Pulmonary Embolism in Sickle Cell Disease: A Case-Control Study.

Pulmonary embolism (PE) is a leading cause of morbidity and mortality in the U.S. We and others previously have shown that PE is increasingly detected with increasing use of spiral computerized tomography (CT), at lower severity of illness, suggesting earlier PE diagnosis. Although thrombosis is increased in those with sickle cell trait (Austin, Blood 2007), data are lacking on the prevalence and predictors of PE among individuals with sickle cell disease (SCD), in whom thrombosis may be driven by hemolysis-associated endothelial dysfunction and platelet aggregation.

We performed a case-control study to compare demographics, co-morbidity, severity of illness, and mortality in SCD cases with PE and SCD controls without PE, by analyzing statewide discharge data (2001-2006) from the Pennsylvania Health Care Cost Containment Council (PHC4). For this analysis “SCD” included ICD-9 codes for HbSS, HbS/C, HbS/thal, and HbS trait. Clinical and laboratory data were obtained on local SCD cases and controls from the Medical Archival Record System (MARS) at the University of Pittsburgh, using the same ICD-9 codes. Cases were matched by age within 5 years, race, gender, year of admission, and (for PHC4 data only) hospital system.

The prevalence of PE among SCD admissions, 2001-2006, was 119/ 20,847 (0.57%), increasing over the period by 41.2%, from 0.51% to 0.72%. By comparison, among non-SCD admissions, the prevalence of PE was 66,440/1,416,109 (4.7%), with a similar rate of increase, 54.0%, from 3.78% to 5.82% over the same period. As compared to controls without PE, SCD PE cases had a longer length of stay, 9 vs. 5 days (p=0.0006), greater severity of illness, two highest scores 11.1% vs. 0%, (p=0.0002), and a higher in-hospital mortality rate, 8.7% vs. 0% (p=0.0011). In the local sample (n=14 cases, n=28 controls), the majority of cases were evaluated by spiral CT scans (92.8%). Co-morbidity rates were no greater in cases than controls, including pneumonia (p=0.227), heart failure (p=0.461), coronary symptoms (p=0.464), or stroke (p=0.183). Risk factors for thrombosis were also similar between groups, including estrogen use (p=0.461), obesity (p= 0.763), hyperlipidemia (p=0.106), hypertension (p=0.146), diabetes (p=0.276), smoking (p=0.069), HIV (p=1.000), and HCV (p=0.667). A similar proportion of cases and controls were in crisis, 35.7% each (p=0.888), and there were no differences in the proportion with HbSS, 35.7% vs. 46.4% (p=0.212), HbS/C, 14.3% vs. 0% (p=0.106), HbS/thal, 0% vs. 3.6% (p=0.667), and sickle trait, 50.0% each (p=1.000). Among cases, the degree of anemia (p=0.276), reticulocytosis (p=0.261), WBC (p=0.257) and platelet count (p=0.254) were similar to that of controls. There was no difference between groups in the proportion receiving hydroxyurea (p=0.167), iron chelation therapy (p=1.000), or red cell transfusions (p=0.262).

Rates of PE are increasing in hospitalized SCD patients in Pennsylvania, with greater severity of illness and higher mortality than in SCD patients without PE. The clinical and laboratory parameters, including severity of hemolysis measured, do not appear to be predictive of PE. Thus, prospective studies are needed to evaluate risk for PE by other markers of disease severity, including tests of platelet activation, tissue factor activity, and hemostatic activation, e.g. thrombospondin, von Willebrand factor, and ADAMTS13.

No relevant conflicts of interest to declare.