• Open Access

Challenges in comparing the methods and findings of cohort studies of oral health: the Dunedin (New Zealand) and Pelotas (Brazil) studies

Authors


Correspondence to:
Professor WM Thomson, Sir John Walsh Research Institute, School of Dentistry, The University of Otago, PO Box 647, Dunedin, New Zealand; e-mail: murray.thomson@otago.ac.nz

Abstract

Objective: to systematically compare methods and some findings from two prospective cohort studies of oral health.

Methods: This paper describes and compares two such population-based birth cohort studies of younger adults: the Dunedin Multidisciplinary Health and Development Study (conducted in New Zealand); and the 1982 Pelotas Birth Cohort Study (conducted in Brazil).

Results: The two cohorts showed socio-demographic similarities and differences, with their gender mixes being similar, but their ethnic compositions differing markedly. There were some important similarities and differences in methods. Overall dental caries experience was higher among the Dunedin cohort. Each of the studies has examined the association between childhood-adulthood changes in socio-economic status and oral health in the mid-20s. Both studies observed the greatest disease experience among those who were of low SES in both childhood and adulthood, and the least among those who were of high SES in both childhood and adulthood. In each cohort, disease experience in the upwardly mobile and downwardly mobile groups lay between those two extremes.

Conclusions and implications: There are important similarities and differences in both methods and findings. While the need for a degree of methodological convergence in future is noted, the two studies are able to use each other as replicate samples for research into chronic oral conditions.

The most prevalent and important oral conditions are cumulative and chronic in nature, developing over many years. The prospective cohort study design underpins the life course approach to oral epidemiology,1 whereby the outcomes under study are considered in relation to the accumulated effects of various detrimental and beneficial exposures through life. Cohort studies require the identification and baseline characterisation of a (usually) representative sample which is then subsequently assessed on at least one further occasion. The most informative cohort studies make repeated assessments over many years, with each subsequent measurement considerably increasing the usefulness of those which preceded it.

However, prospective birth cohort studies are not common in oral epidemiological research. There are a number of reasons for this. First, such studies take time; for a meaningful contribution to be made to life course research, a birth cohort usually needs to be followed for many years. The reality for investigators is that the study is likely to outlast their research careers (with a concomitant need to develop associates who are able to continue the work). Second, prospective cohort studies are expensive, and funding bodies may be reluctant to take on what may be a number of decades of commitment which (in turn) reduce the funds available for other projects. Third, loss to follow-up is a very real problem, and a great deal of effort is required to maintain follow-up rates at an acceptable level. Outweighing these disadvantages are the considerable benefits of the prospective cohort study for life course research. Foremost among these is the prospective measurement of exposure to the relevant antecedent factors; this minimises the possibility of recall bias and enables distinctions to be made between risk factors and risk indicators,2 thus identifying the exposures which really matter. Moreover, the prospective measurement enables the elucidation of the natural history of oral conditions.

A key principle of scientific investigation is the replication of findings.3 It is possible that a particular risk factor's association with an outcome observed in one study may not be found in another (local environmental effects notwithstanding). If it is found, the case for a causal association is strengthened, especially if the social and economic circumstances of the two cohorts are dissimilar.4 The scarcity of prospective cohort studies in oral epidemiology (especially those with at least one childhood, adolescent and adult measurement in a general population sample) means that replication of cohort study findings remains the exception rather than the norm, and makes it important that the methods and findings of any existing cohort studies are compared systematically.

