Background
Methods
Patient population
Search strategy
Study selection
Data extraction
Assessment of risk of bias
Results
Overview of studies
Study, Year, Country | Design and Population | ETS Exposure | Outcome | Results | Bias Assessment/Comment |
---|---|---|---|---|---|
Prospective cohort studies in premature infants
| |||||
Broughton 2005 United Kingdom [19] | Prospective study of 126 premature infants (GA <32 wks; 40% developed BPD) | Maternal smoking during pregnancy | RSV LRTI (41% hospitalized) | aOR, 4.85 (95% CI, 1.61–14.58); P = 0.005 | No significant bias concerns affecting the relationship of ETS and outcome |
Parental smoking in home | RSV LRTI | NS aOR, 0.81 (95% CI, 0.19–3.37); P = 0.771 | |||
Maternal smoking during pregnancy | Hospital admission (all cause; 56% of admissions were RSV LRTI) | NS aOR, 1.19 (95% CI, 0.20–7.07); P = 0.849 | |||
Parental smoking in home | Hospital admission | aOR, 3.39 (95% CI, 1.08–10.63); P = 0.003 | |||
Maternal smoking during pregnancy | Length of hospital stay | NS, P = 0.150 (OR not reported) | |||
Parental smoking in home | Length of hospital stay |
P < 0.001 (OR not reported) | |||
Carbonell-Estrany 2001 Spain [9] | Prospective, longitudinal study of 999 premature infants (GA ≤32 wks) | Days of smoke exposure | RSV hospitalization | aOR, 1.63 (1.05–2.56); P = 0.031 | No significant bias concerns affecting the relationship of ETS and outcome |
Figueras-Aloy 2008 Spain [20] | 2-cohort study of premature infants (GA 32–35 wks); 202 cases hospitalized for RSV and 5239 controls not hospitalized for respiratory illness | Maternal smoking during pregnancy | RSV hospitalization | aOR, 1.61 (95% CI, 1.16–2.25); P = 0.004 | Authors note relatively high loss to follow-up of 12% of children fulfilling inclusion criteria. Both ETS exposure variables were significant in bivariate analysis at P < 0.01, but when included in multivariate model, only prenatal smoking was significant, possibly due to misclassification of ETS exposure |
≤2 smokers in home | RSV hospitalization | NS in multivariate model | |||
Significant in bivariate analysis, OR 1.59 (95% CI, 1.12–2.26); P = 0.01 | |||||
Law 2004 Canada [10] | Prospective cohort study of 1832 premature infants (GA 33–35 wks) | ≥2 smokers in household | RSV hospitalization | aOR, 1.87 (95% CI, 1.07–3.26); P = 0.027 | No significant bias concerns affecting the relationship of ETS and outcome |
Case–control study in premature infants
| |||||
Figueras-Aloy 2004 Spain [21] | Case–control study of premature infants (GA 33–35 wks); 186 cases hospitalized for RSV; 371 controls born at same time as cases | Maternal smoking during pregnancy | RSV hospitalization | NS in multivariate model | No significant bias concerns affecting the relationship of ETS and outcome |
Significant in bivariate analysis OR, 1.62 (95% CI, 1.08–2.42); P = 0.027 | |||||
Maternal smoking at home | RSV hospitalization | NS in bivariate model | |||
OR, 1.49 (95% CI, 1.01–2.18); P = 0.055 | |||||
≥2 smokers at home | RSV hospitalization | NS in bivariate model | |||
OR, 1.41 (95% CI, 0.92–2.14); P = 0.146 | |||||
Prospective cohort studies in the general population
| |||||
Holberg 1991 US [22] | Prospective birth cohort study of 1179 healthy infants followed for 1 year | Maternal smoking | RSV diagnosed in an office visit | NS in multivariate model | No significant bias concerns affecting the relationship of ETS and outcome |
Rate ratio, 1.0 (95% CI, 0.3–3.5) | |||||
von Linstow 2008 Denmark [6] | Prospective birth cohort study of 217 children followed for 1 year | Smoking in household | RSV hospitalization | aOR, 5.06 (95% CI, 1.36–18.76); P < 0.02 | No significant bias concerns affecting the relationship of ETS and outcome; to reduce problems with colinearity, only 1–2 variables from each group of covariates (e.g., social variables, smoking parameters) were included in the multivariate model. |
Maternal smoking during pregnancy | RSV hospitalization | NS in multivariate model (OR not reported) | |||
Significant in univariate model | |||||
OR, 4.19 (95% CI, 1.21–14.53); P = 0.024 | |||||
Case–control studies in the general population
| |||||
