8 hour diet men's health

>> thank you forthat introduction and thank you for coming. i'm very honored to be partof this event which is one of hundreds of eventsthat are going on around the world todaycelebrating international women's day in 2012. of course, the purpose ofinternational women's day is to highlight the achievementsof women, both past and present, and to look forward intothe future opportunities

that await women of the future. i believe that today's the38th official un international women's day, but severalcountries have been celebrating women's days forover 100 years now. so there's a longhistory to this. my focus today willbe on women in the uk in the domestic context, buti've included this picture of a yemeni woman herebecause i realized as i was putting thispresentation together that,

to my great shame, it'sbeen nearly 20 years since i did anything to officially honorinternational women's day, and that's when i was a peacecorps volunteer in the republic of yemen and i organizedan event highlighting the achievements of yemeni women. so i've included this picture of this woman wearingtraditional sunni [assumed spelling] yemeni dressreally just to remind us all

of the great diversityof circumstances that women find themselves inacross the globe as we talk about women and workin the uk today. so this lecture has been framedaround me dispelling myths about work and motherhood, and iwill be providing some evidence that suggests that some of themyths or ideas that we have about work and motherhoodmay not be true in the general population,but i just wanted to note that we are usingpopulation level data here.

we're talking about populationaverages so it's not to say that these myths or ideasaren't sometimes true for individual women. that can very much be the case. but i want to focus ontwo particular myths, if i can call themthat, and the first of those is sometimes calledthe role strain hypothesis, the role conflict hypothesis, sometimes called therole overload hypothesis,

and that basically saysthat combining paid work with motherhood is stressful and will therefore beharmful to women's health. and this is a very old idea. it emerged about 40 years agoin the united states, initially, and mainly in response to women's increasedlabor market participation at that time, and particularlythe increased labor market participation of mothers atthat time which led to concerns

about the impacton women's health of combining theresponsibilities of paid work with family responsibilities. so you might think 40 yearsis a long time and we've moved on a lot since then, and yes,of course we have moved on, but we do still see, as thesemedia representations show, a great concern in the mediaand in public discourse about the potentialharmful impact on women of combining workwith motherhood.

and perhaps that's because,although women's ties to paid work haveincreased fairly dramatically over the past several decades,their responsibilities in terms of unpaid domestic labor haven'treduced to the same extent. so this is an exampleof the amount of minutes that women pay --spend in unpaid labor, over and above theamount of minutes that men spend in unpaid labor. so the female excess inunpaid labor, if you like,

in minutes per day,in oecd countries. so these are relativelyadvantaged countries. so you can see that in placeslike india women are spending, on average, five hours aday more on unpaid labor than men are, but evenin some of the more -- we consider the moregender equal countries, of the nordic countries, womenare spending, on average, about an hour a day morein unpaid labor than men, and the uk's abouttwo hours a day more.

so this is stillan issue for women. so, as i mentioned, thisis a fairly old idea. it has been studied quite a lot,and what studies have tended to find is that womenwho do combine paid work with family responsibilitiestend to actually be healthierthan women who don't. but the vast majority of thesestudies are what we call cross sectional studies, sothey measure work, family, and health at onesingle point in time.

so they're not able toactually provide evidence as to whether combiningpaid work with family responsibilities isbeneficial for women's health or whether, perhaps, womenwho have better health earlier on in their life courses arethose who are more likely to develop careers,form families, and have better healthlater in life. so what we need to reallythis question is information about work, family, and healthacross women's life courses.

and we're very lucky in the uk because the uk is really aworld leader in terms of studies that follow individualsover time, what we call longitudinalstudies. and particularly the ukis a world leader in terms of what we call birthcohort studies. so these are studies that followpeople from cradle to grave, and there are four ofthem currently in the uk with a fifth beingplanned at the moment.

