METHODOLOGY
3.3. MAIN STUDY
3.3.2. SAMPLE SELECTION METHODS
3.3.2.1. Great Britain
The British sample consists of the free-living sample of the NDNS oral health survey, which in turn is a subsample of the respective sample of the main NDNS survey. The free-living sample of the NDNS main survey consisted of around 1250 individuals and was selected using a multi-stage random probability design, with postal sectors as the first stage units. The Postcode Address File (PAF), a computerised list maintained by the Post Office of all addresses and delivery points, was used as the sampling frame. The frame was stratified according to region and 1991 census data for social class. Eighty postal sectors were randomly selected as first stage units, and 375 addresses from each sector were again randomly chosen. The sample consisted of individuals aged 65 years or older living in the selected addresses.
All subjects who completed the various stages of the main survey were contacted for participation in the oral health survey. All dentate subjects that agreed, were included in the oral health survey sample. On the other hand, collecting data from all edentulous subjects that consented to participate would be neither cost effective nor necessary. Consequently, the edentulous subsample of the oral health survey consisted of randomly selected approximately 60 per cent of those agreed to take part. This study consisted only of the free-living persons participating in the NDNS dental survey, thus excluding those living in long-stay institutional accommodation. Apart from that, eligible individuals succeeded in a standard memory exercise aiming to screen for subjects, whose responses would cast doubt on the validity of the data. This Memory Questionnaire (Appendix 1) was administered during the NDNS diet survey.
3 .3 .2.2. G reece
The Greek sample is an opportunity sample, recruited from the people aged 65 years or over living in two municipalities of the capital city, Athens. According to the 1991 Census, the total population of this area was 3,072,922 persons, that is around 30% of the population of the whole country. Of those, 395789 persons were aged 65 years or over. This number represents 12.9% of the total population of Athens and 28.2% of the 1,404,352 people aged 65 years or over in the whole country (National Statistical Service of Greece, 1994).
The municipality of Athens is the highest populated in Greece and covers a wide area of the city, including the city centre. The sample was drawn from the participants of a
holiday camp for elderly people. The participants were in general characterised by a relatively lower socioeconomic status, as priority was given to subjects with low income (under 1.5 million drachmas). As it was not feasible to include everyone in the study, houses of the camp were randomly selected and all subjects belonging to the eligible age group were asked to participate in the study. The two houses with mentally handicapped people were excluded, because the validity of their responses would be limited.
The second chosen municipality (Holargos) is situated in the north suburbs of Athens, and it is a typical area of middle class people. A random sample was selected from the people aged 65 years or over living there. Initially, the map of the municipality led to the identification of all the streets, which comprised the sampling units. A random sample of streets was selected and all the houses and flats were visited in order to identify the people aged 65 years or over that lived in the selected area. Additionally, the list of members of the Centre for the Rehabilitation and Protection of the Elderly (KAPI, Kentro Apokatastasis kai Prostasias Ilikiwmenwn), which is a day centre without membership restrictions, were also used for the identification of the eligible subjects living in the selected streets. Once identified, they were asked to participate in the survey. In case of failure to contact the residents of a flat or house, two more attempts were made in different days and times.
The chosen boroughs and sampling methods should collectively facilitate the selection of a sample that covers a relatively wide range of socioeconomic status. This was confirmed by the Real Estate Values (Antikeimeniki Aksia Akinitwn), an official list reflecting the estimated price of buying a square meter (m^) of land (Ministry of Finance, 1998), which can be used as a proxy measure of the socioeconomic status of an area. This would further facilitate the comparison with the British sample, which being a national sample is bound to guarantee diverse social status representation.
Apart from the exclusion of subjects living in long-stay institutions or being hospital patients at the time of the survey, the only criterion for the selection of subjects was their ability to comprehend the questions. Mentally handicapped subjects were excluded from the study, as a result of the invalidity of their responses. The determination of the subjects’ eligibility for participation was performed through a cognitive test, which consisted of five very simple questions with obvious answers and was administered to all persons that agreed to participate in the study (Appendix 2). Some of the questions of the cognitive test were adopted from the short form of the Orientation-Memory- Concentration Test (Katzman et al, 1983). Subjects that failed to answer correctly two or more of those questions were excluded from the study.