Only two population-based birth cohort studies have investigated the natural history and associations of oral conditions from childhood to adulthood. The Dunedin Multidisciplinary Health and Development Study is a longitudinal study of a birth cohort of 1,037 children born at the Queen Mary Hospital, Dunedin (New Zealand) between 1 April 1972 and 31 March 1973. Periodic health and developmental data collections have been undertaken at ages 3, 5, 7, 9, 11, 13, 15, 18, 21, 26 and 32, with dental assessments conducted at ages 5, 9, 15, 18, 26 and 32 (to date). The 1982 Pelotas birth cohort study is a prospective population-based study of all children born in Pelotas (Brazil) in 1982.5 During the whole of 1982, the three maternity hospitals in the city were visited daily, identifying 6,011 infants born to mothers resident within the Pelotas urban area. From birth registration and from a city census, another 46 children (0.8%) who were delivered at home were identified. A total 5,914 live-born infants constitute the original cohort. Periodic socio-economic, health and developmental data collection were carried out at birth and at ages 1, 2, 3, 13, 15, 18, 19 and 23, with dental examinations in the year when participants turned 15 and again when they turned 24 years old.6,7

The two studies have made important contributions to knowledge of the natural history and antecedents of oral conditions. For example, the Dunedin study has shed new light on the natural history of dental caries8–10 and periodontitis11 and on the risk factors for the latter.12–14 In addition, it has contributed knowledge on the impact of childhood oral health and socio-economic conditions upon adult oral health.15 The Pelotas study has contributed to our understanding of the role of poverty and family socio-economic trajectories during the life course in the occurrence of oral conditions in adolescence6 and in young adulthood.16 Moreover, its findings on the influence of malocclusion on self-perceived appearance, regardless of other physical conditions, indicate the importance of oral health in the psychological aspects of adolescence.7

This paper compares the methods, cohort characteristics and some findings of the Dunedin and Pelotas oral health cohort studies in order to identify similarities and dissimilarities, and to facilitate and inform future comparisons of their findings.

Methods

The Dunedin Multidisciplinary Health and Development Study

Dunedin is a city in the south of New Zealand's South Island. Its population is about 125,000; at the time of the study's commencement, it was just over 111,000. The city's main economic activities are education, research (it has New Zealand's oldest university), manufacturing and technology, and tourism. Dunedin has had a community water fluoridation scheme since 1967.

The Dunedin Multidisciplinary Health and Development Study is a longitudinal study of a birth cohort of children born at the Queen Mary Hospital (Dunedin) between 1 April 1972 and 31 March 1973.17 That institution was Dunedin's only obstetric hospital. The sample that formed the basis for the longitudinal study was 1,037 children, all assessed within a month of their third birthdays, and represents 91.0% of consecutive births. The participants’ families represented the full range of socio-economic status. Periodic socio-economic, health and developmental data collections have since been undertaken (at ages 3, 5, 7, 9, 11, 13, 15, 18, 21, 26 and 32), with dental examinations conducted at ages 5, 9, 15, 18, 26 and 32. The age-26 data collection was undertaken in 1998 and 1999. Information on self-reported oral health, dental service-use and self-care was gathered using a self-completed questionnaire. Dental examinations for caries and missing teeth were conducted using World Health Organization (WHO) criteria18 and undertaken in a conventional dental chair with clinical lighting, and dental examiners were not aware of participants’ questionnaire responses at the time of the examinations.

Made for the first time at age 26, periodontal measurements (not conducted on individuals who reported a history of cardiac valvular anomalies or rheumatic fever) involved two quadrants: the upper right and lower left for those whose study ID number was odd; the upper left and lower right for those with an even ID number; the mix of odd and even numbers was approximately 50:50. Three sites (mesiobuccal, buccal, and distolingual) per tooth were examined. Probing depth (PD; the distance from the probe tip to the gingival margin) and gingival recession (GR; the distance from the gingival margin to the amelocemental junction) were recorded using a National Institute for Dental Research probe. Midbuccal measurements for molars were made at the midpoint of the mesial root. All measurements were rounded down to the nearest whole millimetre at the time of recording. Plaque accumulation on six index teeth was measured using a modification of the Simplified Oral Hygiene Index (OHI-S),19 and the presence of calculus on those index teeth was also recorded. At age 26, three examiners were used, conducting 85%, 10% and 5% of the examinations respectively. Repeat examinations for examiner reliability analyses were not possible because of logistical constraints imposed by the cohort members’ tightly scheduled assessment day, of which the oral health assessment is a small part. However, examiners underwent training before the study began. Ethical approval for the study was obtained from the Lower South Regional Ethics Committee, and written informed consent was obtained from each participant.

