 Before we go ahead and get started, we've given a few extra minutes for the stragglers to comment, which is nice. So, this is Melanie Pickett. Dr. Melanie Pickett, she's just finished up her dissertation. Oh, Duff. You have a recent name change. Congratulations. Is that a marriage? Yes. Okay, wonderful. Melanie Duff is here today. She's just finishing up her doctorate at Harvard. I believe, just submitted her dissertation, defended it a couple weeks ago. She's now on the interview circuit. And one of her options is to come here to UCI to work with Mace. I'm very excited to have her here today. And she's going to be giving her talk on smoke-free air laws, second-hand smoke exposure in health and children and adults. And I will add, actually, Melanie was an undergraduate here at UCI some years ago in the School of Social Ecology. And so it's nice to have you back here. I guess that was in environmental analysis and design. Which, I believe, is social ecology, right? It was. It was. Yeah, there's been some restructuring recently. So, one time, though, is social ecology. And, of course, our MPH program actually grew really out of the School of Social Ecology and a lot more senior faculty from that area. So it's kind of nice to have a whole student kind of return after many years with the doctor and hear what she's been doing since then. You? Yeah. So it's nice to be back here and to be able to present my research. So thank you for this opportunity. Today I'll be talking about smoke-free air laws, second-hand smoke exposure in health and children and adolescents. So this is just a brief outline of my talk. First, I'll give a little bit of background on what we're doing in second-hand school and smoke-free air laws. Next, I'll describe my three papers. And for each paper, the methods, results, conclusions, and how it compares to other studies. Then I'll describe the strengths and limitations together for all three studies. And finally, I'll give an overall conclusion. So please feel free to interrupt me if you have any questions. So second-hand smoke exposure is both a smoke exhaled from smokers called mainstream smoke. And it's produced by the burning end of the cigarette called side-stream smoke. And it has over 250 toxic chemicals, which 60 are normally suspected to cause cancer. And some of them are listed here, including arsenic, benzene, rillium, and others. Second-hand smoke causes a number of health outcomes. In 2006, the Surgeon General's report evaluated all the evidence and concluded that second-hand smoke is sufficient evidence that second-hand smoke in adults has this coronary artery disease, lung cancer, and reproductive sex in women. And in children, there was sufficient evidence that second-hand smoke causes a little ear disease, respiratory symptoms, impaired lung function, sins, or respiratory illnesses, and low birth weight. So a useful way to measure second-hand smoke exposure in non-smokers is to use codeine, which is a metabolite of nicotine, and half-life is about 15 to 20 hours. This is a figure showing the distribution of codeine in the U.S. population. And so on the x-axis is serum-coating levels on the log scale. And so codeine has a bi-mobile distribution so that non-smokers over here tend to have codeine levels less than 10 to 15, whereas smokers tend to have codeine levels greater than 10 to 15 nanograms per milliliter. So the strengths of using codeine as a biomarker of exposure to second-hand smoke are that it can be measured in a variety of mediums, including serums, saliva, urine, air, and toenails. It's specific to second-hand smoke exposure in non-smokers. And the half-life is longer than nicotine. Nicotine's half-life is about half an hour to three hours. There are some limitations with using codeine. It reflects exposure within the past couple of days. There's been shown to be diagonal variations with codeine levels higher later in the day. And it reflects exposure to nicotine, which isn't the most harmful chemical in second-hand smoke. This is a graph using data from the U.S. showing that codeine levels have increased over the years. And so they were first measured in 1988. And this is a geometric mean of codeine levels. So they've decreased quite dramatically over the years. And this figure shows it's just for children aged 4 to 16. And it's stratified by whether or not they lived in the silver. So the red line shows that along children that lived in the silver, there was no change in the geometric mean codeine levels over time. However, the yellow line is children who did not live with a smoker, showing that codeine levels decreased from .12 to .05, suggesting that there's something in the environment that is contributing to this decrease. So smoke-free air laws completely ban smoking. And they were primarily enacted in indoor areas such as workplaces, restaurants, and bars. And they've been increasing over the years. This is a map of the U.S. in 2000, showing that only two states had a smoke-free air law, including California and Utah, and a couple of local laws in California and also in Massachusetts. But by 2005, there were 10 states with a smoke-free air law. And by 2010, there were 34 states covering 74.2% of the U.S. population. So smoke-free air laws, there's been many studies that have shown that smoke-free air laws reduced second-hand smoke exposure in adults. And they've also been shown to change social norms towards smoking, reduce smoking prevalence, and increase the number of smoke-free homes. So these things combined may result in a reduction of second-hand smoke exposure in