so there's a study that's beenfollowing people who were born in 1946, 1958, 1970, and themost recent birth cohort study that we have in the uk isthe millennium cohort study of children born in 2001. and, as i say, there's a fifthstudy currently being planned. so we -- i wouldlike to give thanks to the uk researchcouncils for their foresight in funding these studies,and also to the participants who give up their timeand information which is

so important in these studies. so today we're looking atthe oldest of these studies which is the mrc national studyof health and development, the 1946 birth cohort study. so this was a studythat took people who were born during oneweek in march in 1946 and have followed a randomsubset of those people, over 5,000 of thosepeople, ever since. so they're turning66 this month.

and what i've listed here arethe ages at which a wide variety of information has beencollected on these people in adulthood, and a lot ofinformation collected as well in childhood, and i've justindicated the years along the bottom to provide you with somehistorical context of the lives of people in this study. so this data set isreally the perfect data set for answering thesequestions that we want to look at which are, "whatis the relationship

between long termwork and family roles, and subsequent healthamongst women in midlife?" and, "are relationshipsbetween work and family roles and health explained byearly life predictors?" so health in early life, butalso socioeconomic circumstances in early life, that partlydetermine people's subsequent careers, family situations,and health. so the health outcomesthat we're looking at here in midlife are a measure ofself reported health at age 54.

this is just a general question. self reported health, whichis used quite frequently in epidemiological andsocial science studies. and here women are simplyasked how they would rate their health: excellent,good, fair, or poor. and if they say fair or poorhere we've considered them to have poor selfreported health at age 54. but in addition to thatsubjective measure of health, we also wanted to look at a moreobjective measure of health,

and at age 53 interviewersweighed and measured these people sowe have objectively measured -- measures of body mass index. so we're also lookingat obesity at age 53 as a more objective measureof health in midlife. so i wanted to start justby showing you the marital and parental statuses of thewomen in this study at each age of adult data collection. so this is the proportionof women who are married,

with and withoutchildren, never married, with and without children,or previously married, with and withoutchildren, within each age. and what i reallywant you to take from this is how verytraditional this particular generation of women were interms of their family formation. nearly 85 percent of the women in the study weremarried by the age of 26. two-thirds were mothersby the age of 26.

and the median age offirst birth was 23. so, half of the women in the study weremothers by the age of 23. work by kath kiernan has shown that this particular generationare also the most uniform, in terms of their age ofmarriage, of any generation that we know of,either before or since. so, a really -- a verytraditional group of people in terms of family formation.

this is work statusat the same ages. so you can see at age26 about half of women when they have young -- a lot of these women haveyoung children at home. about half of them are fulltime homemakers at age 26. by age 36 that proportion of full time homemakinghas reduced a little bit, and there's been a big increasein part time employment. so about -- overa third of women

in the study are employedpart time at age 36, and by age 43 most of the womenin the study are in paid work. about 85 percent of thewomen are in paid work, and there's a fairly evensplit between full time and part time employment. so what we wanted to dowas to take that work and family informationfrom across those ages, and try to bring it together. so we've used that information

to create what we're callingwork family histories. so our first group is -- we'vecalled the multiple roles group. and this was the most commonpattern amongst the women in this study. so 38 percent of the women arein the multiple roles group, and these are womenwho are married to one person throughoutthe study, they're mothers, and they have relatively strongties to the labor market. so about half of thewomen in this group were

in paid work throughout -- at each of the ages of datacollection, and about half were at home full time at age 26 and then entered thelabor market subsequently. we have a group of childlesswomen, and about half of the women in thisgroup are married, and they have fairly strongties to the labor market. a group of homemakers who,like the multiple roles women, are married to one personthroughout, they're mothers,

but they have weak tiesto the labor market. so they're either not inpaid work at any of the ages or they might be in parttime employment at one age. we have a group of lone mothers, and in this studythe vast majority of lone mothers are lone mothersas the result of a divorce or separation, and a groupof remarried mothers, so lone mothers who'veremarried by age 53. and then a group that we'vecalled intermittently employed

married mothers. so these women, likethe multiple roles women and the homemakers, are marriedto one person throughout, they're mothers, but theirlabor market attachment is between that of themultiple roles group and the homemaking group. so they kind of pop in andout of the labor market. so this is just showing youthe proportion of poor health, both poor self reported healthat age 54 and obesity at age 53,