The 1982 Pelotas birth cohort study

Pelotas is a city in the extreme south of Brazil (close to the Uruguayan border) with an urban population which has increased from 214,000 inhabitants in 1982 to 340,000 in 2007. The city's main economic activities are agriculture, livestock farming and business, and it is an important university centre.

Pelotas has had a community water fluoridation scheme since 1961.20 Primary dental care for the population is offered from 32 Health Centers linked to the Brazilian Unified Health System; 22 of them are in the urban area. In addition, eight dental offices in public schools provide primary dental care for young students. Data from the Brazilian Dental Council show that 514 dentists were registered in the city in 2007 (http://www.cfo.org.br).

In 1982, a birth cohort study was started in Pelotas. All 5,914 live-born infants born in hospital (over 99% of all deliveries) in 1982 were enrolled. The first interview with the mothers was carried out at the hospital, shortly after birth. Follow-up visits were carried out in 1983, 1984, 1986, 1995, 1997, 2000, 2001 and 2005. Initially, the study focus was on perinatal, infant, and early childhood morbidity and mortality, with the main lines of enquiry related to breastfeeding patterns and nutritional status, as well as social and environmental factors. By mid-childhood, the study had shifted in emphasis to child care, utilisation of health services, selected morbidity indicators, and child development. In adolescence, issues related to sexual and reproductive behaviours (including teenage pregnancies), habits (such as smoking and alcohol use), mental health, and education became the focus of the investigation. In more recent assessments, with cohort members being young adults, the main emphasis has shifted to chronic disease risk factors (including smoking, diet, physical exercise, and being overweight), reproductive history, and mental health. Further information on the cohort is available elsewhere.21,22

In 1997, when the cohort members were 15 years old, 70 census tracts from Pelotas (27% of the total) were systematically selected and every household in these tracts was visited. A total of 1076 cohort members were interviewed, corresponding to a follow-up rate of 72% of those expected to be found in those areas. Of these, 900 cohort members were randomly selected and 888 were examined and interviewed at home for the 1997 Oral Health Study (OHS).

All adolescents participating in the 1997 OHS were visited in their homes in 2006 at age 24 years, when dental examinations and face-to-face interviews were conducted. We managed to examine and interview 720 cohort members. Dental caries was diagnosed using the 1997 WHO criteria.18 A full-mouth periodontal examination was performed, with probing depth measurements made at six sites (mesiobuccal, midbuccal, distobuccal, distolingual, midlingual and mesiolingual) per tooth using a ball-ended periodontal probe (Trinity®). Periodontal probing depth was measured as the distance (in mm) from the free gingival margin to the base of the gingival sulcus. Gingival bleeding and calculus were also recorded (as ‘present’ or ‘absent’). Also investigated were: the use of (and need for) any dental prosthesis (WHO, 1997); the type, size (in mm) and site of any oral mucosal lesions; and the quality of any dental restorations in posterior teeth.23 A headlamp was used to illuminate the mouth during the examination. Dental services utilisation and dental self-care were assessed using a structured questionnaire, together with any episodes of dental pain in the previous four weeks. The fieldwork team comprised six dentists and four advanced dental students, who each examined and interviewed similar numbers of participants. All observers were previously trained and calibrated following methodology described elsewhere.24 A pilot study was carried out before the fieldwork commenced. Examiner reliability was assessed by means of weighted kappa for categorical variables and intra-class correlation coefficients for continuous variables. The lowest value was 0.6 for gingival bleeding, while the majority of values were 1.0. Data quality control was undertaken by the fieldwork supervisor, who re-contacted 10% of the sample by telephone and applied a short version of the original questionnaire. A loss to follow-up was recorded after four unsuccessful home visits, including at least one at the weekend and one during the evening. Cohort members who had moved to cities up to 500 km from Pelotas were contacted by telephone and then visited in their homes. The fieldwork was carried out in twelve weeks between February and April 2006.

