of the operation". Our multiple cause coding is different, however: like England, we do not modify the multiple codes unless necessary to preserve information that otherwise would have been lost. Also, we would insert the code for the cause of the operation.

USA /Julia E Raynor/: We would code toxemia and gangrene as indexed and consider infection the complication of the arterial prosthesis. Since the reason for the arterial prosthesis is not reported, code disease, site of surgery (artery), I779 for uc. Volume 2,page 71,4.2.6.'


14-II-1997 Q1 Malnutrition an obvious consequence of brain disease?

(Rule 3 /UC Chapter IV,VI)

[Coding suggestions from France. A comment from Sweden.]

[The certificate - the application of Rule 3:]

I(a) Cardiac failure
I(b) Anaemia
I(c) Malnutrition
II Machado-Joseph disease (hereditary spino-cerebellar degeneration)

[Suggested coding:]

                   UC         MC

Canada        E46       I509/D649/E46*G318

England        D539     I509/D649/E46*G119

*France       G111     I509/D649/D539*G111

Sweden       E46        I509/D649/E46*G318

USA            E46        I509/D649/E46*G119


*France /Gerard Pavillon/: This is a young man and Malnutrition is quite certainly due to another disease. Machado-Joseph disease is considered as the origin of the malnutrition (Rule 3).

*Sweden /Lars Age Johansson/: It would be fine with some more discussion on when to use Rule 3. I agree with Julia that it is easy to overuse it. If we "correct" a wide variety of certificates through using Rule 3, isn't there a substantial risk that the statistics reflect what we think the certificates should look like, rather than what the certifiers have actually reported? Of course, many certifiers do not understand the difference between Part I and Part II, but Rule 3 should be restricted to obvious cases. To be quite frank, I'm not very happy with some of the instructions on Rule 3 in ICD-10. For instance, we are told to regard bronchopneumonia as a direct consequence of a cerebral thrombosis reported in Part II, with paralysis as the presumed connecting link. One of our medical consultants went fuming when he saw that, since most cerebral thromboses do not result in paralysis. Moreover, we are told (on page 66) not to let an assumed cause influence the coding, except when necessary to accept a sequence in Part I... But since the instruction is there, we will have to obey it.

England /Lin Shane/: We would not assume the anemia and malnutrition to be a direct result of the disease in part 2, and would therefore link the malnutrition with anemia and produce D53.9 as the underlying cause code.

Cleo Rooney's comment: I would like to use Rule 3 to say that malnutrition is due to the disabling/wasting disease, but I am not sure I can. What do other countries think? I do not think you can call it trivial.

USA /Julia E Raynor/: This condition is not indexed in 10th revision as reported. We think G119 is the best code since it is specified as hereditary. Disease, spinocerebellar (hereditary) G119 By 9th revision causal relationship tables, we would select malnutrition by General Principle. In Table D, modification rules, address code 2639 is SDC with 2859 for UC of 2819. We would not consider malnutrition a direct consequence of spino-cerebellar degeneration in Pt II. UC 2819 But, note in ICD-9 Modification Table D that address code, 2639 is a direct sequel of some brain degenerations. Certainly, it is sensible to include 3348 as well.

I agree with John we should try to use the same interpretations for this rule, if possible. For me, personally, it is the most difficult rule to apply. On one hand, there are conditions that certainly seem to be a direct consequence of other conditions. When a decision is made to consider one condition a direct consequence of another, the next decision must be how many other related conditions to also include. I have mixed feelings about these decisions. Many times we are leaving Part 1 to prefer a condition in Part II when the certifier could have reported the condition in a "due to" position in Part 1, but chose not to. Are we trying to "out think" the certifier? In the U.S., these decisions are made by NCHS senior nosologists when the ACME tables are prepared. NCHS and U.S. state coders follow the decisions documented by the ACME tables, therefore, insuring consistency in our data.

I agree with Lars Age that instructions in Vol. 2, pages 39 and 40 are not all inclusive. He has listed some interesting criteria for decision making. Concerning the instruction on using anemia, (D649), malnutrition, (E46), marasmus, (E41), or cachexia, (R64) as a direct consequence of other conditions. We would use as direct consequence even though cachexia, R64 is indexed to ill-defined. We use the rules in order and Rule 3, direct consequences precedes Modification Rule A, ill-defined.