Bulkow 2002 US [12] | Case–control study of Alaska native children aged <3 years with 204 cases and 338 controls | Smoker in household | RSV hospitalization | NS in multivariate model | Unclear risk of ETS exposure misclassification because of high prevalence of smoking and frequency of indoor visiting among households during winter RSV season; low risk of other types |
Significant in bivariate analysis, OR, 1.61; P ≤ 0.018 | |||||
Gurkan 2000 Turkey [23] | Case–control study of 28 cases and 30 controls aged 2–18 months | ▪Nonsmoking parents ▪Only smoker mother ▪Only smoker father ▪Both parents smokers | RSV bronchiolitis admitted to the ED Serum cotinine assessed during ED visit and 1 month later | Significant differences in cases vs. controls (P < 0.05) for all ETS exposure variables; however, only father smoker was more prevalent in the control than case group | No multivariate analysis performed (confounding bias) |
Significant differences in cases vs. controls (P < 0.05) in cotinine levels for both parents smokers vs. both parents nonsmokers and for only mother smoker vs. both parents nonsmokers in the control group | |||||
Hall 1984 US [24] | Case–control study of 29 cases and 58 controls hospitalized with non respiratory acute illness | Smoking in household | RSV hospitalization | Significant difference in smoking in household in cases (76%) vs. controls (40%) (P < 0.05) | No multivariate analysis performed (confounding bias) |
Hayes 1989 American Samoa [25] | Case–control study of children aged <1 year (20 cases and 15 well controls) | Smoker in household | RSV hospitalization (53% laboratory-confirmed) | Significant difference in smoker in household in cases (92%) vs. well controls (53%) (P = 0.04) | No multivariate analysis performed (confounding bias)53% of hospitalizations were laboratory-confirmed RSV |
Nielsen 2003 Denmark [5] | Case–control study of 1252 cases in children aged <2 years and 5 controls for each case | Maternal smoking during pregnancy from the Medical Birth Register | RSV hospitalization | aOR, 1.56 (95% CI, 1.32–1.98) | No significant bias concerns affecting the relationship of ETS and outcome |
Reeve 2006 Australia [11] | Case–control study with 271 cases and 542 controls (median age 6 mo) | Maternal smoking | RSV hospitalization | NS in main multivariate modelBivariate OR not reported CART analysis performed to define groups that are most homogeneous with regard to the outcome of RSV hospitalization. CART analysis found that smoking was a risk factor in children with birthweight >2500 g and single mothers (41.0% hospitalized vs. 26.9% for single nonsmoking mothers)Smoking was not significant for any other group | Analysis was weakened by reliance on a questionnaire that did not seek to quantify ETS exposure and by the absence of laboratory confirmation of ETS exposure.63 participants were excluded due to data unavailability (37 of these were missing the mother’s smoking status and 47 had proven RSV), although the missing data were not statistically significant |
Stensballe 2006 Denmark [26] | Case–control study of 2564 cases and 12 816 controls from birth to 18 months | Any maternal smoking during pregnancy and lactation | RSV hospitalization | aOR, 1.35 (95% CI, 1.20–1.52); P < 0.001 | No significant bias concerns affecting the relationship of ETS and outcome |
Study, Year, Country | Design and Population | Smoke Exposure | Outcome | Results | Bias Assessment |
---|---|---|---|---|---|
Cohort study in premature infants
| |||||
Gavin 2007 US [27] | Retrospective cohort study of 2098 premature infants (GA 32–35 weeks) in the Texas Medicaid program | Maternal smoking status during pregnancy from the birth certificate | Insurance claims for bronchiolitis or RSV hospitalization in the first year of life | NS, aOR, 0.78 (95% CI, 0.38–1.61) | Clinical diagnosis of RSV leading to misclassification could underestimate ETS exposure risk |
Cohort studies in the general population
| |||||
Boyce 2000 US [4] | Retrospective cohort study of children aged <3 years in the Tennessee Medicaid program from 1989–1993, with 248 652 child-years of follow-up | Maternal smoking status during pregnancy from the birth certificate | Insurance claims for bronchiolitis or RSV hospitalization in the first year of life | aRR, 1.3 (95% CI, 1.2–1.4) | Clinical diagnosis of RSV leading to misclassification could underestimate ETS exposure risk |