the community that would result in children being less exposed to the second-hand smoke. And in children, there's been a few studies that have looked at this association. There was one study in Scotland that looked at coating levels in children before and after a smoking ban. And they found a 39% reduction in saliva coating levels. And the study in Wales attempted to replicate what the Scottish study found, and they found a 12% reduction in the smoking. So there's been a lot of studies looking at the impact smoke-free air laws have on heart attacks or acute liocardial infarction in adults. And this is a meta-analysis that came out recently that looked at 11 studies, primarily in the U.S. and Europe. And they found a cool rate ratio of 0.81, suggesting a 19% reduction in AMI hospitalization rates after a smoking ban compared to before a smoking ban. And a similar meta-analysis came out around the same time, and they found a cool rate ratio of 492, suggesting a 8% reduction in AMI. So both these meta-analysis looked at the impact of the smoke-free air law over time. And so on the X-axis is the duration post-ordnance. This one is in months, and this one is in years. And on the Y-axis is the log of the rate ratio, so as the time increases, the rate ratio increases. So the impact of the Massachusetts smoke-free air law on AMI has not been examined until now. Massachusetts has a long history of tobacco control. In 1993, they formed a tobacco control program. In 1994, the first smoke-free air law was implemented in Amherst. In 93, 96, and 2002, there was an increase in the Massachusetts cigarette sales tax. And in 2003, Boston implemented a smoke-free air law, and about a year later, the state implemented a statewide smoke-free air law. So my overall hypothesis is that smoke-free air law is lower setting than smoke exposure, and thus reduce disease in both children and adolescents. So the first paper is called Smoke-free Air Laws and Second Hand Smoke exposure among non-smoking youth, in Haines, 1999 to 2006. These are my co-authors. So the study population is the National Health and Nutrition Examinations Survey, also called in Haines from 1999 to 2006. This is a cross-sectional survey. It's representative of the U.S. population. It uses a complex multi-stage design. There's home interviews and also physical exams in these mobile examination clinics. And there were 117 survey locations or counties sampled between 1999 and 2006. So this describes the complex. Survey design a little bit better. So first counties are selected, and then segments or city blocks are selected, and then households, and finally individuals. So the study population included youth aged 3 to 19 years who are not missing data on protein. We limited our analysis to non-smokers defined by both having a codeine level less than 15 and also by we excluded youth who hadn't smoked in the past five days. And we had a sample size of 11,486. Do you have questions now or are they at the end? Yeah, no, it's fine. So I assume secondhand smoke still has an effect on smokers. So why would you exclude smokers? So we're looking at non-smokers. We're looking at kids. So, right, secondhand smoke does have an impact on smokers. But we want to see the impact of the smoke-free air law on non-smokers. So if we included smokers, we're looking at codeine levels. And so most of their codeine levels would be from them smoking. And smoke-free air laws are primarily enacted to reduce secondhand smoke exposure. And so it would be difficult to see the impact of the smoke-free air law because although smoke-free air laws haven't shown to reduce smoking from less, that's not their main, that's not what they were designed to do. So the effect would be, given the huge base rate of codeine? Right, right. So right, they've seen really large decreases in codeine among non-smokers as a result of these smoke-free air laws. And that's in urine or toenails? Serum. So our exposure is smoke-free air law coverage. So in-hand survey locations are categories. We're put into these three categories. 26 counties were classified as having extensive coverage, meaning there was at least one smoke-free workplace restaurant or bar law at the county or state level. 11 counties were classified as having limited coverage, meaning there was at least one city within the county with a smoke-free workplace restaurant or bar law. And 80 counties were classified as having no coverage, meaning there was no smoke-free air laws coverage at the state, county, or city level. And this tree outlines the exposure a little bit more. So we identified which county each individual was in. Then we found out if they had a state law, and if they did, they were classified as having extensive coverage. If there was no state law, but if there was a county law, that covered the entire county population, they were classified as having extensive coverage. If there was no county law, but a city law, they were classified as having limited coverage. And no city law, they were classified as having no coverage. So our outcome is CODE. And from 1999 until 2006, the limit of detection was lower. So from 1999 until 2000, the limit of detection was .05. But in 2001 to 2002, the limit of detection was lower to .015. So 15% of the sample had the highest limit of detection. And 85% had the lower limit of detection. And from 2003 to 2006, everyone had the lower limit of detection. And in Haines, for values that are below the limit of detection, in Haines imputes a value. That's the limit of detection divided by the square root of 2. So it was .035 or .05 and .011 or .015. About what percentage of your study population actually had levels of lower limit of detection? Was that a large percentage or roughly small fraction? Right, so there was about 45% that had levels below .05 and about 17% that had levels below .015. So the lower limit of detection gives us a lot more data to work with. So we looked at CODE in two different ways. The first was dichotomous using all of the data from 99 to 2006. And we looked at CODE levels greater than or equal to .05 or less than .05. And we used prevalence. And for years 2003 to 2006 in which everyone had the lower limit of detection, we looked at CODE continuously and we naturally long-transform CODE and we looked at geometric means. We used SAS and 2DM software to analyze the data. And the 2DM software takes into account the unequal probability of selection and the response, the multi-stage design, and the correlation among sample persons within clusters. For dichotomous CODE, we looked at public ratios because the outcome was greater than 10%. And for continuous CODE, we looked at the ratio of geometric means. We adjusted for the following covariance. In model 1, we adjusted for age, race, gender, service level, ratio of family income to poverty, region of the country, a number of meals eaten outside the home, which is a proxy for amount of time spent in the restaurant. And we also, so we adjusted for these variables in model 1. And in model 2, we additionally adjusted for adult smoking prevalence. And adult smoking prevalence may be acting as a confounder in that it's associated with smoke for air laws and coating. However, it might also be acting as an intermediate variable, so it's on the pathway between smoke for air law and coating, in which case we wouldn't want to adjust for it. And so we look at it both ways. We also looked at effect modification of my home second-hand smoke exposure status. MS was assessed by, if at least one person smoked inside the home, all members were classified as having home second-hand smoke exposure. So these are some of our results. This is looking at prevalence ratios by all non-smoking youth and stratified by home second-hand smoke exposure. So the prevalence ratio comparing extensive to no-law coverage was 0.59 in model 1. And after additionally adjusting for smoking, adult smoking prevalence, it was estimated at 0.70. And among children from non-smoking homes, the association was stronger. It was the prevalence ratio was 0.51, which was estimated at 0.61. And in model 2, there was no difference between the limited and the no-law coverage categories. And among smoking homes, children from smoking homes, the prevalence ratio was 0.97 in both models, so indicating no association for extensive or limited coverage categories. And this is the distribution of coating on the natural log scale by smoke-free adult coverage. So this is using data from 2003 to 2006. So on the left-hand side are children from non-smoking homes and the right-hand side children from smoking homes. And so the dark black line is at coating at 0.05, so you can kind of see the distribution of those above and below this cut point. And the dotted line are the geometric means. So among children from non-smoking homes, going from extensive to no-law coverage, the distribution moves towards higher coating levels. However, among children from smoking homes, there's very little difference in the distribution of coating by the three smoke-free adult coverage groups. So these are our results. I'm looking at the ratio of geometric means. Among all non-smoking youth, there was a ratio of 0.47, comparing the extensive to the no-law coverage, which is insinuated to 0.58. Amongst children from non-smoking homes, the ratio of geometric means was 0.50, which is insinuated to 0.57. And again, we find no difference between the limited and the no-law coverage categories. And no difference looking at children from smoking homes, comparing the extensive or limited to the no-law coverage category. So in conclusion, among youth, not exposed to second-hand smoke in the home, there was a 39% lower prevalence of defectible coating and a 43% lower geometric mean, and no difference between the limited and no-law coverage categories. Among youth exposed to second-hand smoke in the home, there was no association between smoke-free air-law coverage and encode. So this figure just shows how our results compare to the Scottish study. And so our results are labeled in Haines. So the Scottish study also looked at children stratified by whether or not they lived in the smoker. And so among children in the home, we found similar results as well as stratified by having a smoker in the home. So our second paper looked at what's called smoke-free air-laws and asthma prevalence, sometimes of severity among non-smoking youth. And these are our co-authors. So our study population was very similar. They used data from Haines again from 99 to 2006. We looked at children aged 3 to 15 years. Non-smokers, again, defined by coating and youth who had not smoked in the past five days. And our sample size was 8,800 because we looked at children aged 3 to 15 instead of 3 to 19. Exposure, again, was exposure to smoke-free air-laws. And we combined the low and limited thought coverage categories into the category no smoke-free thought coverage in the county versus having a smoke-free air-law. For the age, what did you choose to do? Right, because a lot of the covariance that we wanted to adjust for were only collected until the age of 15. So we thought it would be better to limit it to 3 to 15 