Canada /Patricia Wood/: I would apply the following ICD-10 codes . . .


Then, accepting a causal relationship between the components of the sequence of events leading to death I would select, as the tentative underlying cause of death, E46. Next, in the absence of a related rule, note or guideline in ICD-10, I would not reselect another condition so I would select, as the underlying cause of death, E46.

While recognizing Dr. Donovan's medical knowledge and expertise and appreciating his statement that the obvious cause of the malnutrition was the disabling disease in Part II, I wonder if a coder, working without such knowledge or expertise, would consider the possibility of a direct consequence on this record.

Finally, I believe that cachexia (R64) would be considered to be an ill-defined condition and subjected to Rule A if selected as the underlying cause of death on a certificate.

Sweden /Lars Age Johansson/: The instructions on Rule 3 (Vol 2, pp39-40) apparently do not cover all situations where the rule will apply - for example, the relationship in Example 31 (p 41) is not mentioned on pp 39-40. So, I suppose that we cannot accuse John of stretching Rule 3 too far if he decides to use it in this case. In Sweden, we have no instruction on Rule 3 and malnutrition, and we would code this case to E46, if it is not possible to get any further information from the certifier.

In Sweden, we will apply Rule 3 in the cases mentioned in the manual, and also if the certificate fulfils either of these two requirements ():

1) "Unavoidable consequence" - the condition on the last used line i Part 1 is a more or less unavoidable consequence of the (presumed) underlying cause in Part 2. We use this criterion mainly for acute or rapidly progressing conditions, such as peritonitis in Part 1 and intestinal gangrene in Part 2, or biliary obstruction in Part 1 and a malignant neoplasm of the head of pancreas Part 2. (How "unavoidable" these consequences are, and the strength of the relationship, might of course be argued.)

2) "Sudden turn" - if the presumed underlying cause in Part 2 is a chronic, potentially seriuous condition, and there is a condition that might be a complication reported on the lowest used line in Part 1, we will assume that the condition in Part 1 is a direct consequence of the chronic disease in Part 2 if, and only if, there is evidence of an acute exacerbation of that disease.

For example: aspiration pneumonia in Part 1 and multiple sclerosis in Part 2, AND further information on the MS like "recent deterioration". Such "further information" is not often found on the certificates, but we use this criterion when we evaluate medical records (we ask for the medical records if we suspect that the true underlying cause has been reported in Part 2).

I think John is certainly right: there are many different interpretations and applications of Rule 3, and for some conditions (e.g. diabetes) differences are so wide as to make the statistics impossible to compare. Clearly a more complete list of situations where Rule 3 applies would be of great use.


14-II-1997 Q2  May an anomaly be due to something else in Chapter XVII?

(Rule D 4.2.2 /UC Chapter XVII)

A reply from France:]

[The certificates - the note on p 68 says that a congenital anomaly may not be due to any other disease:]

1 (a) Multiple malformations
1 (b) Trisomy 18
1 (a) Hypoplasia of lungs
1 (b) Thorax malformation

[Suggested coding of cert #1:]

                    UC           MC

Canada         Q913       Q897/Q913

England         Q913       Q897/Q913

*France        Q913        Q897/Q913

USA             Q913        Q897/Q913

[Suggested coding of cert #2:]

USA             Q336        Q336/Q687

[Comments: ]

*France /Gerard Pavillon/: We accept the sequence even if it is considered as improbable by ICD10.

England /Lin Shane/: In this instance we feel sure that the multiple malformations were the manifestation of the Trisomy 18. We had little hesitation in selecting Trisomy 18 as the underlying cause. We are curious to know what an automated system would have made of this.

Cleo's comment: Must have an exception for chromosomal/genetic basis of multiple malformations? Suggest this is one to refer to the next WHO Heads of Centres meeting.

USA /Julia E Raynor/:

[#1:] We would select multiple malformations, Q897, by Rule 2. Next, by application of Rule D, specificity, code UC to trisomy 18, Q913.

[#2:] We would not accept the sequence. UC Q336, hypoplasia, lung We interpret note (k) on page 68, Vol. 2 as meaning that congenital anomalies are not due to other diseases and that one congenital anomaly is not due to another congenital anomaly.