by each of these workfamily history groups that i've just described. so what you can see is women inthe multiple roles group here in the dark blue barare the least likely to report poor health atage 54, and when we look at obesity it looks like it'swomen who have the weakest ties to the labor market who arethe most likely to be obese. so, homemakers andintermittent employed women. now, this is a similar idea,

but here we're lookingat odds ratios. so, for those of you who aren'tfamiliar with odds ratios, they show us thelikelihood of, in this case, reporting poor health atage 54 for each of the women in the categories,compared with women in the multiple roles group. so women in the multipleroles group are set to one. they're the reference. and the black circlesrepresent the odds ratios

or the likelihood of reportingpoor health for women in each of the other groupscompared to those women so, for example, the odds ratiofor the homemaker group is two. so they're twice as likely toreport poor health at age 54 as women in the multipleroles group. and the black lines around thoseodds ratios are what we call 95 percent confidence intervals. so they mean thatwe're 95 percent sure that the true odds ratio liessomewhere on that black line.

so if the black lineisn't crossing one, then we consider theodds ratio for that group to be significantly different. we're 95 percent sure thatit's different from women in the reference group, womenin the multiple roles group. so what we can see here is thatchildless women, homemakers, and lone mothers are allsignificantly more likely to report poor health atage 54 compared with women but the reason we usethese odds ratios,

and we use regressionmodels to drive them, is that we can then takeaccount of other factors that we think might be importantin explaining the relationships that we're looking at here. and here we're particularlyinterested in those early life predictors. so is early health, early socioeconomiccircumstances explaining these relationships?

so we put those intothe regression models. so here we're adjusting forself reported health at age 26, whether or not womenhave had some kind of mental health episodebetween ages 15 and 32, and father's socialclass at age 11. so what we're looking at now isthe relationship between health and these work family histories, independent of theseearly life factors that we've includedin the model.

so if they explain ourrelationship we'd expect to see no relationship here,but what we actually see is that the relationshipschange very little when we include those earlylife markers into the models. the odds ratios andconfidence intervals for lone mothers decrease alittle bit because they come from more disadvantagedchildhood circumstances. so a little bit of thatrelationship is explained by childhood social class,

but really the relationshipsremain very much as they were prior to adjustingfor early life factors. we're now looking at thesame thing for obesity. so you can see that women inthe homemaker group are more than twice as likely to be obeseat age 53 compared with women and, again, when weadjust for bmi at age 26 and father's socialclass at age 11, and we also ran these modelsadjusting for bmi at age 15 and women's educationalattainment,

we see very much the same thing, that these aren'texplaining our relationship. so homemakers remain twiceas likely to be obese at age 53 compared to womenin the multiple roles group, and once we account for theseearly life factors we actually start to see a significantrelationship for women in the intermittentemployed group as well. so those women tended to come from more advantagedchildhood circumstances.

so once we adjust for father'ssocial class at age 11, a relationship emergesfor that group. so one of the thingsthat we need to think about when we'restudying the same people over time is the fact thatmeasures on the same person over time are correlatedwith one another. so your bmi early in your lifeis likely to be correlated with your bmi laterin your life, and to do this properly we needto use techniques which account

for that correlationwithin individuals. so we used something calledgeneralized estimating equations which account for that withinindividual correlation. so here i'm showingyou mean bmi. so this is average bmi by agefor the two most extreme groups: so the multiple roles groupand the homemaker group, using these generalizedestimating equations to account for that within personcorrelation. and on the y axis i've shownyou the who guidelines in terms

of what's considered to be ahealthy bmi, overweight, obese. so what you can seeis that the women in the two groups startoff in the same place, and unfortunately bmiincreases for everyone with age, but it's increasing at agreater rate, more quickly, for women in thehomemaker group. so by age 36, on average, women in the homemakergroup are starting to enter into that overweightcategory, whereas for women

in the multiple roles group it'shappening a little bit later at age 43. so our conclusions fromthis were that women who combined paid work andstable family relationships over the long term endedup healthier in mid life than full time homemakers, andalso healthier than lone mothers and childless women, and theywere less likely to be obese than women who had relativelyweak ties to the labor market, and this doesn't seem tobe explained by health

or socioeconomic circumstancesearlier on in their life course. so i'd now like to turnto myth number two. when we published the results from the study i've just shownyou, one of the reactions in the media was, "well,that's great for the women, but what about theirpoor children? what kind of detrimental impactis this maternal employment having on the children?" so we wanted to startto look at this.

