Old Topics‎ > ‎


Mortality Forum Apr1997-July1997

14-IV-1997 Q1 Linkage diabetes - renal complication

(Rule C /UC Chapter IV)

An answer from Canada.

I think Augusto Hasiak Santo has raised a very important question here: should we apply only such linkages that are explicitly mentioned in the ICD, or should we also link in medically equivalent cases? The advantage of coding like the Americans is, of course, that you will have fewer arguments about when to link and when not to. On the other hand, it might be difficult to get the users of the statistics to understand why we code "nephropathy due to diabetes" as "diabetes with renal complication", but not "renal failure due to diabetes".

I think this is a general problem - should we apply the instructions in the ICD literally, or should we use them as guidelines, and try to figure out to what similar cases they should also apply? If we decide to use them as guidelines, then there will be some need of international coordination...

Opinions and suggestions are very welcome!]

[Augustos question was: ]

I would like also to propose a discussion on linkages of diabetes and renal involvement, in particular with renal failure, N17, N18 and N19, or, 584, 585 and 586 in the Ninth Revision. The physicians (in Brazil) don't mention diabetes with renal diseases or nephopathy but with renal failure.

The ACME decision tabe D (1992 edition) provides linkages of diabetes (2500 and 2509) with the following codes in order to obtain 250.3: 581.9, 582.6, 582.7, 582.9, 583.9 and 593.9. (582.6 and 582.7 are created codes).

Some examples of death certificates with this problem:

   1. R092/N19/E149
   2. E878/N189/E149
   3. R092/N19*E149

[or, approximately translated into diagnostic statements:

1 (a) Respiratory arrest
1 (b) Renal failure
1 (c) Diabetes mellitus

1 (a) Electrolyte imbalance
1 (b) Chronic renal failure
1 (c) Diabetes mellitus

1 (a) Respiratory arrest
1 (b) Renal failure
2  Diabetes mellitus

[Suggested coding:]

                   UC(#1)            UC(#2)            UC(#3)

*Canada      E149               E149                 N19

England       2503               2503                   ?

Sweden       E142               E142                 N19

USA             E149             E149                  N19

*Canada /Patricia Wood/: (An advantage to falling so far behind with my correspondence with this group is that I have the benefit of reading everyone else's opinion before offering mine!)

As usual I will point out that in Canada we use the NCHS ACME Decision Tables (either manually or in automation) so our approach to the diabetes with renal manifestation situations is as described by Julia. I think that, as you articulate Lars Age, there is a fundamental decision to be made regarding these and all other linkages; do we restrict our applications to those specified in the classification or use them as indications of appropriate coding in similar circumstances? If the latter approach is used internationally consistent decisions would certainly be best.

Australia /John Donovan/: This message of yours and Augusto's question relate to some of the proposals for revision of the renal section of ICD-10 which we sent to Björn [Smedby at the Nordic WHO Centre] in April. If you have not already seen them, please ask him for a copy.

USA /Julia E Raynor/ : The U.S. uses only the linkages provided by ICD. For the 9th revision, they are included in Volume 1, Notes for use in underlying cause mortality coding, pages 713-721, in Volume 1 under the specific categories and in Volume 2. In 10th revision, they are in Volume 2, Notes for use in underlying cause mortality coding, pages 50-65, Volume 1 under the specific categories and in Volume 3. Apparently, the U.S. is more restrictive in using linkages than some other countries.

We do not treat diabetes linkages differently than other conditions with linkage. We will link diabetes, E109, E119, E129, E139 and E149 with renal conditions indexed "diabetic" in Volume 3, indexed "with or due to diabetes" in Volume 3 and those listed in Volume 1, categories E10-E14 with .2 "With renal complications." We have not considered adding any linkages based on other frequently reported renal conditions.

For direct consequence of renal conditions to diabetes, we will use the same renal conditions we used for diabetes linkage.

U.S. codes for examples

Example 1
R092/N19/E149 UC E149

Example 2
E878/N189/E149 UC E149

Example 3
R092/N19*E149 UC N19

This is an example of when we WOULD use linkage with diabetes.

a) nephropathy N289
b) diabetes E149

UC E142 Indexed Diabetes, diabetic, nephropathy E142 Also indexed Nephropathy, diabetic E142

England /Lin Shane/: When manually coding to ICD-9 we would have linked diabetes with any resulting renal connection (except conditions coded to 589 and 590). The ACME decision table D does not agree with the linkages listed in ICD-9 volume 1 for 2503.

Is it true to say that for ICD-10 we should only make the link with diabetes and renal manifestations is the renal manifestation is coded to N083? We will need some guidance on this before we start ICD-10 training.