Canada /Patricia Wood/: Considering the intention of the certifier (and pretty much ignoring note (k) on page 68 of Volume 2) and accepting a causal relationship between the components of the sequence of events leading to death I would select, as the tentative underlying cause of death, Q91.3.

In the absence of any related rule, note or guideline in ICD-10, I would not reselect another condition so I would select, as the underlying cause of death, Q91.3.

I am sure that it is "highly improbable" that congenital anomalies are due to any other disease but on this particular certificate I believe that the certifier is merely stating that the decedent had multiple malformations as part of the manifestation of the Trisomy 18. I realize that this contradicts note (k) but I really do not see this particular sequence as "highly improbable."

I think that we should be cautious in assuming that the instruction in note (k) applies only to congenital malformations reported as due to conditions outside Chapter XVII because it is possible that a certifier, misunderstanding the format of the certificate, could report one congenital anomaly as due to another where the causal relationship is truly "highly improbable."


17-III-1997 Q1 Cerebral palsy an obvious consequence of brain damage?

(Rule 3, C / UC Chapter VI, XVI)

[An American/Swedish suggestion for some modifications to Julia Raynor's first DS list for cerebral palsy. To those not quite familiar with ACME terminology: "DS" stands for "direct sequel" and refers to Selection Rule 3. So, if you have selected cerebral palsy as the "originating antecedent cause", and a condition on the DS list is also mentioned on the certificate, then you will select this condition as the underlying cause of death.]

[The question - a long one - was:]

In ICD-9, there was an instruction in the Notes (Vol 1, page 720) on perinatal conditions and cerebral palsy, saying that 760- 779 excluded residual cerebral paralysis at ages 4 weeks or over (these were coded to 333.7 or 343).

We have not been able to find any similar instruction in ICD-10, except the entry in Volume 3 for

- cerebral (congenital) (infantile) G80.9

Does the change in the Notes mean that the following certificate should be coded to P10.1 rather than to G80.9?

(1) 1 (a) Cerebral palsy
      1 (b) Cerebral haemorrhage (birth injury)

Moreover, does it mean that we should use Rule 3 in a case like the one below, and code that to P52.4?

(2) 1 (a) Residual cerebral palsy (from birth)
      2 Perinatal cerebral haemorrhage

[Suggested coding:]

(1)                   Muiltiple causes       Underlying cause

*Canada          G809/P101              P101

England            G809/P101              P101

USA                G809/P101              P101


*Canada          G809*P524            P524

England           G809*P524             P524

USA                G809*P524            P524

*Canada /Patricia Wood/: I have reviewed the codes in ICD-10 and have taken note of the absence of an instruction in ICD-10 saying, as in ICD-9, that perinatal conditions exclude residual cerebral paralysis at ages 4 weeks or over. I would code each of the two records you present as you suggested and Julia supported . . .

(1) multiple cause (ICD-10) codes: G809/P101 underlying cause (ICD-10)

code: P101

(2) multiple cause (ICD-10) codes: G809*P524 underlying cause (ICD-10)

code: P524

I have also reviewed the proposed list of DS relationships for G800-G809 (Infantile cerebral palsy) and I think that they would be very useful.

Sweden /Lars Age Johansson/ and the US /Julia Raynor/:

Would this list be correct?

G800-G809 [Infantile cerebral palsy]

DS [is a direct consequence of]

G000-G009 MAYBE [Bacterial meningitis]

G030-G039 MAYBE [Meningitis, other and unspecified]

P100-P109 [... due to birth trauma]

P110-P112 [... due to birth trauma]

P119 [... due to birth trauma]

P200-P210 [Intrauterine hypoxia]

P219-P220 [Birth asphyxia]

P228-P229 [Resp distress, newborn]

P520-P529 [Intracranial ... haemorrhage ...]

P910 [Neonatal cerebral ischaemia]

The MAYBEs: the nosologist would have to make sure that the meningitis occurred before the palsy.

