When seeking to prove reduced life expectancy following a delay in diagnosis of cancer, practitioners need to be wary of:
- over-reliance on statistical or epidemiological evidence particularly in cases involving rare cancers where there is a paucity of data.
- applying data or evidence regarding one type of cancer to a case involving a different type of cancer.
The recent Fatal Accident Act case of Hague & Rich v Dalzell &Fish  EWHC 2753 (QB) illustrates these difficulties.
Breach of duty had been admitted. The defendant GPs negligently failed to refer the deceased for investigation of abdominal pain and vaginal bleeding. Had they done so then cervical cancer would have been diagnosed and surgery performed. The delay in treatment was seven months and four months in relation to the failures by the first and second defendant respectively.
After actual diagnosis and treatment the deceased sadly suffered further cervical cancer - the consensus was that this was a recurrence due to the growth of residual cancer cells following the initial surgery. She died just over a year after the recurrence was diagnosed.
Mr Justice Lewis found as fact that at the time when diagnosis ought to have been made, the tumour would have been stage 1b1. At actual diagnosis, as I understand it, the clinical staging was 1b2 but there was agreement amongst the experts that the tumour "would have been" classed as a 2b tumour on the basis of the MRI scan at that time.
The Judge rejected attempts by one of the Claimants' experts, disputed not only by the Defendants' experts but also by the second of the Claimants' experts, to use doubling times to interpolate the size and characteristics of the tumour at the time when diagnosis ought to have been made. He quoted from expert evidence as follows:
" ...it is inadvisable to use tumour doubling times in this particular tumour because there is no reliable data for the volume doubling time of this tumour or any primary cervical carcinoma growing in situ."
It was agreed that the particular cancer affecting the deceased was very aggressive and rare. It was a sarcomatoid squamous cell carcinoma. There were only 17 such cases in the literature, the court was advised. The Judge therefore also rejected the application of studies on survival chances following treatment for other, more common and less aggressive, kinds of cervical squamous cell carcinomas.
That much is uncontroversial.
However the Judge went further in his rejection of the Claimants' use of statistical evidence to show that the deceased would have been cured with earlier diagnosis and treatment.
- He observed that "The figures are intended to show prospectively the likelihood of survival. They are not designed to be applied retrospectively to predict the likelihood of whether any particular individual would have survived if treated by a particular date."
- He commented further that "the statistics are intended to show likely survival rates for a cohort of people ... [evidence that shows] that just under 95 out of a hundred such patients would be alive after 5 years does not assist in determining whether or not a particular patient will be one of the 95 patients who survive or one of the 5 who do not."
Nicholas Peacock, for the Defendants, took the Judge to various judicial comments on the limitations of the use of statistics.
Mr Justice Lewis was persuaded by the expert evidence that "the tumour in the present case was a highly aggressive, rare form of cancer, namely a sarcomatoid squamous cell carcinoma and that such tumours have a poor prognosis and are likely to recur." The Claimants could not persuade him that the deceased would have probably survived more than 5 years on earlier diagnosis.
Whilst statistics are intended to show prospective survival rates they are often based on data of actual survival. The judgment shows the need to scrutinise the data relied upon. Is it a retrospective study of actual mortality?
Of course statistical evidence relating to a cohort can never demonstrate beyond doubt what would have happened to a particular individual, but the test is the balance of probabilities. If, on all the evidence, there is good reason to conclude that the particular patient would have been in the 5% rather than the 95%, then the fact that 95% survive will be irrelevant. Absent such evidence, however, the fact that 95% survive is surely compelling evidence of the likelihood that the particular patient would have survived, assuming that the data is directly relevant to the particular cancer.
It is of interest that the issue was framed for the Court as whether the deceased "would have survived for five years if she had been treated" earlier. It seems that five year survival was treated as a cure and that the case was brought on a "cured or not cured" basis. I cannot find reference to median life expectancy in the judgment. In some cases it is possible to show a reduction of a number of years even if both with and without delay the deceased would not have survived more than five years. Perhaps that was not possible on the evidence in this case.
Damages had been agreed at £225,000 had the causation findings gone in favour of the Claimants, and £15,000 on the findings as they were in fact made. The smaller sum was for the estate's claim, not for the dependants under the FAA 1976.