Sweden /Lars Age Johansson/: We have assumed that "renal manifestations" (as the rubric of 250.3 said in ICD-9) included renal failure as well. Perhaps that was not technically quite correct in ICD-9, since both dagger and asterisks appeared in the rubric and 586 (renal failure) was not one of the listed asterisk categories. In ICD-10, however, the asterisks have been moved to the example terms below the rubric, and I suppose that allows us to include other renal complications as well, eg renal failure. So we would code E14.2 as the underlying cause in the first two examples (we do not use asterisk codes for the underlying cause of death). In the third example, we would not assume that the renal failure is a consequence of diabetes. About ten years ago we returned quite a few certificates of that type (diabetes in Part II, a renal disease that might be a complication to diabetes in Part I) to the certifiers and asked if there was any connection between the diabetes and the renal condition. We found that the diabetes was often fairly mild and/or with a late onset, and that the renal condition in many cases was caused by something that the patient had had before he/she got diabetes, e.g. chronic pyelitis or chronic nephritis. Our conclusion was that it is not safe to assume that a renal condition in Part I is a direct consequence of diabetes in Part II. I do not know if this would apply in other countries as well, the situation in Sweden might be due to the fact that we have a quite old population, and many E11 (non-insulin-dependent diabetes mellitus) cases.

14-IV-1997 Q2 Pulmonary edema an obvious consequence of myocardial infarction?

(Rule C / UC Chapter IX)

[The question, from Sweden:]

1 (a) Cardiac arrest
1 (b) Myocardial infarction
1 (c) Pulmonary oedema

In ICD-9, we would have regarded the pulmonary oedema as ill-defined, disregarded it, and then re-selected the underlying cause of death as if the oedema had not been reported, and thus arrived at myocardial infarction as the underlying cause of death. However, I conclude from the discussions we had earlier in this group on cardiac arrest etc, that most of you would not consider pulmonary oedema an ill-defined cause of death. So what would we get in this case as the underlying cause? Both the infarction and the arrest are reported as consequences of the oedema (and I suppose that is an acceptable sequence, even if another order of events might have been more plausible). That means, as far as I understand, that we cannot use Rule 3 here. Would we end up with the pulmonary oedema as the underlying cause of death?

[An answer from Canada:]

*Canada /Patricia Wood/: multiple causes (ICD-9) codes: 4275 410 514

underlying cause (ICD-9) code: 410

Selected by Rule 1 using ACME Decision Table C where 4275 can be due to 514 but 410 can not be due to 514 therefore eliminating an application of the General Rule, and 4275 can be due to 410 establishing a reported sequence terminating in a condition first entered on the certificate. No further linkages or modifications confirming that the underlying cause code is (ICD-9) 410.

England /Lin Shane/: In ICD-9 we also would have coded this death to the myocardial infarction. This is consistent with the Acme tables. However it is quite different in ICD-10. There is an inconsistency between Volume 3, the index, and Volume 1, the tabular list.

The index reads Edema - Lung - acute - with heart condition or failure, which we would have taken to include myocardial infarction. he exclusion note in the tabular list under J81 reads 'with mention of heart disease NOS or heart failure', which would not include myocardial infarction. The coder may therefore stay with J81 and select the underlying cause as the Pulmonary Oedema.

14-IV-1997 Q3 Which conditions are an obvious cause of pulmonary edema?

(Rule 3 / UC Chapter IX)

[Another question from Sweden:]

Would you consider pulmonary oedema a direct consequence of no other diseases, of some heart diseases, of any heart disease, of any other conditions, or??? In other words: what coding practice do you have as regards pulmonary oedema and Rule 3 (direct consequence)?

[Canada's answer:]

*Canada /Patricia Wood/: Using ACME Decision Table D as a reference pulmonary edema (ICD-9 514) would be considered to be a direct sequel of conditions coded to ICD-9 0010-2899, 2900-319, 3200-4599, 460-4878, 490-5199, 5200-7799 and nature of injury 800-9999. Rule 3 relationship would be applied if pulmonary edema was the tentative underlying cause and if the condition it is a direct sequel of was reported on the same or lower used line of Part I or in Part II.

Using ACME Decision Table D as a reference acute pulmonary edema (ICD-9 5184) would be considered to be a direct sequel of conditions coded to ICD-9 0010-2899, 2900-319, 3200-4281, 4282-4284, 4290-4298, 4300-5131, 515-5199, 5200-7799, and nature of injury 800-9999. Again, Rule 3 would be applied if acute pulmonary edema was the tentative underlying cause and if the condition it is a direct sequel of was reported on the same or lower used line of Part I or in Part II.

Since the ACME Decision Tables are used (either manually or in automation) in Canada these are the direct sequel relationships that we would make with pulmonary edema and acute pulmonary edema.

(One disadvantage of falling so far behind with my correspondence with this group is that I sound so very repetitive in my answers!)

England /Lin Shane/: For ICD-9 coding we would consider conditions in 514, Pulmonary oedema, to be terminal events and apply rule 3. We would not apply the same reasoning to conditions in 5184, Acute pulmonary oedema. The difficulty in ICD-10 is that Pulmonary oedema and Acute pulmonary oedema are both coded to J81.

14-IV-1997 Q4 Underlying cause of surgery without stated reason for the surgery?

[An answer from Patricia Wood:]

[And yet another question from us Swedes:]

On page 71 of Volume 2 there is an instruction on coding of operations without mention of the condition for which the operation was performed. It says that "if the operation does not indicate an organ or a site ... code to ... R99 ... unless there is a mention of a therapeutic misadventure classifiable to Y60-Y84 or a postoperative complication." Where do the postoperative complications come into it? The instructions in the Notes on postoperative complications (eg on I97, page 57) say that codes for postprocedural disorders are not to be used for underlying cause mortality coding.

