Ethical approval for the conventional study was received by the University of Adelaide's Human Research Ethics Committee. Ethics approval for wave III of the ABC study was received by the Ethics Committee of the Northern Territory Department of Health and Community Services and Menzies School of Health Research, which includes an Aboriginal sub-committee with absolute right of veto.
Location, privacy and equipment
Dental examinations in the conventional investigation took place in dental clinics, using customised dental chairs with standard dental lights attached (Table ). Privacy was afforded through dental surgery walls. The examinations were conducted using non-disposable probes and a mirror light kit with disposable mirror heads and disposable plastic sleeves. By comparison, dental examinations in the ABC (non-conventional) study occurred in a range of opportunistic settings including hospital rooms, health centres, community recreation halls, community sheds, school classrooms, women's centres and outdoor spaces. Privacy was ensured through creative use of twine and curtains, and any additional props that may have been present (for example, up-ended tables, racks of second-hand clothing). The dental chair was a portable reclining camp chair (total weight 5 kg) and the dental light a fire-fighter's head torch with rechargeable batteries. The probes used for examination were disposable, as were the mirror heads and plastic sleeves used with the mirror kit (the same as those used in NSAOH). All items were portable and disposable, with a total weight of 8 kg. Logistic constraints to routine sterilisation procedures in the field made it difficult to use non-disposable probes in the ABC study.
Comparison of conventional (NSAOH) with non-conventional (ABC study) dental data collection techniques
Recruitment, number of contacts made and transport
A three-staged, stratified clustered sampling design was implemented in NSAOH to select Australian residents aged 15+ years. The sampling frames were households with listed telephone numbers recorded in an electronic white pages database. The first stage selected postcodes, the second stage selected households within sampled postcodes, and the third stage selected one person aged 15+ years from each sampled household. Approximately 10 days before dialling each sampled telephone number, a primary approach letter explaining NSAOH's purpose was mailed to the address accompanying the sampled telephone numbers. Up to six contact calls were made. At the completion of each telephone interview, participants were requested to take part in a clinical oral epidemiological examination, with appointments made at public dental clinics within or near the postcode in which people were sampled. Participants were expected to make their own way to the clinic. A second appointment was made if participants failed to present at the first appointment.
In wave III of the ABC study, a letter of introduction was sent to each Aboriginal community council explaining the purpose of the study, requesting assistance in locating a suitable space in which to work, and enquiring as to the availability of a local person willing to be employed in one or all roles of locator, translator and advocate. A permit to visit the community was applied for and lists of potential participants were shown to community councils and/or health centres. In a given community, strategies employed to locate participants; primarily through the locator, included canvassing high schools, homes, work places and recreation areas. Participants would usually be collected and dropped off by study members in a vehicle, and were encouraged to bring along any children or supportive family members. In the urban settings, participants were canvassed in a similar manner, although there was a greater reliance on telephone calls and house calls. Snowball techniques were also employed. A considerable number of attempts were made to contact participants, only ceasing when the study team left the community or were given an outright refusal from a study member.
Participants in NSAOH provided verbal consent prior to answering questions in the telephone interview and signed informed consent prior to the oral examination. Signed, informed consent was received for the individual components of the ABC study, including permission to take a photograph for future identification purposes. The male Aboriginal research assistant involved in the study went through a visual enhanced staged consent for male participants, while female study team members and/or female locally employed assistants went through the consent process with female participants.
The dental examinations in both the NSAOH and ABC study recorded presence or absence of oral mucosal lesions; tooth loss; gingivitis, calculus and plaque levels; periodontal status; caries experience; attrition and dental fluorosis (Table ). The collection of buccal mucosal cells (for DNA purposes) and gingival crevicular fluid was implemented only in NSAOH, while dental trauma was assessed only in the ABC study.
Oral epidemiological measures used in NSAOH and the ABC study
A dental assistant was present for all NSAOH examinations to record dental data directly into a screen-based entry system on a laptop computer, in comparison with the ABC study where three dental data recording methods were trialled. When compared with the gold standard of a human voice recorder, the digital recorder scored highly for portability and cost-effectiveness and moderately for general and technical ease-of-use, reliability and time-efficiency (Table ). The computer-based voice recorder with Microsoft Word© facilities scored poorly in general and technical ease-of-use and portability, and very poorly in reliability and time-efficiency. The computer-based voice-recording software using Microsoft Excel© was worst, with low points scored in general and technical ease-of-use, reliability and time-efficiency. Of note the human voice recording method scored poorly for portability and cost-effectiveness, for example, an extra seat on charter planes being necessary thus substantially increasing the cost. The main issue with the digital recorder was time efficiency, with time required to play back the recording and to transcribe the information onto the paper-based data capture forms. However, this was even more so for the computer-based voice recording software, where the operator needed to consistently check the screen for wording, to repeat phrases and to, in turn, transcribe the data onto the paper-based forms. Both computer-based systems proved unreliable under time pressure, particularly when electricity was unavailable or there was not enough computer memory.
Comparison of dental data recording techniques in wave III of the ABC Study
In NSAOH, replicate examinations were conducted with 157 participants to assess reliability of 29 examiners and moderate agreement was reached for measures of decayed dental surfaces (median intra-class correlation = 0.56). As the dental component was only part of an extensive assessment taking over two hours, research fatigue considerations, time and logistical constraints, made it not possible to conduct replicate dental examinations in the ABC study. In the Dunedin Multidisciplinary Health and Development Study, it has also been reported that replicate oral examinations are not possible because of time constraints
], indicating that this is perhaps a common issue in longitudinal life course studies in which oral health data is not the only information gathered and co-operation with future follow-up is being fostered.
There were a number of data cleaning steps undertaken in the ABC study prior to entry into statistical software packages. These included cleaning when data was in its hardcopy form (on-site by a research assistant at the end of each data collection day) and once data had been entered into an electronic spreadsheet. Later double entry of 100 dental data forms was undertaken with a 5% error rate confined to one variable that was corrected in the final database subsequently used for all statistical analysis.