The study included 610 randomly selected, 25-40-year old subjects. All subjects presented as new patients seeking routine dental care (not emergency care) at the University of Toronto Faculty of Dentistry (n = 400) and at the University of Saskatchewan Dental School in Saskatoon (n = 210) between the years 1998 and 2000. These two sites were chosen because they were considered to reflect contrasting conditions - endodontists are abundant in Toronto, whilst the yare scarce in Saskatoon. Randomization was carried out using a random numbers table. To be enrolled in the study, the patient’s chart had to contain a current full-mouth radiographic series or panoramic radiograph. Where neither was present, the patient was excluded and the next one (in the alphabetical list of all patients) selected. Where only a panoramic radiograph was present, supplemental periapical radiographs were taken of all teeth that had root fillings. Because of the nature of the radiology clinics in both institutions, the intraoral films were exposed with a variety of X-ray machines and processed in a variety of automatic processors. However, all periapical radiographs were taken using the paralleling technique.
All radiographs were viewed in a ‘blind’ manner by one investigator. The method of viewing the radiographs was standardized - they were examined in a darkened room using an illuminated viewer box with magnification (3.5x) whilst mounted in a cardboard slit to block off ambient light emanating from the viewer. To avoid the need for interexaminer calibration, the same investigator collected the data at both study sites.
The following information was recorded on a structured form for each subject:

1. number of teeth present and the location of missing teeth;
2. number and location of teeth without root fillings (untreated teeth) having identifiable apical periodontitis lesions (as observed in panoramic or periapical radiographs) and
3. number and location of root-filled teeth.

In addition, the ‘Periapical Index’ (PAI) score (Krstavik et al.1986) was recorded for each one of the root-filled roots. The worst score of all roots was taken to represent the PAI score for multirooted teeth. The scores were then dichotomized into periapically ‘healthy’ (PAI 1 and 2) or ‘diseased’ (PAI 3- 5).The latter was considered to be consistent with apical periodontitis. The principal investigator was calibrated for the use of the PAI index. The kit of 100 reference radiographs and ‘gold standard’ observations (Krstavik et al. 1986) was used, dichotomizing the ‘gold standard’ observations as described above, and the interobserver Kappa score calculated. Observations were recorded at two different times and compared, in order to calculate the intraobserver Kappa score.
The quality of the root filling, with respect to density and length, was assessed and recorded for each root. Density was classified as follows:

1. adequate - uniform radiodensity and adaptation of the filling to the root canal walls, or
2. in adequate - visible canal space laterally along the filling or voids within the filling mass, or identifiable untreated canal.

Length was classified as follows:

1. short - ending more than 2 mm from the radiographic apex;
2. adequate - ending 0-2 mm short of the apex;
3. long - extruding beyond the apex.

The quality of the restoration was also assessed radiographically and classified as inadequate if the restoration was absent or if open margins, overhangs or secondary caries were detectable. In addition, the presence or absence of posts was noted.
All subjects identified to have at least one root-filled tooth were contacted by telephone and invited for an interview. Because the majority of the subjects had been enrolled as active patients at both institutions since their initial screening, these patients were interviewed when they arrived to undergo treatment in the university clinics. In order to increase the response rate, letters were sent to those patients who were not contacted by telephone. When contacted, the subjects were explained about the study, invited to enroll and offered a reimbursement for lost work time and travel expenses. The ethics committees of both institutions approved the study, and all enrolled subjects signed a written informed consent.
During the interview, the subjects answered a structured questionnaire that recorded reasons for extraction of missing teeth (if applicable), approximate date endodontic treatment was rendered and whether this was provided by a generalist or an endodontist. Also included were additional general questions intended to define the socio-demographic characteristics of the studied populations. In addition to the questionnaire, all root- filled teeth were clinically examined, with particular attention to detecting possible cracks or fractures of the coronal tooth structure, open margins, overhangs or secondary caries. The type of the restoration (temporary or permanent, intracoronal or extracoronal) was also noted.
Raw data were entered in Excel databases and statistical tests were carried out using the spss statistical package. Where appropriate, data were pooled for each subject and then for each of the two study populations. Chi-square and independent t-tests were used to analyse differences between and within groups. The distribution of each continuous outcome variable was checked for normality; for nonparametric distributions, Mann- Whitney U-test was used to confirm significance.
The prevalence of apical periodontitis was compared between, and within the two studied populations, between teeth treated by generalists and those treated by specialists and between teeth treated by generalists in Toronto and those treated by generalists in Saskatoon. Further comparisons were carried out between the different categories of root filling and restoration, and combinations thereof. All tests were two-tailed performed at the 5% significance level.