*Canada /Patricia Wood/: When I try to interpret the instruction on page 71, Volume 2, ICD-10 I complete it like so, ". . . if the operation does not indicate an organ or site code to R99 unless there is a mention of a therapeutic misadventure classified to Y60-Y82, in which case code the underlying cause to the therapeutic misadventure, or there is mention of a postoperative complication, in which case code the underlying cause to the surgery (as the cause of abnormal reaction or later complication, without mention of misadventure at the time of the procedure) Y83."

I would only code the underlying cause to R99 if an operation that does not indicate the organ or site is certified as the cause of death with no therapeutic misadventure and no postoperative complication mentioned.

USA /Julia E Raynor/ :We interpret instruction in Volume 2, page 71, Operations, as meaning if surgery, such as laparotomy, that does not indicate an organ or site is reported without a complication, code UC to R99.

Example a) laparotomy UC R99

If there is mention of a complication indexed to T code, code E code for surgery as UC.

Example a) septicemia T814
              b) laparotomy Y838 UC Y838

If complication of surgery is indexed to postprocedural disorder in disease chapters, code E code for surgery as UC.

Example a) renal failure N990
               b) laparotomy Y838 UC Y838

Volume 2, page 66, Table 2 lists all the postprocedural categories, not just I97, as invalid for UC.

England /Lin Shane/: The note on page 57 of Volume 2 refers to 'postprocedural disorders of circulatory system', not all or any postprocedural condition. There are other postprocedural conditions which it is acceptable to use for underlying cause if no further information is available. Surely the only real solution is to make further enquiries of the certifier.

[My comment:] The same instruction is given in the Notes for E89, G97, H59, H95, I97, J95, K91, M96 and N99. Is the intention that we should code, eg Y83.9/T81.4 if a patient dies in a postsurgical septicaemia, but R99 if s/he dies in a postsurgical renal failure?

14-IV-1997 Q5 Code for "acute cerebrovascular lesion"?

(UC Chapter IX)

[An answer from Patricia Wood:]

[And yet a very Swedish question:]

Acute cerebrovascular lesion

Most conditions coded to 436 in ICD-9 are now coded to I64. There are a few exceptions, notably "acute cerebrovascular disease" (I678) and "cerebrovascular leasion" (I679). What should we do with "acute cerebrovascular lesion"? Should we put the stress on "acute" and code it to I678, or on "lesion" and code it to I679 as well? (We're well aware that this is no major point, but it is quite possible to spend much time debating even small details...)

[Suggested codes for "Acute cerebrovascular lesion":

*Canada   I67.9

England     I67.9

USA         I67.9

*Canada /Patricia Wood/: When coding "acute cerebrovascular lesion" I would use "lesion" as the lead term in the index and I would find the modifier "cerebrovascular" but no subsequent modifier "acute" so I would select ICD-10 code I679.

USA /Julia E Raynor/: If index does not provide a code for the modifier "acute", we usually assign NOS code. Lesion, cerebrovascular, acute I679

England /Lin Shane/: Our feeling is that we would code Acute cerebrovascular lesion to I67.9.

21-IV-1997 Q1 When to use Chapter XVI (P codes)?

(UC Chapter XVI)

[An answer from England:]

[The question, from Sweden, was:]

We are not quite sure when to use the special perinatal codes (Chapter XVI, P00-P96), and when to use the "ordinary" codes from the other chapters. The introduction to Chapter XVI says that it covers "conditions that have their origin in the perinatal period". Does that mean that ALL conditions in the perinatal period should be coded to Chapter XVI, even if the origin is not something directly related to the delivery, but, e.g. an accident or surgery? Should we use a Chapter XVI code even if there is no specific code for the condition?

In other words, how would you code the following certificates (they are both for perinatal deaths):

      1 (a) Mesenteric infarction
      1 (b) Thrombosis of abdominal aorta

      1 (a) Myocardial infarction
      1 (b) Surgery for transposition of great arteries

[Suggested coding of cert #1:]

                           UC                 MC
Canada               I740                K550/I740

*England             I740                K550/I740

USA                   I740                K550/I740

[Suggested coding of cert #2:]

                           UC                 MC
Canada               Q203              T814/Y839 Q203

*England             Q203             T818/Y839/Q203

USA                   Q203              T818/Y839 Q203


*England /Lin Shane/ : We also would only use a 'P' code if the condition is indexed to Chapter XVI.

Canada /Patricia Wood/: Since none of these conditions are indexed to the Chapter XVI Certain conditions originating in the perinatal period I did not use a "P" code for any of them. I will use "P" codes if the condition I am coding is indexed to Chapter XVI or is indexed with a modifier "neonatal" or "perinatal" and there is an indication on the certificate that the condition did arise in the perinatal period (i.e. the decedent is a perinate or the duration of the condition and the age of the decedent indicate that the condition originated in the perinatal period). Our processing system will flag "P" codes assigned as the underlying cause of death for any decedent who is over one year old so the record will be checked but if an indication that the condition arose in the perinatal period is confirmed the underlying cause will not be changed. So, we can have a 90 year old with an underlying cause of death code from Chapter XVI.

