Clinical epidemiology—how important now?

Chronic diseases have become an increasing portion of modern medicine in the Western world and the goals for treatment are not only to relieve symptoms and their influence on daily life for patients but also to avoid long term complications of the disease and of the treatment given.

The rules for obtaining evidence of the benefit of a new treatment have been widely accepted, with the randomised controlled study as the “gold standard”. Such studies give some guarantee of the effect of a given treatment modality. The possible long term side effects however cannot be secured against in studies carried out over a few months, the usual length of a controlled trial.

The natural history of a chronic disease—how will the disease proceed if no medical intervention is carried out—should ideally be the background for any therapeutic trial. For the immediate short term course, the placebo arm of a controlled study is sufficient. Knowledge of the long term natural course of the disease however does not exist as both the medical profession and patients themselves have continuously interfered and tried to relieve the consequences of the disease. Gradually therefore, the meaning of “natural course” has changed to “the course of the disease when treated in accordance with established and well defined treatment policies”. Studies of this kind, systematic collection of clinical data, have extended the subject of clinical epidemiology from estimation of incidence and prevalence of diseases in a population, to long term follow up of cohorts of patients. Good epidemiological studies imply avoidance of selection of patients using regional patient cohorts, securing complete follow up with a minimum of untraced patients, and use of well described and unambiguous definitions of diagnosis and clinical parameters. If the pitfall of selection is not avoided, the created patient group may be unrecognisable, and the results obtained become misleading.

Modern information technology has facilitated the creation of large databases which, combined with epidemiologically correct data collection, give the possibility of depicting prognosis and the long term course of chronic diseases during a given treatment regimen. Although it is both costly and time consuming to maintain such databases, it is our professional responsibility to cumulate clinical experience in order to inform patients properly and to find strategies for improvements in treatment.

With regard to inflammatory bowel disease, the above mentioned goals for medical management of these chronic diseases are mandatory. The development of new biological treatment modalities, such as infliximab, the tumour necrosis factor α inhibitor with strong immunosuppressive properties and a well documented effect on Crohn’s disease, has been the major achievement within the last half decade. The long term effect however, and in particular the fear of increased risk of lymphoma, has been a matter of concern. In this issue of Gut, a Swedish group¹ of very experienced clinical epidemiologists have used their two regional population based cohorts of patients with ulcerative colitis and Crohn’s disease—more than 8000 patients—diagnosed between 1955 and 1990, to examine the occurrence of haematopoietic disease during the course of IBD (see page 617). Furthermore, they added a nationwide inpatient register of 45 000 patients diagnosed between 1964 and 2000. All patients were identified by an individual national registration number and linked with the Swedish cancer register. Patients from the regional cohorts and national inpatient register were analysed separately, but the results were very similar—namely, a non-significant, borderline increased risk of lymphoma in patients with Crohn’s disease and a borderline increased risk of myeloid leukaemia in ulcerative colitis. Interestingly, the lymphoma risk appeared within the first five years of Crohn’s disease. Most of the observation period was before the era of infliximab treatment and immunosuppression with azathioprine but a limitation of the study was the lack of information about pharmacotherapy in the patients. None the less, the study is important and calls for close follow up of future patients with IBD, especially those receiving new biological treatment modalities. Such a study, also from Sweden, was published very recently by Ljung and colleagues.² IBD patients from Stockholm county were all treated with infliximab within a two year period and severe adverse events were noted in 19% of patients and a 1.5% annual incidence of lymphoma. This study however was hampered by not having any controls.

The tool for monitoring new treatment modalities in relation to already established management of patients will be ongoing previous and future studies based on lifelong close follow up of patients in outpatient clinics. Such studies reveal a detailed picture of the disease course in each patient and in subgroups of patients, classified according to clinical characteristics and treatment. There is of course much more that characterises and influences the long term course of inflammatory bowel disease than just the complication lymphoma. Studies from Denmark in ulcerative colitis, carried out over 3–4 decades,^3,4 have revealed survival not different from that of the background population from one year after diagnosis to 36 years. Further analysis showed that the subgroup of patients with extensive colitis and late onset of disease (>50 years), who demanded surgical intervention within the first year of their disease, were at a higher risk of dying from postoperative complications or comorbidity. Leukaemia and lymphoma were not found more frequently than expected in the population in general. This cohort of patients was treated in the pre-immunosuppressive era and had only short courses of corticosteroids at flare ups.

In Crohn’s disease, slightly increased mortality was observed in women with Crohn’s disease late in the disease course, 20–25 years after diagnosis, associated with severe inflammatory bowel disease.^5,6 No lymphomas were found, and in general the risk of extraintestinal cancer was not different from those without Crohn’s disease. Cohort studies with continuous clinical follow up over long periods of time have been published in Sweden,^7,8 Italy,⁹ and the USA,¹⁰ all necessary background for further studies which need to be carried out as more aggressive and more effective medical treatment modalities become more common. There is no evidence that inflammatory bowel disease per se carries an increased risk of lymphoma.

Electronic patient records will apparently be generally available within the coming years, at least for outpatients, but soon for all patient care, in order to secure the quality of treatment. It must be borne in mind that information technology is only a tool, especially for administrators, and it can never replace the knowledge and expertise of clinical epidemiology and good clinical practice. It is our responsibility that these skills are implemented in the IT future of medicine.