Low cigarette smoking prevalence in peri-urban Peru: results from a population-based study of tobacco use by self-report and urine cotinine

While tobacco smoking has declined dramatically in high-income countries, data suggests that it may be increasing in low- and middle-income countries (LMICs). This rise in smoking stands to accelerate countries’ epidemiologic transitions, contributing to increasing rates of chronic disease morbidity and mortality [1, 2]. In these countries, accurate tobacco use data is essential for public health practitioners designing and implementing interventions to curb smoking and prevent chronic disease.

This work focuses on Peru as a case study. Peru is an LMIC showing early signs of increasing chronic disease, but an uncertain smoking landscape. Earlier studies that have attempted to estimate the prevalence of smoking in Peru have largely focused on urban areas, overlooking peri-urban shantytowns. The CARMELA study reported that the prevalence of current tobacco use – roughly, anyone who reported current daily or occasional inhaled tobacco use – among those aged 25–64 years in urban Lima, Peru was 26.6% in 2005 [3]. The same year, the Center for Information and Education for the Prevention of Drug Abuse in Peru reported a 27.7% prevalence of all smoking and a 7.7% prevalence of daily smoking among those aged 12–64 years, also in urban Lima [4]. Meanwhile, our group estimated in a 2012 study that smoking prevalence may be substantially lower in peri-urban Lima, with a prevalence estimate of 16% for cigarette smoking among adults ≥40 years old [5]. Importantly, only 2% of the sample were daily smokers with the balance reporting occasional smoking [5]. This finding, if true, has important public health implications for Peru. Given the increasing proportion of Peruvians residing in peri-urban zones, high-quality surveillance must be directed to these areas to derive estimates of chronic disease prevalence and risk.

Our first objective was to describe the tobacco-using population of peri-urban Peru by validating our low self-reported finding smoking prevalence with urine cotinine. Cotinine, the primary metabolite of nicotine, is a reliable [6] and valid [7, 8] biomarker that has been long used as a surrogate for tobacco smoke exposure in the previous 48 h [9‐22]. We compared the prevalence of smoking by self-report with that based on a laboratory analysis of urine cotinine in a population-based sample of 382 adults from low-income peri-urban Lima and semi-urban Tumbes, Peru. Our second objective was to characterize the chronic disease burden in Peru, especially related to smoking status. To accomplish this, we compared chronic disease diagnoses, respiratory symptoms, biomarkers, and lung function scores over a 40-month follow up period between self-reported daily, occasional, and non-smokers over four separate field sites. For this analysis, we utilized our full cohort of 2978 adult participants.

Systematic surveillance of tobacco use is an essential step in the formation of programs to reduce smoking-related mortality, especially for those who joined the WHO’s Framework Convention on Tobacco Control, which obliges signatory nations, including Peru, to implement policy changes to reduce tobacco use [26]. In our primary analysis, based on urine cotinine, we validate a lower prevalence of smoking in low-income peri-urban areas than was reported in urban areas in previous large studies. It appears that Peru is part of a pattern seen in other Latin American countries, with the exception of the Southern Cone, that is characterized by smoking practices favoring occasional over daily smoking and fewer cigarettes consumed. The prevalence of self-reported smoking in our entire sample (n = 2978) was 3.3% daily, 8.9% occasional, and 12.2% current smokers. These findings are compatible with GATS data from Mexico [40], Costa Rica [41], and Panama [42]: all of which feature low smoking prevalence and daily smoking equal to or less than half of current smoking. This pattern of reduced tobacco use was previously noted by Tapia-Conyer et al., who found that daily smoking in Mexico decreased from 64% to 52% between 1988 and 1998 while among daily smokers, those who smoked five or fewer cigarettes/day increased from 49% to 74% [43]. Earlier, Samet et al. found that Hispanics in New Mexico tended to consume about half as many cigarettes as Whites, though prevalence of smoking and years smoked were similar [44]. This pattern contrasts with the Southern Cone countries, which have demonstrated higher smoking prevalence and frequency. Both the PLATINO and CARMELA studies report current smoking prevalence of nearly 40% in Chile and Argentina (WHO reports a drop to 29% in Argentina) while figures from GATS show a lower current smoking prevalence of 25% for Uruguay but notes that 80% of current smokers are also daily smokers [3, 45‐47]. Menezes et al. note that only 2 of 1626 (0.12%) current smokers in the PLATINO study were not daily smokers, contrasting remarkably with our findings in Peru [45].