instead of 3 to 19. So we looked at, we were able to look at a number of different outcomes. So we looked at prevalent asthma using two different definitions. The first was self-reported current asthma. So youth who responded positively to both has a doctor or other health professional ever told you you have asthma and do you still have asthma who are classified as having self-reported current asthma? And we looked at the second definition called ever-asmo plus current symptoms. So youth responded positively to this first question and they had at least one of these following symptoms. Persistent weeks, chronic night talk or weeks of education use. They were classified as having ever-asmo plus current symptoms. We also looked at asthmatic symptoms alone. So if you had at least one of these symptoms, they were classified as having asthmatic symptoms. We looked at persistent ear infection and asthma severity having an asthma attack in the past year or we were just doing a visit for asthma. And we used SAS and CDM software to analyze the data. And we looked at effect modification by home second-hand smoke exposure and age. So we looked at the following covariates. In this column, we looked at some of the same covariates as in the previous analysis. Age, gender, race, ratio of family income to poverty to medical region. In this column, we looked at some additional covariates that have been found to be associated with asthma. So household size, health insurance, DMI, mothers of birth, mothers smoke during pregnancy, low birth weight and ever attend daycare at preschool. These are some of our results. And so the odds ratio is on the Y axis. And so for prevalent asthma using the definition of self-reported current asthma, we found an odds ratio of 1.08 indicating no association. However, looking at ever asthma plus current symptoms, we found an odds ratio of 0.74 indicating a reduced odds of having ever asthma plus current symptoms if you have less coverage of the smoke rate a lot. Looking at the symptoms total, so having at least one of these symptoms, we found an odds ratio of 0.67 and similar odds ratios looking at the symptoms individually. And for ear infection, we found an odds ratio of 1.01 suggesting no association. This figure looks at odds ratios for asthma attack and ER visit for asthma. And we found similar odds ratios of 0.66 and 0.55. The figure looks at the association between smoke free airlock coverage and ever asthma plus current symptoms and asthmatic symptoms stratified by home second-hand smoke exposure. And so for ever asthma plus current symptoms it looks like there might be a stronger association among children who do not live with a smoker. P value for the interaction term was 0.36. And for asthmatic symptoms it looks like there's no difference in the association between the smoke free airlocks and asthmatic symptoms whether or not you live with a smoker. This is a similar figure stratified by age and so for both outcomes it looks like there might be a stronger association among children aged 5 to 12 years of age. So that might reflect the prevalence of asthma by age that most asthma gets diagnosed at around 5. And then typically it would resolve around teenage years so that in terms of the number of people aged 5 to 12 that might be likely to report to having asthma. Right, we could get into smaller sample sizes. So in conclusion smoke free airlock is primarily reduced to symptoms associated with asthma but not the prevalence. And we found an automation of 0.67, 0.66 and 0.55. Asthmatic symptoms, asthma attack and we are visiting asthma spectrum. Compare the odds of the clinical outcomes comparing children that live with smokers versus children that don't live with smokers. I mean, ideally I hypothesize anyways that there is a stronger effect to the three posts among them the three posts locked among each other and adult by the smokers. Right, right. I did look at that I'm remembering the association between second between children who do live with second-hand smoke exposure and current asthma and it was similar, there was no association but for the for the symptoms there was a stronger there wasn't association between among kids who lived with the smoker had more asthma attack symptoms than kids who did not live with the smoker. So there's only been one other study that has looked at the impact of smoker airlock on children's health impacts on children and this was a study conducted in Kentucky that looked at emergency room visits for asthma before and after the implementation of the smoker airlock and they found an 18% reduction and we found a 45% lower odds of going to the York Asthma County with the smoker airlock compared to County without the smoker airlock and since there hadn't been a lot of studies looking at this association this figure looks at studies that have looked at the association between exposure in these health outcomes so for persistent weeds the Surgeon General's report found a hold odds ratio of 1.26 and a previous in pain study that looked at some more outcomes on an odds ratio of 1.3 comparing high coating levels to low coating levels and 1.1 comparing medium coating levels to low and our odds ratio is 1.72 and this is a similar figure about looking at prevalent asthma so the Surgeon General's report found an odds ratio of 1.23 and the previous in pain study found odds ratios of 1.5 and 1.1 and our studies found an odds ratio of 0.93 looking at current asthma and 1.35 looking at ever asthma less current symptoms so the last paper looks at the impact of Massachusetts smoke pre-workplace