- Email: [email protected]
Associations between cadmium levels in blood and urine, blood pressure and hypertension among Canadian adults
Environmental Research 155 (2017) 64–72
Contents lists available at ScienceDirect
Environmental Research journal homepage: www.elsevier.com/locate/envres
Associations between cadmium levels in blood and urine, blood pressure and hypertension among Canadian adults
MARK
⁎
Rochelle E. Garnera, , Patrick Levalloisb,c a b c
Health Analysis Division, Statistics Canada, Ottawa, Ontario, Canada Direction de la santé environnementale et de la toxicologie, Institut National de Santé Publique du Québec, Québec City, Québec, Canada Axe santé des populations et pratiques optimales en santé, Centre de Recherche du CHU de Québec-Université Laval, Québec City, Québec, Canada
A R T I C L E I N F O
A BS T RAC T
Keywords: Cadmium Blood pressure Hypertension Smoking Body mass index Canadian Health Measures Survey
Background: Cadmium has been inconsistently related to blood pressure and hypertension. The present study seeks to clarify the relationship between cadmium levels found in blood and urine, blood pressure and hypertension in a large sample of adults. Methods: The study sample included participants ages 20 through 79 from multiple cycles of the Canadian Health Measures Survey (2007 through 2013) with measured blood cadmium (n=10,099) and urinary cadmium (n=6988). Linear regression models examined the association between natural logarithm transformed cadmium levels and blood pressure (separate models for systolic and diastolic blood pressure) after controlling for known covariates. Logistic regression models were used to examine the association between cadmium and hypertension. Models were run separately by sex, smoking status, and body mass index category. Results: Men had higher mean systolic (114.8 vs. 110.8 mmHg, p < 0.01) and diastolic (74.0 vs. 69.6 mmHg, p < 0.01) blood pressure compared to women. Although, geometric mean blood (0.46 vs. 0.38 µg/L, p < 0.01) and creatinine-adjusted standardized urinary cadmium levels (0.48 vs. 0.38 µg/L, p < 0.01) were higher among those with hypertension, these differences were no longer significant after adjustment for age, sex and smoking status. In overall regression models, increases in blood cadmium were associated with increased systolic (0.70 mmHg, 95% confidence interval [CI]=0.25–1.16, p < 0.01) and diastolic blood pressure (0.74 mmHg, 95% CI=0.30–1.19, p < 0.01). The associations between urinary cadmium, blood pressure and hypertension were not significant in overall models. Model stratification revealed significant and negative associations between urinary cadmium and hypertension among current smokers (OR=0.61, 95% CI=0.44–0.85, p < 0.01), particularly female current smokers (OR=0.52, 95% CI=0.32–0.85, p=0.01). Conclusion: This study provides evidence of a significant association between cadmium levels, blood pressure and hypertension. However, the significance and direction of this association differs by sex, smoking status, and body mass index category.
1. Introduction Cadmium is a heavy metal that occurs naturally in the environment. Whereas certain occupational subgroups are at elevated risk of cadmium exposure, including sawmill and wood preservation workers, auto repair workers, and welders (CAREX Canada, 2014), nonoccupational exposure to cadmium is generally the result of cigarette smoking and ingestion of cadmium-high foods (CAREX Canada, 2015; Garner and Levallois, 2016). The International Agency for Research on Cancer considers there to be sufficient evidence of the carcinogenic effect of cadmium and its compounds among humans (IARC Working Group on the Evaluation of Carcinogenic Risks to Humans, 2009).
⁎
Furthermore, other evidence indicates that prolonged or increased exposure to cadmium could have significant health consequences, including kidney dysfunction, skeletal damage, and possible cardiovascular effects (Järup et al., 1998; United Nations Environment Programme, 2010). Blood levels of cadmium are valid markers of recent exposure (Järup and Åkesson, 2009; Järup, 2003). However, even after exposure cessation, cadmium levels in the blood can remain elevated, and are therefore also reflective of long-term exposure (Järup et al., 1998). Urinary cadmium levels reflect long-term (cumulative) exposure as well as cadmium levels in the kidney (Centers for Disease Control and Prevention, 2009, p. 202; Järup, 2003). Animal model studies have demonstrated that higher levels of
Correspondence to: Health Analysis Division, Statistics Canada, 100 Tunney's Pasture Driveway, RH Coats Building, 24th floor, Ottawa, Ontario, Canada K1A 0T6. E-mail address: [email protected] (R.E. Garner).
http://dx.doi.org/10.1016/j.envres.2017.01.040 Received 3 October 2016; Received in revised form 5 January 2017; Accepted 31 January 2017 0013-9351/ Crown Copyright © 2017 Published by Elsevier Inc. All rights reserved.
Environmental Research 155 (2017) 64–72
R.E. Garner, P. Levallois
2.2. Measures
cadmium are associated with increased blood pressure and hypertension (Perry and Erlanger, 1974; Perry et al., 1977, 1979; United Nations Environment Programme, 2010). Observational studies, however, have found mixed results. Studies vary in the direction or significance that cadmium body stores have on blood pressure or hypertension, with results seeming to vary by type of cadmium biomonitoring sample (blood vs urine), sex, smoking status, ethnicity and medication levels (Tellez-Plaza et al., 2008; Whittemore et al., 1991; Agarwal et al., 2011; Scinicariello et al., 2011; Lee and Kim, 2012; Lee et al., 2016). In their recent systematic review and metaanalysis of the associations between blood and urine cadmium, blood pressure and hypertension, Gallagher and Meliker (2010) concluded that there was evidence of a positive association between blood cadmium (BCd) and blood pressure among women, but not among men. Furthermore, their analysis indicated that there was an inverse relationship between urinary cadmium (UCd) and the likelihood of hypertension. However, Gallagher and Meliker pointed out that few studies included representative, population-based samples of never smokers, and that previous studies have varied in their outcome definitions, thereby limiting interpretation of meta-analysis findings (Gallagher and Meliker, 2010). The present study seeks to add to the existing literature by examining the association between cadmium levels of adult Canadians and measurements of blood pressure and the presence of hypertension. This study will examine the association of cadmium levels in both blood and urine, and will also examine whether the significance or magnitude of any association between cadmium and blood pressure differs according to sex or smoking status.
2.2.1. Cadmium measures Blood and urine samples were collected from all eligible respondents at the mobile examination clinic. Blood samples were collected by a phlebotomist using a standardized venipuncture technique. These were subsequently centrifuged and aliquoted into smaller tubes. Respondent-provided urine samples were also aliquoted into smaller tubes. Urine samples were collected using the first catch urine in cycle 2, as opposed to the mid-stream urine collected in cycle 1. Cycle 2 respondents were also asked to abstain from urinating two hours prior to their appointment. This may have resulted in a shift in creatinine levels when multiple cycles are compared. This, in turn, could affect creatinine-adjusted levels of cadmium. Blood and urine sample tubes were placed in shipping trays and stored in the mobile examination clinic laboratory's refrigerator or freezer, depending on the test. All specimens were stored as soon as processing was complete to ensure the quality of the samples’ integrity, which was achieved within two hours from the time of collection for most samples: a four hour ceiling from the point of collection was placed on the time for blood samples to be processed and stored (Statistics Canada, 2015a). The limit of detection (LOD) for BCd was 0.045 µg/L (0.4 nmol/L) in cycles 1 and 2, and 0.080 µg/L (0.71 nmol/L) in cycle 3. The LOD for UCd was 0.09 µg/L (0.8 nmol/L) for cycle 1 and 0.07 µg/L (0.6 nmol/ L) for cycle 2. Consistent with others who suggest excluding urine samples that are too dilute (Alessio et al., 1985), urine samples with creatinine levels below the LOD (0.31 mmol/L for cycle 1, 0.44 mmol/ L for cycle 2) were excluded from analyses (n=10). For remaining samples, observations below the LOD for the particular measure were imputed with values at half the LOD. When the LOD for a particular test differed between cycles, the highest LOD was used (BCd, 0.08 µg/ L; UCd, 0.09 µg/L). All values were rounded to two significant digits (Statistics Canada, 2015b), and were converted from Système International units to conventional units for the study. For UCd analyses, it was necessary to take into account the dilution level of the urine sample. This is generally accomplished by adjusting for urinary creatinine levels. Greater attention is being paid to the method used to adjust urinary biomarkers for creatinine, and the analytic impact that adjustment has. The present study used creatinine-adjusted standardization, an adjustment technique recommended by O'Brien et al. (2016). First, a model was fit for the natural logarithm of creatinine as a function of age, sex, body mass index (BMI) category, use of anti-hypertensive medication, diabetes diagnosis, presence of chronic kidney disease, and CHMS cycle. Next, the observed creatinine level was divided by the predicted creatinine level (exponentiated to yield the original metric) to generate a creatinine ratio. Finally, the observed UCd level was divided by this creatinine ratio to yield a creatinine-adjusted standardized UCd measure (CAS-UCd).
2. Materials and methods 2.1. Data source Data for the present study were drawn from multiple cycles of the Canadian Health Measures Survey (CHMS), which collects both survey-based information as well as direct physical health measures for Canadians. The present study pools information from Cycle 1 (2007–2009) and Cycle 2 (2009–2011) for analyses concerning UCd measures, and further includes Cycle 3 (2012–2013) data for analyses concerning BCd measures: UCd was not measured in Cycle 3 of the CHMS. The CHMS includes the population ages 3 through 79 (6 through 79 in Cycle 1) living in the ten provinces, but excludes the following individuals: persons living in the three territories; persons living on reserves and other Aboriginal settlements in the provinces; full-time members of the Canadian Forces; the institutionalized population and residents of certain remote regions. These exclusions are thought to represent approximately 4% of the target population (Statistics Canada, 2015a). The study sample was limited to respondents 20–79 years of age, and excluded pregnant women (n=106) and those who were missing relevant measures of cadmium (blood, urine) or blood pressure. This left a final sample of 10,099 respondents for BCd analyses, and 6988 respondents for UCd analyses.
2.2.2. Blood pressure and hypertension A series of blood pressure readings were taken at one-minute intervals for each respondent, with the last five readings being averaged to determine systolic (SBP) and diastolic blood pressure (DBP) measures. Respondents with any of the following characteristics were classified as hypertensive: (i) SBP≥140 mmHg; (ii) DBP≥90 mmHg; (iii) self-reported doctor-diagnosed high blood pressure; or (iv) use of anti-hypertensive medications, which was determined based on the Anatomical Therapeutic Chemical (ATC) code assigned to respondents’ medications. Based on previous work with CHMS data (Bushnik et al., 2014; Wilkins et al., 2012), codes indicative of anti-hypertensive medications were: beta blockers (ATC codes C07, excluding C07AA07, C07AA12 and C07AG02); agents acting on the reninangiotensin system (ATC codes C09); thiazide diuretics (ATC codes C03, excluding C03BA08 and C03CA01); calcium channel antagonists (ATC codes C08); and miscellaneous anti-hypertensives (ATC codes C02, excluding C02KX01).