and, as you can see fromthese media cutouts here, this is something we hear a lotabout, as well, in the media and just in popularconversation: the impact of maternalemployment on children. but what does the evidence say? [inaudible] quite a lot. a lot of the work in this areahas focused on educational and cognitive outcomesfor children. so: reading scores, mathscores, that kind of thing.

and the results of those studieshave really been quite mixed. some of them suggestthat it depends on the socioeconomiccircumstances of the household, the age of the child whenthe mother goes to work, the number of hoursthat the mother does, but it's a real mixedbag of results. nothing terriblyconclusive there. there have been a few studiesnow linking maternal employment with child overweightand obesity.

we've seen that inseveral studies now. we wanted to lookat the relationship between maternal employment andchild psychological well being, what we call socioemotional well being. so our research questions were:are children whose mothers are in paid work in the early years,over those first five years of life, more likely thanchildren whose mothers are at home full timeto show emotional or behavioral symptomsat age five?

and are children moresensitive to the effects of maternal employment inthat first year of life because that's beensuggested by some of the studies thathad been done? so here we're using the mostrecent british birth cohort study, the millenniumcohort study, and this is a national studyof nearly 19,000 children who were born in the ukbetween 2000 and 2001, and so far information hasbeen collected at 9 months,

3 years, 5 years, and 7 years. today i'm just going to showyou information up to age 5. so first this isjust the prevalence of maternal employmentat those first three ages of data collection. so 9 months, 3 years,and 5 years. whether or not mothersare at home full time, working part time in paid work,or full time in paid work. so at 9 months you can seethat a little over half

of the mothers are in paid work. most of the paid workis part time employment, and with each subsequent age, as children are growing a littlebit older, there's a little bit of an increase in thematernal employment, but not big increases. about 3 percentagepoints between each sweep, and most of that increaseis in part time employment. and i wanted to showyou the prevalence

of maternal employment by someof these other key factors that are likely to be animportant part of this story. so maternal educationalqualifications, household income, maternaldepressive symptoms, and partner's work status. so, as you mightexpect, the prevalence of maternal employment increases with maternal educationalqualifications. so nvq5 is equivalent to atleast some higher education.

maternal employment increaseswith increasing household income until we get to thehighest quintile of income when it drops off a little bit. and mothers who are inpaid work are less likely to have depressive symptoms thanmothers who aren't in paid work. and mothers who areliving with a partner who's in paid work are mostlikely themselves to also be in paid work. mothers who are livingwith a partner who isn't

in paid work are actuallythe least likely to be in paid work themselves. okay. to measure a child's socioemotional behavior we're using the strengths and difficultiesquestionnaire, the sdq. this is a standard questionnairethat's used in a lot of the big surveys tomeasure child socio emotional well being. it has 20 items whichi've briefly listed her, and covers these 4 domains.

so emotional symptoms, conductproblems, hyperactivity, and peer relationships. so here we're using the topdecile, the top 10th percentile of scores on the sdq,as our definition of behavioral difficulties. so here i'm showingyou the prevalence of behavioral difficultiesat age 5. first, in the mauvebars, without looking at maternal employment, justlooking at gender differences,

you can see thatboys are more likely to have reported behavioraldifficulties at age 5 than girls, and this is a wellknown finding in the literature. and then looking atbehavioral difficulties by maternal employmentat 9 months. so trying to look at theimportance of that first year of life in terms ofmaternal employment. and what you can seeis that the prevalence of behavioral difficultiesincreases

