Обзор американских кардиологических журналов за неделю (ссылки)

[Chevychelov]

New Tool Predicts VTE Risk

By Todd Neale, Senior Staff Writer, MedPage Today
August 17, 2011

Explain that a risk prediction model for venous thromboembolism (VTE) predicted events at one and five years in both prediction and validation cohorts.

The risk prediction model included information readily available from patients or general practice records.
Review
A new risk prediction model may help to estimate an individual patient's risk of venous thromboembolism up to five years out, researchers found.

The model, which includes 11 variables for men and 14 variables for women, yielded an area under the receiver operating characteristic curve (AUC) of 0.75, according to Julia Hippisley-Cox, MD, and Carol Coupland, PhD, of the University of Nottingham.

And comparing the predicted versus the observed risks at one and five years indicated that the algorithm was well calibrated, they reported online in BMJ.

"The algorithm is based on simple clinical variables which the patient is likely to know or which are routinely recorded in general practice records," they wrote.

"The algorithm could be integrated into general practice clinical computer systems and used to risk assess patients before hospital admission or starting medication which might increase the risk of venous thromboembolism," including oral contraceptives, antipsychotics, and hormone replacement therapy, they wrote.

The researchers added that the tool, which is called Qthrombosis and is available online, is not meant to assess the current risk of venous thromboembolism in symptomatic patients.

Hippisley-Cox and Coupland developed the model by examining data from patients treated at 564 general practices in England and Wales that participated in the QResearch database. None of the patients had been pregnant in the preceding 12 months, had a previous deep vein thrombosis or pulmonary embolism, or were taking oral anticoagulants at baseline.

The cohort used to derive the risk prediction model included about 2.3 million patients. During the study, there were 14,756 incident cases of venous thromboembolism -- a rate of 14.6 per 10,000 person-years.

The validation cohort included about 1.2 million people. In this group, there were 6,913 incident cases -- a rate of 14.9 per 10,000 person-years.

For both sexes, independent predictors of venous thromboembolism that were included in the risk prediction model were age, body mass index, smoking status, varicose veins, congestive heart failure, chronic renal disease, cancer, chronic obstructive pulmonary disease, inflammatory bowel disease, hospital admission in the past six months, and current prescriptions for antipsychotic drugs.

For women, additional factors were use of oral contraceptives, tamoxifen, and hormone replacement therapy.

In the validation cohort, the risk prediction equation explained 33% and 34% of the variation in the time to venous thromboembolism in women and men, respectively.

The D statistic was 1.43 for women and 1.45 for men at five years, with an AUC of 0.75 for both sexes.

Good calibration of the model was confirmed by comparing the mean predicted risks at one and five years with the observed risks. For example, for women in the top 10% of predicted risk, the mean predicted five-year risk was 2.78% and the observed risk was 2.7%. For men, the corresponding values were 2.46% and 2.35%.

The authors acknowledged some limitations of their analysis, including the lack of formally adjudicated outcomes, a reliance on information provided by the primary care physicians, a lack of genetic data, the potential for missing information, and residual confounding.