Intermittent smoking and chronic kidney disease

In this issue of Nephrology Dialysis Transplantation, Franceschini et al. [1] report on the differential associations of daily, intermittent and past smoking, as compared to never smoking, with prevalent chronic kidney disease (CKD), among a healthy, population-based cohort of US Hispanics ages 18–74 years from the Hispanic Community Health Study/Study of Latinos (HCHS/SOL). The authors found that, of the current smokers in the study, about one-third reported being intermittent versus daily smokers. Relative to never smokers, both intermittent and daily smokers had a higher prevalence of CKD, as defined by either microalbuminuria or reduced kidney function, at the time of the survey. While the study has substantial limitations, including its cross-sectional design, self-reported smoking status and single measurements of albuminuria and kidney function, the results are nonetheless suggestive of an association between both daily and intermittent smoking and CKD in this population. These results add to the current pool of observational evidence that smoking may be an independent risk factor for CKD, although the evidence is mixed and may differ by CKD etiology [2, 3]. Furthermore, much of the previous research linking smoking to CKD has been performed in predominantly non-Hispanic white cohorts [3], thus this exploration of the association of smoking patterns with CKD among the growing US Hispanic population [1] is an essential first step toward targeted public health planning.

Because intermittent smoking has rarely, if ever, been examined in the setting of CKD, the results reported by Franceschini et al. [1] generate many new hypotheses, which could lead to future studies in the HCHS/SOL and beyond. Perhaps the most interesting observation made by the authors was the potential effect modification of the association of smoking pattern with CKD by pack-years. Pack-years are defined as the average number of cigarette packs smoked per day multiplied by the number of years smoked and can be thought of as a measure of cumulative lifetime exposure to smoking. By definition, pack-years are higher in daily versus intermittent smokers, given the same number of years smoked and the same maximum number of cigarettes smoked in a day (see Table 1; e.g. A versus C, B versus D). This would suggest that daily smokers would be at higher risk of poor outcomes and that these outcomes would occur earlier in the course of smoking exposure compared with intermittent smokers. However, the authors found that the estimated relative prevalence of CKD for intermittent versus never smokers was more pronounced at lower pack-years of exposure compared with the estimated relative CKD prevalence for daily versus never smokers. For example, the authors found that daily smoking resulting in 40 pack-years of cigarette exposure, relative to never smoking, was associated with prevalent CKD at an odds ratio of 1.4; the same odds ratio was seen in intermittent versus daily smokers at only 10 pack-years of exposure [1].

These results suggest several testable hypotheses. First, the number of years smoked (duration of smoking) may inform CKD risk more than pack-years (cumulative exposure or ‘dose’ of smoking), regardless of smoking pattern. Second, while the authors stated that the majority of CKD cases in the cohort were due to albuminuria, results with albuminuria and reduced kidney function as separate outcomes were not presented. Such analyses might reveal differential associations of interest: e.g. if results in intermittent smokers were mostly attributable to albuminuria (which may appear and resolve over time without subsequent CKD progression) whereas results in daily smokers were often attributable to reduced kidney function (which may take many years to manifest). Finally, these results may simply be the result of bias, including (i) misclassification bias, if daily smokers who were less healthy and at greater risk for CKD were more likely to report their smoking as intermittent, or to have decreased frequency of smoking from daily to intermittent due to declining health prior to the survey; (ii) survival bias, if daily smokers were more likely than intermittent smokers to die of other causes prior to developing CKD and were therefore not captured in this cross-sectional survey; and (iii) residual confounding by unmeasured or unknown factors, such as other high-risk behaviors associated with CKD and smoking.

