Patients'expectations of adjuvant cytotoxic chemotherapy

[j.vaidya@ucl.ac.uk]

Christine Gratus 

(http://www.thelancet.com/newlancet/sub/issues/vol352no9127/correspondence581_3.html)

gives a graphic illustration of patient's plight when prognosis of a life threatening 

disease such as breast cancer is discussed.

Firstly, one must realise the difference between probability and event. A probability is 

the chance of an individual actually experiencing an event in future- it is between 0 

and 1; after the event has already occurred, its incidence is no longer a probability. 

It has either occurred or not occurred.  It is amazing how frequently the latter is used 

naively to disprove the former.

Secondly, in terms of natural history of breast cancer, it is known that women 

can die of breast cancer even 30 years after the primary diagnosis: so the chance of 

living 5 years, 10 years and natural life span, will always reduce in that order. 

Clinicians quote the figures reported in clinical trials which are usually in the form 

of 5 and 10 year survival rates. Using a new method of survival analysis(2) the 

probability of living the full life span can also be estimated and it appears that this 

is preferred by patients(3). The different estimates that Gratus was given were 

probably because exact estimates for each individual tumour are not easy to find and 

each is an projected estimate from known values.

When one says that there is a 10% improvement in survival after adjuvant therapy, it 

means either a) every woman receiving the treatment has a 10% improvement in her 

survival, or b) some women have a 20% improvement, some have a 10% improvement and 

some have no improvement at all-together averaging to the 10% figure. The rationale for 

suggesting the second scenario is the possibility of a very variable individual 

responses to treatment. The only way to know which of these is true is to have every 

woman a control for herself, which of course is not practically possible. However, 

confidence intervals do tell us a lot about how much variability there can be within the 

groups. In Gratus' case,  when the improvement was expressed as 1-2.5%, these 

probably represented the  95% confidence intervals: which mean that the improvement for 

95% of women would fall within that range and only 5% of patients would fall outside it. 

Estimates based on neural networks might be able to narrow down the confidence 

intervals, but a replacement of the crystal ball is yet beyond science.

	

Jayant S Vaidya	 Michael Baum



References:

1. Gratus C. Patients' expectations of adjuvant cytotoxic chemotherapy. Lancet 15 August 

1998; 352: 581

2. Vaidya JS, Mittra I. Fraction of Normal remaining life: a new method to express 

survival in chronic diseases. BMJ 1997; 314:1682-1684.

3. Vaidya JS,  Douek M, Lokare VG, Secker-Walker J, Fallowfield L, Taylor I, Baum M. 

Communication of breast cancer prognosis: Women wish to know their chance of cure 

[abstr]. Eur J Surg Oncol 1998; 24:345.