Consent for interviews and examinations was obtained and both projects were approved by the Pelotas Federal University Ethics Committee. All individuals who presented with dental treatment needs or with soft-tissue lesions needing additional diagnostic procedures were referred to the Dental School of Pelotas Federal University.

Results

Information on the methodological characteristics of the two cohort studies and their respective mid-20s assessment phases is presented in Table 1.

Table 1.  Comparison of main characteristics and methodological aspects of the Dunedin and Pelotas birth cohorts.
 Dunedin StudyPelotas Study
  1. Note:

  2. a. This is the proportion of non-Maori.

  3. b. According to the Brazilian census categories: Whites, Dark-skinned black people (Pretos), Lighter-skinned black people (Pardos), yellow skin colour (Asian descent) and Amerindian descent.

  4. c. This is the proportion with a university degree or diploma.

Birth period for cohort1 April 1972 to 30 March 19731 January 1982 to 31 December 1982
Number in dental sample at baseline1,037900
Details of mid-20s assessments
Year(s) of data collection1998–992006
Age of participants2624
% female49.147.4
% white92.2a71.2b
% with maternal schooling 12+ years10.8c12.3
Number completing questionnaire973 (93.8%)720 (80.0%)
Number dentally examined930 (89.7%)720 (80.0%)
Number periodontally examined918 (88.5%)720 (80.0%)
Number of dental examiners used310
Type of dental examinersDentistsDentists (6) and dental students (4)
Location of dental examinationClinicHome
Time available for each dental examination12 minutes per participant12 minutes per participant
Method of periodontal examinationHalf-mouth; 3 sites/toothFull-mouth; 6 sites/tooth
Periodontal information recordedGingival recession, probing depth, bleeding on probingProbing depth, bleeding on probing
Method of obtaining self-report informationSelf-completed questionnaireFace-to-face interview

The Dunedin cohort was 26 years old at the time of the assessments conducted in 1998 and 1999, while the Pelotas cohort was 24 years old at the time of the 2006 assessments. The two cohorts showed socio-demographic similarities and differences, with their gender mixes being similar, but their ethnic compositions differing markedly. There were some important differences in methods, with oral examinations undertaken in participants’ homes in the Pelotas study, but in the dental clinic setting in the Dunedin study; the former used, besides dentists, senior dental students as examiners and interviewers (face-to-face), while the latter used only experienced dentists (and self-completed questionnaires). Periodontal examinations in the Pelotas study were undertaken using full-mouth measurement and recording (at six sites per tooth), while those in the Dunedin study used diagonal half-mouth measurement and recorded three sites per tooth (full-mouth recording was used subsequently, at the age-32 assessment).

Summary information on the main oral health characteristics of the two cohorts in their mid-20s is presented in Table 2.

Table 2.  Comparison of main oral health characteristics of the Dunedin and Pelotas birth cohorts in their mid-20s (brackets contain 95% confidence interval).
 Dunedin StudyPelotas Study
  1. Note:

  2. a. The number of sound teeth and the DMFT do not add up to 32 because of teeth being absent for reasons such as non-eruption, congenital absence, orthodontic extractions or trauma.

  3. b. This estimate was obtained by using the same half-mouth approach as that used in the Dunedin Study; the full-mouth estimate was 3.3 (2.0–4.6).

