Page 11 - SPEMD_59-4

rev port estomatol med dent cir maxilofac . 2018;59(4):181-190 183

the Ethical Board of the Portuguese Northern Regional Health or SROH. The quality of logistic regressions was evaluated us-
Administration. The permission to examine the patients’ ing the area under the curve (AUC), and indicated the adjust-
mouth and record the data was obtained from authorities. ment of the model of SRH and SROH to clinical variables.
Adult volunteers attending the FHC were randomly invited
to participate by telephone. For those who accepted, an ap-
pointment including oral examination was performed to eval- Results
uate if they met the inclusion criteria. Patients with DM2 were
invited for the disease group of the study (DM2 group) and Most participants were female (56.9%), and the mean age was
healthy volunteers with no DM2 diagnosis for the control 63.9±12.8 years old. A significantly (p=0.003) higher percent-
34
group (nDM2 group). The pairing mode between cases and age of women was in the nDM2 group (62.8%). The BMI higher
controls was partially performed based on gender and age. classes – pre-obesity (47.5%) and obesity (37%), were signifi-
A sample size of at least 656 individuals (328 in each group) cantly more frequent (p<0.001) in the DM2 group than in the
was calculated for an expected prevalence of dental caries of nDM2. Significantly fewer smokers (7%) and a higher preva-
65%±5%, an expected difference of 5%, a 95% CI and a power lence of clinical comorbidities were found in the DM2 group.
of 90%, based on the III National Study’s results on the preva- A significantly greater part of the DM2 group (p<0.001) were
lence of dental caries in adults. 35 standard 34,37 controlled DM2 individuals (70%) (Table 1).
Data on haemoglobin A1c (HbA1c), body mass index (BMI), No differences were found in the reported daily oral hygiene
duration of illness and presence of complications were ob- between the nDM2 and DM2 groups. The DMFT index was sig-
tained by consulting the individual electronic or paper clinical nificantly higher (p=0.005) in the nDM2 (17.7±8.3) than in the DM2
process. The clinical analytic data considered was the last one group (15.9±7.8), while dental caries experience was not signifi-
available in the 12 months prior to the interview. cantly different. The DM2 group had a higher prevalence of peri-
The oral-health status was recorded following the WHO odontal disease (99.3% vs. 95.2% in the nDM2 group, p=0.006),
36
oral health criteria through a questionnaire and clinical eval- with a higher severity status. The prevalence of periodontal pock-
uation. The questionnaire was administered face-to-face to ets >3mm was 46% for the DM2 and 37.9% for the nDM2 group.
the volunteers, by asking them to rate their own GH and OH Gingival recession prevalence was 95.7% for the DM2 vs. 87.7%
on a Likert 5-point scale (1 – very bad, 2 – bad, 3 – median, 4 – for the nDM2 group (p<0.001). Total edentulism showed a signifi-
good and 5 – very good). cantly (p<0.001) higher prevalence in DM2 patients, being 1.86
The oral examination was conducted by a single experi- times more prevalent (15.7%) than in the nDM2 group (Table 1).
enced examiner, who was trained and calibrated (Cohen-Kap- Regarding the SRH and SROH results, described in Table 1,
pa of 0.80 was the minimal value). This examination was done the DM2 group classified SRH significantly (p<0.001) worse than
under natural daylight and dental auxiliary light (OSRAM the nDM2, with “bad” or “very bad” perceptions. The nDM2 par-
DECOSTAR), using a plain mirror #4 and a periodontal PDT ticipants (Table 2) who rated SRH as “good” or “very good”
sensor probe. The oral-health status was recorded based on (59.5%) also did it for SROH. Similarly, 53.1% of the DM2 patients
the decayed, missing and filled teeth (DMFT) index and respec- classified both SRH and SROH as “very bad” or “bad.” These
tive components (decayed teeth (DT), missing teeth (MT) and results show a significant agreement in these classifications.
36
filled teeth (FT)), the Community Periodontal Index (CPI); and Table 3 reports the association between median and inter-
the use of a prosthesis, if applicable. quartile range and the socio-demographic and clinical vari-
®
©
Statistical analysis was performed using the IBM SPSS ables. The higher the median value, the higher the perception
©
Statistics v.25 (IBM Corporation). The mean values of the of positive impacts on GH and OH for the DM2 and nDM2
nDM2 and DM2 groups were compared using the Student’s groups. Although this was a rare condition, almost all DM2 and
independent t-test, while categorical variables were compared nDM2 participants with no caries experience (DMFT=0) clas-
using mostly the chi-square test. The binomial test was used sified their OH status as “good” or “very good,” with only 1% of
to estimate the distribution of the controlled HbA1c value. The these DM2 patients classifying their OH as “median.”
relationship of the perceived status of both GH and OH be- MLR identified variables significantly and independently
tween groups was assessed using chi-square tests. The Bon- associated with “bad” or “very bad” SRH or SROH in both groups
ferroni correction was used for proportion comparisons of (Table 4). In the nDM2 group, “bad” or “very bad” SRH was not
more than two categories. The dependent variables of the per- associated with any GH clinical variable, while the CPI was
ceived status of GH and OH were compared between groups found to be a risk factor for that outcome and having calculus
based on the median scores of the categories of relevant co- was a significant risk factor of “bad” health (OR=3.21, 95%CL:1.03-
variates, using the nonparametric Mann-Whitney’s test or the 9.99; p=0.044). In the DM2 group, the CPI was also found to be a
Kruskal-Wallis test. Whenever the Kruskal-Wallis test showed risk factor for”bad” or “very bad” SRH (p=0.015), and arterial hy-
significant differences, the multiple comparisons were per- pertension (HTA) and dyslipidaemia were other significant risk
formed considering the Bonferroni correction. The significance factors (OR =16.62, p<0.001 and OR =5.17, p<0.001).
HTA Dysl
level was set at 0.05 for all inferences. Regarding the “bad” or “very bad” SROH, the CPI was iden-
Multivariable binary logistic regression (MLR) models (Wald tified as a risk factor in the nDM2 group, together with cal-
backward stepwise method, p=0.05 for covariate inclusion and culus and pockets >4mm (OR=3.55, p=0.049, and OR=4.32,
p=0.10 for exclusion) were used to predict associations (as risk p=0.025, respectively). On the other hand, having at least 20
or protective factors) between covariables identified in previ- teeth was found to be a protective factor (OR=0.22, p=0.001),
ous analyses and participants having “bad” or “very bad” SRH decreasing the chance of that outcome in 78%. In the DM2