Sweden /Lars Age Johansson/: We would accept the P code for any age, if it is evident from the certificate that the condition did indeed arise in the perinatal period. On the other hand, we will check all cases coded to Chapter XVI if the deceased was more than 1 year old, to make sure that the code is correct.

Finland / Martti Virtanen, Stakes, Finland /: This problem has special interest to me as a neonatologist as well as a problem with classification especially with regard to DRG. The question how long the codes in Chapter XIV can be used is relevant also for morbidity coding.

So: How long would you accept the P codes after the neonatal period? Some of the children with severe birth injuries for example may live many years with severe handicap.

The DRG-systems have set the limit to 12 months age. Our old system had a limit of 24 months. We have not changed the rules, so these codes can be used as the actual principal diagnosis only during the two first years of life. Thereafter they are accepted only as secondary diagnoses. For deaths this could mean that basic cause of death could be a perinatal diagnosis, but the immediate cause would need to be something else.

USA /Julia E Raynor/: If age of decedent is less than 28 days, and reported condition is indexed neonatal or newborn, we assign newborn code (Chapter XVI). Otherwise, we use "ordinary" or NOS code.

None of these conditions [on cert # 1-2] are indexed to a newborn category, therefore, NOS codes are used. We interpret Chapter XVI inclusion note to mean codes from Chapter XVI can be assigned if age is 28 days or more if there is indication the condition was present during the perinatal period.

age 31 days
(a) anoxia duration 2 weeks

We would code anoxia, newborn, P219, even though age is over 28 days. Death occurred after newborn period but duration of anoxia indicates the condition originated during the perinatal period.

2-VI-1997 Q1 Interpretation of "tumour"

(UC Chapter II)

[This was the first of three questions from Mary Heanue:]

Tumours, all sites:

Can these be automatically taken as malignant? At the moment we take tumours of the Brain and the Lung as automatically malignant.

[I think we have found one more point where coding practice differs considerably, as you can see from the answers - i.e., should we try to correct the certificates using the knowledge we have of the physicians' certification habits, or should we accept them at face value?]

*England /Lin Shane/: We agree entirely with Patricia. In England & Wales we would make no assumptions regarding malignancy and without further clarification would code as indicated in the index. What an interesting international comparison!

Brazil /Augusto Hasiak Santo/: Regarding the question from Mary Heanue, Ireland, about assuming tumors as malignant, I would like to inform that in Brazil, since the 8th Revision, the words "tumor" and "neoplasm", are taken as "malignant tumor" and as "malignant neoplasm" when appearing in death certificates. This decision was done after the results of the Interamerican Studies of Mortality (Adults in 1962-1964 and Childhood in 1968-1969), in which the death certificates were revised with data from the families of the dead, hospital records, autopsies data and all other available information, providing a new and "more correct" underlying cause of death. It was found that when the physicians mentioned tumor or neoplasm, they really wanted to mean malignant tumor or neoplasm. It was noticed that when a benign neoplasm caused the death, this fact was so unexpected that the physicians mentioned them as benign or informed the morphological type, such as "adenoma" or "meningioma". After those two investigation, many other studies have given the same results, with the same methodology, that we use to call "Puffer Method", after Ruth Rice Puffer, that coordinated them.

Canada /Patricia Wood/: We classify all tumours (of all sites) as indexed. We make no assumptions of nature unless there is an implication of malignancy (as outlined on page 728, Volume I, ICD-9). In Canada, brain tumour (without any further specification and no implication of malignancy) would be assigned ICD-9 code 239.6 and lung tumour would be assigned ICD-9 code 239.1.

Sweden /Lars Age Johansson/: We code the neoplasms as reported, i.e. as "unspecified" if they are described as "tumours", and there is no other indication of malignancy, such as metastases. On the other hand, we will send a query to the certifier if the deceased was less than 75 years old, and for all ages if there was an autopsy or the patient had had recent surgery. The decision not to code "tumour" as "malignant neoplasm" was based on a Swedish study of certificates with "brain tumour", where it was found that a substantial part of the cases were in fact benign (meningiomas). I do not know if that result would hold for other sites as well, Augusto's answer indicates that it would not.

2-VI-1997 Q2 Code for "ischaemic leg"

(UC Chapter IX)

[Mary's second question was:]

Ischemic Leg

There doesn't appear to be a code for this. Perhaps 440.9

Artheriolosclerosis would do?


*England /Lin Shane/: We also would code 'ischemic leg to ICD-9 4439. This has been the practice since ICD-9 was introduced and was coded similarly even to ICD-8. The Micar dictionary in the automated system also assigns 4439 to this condition.

Brazil /Augusto Santo/: Regarding the diagnoses of "Ischemic leg", maybe it would be wise to send queries to the physicians. Ischemic leg does not appear in Brazilian death certificates or hospital records.

Canada /Patricia Wood/: An errata published by Statistics Canada's Nosology Reference Centre in November, 1984 classifies Ischemia, extremity (lower) (upper) to ICD-9 code 443.9 and, if specified as arteriosclerotic to ICD-9 code 440.9.