Dental caries
Number examined931720
Mean DMFT7.7 (7.4–8.0)5.6 (5.3–5.9)
Mean DT2.1 (1.9–2.3)1.5 (1.3–1.7)
Mean FT5.4 (5.1–5.7)3.3 (3.1–3.5)
Mean MT0.2 (0.2–0.2)0.8 (0.7–0.9)
Mean number of sound teetha20.6 (20.3–21.0)24.9 (24.6–25.2)
% with 1+ teeth missing due to caries9.0 (7.2–10.8)41.8 (38.2–45.4)
Periodontal conditions
Number examined918720
% with 1+ sites with 4+mm PD15.3 (13.0–17.6)2.0 (0.6–3.4)b
% with 1+ sites with 6+mm PD2.8 (1.7–3.9)0.0 (—)
Mean extent of bleeding on probing29.7 (28.1–31.4)37.5 (34.0–41.0)
% with ≥1 sites with calculus49.0 (45.8–52.2)87.4 (84.9–90.0)
% current smokers39.5 (36.3–42.7)26.6 (22.8–30.4)
Other characteristics
% reporting dental painNot asked23.0 (20.0–26.1)
% visiting a dentist in the previous year51.2 (48.0–54.4)55.3 (51.6–59.0)
% who usually visit for a check-up46.4 (43.3–49.5)22.0 (19.0–25.0)
% reporting toothbrushing 2+ times per day57.0 (53.9–60.1)94.3 (92.7–96.1)
% reporting oral health ‘better than average’41.9 (38.7–45.1)Not asked

Overall mean DMFT scores (and the DT and FT component scores)* were higher among the Dunedin cohort. There were similarities and differences in the relative contributions of the various components to the overall DMFT score, with mean DT contributing 27% of the overall score in each cohort, while the FT component contributed 70% of the score for the Dunedin cohort but 59% of that for the Pelotas cohort. The contribution of the MT component was 14% for the latter, and almost 3% for the former, and there was a large difference between the two in the prevalence of 1+ teeth missing due to dental caries. The periodontal differences were more marked, with the prevalence of attachment loss being greater in the Dunedin cohort, but the extent of bleeding on probing and the prevalence of calculus higher in the Pelotas cohort. Current smoking was observed in approximately four out of ten Dunedin and just over one in four Pelotas study participants. There was no difference in the proportion reporting a dental visit in the previous year, with just over half of each cohort reporting having done so. The proportion who usually visited for a check-up (instead of a problem) was higher among the Dunedin cohort. The prevalence of twice-daily (or better) tooth brushing was significantly higher among the Pelotas cohort.

Similarities in findings: the example of socio-economic trajectories and oral health

Each of the studies has examined the association between childhood-adulthood changes in socio-economic status and oral health in the mid-20s, with the Dunedin study reporting on dental caries, tooth loss and periodontal disease experience by age 26, and the Pelotas study reporting on the number of unsound teeth (associated with caries experience) by age 24.15,16 Having separate measures of early childhood (‘origin’) and adult (‘destination’) SES allowed each study to allocate its cohort members to one of four socio-economic trajectories. In the Dunedin study, this was done using an occupational measure of SES; in the Pelotas study, SES was determined using household income. In each study, those who were in the high-SES group in both childhood and adulthood were categorised as the ‘high-high’ group; those in the low-SES group in both childhood and adulthood were designated the ‘low-low’ group; the remainder were allocated to either the ‘high-low’ or the ‘low-high’ group comprised those who (respectively) were downwardly and upwardly mobile. Each study found that the greatest disease experience was observed among those who were in the low-low group, and the least was seen in the high-high group (Table 3). In each cohort, disease experience in the upwardly mobile and downwardly mobile groups lay between those two extremes.

Table 3.  Comparison of the two studies’ oral health gradientsa by SES trajectories.
 Pelotas StudyDunedin Study
 Proportion of cohortMedian number of sound teethProportion of cohortPer cent missing 1+ teeth due to cariesMean DS
  1. Note:

  2. a. For the Pelotas study, the dependent variable is a measure of the absence of caries, while those for the Dunedin study represent the adverse effects of the condition.

  3. b. From childhood to adulthood.

SES trajectoryb
Always high-SES46.1%2720.5%3.01.3
Upwardly mobile13.1%2621.4%5.51.6
Downwardly mobile18.2%2619.1%6.41.9
Always low-SES22.6%2538.9%12.52.1

Discussion

This paper is the first comparison of the main methodological characteristics and findings from what are, to our knowledge, the only two population-based birth cohort studies of oral health among young adults which also have child and adolescent data points. In general, we found differences (some minor, some major) in methodological aspects such as the information collected, the dental examination protocols, and the number and experience of examiners. Thus, comparisons of the oral health findings need to be made carefully. In the following paragraphs, we account for differences (and similarities) in the main findings of the two studies, partly in order to provide guidance for those planning similar studies in the future. We examine first the findings on oral self-care and the use of dental services, followed by those on the oral health of the two cohorts (because discussion of the oral health findings is informed by the former). Finally, we discuss the findings relating to socio-economic mobility and oral health.

