Recent Advances in Thrombosis and Hemostasis—Part IX
Sam Schulman
DOI: https://doi.org/10.1055/s-0043-1772839
2023-08-29
Seminars in Thrombosis and Hemostasis
Abstract:It is difficult to make predictions, especially about the future.— Attributed to many: Niels Bohr, quantum physicist, 1885–1962 Yogi Berra, baseball player and philosopher, 1925–2015 Questions are often raised whether one should use the term "risk factor" or "predictor," and sometimes they are used interchangeably. The general opinion is that "risk factor" is a broader term and poorly defined, to the extent that many epidemiologists recommend avoiding it.[1] In this compilation issue, we have three contributions that use the term "risk factor(s)," one that uses "predictor," and another one with the similar term "predictive score." If we had an epidemiologist as the Guest Editor for this issue, maybe the authors of the manuscripts using "risk factor" would have been criticized. I have restricted my general critique regarding the statistics and epidemiology to some better-defined areas. First, we see many submissions with p -values and proportion numbers containing too many active digits. When the denominator, for example, is less than 100, there is no point providing a proportion of 59.9% or a p -value of 0.0376. That will give a false impression of very high precision, which is not possible with such a small denominator. The results have to be rounded off. Some journals will not accept more than one active digit for p -values that are nonsignificant; for example, p = 0.44 must be rounded off to 0.4, since the result will not be more nonsignificant by adding digits. Second, distributions of data need to be checked for normality, so that skewed distributions are reported with median and interquartile range (or range) and analyzed with the appropriate test. Third, the term "multivariate" is often wrongly used. When we are analyzing different variables and testing which one(s) might be independent for an outcome, we should use "multivariable." On the other hand, "multivariate" is used when we analyze more than one outcome. Let us now take a look at the contributions to this issue of Seminars in Thrombosis and Hemostasis , starting with some studies focusing on predictors. In patients with dilated cardiomyopathy, there is reduced wall motion and thereby a tendency to blood stasis in the left ventricle. A substantial minority of those patients develop a mural thrombus, with subsequent risk for embolization and stroke. Wu and colleagues reviewed retrospectively 3,134 patients with dilated cardiomyopathy and found that among 15 different variables, elevated D-dimer and reduced ejection fraction appeared to be independent risk factors for thrombus formation.[2] It is by now well known that patients with arterial thromboembolic disorders have an increased risk of venous thromboembolism (VTE)—and vice versa. Hu and colleagues have here investigated which cardiovascular risk factors are associated with increased or decreased risk of VTE in general, deep vein thrombosis (DVT), and pulmonary embolism.[3] They had access to genome-wide association study data linked to different risk factors from several biobanks and utilized these data for Mendelian randomization analysis. They found that body mass index (BMI) was associated with all three venous outcomes, whereas physical activity or higher systolic blood pressure might be protective, although differently for VTE in general, DVT, and pulmonary embolism. To continue on the same theme, but in the opposite direction, Noumegni and colleagues used data from Program d'Etude des Déterminants et Interactions de la THrombose veineuse (EDITH) cohort study on patients with VTE to identify predictors for recurrent thrombotic events on both the arterial and the venous side.[4] They analyzed more than 30 variables, when available, from 2,011 patients. Of those, 801 had follow-up for at least 3 months after stopping anticoagulation, and the overall median follow-up was 92 months. The strongest independent predictor for VTE or arterial events during anticoagulant treatment was cancer-associated VTE, followed by unprovoked VTE. However, looking at VTE and arterial events separately, the predictors differed between them. For the study period after discontinuation of anticoagulation, the strongest independent predictor for VTE or arterial events was increasing age, followed by unprovoked VTE. Again, there were differences regarding individual predictors for venous versus arterial events. Building on observation that mitochondrial dysfunction might be implicated in the pathogenesis of thrombosis, the mitochondrial gene regulators are of certain interest. The long noncoding 7S RNA has been identified as such a regulator in vitro as well as in cultured human cells, and therefore Wang and colleagues measured 7S RNA in plasma from 53 patients with and 184 without DVT from a previous study on diagnostic strategy.[5] There was an i -Abstract Truncated-
peripheral vascular disease,hematology