Sweden /Lars Age Johansson/: In ICD-9, we coded "ischaemia of leg" to 443.9.

2-VI-1997 Q3 Code for "immunosuppression" = AIDS?

(UC Chapter III)

[Mary's third question:]


Can this be automatically considered as AIDS? Can it also be related to renal transplants?

[The answers:]

*England /Lin Shane/: Without any further clarification we would use a code for an immunosuppressive drug (ICD-9 E 9331). We would certainly not assume it was aids.

Brazil /Augusto Santo/: Regarding "immunosuppression" appearing alone in a death certificate, it would not be considered automatically as AIDS since it can be a complication of the treatment of many diseases, like malignant neoplasms as well as transplants procedures, as mentioned by Mary.

Canada /Patricia Wood/: We do not automatically consider Immunosuppression as AIDS. It can be certified as part of the sequence of events leading to the death of an organ transplant recipient as well as a condition related to HIV Infection so we try to classify it in whatever context we see it in.

Sweden /Lars Age Johansson/: We have had cases were the term "Immunosuppression" was indeed used as a euphemism for AIDS, but we would not take that for granted - the term is also used for immunosuppressive treatment. So we would try to get more information from the physician.

10-VI-1997 Q1 Why is self-inflicted poisoning coded differently from other self-inflicted injuries?

(UC Chapter XX)

[Sue Walker's (Australia) question has revealed quite important differences between the countries, so here we have another candidate for clarification. Does anyone know something more about the reason behind the instruction?]

I have an honours student working with me at the moment, writing a dissertation on suicide coding. We are puzzled by the note on page 628 of the ICD-9 which states:

"Categories E980-E989 are for use when it is stated that an investigation by a medical or legal authority has not determined whether the injuries are accidental, suicidal or homicidal. They include self-inflicted injuries, *but not poisoning*, when not specified whether accidental or with intent to harm."

Our question is - what is the reason behind excluding poisonings?

*England /Lin Shane/: If presented with the text 'self inflicted poisoning' and a coroner's verdict was 'Open', then we WOULD use codes E980-E989. Although this appears to go against the ICD-9 instruction, the note on page 628 of Volume 1 could be interpreted as referring to morbidity.

The exception would be 'self inflicted poisoning' with a mention of drug abuse. Even with a coroners 'Open' verdict we would assign an underlying cause code 304.

*USA /Julia E Raynor/: In the U.S. , both poisoning and self-inflicted poisoning are coded as accidental. We assign the accidental code since E9800-E9809 excludes self-inflicted poisoning and poisoning NOS is classified to accidental. There is not (or at least we cannot find one) a direct instruction on how to code self-inflicted poisoning. Suicide categories E950-E959 includes self-inflicted injuries specified as intentional. Our instruction in "Instruction Manual, Part 2b " states to assume poisoning (self-inflicted) by a substance to be accidental unless otherwise indicated. We have used this instruction since the beginning of the 9th revision.

In the 10th revision, the problem appears to be solved by the index. In Volume 3, Self-inflicted, poisoning NEC X49. X49 is an accidental category. I agree the international feedback will be interesting.

Canada /Patricia Wood/: I agree with your interpretation of the ICD-9 index and that "unspecified poisoning" should be coded as accidental poisoning. However, if poisoning, not specified whether accidental or with intent to harm, is specified as "self-inflicted" I would code it as suicidal poisoning. In Canada the "self-inflicted" specification of poisoning, not specified whether accidental or with intent to harm, makes the difference between coding poisoning as accidental or suicidal. Would poisoning, not specified whether accidental or with intent to harm, but specified as "self-inflicted" be coded as accidental poisoning in Sweden? If so, we are consistent in that self-inflicted poisoning, when not specified whether accidental or with intent to harm, is excluded from ICD-9 code block E980-E989, but inconsistent in where it is coded. (It seems that there is no light shed by referring to ICD-10 code block Y10-Y34 as the wording of the note regarding self-inflicted injuries is the same as in ICD-9!)

You are right when you describe this as a very interesting discovery and now I wonder how other countries have, as you say, filled the gap in this situation. Sue's student's question may have set off an international coding debate. Perhaps she will include this in her dissertation paper!

Sweden /Lars Age Johansson/: In Sweden, we have coded unspecified poisonings as accidental, not as suicides. The reason is that the Alphabetical Index has "Poisoning (accidental)" (Vol 2, p 567), so we coded "Poisoning NOS" in the same way as "Poisoning, accidental". Do you think that was wrong? I have tried to find some instruction in the ICD-9 on how to code the unspecified poisonings, but have not found anything except that they are excluded from E98x. Quite clearly ICD-9 had an important gap here, and perhaps different countries filled that gap in quite different ways. A very interesting discovery - you learn much from this email network!

Canada /Patricia Wood/: I will be very interested in the international feedback to Sue's student's question. I remember having many discussions with Elizabeth Taylor about this exact issue ... why is self-inflicted poisoning, when not specified whether accidental or with intent to harm, not included with other self-inflicted injuries, when not specified whether accidental or with intent to harm, in ICD-9 code block E980-E989? Although Elizabeth has recently retired (and is enjoying herself very much) I talked it over with her again last week when this email came in.

