We found that students' classification of conditions that they are likely to encounter during their professional lives as "disease", or not, was concordant > 80% across year cohorts for 16 conditions, each of which might be encountered during the first part of a medical programme. Similarly there was concordance < 20% about which conditions classified as "non-disease" for four - baldness
, fractured skull
and heat stroke
. To determine what showed 'good' concordance and 'poor' concordance, there is no universally agreed definition of "good agreement". However, one standard statistical text suggests that for kappa "good agreement" is 0.60 to 0.80 and this seems a reasonable pair of arbitrary cut-points to adopt [5
The 16 conditions with poor concordance (concordance between 21 and 79%) are of interest. They suggest that the criteria for classifying conditions as "disease" are unclear. For each, one could imagine arguments mounted both for and against. The decision to label something as "disease" clearly has health implications for individuals, but could alternatively be thought of as a societal problem with causes (related to urban design, the economics of food availability and so on) remote from any direct clinical influence.
Some conditions are clearly controversial, such as chronic fatigue syndrome
and attention deficit hyperactivity disorder
. Clinicians' opinions are divided on whether these conditions have an organic cause or are linked to personality, or a societal aetiology and our students' ratings may be a reflection of this. A study of 1250 US Healthcare providers showed this with 20% agreeing that "Chronic fatigue syndrome is only in the patients head", with 51% disagreeing but a further 29% answering they did not know [6
]. A qualitative study of Australian General Practitioners attitudes and practices to attention deficit hyperactivity disorder showed that those that participated in the focus groups emphasised social, family and environmental factors, however the evidence emphasises the neurobiological nature of attention deficit hyperactivity disorder [7
The prescription of drugs for attention deficit hyperactivity disorder is managed several times more often in some parts of Australia compared to others, suggests that such controversial classifications can contribute to differences in clinical management [8
]. Labelling a condition has implications for the way it is managed. For example, having bronchitis
justifies the need for antibiotics, whereas the same clinical presentation described as "a flu"
does not [9
]. Even for more 'objective' conditions there may be disagreement. For example, there is confusion on how to treat ductal carcinoma in situ. Women may be given conflicting information on whether to treat or not based on the labelling of this condition. At a recent conference in an audience of 100 surgeons and oncologists, the audience was asked to vote on how they might describe ductal carcinoma in situ to a patient, 50% voted "might become breast cancer," 25% voted "breast cancer at an early stage," and 25% voted "breast cancer and can't be left" [10
Several previous studies have looked at agreement about labelling and the impact of a medical versus a non- medical label. Participants were psychology and medical students. Both studies resulted in participants considering the medical label more serious, less common and therefore more 'disease-like' than when the condition had the non medical label [11
Medical labels were classified as "disease" more than their lay descriptions in two of three duplicate candidate conditions. There are several possible reasons. Assuming students did not appreciate the conditions were synonyms, they may have been unfamiliar with the more technical description (although for this we might have expected concordance to have increased with the more senior students). Perhaps students simply worked quickly through the list using a heuristic which classified anything technical as disease and anything in common language as "non-disease". This might explain the otherwise non-intuitive classifications of lead poisoning, heat stroke, and fractured skull as "non-disease".
As can be seen from Figure , there was little difference between years of students, but some differences when compared to the earlier Campbell survey of secondary school students. For the 19 conditions the surveys had in common, there was no statistically significant difference for 8. Of the 11 where there was a statistically significant difference, the differences were mostly small except for four conditions: depression, schizophrenia, epilepsy, and haemophilia. While these may reflect societal changes in attitudes, particularly to mental illness, they may also simply reflect sampling differences.
Our methods had shortcomings. The small numbers and multiple comparisons mean that any trends should be interpreted with caution. They may be biased, in that only the years 1, 2 and 3 cohorts were surveyed and only those that attended the lectures. Despite our good response rate, we surveyed students from only one university in Australia who may not be representative of all medical students. There were difficulties in accessing more experienced medical students because they were out on rotations. Surveys are often difficult to interpret, as well as including items from too small a sample of possible conditions. We could look at a different set of conditions in the future as well as a more representative sample of students and practising doctors.