2.1.1. Ethics approval for the CHMS All processes of the CHMS were reviewed and approved by the Research Ethics Boards of two federal agencies (Health Canada, Public Health Agency of Canada) to ensure that internationally recognized ethical standards for human research were met and maintained. In addition, protocols were developed through extensive consultation with recognized experts and were performed by accredited health professionals in conformance with universal precautions. 65
Environmental Research 155 (2017) 64–72
R.E. Garner, P. Levallois
Table 1 Characteristics of respondents aged 20 through 79 years old, Canadian Health Measures Survey, 2007–2013.
Continuous variables Age Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Body mass index, measured (kg/m2) Blood cadmium (µg/L) Urinary cadmium, creatinine-adjusted standardized (µg/L) Categorical variables Age 20–39 40–59 60–79 Has hypertension No Yes Anti-hypertensive medication use No Yes Smoking status Never smoker Former smoker Current smoker Body mass index category, measured Underweight Acceptable weight Overweight Obese Drinking status Never drinker Former drinker Current drinker Diabetes No Yes Chronic kidney disease No Yes Daily or near-daily exposure to second hand smoke No Yes Cycle Cycle 1 Cycle 2 Cycle 3
p-value*
Overall (n=10,099)
Males (n=4850)
Female (n=5249)
Meana 46.0 112.8 71.8 27.3 0.40 0.41 n
SE 0.1 0.4 0.3 0.2 0.01 0.01 %
Meana 45.5 114.8 74.0 27.6 0.35 0.41 n
SE 0.2 0.5 0.4 0.2 0.01 0.01 %
Meana 46.5 110.8 69.6 27.1 0.45 0.41 n
SE 0.1 0.4 0.3 0.2 0.01 0.01 %
< 0.01 < 0.01 < 0.01 0.01 < 0.01 0.74
3404 3504 3191
36.9 41.1 22.1
1567 1732 1551
37.9 40.9 21.1
1837 1772 1640
35.8 41.2 23.0
< 0.01 0.26 < 0.01
7201 2898
74.7 25.3
3354 1496
72.4 27.6
3847 1402
76.9 23.1
< 0.01
7894 2205
81.4 18.6
3752 1098
80.6 19.4
4142 1107
82.3 17.7
0.19
4845 2972 2264
48.8 27.3 23.9
2070 1525 1241
44.4 28.4 27.2
2775 1447 1023
53.2 26.2 20.6
< 0.01 0.16 < 0.01
130 3451 3746 2753
1.6 36.6 36.2 25.6
31 1341 2166 1305
0.8b 30.0 43.0 26.2
99 2110 1580 1448
2.4b 43.2 29.4 25.0
< 0.01 < 0.01 < 0.01 0.43
598 1096 8402
6.5 10.0 83.5
182 476 4192
4.4 9.4 86.2
416 620 4210
8.6 10.7 80.7
< 0.01 0.26 < 0.01
9190 909
91.9 8.1
4355 495
90.5 9.5
4835 414
93.3 6.7
< 0.01
9338 761
94.1 5.9
4525 325
94.7 5.3
4813 436
93.4 6.6
0.05
7446 2653
69.6 30.4
3447 1403
66.3 33.7
3999 1250
72.9 27.1
< 0.01
3420 3565 3114
33.4 41.4 25.2
1622 1673 1555
33.4 41.4 25.2
1798 1892 1559
33.4 41.4 25.2
0.91 0.99 0.92
Note. All estimates in table are based on pooled analysis of CHMS cycles 1 through 3, expect urinary cadmium levels which are based on cycles 1 and 2 only. a Estimates are geometric mean values for measures of cadmium, arithmetic means otherwise. b Due to a moderate coefficient of variation (between 0.167 and 0.333), the estimate should be used with caution. * p-value for comparison between male and female estimates based on t-tests for continuous variables and chi-square tests for categorical variables.
among men (n=31), this category could not be retained as a separate group for stratified models, though underweight respondents were retained in other models. Alcohol consumption status was based on self-report and was categorized as never, former and current drinkers. The presence of diabetes was based on respondents’ self-report of a diagnosis by a doctor. Chronic kidney disease was based on either self-report or was defined as an estimated glomerular filtration rate (GFR) of less than 60 mL/min/1.73 m2. GFR was estimated as 1.75*(serum creatinine in mg/dL)−1.154*(age)−0.203*(0.742 if female)*(1.212 if cultural or racial background is black) (Coresh et al., 2007). Lastly, in regression models, dummy variables were used as indicators of the CHMS Cycle in which the respondent participated: Cycle 2 was the reference category.
2.2.3. Other factors Respondents’ smoking status was based on self-report and was categorized as never smokers, former smokers, and current smokers. Individuals who self-reported as never smokers or former smokers with urinary cotinine levels greater than 50 ng/mL were re-classified as current smokers (n=57 and n=216, respectively; Jarvis et al., 1988; SRNT Subcommittee on Biochemical Verification, 2002). Second hand smoke exposure was dichotomized to identify individuals who reported being exposed on a daily or near-daily basis: current smokers were asked to exclude their own smoking when reporting second hand smoke exposure. An individual's BMI was based on their measured height and weight, and is calculated as weight in kilograms divided by height in metres squared. In some analyses, to minimize the number of model covariates and preserve degrees of freedom, BMI was used as a continuous measurement, centred at 21.75 (the mid-point of the acceptable weight category). In stratified analyses, BMI was used categorically as acceptable weight (18.5≤BMI < 25), overweight (25≤BMI < 30), and obese (BMI≥30). Because so few respondents were classified as underweight (BMI < 18.5, n=130), particularly
2.3. Analysis The association between cadmium levels and other covariates with SBP, DBP and hypertension were first examined descriptively, then were assessed using regressions models. Linear regression models were used to examine the association between cadmium levels and blood 66
Environmental Research 155 (2017) 64–72
R.E. Garner, P. Levallois
models run for the sample overall, as well as those run separately by sex, smoking status and BMI category. In the overall sample, increasing BCd levels were significantly associated with higher blood pressure (SBP and DBP; Figs. 1 and 2), whereas CAS-UCd levels were not. In smoking-stratified models, BCd levels were significantly and positively associated with SBP among former smokers (Fig. 1) and with DBP among never smokers (Fig. 2). In BMI-stratified models, higher BCd levels were associated with significantly higher blood pressure among those who were overweight: associations among those who were of acceptable weight or who were obese were not statistically significant (Figs. 1 and 2). Among those who were classified as obese, higher levels of CAS-UCd were associated with statistically lower SBP (Fig. 3) and DBP (Fig. 4): CAS-UCd showed no association with blood pressure among non-obese. Further stratifying models by sex revealed certain important differences, particularly in smoking- and BMI-stratified models. Most notably, the difference in the association between CAS-UCd and blood pressure among current smokers, and between BCd and blood pressure among those who were overweight. Among respondents classified as overweight, there was a statistically significant and positive association between men's BCd levels and their SBP (Fig. 1) and DBP (Fig. 2), whereas the association was effectively null among overweight women. Examining the association between CAS-UCd among current smokers, there was no association between CAS-UCd levels and blood pressure for men, whereas there was a significant and negative association among women, whereby higher levels of CAS-UCd among current smoking women were associated with significantly lower blood pressure. Exclusion of CHMS respondents taking anti-hypertensive medications did not generally alter the association between CAS-UCd and blood pressure (see Supplemental Table A1). Point estimates were generally very similar between models run on samples including and excluding respondents who were currently using anti-hypertensive medications, although variance around the estimates was usually broader in the latter. In some cases, this increased variance resulted in a loss of statistical significance upon exclusion of respondents currently taking anti-hypertensive medications. Conversely, the association between BCd and blood pressure among current smokers became statistically significant and positive in direction after excluding treated hypertensives (SBP: beta=1.10, p=0.02; DBP: beta=0.78, p=0.02; Supplemental Table A1).
pressure (SBP and DBP), while logistic regression models were used in the case of hypertension. Cadmium levels were transformed using the natural logarithm and were entered as a continuous measure in regression models. Models were first run for the sample overall, then were run separately by sex and smoking status. Regression models (linear and logistic) were adjusted for the following factors: age (linear and quadratic effects), sex (in overall model only), smoking status (never, former, current), use of antihypertensive medications (full sample only), BMI (continuous, centred at 21.75), alcohol consumption status (never drinker, former drinker, current drinker), presence of diabetes, presence of chronic kidney disease, daily or near-daily exposure to second hand smoke, and indicator(s) for CHMS cycle. During the model building process, it was noted that the inclusion of BMI in models sometimes affected the cadmium estimates. Therefore, models were also run separately by BMI category (acceptable weight, overweight, obese). While interactions between cadmium levels and the stratifying variables were not always statistically significant in the overall model, stratification was deemed an appropriate strategy as it allowed the impact of all covariates to differ by the stratifying variable, not just the cadmium measure. Given the limited degrees of freedom available in analyses, it was not possible to include all potential interactions in one model. Therefore, while stratification does not allow for a direct comparison of the impact of cadmium across models, it does allow for an examination of potential differences in effect across levels of the stratifying variable, i.e. sex, smoking status, and BMI. Analyses pooling cycles 1 through 3 used 35 denominator degrees of freedom, whereas analyses pooling only cycles 1 and 2 used 24 degrees of freedom. All analyses were conducted in SAS-callable SUDAAN (version 11.0.1), using the appropriate pooled sample weights, and were bootstrapped to account for the sample design (Statistics Canada, 2015b). 3. Results 3.1. Descriptive analyses Based on the pooled sample of CHMS cycle 1 through 3 respondents, the respondents were evenly split between males and females (50.2% vs. 49.8%). The average age of respondents was 46.0 years, with females being a year older on average than males in the sample (Table 1). Men had significantly higher SBP and DBP, and were more likely to be hypertensive, compared to women (Table 1). Women had higher average BCd levels than men, where CAS-UCd levels were similar between males and females (Table 1).
3.4. Multivariate analyses: hypertension Regarding hypertension, the overall models showed no statistically significant association between hypertension and levels of BCd (Fig. 5) or CAS-UCd (Fig. 6), although the change in the odds of hypertension associated with increasing levels of BCd just marginally failed to reach statistical significance (OR=1.13, 95% CI=0.99–1.29, p=0.06). In smoking-stratified models, there was no association between hypertension and BCd levels. However, there was a significant and negative association between CAS-UCd levels and hypertension among current smokers (OR=0.61, 95% CI=0.44–0.85, p < 0.01). Stratification by BMI did not reveal further associations overall. All ORs and 95% confidence intervals are presented in Supplemental Table A2. Once again, further stratifying models by sex revealed different associations between cadmium levels and hypertension. Although none of the sex-specific estimates for the association of BCd levels with the hypertension were statistically significant (Fig. 5), point estimates varied between men and women among former and current smokers, and among those of acceptable weight. Examining the estimates associated with CAS-UCd, higher cadmium levels among female current smokers were associated with a lower odds of having hypertension (OR=0.52, 95% CI=0.32–0.85, p=0.01), whereas the association was null for males (OR=0.67, 95% CI=0.41–1.08, p=0.10; Fig. 6).
