Changes to coding of diabetes in the Admitted Patient Data in NSW since July 1998 and the impact of these changes on the analysis of diabetes hospitalisation trends

• The coding of diabetes as a principal diagnosis was stable between July 2000 and June 2010. The records within this period are expected to reliably reflect hospitalisation for diabetes (with and without complications) as the primary reason for admission between 2000-01 to 2009-10 inclusive. A substantial change was however introduced in July 2010.

• The coding of diabetes as an additional diagnosis has been very unstable from 1989-90 and the coding has changed 8 times between July 1998 and July 2012. Consequently, monitoring the trends in prevalence of diabetes in the community, or monitoring hospital admission practices by using hospital records of diabetes in additional diagnosis field is not appropriate.

The period after July 2012, that is a period starting in financial year 2012-13, is the time when all admissions with diabetes are accounted for in both primary and additional diagnoses, as long as diabetes is documented in the patient file. In that period, the distinction between diabetes with and without complications reflects the true status of diabetes, as assessed during hospitalisation. Consequently, data from 2012-13 and following years can be used to estimate the burden of diabetes in the community and the distribution of hospitalisations for diabetes with and without complications in NSW.

There have been seven changes concerning the coding of diabetes in the Combined Admitted Patient Epidemiology Data (CAPED) in NSW since 1998.

A precursor to these changes was the change of the version of the classification system from International Classification of Diseases version 9 (NCC 1996) to version 10 (NCCH 1998) in July 1998. The coding of diabetes was not affected by this change.

1. The first change was introduced in July 1999. It required that diabetes (with or without complications) was included in the case record as an additional diagnosis only if it affected the hospital admission (see in 1.1 below). This requirement applied to all conditions already from 1998 with the exception of hypertension, Parkinson disease and diabetes. In Australia in July 1999, this exception was removed [Australian Coding Standard (ACS) 0002] by the National Centre for Classification in Health.

1.1. The Australian Coding Standard 0002, volume 5 [each edition from 1998 up to and including 2010 (NCCH 2010)]:

For coding purposes, additional diagnoses should be interpreted as conditions that affect patient management in terms of requiring any of the following:

• therapeutic treatment, (from July 2008 the following has been added: ‘commencement, alteration or adjustment of therapeutic treatment’ see point 6 for discussion)

• diagnostic procedures

• increased nursing care and /or monitoring.

1.2 This change had some effect and the rate of diabetes in additional fields (both with and without complications) was reduced in NSW. At that time 20 diagnosis fields were available in the electronic Admitted Patient Data system in NSW.

2. The second change, in July 2000 concerned only diabetes with complications (in principal and additional diagnosis fields). It stated (ACS 0401) that it was no longer necessary for clinicians to make a clear ‘cause – effect’ connection between diabetes and another co-existing condition for medical coders to be required to code diabetes with complication, as long as the complication was on the ‘approved ICD list in the Alphabetic Index’. The list comprises conditions classified to categories of diabetes E09-E14 in ICD-10-AM. A separate code for the co-existing condition was also to be used in order to describe the condition more fully.

2.1. For example, a statement in clinical notes to the effect that there is increased intra-abdominal fat deposition leads to the use of code E1x.72: ‘diabetes mellitus with features of insulin resistance’. Renal diseases must be matched with ‘diabetes with renal complication’ (E1x.2x).

2.2. This change had a considerable effect on rates. The rates for diabetes with complications (both in principal and additional diagnosis fields) rose and the rates for diabetes without complication decreased (both in principal and additional diagnosis fields).

2.3. This effect, where records flowed from diabetes without complications to diabetes with complications resulted in improving the scope for monitoring the prevalence of cases of hospitalisations for diabetes with complications (co-existing, not necessarily caused by diabetes as well as those caused by diabetes, which often could not be definitively ascertained).

3. The third change in July 2004 involved only NSW hospital records. It reversed the changes from July 1998 and July 1999 concerning the coding of additional diagnoses. That is, from July 2004 in NSW, all conditions diagnosed in a patient and recorded in a patient file when hospitalised, including diabetes, were coded in the hospital record regardless whether they impacted on the hospitalisation. At that time there were 55 diagnosis fields available in the electronic Admitted Patient Data system in NSW (although only the first 15 or so were ever filled in).

3.1. This change had a dramatic effect on the rates for diabetes with complications (which increased) and without complications (which decreased) in additional diagnosis fields in NSW. The change achieved an improved accounting of diabetes with complications in hospitalised persons in NSW.

3.2. In summary, prior to July 2004, coders coded diabetes without complications in cases where diabetes itself and not a co-existing complication was attended to during a hospitalisation and the hospitalisation occurred for another reason altogether, recorded as principal diagnosis, neither diabetes nor its complication.

3.3 For example, a patient could have had insulin levels measured and medication could have been dispensed accordingly, but renal, neurological or any other complications were not attended to directly. Before July 2004, this was regarded as ‘diabetes without complications’ from the point of view of hospital stay. This confusing coding practice was employed to fulfil the requirement that only a diagnosis which impacted on a hospital stay should be acknowledged in the record.

3.4. After July 2004, the increase in the number of records with diabetes with complications as an additional diagnosis included both records previously coded as diabetes without complications as well as additional records of diabetes with complications in cases where diabetes was previously excluded as not relevant to a hospital stay.

4. The fourth change, in July 2005, reversed the change from July 2004 and concerned additional diagnoses only. The reversal was due to problems in distinguishing diagnoses of diabetes coded as relevant from those coded as not relevant to a hospitalisation.

