A method of calculating diagnostic indexes for possible cancer symptoms in general practice
AuthorHoltedahl, Knut Arne
Most medical encounters start with the presentation of symptomns that lead to hypotheses about the diagnosis. It might be useful to know the probability of a disease when a symptom of it is presented in a consulation with a general practitioner, i.e., the positive predictive value (PPV) of a particular symptom in relation to a disease. A diagnostic index of this kind would need to be based on data about symptom prevalence and disease prevalence in the population studied, and on sensitivity of the symptom as an indicator of the disease at the time of the encounter. Likelihood ratio (LR), the probability of a given symptom when a particular disease is present divided by the probability of finding the symptom without the disease, could also be calculated on the basis of such data. The calculation of diagnostic indcxes for colorectal cancer and lung cancer is shown. A Norwegian general practice study of the seven warning signals of cancer has yielded data about symptom prevalence. This material has been combined with information on local groups of cancer patients, cancer statistics and population statistics to produce necessary estimates of cancer prevalence and of sensitivity. The PPV of the warning signal "indigestion" in relation to colorectal cancer increases with age from <0.1 % to2.5%-4.4% for men 70 years old or more. and 3.4%-6.1% for women in this high age group. LR varies from 7 to 46 and has slightly lower values for the highest age groups. For ''cough/hoarseness'' the PPV range for men 60-69 years old is 3.9%-9.0%. There were insufficient data for males in other age groups. For women, PPV for the different age groups varies from 0.1-0.2% to 0.6-1.5%, with a slight increase with age. LR in the age group 60-69 years is 22-55 for men and 13-30 for women. There is a slight tendency to lower LR values with increasing age in women. The value of a quantitative approach to cancer diagnosis in general practice is considered. Single symptoms may be a good starting point, but they rarely furnish a sufficient basis for decisions. The external validity of the figures is limited because some of the base estimates may vary with time and from place to place. However, the changes from one age group to another and differences between males and females probably would show limited variation. The method may serve as an example of a local quantitative approach to diagnostic thinking in general practice.