3.2. Crude and adjusted cadmium levels by sample characteristics Individuals with high SBP and DBP had significantly higher levels of CAS-UCd than those with lower levels of blood pressure (Table 2). Levels of BCd were also significantly higher among those with high SBP compared to those with low SBP, whereas this association was not found for DBP. However, differences in cadmium levels were no longer statistically significant after adjustment for age, sex and smoking status (Table 2). However, cadmium levels did differ statistically, even after adjustment, by smoking status, BMI category (BCd only), drinking status, and diabetes (CAS-UCd only). Furthermore, after adjustment, respondents taking anti-hypertensive medications had significantly lower CAS-UCd levels compared to those not taking such medications: this association was not significant for BCd levels (Table 2). 3.3. Multivariate analyses: blood pressure Figs. 1 through 4 show the estimated beta coefficients and their 95% confidence intervals for the associations between natural logarithm transformed cadmium levels and blood pressure from regression 67
Environmental Research 155 (2017) 64–72
R.E. Garner, P. Levallois
Table 2 Crude and adjusted geometric mean cadmium levels (blood, urinary creatinine-adjusted standardized) by select respondent characteristics, Canadian Health Measures Survey. Blood cadmium (µg/L), Cycles 1 through 3 Adjusteda
Crude Factor
Sex Male Female Age 20–39 40–59 60–79 Systolic blood pressure Low High Diastolic blood pressure Low High Anti-hypertensive medication use No Yes Hypertension No Yes Smoking status Never smoker Former smoker Current smoker Body mass index category, measured Underweight Acceptable weight Overweight Obese Drinking status Never drinker Former drinker Current drinker Diabetes No Yes Chronic kidney disease No Yes Daily or near-daily exposure to second hand smoke No Yes Cycle Cycle 1 Cycle 2 Cycle 3 a b
Urinary cadmium, creatinine-adjusted standardized (µg/L), Cycles 1 and 2
Geo. mean
SE
p
Geo. mean
0.35 0.45
0.01 0.01
Ref. < 0.01
0.31 0.46 0.46
0.01 0.01 0.01
0.39 0.49
SE
Adjusteda
Crude p
Geo. mean
SE
p
Geo. mean
SE
p
n/a n/a
0.41 0.41
0.01 0.01
Ref. 0.75
n/a n/a
< 0.01 Ref. 0.92
n/a n/a n/a
0.31 0.44 0.53
0.01 0.01 0.02
< 0.01 Ref. < 0.01
n/a n/a n/a
0.01 0.03
Ref. < 0.01
0.40 0.41
0.01 0.02
Ref. 0.62
0.40 0.53
0.01 0.02
Ref. < 0.01
0.41 0.43
0.01 0.02
Ref. 0.20
0.40 0.41
0.01 0.04
Ref. 0.62
0.40 0.44
0.01 0.04
Ref. 0.20
0.40 0.50
0.01 0.03
Ref. < 0.01
0.40 0.46
0.01 0.03
Ref. 0.05
0.38 0.46
0.01 0.02
Ref. < 0.01
0.40 0.40
0.01 0.01
Ref. 0.86
0.39 0.48
0.01 0.01
Ref. < 0.01
0.41 0.38
0.01 0.01
Ref. 0.01
0.38 0.46
0.01 0.02
Ref. < 0.01
0.40 0.40
0.01 0.01
Ref. 0.59
0.38 0.48
0.01 0.01
Ref. < 0.01
0.41 0.40
0.01 0.01
Ref. 0.66
0.22 0.33 1.63
0.01 0.01 0.06
Ref. < 0.01 < 0.01
0.22 0.30 1.73
0.01 0.01 0.06
Ref. < 0.01 < 0.01
0.33 0.45 0.56
0.01 0.01 0.02
Ref. < 0.01 < 0.01
0.34 0.41 0.59
0.01 0.01 0.02
Ref. < 0.01 < 0.01
0.61b 0.42 0.38 0.38
0.11 0.02 0.02 0.01
0.05 Ref. 0.16 0.12
0.47 0.43 0.38 0.37
0.04 0.01 0.01 0.01
0.32 Ref. 0.02 0.01
0.35 0.39 0.42 0.43
0.04 0.01 0.01 0.01
0.43 Ref. 0.02 0.03
0.39 0.41 0.40 0.40
0.04 0.01 0.01 0.01
0.68 Ref. 0.58 0.58
0.34 0.47 0.39
0.03 0.03 0.01
Ref. < 0.01 0.14
0.53 0.43 0.38
0.04 0.02 0.01
Ref. 0.01 < 0.01
0.46 0.51 0.39
0.03 0.03 0.01
Ref. 0.21 0.03
0.56 0.47 0.39
0.04 0.03 0.01
Ref. 0.03 < 0.01
0.39 0.44
0.01 0.03
Ref. 0.21
0.40 0.37
0.01 0.01
Ref. 0.08
0.40 0.56
0.01 0.02
Ref. < 0.01
0.40 0.47
0.01 0.02
Ref. < 0.01
0.39 0.51
0.01 0.04
Ref. < 0.01
0.39 0.45
0.01 0.02
Ref. < 0.01
0.40 0.56
0.01 0.03
Ref. < 0.01
0.40 0.45
0.01 0.03
Ref. 0.04
0.33 0.60
0.01 0.03
Ref. < 0.01
0.38 0.43
0.01 0.01
Ref. < 0.01
0.39 0.44
0.01 0.01
Ref. < 0.01
0.40 0.43
0.01 0.01
Ref. 0.03
0.41 0.37 0.42
0.02 0.02 0.02
0.10 Ref. 0.10
0.41 0.38 0.41
0.01 0.02 0.01
0.16 Ref. 0.16
0.38 0.43
0.01 0.02
Ref. < 0.01
0.37 0.44
0.01 0.02
Ref. < 0.01
Cadmium levels are adjusted for age, sex, and smoking status; estimates by smoking status categories are only adjusted for age and sex Due to a moderate coefficient of variation (between 0.167 and 0.333), the estimate should be used with caution
only among current smokers, particularly among women. The mechanism by which cadmium is thought to affect blood pressure is not clearly understood, although one hypothesized mechanism is via kidney damage (Buchet et al., 1990; Roels et al., 1991). Individuals with decreased kidney function are at increased risk of hypertension (Kalra, 2007). In addition, kidney damage may also be a sequelae of hypertension (Kazancioğlu, 2013; Lea and Nicholas, 2002). The current study included an indicator of chronic kidney disease in regression models to account for this relationship. In fact, the current study may have over-adjusted the relationship between cadmium and blood pressure by including a measure of decreased kidney function. However, inclusion of this indicator did not appear to significantly alter this relationship. A novel finding from the present study was the differential association between cadmium levels, blood pressure and hypertension by BMI category. While higher BMI was associated with higher blood
4. Discussion After controlling for other pertinent factors, higher BCd levels were associated with higher blood pressure levels (SBP and DBP) and a greater likelihood of hypertension, whereas there was no association for CAS-UCd levels overall. Subgroup analyses indicated that CAS-UCd levels were negatively associated with blood pressure and the likelihood of hypertension among currently smoking women. In their meta-analysis of the existing literature, Gallagher and Meliker (2010) concluded that there was evidence of a positive association between BCd and blood pressure among women, and of an inverse association between UCd and hypertension. The present study found a positive relationship between BCd and blood pressure overall and a similar association, although not statistically significant, in sex-specific analyses. Regarding the association between UCd and hypertension, an inverse relationship was found in the present study 68
Environmental Research 155 (2017) 64–72
R.E. Garner, P. Levallois
Fig. 1. Beta coefficients (95% confidence intervals) showing change in systolic blood pressure associated with unit change in natural logarithm transformed blood cadmium from linear regression models, overall and by sex, smoking status, and body mass index category.
Fig. 2. Beta coefficients (95% confidence intervals) showing change in diastolic blood pressure associated with unit change in natural logarithm transformed blood cadmium from linear regression models, overall and by sex, smoking status, and body mass index category.
cadmium exposure sources may explain these differences. These nonsmoking exposure sources may be associated with long-term behaviours, as the BMI-specific associations with blood pressure were found only for UCd, which is an indicator of overall body burden of cadmium, not BCd, which is more sensitive to short-term exposure. In their analysis of NHANES II data, Whittemore et al. (1991) found that a significant and positive association between UCd and blood pressure was rendered non-significant after respondents receiving anti-hypertensive treatment were excluded from the sample. This exclusion was based on the premise that use of certain anti-hypertensive medications increased the urinary excretion of cadmium (McKenzie and Kay, 1973; Wester, 1973). In the current study, exclusion of CHMS respondents taking anti-hypertensive medications did not generally alter the association between CAS-UCd and blood pressure, although some of the measures of association had wider confidence intervals. Given the cross-sectional nature of CHMS data, it
pressure and a greater probability of hypertension in the present study, results from the sex- and BMI-stratified models showed that, within BMI categories, higher CAS-UCd levels were associated with significantly higher blood pressure levels among overweight (DBP only) and obese women (SBP and DBP). These associations are controlled for the impact of smoking status. Others have shown that levels of cadmium decrease with increasing BMI (Garner and Levallois, 2016; Padilla et al., 2010), a relationship that has been shown with other heavy metals as well (Padilla et al., 2010). However, to our knowledge, other studies have not reported differential associations between cadmium levels and blood pressure measures by BMI categories. Although BMI was included as a covariate in the adjustment and standardization of UCd levels for creatinine, other BMI-related factors that may affect the excretion of cadmium may be responsible for the differential association of CAS-UCd across BMI categories. Furthermore, given that BMIstratified models controlled for the effect of smoking status, other
69
Environmental Research 155 (2017) 64–72
R.E. Garner, P. Levallois
Fig. 3. Beta coefficients (95% confidence intervals) showing change in systolic blood pressure associated with unit change in natural logarithm transformed urinary cadmium (creatinine-adjusted standardized) from linear regression models, overall and by sex, smoking status, and body mass index category.
