Lost cause?

Previously I wrote about how mortality rates by cause of death vary by deprivation index (and, by implication, socio-economic group). This substantially complicates any attempt to use cause-of-death data to make projections of mortality for annuity portfolios and defined-benefit pension schemes.

Another challenge lies in changes in classification over time. The most obvious change is that of the classification system itself: up until 2000 England & Wales used the ICD 9 system, whereas the ICD 10 system has been in use since 2001. Mapping from one to the other isn't always straightforward, not least because ICD 10 has many more causes of death than ICD 9.

One quick and obvious solution is to look at periods covered by the same classification system. For example, ICD 9 was used between 1979 and 2000 in England & Wales. Table 1 lists the top six individual ICD 9 codes for deaths amongst males aged 70-74 in 1979, and their corresponding position in 2000.

Table 1. Frequency of top six causes of death cited for males aged 70-74 in England and Wales. Source: 20th Century Mortality.

Table 1 contains no surprises for the top two causes of death for males in this age group: heart attacks were the No. 1 killer, followed by lung cancer. This was the same in both 1979 and 2000. However, bronchopneumonia had slipped from third place in 1979 to fifth place in 2000. The most obvious reason is the sharp rise in importance of two ICD 9 codes for ischaemic heart disease. However, a subtler part of the answer can be found in the note accompanying the data:

"Trend analysis spanning the years either side of 1984 and 1993, must take into account some important coding changes. There is a large increase in mortality from chronic diseases [...] between 1984 and 1993. This is an artefact due to changes in the way ICD-9 rules for selecting the underlying cause of death were interpreted in England and Wales. [...] As a result, some deaths for which bronchopneumonia in Part I of the certificate would previously have been coded as the underlying cause of death were coded to a condition mentioned elsewhere in Part I or Part II."
 

Source: Dr Paul Aylin, Medical Epidemiologist, Office for National Statistics.

In other words, some of the "trends" by cause of death are a result of changes in classification methodology. This means that restricting our attention to periods using the same ICD system does not remove all the problems we thought it might.

When doing projections by cause of death we must not only take great care with socio-economic differentials, but we must also take note of uncertainty surrounding the classification of cause of death. Perhaps as a gentle reminder of this, the six causes of death in Table 1 contain one occurrence of the phrase "ill-defined" and three uses of the word "unspecified".