laws on acute cardio function deaths these are so this paper looks at a different data set looks at Massachusetts adults aged 35 years and older and there are 351 cities and towns in Massachusetts and before Massachusetts implemented a state law that already implemented a very similar law which is about 25% of the population so we were able to look at data from 1999 until 2006 and Boston implemented the smoking ban in 2003 and about a year later Massachusetts and the smoking ban July 5th and 4th so our exposure was exposure to the Massachusetts state smoke for airlock which banned smoking in all imposed public places and workplaces including restaurants and bars and there were some exceptions to this law and they're just listed here including private residences hotel rooms, tobacco stores and others the outcome was death from acute cardio function used primary cod ICD 10 code I21 this was obtained from a registry of death certificates and it was recorded as a number of deaths per day per town we had four questions that we wanted to look at with this data set and so we used Passan regression models and the first question was we wanted to know the overall impact of this state law so we used an indicator variable for the state law among Massachusetts as a whole and by age, gender and whether or not you had a prior smoking ban secondly we wanted to see if there was an impact on the local laws so we restricted the population to just those towns that had implemented a local law and time for the state law and we used an indicator variable again for the local law next we wanted to see if there was an interaction between the state law and a prior local law so our hypothesis was that among towns that had already implemented a local ban there would be no impact on the state law but among towns that had not implemented a local law there would be an impact on the state law and finally we wanted to look at the impact on the state law over time so we adjusted for the following covariance linear term for time season using annual and semi-annual sony cosine terms ambient PM2.5 weekly number of influenza cases city and town specific demographic characteristics obtained from the census listed here a random mix up for each city and the population age 35 or greater was included as an offset in the Poisson model so this figure shows time on the x-axis and the number of people covered by a a smoke fair law in Massachusetts on the y-axis and so in 1999 there were very few people covered by a local smoking ban and then after Boston implemented their smoking ban this number increased and it increased a little bit more because when the state law was implemented it increased even more Question on that Do I see in two months before the state law there were big increases or is that just or is it really that the month of the state law implementation in June 2004 you had this humongous jump or are there some data points months before then where you see the large jumps too Right here? Can you close it? What's the first month in which you see the slope of the roof? Oh it's July 2000 So there's no lead? No, this is just the state law because everyone was covered here Just to connect the dots So this shows the left y-axis the AMI mortality rate for 100,000 and so the pink line is the observed mortality rate and the black line is the predicted AMI mortality rate using data before the state law and adjusted for time and season and so about a year after the state law the observed rates looked to be below the predicted rates So this table shows the AMI mortality rates in total at about 3.3 million people and we had an adjusted 7.4% decrease in AMI mortality after the ban compared to before the ban and we found no significant difference by age or gender however when we looked at prior local smoking ban status we found that there was a decrease of 9.9% because it had not implemented a local ban and no decrease among towns that had already implemented a local ban similar to the state law I'm not sure everyone will know what those peaks and valleys in the previous slide So AMI had a seasonal pattern with rates greater in the winter than in the summer and so that's why there's these why is it greater in the winter I'm not quite sure I think it may be because there's more flu cases, more respiratory illnesses in the winter or So that begs the question whether you see the same peaks in California you see a seasonal pattern but not not as strong so we looked at the impact of local laws so we restricted the population to towns with the local law and towns before the state law and we found that AMI mortality rate decreased by 4.9% after the local law and before the law but this is not statistically significant the p-value is 0.32 and this looks at the interaction between the state and the local laws so this interaction term was statistically significant and so among towns that had implemented a state law there was no difference or had implemented a local law there was no difference at AMI mortality rate after the state law but among towns that had not implemented a local law there was a 9.2% decrease in AMI mortality after the state law compared to other states so this chart looks at time on the x-axis and the cumulative sum of the difference between the predicted and the observed on the y-axis and so the dark gray line looks at towns that had not implemented a local smoking ban so it hovers around zero meaning there's no difference between the observed and predicted until here in the state law there's a little decrease and then a sharper decline about a year after the