Sex and the Incidence and Prevalence of Kidney Disease

The Global Burden of Disease, Injuries and Risk Factors study estimates the worldwide number of men with impaired kidney function at 336 million and the number of women at 417 million, a ratio of 0.81; the number of men treated with dialysis at 1.7 million, and women at 1.3 million, a ratio of 1.3; and the number of men with a functioning kidney transplant at 0.4 million, compared with 0.3 million at women, a ratio of 1.3 (1).

“Women get sick and men die:” the apparent contrast between higher morbidity in women and higher mortality in men was described by Fairfield in 1926. The same paradox appears in available data on kidney disease: women experience a higher prevalence of CKD, but men are more likely to suffer kidney failure, at least as assessed by rates of initiation of kidney replacement therapy (KRT), defined as initiation of dialysis or kidney transplantation.

Rigorously collected, nationally representative data from the US National Health and Nutrition Examination Survey between 1988 and 2012 showed the prevalence of low GFR and of albuminuria to be higher in women than in men (2). For example, the prevalence of stage G3–5 nondialyzed CKD in the 2011–2012 survey was 5.9% in men and 7.8% in women (ratio 0.75), and including albuminuria in the definition resulted in a prevalence of 12.8% and 15.4% in men and women, respectively (ratio 0.83). Throughout this period, the incidence and prevalence of KRT in American men has exceeded that in American women. In Canada, from 2001 to 2009, the ratio of men to women starting dialysis (unadjusted for the sex distribution in the population) was 1.2; in both Canada and the United States, the sex ratio in prevalent patients is similar to this figure (3). Taken together, these data suggest that in North America, where access to KRT is free at the point of use, the pattern of higher prevalence of CKD, but lower KRT incidence and prevalence, in women compared with men, is identical to the global estimates.

This brings us to the new work by Antlanger et al. (4). To address the question of whether injustice—unequal access to KRT, biased against the acceptance of women—might account for the higher incidence of KRT in men, they examined the sex distribution of incidence and prevalence of KRT in nine different countries within the European Renal Association – European Dialysis and Transplant Association Registry. They argued that social factors leading to biased uptake were unlikely to be constant over nine disparate regions and four decades of observation, and sought secular trends to inform this question. They showed that as KRT incidence has increased, most markedly in the elderly, the ratio of men to women in both incident and prevalent KRT (despite consistently exceeding one) is remarkably constant over time. For example, that ratio is 1.2 in patients aged 0–19 years and 2.4 in patients aged ≥75 years in 2005–2015, compared with 1.3 and 2.5, respectively, in 1975–1984. The same pattern was observed for prevalence. The data are robustly collected, internally consistent, and convincing: there is a male preponderance in the incidence and prevalence of KRT that is more marked in more elderly people and is remarkably stable over time.

What can account for this paradox? One possibility is that the estimation of GFR or our definition of normal GFR is flawed in a way that leads to the overdiagnosis of CKD in women. Alternatively, CKD may progress more slowly in women, or they may die disproportionately of other causes before reaching KRT. Finally, cultural and social factors may result in women disproportionately deciding in favor of conservative management or failing to be considered for KRT.

It seems unlikely that estimation of GFR is responsible. The CKD Epidemiology Collaboration (CKD-EPI) equation underestimates iothalamate-measured GFR by 2.2 ml/min per 1.73 m² with no difference between men and women (5). Studies in other data sets have generally confirmed this finding, with the exception of the validation study performed in the Multi-Ethnic Study of Atherosclerosis (MESA), a study of randomly sampled participants free of cardiovascular disease at baseline. This showed that, in both white and black women, CKD-EPI overestimated GFR by 12 and 16 ml/min per 1.73 m², respectively, compared with a respective overestimation of 1.5 and 5.8 in white and black males (6).