The use of dental services in the previous year was similar in both birth cohorts, with just over half reporting having visited a dentist. In Pelotas, as in most parts of Brazil, the majority of dental care is offered by private dentists. People from low socioeconomic status groups cannot afford dental treatment and, consequently, have had mainly extractive care provided by the Brazilian Unified Health System (SUS) until 2002 (there have recently been improvements in access to dental rehabilitation, with endodontic treatment and dental prostheses now provided by SUS and this pattern is expected to change in the future). Paradoxically, the dentist-to-population ratio is much higher in Pelotas (1 dentist for every 661 people) than in Dunedin (1:1984), suggesting wide inequality in the pattern of dental attendance in Brazil. The FT component of the DMFT made a greater contribution to the overall DMFT score in the Dunedin cohort, and this may be explained by the two cohorts differing in the reason for seeking dental care, with a considerably greater proportion of the Pelotas cohort than the Dunedin cohort seeking dental treatment episodically (that is, in response to a dental problem). These differences are similar to those observed in both countries’ populations at large, with fewer than one-third of Pelotas adults having attended for a check-up in the previous year,25 while the equivalent New Zealand estimate was 51%.26

The difference in accumulated dental caries experience (represented by mean DMFT scores) between the Dunedin and Pelotas cohorts is most likely due to a mix of cohort and period effects.27 The age-26 dental examinations for the Dunedin cohort were conducted in 1998–99; this means that, at the time that the Pelotas age-24 dental examinations were conducted (2006), the Dunedin cohort members were almost a decade older. A secular trend of reducing dental caries experience in children, adolescents and young adults has been identified in developed countries in recent decades.28 Such a decline has also been observed in Brazil, and it is considered to be attributable to the expansion of community water fluoridation coverage (available in Brazil since the 1950s, and since the 1960s in Pelotas) and to a marked increase in the market dominance of fluoride dentifrices since 1989. Before September 1988, about 25% of dentifrices sold in Brazil contained fluoride; this had increased to 90% in 1990 and is currently 95%. Brazil's daily consumption of fluoride dentifrice (which ranged from 0.6 g per capita in 1982 to the current 1.4 g per capita) ranks it as the third highest among countries reporting dentifrice consumption.29 On the other hand, it is more difficult to account for the finding that caries-associated tooth loss was considerably greater among the Pelotas cohort than among the Dunedin cohort, although the abovementioned differences in dental utilisation are likely to have played an important part.

While there were differences between the two cohorts in the prevalence of periodontal pocketing, it should be remembered that caution should be exercised when comparing periodontal data from two or more populations at different times or resulting from different sampling or measurement procedures.30 Both of these apply in considering the periodontal estimates. First, the Dunedin cohort was examined about 10 years before the Pelotas sample, and it may be that secular changes in periodontitis occurrence have taken place (much as with dental caries) which contribute somewhat to the observed differences, given that there is a limited amount of evidence for a minor reduction in prevalence over time.31 The differences may be partly methodological. In the Dunedin study, the dental examinations were performed in the clinical setting, while the Pelotas examinations were carried out in each participant's home. There is little available information on the relative efficacy of clinic-based and home-based examinations, although a recent investigation of this issue using older people showed that, in general, clinic-based examinations detected more disease.32 It may be that the differences between the two approaches may be more apparent among younger than older adults, since the number of sites to be examined in the former is much higher than in the latter; on the other hand, however, there is less disease to detect. Leaving aside the issue of the possible extent of any under-estimation in the Dunedin cohort and the home examinations approach used in Pelotas, the fact remains that the prevalence of probing depth greater than 3 mm was considerably higher in the Dunedin group than in the Pelotas cohort.

