SAPALDIA is a multicenter, population-based prospective cohort study consisting of a random sample of 9,561 adults who were 18–60 years of age when they were recruited from eight regions in Switzerland (Martin et al. 1997
). The baseline survey was conducted in 1991 when participants were administered medical examinations, including spirometry testing, and a detailed health questionnaire. The second assessment (SAPALDIA 2) of 8,047 study participants (84.2%) was conducted from 2001 to 2003 and also included HRV measurements and special questionnaires on work-related exposures (Ackermann-Liebrich et al. 2005
). From these participants who were ≥ 50 years of age at the time of SAPALDIA 2 (n
= 4,645), 1,846 individuals (955 women, 891 men) were randomly selected for 24-hr electrocardiogram (ECG) monitoring to assess HRV (Felber Dietrich et al. 2006
In addition, a detailed questionnaire on household cleaning activities was administered to all SAPALDIA 2 participants who responded positively (n
= 3,255) to the following question from the health questionnaire, “Have you been the person doing the cleaning and/or washing in your home in the last ten years?” This cross-sectional analysis was restricted to 851 individuals ≥ 50 years of age who had valid HRV measurements and who had completed the household cleaning questionnaire [for a flow chart describing participation, see Supplemental Material, Figure 1
)]. Of these 851 participants, 188 were excluded for reporting either occupations that used cleaning products at work (n
= 166) or that involved metalworking, welding, or soldering (n
= 22). After further exclusion of participants with insufficient exposure or covariate information (n
= 82), a total of 581 participants contributed to the analyses. The distributions of basic characteristics were similar between the 581 participants included in this analysis and the 808 nonparticipants, who were also ≥ 50 years of age and reported cleaning activities at their homes, but who were not selected for HRV assessment (see Supplemental Material, Table 1
Ethical approval for the study was given by the central Ethics Committee of the Swiss Academy of Medical Sciences and the Cantonal Ethics Committees for each of the eight examination areas (Aarau, Basel, Davos, Geneva, Lugano, Montana, Payerne, and Wald, Switzerland) and participants signed an informed consent at the examination.
HRV measurements and analyses.
Holter recordings, described elsewhere by Felber et al. (2006), were made between August 2001 and March 2003. Recorders were placed on participants who had given consent after a detailed health interview. Participants were asked to follow their regular daily routine during the recording period. To avoid a biased result due to methacholine challenge, which was part of the SAPALDIA lung function testing and which, for practical reasons, was performed before the Holter recording, we excluded the first 2 hr of all recordings. The mean ± SD duration of the Holter recordings was 22.4 ± 2.1 hr. The summary measures of HRV were selected as the primary outcomes of interest in this analysis and included the 24-hr value of the SD of all normal RR (NN) intervals (msec 24-hr SDNN), and the following frequency domain variables: total power (TP; ≤ 0.40 msec2
/Hz), low-frequency (LF) power (0.04–0.15 msec2
/Hz), and high-frequency (HF) power (0.15–0.40 msec2
/Hz). The evaluation of SDNN and TP was also limited to nighttime, which was defined as the time when subjects indicated in the diary that they where sleeping (see Felber et al. 2008; Probst-Hensch et al. 2008
). To improve normality of the residuals, each HRV parameter was log transformed in this analysis.
The spirometry protocol was equivalent to that of the European Community Respiratory Health Survey (ECRHS) (Burney et al. 1994
). No bronchodilation was applied. Participants performed three to eight forced expiratory lung function maneuvers with the spirometer (model 2200; Sensormedics Yorba Linda, CA, USA), and at least two acceptable measurements of forced vital capacity (FVC) and forced expiratory volume in 1 sec (FEV1
) were obtained, complying with the American Thoracic Society criteria (American Thoracic Society 1995
Respiratory symptoms and medication use. Presence of asthma was based on positive responses to the questions “Have you ever had asthma?” and, if yes, “Was this confirmed by a doctor?” Shortness of breath was defined as a positive response to the question “Are you troubled by shortness of breath when hurrying on level ground or walking up a slight hill?” Chronic bronchitis was defined as self-reported cough or phlegm during the day or at night on most days for as much as 3 months each year for ≥ 2 years. Medication use for asthma or breathing problems was defined by a positive response to either of the following questions: “Has your doctor ever prescribed medicines, including inhalers, for your breathing?”; “Are you currently taking any medicines, including inhalers, aerosols, or tablets for asthma?”; or “Have you taken medicine for asthma during the last 3 days?”