However, it remains possible that the definition of abnormal is not correctly calibrated in women. Adjustment for size by body surface area may be insufficient to calibrate GFR in women against that of men, resulting in misclassification of women (in the direction of overdiagnosing disease) and overestimating severity across the spectrum of disease. Again, in the MESA study, iothalamate GFR was measured in a subgroup of 294 people, with correction for body surface area, and was lower in women than in men (e.g., at a mean age of 71 years, mean GFR was 67 ml/min per 1.73 m² in women compared with 78 ml/min per 1.73 m² in men). In principle, body surface area appears to be a useful method of accounting for the differences in size between men and women, with the expectation that, in health, body size correlates with kidney function. However, is it possible that there are true biologic differences between men and women beyond size? If lower GFR in women were relatively more normal, we might expect a lower prevalence of metabolic and hematologic abnormalities at any given GFR. There is limited support for this in data from the CKD Prognosis Consortium: across the range of kidney function, potassium was higher in men than women, and bicarbonate lower; hemoglobin was, unsurprisingly, higher in men, and the relationship between parathyroid hormone and GFR did not differ by sex (7). Alternatively, there are many possible reasons for a true, biologic increase in prevalence of CKD in women compared with men, including complications of pregnancy and increased incidence of lupus and other connective tissue diseases.

If severity of CKD were incorrectly calibrated in women, one would anticipate a lower rate of progression than observed in men. Other, alternative hypotheses to account for a reduced risk of progression in women, compared with men, include the putative protective effect of estrogen or deleterious effects of testosterone, and perhaps unhealthier lifestyles in men compared with women (4). In the CKD Prognosis Consortium individual-patient meta-analyses, the risk of KRT is lower in women than in men at any level of GFR (8). (We are not suggesting that low GFR is unimportant in women: as the authors of this study note, the gradient of the relationship between GFR and risk is actually steeper in women than in men, and at very low GFR the risk in women approaches that in men.) A caveat in accepting this as evidence for differential risk of progression is that the KRT outcome is potentially subject to bias or injustice. Data that do not rely on a KRT outcome are limited. In administrative data from the United Kingdom, overall risk of progression was very low, with >70% of men and women losing <2 ml/min per 1.73m² per year of GFR, but there was, in each age group, an increased proportion who experienced rapid progression (>5 ml/min per 1.73m² per year) in men compared with women (9). A recent meta-analysis of four Italian cohort studies shows increased risk of progression to KRT in men, and also a higher rate of progression determined by change in GFR, in men compared with women (−2.1 versus −1.8 ml/min per 1.73m² per year, respectively). In both analyses, the risk associated with male sex was increased in the presence of proteinuria (10).

Could the competing risk of death before reaching KRT account for the discrepancy in the incidence of KRT? The CKD Prognosis Consortium sheds light on this issue too: risk of death is lower in women than in men across the whole spectrum of GFR (8).

The final possibility is one of differential treatment choices or access to care. This could occur if women were less likely to be referred or were more likely to choose, or be selected for, conservative management. Before eGFR reporting was widespread, women were less likely to be referred than men (9) and, in a small study from the United Kingdom, women were more likely than men to choose maximal conservative management after education about options.

The causes of sex differences in incidence and progression of kidney disease are important and deserve further study. Population-based data can be used to establish normal ranges for GFR using a frequentist approach, and whether kidney disease, defined in this way, shows true biologic differences between men and women can be studied. Using GFR-based outcomes other than KRT (e.g., slope analyses, percentage loss of GFR), although not without issues, will complement studies using KRT as an outcome. Registries should routinely conduct sex-stratified analyses to illuminate secular trends in incidence and prevalence, should collect data on patients with GFR below the thresholds of 15 and 10 ml/min per 1.73 m² and on patients who choose maximal conservative management, and should consider special studies on processes of care that might differ by sex or gender. In the meantime, the idea of bias or injustice in kidney care is difficult to categorically refute. However, the stability in the ratios observed in the important international study by Antlanger et al. (4), over four decades that were attended by marked social changes, is strong evidence that biology, rather than discrimination, may be responsible.

Disclosures

Dr. Clase reports receiving grants from Astellas as a coinvestigator for the investigator-initiated, industry-funded REPORT study and from Boehringer-Ingelheim to fund the OnTarget and Transcend studies, on which she is involved in secondary analyses. Dr. Clase also reports receiving honoraria and/or personal fees for advice given to Amgen, Astellas, Leo Pharma, Janssen, and the Ministry of Health Ontario. Dr. Tomlinson has nothing to disclose.