BERKELEY, CA (UroToday Inc.) - Assessment of patient life expectancy is an important variable in considering treatment alternatives or deciding on delivery of any treatment to an individual patient for a disease process, in particular cancer. For example, radical prostatectomy is usually reserved for those with a 10-year, or perhaps (in Scandinavia) a 15-year life expectancy.

Dr. James Wilson and associates from the Taunton and Somerset Hospital, Taunton, UK and Beatson Oncology Center in Glasgow, UK assessed the ability of urologists and medical oncologists to accurately predict 10-year survival. Their findings are reported in the April 2005 issue of BJU International.

Several indices for assessment of patient life expectancy are available. These authors employed actuarial data, as used by insurance companies, to assess policyholders as an initial step to adjust patient comorbidities for age. This age-adjusted number was then used in conjunction with the Charlson Index and applied to 70 fictional case scenarios, of which 57 were single use and 13 were repeated. A case example is that of a "74 year old individual with lower urinary tract symptoms, blood pressure found to be 170/90 and mild gout".

The physician population consisted of 18 doctors: 4 consultant urologists, 2 consultant oncologists, 4 urology specialist registrars, 4 surgical senior house officers and 4 surgical pre-registration house officers.

These physicians reviewed the 70 case scenarios. Their assessment of 10-year life expectancy was compared to those calculated using the actuarial data combined with Charlson index. The 13 repeat cases were used to assess intra-doctor reliability.

Compared with actuarial predictions, doctors tended to underestimate the probability of 10-year survival, with considerable variability within and between doctors; this was particularly marked amongst the more junior doctors. Fifteen doctors predominantly underestimated 10-year survival with 10 of these underestimating over 75% of the cases.

The test-retest reliability of the 13 repeated cases showed overall intra-doctor reliability of 0.74.

Application of the results can be seen as follows: given a situation where guidelines suggest that treatment should be proffered if the estimated probability of 10-year survival is >50%. Of the 70 case scenarios, actuarial estimates would suggest treatment for 53 of them under this situation. For the 53 cases, the 18 doctors would on average recommend treatment for only 66% and deny offering treatment for 34%. Conversely, the doctors would concur with 76% of the 17 cases where actuarial data would suggest withholding treatment, thereby 'inappropriately' recommending treatment in 24%.

These data are insightful in demonstrating the inability of a select group of physicians to accurately combine patient age and comorbidites for determination of 10-year life expectancy. While doctors of all seniority levels in this study need more education and training in this regard, it is most pronounced in the more junior physicians in training.

BJU Int 2005;95:794-798