Mortality due to pulmonary embolism, myocardial infarction, and stroke among incident dialysis patients: a rebuttal.

This letter discusses the the paper by Ocak et al. The authors found that the age- and sex-standardized mortality rate (SMR) for pulmonary embolism (PE) was 12.2 times higher in dialysis patients than in the general population. Such SMR was even higher than that of myocardial infarction (MI) (11.0), stroke (8.4), and other cardiovascular disease (8.3). The association between the different grade of impaired renal function and MI received important confirmation, and a recent study from our group conducted on a large regional database of hospital admissions found that mortality rates during hospitalization in patients admitted with MI was significantly higher in patients with end-stage renal disease (ESRD) and chronic kidney disease (CKD) compared with patients without renal dysfunction. On the other hand, the mortality risk for PE in CKD and dialysis patients is not so firmly established. Epidemiological data in this population of patients are very limited, are often based on different codification systems, and originate from different populations (e.g. outpatients or inpatients). We evaluated the effect of renal dysfunction on in-hospital mortality for PE, considering 24,690 cases of PE recorded in a large regional database from 1999 to 2009. In-hospital mortality for PE was not different in patients without renal dysfunction, with CKD, or ESRD, and was independently associated with age (OR 1.045), female sex (OR 1.322), hypertension (OR 1.096), diabetes mellitus (OR 1.120), dementia (OR 1.171), peripheral vascular disease (OR 1.349) and malignancy (OR 1.065). In patients hospitalized for PE, age and comorbidity were associated with in-hospital mortality for PE, whereas CKD did not appear to be an independent predictor of adverse outcomes. Therefore, differently from MI patients, we failed to detect significant differences in the prevalence of PE-related death in ESRD and CKD patients compared with patients without renal dysfunction (24.0% vs. 23.6% vs. 18%, respectively). There are several differences between the study by Okac et al and our study that may explain the different results. First, the choice of identification codes. Death attributable to PE was defined by ERA-EDTA code 21 in the study by Okac et al, and we utilized the ICD-9-CM code 415.1 to identify all subjects with emergency hospital admissions for PE. Second, in the study by Okac et al older age at the beginning of renal replacement therapy (equal or greater than 85 years) was associated with an increased risk of PE. However, the authors did not include comorbidities in the analysis, and it is likely that these were more prevalent in dialysis patients. In agreement with Okac et al, we found that age, female sex, diabetes mellitus (and also hypertension, dementia, peripheral vascular disease and malignancy) were independently associated with mortality. Of course, a pivotal important difference is that the study by Ocak et al studied age- and sex-mortality in registry ambulatory patients, whereas we focused on in-hospital PE-related death. This may represent a major issue, given that the diagnosis of PE during hemodialysis is difficult, and PE ranks second position, after coronary heart disease (7.5% of cases), among necropsy-confirmed causes of out-of-hospital sudden deaths in the general population. Although a reduced renal function undoubtedly represents an important risk factor for increased cardiovascular mortality, we believe that the evaluation of PE mortality in CKD and dialysis patients should consider which type of population is enrolled in the different studies, the evaluation of in-hospital subjects underlines the importance of comorbidities.