Carroll 2007 US [28] | Retrospective cohort study of 101 245 term infants enrolled in the Tennessee Medicaid program | Maternal smoking status during pregnancy from the birth certificate | Insurance claims for bronchiolitis or RSV pneumonia in the first year of life | ▪Hospitalization aOR, 1.28 (95% CI, 1.20–1.36) | Clinical diagnosis of RSV leading to misclassification could underestimate ETS exposure risk |
▪ED visit aOR, 1.22 (95% CI, 1.13–1.31) | |||||
▪Clinic visit aOR, 1.06 (95% CI, 1.01–1.12) | |||||
▪Bronchiolitis diagnosis aHR, 1.14 (95% CI, 1.10–1.18) | |||||
Koehoorn 2008 Canada [29] | Retrospective cohort study of 93 058 infants aged 2–12 months | Maternal smoking status during pregnancy from perinatal database | Diagnostic codes for bronchiolitis for outpatient visits or hospitalizations | ▪Inpatient onlyaHR, 1.47 (95% CI, 1.27–1.69) | Clinical diagnosis of bronchiolitis leading to misclassification could underestimate ETS exposure risk |
▪Outpatient or inpatient NS in multivariate model, aHR, 1.03 (95% CI, 0.97–1.09) | Maternal smoking during pregnancy was significant in bivariate analysis for both case definitions, but when included in the multivariate models, it was significantly associated only with the inpatient (more severe) case definition | ||||
▪Significant in bivariate analysis, OR 1.14 (95% CI, 1.08–1.21); no P value reported | |||||
Marbury 1996 US [30] | Prospective cohort study of 1424 children with private insurance followed to age 2 years | Maternal smoking status | Diagnosis of bronchiolitis from electronic medical records | NS, aRR, 1.3 (95% CI, 0.8–2.2); no P value reported | Clinical diagnosis of bronchiolitis leading to misclassification could underestimate ETS exposure risk The authors noted that smokers were less likely to participate in the study (the Indoor Air and Children’s Health Study) and that smokers who participate may differ from those who do not. They also noted the possibility of underreporting of smoking |
Reese et al., 1992 Australia [18] | Retrospective cohort study of 491 patients up to age 17 years admitted to an Australian children’s hospital June-Dec. 1987 for whom urinary cotinine levels were available. | Urinary cotinine level, analyzed without knowledge of exposure status or diagnosis | Hospitalization with diagnosis of bronchiolitis vs. a non-respiratory diagnosis (limited to patients aged 5–15 mos in the nonrespiratory illness group) | Elevated cotinine levels found in bronchiolitis vs. nonrespiratory illness group (P < 0.02) Subanalysis of the bronchiolitis group by RSV status found no significant difference between RSV-positive (n = 16) and RSV-negative (n = 23) patients; both subgroups had elevated cotinine vs. the nonrespiratory illness group | Risk of confounding not clear; regression analysis appears to have been performed but was insufficiently reported (no aORs reported) |
Among those with cotinine levels, 41 patients (aged 5–15 mos.) were diagnosed with bronchiolitis | |||||
Case–control studies in the general population
| |||||
Holman 2003 US [3] | Case–control study of 224 infants aged <1 year who died from bronchiolitis, and 2336 controls | Maternal smoking status during pregnancy | Bronchiolitis death from death certificate data | aOR, 1.6(95% CI, 1.0–2.6) | Clinical diagnosis of bronchiolitis from death certificate leading to misclassification could underestimate ETS exposure risk |
McConnochie 1986 US [31] | Case–control study of 53 cases of bronchiolitis and 106 controls in children aged <2 years presenting to a physician’s officeBivariate analysis included 3 ETS exposure variables, but only “passive smoking” was included in the multivariate analysis | Any passive smoking | Bronchiolitis from diagnostic registry and record review | aOR, 3.87 if no family history of asthma (no CI or P value reported)aOR, 4.03 if family history of asthma (no CI or P value reported) | Clinical diagnosis of bronchiolitis leading to misclassification could underestimate ETS exposure risk Interviews related to smoking status were conducted approximately 7.8 years after the bronchiolitis episodes; current and former smokers at the time of the interview were assumed to be smoking at the time of the bronchiolitis episode |