To test such hypotheses, assessment of intermittent smoking should be considered carefully [4]. Patterns of smoking that individuals report as ‘intermittent’ could vary widely: as shown in Table 1, 40 years of intermittent smoking might mean cumulative exposures of <1 pack-year to >30 pack-years. Current survey items may not capture the heterogeneity within intermittent smoking patterns. Items similar to ‘Of the entire time you have smoked, on average how many cigarettes do you smoke per day?’ are generally used to estimate pack-years. But for the intermittent smokers in Table 1 (individuals C–H), this may be difficult to answer or may even be misinterpreted as the number of cigarettes smoked on the days during which they smoked (versus averaged over all smoking and non-smoking days). Furthermore, individuals with less regular patterns than those depicted in Table 1 (e.g. if patterns vary over weeks or months) might find such a question even more difficult to answer accurately. Methods that involve more frequent assessments of multiple events, closer in time to events, over a longer period, such as Ecological Momentary Assessment [5], may be necessary for more accurate, complete assessment of intermittent smoking in a single survey period. Longitudinal or even retrospective assessment could capture important changes over time in intermittent smoking status, either from never, past or daily smoking to intermittent smoking or from intermittent smoking to past or daily smoking. Finally, as with any self-reported assessment of smoking, social desirability bias is always a potential issue to consider, which may differ by age, sex and race/ethnicity; for US Hispanics, both sex and level of acculturation may affect the level of this bias [6].

Variations in actual exposure to the multiple toxins present in cigarettes and the overall ‘dose’ of these toxins would be important to consider in assessing intermittent versus other patterns of smoking. Many factors could contribute to differential toxic exposures, including brand-to-brand variation in cigarette ingredients; exposure to other tobacco products (pipes, cigars, chewing tobacco); use of electronic nicotine delivery systems, which are currently unregulated by the U.S. Food and Drug Administration and the health effects of which are relatively unknown; secondhand smoke exposure [7] and even exposure to marijuana and other smoked substances. Levels of certain biomarkers could be used to augment self-reported estimation of exposure, as well as to validate the accuracy of self-reported smoking. Combinations of biomarkers that measure relatively recent smoking (e.g. cotinine) [8] and that measure cumulative exposure over time in the kidneys (e.g. cadmium) [9, 10] may be helpful, if not always feasible to measure.

Studies that examine not only intermittent smoking but also other important factors associated with intermittent smoking, such as the reasons for smoking at particular times, are essential to fully inform public health prevention efforts to decrease population-level smoking and poor outcomes potentially associated with smoking, including CKD. For example, the terms ‘intermittent smoker’ and ‘social smoker’ are often used interchangeably, but recent studies have shown that most intermittent smoking events in adults do not occur in the presence of others [11]. Additionally, intermittent smokers may not display the ‘classically dependent’ patterns usually seen in daily smokers and may be cued by stimuli other than temporal cycles of cravings [12]. Effective strategies that lower smoking rates at the population level, such as taxes on tobacco products [13], may be less effective among those who smoke less frequently. Additionally, while it is important to target adolescents, 25% of whom believe that there are no adverse health effects to intermittent smoking [14], it is also important to consider that patterns of smoking may change from adolescence to adulthood and reasons for smoking may also change during this transition [15].

As daily smoking declines in the USA, intermittent smoking may be increasing [16], making the assessment of intermittent smoking increasingly important when examining the association of smoking with CKD development and outcomes. As noted by Franceschini et al. [1], Hispanics who smoke are likely to be intermittent smokers [17], making the HCHS/SOL an ideal population for their study. While one-third of smokers in the HCHS/SOL were intermittent smokers, other recent studies have shown that as many as 80% of Hispanic smokers are intermittent smokers [16]. Similarly, up to 74% of black smokers may also be intermittent smokers, compared with up to 49% of white smokers [16]. Thus, intermittent smoking is increasing overall and is disproportionately prevalent among not only Hispanic but other minority populations who may also be at increased risk of CKD progression and poor outcomes relative to whites [18]. Furthermore, the study presented in this issue of Nephrology Dialysis Transplantation [1] suggests that intermittent smoking may be associated with increased risk of CKD among Hispanics and that this association may exist at fairly low levels of cumulative exposure. Thus, further exploration of the relationship between intermittent smoking and CKD across multiple populations, with careful attention to the assessment of and reasons for intermittent smoking, is warranted to inform prevention efforts.

CONFLICT OF INTEREST STATEMENT

The authors have no conflicts of interest to declare. (See related article by Franceschini et al. Smoking patterns and chronic kidney disease in US Hispanics: Hispanic Community Health Study/Study of Latinos. Nephrol Dial Transplant 2016; 31: 1670–1676)

REFERENCES