Fig. 4. Beta coefficients (95% confidence intervals) showing change in diastolic blood pressure associated with unit change in natural logarithm transformed urinary cadmium (creatinine-adjusted standardized) from linear regression models, overall and by sex, smoking status, and body mass index.
smoking behaviours, they represent different exposure timings. BCd levels among current smokers likely represent current exposure, proportional to the number of cigarettes smoked (Järup et al., 1998). BCd levels will decrease rapidly upon smoking cessation, although levels may not drop to pre-exposure levels (Järup et al., 1998). Therefore, BCd levels among formers smokers may reflect time since cessation. It has been shown that blood pressure levels will increase upon smoking cessation, and that they may remain elevated even as time since cessation increases (Lee et al., 2001). An association between cadmium levels and hypertension was found only with UCd, and only among current smokers. In this group, increased CAS-UCd was associated with a decrease in the likelihood of having hypertension (OR=0.61, 95% CI=0.44–0.85; Fig. 6 and Supplemental Table A2). This finding is consistent with the association found with blood pressure in this study, particularly DBP. It may also be the result of a “healthy smoker” effect, whereby current smokers are those who are sufficiently healthy to continue to smoke. For example, individuals may cease smoking, and therefore become former smokers,
is not clear the direction of association between cadmium levels and the use of anti-hypertensive medications. For example, are those with higher cadmium levels more likely to require the use of anti-hypertensive medications? Or, conversely, does the use of anti-hypertensive medications increase an individuals’ cadmium levels? In the present study, respondents taking anti-hypertensive medications had significantly higher BCd (0.46 µg/L vs. 0.38 µg/L; p < 0.01) and CAS-UCd levels (0.48 µg/L vs. 0.39 µg/L; p < 0.01) compared to those not taking anti-hypertensive medications (Table 2). Excluding those using antihypertensive medications would mean excluding a large proportion of respondents (18.6% overall), as well as those at the higher end of the cadmium levels. Therefore, exclusion of those on such medications may be unnecessarily restrictive. Increasing BCd levels among former smokers was associated with an increase in SBP. Although this association was not initially statistically significant among current smokers, it became significant after exclusion of those taking anti-hypertensive medications. Although BCd levels among former and current smokers are likely a result of 70
Environmental Research 155 (2017) 64–72
R.E. Garner, P. Levallois
Fig. 5. Odds ratios (95% confidence intervals) of association between natural logarithm transformed blood cadmium and hypertension from logistic regression models, by sex, smoking status, and body mass index category.
ability to determine causality. Second, the method used in the present study to adjust UCd levels for creatinine differed from that used in other studies, thereby potentially affecting the ability to compare the current results to those from other studies in the literature. However, the adjusted and standardized method has been shown to produce the least biased results in simulation studies (O'Brien et al., 2016). Lastly, the response rates of the combined cycles were low: 52.9% for cycles 1 through 3, and 53.5% for cycles 1 and 2. This largely reflects a low response rate among households initially contacted for participation; households that agreed to provide a household roster had high response rates (more than 78%) to the interview and mobile examination centre components of the survey. It is not known if households that initially declined to participate differed in important ways from participating households, or how this may impact the current study's results.
after a negative health event or diagnosis, including the diagnosis of hypertension (Neutel and Campbell, 2008). Therefore, those with morbidity are shifted from the exposed current smoker group and moved into the less healthy, but less exposed, former smoker category. Furthermore, others have shown that smoking quitters have a higher risk of incident hypertension compared to current smokers (Lee et al., 2001), which is also consistent with our current study. A strength of the present study is its ability to examine both BCd and UCd levels in a population-based sample. It should be noted, however, that the analytic sample was larger for the BCd analyses as UCd was not collected in cycle 3 of the CHMS. Furthermore, Gallagher and Meliker (2010) had previously noted in their meta-analysis that the literature was lacking population-based studies of never-smokers: the present study serves to reduce that literature gap by reporting results among never, former and current smokers. Despite its strengths, the present study has certain limitations that should be considered. First, as previously noted, the cross-sectional nature of the data precludes the
Fig. 6. Odds ratios (95% confidence intervals) of association between natural logarithm transformed urinary cadmium (creatinine-adjusted standardized) and hypertension from logistic regression models, by sex, smoking status, and body mass index category.
71
Environmental Research 155 (2017) 64–72
R.E. Garner, P. Levallois
Järup, L., 2003. Hazards of heavy metal contamination. Br. Med. Bull. 68 (1), 167–182. Järup, L., Åkesson, A., 2009. Current status of cadmium as an environmental health problem. Toxicol. Appl. Pharmacol. 238 (3), 201–208. Järup, L., Berglund, M., Elinder, C.G., Nordberg, G., Vahter, M., 1998. Health effects of cadmium exposure – A review of the literature and a risk estimate. Scand. J. Work Environ. Health 24 (Suppl. 1), 1–51. Jarvis, M.J., Russell, M.A.H., Benowitz, N.L., Feyerabend, C., 1988. Elimination of cotinine from body fluids: implications for noninvasive measurement of tobacco smoke exposure. Am. J. Public Health 78 (6), 696–698. Kalra, P.A., 2007. Renal specific secondary hypertension. J. Ren. Care 23 (1), 4–10. Kazancioğlu, R., 2013. Risk factors for chronic kidney disease: an update. Kidney Int. Suppl. 3, 368–371. Lea, J.P., Nicholas, S.B., 2002. Diabetes mellitus and hypertension: key risk factors for kidney disease. J. Natl. Med. Assoc. 94 (8 Suppl), 7S–15S. Lee, B.K., Kim, Y., 2012. Association of blood cadmium with hypertension in the Korean general population: analysis of the 2008–2010 Korean national health and nutrition examination survey data. Am. J. Ind. Med. 55 (11), 1060–1067. Lee, B.K., Ahn, J., Kim, N.S., Lee, C.B., Park, J., Kim, Y., 2016. Association of blood pressure with exposure to lead and cadmium: analysis of data from the 2008–2013 Korean National Health and Nutrition Examination Survey. Biol. Trace Elem. Res. 174 (1), 40–51. Lee, D.H., Ha, M.H., Kim, J.R., Jacobs, D.R., 2001. Effects of smoking cessation on changes in blood pressure and incidence of hypertension: a 4-year follow-up study. Hypertension 37 (2), 194–198. McKenzie, J.M., Kay, D.L., 1973. Urinary excretion of cadmium, zinc, and copper in normotensive and hypertensive women. N.Z. Med. J. 78 (495), 68–70. Neutel, C.I., Campbell, N.R.C., 2008. Changes in lifestyle after hypertension diagnosis in Canada. Can. J. Cardiol. 24 (3), 199–204. O'Brien, K.M., Upson, K., Cook, N.R., Weinberg, C.R., 2016. Environmental chemicals in urine and blood: Improving methods for creatinine and lipid adjustment. Environ. Health Perspect. 124 (2), 220–227. Padilla, M.A., Elobeid, M., Ruden, D.M., Allison, D.B., 2010. An examination of the association of selected toxic metals with total and central obesity indices: NHANES 99-02. Int. J. Environ. Res. Public Health 7 (9), 3332–3347. Perry, H.M., Erlanger, M.W., 1974. Metal-induced hypertension following chronic feeding of low doses of cadmium and mercury. J. Lab. Clin. Med. 83 (4), 541–547. Perry, H.M., Erlanger, M.W., Perry, E.F., 1977. Elevated systolic pressure following chronic low-level cadmium feeding. Am. J. Physiol. 232 (2), H114–H121. Perry, H.M., Erlanger, M.W., Perry, E.F., 1979. Increase in the systolic pressure of rates chronically fed cadmium. Environ. Health Perspect. 28, 251–260. Roels, H.A., Lauwerys, R.R., Bernard, A.M., Buchet, J.P., Vos, A., Overstyens, M., 1991. Assessment of the filtration reserve capacity of the kidney in workers exposed to cadmium. Br. J. Inductrial Med. 48 (6), 365–374. Scinicariello, F., Abadin, H.G., Murray, H.E., 2011. Association of low-level blood lead and blood pressure in NHANES 1999–2006. Environ. Res. 111 (8), 1249–1257. SRNT Subcommittee on Biochemical Verification, 2002. Biochemical verification of tobacco use and cessation. Nicotine Tob. Res. 4 (2), 149–159. Statistics Canada, 2015a. Canadian Health Measures Survey (CHMS) Data User Guide: cycle. Statistics Canada, Ottawa, 3. Statistics Canada, 2015b. Instructions for Combining Multiple Cycles of Canadian Health Measures Survey (CHMS) Data. Statistics Canada, Ottawa. Tellez-Plaza, M., Navas-Acien, A., Crainiceanu, C.M., Guallar, E., 2008. Cadmium exposure and hypertension in the 1999–2004 National health and Nutrition Examination survey (NHANES). Environ. Health Perspect. 116 (1), 51–56. United Nations Environment Programme, 2010. Final Review of Scientific Information on Cadmium. Geneva. Wester, P.O., 1973. Trace elements in serum and urine from hypertensive patients before and during treatment with chlorthalidone. Acta Med. Scand. 194 (6), 505–512. Whittemore, A.S., DiCiccio, Y., Provenzano, G., 1991. Urinary cadmium and blood pressure: results from the NHANES II survey. Environ. Health Perspect. 91, 133–140. Wilkins, K., Gee, M., Campbell, N., 2012. The difference in hypertension control between older men and women. Health Rep. 23 (4), 3–10.
5. Conclusions In conclusion, the current large, cross-sectional study provides supporting evidence of an association between cadmium exposure and blood pressure. This association varied by marker of cadmium levels (blood versus urine) and differed by subpopulations. The present study also confirms, as has been previously reported, a significant association between cadmium levels and blood pressure and hypertension among women, particularly among never smokers. A new finding is the possible modification effect of body weight. Such results demonstrate the need for further investigation of this topic, particularly in prospective studies. Funding This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors. Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.envres.2017.01.040. References Agarwal, S., Zaman, T., Tuzcu, E.M., Kapadia, S.R., 2011. Heavy metals and cardiovascular disease: results from the national health and nutrition examination survey (NHANES) 1999–2006. Angiology 62 (5), 422–429. Alessio, L., Berlin, A., Dell'Orto, A., Toffoletto, F., Ghezzi, I., 1985. Reliability of urinary creatinine as a parameter used to adjust values of urinary biological indicators. Int. Arch. Occup. Environ. Health 55, 99–106. Buchet, J.P., Lauwerys, R., Roels, H., Bernard, A., Bruaux, P., Claeys, F., Ducoffre, G., DePlaen, P., Staessen, J., Amery, A., Lijnen, P., Thijs, L., Rondia, D., Sartor, F., Saint Remy, A., Nick, L., 1990. Renal effects of cadmium body burden of the general population. Lancet 336 (8717), 699–702. Bushnik, T., Levallois, P., D’Amour, M., Anderson, T.J., McAlister, F.A., 2014. Association between blood lead and blood pressure: results from the canadian health measures survey (2007–2011). Health Rep. 25 (7), 12–22. CAREX Canada, 2014. Cadmium – Occupational Estimate. 〈www.carexcanada.ca/en/ cadmium/occupational_estimate〉( accessed 11.05.16). CAREX Canada, 2015. Cadmium – Substance profile. 〈www.carexcanada.ca/en/ cadmium/occupational_estimate〉( accessed 11.05.16). Centers for Disease Control and Prevention, 2009. Fourth National Report on Human Exposure to Environmental Chemicals. Department of Health and Human Services, Centers for Disease Control and Prevention, Bethesda, MD. Coresh, J., Selvin, E., Stevens, L.A., Manzi, J., Kusek, J.W., Eggers, P., Van Lente, F., Levey, A.S., 2007. Prevalence of chronic kidney disease in the United States. J. Am. Med. Assoc. 298 (17), 2038–2047. Gallagher, C.M., Meliker, J.R., 2010. Blood and urine cadmium, blood pressure, and hypertension: a systematic review and meta-analysis. Environ. Health Perspect. 118 (2), 1676–1684. Garner, R., Levallois, P., 2016. Cadmium levels and sources of exposure among Canadian adults. Health Rep. 27 (2), 10–18. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. 2009. A review of human carcinogens: Part C – Arsenic, metals, fibres, and dusts. IARC Monographs on the Evaluation of Carcinogenic Risk to Humans, 100C. Lyon, France.