Our self-reported smoking prevalence mirrors that published by Weygandt et al. but was substantially less than what the CARMELA study reported from Lima [3, 5]. In one comparison, we found 15.5% prevalence of current smoking in peri-urban Lima while CARMELA reported 26.6% prevalence for the 45–65 year age range [3]. For all ages in Lima, we found a current smoking prevalence of 14.8% and for all sites, 12.2%. Additionally, CARMELA reported finding a mean of around 7 cigarettes consumed daily among currently smoking men and around 5 for women [3]. Our study found a mean of 1.8 cigarettes per day for men and 1.1 for women among current smokers in peri-urban Lima. However, it is important to note that these sets of estimates are not directly comparable: generally, CARMELA reported use of all inhaled tobacco in an urban sample of age 25–65 years while we report cigarette use in adults aged ≥35 years living in resource-poor settings [3]. One should also note that the relationship between nicotine intake and cigarettes smoked is not completely linear. Blackford et al. report that, in an international sample, participants tended to titrate smoking to their preferred nicotine level and not number of cigarettes smoked. Subsequently, those who smoked fewer than 20 cigarettes per day (99.5% of our sample) consumed more nicotine per cigarette than those who smoked more [11].

In our secondary analysis, we attempted to characterize the smoking populations of low-income areas using longitudinal data. It is difficult to come to any conclusions about the long-term risk of chronic disease between groups, perhaps because the use of tobacco is so low, even among daily smokers. Only 3% of our entire cohort reported a history of ten or more pack-years. Considering the general guidelines for lung cancer screening (loosely, ≥55 years old and ≥30 pack-years smoking), only 0.75% of our sample qualified [48‐51]. Furthermore, there was no difference between mean pack-years for the occasional and non-smoking groups, indicating that the occasionally smoking group had a similar history to the then non-smokers.

Considering this, the relationships between lung function, obesity, and smoking status are of note. Unadjusted means show a greater loss of functional volume (FVC) and ability to exhale (FEV₁) for daily smokers compared to occasional and non-smokers and the gap was made more apparent by linear regression with age, sex, site, and wealth index, although the difference was not statistically significant. The same pattern was found when observing changes in BMI, with daily smokers experiencing a greater percentage rise in BMI than occasional or non-smokers. That pattern was also exacerbated in linear modeling though, again, it was not significant. No association was found with smoking status and change in systolic or diastolic blood pressures. This indicates that, even with relatively low levels of smoking, daily smokers in this sample may have a worsening risk profile for chronic disease than occasional or non-smokers. However, taking into account the rest of our data, the groups may not be different in a clinical way, which presents opportunities for the Peruvian government to intervene and prevent the public health situation from getting worse.

As an observation, multi-center surveys of smoking prevalence such as PLATINO did not distinguish between current and daily smoking, while those using the GATS protocol, such as CARMELA and our group, did [3, 5, 45]. Under the usual definition of current smoking – a small lifetime history and any reported smoking within a month or more – no value is given to the frequency of cigarette smoking, which, as shown by our sample, may be very low. The definition of active smoking from the PLATINO study would have included all of our self-reported daily smokers and 82% of our low-use occasional smokers, inappropriately characterizing our sample. As Weygandt et al. argue, including those who smoke as infrequently as they do in our sample represents an over-estimation of the actual active smoking prevalence in a population [5].

Strengths of this study include high-quality data collection methods and surveillance of disease outcomes, nested within a well-designed population-based study, and standardized laboratory methods for measuring biomarkers. Limitations of this study include the small sample of daily smokers, which limited statistical power in attempts to characterize the group, and the larger number of participants lost to follow up in the Puno field sites, which may have skewed our health outcomes results. Additionally, the varying time between interview and biological sample question could change our associations but, since people in our sample rarely changed smoking status (our main exposure), we don’t anticipate that the lags had any noticeable effect.

In summary, we used creatinine-adjusted urine cotinine to validate observed low prevalence of self-reported active smokers and low overall frequency of smoking in low-income areas of Lima and Tumbes, Peru. This places Peru within an emerging pattern of low prevalence and frequency of smoking, favoring occasional smoking, seen in other Latin American countries except for those in the Southern Cone. As Peru is party to WHO’s FCTC as of 2004, it is imperative that national policymakers have access to accurate epidemiological data on tobacco smoking in order to properly implement the MPOWER program.

In investigating longitudinal health outcomes data, we did not observe significant differences between groups of self-reported smokers, perhaps due to low levels of smoking across the population. However, even with low cigarette consumption, patterns can be seen that favor worsening lung function and BMI over time for daily smokers compared to occasional and non-smokers.

As a final observation, we recommend the use of the Global Adult Tobacco Survey protocol – which distinguishes between daily and current smoking – for tobacco use surveillance in future epidemiological studies. This would be especially useful in characterizing regions such as Peru, where the sole use of current smoking as an indicator overestimates the real prevalence of active smoking.