Prospective community screening for aortic conditions--true incidence or just a better guess?

Introduction

For the clinical epidemiologist with an interest in dissection and aneurysm of the aorta, a well‐executed, general population screening study – combined with accurate risk factor analysis – is a dream come true; the paper by Smith and colleagues reported today is such a study. Community screening of 30 412 individuals participating in the Malmö Diet and Cancer survey produced reliable measures of the best available (true) event rate of treated aortic disease in a western population. The reported rates of 15/100 000 patient‐years for aortic dissection, 9 for thoracic aortic aneurysm, and 27 for abdominal aortic aneurysm (supported by over 20 years of follow‐up)[1] is important new data which has been desperately waited for. In addition, results are reported in a meaningful way; the confusing combination of community and hospital‐based numbers often seen in the aortic literature now appear to suggest incidences of less than one‐third of the real figures.[2-5] Thus, a great deal of nebulous assumptions are being clarified.

The authors are to be congratulated for making the most of a solid Swedish database of community morbidity/mortality reporting to derive meaningful information on high‐impact but low‐incidence disease conditions such as aortic dissection and aneurysm; previous reports from Sweden have already heralded an increasing incidence of chronic and acute aortic conditions.[6] The only comparable effort has been reported for Oxfordshire in the OXVASC project with similarly realistic figures way beyond figures reported by IRAD.[7]

Moreover, the Scandinavian report confirms a sobering high in‐hospital mortality of 39% for dissection of the aorta, and 34% and 41% for ruptured abdominal and thoracic aortic aneurysm, respectively. The usual risk factor suspects are confirmed in the paper – most obviously untreated hypertension in 86% of subsequent dissections smoking, along with ApoB/ApoA1 ratio for abdominal aneurysm; smoking was also confirmed as a risk factor for both dissection and thoracic aneurysm.

It could be argued that the risk factor findings in this Swedish population‐based cohort analysis simply confirm existing knowledge. We believe this would be an underestimate as the most convincing elements of this paper are the solid database and the undisputed findings. Another valuable asset is that the risk factor analysis stems from a population‐based rather than hospital‐based cohort of middle‐aged men and women with long follow‐up – allowing for true estimates of incidence and attributable risks. Though the general information is not new or disruptive, in the era of personalized medicine, the results surely justify further exploration of these risk factors. It is also interesting to note, that there was no reverse association between diabetes and the risk of aortic dissection; in other words, the solid database of the Swedish obligatory reporting system failed to show any “protective effect” of diabetes[8-9] on the occurrence of aortic dissection, and helps clarify controversial previous suggestions.[10-11] If diabetes played any significant pathophysiological role in aortic dissection this should have surfaced in a population‐based sample, whereas in a hospital‐based sample diabetes is more likely to be overrepresented.[10,12]

A weakness of the data is that the incidence rates for AAA and TAA are accurate for intervention only; asymptomatic patients were not screened for either condition. This means that the incidences of these two diseases are likely higher than reported – although we do not know what diameter thresholds were used for treatment.

Thus, properly followed up, population‐based data on the incidence of a wide variety of medical conditions, and observed morbidity and mortality rates are certainly highly valuable and may serve as a better benchmark of disease impact than the traditional, hospital‐based assumptions we are all used to seeing. Perhaps eventually we will gather and use such data on all conditions in teaching texts. Against this, such data are only available in selected countries that provide and allow access to unselected population‐based medical information under specific medico‐juridical context, in an appropriate national legal framework. Most industrialized countries in Europe and the United States do not provide such data – documentation and reporting is usually voluntary and subject to data security and privacy protection. Finally, it needs to be shown whether the population‐based statistics from Sweden can be extrapolated to other societies and ethnic groups in a reliable way.