72
Contents lists available at ScienceDirect
Environmental Research journal homepage: www.elsevier.com/locate/envres
Associations between cadmium levels in blood and urine, blood pressure and hypertension among Canadian adults
MARK
⁎
Rochelle E. Garnera, , Patrick Levalloisb,c a b c
Health Analysis Division, Statistics Canada, Ottawa, Ontario, Canada Direction de la santé environnementale et de la toxicologie, Institut National de Santé Publique du Québec, Québec City, Québec, Canada Axe santé des populations et pratiques optimales en santé, Centre de Recherche du CHU de Québec-Université Laval, Québec City, Québec, Canada
A R T I C L E I N F O
A BS T RAC T
Keywords: Cadmium Blood pressure Hypertension Smoking Body mass index Canadian Health Measures Survey
Background: Cadmium has been inconsistently related to blood pressure and hypertension. The present study seeks to clarify the relationship between cadmium levels found in blood and urine, blood pressure and hypertension in a large sample of adults. Methods: The study sample included participants ages 20 through 79 from multiple cycles of the Canadian Health Measures Survey (2007 through 2013) with measured blood cadmium (n=10,099) and urinary cadmium (n=6988). Linear regression models examined the association between natural logarithm transformed cadmium levels and blood pressure (separate models for systolic and diastolic blood pressure) after controlling for known covariates. Logistic regression models were used to examine the association between cadmium and hypertension. Models were run separately by sex, smoking status, and body mass index category. Results: Men had higher mean systolic (114.8 vs. 110.8 mmHg, p < 0.01) and diastolic (74.0 vs. 69.6 mmHg, p < 0.01) blood pressure compared to women. Although, geometric mean blood (0.46 vs. 0.38 µg/L, p < 0.01) and creatinine-adjusted standardized urinary cadmium levels (0.48 vs. 0.38 µg/L, p < 0.01) were higher among those with hypertension, these differences were no longer significant after adjustment for age, sex and smoking status. In overall regression models, increases in blood cadmium were associated with increased systolic (0.70 mmHg, 95% confidence interval [CI]=0.25–1.16, p < 0.01) and diastolic blood pressure (0.74 mmHg, 95% CI=0.30–1.19, p < 0.01). The associations between urinary cadmium, blood pressure and hypertension were not significant in overall models. Model stratification revealed significant and negative associations between urinary cadmium and hypertension among current smokers (OR=0.61, 95% CI=0.44–0.85, p < 0.01), particularly female current smokers (OR=0.52, 95% CI=0.32–0.85, p=0.01). Conclusion: This study provides evidence of a significant association between cadmium levels, blood pressure and hypertension. However, the significance and direction of this association differs by sex, smoking status, and body mass index category.
1. Introduction Cadmium is a heavy metal that occurs naturally in the environment. Whereas certain occupational subgroups are at elevated risk of cadmium exposure, including sawmill and wood preservation workers, auto repair workers, and welders (CAREX Canada, 2014), nonoccupational exposure to cadmium is generally the result of cigarette smoking and ingestion of cadmium-high foods (CAREX Canada, 2015; Garner and Levallois, 2016). The International Agency for Research on Cancer considers there to be sufficient evidence of the carcinogenic effect of cadmium and its compounds among humans (IARC Working Group on the Evaluation of Carcinogenic Risks to Humans, 2009).
⁎
Furthermore, other evidence indicates that prolonged or increased exposure to cadmium could have significant health consequences, including kidney dysfunction, skeletal damage, and possible cardiovascular effects (Järup et al., 1998; United Nations Environment Programme, 2010). Blood levels of cadmium are valid markers of recent exposure (Järup and Åkesson, 2009; Järup, 2003). However, even after exposure cessation, cadmium levels in the blood can remain elevated, and are therefore also reflective of long-term exposure (Järup et al., 1998). Urinary cadmium levels reflect long-term (cumulative) exposure as well as cadmium levels in the kidney (Centers for Disease Control and Prevention, 2009, p. 202; Järup, 2003). Animal model studies have demonstrated that higher levels of
Correspondence to: Health Analysis Division, Statistics Canada, 100 Tunney's Pasture Driveway, RH Coats Building, 24th floor, Ottawa, Ontario, Canada K1A 0T6. E-mail address: [email protected] (R.E. Garner).
http://dx.doi.org/10.1016/j.envres.2017.01.040 Received 3 October 2016; Received in revised form 5 January 2017; Accepted 31 January 2017 0013-9351/ Crown Copyright © 2017 Published by Elsevier Inc. All rights reserved.
Environmental Research 155 (2017) 64–72
R.E. Garner, P. Levallois
2.2. Measures
cadmium are associated with increased blood pressure and hypertension (Perry and Erlanger, 1974; Perry et al., 1977, 1979; United Nations Environment Programme, 2010). Observational studies, however, have found mixed results. Studies vary in the direction or significance that cadmium body stores have on blood pressure or hypertension, with results seeming to vary by type of cadmium biomonitoring sample (blood vs urine), sex, smoking status, ethnicity and medication levels (Tellez-Plaza et al., 2008; Whittemore et al., 1991; Agarwal et al., 2011; Scinicariello et al., 2011; Lee and Kim, 2012; Lee et al., 2016). In their recent systematic review and metaanalysis of the associations between blood and urine cadmium, blood pressure and hypertension, Gallagher and Meliker (2010) concluded that there was evidence of a positive association between blood cadmium (BCd) and blood pressure among women, but not among men. Furthermore, their analysis indicated that there was an inverse relationship between urinary cadmium (UCd) and the likelihood of hypertension. However, Gallagher and Meliker pointed out that few studies included representative, population-based samples of never smokers, and that previous studies have varied in their outcome definitions, thereby limiting interpretation of meta-analysis findings (Gallagher and Meliker, 2010). The present study seeks to add to the existing literature by examining the association between cadmium levels of adult Canadians and measurements of blood pressure and the presence of hypertension. This study will examine the association of cadmium levels in both blood and urine, and will also examine whether the significance or magnitude of any association between cadmium and blood pressure differs according to sex or smoking status.
2.2.1. Cadmium measures Blood and urine samples were collected from all eligible respondents at the mobile examination clinic. Blood samples were collected by a phlebotomist using a standardized venipuncture technique. These were subsequently centrifuged and aliquoted into smaller tubes. Respondent-provided urine samples were also aliquoted into smaller tubes. Urine samples were collected using the first catch urine in cycle 2, as opposed to the mid-stream urine collected in cycle 1. Cycle 2 respondents were also asked to abstain from urinating two hours prior to their appointment. This may have resulted in a shift in creatinine levels when multiple cycles are compared. This, in turn, could affect creatinine-adjusted levels of cadmium. Blood and urine sample tubes were placed in shipping trays and stored in the mobile examination clinic laboratory's refrigerator or freezer, depending on the test. All specimens were stored as soon as processing was complete to ensure the quality of the samples’ integrity, which was achieved within two hours from the time of collection for most samples: a four hour ceiling from the point of collection was placed on the time for blood samples to be processed and stored (Statistics Canada, 2015a). The limit of detection (LOD) for BCd was 0.045 µg/L (0.4 nmol/L) in cycles 1 and 2, and 0.080 µg/L (0.71 nmol/L) in cycle 3. The LOD for UCd was 0.09 µg/L (0.8 nmol/L) for cycle 1 and 0.07 µg/L (0.6 nmol/ L) for cycle 2. Consistent with others who suggest excluding urine samples that are too dilute (Alessio et al., 1985), urine samples with creatinine levels below the LOD (0.31 mmol/L for cycle 1, 0.44 mmol/ L for cycle 2) were excluded from analyses (n=10). For remaining samples, observations below the LOD for the particular measure were imputed with values at half the LOD. When the LOD for a particular test differed between cycles, the highest LOD was used (BCd, 0.08 µg/ L; UCd, 0.09 µg/L). All values were rounded to two significant digits (Statistics Canada, 2015b), and were converted from Système International units to conventional units for the study. For UCd analyses, it was necessary to take into account the dilution level of the urine sample. This is generally accomplished by adjusting for urinary creatinine levels. Greater attention is being paid to the method used to adjust urinary biomarkers for creatinine, and the analytic impact that adjustment has. The present study used creatinine-adjusted standardization, an adjustment technique recommended by O'Brien et al. (2016). First, a model was fit for the natural logarithm of creatinine as a function of age, sex, body mass index (BMI) category, use of anti-hypertensive medication, diabetes diagnosis, presence of chronic kidney disease, and CHMS cycle. Next, the observed creatinine level was divided by the predicted creatinine level (exponentiated to yield the original metric) to generate a creatinine ratio. Finally, the observed UCd level was divided by this creatinine ratio to yield a creatinine-adjusted standardized UCd measure (CAS-UCd).
2. Materials and methods 2.1. Data source Data for the present study were drawn from multiple cycles of the Canadian Health Measures Survey (CHMS), which collects both survey-based information as well as direct physical health measures for Canadians. The present study pools information from Cycle 1 (2007–2009) and Cycle 2 (2009–2011) for analyses concerning UCd measures, and further includes Cycle 3 (2012–2013) data for analyses concerning BCd measures: UCd was not measured in Cycle 3 of the CHMS. The CHMS includes the population ages 3 through 79 (6 through 79 in Cycle 1) living in the ten provinces, but excludes the following individuals: persons living in the three territories; persons living on reserves and other Aboriginal settlements in the provinces; full-time members of the Canadian Forces; the institutionalized population and residents of certain remote regions. These exclusions are thought to represent approximately 4% of the target population (Statistics Canada, 2015a). The study sample was limited to respondents 20–79 years of age, and excluded pregnant women (n=106) and those who were missing relevant measures of cadmium (blood, urine) or blood pressure. This left a final sample of 10,099 respondents for BCd analyses, and 6988 respondents for UCd analyses.
2.2.2. Blood pressure and hypertension A series of blood pressure readings were taken at one-minute intervals for each respondent, with the last five readings being averaged to determine systolic (SBP) and diastolic blood pressure (DBP) measures. Respondents with any of the following characteristics were classified as hypertensive: (i) SBP≥140 mmHg; (ii) DBP≥90 mmHg; (iii) self-reported doctor-diagnosed high blood pressure; or (iv) use of anti-hypertensive medications, which was determined based on the Anatomical Therapeutic Chemical (ATC) code assigned to respondents’ medications. Based on previous work with CHMS data (Bushnik et al., 2014; Wilkins et al., 2012), codes indicative of anti-hypertensive medications were: beta blockers (ATC codes C07, excluding C07AA07, C07AA12 and C07AG02); agents acting on the reninangiotensin system (ATC codes C09); thiazide diuretics (ATC codes C03, excluding C03BA08 and C03CA01); calcium channel antagonists (ATC codes C08); and miscellaneous anti-hypertensives (ATC codes C02, excluding C02KX01).