Some thoughts . . . Can it be that self-inflicted poisoning implies a certain amount of intent by its very nature? A self-inflicted gun shot can quite feasibly be an accident (several possible scenarios come to mind) so when not specified whether accidental or with intent to harm an assumption is not made either way and the self-inflicted gun shot is coded as undetermined. On the other hand a self-inflicted poisoning implies that the decedent consumed, by his own hand, enough of whatever substance to suffer toxic effects. While not inconceivable, an accidental self-inflicted poisoning is probably far less likely than a suicidal one. Perhaps this is why a self-inflicted poisoning, when not specified whether accidental or with intent to harm, is not coded as undetermined but rather as suicidal.

2-VII-1997 Q1 Linkage pneumonia - measles - leukemia

(Rule 3 / UC Chapter I, X, III)

[We had this question from Gerard Pavillon:]

Here is a death certificate we would like to submit to the group. Two questions:

1a. Do you agree with leukaemia as the underlying cause ?
1b. What underlying cause an automatic coding system would select and which rules would be applied? In particular, how many times rule 3 would apply?

1 (a) Cardiocirculatory arrest
1 (b) Interstitial pneumonia, measles
II. Immunosuppression (chemotherapy), acute lymphoblastic leukaemia

[Suggested coding:]

                        Multiple causes                                        Underlying cause

Brazil               [R092/J849 B059*Y433 D848 C910        J849]

                       R092/J189 B059*Y433 D848 C910          C910

*Canada         I469/J849 B058*C910                                J849

Finland                                                                             C910

Sweden          I469/J849B059*D848 Y433 C910            B052

USA                                                                                 [J849]

[Comments from Canada:]

*Canada /Patricia Wood/: I would code this record to I469/J849 B058*C910 (ICD-9: 4275/5168 0559*2040). Since arrest, cardiac and arrest, circulatory are both indexed to I469 I reasoned that cardiocirculatory arrest could be coded there. Since interstitial pneumonia is not indexed as a complication of measles (like pneumonia NOS is) I did not code it as such. I also did not code it as an adverse effect of the immunosuppression because a "due to" relationship was not specified as would seem to be required when coding complications of drug therapy, (ICD-10, Volume 2, Page 84), if I am interpreting this guideline properly. Making a manual selection of underlying cause I would apply Rule 1 for an underlying cause of J849. I did not apply Rule 3 as I did not see the interstitial pneumonia as "obviously a direct consequence" of another reported condition. I believe that ACME for ICD-9 would select the same underlying cause of death with multiple cause codes 4275/5168 0559*2040 as input.

I understand, and agree with, the comments that coding should reflect clinical reality but I think that many coders do not know whether, clinically speaking, interstitial pneumonia can be considered a complication of measles and/or immunosuppression or not, and will rely on the format of the conditions certified on the record and the way they are indexed in the ICD to make their coding decision. The clearer the rules and guidelines the more consistent the coding will be.

Incidentally, in Canada, multiple subsequent applications of Rule 3 are handled the same way that Julia indicated in her response to this question.

Sweden /Lars Age Johansson/: I agree with Martti that we must take current clinical thinking, and the purpose of the statistics, into account when we make our decisions on how to code. But it is also very important that mortality coding is internationally consistent. As far as I can see, the only way to achieve that is to have quite detailed international coding instructions which are uniformly applied by coders all over the world, and that does not leave too much room for the individual coder's personal judgement. Let's see if we can find some mechanism for bringing clinical thinking and international coding instructions closer to each other. I suspect, though, that we would see as much differences in opinion among clinicians as among coders.

But not in this case, I suppose, and perhaps we should try to have a clarification of Rule 3 and the "highly improbable" section on this.

Julia is right: my coding of Part II is inconsistent. Since I did not select the leukemia, I should not have coded the immunosuppression as an adverse effect. I suppose I did so because I felt that the leukemia was "really" the underlying cause. But, as I said before, you should not let your personal opinion influence the coding...

Finland /Martti Virtanen/: I would like to comment this case from the clinical point of view. I have the feeling that the rules should be changed according to the reality and not vice versa.

Why did the child die? In developed countries like France children rarely die in measles. This is not true with malnourished children. They very easily get serious complications for example from measles. Even measles pneumonia usually does not cause fatality in otherwise healthy children. Therefore one should in most cases have some other underlying cause than measles ( for example malnutrition or as here immunosuppressive therapy).

Why do we collect these data?

Because the society needs to know how successful the treatment of leukemia is?! In that case we have to be able to tell how many died because of the leukemia in spite our therapeutic efforts. Therefore a traffic accident during the therapy should not be counted as due to leukemia but measles, varicella and various sepsis deaths definitely should. Or to know how many measles deaths occur? In third world countries measles is an important killer! However it would be false to mix the malnutrition related deaths of third world with deaths related to therapeutic immunosuppression.