with mothers decreasingattachment to the labor market. so boys and girlswhose mothers are at home full timeare more likely to have behavioraldifficulties reported than children whosemothers are in paid work. so this is -- these areunadjusted relationships, it's important to remember,so they don't take account of the things thatwe've seen are linked with maternal employmentlike maternal education,

household income, butthey certainly suggest that there isn't adetrimental effect of maternal employment, anyway. we then wanted to go on and look at cumulative maternalemployment across those first 5years, and we've done that with this really rathercrude measure, i have to say, of cumulative maternalemployment over the three ages. so simply or not mother is inpaid work at all three ages,

two of the three,one of the three, or none of the agesof data collection. so this includes both fulland part time employment. so a little over 40 percentof the mothers have been in paid work at all three ages. about a quarter havenot been in paid work at any of the three ages. and 16 to 17 percent have beenin paid work at one or two of the ages of data collection.

so here, again, areour odds ratios, and these are the likelihood ofbehavioral difficulties at age 5 by that cumulative measureof maternal employment. so what you can see isthe likelihood of reports in behavioral difficultiesat age 5 is increasing so, for example, boyswhose mothers haven't been in paid work at any of theages are twice as likely to have reported behavioraldifficulties compared with boys whose mothershave been

and the relationship ismuch stronger for girls than for boys, and this is asignificant gender difference. so for girls whose mothershaven't been in paid work for any of the ages they'resix times more likely with girls whosemothers have been in paid work at allthree sweeps. but, again, these areunadjusted models. so when we include those factorsthat i showed you are important and linked with maternalemployment you can see

that they do explain therelationship, at least for boys. so once we include thingslike maternal education, what we see is thatpotential beneficial effect of maternal employment forboys is actually explained by the fact thatthose are the mothers who have higher educationalqualifications and live in higher income households. but for girls even oncewe take those factors into account girls whose mothershaven't been in paid work at any

of the ages remaintwice as likely to report behavioraldifficulties. they're not reportingthem, i should say. compared with girlswhose mothers have been so, despite the factof accounting for maternal education,household income, maternal depressive symptoms, there's still thisresidual beneficial effect of maternal employmentfor girls, it seems.

so our conclusions are thatwe don't see any evidence of a detrimental effect of maternal employmentin the early years. if anything, there'ssome suggestion of a beneficial effect, but thateffect is explained for boys by the fact that thosemothers are those that have highereducational qualifications, live in higher incomehouseholds, but for girls it'snot entirely explained

and we don't yetknow why this is. so i just want to finish by revisiting the first studyi showed you very briefly, and talking a littlebit about the work that we're doing nowmoving on from that. so i showed you for those womenwho were born in 1946 the women who had stronger tiesto the labor market, who had stable familyrelationships, were those who hadbetter health in mid life,

but we need to remember thatthose women were prior to a lot of the big social changes thatwe've seen in relation to work and family and genderover the past few decades. so they started theircareers and began to form their familiesbefore we had gender equality in education, as we do now, before the continuedstrengthening of women's ties to the labor market,particularly for those women who have those highereducational qualifications,

also before the increasesthat we've seen in household inequality. and we've also seen increasesin inequality amongst women, between those women who aregetting those higher educational qualifications andthose women who aren't, and they're really facingvery different circumstances. also prior to the bigdiversification we've seen in family forms. so you saw how traditionalthis generation were in terms

of their family formation,and they were prior to the big increases thatwe've seen in cohabitation, for example, birthsoutside marriage, etcetera. and also potentially prior tosome increases that we've seen in paternal participationin family life. so what we want to know nowis are those relationships that we've seen in thatparticularly traditional generation of women-- do they hold true for more recent generationsof women?