The questionnaire module on cleaning and washing in the home, which was adopted from the ECRHS, asked about the frequency of using of 16 different products for domestic cleaning and washing over a period of at least 3 consecutive months since the baseline survey in 1991 (ECRHS 2002). In a previous analysis of Spanish housewives, Medina et al. (2000)
compared the frequency responses in this module with a 1-week diary as the gold standard, and the median specificity was 94% across the different cleaning products. We hypothesized that use of products with spray application would better facilitate respiratory exposure to irritants than would nonspray products. Thus, we mainly focused on several spray products used for cleaning glass, furniture, rugs/curtains/carpets, or ovens and on products for ironing, air freshening, and other unspecified purposes. We also examined the use of scented products, which could either be in spray or nonspray form. For each product, the frequency of use was recorded as never, < 1, 1–3, or 4–7 days/week and assigned a score from 0 to 3, respectively. In a preliminary factor analysis, it was determined that the use of cleaning sprays for glass, furniture, and rugs/carpets/curtains contributed to most of the variation in the reported use of spray products in the study sample. A composite score variable for cleaning sprays was subsequently constructed, which was the sum of individual frequency scores for using glass, rug/carpet/curtain, and furniture cleaning sprays with a value ranging from 1 to 9, and divided into four categories (1, 2, 3, ≥ 4). To evaluate the number of sprays used weekly (accounting for all types of sprays, including air freshening sprays), another composite score variable was developed with a value of 1–3 (1, any spray < 1 day/week; 2, 1 spray ≥ 1 day/week; 3, ≥ 2 sprays ≥ 1 day/week).
Statistical analyses were performed using SAS software (version 9.2; SAS Institute Inc., Cary, NC, USA). Log-transformed 24-hr SDNN, TP, LF, and HF were regressed separately against the different categorical variables of cleaning spray, air freshening spray, scented products, and number of different sprays used weekly in multiple linear regression (PROC GLM). Effect estimates for each exposure frequency category were first expressed as geometric mean ratios, with unexposed participants as the reference group, and then converted into average percent changes. We also evaluated ordinal exposure–response trends by treating exposure variables as continuous, where unexposed participants were assigned a score of zero. Because 24-hr SDNN and TP are in theory mathematically correlated, the Wilks’ lambda test was used to evaluate the overall association between exposure and both outcomes 24-hr SDNN and TP using the MANOVA procedure, which handles multiple correlated outcomes (Scheiner 2001
); only p
-values indicating statistically significant deviation (p
< 0.05) from the null hypothesis of no association are reported.
All models were adjusted for individual-level covariates that were considered potential confounders of the association between long-term use of household sprays and scented products and HRV including sex (female as reference), age (years), age2
, body mass index (BMI; kilograms per meter squared), BMI2
, smoking status [former, current, never (reference)], tertiary education level [high, medium, low (reference)], employment status [retired, sick/disabled, or other; fully/partially employed, in military, or student; unemployed housewife/househusband (reference)], weekly physical activity [to the point of getting out of breath or sweating for < 30 min (reference), between 30 min and 2 hr, or > 2 hr], daily alcohol consumption [≥ 1, < 1 drink (reference)], daily exposure to environmental tobacco smoke [ETS, < 3, ≥ 3, 0 hr (reference)], uric acid concentration measured in serum (micromoles per liter), current cardiovascular medication intake [yes, no (reference)], seasonal effects (based on sine and cosine function of day of examination), street-related noise, train-related noise, average traffic-related particulate matter with aerodynamic diameter < 10 μm (PM10
) concentration, and study area. The measurement and analysis of personal noise and traffic-related PM10
exposures have been described in detail elsewhere (Dratva et al. 2012
; Künzli et al. 2009
; Liu et al. 2007
Having ever smoked, obesity (BMI ≥ 30 kg/m2), cardiovascular medication intake, and markers or symptoms of obstructive lung disease (OBS) were evaluated as potential effect modifiers. We constructed multiplicative interaction terms between each effect potential modifier and ordinal exposure variables (e.g., exposure scores modeled as continuous variables), and included them in separate multiple linear regression models. Only interactions with p-values < 0.05 are reported. In addition, we evaluated interactions with 24-hr SDNN and TP as a combined outcome using the MANOVA procedure described above. OBS was defined as the presence of any of the following markers or symptoms: ratio of forced expiratory volume in 1 sec over forced vital capacity (FEV1:FVC) < 0.70, self-reported symptoms of chronic bronchitis, or self-reported shortness of breath. To evaluate effect modification by OBS as distinct from asthma, we excluded all participants who reported an occurrence of asthma or asthma medication intake from the analysis. We did not evaluate self-reported asthma, diabetes, or heart disease for effect modification because of insufficient numbers of observations for statistical comparisons.
Secondary analyses. Specific cleaning activities were not recorded in the time activity diaries; thus, we were not able to evaluate the acute effect of household sprays and scented products on HRV. Because HRV during nighttime is less likely to be influenced by short-term disturbances, we estimated adjusted average percent changes of (log-transformed) nighttime SDNN and TP in association with the frequency of use of each product type in multiple linear regression. Linear regression models were also repeated with the reference category for each exposure variable comprising both unexposed participants and those who used the product of interest < 1 day/week.