USA /Julia E Raynor/:

a) 4275
b) 5168 0559
II 2040

UC 5168 Rule 1

In the 9th revision, the U.S. treats interstitial pneumonia, 5168, differently than pneumonia. We use interstitial pneumonia somewhat as an idiopathic condition, accepting it as due to very few conditions. The U.S. automated system (ACME) would not apply Rule 3 and code interstitial pneumonia as a direct sequel of measles or leukemia. In Part II, we would not assign a code for immunosuppression (chemotherapy) since we do not consider there is a complication or adverse effect of the therapy.

In the 10th revision, we will continue to treat interstitial pneumonia and pneumonia as we do in the 9th, at least for the first year. Conversion of our automated systems and implementation of the 10th revision must take priority over making changes not directly related to the revision itself.

It is correct, as Lars stated, that it is possible to apply Rule 3 more than one time. We do have an established priority order if a selected condition can be considered a direct sequel of two or more conditions. The order is from left to right (1) on the same line, (2) on lower line in Part 1, and (3) in Part II. This instruction is included in the NCHS Instruction Manual Part 2a, page 23.

Brazil /Augusto Santo/: The death certificate that Gerard submitted to discussion is very interesting as well are important the questions he has addressed. Before discussing these points, some assumptions must be done:

1 - The death certificate coded according to the Ninth Revision guidelines would have "measles with pneumonia" (055.1) as the underlying cause of death.

2 - For the Tenth Revision, we should consider the note (a) on Infectious diseases and malignant neoplasms that appear on page 84 of the Instruction Manual (Volume 2) that states : "Owing to the effect of chemotherapy on the immune system, some cancer patients become prone to infectious diseases and die of them. Therefore, any infectious disease classified to A00-B19 or B25-B64 reported as due to cancer will be an acceptable sequence whether in Part I or II."

3 - The ACME decision table D (1992 edition), regarding modifications for the code for interstitial pneumonia (516.8), only has provision for a possible specification with mention of pneumocystosis (136.3), giving preference for this one.

The proposed death certificate was coded for our SCB System as follows: R092/J849 B059*Y433 D848 C910 that leads to interstitial pneumonia as the underlying cause because of lack of provision for modifications.

Nevertheless, interstitial pneumonia was changed for pneumonia, as follows:

R092/J189 B059*Y433 D848 C910

that has provided Acute Lymphoblastic leukaemia (C910) as the underlying cause.

And now we can answer Gerard's questions.

Question 1 - The underlying cause of death is acute lymphoblastic leukaemia, but we do not totally agree with the guidelines that directs to this decision. It is not peaceful the conception that ANY infectious disease classified to A00-B19 or B25-B64 reported as DUE TO cancer will be an acceptable sequence whether in Part I or II. It is difficult to accept leprosy, poliomyelitis, dengue, leptospirosis, malaria and trypanosomiasis (Chagas disease), only to mention some examples, as due to cancer. On the other hand, sequences are admitted only in Part I (see last paragraph on page 22 of the Intructiona Manual). We may state that this guideline is a serious candidate for clarification. Maybe someone might forward this message to Connie so we could share her sights about this matter.

Question 2 - Considering the above assumptions, an automatic coding system should select acute lymphoblastic leukaemia as the underlying cause. Celso Escobar Pinheiroand I will include for interstitial pneumonia (J84.9) the appropriate modifications in our SCB system and consider the modifications that prevail to pneumonia in preparing them.

Regarding the rules that were applied, we can distinguish between automatic and manual coding. In automatic processing, Rule 1 selected interstitial pneumonia, then a Direct Sequel Combination was applied and measles complicated with pneumonia was reselected (DSC means Rule 3 followed by Linkage with Mention resulting in a Combination code) and then a new Rule 3 finally identified acute lymphoblastic leukaemia. In manual coding, Rule 1 selects interstitial pneumonia, Rule 3 reselects measles, Rule C links measles complicated by pneumonia, then another Rule 3 for immunossuppression, another Rule 3 for chemotherapy and finally another Rule 3 for acute lymphoblastic leukaemia.

Sweden /Lars Age Johansson/: Our automated coding system (which is still in its cradle) would produce multiple codes somewhat different from Augusto's, and it will be interesting to see how ACME will handle them. Perhaps we will have to make some modifications to our software.

When selecting the underlying cause, we would first use Rule 1 to select the pneumonia, and then, as Augusto says, apply Rule 3 to get measles with pneumonia (B05.2+ J17.1*). There is a parallell to "conflict in linkage" here, since we could apply Rule 3 either to the measles or to any of the conditions in Part II. There are no instructions in Vol 2 on how to solve such a conflict, but if we use the same method as for "conflict in linkage" (Vol 2, p 43), we will arrive at the measles.

Then I'm getting a bit uneasy, because even if I would very much like to code the leukemia as the underlying cause of death, I cannot find any instruction in either Vol 1 or 2 that tells us to regard measles as a direct consequence of either immunosuppression or neoplasms. It is true that the instructions (Vol 2, p 67 and p 84) say that infections should be accepted as *due to* malignant neoplasms, but this certificate does not state that the infection was due to either the therapy or the leukemia. As far as I can see, I cannot get any further but have to accept the measles as the underlying cause of death. I'm very probably wrong, and I hope that someone can correct me. I agree wholeheartedly with Augusto: this is indeed a serious candidate for clarification.