This begs the question of what else could explain the difference. The most well-known behavioural risk factor for periodontitis is the smoking of tobacco,13,33,34 and this was certainly more prevalent among the Dunedin cohort. The population attributable risk for periodontitis from tobacco smoking in the Dunedin cohort was recently estimated to be between 34% and 67%, depending on which case definition was used for incident attachment loss (albeit for the incidence of new attachment loss between ages 26 and 32).13 The Dunedin cohort's greater exposure to tobacco may indeed account for a considerable amount of the observed difference, but it may not have contributed to all of it. In order to test this, we computed the unadjusted odds ratio for the prevalence of one or more sites with 4+ mm probing depth: for the Pelotas cohort, it was 2.6 (95% CI 1.1–6.0); for the Dunedin cohort, it was 1.7 (95% CI 1.2–2.5). That these were not significantly different suggests that perhaps the latter's greater smoking was indeed the main factor involved.

Gingival bleeding and dental calculus were both more prevalent in the Pelotas cohort. There are two possible explanations for these differences. First, ineffective tooth brushing is the main contributor to gingival bleeding and dental calculus which, in turn, are more common among less-educated groups.35 Second, the Dunedin cohort's access to a free school-based dental care program during childhood and adolescence36 may have instilled positive dental self-care behaviours which have persisted into adulthood to the extent that they are more common than in the Pelotas cohort. The findings also suggest the weakness of self-reported tooth brushing as a measure in the Pelotas study, and a need for an additional clinical measure of plaque accumulation in order to supplement the self-reported self-care data, as has been done in the Dunedin study, which has used a modification of the Simplified Oral Hygiene Index19 at several assessment phases.

There were considerable difficulties in comparing the socio-demographic characteristics of the two cohorts. The comparison of ethnic characteristics is a case in point. Self-reported ethnic identification is used in New Zealand,37 while skin colour and race are used in Brazil. Investigations of ethnic-racial inequalities present several methodological problems because the attribution of colour/race is complex and controversial, and it involves a number of factors. In Brazil, racial identification is based on a combination of physical characteristics, such as skin colour, nose and lip shape and hair type; the attribution of colour/race is based on external observation or self-classification based upon those characteristics. In accordance with international practice, the Brazilian Institute for Geography and Statistics (Instituto Brasileiro de Geografia e Estatística, IBGE) trains its interviewers for the ten-year Census to record race as declared by the interviewers, rather than the interviewees.38 However, racial measurement needs to be in accordance with the objectives of each study in particular. The way in which data are collected in a skin-colour-based racial classification may relate to the type of exposure to health risks that is to be studied.39 It has been argued that colour/race should be attributed by the interviewer when estimating the social exposure to health risks, and self-classified when estimating the exposure to risky behaviours which is attributable to cultural factors.40 Thus, in the 1982 Pelotas cohort study, both self-classified and interviewer-attributed categorisations of colour/race were collected in order to test hypotheses about racial differences in health outcomes.

Information on the validity and reliability of measurements (whether clinical or self-report) needs to be reported whenever possible, in order to allow judgments to be made about the quality of the data and the internal validity of the study. However, this is not always possible in multidisciplinary cohort studies, where logistical and ethical reasons preclude this. In the Dunedin study, for example, the very busy assessment protocol undergone by cohort members means that there is no time for repeat examinations (and, indeed, the conditions of the ethical approval for the study specifically preclude this). Thus, any examinations of intra- and inter-examiner reliability are conducted on a separate sample of volunteers of similar age. In any case, requiring cohort members to undergo repeat examinations would inevitably affect the study's follow-up rate, a core issue in prospective studies. Both the Pelotas and Dunedin studies therefore undertake strict training and calibration exercises parallel to the cohort data collection period with a different population. In Pelotas, for nonclinical data, a short version of the main questionnaire was used through a telephone interview with 10% of the sample, allowing examination of the reliability of variables such as schooling, self-reported skin colour, time since the last dental visit, and smoking. To date, the Dunedin study's sociodental questionnaire has not undergone the same scrutiny.

