Background: Levocarnitine deficiency in hemodialysis patients is common. Although the effect of intravenous levocarnitine therapy was studied in small trials, the effect on global outcomes in larger populations is unclear.
Study design: Retrospective observational study.
Setting & participants: Centers for Medicare & Medicaid Services data; prevalent hemodialysis patients, 1998 to 2003.
Predictor: Intravenous levocarnitine use, clinical characteristics, comorbid conditions.
Outcomes & measurements: Effect of 1 g or greater per dialysis session of levocarnitine for 10 or more sessions during a month on subsequent hospitalization days. Repeated-measures and marginal structural models were fit, the latter to account for time-dependent confounding.
Results: Of the study population, 3% to 7% received levocarnitine for 1 month per year or more. Treated patients were older with more severe comorbidity and larger erythropoietin doses than untreated patients. In repeated-measures model analysis adjusted for demographic characteristics and disease severity, 1 g or greater per dialysis session of levocarnitine for 10 or more sessions during a month was associated with a 10.8% (95% confidence interval, 9.7 to 11.9; P < 0.01) subsequent-month decrease in hospitalization days. In marginal structural model analysis, levocarnitine therapy was associated with a 21.7% (95% confidence interval, 18.4 to 24.9; P < 0.01) decrease in hospitalization days.
Limitations: Algorithm for identifying comorbid conditions from claims validated only for diabetes; biochemical marker levels unavailable in Medicare claims; levocarnitine therapy quantified only while patients were not hospitalized.
Conclusion: Because hemodialysis patients are hospitalized about 15 days yearly, the association of monthly levocarnitine regimen with lower hospitalization rate is clinically significant. The causality of this association must be confirmed by randomized clinical trials.