2.1.1. Ethics approval for the CHMS All processes of the CHMS were reviewed and approved by the Research Ethics Boards of two federal agencies (Health Canada, Public Health Agency of Canada) to ensure that internationally recognized ethical standards for human research were met and maintained. In addition, protocols were developed through extensive consultation with recognized experts and were performed by accredited health professionals in conformance with universal precautions. 65
Environmental Research 155 (2017) 64–72
R.E. Garner, P. Levallois
Table 1 Characteristics of respondents aged 20 through 79 years old, Canadian Health Measures Survey, 2007–2013.
Continuous variables Age Systolic blood pressure (mmHg) Diastolic blood pressure (mmHg) Body mass index, measured (kg/m2) Blood cadmium (µg/L) Urinary cadmium, creatinine-adjusted standardized (µg/L) Categorical variables Age 20–39 40–59 60–79 Has hypertension No Yes Anti-hypertensive medication use No Yes Smoking status Never smoker Former smoker Current smoker Body mass index category, measured Underweight Acceptable weight Overweight Obese Drinking status Never drinker Former drinker Current drinker Diabetes No Yes Chronic kidney disease No Yes Daily or near-daily exposure to second hand smoke No Yes Cycle Cycle 1 Cycle 2 Cycle 3
p-value*
Overall (n=10,099)
Males (n=4850)
Female (n=5249)
Meana 46.0 112.8 71.8 27.3 0.40 0.41 n
SE 0.1 0.4 0.3 0.2 0.01 0.01 %
Meana 45.5 114.8 74.0 27.6 0.35 0.41 n
SE 0.2 0.5 0.4 0.2 0.01 0.01 %
Meana 46.5 110.8 69.6 27.1 0.45 0.41 n
SE 0.1 0.4 0.3 0.2 0.01 0.01 %
< 0.01 < 0.01 < 0.01 0.01 < 0.01 0.74
3404 3504 3191
36.9 41.1 22.1
1567 1732 1551
37.9 40.9 21.1
1837 1772 1640
35.8 41.2 23.0
< 0.01 0.26 < 0.01
7201 2898
74.7 25.3
3354 1496
72.4 27.6
3847 1402
76.9 23.1
< 0.01
7894 2205
81.4 18.6
3752 1098
80.6 19.4
4142 1107
82.3 17.7
0.19
4845 2972 2264
48.8 27.3 23.9
2070 1525 1241
44.4 28.4 27.2
2775 1447 1023
53.2 26.2 20.6
< 0.01 0.16 < 0.01
130 3451 3746 2753
1.6 36.6 36.2 25.6
31 1341 2166 1305
0.8b 30.0 43.0 26.2
99 2110 1580 1448
2.4b 43.2 29.4 25.0
< 0.01 < 0.01 < 0.01 0.43
598 1096 8402
6.5 10.0 83.5
182 476 4192
4.4 9.4 86.2
416 620 4210
8.6 10.7 80.7
< 0.01 0.26 < 0.01
9190 909
91.9 8.1
4355 495
90.5 9.5
4835 414
93.3 6.7
< 0.01
9338 761
94.1 5.9
4525 325
94.7 5.3
4813 436
93.4 6.6
0.05
7446 2653
69.6 30.4
3447 1403
66.3 33.7
3999 1250
72.9 27.1
< 0.01
3420 3565 3114
33.4 41.4 25.2
1622 1673 1555
33.4 41.4 25.2
1798 1892 1559
33.4 41.4 25.2
0.91 0.99 0.92
Note. All estimates in table are based on pooled analysis of CHMS cycles 1 through 3, expect urinary cadmium levels which are based on cycles 1 and 2 only. a Estimates are geometric mean values for measures of cadmium, arithmetic means otherwise. b Due to a moderate coefficient of variation (between 0.167 and 0.333), the estimate should be used with caution. * p-value for comparison between male and female estimates based on t-tests for continuous variables and chi-square tests for categorical variables.
among men (n=31), this category could not be retained as a separate group for stratified models, though underweight respondents were retained in other models. Alcohol consumption status was based on self-report and was categorized as never, former and current drinkers. The presence of diabetes was based on respondents’ self-report of a diagnosis by a doctor. Chronic kidney disease was based on either self-report or was defined as an estimated glomerular filtration rate (GFR) of less than 60 mL/min/1.73 m2. GFR was estimated as 1.75*(serum creatinine in mg/dL)−1.154*(age)−0.203*(0.742 if female)*(1.212 if cultural or racial background is black) (Coresh et al., 2007). Lastly, in regression models, dummy variables were used as indicators of the CHMS Cycle in which the respondent participated: Cycle 2 was the reference category.
2.2.3. Other factors Respondents’ smoking status was based on self-report and was categorized as never smokers, former smokers, and current smokers. Individuals who self-reported as never smokers or former smokers with urinary cotinine levels greater than 50 ng/mL were re-classified as current smokers (n=57 and n=216, respectively; Jarvis et al., 1988; SRNT Subcommittee on Biochemical Verification, 2002). Second hand smoke exposure was dichotomized to identify individuals who reported being exposed on a daily or near-daily basis: current smokers were asked to exclude their own smoking when reporting second hand smoke exposure. An individual's BMI was based on their measured height and weight, and is calculated as weight in kilograms divided by height in metres squared. In some analyses, to minimize the number of model covariates and preserve degrees of freedom, BMI was used as a continuous measurement, centred at 21.75 (the mid-point of the acceptable weight category). In stratified analyses, BMI was used categorically as acceptable weight (18.5≤BMI < 25), overweight (25≤BMI < 30), and obese (BMI≥30). Because so few respondents were classified as underweight (BMI < 18.5, n=130), particularly
2.3. Analysis The association between cadmium levels and other covariates with SBP, DBP and hypertension were first examined descriptively, then were assessed using regressions models. Linear regression models were used to examine the association between cadmium levels and blood 66
Environmental Research 155 (2017) 64–72
R.E. Garner, P. Levallois
models run for the sample overall, as well as those run separately by sex, smoking status and BMI category. In the overall sample, increasing BCd levels were significantly associated with higher blood pressure (SBP and DBP; Figs. 1 and 2), whereas CAS-UCd levels were not. In smoking-stratified models, BCd levels were significantly and positively associated with SBP among former smokers (Fig. 1) and with DBP among never smokers (Fig. 2). In BMI-stratified models, higher BCd levels were associated with significantly higher blood pressure among those who were overweight: associations among those who were of acceptable weight or who were obese were not statistically significant (Figs. 1 and 2). Among those who were classified as obese, higher levels of CAS-UCd were associated with statistically lower SBP (Fig. 3) and DBP (Fig. 4): CAS-UCd showed no association with blood pressure among non-obese. Further stratifying models by sex revealed certain important differences, particularly in smoking- and BMI-stratified models. Most notably, the difference in the association between CAS-UCd and blood pressure among current smokers, and between BCd and blood pressure among those who were overweight. Among respondents classified as overweight, there was a statistically significant and positive association between men's BCd levels and their SBP (Fig. 1) and DBP (Fig. 2), whereas the association was effectively null among overweight women. Examining the association between CAS-UCd among current smokers, there was no association between CAS-UCd levels and blood pressure for men, whereas there was a significant and negative association among women, whereby higher levels of CAS-UCd among current smoking women were associated with significantly lower blood pressure. Exclusion of CHMS respondents taking anti-hypertensive medications did not generally alter the association between CAS-UCd and blood pressure (see Supplemental Table A1). Point estimates were generally very similar between models run on samples including and excluding respondents who were currently using anti-hypertensive medications, although variance around the estimates was usually broader in the latter. In some cases, this increased variance resulted in a loss of statistical significance upon exclusion of respondents currently taking anti-hypertensive medications. Conversely, the association between BCd and blood pressure among current smokers became statistically significant and positive in direction after excluding treated hypertensives (SBP: beta=1.10, p=0.02; DBP: beta=0.78, p=0.02; Supplemental Table A1).
pressure (SBP and DBP), while logistic regression models were used in the case of hypertension. Cadmium levels were transformed using the natural logarithm and were entered as a continuous measure in regression models. Models were first run for the sample overall, then were run separately by sex and smoking status. Regression models (linear and logistic) were adjusted for the following factors: age (linear and quadratic effects), sex (in overall model only), smoking status (never, former, current), use of antihypertensive medications (full sample only), BMI (continuous, centred at 21.75), alcohol consumption status (never drinker, former drinker, current drinker), presence of diabetes, presence of chronic kidney disease, daily or near-daily exposure to second hand smoke, and indicator(s) for CHMS cycle. During the model building process, it was noted that the inclusion of BMI in models sometimes affected the cadmium estimates. Therefore, models were also run separately by BMI category (acceptable weight, overweight, obese). While interactions between cadmium levels and the stratifying variables were not always statistically significant in the overall model, stratification was deemed an appropriate strategy as it allowed the impact of all covariates to differ by the stratifying variable, not just the cadmium measure. Given the limited degrees of freedom available in analyses, it was not possible to include all potential interactions in one model. Therefore, while stratification does not allow for a direct comparison of the impact of cadmium across models, it does allow for an examination of potential differences in effect across levels of the stratifying variable, i.e. sex, smoking status, and BMI. Analyses pooling cycles 1 through 3 used 35 denominator degrees of freedom, whereas analyses pooling only cycles 1 and 2 used 24 degrees of freedom. All analyses were conducted in SAS-callable SUDAAN (version 11.0.1), using the appropriate pooled sample weights, and were bootstrapped to account for the sample design (Statistics Canada, 2015b). 3. Results 3.1. Descriptive analyses Based on the pooled sample of CHMS cycle 1 through 3 respondents, the respondents were evenly split between males and females (50.2% vs. 49.8%). The average age of respondents was 46.0 years, with females being a year older on average than males in the sample (Table 1). Men had significantly higher SBP and DBP, and were more likely to be hypertensive, compared to women (Table 1). Women had higher average BCd levels than men, where CAS-UCd levels were similar between males and females (Table 1).
3.4. Multivariate analyses: hypertension Regarding hypertension, the overall models showed no statistically significant association between hypertension and levels of BCd (Fig. 5) or CAS-UCd (Fig. 6), although the change in the odds of hypertension associated with increasing levels of BCd just marginally failed to reach statistical significance (OR=1.13, 95% CI=0.99–1.29, p=0.06). In smoking-stratified models, there was no association between hypertension and BCd levels. However, there was a significant and negative association between CAS-UCd levels and hypertension among current smokers (OR=0.61, 95% CI=0.44–0.85, p < 0.01). Stratification by BMI did not reveal further associations overall. All ORs and 95% confidence intervals are presented in Supplemental Table A2. Once again, further stratifying models by sex revealed different associations between cadmium levels and hypertension. Although none of the sex-specific estimates for the association of BCd levels with the hypertension were statistically significant (Fig. 5), point estimates varied between men and women among former and current smokers, and among those of acceptable weight. Examining the estimates associated with CAS-UCd, higher cadmium levels among female current smokers were associated with a lower odds of having hypertension (OR=0.52, 95% CI=0.32–0.85, p=0.01), whereas the association was null for males (OR=0.67, 95% CI=0.41–1.08, p=0.10; Fig. 6).