4.1. It should be noted that the 2004 change was contrary to the Australian Coding Standards and in the absence of appropriate filtering of records the NSW health system could not fulfil its national reporting obligations. The change introduced in 2005 was therefore necessary, to ensure that the data collection was properly supporting the system of Diagnosis Related Groups and financial monitoring although it was at the expense of improved reporting of diabetes prevalence in the population.

4.2. The change from July 2005, that is the one that returned the NSW diabetes coding in hospital statistics to the national standard, did not bring the expected reduction in the numbers and rates of diabetes in additional diagnosis in the NSW hospital statistics. The rates in 2005-06 were higher than in 2004-05 and continued to rise afterwards.

4.3. The reason for this lack of change in the trend was that the diabetes was still coded even if its impact on hospital stay was limited to monitoring. That is even if only blood glucose test was performed, diabetes was still coded as relevant to hospital stay.

5. In July 2008 the Australian Coding Standard 0002 concerning additional diagnoses was revised (see the first dot point in point 2.1 above) and the requirement for relevancy to hospital stay was strengthened. This was the sixth change. The new definition of additional diagnosis directs coders to include diabetes as an additional diagnosis only if a substantial alteration to the treating regime took place (NCCH 2008). This change had a dramatic effect on the numbers and rates of diabetes in additional diagnosis field in the NSW hospital statistics, with a 54% reduction in comorbidity rates between 2007-08 and 2008-09.

6. In July 2010 the Australian Coding Standard 0401 concerning diabetes mellitus and impaired glucose regulation was revised and a major change was introduced. The coding of diabetes was changed to follow the rules of coding of the principal (ACS 0001) and additional diagnoses (ACS 0002) applying to all other hospitalisations thus removing an anomaly known as ‘a reverse index pathway’ coding concerning diabetes up to that point. This change meant that the coding of the principal diagnosis was to take place first and additional diagnoses were to be coded then in all hospitalisations that involved diabetes. This change affected both the coding of the principal diagnosis and additional diagnoses. This was the sixth change.

6.1. For example, a cataract or ulcer being treated as the main reason for the hospitalisation in a person with a history of diabetes would have previously been coded as diabetes with a matching complication in the principal diagnosis and a cataract or ulcer would have been included in additional diagnosis (‘a reverse index pathway’). Since July 2010 a cataract or ulcer are coded as principal diagnoses and diabetes is included as additional diagnosis if it fulfils the criteria of additional diagnosis under the rules of ACS 0002, that is if it substantially impacts on admission (as explained in point 2.1 and 5 above) or if a ‘cause and effect’ relationship between diabetes and a complication is clearly established in medical notes (see points 3. and 3.1) and diabetes can be classified as a problem or underlying condition to the condition in the principal diagnosis.

6.2. This change affected diabetes in both principal and additional diagnoses. In the principal diagnosis the change had a dramatic effect on hospitalisation rates, with an almost 60% drop in rates between 2009-10 and 2010-11.

6.3. This change had a negative impact on the use of hospital data to monitor the burden of diabetes in the population. The long term trend of hospitalisations for diabetes in the principal diagnosis no longer reflected the rising prevalence of Type 2 diabetes. Many cases of complications of diabetes could be still ascertained via additional diagnoses. However, many complications had not been included as additional diagnoses since July 2010. These complications were, for example, where the ‘cause and effect’ had not been specifically mentioned in the clinical notes, but did exist, or conditions where the ‘cause and effect’ did not apply but the principal cause of the hospitalisation had been worsened by co-existing diabetes.

7. In July 2012 the rules of coding of diabetes in additional diagnosis were changed again. The direction to coders has been since that ‘Diabetes mellitus and intermediate hyperglycaemia should always be coded when documented’ (Rule 1 of ACS 0401). This was the eighth change to the coding of diabetes since 1998. This change reversed the sixth change contained in ACS 0002 in relation to diabetes (see point 6).

7.1. The change in July 2012 has affected diabetes as an additional diagnosis by increasing the number of hospitalisations where the diabetes is coded in the record of hospitalisation. There was an almost 4.5 fold increase between 2011-12 and 2012-13. Before the July 2012 change, many cases of diabetes were not coded because they did not have a sufficient impact on treatment provided during a hospital stay (ACS 0002). After the change, diabetes is coded regardless of its impact on the course of hospitalisation (ACS 0401, Rule 1).

7.2. From July 2012, the distinction between diabetes with and without complications reflects the true status of diabetes, as assessed during hospitalisation. The equivalent period in the first half of 2010s was the financial year 2004-05 (and only 2004-05). Data from this year can be used for comparisons with 2012-13 and subsequent years, if desired.

7.3 The July 2012 change affected the coding of diabetes as an additional diagnosis and there is no evidence of an effect on the numbers and rates of hospitalisations for diabetes as a principal diagnosis in the first year after the change.

National Centre for Classification in Health. The International statistical classification of diseases and related health problems, 10th Revision, Australian Modification (ICD-10-AM). Seventh edition. Australian Coding Standards Sydney: NCCH, 2010.

National Centre for Classification in Health. The International statistical classification of diseases and related health problems, 10th Revision, Australian Modification (ICD-10-AM). Sixth edition. Australian Coding Standards Sydney: NCCH, 2008.

National Centre for Classification in Health. The International statistical classification of diseases and related health problems, 10th Revision, Australian Modification (ICD-10-AM). First edition. Sydney: NCCH, 1998.

National Coding Centre. The Australian version of the International Classification of Diseases, 9th revision, clinical modification (ICD-9-CM). Sydney: University of Sydney, 1996.

University of Wollongong. Casemix and Coding Matters Newsletter