A shaky foundation?

As with anything that must combine reliable data with hard maths and sound judgement, forecasting mortality is difficult. When using stochastic projection models, reliable data is critical, since without it, we are building an ambitious structure on a shaky foundation. Even with seemingly straightforward all-cause data, problems can and do exist, but when working with mortality data classified by cause-of-death, the challenges are both numerous and difficult to navigate.

A key extra step in the generation of cause-of-death data is the classification process itself. We've previously discussed issues around changes in the classification systems used through time making it difficult to create a consistent time-series. One issue we've not discussed before involves shifts in the information upon which death classification is based. This year, Turnbull et al (2015) reported on the huge decline in hospital autopsies in the UK and other countries. In 1960 we'd have seen around 40% of hospital deaths subjected to post-mortem examination, whilst today we see less than 1% of hospital deaths autopsied across the UK. Since the majority of UK deaths are recorded in a hospital setting, this is clearly a dramatic change.

With the greater prevalance of diagnostic imaging it isn't obvious what impact the fall in the use of post-mortem histology will have made. To what extent might changes to the information available to a medical examiner influence the conclusion they reach? This question was considered Nashelsky & Lawrence (2003) who found two medical examiners blinded to autopsy results assigned a cause-of-death that was wrong in 28% of cases. This general finding seems supported by the meta-analysis from Roulson et al (2005) who examined discrepancies between autopsy and clinical diagnoses.

"Clinicians appear to be more accurate in diagnosing the main admitting condition, with discrepancy rates ranging from 15 to 30% and lower (612%) when they were confident of the diagnosis. The cause of death produces more discrepancies than the main diagnosis, with rates of 30% and above."


There are many reasons why practitioners may want to forecast mortality by cause-of-death. The marked changes in quality and type of information available to the classification process raise doubts about whether cause-of-death forecasts can credibly stand alone. Given this aspect and others, the prudent conclusion seems to be that the results from cause-of-death projections require strong corroboration using other methods before forming part of the justification for a mortality basis.

References

Turnbull, A., Osborn, M. and Nicholas, N. (2015) Hospital autopsy: Endangered or extinct?, J Clin Pathol doi:10.1136/jclinpath-2014-202700

Mooney, H. (2013) Why we are still dying at hospital, not home, Health Service Journal, 7th June 2013.

Nashelsky, M. B. and Lawrence, C. H. (2003) Accuracy of cause of death determination without forensic autopsy examination, Am J Forensic Med Pathol. 2003 Dec;24(4):313-9.

Roulson et al (2005) Discrepancies between clinical and autopsy diagnosis and the value of post mortem histology; a meta-analysis and review. Histopathology. 2005 Dec;47(6):551-9.

 

Comments

captcha

Find by key-word


RECENT POSTS

The upcoming EU General Data Protection Regulation places focus on ... Read more
Assume we have a random variable, \(X\), with expected value ... Read more
Our new book, Modelling Mortality with Actuarial Applications , describes ... Read more
Gavin Ritchie
Gavin Ritchie is the IT Director of Longevitas