state law and the lighter colored line looks at towns that had implemented a local smoking ban between May 2003 and June 2004 so this was when most of the local bans were implemented and also hovers around zero until the Boston law was implemented here and then later there was a decline sort of a parallel decline here after the state law so you would expect that eight months later those two lines will converge because it took Boston about eight months to move at some point right there it just seems longer then it took Boston to a definite effect right I mean 2007 data is now available that we can look at to see the trends continue when there's a decrease or a decline unless of course it implies that in places where the laws were forced by the state maybe compliance is in all that I think that was a question that gets at a larger point what are the carrots and the sticks of some local versus state laws I mean what's the strength of the Boston law versus the Massachusetts law is there any way that we can think of you've done a great job of separating them out and interacting with them to have any idea of what part of the team the cost for a bar is it thousands of dollars or is it a hundred dollars is there some legal ramifications I'm not quite sure about the penalties I think maybe at first there was a warning and then if you did it again you got fine but I think for the difference between the state and the local laws for local laws such as one that was implemented in Boston at the time you could smoke in bars and restaurants in Cambridge which is just across the river miles away and so if you wanted to smoke at a bar or restaurant you could just go to a different city whereas if you implemented the state law then it's much different to go to another state and so we looked at the impact of our time in the Russian models as well and so zero to 12 months after the state law there was a 1.6% decrease and zero to 12 months after the state law there was an 18.6% so in conclusion there was an overall 7.4% decrease in AMI mortality rates after the state's smoking law compared to before the law among towns without a local smoking ban there was a 9.9% decrease and steeper reductions were seen one year after the state law so prior to the state law about 25% of the population was covered by a local smoking ban and local bans were associated with a 4.9% decrease in AMI mortality and the state's smoking workplace law further reduced AMI mortality in these towns so this table compares our results in a to a similar study conducted in New York and among the 11 or so studies that have looked at this association ours was most similar to the New York study in geography in time as well as in New York they had implemented a lot in local counties more for air laws before they implemented their state law there were big differences we looked at AMI mortality in New York we looked at AMI hospitalizations but we saw a similar we saw 7.4% decrease in New York saw an 8% decrease the statements are technically accurate but they kind of do you believe that this is attributable to the smoking law but you saw that the overall marginal effect was a decrease in mortality associated with AMI at the time as you were going along so do you have just the attributable fact that you would estimate which would be what your model would have assumed pre-post the law okay I calculated it's in the backup slide I can try to find it I am interpreting that correctly though this includes including both the temporal trend in mortality as well as the change in the law right but we do adjust for the temporal trend in mortality so we include a term for time that adjusts for a linear downward trend in AMI mortality so we do adjust for that downward trend in these figures we'll talk a little bit about the strengths and limitations first of all to talk about in-gain data and so some strengths are first of all talked about the limitations there is general misclassification of smoke-free air-law coverage so some of my live in a county with a law but visit a county without a law or a child may live in a county but individual places may have implemented their own laws that we want to talk about there could be misclassification of current asthma status it was based on some report and not validated with medical records the cross-sectional survey so examining the timing of them can be tricky and there could be residual confounding we didn't have information on parental asthma status which is a strong risk factor for children with asthma some strengths so we had a large sample size we had an objective measure of second-hand smoke exposure from non-smokers at Kotlin we were able to look at Kotlin two different ways so we found similar outcomes and we were able to examine a number of respiratory outcomes for the Massachusetts study some limitations include misclassification of the local smoking bands as I had just mentioned there could be inaccuracies in using death certificates as a measure of death from AMI we did not account for the changing population over time so we used the same population since this population from 99 to 2006 and we did not have information on individual smoking history some strengths are that it was a large data set we were able to adjust for many total confounders and we were able to compare times with and about a local smoking band so our overall conclusions are in children and adolescents from non-smoking homes there was an association between smoke-free air loss and decreased exposure to the second-hand smoke as measured by a certain coating among children from