3.2. Crude and adjusted cadmium levels by sample characteristics Individuals with high SBP and DBP had significantly higher levels of CAS-UCd than those with lower levels of blood pressure (Table 2). Levels of BCd were also significantly higher among those with high SBP compared to those with low SBP, whereas this association was not found for DBP. However, differences in cadmium levels were no longer statistically significant after adjustment for age, sex and smoking status (Table 2). However, cadmium levels did differ statistically, even after adjustment, by smoking status, BMI category (BCd only), drinking status, and diabetes (CAS-UCd only). Furthermore, after adjustment, respondents taking anti-hypertensive medications had significantly lower CAS-UCd levels compared to those not taking such medications: this association was not significant for BCd levels (Table 2). 3.3. Multivariate analyses: blood pressure Figs. 1 through 4 show the estimated beta coefficients and their 95% confidence intervals for the associations between natural logarithm transformed cadmium levels and blood pressure from regression 67
Environmental Research 155 (2017) 64–72
R.E. Garner, P. Levallois
Table 2 Crude and adjusted geometric mean cadmium levels (blood, urinary creatinine-adjusted standardized) by select respondent characteristics, Canadian Health Measures Survey. Blood cadmium (µg/L), Cycles 1 through 3 Adjusteda
Crude Factor
Sex Male Female Age 20–39 40–59 60–79 Systolic blood pressure Low High Diastolic blood pressure Low High Anti-hypertensive medication use No Yes Hypertension No Yes Smoking status Never smoker Former smoker Current smoker Body mass index category, measured Underweight Acceptable weight Overweight Obese Drinking status Never drinker Former drinker Current drinker Diabetes No Yes Chronic kidney disease No Yes Daily or near-daily exposure to second hand smoke No Yes Cycle Cycle 1 Cycle 2 Cycle 3 a b
Urinary cadmium, creatinine-adjusted standardized (µg/L), Cycles 1 and 2
Geo. mean
SE
p
Geo. mean
0.35 0.45
0.01 0.01
Ref. < 0.01
0.31 0.46 0.46
0.01 0.01 0.01
0.39 0.49
SE
Adjusteda
Crude p
Geo. mean
SE
p
Geo. mean
SE
p
n/a n/a
0.41 0.41
0.01 0.01
Ref. 0.75
n/a n/a
< 0.01 Ref. 0.92
n/a n/a n/a
0.31 0.44 0.53
0.01 0.01 0.02
< 0.01 Ref. < 0.01
n/a n/a n/a
0.01 0.03
Ref. < 0.01
0.40 0.41
0.01 0.02
Ref. 0.62
0.40 0.53
0.01 0.02
Ref. < 0.01
0.41 0.43
0.01 0.02
Ref. 0.20
0.40 0.41
0.01 0.04
Ref. 0.62
0.40 0.44
0.01 0.04
Ref. 0.20
0.40 0.50
0.01 0.03
Ref. < 0.01
0.40 0.46
0.01 0.03
Ref. 0.05
0.38 0.46
0.01 0.02
Ref. < 0.01
0.40 0.40
0.01 0.01
Ref. 0.86
0.39 0.48
0.01 0.01
Ref. < 0.01
0.41 0.38
0.01 0.01
Ref. 0.01
0.38 0.46
0.01 0.02
Ref. < 0.01
0.40 0.40
0.01 0.01
Ref. 0.59
0.38 0.48
0.01 0.01
Ref. < 0.01
0.41 0.40
0.01 0.01
Ref. 0.66
0.22 0.33 1.63
0.01 0.01 0.06
Ref. < 0.01 < 0.01
0.22 0.30 1.73
0.01 0.01 0.06
Ref. < 0.01 < 0.01
0.33 0.45 0.56
0.01 0.01 0.02
Ref. < 0.01 < 0.01
0.34 0.41 0.59
0.01 0.01 0.02
Ref. < 0.01 < 0.01
0.61b 0.42 0.38 0.38
0.11 0.02 0.02 0.01
0.05 Ref. 0.16 0.12
0.47 0.43 0.38 0.37
0.04 0.01 0.01 0.01
0.32 Ref. 0.02 0.01
0.35 0.39 0.42 0.43
0.04 0.01 0.01 0.01
0.43 Ref. 0.02 0.03
0.39 0.41 0.40 0.40
0.04 0.01 0.01 0.01
0.68 Ref. 0.58 0.58
0.34 0.47 0.39
0.03 0.03 0.01
Ref. < 0.01 0.14
0.53 0.43 0.38
0.04 0.02 0.01
Ref. 0.01 < 0.01
0.46 0.51 0.39
0.03 0.03 0.01
Ref. 0.21 0.03
0.56 0.47 0.39
0.04 0.03 0.01
Ref. 0.03 < 0.01
0.39 0.44
0.01 0.03
Ref. 0.21
0.40 0.37
0.01 0.01
Ref. 0.08
0.40 0.56
0.01 0.02
Ref. < 0.01
0.40 0.47
0.01 0.02
Ref. < 0.01
0.39 0.51
0.01 0.04
Ref. < 0.01
0.39 0.45
0.01 0.02
Ref. < 0.01
0.40 0.56
0.01 0.03
Ref. < 0.01
0.40 0.45
0.01 0.03
Ref. 0.04
0.33 0.60
0.01 0.03
Ref. < 0.01
0.38 0.43
0.01 0.01
Ref. < 0.01
0.39 0.44
0.01 0.01
Ref. < 0.01
0.40 0.43
0.01 0.01
Ref. 0.03
0.41 0.37 0.42
0.02 0.02 0.02
0.10 Ref. 0.10
0.41 0.38 0.41
0.01 0.02 0.01
0.16 Ref. 0.16
0.38 0.43
0.01 0.02
Ref. < 0.01
0.37 0.44
0.01 0.02
Ref. < 0.01
Cadmium levels are adjusted for age, sex, and smoking status; estimates by smoking status categories are only adjusted for age and sex Due to a moderate coefficient of variation (between 0.167 and 0.333), the estimate should be used with caution
only among current smokers, particularly among women. The mechanism by which cadmium is thought to affect blood pressure is not clearly understood, although one hypothesized mechanism is via kidney damage (Buchet et al., 1990; Roels et al., 1991). Individuals with decreased kidney function are at increased risk of hypertension (Kalra, 2007). In addition, kidney damage may also be a sequelae of hypertension (Kazancioğlu, 2013; Lea and Nicholas, 2002). The current study included an indicator of chronic kidney disease in regression models to account for this relationship. In fact, the current study may have over-adjusted the relationship between cadmium and blood pressure by including a measure of decreased kidney function. However, inclusion of this indicator did not appear to significantly alter this relationship. A novel finding from the present study was the differential association between cadmium levels, blood pressure and hypertension by BMI category. While higher BMI was associated with higher blood
4. Discussion After controlling for other pertinent factors, higher BCd levels were associated with higher blood pressure levels (SBP and DBP) and a greater likelihood of hypertension, whereas there was no association for CAS-UCd levels overall. Subgroup analyses indicated that CAS-UCd levels were negatively associated with blood pressure and the likelihood of hypertension among currently smoking women. In their meta-analysis of the existing literature, Gallagher and Meliker (2010) concluded that there was evidence of a positive association between BCd and blood pressure among women, and of an inverse association between UCd and hypertension. The present study found a positive relationship between BCd and blood pressure overall and a similar association, although not statistically significant, in sex-specific analyses. Regarding the association between UCd and hypertension, an inverse relationship was found in the present study 68
Environmental Research 155 (2017) 64–72
R.E. Garner, P. Levallois
Fig. 1. Beta coefficients (95% confidence intervals) showing change in systolic blood pressure associated with unit change in natural logarithm transformed blood cadmium from linear regression models, overall and by sex, smoking status, and body mass index category.
Fig. 2. Beta coefficients (95% confidence intervals) showing change in diastolic blood pressure associated with unit change in natural logarithm transformed blood cadmium from linear regression models, overall and by sex, smoking status, and body mass index category.
cadmium exposure sources may explain these differences. These nonsmoking exposure sources may be associated with long-term behaviours, as the BMI-specific associations with blood pressure were found only for UCd, which is an indicator of overall body burden of cadmium, not BCd, which is more sensitive to short-term exposure. In their analysis of NHANES II data, Whittemore et al. (1991) found that a significant and positive association between UCd and blood pressure was rendered non-significant after respondents receiving anti-hypertensive treatment were excluded from the sample. This exclusion was based on the premise that use of certain anti-hypertensive medications increased the urinary excretion of cadmium (McKenzie and Kay, 1973; Wester, 1973). In the current study, exclusion of CHMS respondents taking anti-hypertensive medications did not generally alter the association between CAS-UCd and blood pressure, although some of the measures of association had wider confidence intervals. Given the cross-sectional nature of CHMS data, it
pressure and a greater probability of hypertension in the present study, results from the sex- and BMI-stratified models showed that, within BMI categories, higher CAS-UCd levels were associated with significantly higher blood pressure levels among overweight (DBP only) and obese women (SBP and DBP). These associations are controlled for the impact of smoking status. Others have shown that levels of cadmium decrease with increasing BMI (Garner and Levallois, 2016; Padilla et al., 2010), a relationship that has been shown with other heavy metals as well (Padilla et al., 2010). However, to our knowledge, other studies have not reported differential associations between cadmium levels and blood pressure measures by BMI categories. Although BMI was included as a covariate in the adjustment and standardization of UCd levels for creatinine, other BMI-related factors that may affect the excretion of cadmium may be responsible for the differential association of CAS-UCd across BMI categories. Furthermore, given that BMIstratified models controlled for the effect of smoking status, other
69
Environmental Research 155 (2017) 64–72
R.E. Garner, P. Levallois
Fig. 3. Beta coefficients (95% confidence intervals) showing change in systolic blood pressure associated with unit change in natural logarithm transformed urinary cadmium (creatinine-adjusted standardized) from linear regression models, overall and by sex, smoking status, and body mass index category.
