Is Female Sex Really a Risk Factor for Infectious Death in Peritoneal Dialysis?

Infection is a major cause of morbidity and technique failure in peritoneal dialysis (PD) patients (1). It is also an important cause of mortality (1,2). Compared with hemodialysis (HD), PD is associated with a significantly increased risk of infectious death, primarily because of an increased risk of death from peritonitis (3).

Since the early 2000s, a large number of risk factors for peritonitis have been identified (4-8). Such risk factors can be divided into

• patient factors: age, racial minorities, obesity, protein-energy wasting, smoking, diabetes mellitus, chronic lung disease, coronary artery disease, congestive heart failure, hypertension, immunosuppression, non-use of vitamin D supplements, lupus nephritis, lower socio-economic and education status, remote residence, prior HD, enforced PD against patient preference;

• health care-related factors: unit size, after-hours staffing, staff experience; and

• environmental factors: climate and season.

Although only some of the factors are modifiable, identifying patients at heightened risk of PD-associated peritonitis and infectious death is crucial to developing strategies for improving outcomes in PD patients.

In this issue of Peritoneal Dialysis International, Ros et al. (9) use data from all 12 PD units in Andalusia, Spain, (SICATA registry) to report a retrospective observational cohort analysis of 1458 PD patients more than 13 years of age (57% men) who commenced PD as their first renal replacement therapy between 1999 and 2010 and who were followed for a maximum of 5 years. The key finding of the study is that, compared with male PD patients, female PD patients experienced a significantly higher infection-related mortality rate [3.45 vs 1.67 deaths per 100 patient-years; unadjusted hazard ratio (HR): 1.92; 95% confidence interval (CI): 1.15 to 3.20; p = 0.01), which persisted after adjustment for age, primary renal disease, dialysis commencement era, and initial PD modality (adjusted HR: 1.76; 95% CI: 1.03 to 3.01). A sensitivity analysis considering the competing risks of renal transplantation and HD transfer yielded similar results (adjusted HR: 1.82; 95% CI: 1.08 to 3.06).

Analysis of the specific causes of infection-related mortality demonstrated significantly increased risks of both sepsis and peritonitis in female PD patients. Despite the increased occurrence of fatal infection in female compared with male PD patients, no significant differences were observed between the groups with respect to overall mortality (adjusted HR: 0.91; 95% CI: 0.72 to 1.15; p = 0.46), cardiovascular mortality (adjusted HR: 0.76; 95% CI: 0.52 to 1.09), or non-cardiovascular mortality (adjusted HR: 0.79; 95% CI: 0.54 to 1.16; p = 0.24). The investigators speculate that the heightened risk of infection-related mortality in female PD patients may be related to either ascending gram-negative infections from the female genitourinary tract or impaired immune response secondary to uremic hypogonadism. They further conclude that “targeted sex-specific care is likely necessary” to mitigate the increased risk of fatal infections in female PD patients.

The study’s main strengths are its large sample size, use of competing-risks analyses, and inclusion of data from 12 PD centers; its key limitations are its relatively small number of infection-related death events (41 in women, 24 in men), limited multivariable adjustment (which did not include comorbidity), exclusion of patients who experienced renal recovery and who had already undergone HD or renal transplantation, restriction of follow-up to a maximum period of 5 years, and lack of detailed data regarding the microbiology and treatment of PD infections.

The findings of Ros et al. are supported by a limited number of earlier, predominantly single-center, studies. Pérez Fontan et al. (10) observed that female PD patients in a Spanish center had a higher risk of peritonitis-related mortality (relative risk: 2.13; 95% CI: 1.24 to 4.09; p = 0.02), and an Australian center (11) reported an increased risk of PD-associated peritonitis in women (odds ratio: 1.91; 95% CI: 1.20 to 3.01), with many of the episodes seeming to be related to gram-negative bacteria. Another center in Turkey (12) found a significantly higher proportion of women among the PD patients experiencing multiple peritonitis episodes, albeit on univariable analysis only. An analysis of the Baxter POET database from 25 Canadian PD centers found a significant interaction between sex, diabetes mellitus, and peritonitis—whereby, after multivariable adjustment, female PD patients with diabetes experienced a 27% increased risk of peritonitis (13). Furthermore, in the US Renal Data System (USRDS) Annual Data Report for 2012, female compared with male patients experienced similar or higher unadjusted infection-related mortality rates for septicemia (1.51 vs 1.27 deaths per 100 patient-years), pulmonary infection (0.23 vs 0.3 deaths per 100 patient-years), and other infections (0.61 vs 0.45 deaths per 100 patient-years).