As to Gerard's second question, I think it is possible to apply Rule 3 several times on a certificate. There is no specific instruction on that, but if you do not do it, it would mean that you would never arrive at the originating cause if the physician has reported a disease process with more than one step.

2-VII-1997 Q2 – Q4

2-VII-1997 Q2 Malignant neoplasm due to ischaemic heart disease (Rule GP / UC Chapter III)

2-VII-1997 Q3 Ischaemic heart disease in Part I, malignant neoplasm in Part II (Rule 3 / UC Chapter IX)

2-VII-1997 Q4 Carcinoid due to carcinoid syndrome (Rule GP / UC Chapter IX)

[Mary Heanue has three questions on ICD-9:]

We have some cancer medical certs here that we could do with some advise on. We only code to underlying cause in Ireland and we do not do multiple cause coding. The replies to the last set of questions were most helpful. We are now querying all tumours to see if they are malignant or benign.

[Question 2:]
1 (a) Cancer of the Bowel (2 months)
1 (b) Ischaemic Heart Disease/(Myocardial Infarction)

Would you code this as 414.9 or a 159.0?

[Question 3:]
1 (a) Acute myocardial infarction
1 (b) Ischaemic Heart Disease (eight years)
2 Carcinoma of the rectum (one week)

Would you take into account the cancer in part 2 or just leave it as 410? If not, would you take the cancer into account if the duration of the cancer was longer?

[Question 4:]
1 (a) Carcinoid tumour of liver
1 (b) Carcinoid syndrome (259.2)
2 Congestive Cardiac Failure/atrial fibrillation associated with mitral regurgitation

How would you code this one?

[Suggested coding:]

                    Multiple causes                              Underlying cause
Brazil                                                                1590
Brazil                                                                C260
*Canada    1590/4149 410                               1590
Sweden                                                            1590
US                                                                   1590

Brazil                                                                 410
Brazil                                                                I219
*Canada    410/4149*1541                              410
Sweden                                                             410
US                                                                    410

Brazil          2353/2592*4280 4273 4240        2353
Brazil          C229/E340*I500 I48 I340           C229
*Canada     2353/2592*4280 4273 4240        2353
Sweden      2353/2592*4280 4273 4240        2353
US             2353/2592                                    2353


*Canada /Patricia Wood/:

2-VII-1997 Question 2

I would code this record to 1590/4149 410 and making a manual selection of underlying cause, I would consider ICD-9, Volume 2, page 721 B. Interpretation of highly improbable ... (b) a malignant neoplasm reported as "due to" any other disease, and I would apply Rule 2 for an underlying cause of 1590.

2-VII-1997 Question 3

I would code this record to 410/4149*1541 and making a manual selection of underlying cause I would apply the General Rule for a tentative underlying cause of 4149 and then I would apply Rule 7 to reselect the underlying cause as 410. Considering ICD-9, Volume 2, page 721 B. Interpretation of highly improbable ... (h), I would not apply Rule 3 for a direct sequence to the cancer regardless of its duration.

2-VII-1997 Question 4

I would code this record to 2353/2592*4280 4273 4240 and making a manual selection of underlying cause I would select 2353 considering the coding instruction in ICD-9, Volume 1, page 154 re: carcinoid syndrome not being used as the primary code for carcinoid tumour.

US /Julia Raynor/:

Question 2

The U.S. would code the underlying cause to cancer bowel, 1590, based on the "highly improbable " list, Volume 1, pages 721-722, (b).

Question 3

The U.S. would code the underlying cause to 410 using General Rule and Rule 7. We would code 410 as U.C. even if duration of cancer in part II was longer.

Question 4

a) 2353
b) 2592
Underlying Cause 2353

Brazil /Augusto Hasiak Santo/:

Regarding question 2 (first of Mary's), the underlying cause of death is coded as cancer of the bowel (159.0 in ICD9 and C26.0 in ICD10). The interpretation of *highly improbable* are practically the same in both Revisions. It should be regarded as highly improbable ...a malignant neoplasm reported as due to any other disease - added in ICD-10 - except human immunodeficiency virus [HIV] disease.

Regarding question 3 (second of Mary's) the underlying cause would be left as acute myocardial infarction (410 in ICD-9 and I21.9 in ICD-10) even if the duration of the cancer was longer and the carcinoma mentioned on line c of Part I. The interpretations of *highly improbable* are also the same for both revisions. It should be regarded as *highly improbable* ... chronic ischaemic heart disease (412-414 or I20, I25) reported as due to any neoplasm.

Regarding the last question  (4) (third of Mary's) and considering that the cross-reference in the Index for the Ninth Revision ( "see also Neoplasm, uncertain behaviour") differs from the one of the Tenth Revision ( "see also Neoplasm, malignant"), the underlying cause of death would be carcinoid tumour of liver (235.3 in ICD-9 and C229 in ICD-10) and the death certificate

would be coded as follows:

2353/2592*4280 4273 4240 in ICD9, or

C229/E340*I500 I48 I340 in ICD10".