Fig. 4. Beta coefficients (95% confidence intervals) showing change in diastolic blood pressure associated with unit change in natural logarithm transformed urinary cadmium (creatinine-adjusted standardized) from linear regression models, overall and by sex, smoking status, and body mass index.
smoking behaviours, they represent different exposure timings. BCd levels among current smokers likely represent current exposure, proportional to the number of cigarettes smoked (Järup et al., 1998). BCd levels will decrease rapidly upon smoking cessation, although levels may not drop to pre-exposure levels (Järup et al., 1998). Therefore, BCd levels among formers smokers may reflect time since cessation. It has been shown that blood pressure levels will increase upon smoking cessation, and that they may remain elevated even as time since cessation increases (Lee et al., 2001). An association between cadmium levels and hypertension was found only with UCd, and only among current smokers. In this group, increased CAS-UCd was associated with a decrease in the likelihood of having hypertension (OR=0.61, 95% CI=0.44–0.85; Fig. 6 and Supplemental Table A2). This finding is consistent with the association found with blood pressure in this study, particularly DBP. It may also be the result of a “healthy smoker” effect, whereby current smokers are those who are sufficiently healthy to continue to smoke. For example, individuals may cease smoking, and therefore become former smokers,
is not clear the direction of association between cadmium levels and the use of anti-hypertensive medications. For example, are those with higher cadmium levels more likely to require the use of anti-hypertensive medications? Or, conversely, does the use of anti-hypertensive medications increase an individuals’ cadmium levels? In the present study, respondents taking anti-hypertensive medications had significantly higher BCd (0.46 µg/L vs. 0.38 µg/L; p < 0.01) and CAS-UCd levels (0.48 µg/L vs. 0.39 µg/L; p < 0.01) compared to those not taking anti-hypertensive medications (Table 2). Excluding those using antihypertensive medications would mean excluding a large proportion of respondents (18.6% overall), as well as those at the higher end of the cadmium levels. Therefore, exclusion of those on such medications may be unnecessarily restrictive. Increasing BCd levels among former smokers was associated with an increase in SBP. Although this association was not initially statistically significant among current smokers, it became significant after exclusion of those taking anti-hypertensive medications. Although BCd levels among former and current smokers are likely a result of 70
Environmental Research 155 (2017) 64–72
R.E. Garner, P. Levallois
Fig. 5. Odds ratios (95% confidence intervals) of association between natural logarithm transformed blood cadmium and hypertension from logistic regression models, by sex, smoking status, and body mass index category.
ability to determine causality. Second, the method used in the present study to adjust UCd levels for creatinine differed from that used in other studies, thereby potentially affecting the ability to compare the current results to those from other studies in the literature. However, the adjusted and standardized method has been shown to produce the least biased results in simulation studies (O'Brien et al., 2016). Lastly, the response rates of the combined cycles were low: 52.9% for cycles 1 through 3, and 53.5% for cycles 1 and 2. This largely reflects a low response rate among households initially contacted for participation; households that agreed to provide a household roster had high response rates (more than 78%) to the interview and mobile examination centre components of the survey. It is not known if households that initially declined to participate differed in important ways from participating households, or how this may impact the current study's results.
after a negative health event or diagnosis, including the diagnosis of hypertension (Neutel and Campbell, 2008). Therefore, those with morbidity are shifted from the exposed current smoker group and moved into the less healthy, but less exposed, former smoker category. Furthermore, others have shown that smoking quitters have a higher risk of incident hypertension compared to current smokers (Lee et al., 2001), which is also consistent with our current study. A strength of the present study is its ability to examine both BCd and UCd levels in a population-based sample. It should be noted, however, that the analytic sample was larger for the BCd analyses as UCd was not collected in cycle 3 of the CHMS. Furthermore, Gallagher and Meliker (2010) had previously noted in their meta-analysis that the literature was lacking population-based studies of never-smokers: the present study serves to reduce that literature gap by reporting results among never, former and current smokers. Despite its strengths, the present study has certain limitations that should be considered. First, as previously noted, the cross-sectional nature of the data precludes the
Fig. 6. Odds ratios (95% confidence intervals) of association between natural logarithm transformed urinary cadmium (creatinine-adjusted standardized) and hypertension from logistic regression models, by sex, smoking status, and body mass index category.
71
Environmental Research 155 (2017) 64–72
R.E. Garner, P. Levallois
Järup, L., 2003. Hazards of heavy metal contamination. Br. Med. Bull. 68 (1), 167–182. Järup, L., Åkesson, A., 2009. Current status of cadmium as an environmental health problem. Toxicol. Appl. Pharmacol. 238 (3), 201–208. Järup, L., Berglund, M., Elinder, C.G., Nordberg, G., Vahter, M., 1998. Health effects of cadmium exposure – A review of the literature and a risk estimate. Scand. J. Work Environ. Health 24 (Suppl. 1), 1–51. Jarvis, M.J., Russell, M.A.H., Benowitz, N.L., Feyerabend, C., 1988. Elimination of cotinine from body fluids: implications for noninvasive measurement of tobacco smoke exposure. Am. J. Public Health 78 (6), 696–698. Kalra, P.A., 2007. Renal specific secondary hypertension. J. Ren. Care 23 (1), 4–10. Kazancioğlu, R., 2013. Risk factors for chronic kidney disease: an update. Kidney Int. Suppl. 3, 368–371. Lea, J.P., Nicholas, S.B., 2002. Diabetes mellitus and hypertension: key risk factors for kidney disease. J. Natl. Med. Assoc. 94 (8 Suppl), 7S–15S. Lee, B.K., Kim, Y., 2012. Association of blood cadmium with hypertension in the Korean general population: analysis of the 2008–2010 Korean national health and nutrition examination survey data. Am. J. Ind. Med. 55 (11), 1060–1067. Lee, B.K., Ahn, J., Kim, N.S., Lee, C.B., Park, J., Kim, Y., 2016. Association of blood pressure with exposure to lead and cadmium: analysis of data from the 2008–2013 Korean National Health and Nutrition Examination Survey. Biol. Trace Elem. Res. 174 (1), 40–51. Lee, D.H., Ha, M.H., Kim, J.R., Jacobs, D.R., 2001. Effects of smoking cessation on changes in blood pressure and incidence of hypertension: a 4-year follow-up study. Hypertension 37 (2), 194–198. McKenzie, J.M., Kay, D.L., 1973. Urinary excretion of cadmium, zinc, and copper in normotensive and hypertensive women. N.Z. Med. J. 78 (495), 68–70. Neutel, C.I., Campbell, N.R.C., 2008. Changes in lifestyle after hypertension diagnosis in Canada. Can. J. Cardiol. 24 (3), 199–204. O'Brien, K.M., Upson, K., Cook, N.R., Weinberg, C.R., 2016. Environmental chemicals in urine and blood: Improving methods for creatinine and lipid adjustment. Environ. Health Perspect. 124 (2), 220–227. Padilla, M.A., Elobeid, M., Ruden, D.M., Allison, D.B., 2010. An examination of the association of selected toxic metals with total and central obesity indices: NHANES 99-02. Int. J. Environ. Res. Public Health 7 (9), 3332–3347. Perry, H.M., Erlanger, M.W., 1974. Metal-induced hypertension following chronic feeding of low doses of cadmium and mercury. J. Lab. Clin. Med. 83 (4), 541–547. Perry, H.M., Erlanger, M.W., Perry, E.F., 1977. Elevated systolic pressure following chronic low-level cadmium feeding. Am. J. Physiol. 232 (2), H114–H121. Perry, H.M., Erlanger, M.W., Perry, E.F., 1979. Increase in the systolic pressure of rates chronically fed cadmium. Environ. Health Perspect. 28, 251–260. Roels, H.A., Lauwerys, R.R., Bernard, A.M., Buchet, J.P., Vos, A., Overstyens, M., 1991. Assessment of the filtration reserve capacity of the kidney in workers exposed to cadmium. Br. J. Inductrial Med. 48 (6), 365–374. Scinicariello, F., Abadin, H.G., Murray, H.E., 2011. Association of low-level blood lead and blood pressure in NHANES 1999–2006. Environ. Res. 111 (8), 1249–1257. SRNT Subcommittee on Biochemical Verification, 2002. Biochemical verification of tobacco use and cessation. Nicotine Tob. Res. 4 (2), 149–159. Statistics Canada, 2015a. Canadian Health Measures Survey (CHMS) Data User Guide: cycle. Statistics Canada, Ottawa, 3. Statistics Canada, 2015b. Instructions for Combining Multiple Cycles of Canadian Health Measures Survey (CHMS) Data. Statistics Canada, Ottawa. Tellez-Plaza, M., Navas-Acien, A., Crainiceanu, C.M., Guallar, E., 2008. Cadmium exposure and hypertension in the 1999–2004 National health and Nutrition Examination survey (NHANES). Environ. Health Perspect. 116 (1), 51–56. United Nations Environment Programme, 2010. Final Review of Scientific Information on Cadmium. Geneva. Wester, P.O., 1973. Trace elements in serum and urine from hypertensive patients before and during treatment with chlorthalidone. Acta Med. Scand. 194 (6), 505–512. Whittemore, A.S., DiCiccio, Y., Provenzano, G., 1991. Urinary cadmium and blood pressure: results from the NHANES II survey. Environ. Health Perspect. 91, 133–140. Wilkins, K., Gee, M., Campbell, N., 2012. The difference in hypertension control between older men and women. Health Rep. 23 (4), 3–10.
5. Conclusions In conclusion, the current large, cross-sectional study provides supporting evidence of an association between cadmium exposure and blood pressure. This association varied by marker of cadmium levels (blood versus urine) and differed by subpopulations. The present study also confirms, as has been previously reported, a significant association between cadmium levels and blood pressure and hypertension among women, particularly among never smokers. A new finding is the possible modification effect of body weight. Such results demonstrate the need for further investigation of this topic, particularly in prospective studies. Funding This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors. Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.envres.2017.01.040. References Agarwal, S., Zaman, T., Tuzcu, E.M., Kapadia, S.R., 2011. Heavy metals and cardiovascular disease: results from the national health and nutrition examination survey (NHANES) 1999–2006. Angiology 62 (5), 422–429. Alessio, L., Berlin, A., Dell'Orto, A., Toffoletto, F., Ghezzi, I., 1985. Reliability of urinary creatinine as a parameter used to adjust values of urinary biological indicators. Int. Arch. Occup. Environ. Health 55, 99–106. Buchet, J.P., Lauwerys, R., Roels, H., Bernard, A., Bruaux, P., Claeys, F., Ducoffre, G., DePlaen, P., Staessen, J., Amery, A., Lijnen, P., Thijs, L., Rondia, D., Sartor, F., Saint Remy, A., Nick, L., 1990. Renal effects of cadmium body burden of the general population. Lancet 336 (8717), 699–702. Bushnik, T., Levallois, P., D’Amour, M., Anderson, T.J., McAlister, F.A., 2014. Association between blood lead and blood pressure: results from the canadian health measures survey (2007–2011). Health Rep. 25 (7), 12–22. CAREX Canada, 2014. Cadmium – Occupational Estimate. 〈www.carexcanada.ca/en/ cadmium/occupational_estimate〉( accessed 11.05.16). CAREX Canada, 2015. Cadmium – Substance profile. 〈www.carexcanada.ca/en/ cadmium/occupational_estimate〉( accessed 11.05.16). Centers for Disease Control and Prevention, 2009. Fourth National Report on Human Exposure to Environmental Chemicals. Department of Health and Human Services, Centers for Disease Control and Prevention, Bethesda, MD. Coresh, J., Selvin, E., Stevens, L.A., Manzi, J., Kusek, J.W., Eggers, P., Van Lente, F., Levey, A.S., 2007. Prevalence of chronic kidney disease in the United States. J. Am. Med. Assoc. 298 (17), 2038–2047. Gallagher, C.M., Meliker, J.R., 2010. Blood and urine cadmium, blood pressure, and hypertension: a systematic review and meta-analysis. Environ. Health Perspect. 118 (2), 1676–1684. Garner, R., Levallois, P., 2016. Cadmium levels and sources of exposure among Canadian adults. Health Rep. 27 (2), 10–18. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. 2009. A review of human carcinogens: Part C – Arsenic, metals, fibres, and dusts. IARC Monographs on the Evaluation of Carcinogenic Risk to Humans, 100C. Lyon, France.
72