År 1997 påbörjades ett världsunikt projekt, ABIS-studien (Alla Barn i Sydöstra Sverige) på Hälsouniversitetet i Linköping (Linköpings Universitet). Av alla barn som föddes under perioden 1 oktober 1997 till 1 oktober 1999 i Blekinge, Småland, Öland och Östergötland har ca 17 000 följts med jämna mellanrum över åren: barnets födelse, ettårsuppföljning, 2-3-årsuppföljning, 5-6-årsuppföljning samt åttaårsuppföljning. Barnen/familjerna har inför varje moment i studien lämnat in biologiska prover och svarat på frågeformulär. Frågeformulären innehåller frågor av varierade typ och denna uppsats tar hänsyn till de sociodemografiska variablerna samt de variabler som användes för att ”mäta” stress hos föräldrarna och i vissa fall barnen. Genom åren har antalet frågeformulär som fyllts i minskat kraftigt från 16051 till 4030 i åttaårsuppföljningen. Med hjälp av den här uppsatsen vill vi undersöka om bortfallgruppen har speciella kännetecken och om orsaken till bortfall kan förklaras. Som underlag arbetar vi med de data som registreras genom frågeformulären med startpunkt i födelseformuläret, där alla som svarat antas vara populationen och de som därefter försvinner inför kommande uppföljningar är bortfall. För att kunna angripa problemet har multidimensionell skalning, klusteranalys och logistisk regression använts. Ingen av metoderna resulterade i att vi kunde särskilja observationer i två olika grupper som överensstämde med grupperna svarande och bortfall. Därför kan vi inte heller beskriva eller förklara bortfallet utifrån de variabler som valdes, mot sociodemografiska variabler och stressvariabler. / In 1997 began a unique project, the ABIS study (All Babies in Southeast Sweden) at the Faculty of Health Sciences in Linköping (Linköping University). Of all babies born during the period 1 October 1997-1 October 1999 in the counties Blekinge, Småland, Öland and Östergötland have about 17 000 been followed-up at regular intervals over the years: at birth, after one-year, after 2-3 years, after 5-6 years and after eight years. Children/families have for each moment in the study submitted biological samples and responded to questionnaires. The questionnaires contain questions of varying types; this paper takes into account the socio-demographic variables and the variables that were used to "measure" stress with the parents and to some extent with the children. Through the years the number of participants declined sharply from 16 051 filling out the first questionnaire to 4030 at the eight-year follow-up. With this essay we will investigate if the nonresponse group has specific characteristics and if the cause of the nonresponse can be explained. As a basis we work with the data recorded by the questionnaires, with starting point at the birth in which all who responded are assumed to be the population and then the ones who leave the study into the next follow-ups constitute the nonresponse group. In order to tackle the problem, multidimensional scaling, cluster analysis, and logistic regression have been used. None of the methods however made it possible distinguish observations in two different groups that correspond with the groups of respondents and dropouts. Therefore, we cannot describe or explain the nonresponse from the variables that have been chosen, i.e. socio-demographic and stress variables.

Passive smoking has been recognised as a health hazard, and chidren are especially vulnerable. The general aim of this thesis was to describe and analyse the importance of parents’ smoking and smoking behaviour for children’s tobacco smoke exposure. The studies were conducted in the South-East part of Sweden and pre-school children and their parents constituted the study samples. Five studies are described in six papers. Smoking prevalence among parents (14%) and commonly used measures of protection were surveyed. An instrument designed to measure children’s tobacco smoke exposure in the home was developed and validated. It was used on 687 families with a smoking parent and a child 2½-3 years old, included in a prospective cohort study on environmental variables of importance for immun-mediated diseases ABIS (All Babies in South-East Sweden). Almost 60% of the parents stated that they always smoked outdoors with the door closed, 14% mixed this with smoking near the kitchen fan, 12% near an open door, 7% mixed all these behaviours and 8 % smoked indoors without precautions. The smoking behaviours were related to the children’s creatinine adjusted urine cotinine. All groups had significantly higher values than had children from non-smoking homes, controls. Outdoor smoking with the door closed seemed to be the best, though not a total, measure for tobacco smoke protection in the home. Most parents were aware of the importance of protecting children from tobacco smoke exposure but all were not convinced of the increased risk for disease for exposed children. The majority of parents were not satisfied with the smoking prevention in health-care and 50% did not think that their smoking was of any concern to the child health care nurse. Further research is warranted to describe if the difference in exposure score related to smoking behaviours is related to different prevalence of disease. Efforts are needed to convince those who still smoke indoors that tobacco smoke exposure influence children’s health and that consequent outdoor smoking with the door closed seemed to give the best protection. / Article I: copyright (2003), with permission from Oxford University Press. On the day of the public defence the status of article III was: Submitted and the status of article VI was: Revised and resubmitted and the original title was: Attitudes to children’s tobacco smoke exposure among smoking and non-smoking parents and their opinions on how the issue is handled in health care.