England /Lin Shane/: Subject: 17 III 1997 Question 1

We agree that because there is no corresponding note in ICD-10 to that on page 720 of Vol. 1 ICD-9, we would code the first record to the cerebral hemorrhage. From the ICD-10 index it seems that P10.1 only applies to fetus or newborn, and that there does not appear to be a code for cerebral hemorrhage due to birth injury in older people. However we have considered the inclusion note at the beginning of Chapter XVI 'Includes: conditions that have their origin in the perinatal period even though death or morbidity occurs later.' It therefore follows on that if we logically accept this as a reasonable sequence the second record should also be assigned to the perinatal cerebral hemorrhage.

USA /Julia /:
(1) agree the underlying cause is P101 by general principle. Evidently, the intent in the 10th is to prefer the condition causing the palsy, rather than the palsy. I could not find an instruction indicating otherwise.

(2) I agree the cerebral palsy should be considered a direct consequence of cerebral hemorrhage. After that decision is made, as we have discussed in the past, how many other categories should we include also? Would it be correct to DS G800-G809 ( infantile cerebral palsy) to P520-P529 (intracranial nontraumatic hemorrhage of fetus and newborn)

P100-P109 (intracranial laceration and hemorrhage due to birth injury)

P110-P119 (other birth injuries to central nervous system)

P159 (birth injury, unspecified)

Some of these might need to be "maybe" in our automated tables so an experienced nosologist could review the type of palsy and central nervous system condition reported on the certificate and make a decision about the relationship between the conditions.


23-III-1997 Q1 Coding of drug dependence

(UC Chapter V)

[Here is a question on drug coding in ICD9 from Mary Heanue, Ireland:]

I have a question on ICD9 which I would appreciate your input on. How do you code suicide by drug overdose by a drug dependent person? Do you code it as a 304 with a suicide ecode or simply 304? Also, how do you distinguish in your coding that the drugs were prescribed?

In the manual notes on 960-999, deaths due to dependence producing drugs are excluded if drug dependence is mentioned, whether accidental or purposeful. By purposeful, does it mean murder only or both murder and suicide?

[A new answer from Canada:]

* Canada /Patricia Wood/: In Canada the underlying cause code for "suicide by drug overdose by a drug dependent person" would be E950.- . The drug dependence (304.-) would be coded as one of the multiple causes but would not be selected as the underlying cause of death even if certified as the cause of the drug overdose. Deaths certified as drug poisoning in a drug dependent person are coded to 304.- if the drug poisoning is accidental (E850 - E858) or undetermined whether accidentally or purposely inflicted (E980). This is based on the note on page 721 of Volume I, ICD-9.

I think that we are all in agreement that there is no way in the ICD-9 to code a distinction between drugs that were prescribed to the decedent and ones that were not.

To remark on the discussion regarding accidents due to disease conditions . . . we use the relevant NCHS instructions (found in Manual 2a) as our guideline for what diseases can be accepted as the cause of accidents. I agree with you Lars Age that international consistency would be promoted by including related instructions in the "clarifications of the mortality coding rules and guidelines."

England /Lin Shane/: Underlying cause Multicause

E95- 304-

Deaths with a verdict of suicide will be coded to E95* even if there is also mention of abuse or dependence of the poisoning drug. We would underlying cause code a suicide death by a drug dependent person to the appropriate E code for suicide poisoning. We would also assign a secondary code from Chapter XVII to identify the drug if known, or to 9779 if not specified. ICD 304 with the appropriate 4th digit would be assigned as a multicause code. We are not able to distinguish by coding whether the drug was prescribed. Deaths from accidents, violence or poisoning or directly due to drug abuse or dependence can only be certified by a coroner, following a legal inquest in England and Wales. Manner of death generally equates in E&W to coroners verdict from inquest, i.e. accident, suicide, homicide, open etc. Information on whether the subject abused drugs or was dependent on them most commonly comes from a coroners narrative verdict, such poisoning due to drug abuse, or accidental overdose in a known drug addict. These verdicts are used in assigning the underlying cause of death. For example, a death certified as due to overdose of heroin can be coded to E8500 (accident), E9500 (suicide), E9800 (open verdict i.e. the intent could not be determined by inquest), 3040 (drug dependence or addiction ) or 3055 (drug abuse). If drug abuse or dependence is mentioned together with an open or accident verdict, in a death from drug overdose, the death is coded to 304* or 305*, not to an E code. This is in accordance with ICD 9 mortality coding rules (see page 721 vol I ICD 9, Notes for use in underlying cause mortality coding: E850 E858 and E980 excludes deaths due to dependence producing drugs if drug dependence is mentioned Äcode instead toÅ 304*), and should be the same in all countries using ICD 9. Many of the categories listed here should not therefore exist in mortality data.