smoking homes there was no association between smoke-free air loss and combined we saw an association between smoke-free air loss and decreased osmotic symptoms and in adults we found an association between the Massachusetts state smoke-free air loss and reduced AMI mortality rate and a larger impact one year after the ban and we found a smaller association between local smoke-free air loss and a reduced AMI mortality rate and these are the contributors that helped greatly to this research from my research committee we collaborated with the Massachusetts Department of Public Health the National Center for Health Statistics and other Harvard School of Public Health so thank you I have a couple the first one is excellent presentation two questions so I'm thinking of other state or county of factors that could influence smoking levels but not be caused by the law that might coincide with the law so one example is prices of cigarettes so we know that changing the state tax on cigarettes can dramatically change consumption do you know anything about the concurrence of state laws or perhaps state laws with cigarette taxes or with general educational efforts that seem to kind of be mustard with this the state law so anyway you can try to control for that right we did look at the impact of cigarette sales tax so in 2003 Massachusetts increased the cigarette sales tax and I looked at how that impacted cigarette consumption and so there was a sharp decrease in cigarette consumption right after the Massachusetts cigarette sales tax but then it kind of plateaued and so it didn't impact so at the time of the state law it was in that plateau okay so it proceeded a little bit second question is kind of what an economist would ask and I'm not an economist but you know we've seemed to convince in the show that there is a reduction in second hand smoke exposure among children in non-smoking homes so which leads to the conclusion that smokers are smoking less in these areas that previously were cannot ban but now they're banned well given what we know about addiction and smoking behaviors it makes me think well where are they going to smoke them because maybe their overall smoking models are they reduced or maybe they're just smoking somewhere else so they could smoke in another city like you mentioned they could smoke at home this interaction effect of smoking homes and non-smoking homes and we found that in smoking homes we didn't see codeine going down on the children and what I've seen in some of the economics literature is in fact kind of a perverse effect of the law and then smoking homes have increased the children have increased codeine not with codeine but there's increased smoking exposure at home because they're not going elsewhere what do you make of that explanation is that something you think could be viable right so I have that is a big concern that people if they can't smoke at bars or restaurants are going to smoke more in homes around their kids but there's been several studies I haven't looked at an economics study but there's been several public health studies that have looked at at that in Ireland and the Scottish study that I mentioned looked at it and they found no difference after the ban in the the number of people who smoked in the home so it was a big concern like tobacco control community but the studies that have looked at it have found it hasn't been the case could you have though an increase in the number of cigarettes being smoked at home but actually the same number of people the way you worded it I'm not saying criminals could go out it could be the same number of people smoking in the home maybe they're smoking their entire pack per day in the home as opposed to half the bar in Santa Monica it's just about to ban more smoking in apartment complexes so that basically some even in your home complex is getting a cigarette smoked and they could ban it in that area and public health has done an amazing job at changing norms for smoking but I just wonder how where are we pushing for people that have a perhaps a genetic assumption that we're changing the structure of who's smoking but the intensity is also a question my impression that I don't know the literature but the impression that there's also been a shift towards even addicted smokers smoking outside more as opposed to inside the home as education has kind of increased I don't know if you know but I think you're right like he said like a lot of changing social norms so smokers are more aware of the harmful impacts that it's having on them so they're trying to smoke like you said outside away from their kids I'm still a little concerned with what Dan was talking about because of the time trends even though you even though you adjusted for the the temple trend and the temple trend had to decrease a little high you said that you said that the population the base population was still 2,000 census so this was a Poisson regression did you put population offsets in the regression? yeah so each there was a population for each town each town had a population that was the offset in the wild but it was all from the 2,000 census and I looked at how the the population changed from 2,000 to 2,006 and it didn't change very much it was kind of a steady 0.6, 0.7 percent increase over that time period and there weren't any large chunks no big emigration or migration immigrants or anything like that okay you go back to that on the time trends slide that was interesting just one more time before we jump off sure that could be a concern I mean you do have a little bit of empirical data