Smoking in household | Bronchiolitis | Bivariate OR, 3.21 (95% CI, 1.42–7.25) | |||
Mother smokes | Bronchiolitis | Bivariate OR, 2.33 (95% CI, 1.19–4.57) | |||
Father smokes | Bronchiolitis | NS in bivariate model, OR 1.71 (95% CI, 0.87–3.33) |
Study, Year, Country | Design and Population | Smoke Exposure | Outcome | Results | Bias Assessment |
---|---|---|---|---|---|
Risk of hospitalization among premature children with RSV illness
| |||||
Groothuis et al., 1988 US [32] | Prospective cohort study of 30 premature infants aged <2 years with BPD receiving home oxygen therapy; participants followed for 5 mos (Dec-Apr) | Smokers in home | Risk of hospitalization (11 of 16 with RSV hospitalized) vs. outpatient treatment | NS; bivariate analysis reported and P value not given | No multivariate analysis performed (confounding bias) |
Risk of hospitalization among children in the general population with RSV illness
| |||||
Al-Shehri 2005 Saudi Arabia [33] | Case–control study; 51 children aged ≤5 years hospitalized for bronchiolitis (cases) and 115 children with bronchiolitis but not hospitalized (controls); 40% of cases were RSV | History of exposure to smoking | Risk of hospitalization vs. outpatient treatment for bronchiolitis | aOR, 2.51 (95% CI, 2.11–3.73); P = 0.05 | Risk of participant selection bias because both cases and controls had bronchiolitis |
Hall 2009 US [2] | 919 children aged <5 years with laboratory-confirmed RSV infections | Smoking in household | Risk of hospitalization vs. outpatient treatment | NS in multivariate analysis (no aOR, CI, or P value)NS in bivariate analysis (P = 0.43) | No significant bias concerns affecting the relationship of ETS and outcome |
Mother smokes | Risk of hospitalization vs. outpatient treatment | Not included in multivariate analysis NS in bivariate analysis (P = 0.21) | |||
Somech 2006 Canada [34] | Prospective study of 195 infants (mean age 3.8 months) with laboratory-confirmed RSV infection | Exposure to smoke from at least one family member | Hospitalization (113) vs. outpatient treatment (82) of RSV | ETS exposure was unrelated to hospitalization (P value not reported) | No multivariate analysis performed (confounding bias) |
Disease severity in children in the general population hospitalized with RSV illness
| |||||
Al-Sonboli 2006 Yemen [35] | Prospective study of 325 children aged ≤2 with acute respiratory illness seeking emergency or outpatient services at a hospital (82% RSV) | Smoking in household | Severe hypoxia among RSV-positive group | aOR, 3.8 (95% CI, 1.5–9.8); P = 0.002 | No description of how smoke exposure or other family characteristics were ascertained (exposure bias) |
Bradley 2005 US [36] | Prospective evaluation of 206 infants hospitalized with their first episode of severe RSV bronchiolitis | Current maternal smokingMaternal smoking status during pregnancy | Lowest oxygen saturation rate | Current maternal smoking was associated with lower oxygen saturation, P =0.05No effect of smoking during pregnancy only (n = 10) | No significant bias concerns affecting the relationship of ETS and outcome |
Chatzimichael 2007 Greece [37] | Prospective study of 240 children aged 6–24 months hospitalized for bronchiolitis | Exposure to >5 cigarettes per day in the home; children with prenatal exposure were excluded | Disease severity measured with a clinical rating tool that included hypoxemia | aOR, 2.2 (95% CI, 1.1–3.6); P = 0.003 | Unclear RSV disease classification; severity tool used |
Sritippayawan 2006 Thailand [38] | Study of 19 children (median age 9 months) admitted to the hospital with laboratory-confirmed RSV LRTI | Exposure measured by urinary cotinine | Hypoxemia (oxygen saturation <92%) | Cotinine was detected in 100% of infants with hypoxia vs. 33% of those without hypoxia; P = 0.01 | High risk of selective reporting biasRisk of confounding not clear; regression analysis appears to have been performed but was insufficiently reported so it was not possible to tell which factors were controlled for |