USA /Julia E Raynor/: When suicidal or homicidal drug poisoning is reported on same certificate with drug dependence, we would first apply selection rules. The U.S., like Sweden, does not accept suicide or homicide as due to drug abuse.

Therefore, if suicidal or homicidal drug poisoning is reported due to drug dependence or as first condition on same line with drug dependence, we would code underlying cause to poisoning. When accidental drug poisoning (E850-E858) or poisoning undetermined whether accidentally or purposely inflicted (E980) is reported with drug dependence (3049) , we follow instruction in Volume 1, page 721. This instruction states that E850-E858 and E980 excludes poisoning by dependence producing drugs if drug dependence is mentioned (304). Underlying cause would be assigned to 304. We interpret E9800-E989 as deaths that have not been determined whether they are accidental, suicidal or homicidal. In the U.S., there is not a way to distinguish if drugs were prescribed.

The U.S. coding does differ from Sweden's coding with accidents due to diseases. We do not accept accidents due to diseases such as cerebral hemorrhage, heart attack, alcoholism, diabetic coma unless there is evidence death occurred prior to accident. Our exceptions to this rule of not accepting accidents due to diseases are

1. asphyxia from aspiration of mucus or vomitus as a result of a disease.
2. a fall from a pathological fracture or disease of the bone.
3. aspiration of food due to diseases which affect the ability to swallow, such as throat ca.
4. accidents resulting from epilepsy.

The instruction for accidents due to epilepsy is in Vol. 1, page 714. I have not found an instruction in ICD-9 to support our other exceptions. In 10th revision, Volume 2, page 68, accidents due to any other cause outside the chapter except epilepsy is included on improbable list. We will use our 9th revision exceptions for 10th revision also. Julia Raynor

Sweden /Lars Age Johansson/: It is good that ICD-10 has an instruction on how to deal with external causes due to diseases, there were many differences between countries on that in ICD-9. But can anyone explain to me why ICD-10 still has an exception for epilepsy? It is quite difficult to explain to, eg. forensic pathologists, that if a person stands on a ladder, gets an epileptic seizure, falls from the ladder and breaks his neck, then the death will be coded to epilepsy; but if he stands on a ladder , has a stroke, falls and breaks his neck, then the death will be coded to the fall. But this is the fascination of ICD coding, the uninitiated can never guess what the underlying cause of death will be...

I think it is a very good idea to make exceptions for aspirations and pathological fractures. Perhaps we should try to have that included in the "clarifications of the mortality coding rules and guidelines" that will be discussed at the next Centre Head meeting?

Scotland /Ian/:

1. In Scotland we would code suicide by drug overdose to the appropriate E Code as the underlying cause and 304 as the secondary cause.

2. We are unable to distinguish if drugs were prescribed.

3. With deaths through drugs by murder or suicide we would code as with 1 above, that is to the appropriate E Code as the underlying cause and 304 as the secondary cause.

Sweden /Lars Age Johansson/: In the case you describe, I would code suicide (E950.-) as the underlying cause of death, and drug dependence (304.- ) as a contributory cause. This is Swedish coding practice, of course, and not something based on ICD-9 instructions - there is nothing in ICD-9 that prevents an accident, or a suicide, to be "due to" a disease, but in our coding practice we would not accept such a sequence, or a murder due to disease. So, we would code a certificate like this one

1 (a) Suicide
1 (b) Drug abuse

to suicide. (On the other hand, we did accept that accidents could be due to diseases, e.g. a fall due to a myocardial infarction.)

As far as I understand, the codes E980-E989 include both cases which might be either suicides or accidents, as well as cases which might be either murders or accidents.

I do not know of any good way to indicate that a drug was prescribed. Perhaps an extra variable (non-ICD) should be used for that - "Drug poisoning - prescribed drugs? YES/NO"?