The two studies’ findings on socio-economic trajectories and oral health were largely consistent, with the poorest oral health observed among those who had followed a consistently low socio-economic trajectory; by contrast, those who followed a consistently high socio-economic trajectory were dentally advantaged. This replication of findings (despite differences in the way in which SES was operationalised in each study) suggests that the means by which SES affects oral health are broadly similar in different cultures and populations. Moreover, the two studies’ findings support the concept, first applied to general chronic diseases (such as cardiovascular disease, diabetes and chronic bronchitis), that socioeconomic circumstances throughout the life course affect adult health and disease risk,41 and that preventing chronic conditions (whether oral or systemic) requires a focus on the longer term and on the life course. Quick and easy solutions are unlikely; a common risk factor approach is likely to be more effective than a disease-oriented one.

Challenges for the future

In order to enhance future comparisons of findings from these cohort studies, we intend to work towards greater convergence in the assessment procedures. The challenge in doing so will be not to compromise intra-study comparability with earlier assessments. For example, future recording of periodontal attachment loss in the Pelotas study will include the collection of data on gingival recession while, at the same time, continued collection of probing depth data will enable determination of periodontal changes since age 24. The former will enhance comparisons with the Dunedin study's periodontal estimates. Similarly, the Dunedin study will collect data on the experience of dental pain in the previous four weeks, in order to facilitate comparisons with the Pelotas estimates. These are illustrative examples only, but they indicate the careful planning that will be needed to promote the replicability of findings in the two studies (and therefore their wider value).

Similarly, others who are planning a dental cohort study are advised to consider the issues of replicability and comparability very early in their protocol design. Scientists who have been in similar situations in the past and have learned from experience may have useful advice to offer.

In conclusion, oral health cohort studies are rare but make valuable contributions to understanding of the antecedents and natural history of oral health and ill-health. Systematic comparison of their methods and findings is an important step in the accumulation and assessment of that knowledge. The similarities in findings from cohort studies in two quite different countries suggest commonalities in the mechanisms which determine oral health through the life course. Whether these are universal or specific to those two countries requires further investigation.

Footnotes

  • *

    DMFT represents Decayed, Missing or Filled Teeth; DT represents Decayed Teeth; FT represents Filled Teeth

Acknowledgements

The oral health component of the Dunedin Multidisciplinary Health and Development Study is supported by Grant R01 DE-015260 from the National Institute of Dental and Craniofacial Research, National Institutes of Health, Bethesda, Maryland, 20892, USA, and by the Health Research Council of New Zealand. The Dunedin Multidisciplinary Health and Development Research Unit is supported by the Health Research Council of New Zealand. The Study members, their families and their friends are thanked for their continuing support.

The 1982 Pelotas birth cohort was supported by the Wellcome Trusts initiative entitled Major Awards for Latin America on Health Consequences of Population Change. Earlier phases of the 1982 cohort study were funded by the International Development Research Center (Canada), the World Health Organization (Department of Child and Adolescent Health and Development, and Human Reproduction Programme), the Overseas Development Administration (United Kingdom), the United Nations Development Fund for Women, the National Program for Centres of Excellence (Brazil), the Brazilian National Council for Scientific and Technological Development (CNPq – Brazil) and the Ministry of Health (Brazil). Dr. Marco Aurélio Peres was supported by an Overseas Senior Research grant (number 201291/2008–8) from the Brazilian National Council for Scientific and Technological Development (CNPq – Brazil). Dr. Karen G. Peres received grants from the Brazilian National Council for Scientific and Technological Development (CNPq – Brazil) to develop the late phase of the oral health study (number 47698520045) and was supported by a productivity in research grant from the CNPq (Conselho Nacional de Desenvolvimento Científico e Tecnológico) and CAPES (Brazilian Coordination for Coordination for the Improvement of Higher Education Personnel), number 2580/08–0.

Ancillary