On the other hand, a larger number of studies have not identified an association between female sex and peritonitis (4,5,14-24) or infection-related mortality (3), including several large registry studies from the United States (14), Brazil (15), and Australia and New Zealand (3-5,20-24). Moreover, detailed analyses of organism-specific peritonitis have not identified any associations between sex and either the risks or outcomes of peritonitis related to Streptococcus (25), Staphylococcus aureus (26), coagulase-negative Staphylococcus (27), Enterococcus (28), Corynebacterium (29), culture-negative episodes (30), Pseudomonas (31), non-pseudomonal gram-negative organisms (32), polymicrobial episodes (33), or fungi (34).

Because some of the apparent disparity in findings between those studies and that of Ros et al. (9) may relate to the latter’s restrictive inclusion criteria and limited multivariable adjustment, we replicated their analyses in 2251 de novo incident PD patients more than 13 years of age who commenced PD in Australia and New Zealand between 2007 and 2010. Female sex was not associated with infection-related mortality on either univariable analysis (HR: 0.99; 95% CI: 0.52 to 1.87; p = 0.97; Figure 1) or after adjustment for age, primary renal disease, dialysis commencement era, and initial PD modality (adjusted HR: 1.10; 95% CI: 0.58 to 2.10). Similar findings were observed in a competing-risks analysis (adjusted HR: 0.98; 95% CI: 0.52 to 1.86). Interestingly, in keeping with the findings of Ros et al. (9), we observed significant differences between male and female PD patients with respect to their primary renal disease, such that diabetic nephropathy and renovascular or hypertensive nephropathy were significantly more common in men. More importantly, diabetes mellitus, coronary artery disease, cerebrovascular disease, and peripheral vascular disease were also significantly more common in men than in women. Given that comorbidity data were not available in the SICATA registry study (and therefore unable to be adjusted for), the finding of female sex as a risk factor for fatal infection by Ros et al. (9) might potentially be explained by limited multivariable adjustment with residual confounding. Alternative possibilities include a type 1 statistical error (chance finding), coding or classification bias (because of data collection from medical history and death records), or co-intervention bias, whereby female patients may have been treated differently than male patients. The latter possibility was suggested by the fact that female patients in Andalusia were less likely than male patients to receive automated PD (42.1% vs 47.6%, p = 0.04); no such difference was identified in Australia and New Zealand (25.8% vs 25.8%, p = 0.98).

Figure 1 —

Kaplan-Meier survival curve for infection-related mortality in 2251 de novo incident peritoneal dialysis (PD) patients in Australia and New Zealand, 2007-2010.

So, is female sex really a risk factor for infectious death in PD?

The findings of Ros et al. (9) seem to strongly suggest such a risk, but the lack of a robust association between female sex and either peritonitis or fatal infection across different countries and studies argue against it. Nevertheless, it is clear that female PD patients experience a greater burden of infection in some centers and regions, and that this difference may be related to specific local practices, either in relation to selection of female patients for PD or to their subsequent PD management. Hopefully, the reasons that female sex is an important risk factor for PD-associated infection in some centers, but not others, will be able to be teased out from the ongoing Peritoneal Dialysis Outcomes and Practice Patterns Study (P-DOPPS).

Disclosures

DWJ has received research funds, consultancy fees, speakers’ honoraria, and travel sponsorships from Baxter Healthcare and Fresenius Medical Care. RM has received grant support or honoraria from DaVita and Baxter Healthcare. YC has no financial conflicts of interest to declare.