that says that children from homes where there are smokers the point is that you're still about one correct right, pretty close to law that right for clinical symptoms clinical asthma symptoms I'm sorry there was no difference that's right they're all about 0.971, 1.0 that's surprising at all when you do actual air monitoring inside a home where there are smokers the exposures are just way more than you would get just about anywhere else unless you go to a smoky bar when you ask these kids and they're exposed when we do continuous air monitoring inside a home that we thought was a non-smoking home we can see that it is a non-smoking home which we dislike in concentration and those concentrations especially when it's cool remain throughout the evening all night and all morning so these kids are not only exposed during the day they're exposed at night when everyone's asleep because the part of the film go away they remain suspended for many hours so I mean I'm not surprised at all I would expect that a smoking home would just overwhelm any kind of ban elsewhere because these kids are going, if they're kids they're going to school they're not going to school anyway they may go to a restaurant where else are they going to go so it could have been an adult who's exposed at their work in workplace but not a kid so I'm not surprised at all that's a logical finding and the studies have found that the home is not as critical as the company levels you mentioned that there was this cigarette tax put in in Massachusetts in 2003 was it? I have a figure if you'd like to see it that charts the cigarette consumption over time I don't believe you that it plateaus I was just wondering though if you can pick that up at all in these chain scores if that could be part of that because you might expect if you see that decrease in consumption there's really a length that would have had some effect on the time trend I think it was 2002 I don't know if that's easier to see on this or on the sinusoidal trend slide I was just wondering if you can actually see that in the MI pattern like any impact of the cigarettes or if there appears to be an indication of the decline there was definitely a decrease in cigarette consumption because you said you might expect from this that it could be 6 to 12 months after the tax it's put into place to see some kind of drop I can't remember the exact month I have it on another slide I was just curious did you think a little bit about trying to actually put that in the model to see if you could pick up that component of the trend like if there's a change in the linear trend like a spine or something can you tell us about that time of tax yeah, no I didn't look at that but it would be something if you could actually go back to that sinusoidal slide I just wanted to come to a follow-up question I guess it kind of speaks to a larger issue of I guess you assumed a linear decline for the trend so you don't need to part your get kind of founded with the post thing on kind of estimates so very sensitive I guess and part your change points in the linear trend that might be caused by something even if it's not at the exact same time I think if there was another like a change in the linear trend after 2003 after the cigarette tax I guess the argument is that it goes down and you maybe see the trend only last for a year or so after the sales tax so maybe you're okay I mean there are other ways to model the trend in linear models are good at looking at seasonality so I mean here's reasonable just looking at the data it's just but there has been a study I think it was conducted in Italy that association between the smoke rate allowed in AMI hospitalizations and yes they looked at AMI hospitalizations which is a little bit different than mortality but they found when they modeled time linearly they found an association but when they used a spline they didn't find an association so I think it would be interesting to use the spline on their data and see what we find it would be interesting to look out past the in two more years because there's a peak section the one peak that you have there in winter there's one that still spikes up above the predicted then the next year it does not I mean I'm not saying there's anything per se in that but if you have a smoking ban especially in Massachusetts very cold I mean it's very very cold and you go out or you're in any building everything gets sealed up in the winter so that's when you see the biggest impact really is in the winter when there's low infiltration when there's low infiltration rates it's a good point maybe it's just a lot of snow that winter it's a lot of shoveling it's a lot of shoveling yeah it's just one of the limitations it's not a trend it's hard to really stop that so it's been four or five years since this law there's still some exposure I mean children are knocked down at zero even those who don't believe in smokers is there something in your study that could support initiatives in California to extend the laws to outdoor public places like the parks from the beach I'm going to add the proposal right I think yeah there has been some laws implemented in California that bans smoking at beaches and parks but we didn't find a decrease you know we found no decrease in kids and so I think it would be helpful to look at interventions to try to help exposure in kids by reducing the amount of smoking in the home so you know possibly no pediatricians trying to reduce parental snowfall we're going to walk down to that veggie grill for anyone who would like to join us you're welcome