and also how are these socialchanges impacting on the health of men and children becausethis area has tended to focus on the health of women, and as gender roles arechanging it'd be interesting to see the impact of thaton men's health as well. so we're starting to lookat these relationships now in those more recentbritish birth cohorts, to look at whether thereare generational differences in the relationships that we'veseen, and just a quick plug:

i'll be advertising a phdstudentship any day now to work on some of this stuff. so if anyone thinks they'reinterested in working on it, please come and speak with me. so that just leaves me to thankcolleagues who've worked with me on this work: professoryvonne kelly, professor mel bartleywho's director of the international centre forlifecourse studies in society and health where i work,

professor diana kuhwho's director of the 1946 birthcohort, and trish crowley who was instrumental in today,and particularly the economic and social research council whofund the icls research center. so thank you. [ applause ] >> thank you very much, anne, for sharing theseexciting analyses. we have time for a few questions

if there are anyin the audience. there's a mic coming your way. >> thank you forthis presentation. i am wondering about --in the second analysis, about the health of child. you're taking into account theincome, but you don't take it into account in thefirst analysis, and i was wondering whetherincome could have an impact -- >> i didn't accountfor -- i'm sorry?

>> income. the income in the -- >> of the household. for the children. >> no. you take it into accountfor the children's health, but not for the mother -- >> not for the mothers. yes. that's a good point. the measure of income inthat particular study isn't

brilliant, and wedid look at it, but it had an even weaker effectthan occupational class even. so i haven't shownthose results. but yeah. no. it's -- we did includeit at one stage. >> why was bmi used as ameasure of women's health? that seems quite limited. >> it is limited. we could have -- there were onlytwo objective measures of health

that we could haveincluded over the lifecourse because not all themeasures have been collected at every age. so it was either bmior blood pressure. we could have lookedat blood pressure. i chose bmi because icould imagine, hypothesize, kind of psychosocial,psychological pathways through which perhaps if womenwere unhappy with their family or work circumstances thatmight lead to greater bmi,

but certainly we -- it is limited, and we couldlook at blood pressure. we should do that, probably. >> yeah. wouldn'thomemakers have more time to report ill health than peoplewho are at work, you know, because, you know, if you had,like, the flu or something and you were at work maybeyou wouldn't report it as much as someone who's at home allthe time and had the time to go? >> yeah. do you meanin terms of like going

to the doctor andhaving time to -- [ inaudible response ] >> in the reporting. we can't really look at whetherthere's a reporting bias, but these are allcollected in the same way so they're all collected by an interviewer goingto people's houses. so it's not that it's comefrom health service use or anything like that.

>> thank you. those prevalence charts of womenwho -- maternal employment? do they take into accountmaternity leave, in the sense that if, you know -- ifthey were in paid work does that also mean they couldhave taken a year off for maternity leave? >> so they're taken at 9months, and women who are on maternity leave areincluded in the statistics. so they would be consideredto be in paid work.

yeah. >> have you taken intoaccount the number of children that the women have? >> we did for theobesity finding because someone suggested to us that this homemakereffect could be because they have more children. so we did look at that, andit did attenuate a little bit, but certainly notenough to -- yeah.

it didn't explainthe relationship. so the homemakers did havemore children than women in the multiple rolesgroup, but when we adjust for that it doesn'texplain the relationship. >> i'm intriguedby the differences between boys and girls. do you think there's anythingto do with the role models so when the womenare in employment that it has a positiveeffect on the girls?

>> it could be. we can't look at that, idon't think, in this study. we did speculatethat in the paper. and we also -- what ihaven't shown is we looked at the employmentsituation of both parents where there were twoparents in the household, and we saw that for the femaleheaded household the boys were doing worse, but in thetraditional male breadwinner households the girlswere doing worse.

so there were other suggestionsof gender differences there for the children, as well. so it'll be really interestingto try to figure that out. >> thank you for the talk. and i was just wonderingif you thought about the policyimplications of your work. >> well, that's an easy one. >> i would say that theresults here suggest that it's certainlynot detrimental,

and probably quite beneficial,to support women to be able to combine paid workwith motherhood, and finding better waysof doing that in terms of childcare provisionand -- yeah. so domestic labor, all kindsof things that hold women back from paid work, absolutely. but i haven't worked outspecific policies yet. >> this is probably a goodtime to end this talk. so please join me inthanking anne, again,

for sharing these analyses. we look forward to them.