On Statin Treatment to Prevent Sepsis in Dialysis Patients

What Did This Important Study Show? 

From October 1995 to June 1998, 1,041 patients who had initiated outpatient hemodialysis or peritoneal dialysis in the previous 3 months at 81 US not-for-profit outpatient dialysis clinics were enrolled in a prospective cohort study, with follow-up until January 2005. Statin use was determined by medical record review. To assess for an association between statin use and sepsis, the investigators performed 2 analyses: (1) a multivariable Poisson regression analysis on the entire cohort, comparing occurrence of sepsis in statin users versus nonusers, and (2) an analysis comparing occurrence of sepsis in patients who used statins (n = 107) versus a control group of patients who did not use statins (n = 107) selected based on their propensity to have been prescribed a statin. The primary endpoint, hospitalization for sepsis, was determined from US Renal Data System hospital records using International Statistical Classification of Diseases and Related Health Problems, Ninth Revision (ICD-9) codes for septicemia (038.0 to 038.9) and bacteremia (790.7). At baseline, 14% of patients (n = 143) were taking statins. During a mean follow-up period of 3.4 years, there were 303 hospitalizations for sepsis in 165 patients. Statin users were 63% less likely to be hospitalized for sepsis than were nonusers (crude sepsis incidence rates 41/1,000 patient-years v 110/1,000 patient-years, respectively; hazard ratio 0.41; 95% Confidence Interval (CI), 0.25-0.68; P < 0.001; the number of dialysis patients needed to treat with statins to prevent 1 episode of sepsis is approximately equal to 14). After adjustment for demographic characteristics, dialysis modality, comorbid conditions, and laboratory values, the risk of hospitalization for sepsis was 62% lower in statin users. When patients with prior episodes of sepsis were excluded from the multivariable analyses, statin users were still 60% less likely to be hospitalized for sepsis. Statin use appeared more protective in the subanalyses that utilized propensity score matching (incidence rate ratio, 0.24; 95% CI, 0.11-0.49). The authors concluded that use of statins was strongly and independently associated with a reduction in the risk of hospitalization for sepsis. They stated “[t]o our knowledge, this is the first study to show a strong and significant effect of a medication administered long term on lower rates of sepsis among patients with chronic kidney disease.”

As pointed out by the authors the study has important limitations. It was not a randomized trial, so causality cannot be proved. Observational studies may not be able to control adequately for confounding by indication for treatment. Additionally, patient and treatment factors were assessed at baseline and changes over time were not accounted for. Importantly, there was no information regarding statin discontinuation or initiation, treatment cessation, and new uptake. Outcome data relied on hospital bills and accurate coding of administrative data. Previously, such billing data have been demonstrated to be 75% to 88% sensitive in diagnosing sepsis in comparison to a review of the medical record.5

How Does This Study Compare With Prior Studies? 

Studies in animals suggest that statins might prevent sepsis and 2 smaller observational reports provide evidence to support this notion.6 A recent, large, population-based retrospective cohort study by Hackam et al7 found a reduced incidence of sepsis among patients who were prescribed a statin after hospitalization for a cardiovascular event (0.71 v 0.88 per 100 patient-years in users and nonusers, respectively [hazard ratio 0.81; 95% CI, 0.72-0.91]); benefits persisted in the subgroup of individuals in this study with chronic kidney disease as defined by ICD-9 codes. The incidence rates were not given for the 7,169 chronic kidney disease patients,7 but it should be noted that the incidence rates of sepsis were much higher in the study by Gupta et al than in the study by Hackam et al. Overall, it is difficult to compare results of statin use in these 2 cohorts.

In a secondary analysis, we investigated the effect of statins on fatal infections in the 4D (Die Deutsche Diabetes Dialyse) study. In this study, 1,255 hemodialysis patients with type 2 diabetes were randomly assigned to receive atorvastatin 20 mg/d or placebo. During a mean follow-up period of 4 years, 37% of patients reached the primary endpoint (death from cardiac causes), with no significant difference between atorvastatin and placebo (relative risk, 0.92; 95% CI, 0.77-1.10; P = 0.4).8 Three independent clinicians from the endpoint committee adjudicated the diagnosis of sepsis on the basis of prospectively collected and extensively documented material and hospital records. During the trial, 128 patients died from infection during 4 years of follow-up (22% of all deaths in the 4D study; Fig 1), with no significant difference in the number of deaths in the atorvastatin and placebo groups (68 v 60 cases, respectively). We did not analyze the causes leading to hospital admission in the 4D study but the number of adverse events requiring hospitalization (949 and 942 events, respectively) was similar, as was the number of hospital days in the atorvastatin and placebo groups (2,392 and 2,336 days, respectively). Although secondary analyses from randomized controlled trials do not have the same strength of evidence as the primary analyses, these results do not support the hypothesis that statins reduce the risk of sepsis in hemodialysis patients.

View Large Image Download to PowerPoint

Figure 1. Causes of death in the 4D (Die Deutsche Diabetes Dialyse) study. Abbreviations: MI, myocardial infarction; CHF, congestive heart failure; CHD, coronary heart disease.

What Should Clinicians and Researchers Do? 

The findings of Gupta et al are potentially important and may have many implications, but should we change our clinical care? The observed effect size was impressive, and although subject to confounding, it was large enough to be interpreted as causal by many physicians and researchers. My conclusion is that the strength of evidence is not yet sufficient to warrant a change in practice. Several limitations of the study by Gupta et al are important to consider.

First, a prospective cohort study of dialysis patients is prone to multiple methodological problems, including unquantifiable bias of indeterminate direction. While Gupta et al adjusted for many possible confounding factors, adjustment can only be made for those factors that are observed. Thus, it is difficult to assess the extent of bias. The same limitation also applies to an analysis of the US Renal Data System’s Dialysis Morbidity and Mortality Wave 2 study of the association of statin use with mortality in patients treated by peritoneal dialysis by Goldfarb-Rumyantzev et al in this issue of AJKD.9

Second, there is uncertainty in the definition and meaning of “bacteremia,” “sepsis,” or “systemic inflammatory response syndrome” ascertained from administrative data. Less than 20% of physicians seem to agree on any one definition of sepsis.10 On the other hand, vagueness in defining the disease (bacteremia and septicemia) would tend to bias the analysis towards the null and would yield underestimates of effect size.

Third, the authors stressed the potential pleiotropic clinical effects of statins beyond their effects on lipids. The evidence for cholesterol-independent effects has accumulated in the literature and is reflected in studies demonstrating statin-associated reductions in the risk of dementia, Alzheimer disease, ischemic stroke, elevated blood pressure, osteoporosis, breast and lung cancer, and diabetic foot ulceration.11, 12, 13, 14 Many clinicians, especially lipidologists, find it difficult to embrace the concept of statin pleiotropy for a number of reasons.11 While a recent review titled “Ongoing clinical trials of the pleiotropic effects of statins” lists the 4D study among the potential trials,15 no pleiotropic effects of atorvastatin were noted in the 4D study, including no effect to decrease the serum level of C-reactive protein (CRP; unpublished data). Interestingly, the CHOICE study had similar baseline levels of albumin, CRP, and IL-6 in statin users and controls, also arguing against effects on inflammation.

Can statins really be linked to lower incidence of sepsis in dialysis patients? Large placebo-controlled blinded randomized trials in dialysis patients are needed to further investigate this potentially exciting preventive approach.