Diuretics and mortality reduction in dialysis patients: a two-year observational study

We assessed the impact of exposure to loop diuretics on survival, specific causes of death, and hospitalizations in the 2 years after onset by comparing data from the nationwide REIN registry with the SNDS administrative database.

Overall, diuretic exposure was associated with a significantly lower mortality rate after two years. However, the duration of loop diuretic exposure had a significant impact on outcomes in dialysis patients. Those with longer exposure (>80%) had a significantly higher risk of death from all causes both one and two years after onset. This effect was linear in time and quantitatively significant, with a mortality rate almost three times higher than that of the average dialysis patient. Importantly, these results were consistent across both competing risk models and IPW propensity score analyses. In addition, prolonged exposure to loop diuretics was associated with increased hospitalizations for heart failure and other causes, an effect observed in both hemodialysis and peritoneal dialysis patients.

In the DOPPS study, Bragg-Gresham et al. examined the association between self-reported diuretic consumption and clinical outcomes in 16,420 hemodialysis patients on three continents⁶. Diuretic use was associated with lower odds of interdialytic weight gain and hyperkalemia, higher odds of receiving RKF at 1 year, and lower cardiac mortality, with a trend toward lower all-cause mortality. The use of diuretics and the presence of RKF were correlated with 32.7% of patients with RKF using diuretics versus 18.8% of patients without RKF. In a stratified analysis, the association between diuretic use and all-cause mortality was similar in patients with and without RKF, whereas diuretic use was associated with lower cardiac mortality in patients without RKF. These results are somewhat surprising, as another study in patients with anuric Huntington's disease showed that neither low nor high doses of furosemide had any significant effects on central cardiac hemodynamics¹¹. Furthermore, these results could be confounded by indication bias, as patients receiving diuretics were more likely to have residual renal function¹².

Sibbel et al. examined the effects of diuretic continuation or discontinuation on survival, hospitalizations, and intradialytic hypotension in the first year of dialysis. These authors found that continued diuretic therapy after one year resulted in a significant reduction in hospitalizations and intradialytic hypotension, but not in deaths⁷. A limitation in this study is that patients with RKF were more likely to be prescribed diuretics. To address confounding by indications, the authors repeated their analyzes only in the subgroup of individuals with RKF and found similar associations to those in the primary analysis. Residual confounding was also possible in these analyses, as users of loop diuretics differed from nonusers. In addition, diuretic users were more likely to receive nephrology care before starting dialysis, which may have resulted in different decisions regarding dialysis initiation, peridialysis management, and long-term outcomes¹³.

Our study found a lower 2-year mortality rate associated with exposure to loop diuretics, consistent with previous findings. What is particularly interesting, however, is the connection between the exposure pattern and the results. We observed that patients with longer diuretic exposure had significantly higher mortality rates, higher numbers of hospitalizations, and longer cumulative hospitalization times compared to patients with less diuretic exposure. This discrepancy may be due to indication bias. We anticipate that treatment with loop diuretics will be continued in patients with persistent fluid overload or in centers aiming to preserve residual renal function despite chronic volume expansion. This hypothesis is supported by the fact that patients with exposure ≥80% had a higher history of coronary artery disease, heart failure, and arrhythmias. This may also explain the high proportion of deaths from cardiovascular and fluid overload causes (e.g., pleural effusion, pulmonary disease) in this group. Importantly, the increased mortality was not limited to patients with heart failure and was observed regardless of a history of heart failure. Alternatively, high diuretic doses could have been maintained in patients with unstable peridialytic hemodynamics in whom high ultrafiltration rates were not possible.

Finally, we found similar results regarding diuretic exposure in both hemodialysis and peritoneal dialysis patients. Within each diuretic exposure group, mortality rates were slightly lower in peritoneal dialysis patients, which may be explained by competing risks and a higher likelihood of transplantation in these patients.

Limitations and strengths of the study

Our study has several limitations that should be taken into account. As an observational study, it is difficult to conclude causal relationships. There may have been biases in the indications for loop diuretics that could lead to some paradoxical results. In particular, it may be difficult to distinguish between diuretic indications in fluid overload and heart failure with preserved ejection fraction. However, our stratified analysis based on a history of heart disease showed similar results.

For matching, we used an “iterative deterministic” method that allows strong linkage between two databases without directly identifying patients. The validity of this matching method has been demonstrated specifically for these two databases⁸. Some administrative coding for hospital admissions may be subject to question. For example, in our study, some hospitalizations of dialysis patients were simply coded as “CKD without dialysis,” making it impossible to determine the actual reason for the hospitalization.

Furthermore, our study did not consider individual diuretic doses, although nearly 80% of patients taking diuretics were prescribed the 500 mg dose of furosemide. We also did not take concomitant therapies such as ACE inhibitors or beta blockers into account, which could have better identified the patient population with heart failure. However, these drugs are also commonly used as antihypertensive drugs, and it is uncertain whether the distinction between these groups could be made on pharmacological grounds.

Finally, we had no information on residual diuresis and interdialytic weight gain. However, we can reasonably assume that patients with >80% loop diuretic exposure had residual diuresis over the 2-year follow-up period, while patients with >80% exposure to loop diuretics had residual diuresis

Despite these limitations, our study has several strengths. First, it was based on a comprehensive contemporary cohort from the nationwide REIN registry in which structured data were collected prospectively at baseline and during dialysis follow-up. We made adjustments to avoid confounding or indication bias by stratifying by history of heart failure and dialysis modality and confirmed our results using IPW propensity score analysis and competing risk models. Finally, our study is the only large study reporting the use of diuretics and their effects in peritoneal dialysis patients.