Student's t-test was used to compare the means of the outcome variables following normal distribution. Kolmogorov-Smirnov test was used to estimate normality of the distribution, and Levene's test the equality of variances. The distribution-free Mann-Whitney U-test was applied to compare the medians when the values were not normally distributed 25. The results are given as mean, standard error (SE), and the number of subjects in the group, unless otherwise indicated. The 95% confidence intervals and exact two-tailed p-values are given for each comparison.

In order to examine relationships between subjects' background variables and the number of periodontal pockets and PIBI, linear regression analysis was performed separately for the statin using patients and patients without statin.

Statistical calculations were performed using SPSS version 13.0 software (SPSS Inc., Chicago, Illinois) when appropriate.

The demographic data of study subjects are presented in Table 1. Of the 100 consecutive patients referred to the clinic for periodontal treatment, three did not fulfil the inclusion criteria of having at least 50% of the dentition remaining. Of the eligible subjects (n = 97), 44 were women. The mean age of all study subjects was 53 years (range 42–69 years, Table 1).

The percentage of smokers was greater in statin users (28.6%) than in patients not receiving statin (18.4%). Diabetes was more frequent in statin users (28.6% vs. 3.9%). The percentage of patients with rheumatic disease was similar in both groups.

The number of remaining teeth of statin-using patients was 24.3 (SE 0.7) when that of non-using patients was 26.1 (SE 0.4). All subjects had poor oral hygiene, evident as visible dental plaque and gingival bleeding on probing.

Twenty-one subjects were on statin medication: 11 used atorvastatin, 7 simvastatin, 2 fluvastatin, and 1 rosuvastatin. All statin-using patients were receiving a conventional low-dose statin therapy.

Numbers of periodontal pockets in subjects with and without statin medication are shown in Figure 1 (means/medians and 95% CIs). In patients not using statins, the number of at least 4 mm deep periodontal pockets was 79.1 (SE 3.7; n = 76) (Figure 1A). In patients using statins, the corresponding number was 50.1 (SE 7.0; n = 21), being 37% lower than non-users. The difference of means was 29.0 (95% CI 12.9–45.0; P = 0.00043) (Table 2).

Data were analysed separately for the number of moderate periodontal lesions and advanced periodontal lesions. In patients without statins, N_modPPD indicating moderate periodontal lesions was 50.7 (SE 2.5; n = 76). In statin users, N_modPPD was 36.0 (SE 5.0; n = 21), being 29% lower than in the non-using patients. The difference of means was 14.7 (95% CI 3.2–26.2; P = 0.0087) (Table 2).

Advanced periodontal inflammatory injury was measured as the number of periodontal pockets at least 6 mm deep. For patients without statin, N_advPPD was 28.4 (SE 2.6; n = 76). The corresponding number for statin using patients was 14.2 (SE 3.8; n = 21), being 50% lower than in patients not using statins. The difference of means was 14.2 (95% CI 4.9–23.5; P = 0.0034). The N_advPPD data indicating advanced periodontal lesions did not follow normal distribution. We therefore also used the distribution-free Mann-Whitney U-test to compare the population medians. The median was 20 for non-users and 11 for statin users. By the method of Campbell and Gardner 25, the estimated difference in population medians was 11 and the approximate 95% CI for the difference in population medians 4–20 (P = 0.0010) (Figure 1B, Table 2).

The Periodontal Inflammatory Burden Index (PIBI) combines and unifies data on PPD. In patients not taking statins PIBI was 107.6 (SE 5.8; n = 76) and in statin-using patients 64.3 (SE 10.0; n = 21) (Figure 1C). The difference of the means was 43.1 (95% CI 19.5–66.7; P = 0.00069). In statin users, PIBI was 40% smaller than in other periodontitis patients. For the non-smoking patients, PIBI was 64% lower for statin users when compared to patients without statin (48.9 vs. 105.2).

The comparison of PIBI in subjects using simvastatin and atorvastatin is presented in Figure 2. Subjects on simvastatin as well as those on atorvastatin showed 40% lower mean PIBI values than subjects without statin. We observed no difference in PIBI between subjects on simvastatin or atorvastatin. The small number of subjects on fluvastatin or rosuvastatin did not allow their inclusion in the comparison.

To describe average periodontal injury, we divided the PPD values and PIBI by the number of teeth. The tooth-adjusted means of all of these clinical periodontal inflammatory signs were consistently lower in statin-users than in non-users (Table 2).

Linear regression analysis of the effect of background variables on the outcome variables revealed that smoking and diabetes were associated with a higher number of periodontal pockets among statin using patients, but not in those without statin (Table 3). The subject's age did not have an effect on the observed differences in the number of periodontal pockets between statin users and non-users. Age had no effect on PIBI in statin users, but explained 5% of the variation in PIBI of subjects not having a statin medication (Table 3).

The primary data of this cross sectional, retrospective study based on clinical records of periodontitis patients can be considered reliable due to the extensive clinical examination protocol used for educational purposes in Finnish dental schools. The accuracy of the entries and the thoroughness of clinical examinations are controlled by senior staff members. The retrospective nature of the study places limits to external calibration of the measurements, and limits the interpretation of our measurements to the realm of a university dental clinic and the results are not necessarily applicable to a local dentist's practice.

Our subjects are Finnish adults from Helsinki area with advanced chronic periodontitis, either receiving regular statin medication or without statin medication. The number of teeth 26 as well as the percentage of statin users 27 in our subjects coincides well with that reported for this age group in Finland.

Clinical signs of periodontal tissue destruction have been reported to increase with age 28. In our study this effect of age was evident for subjects not using statins. Interestingly, however, the effect of age could not be detected in statin users.

Diabetes is frequent among CVD patients 29. This explains the higher percentage of diabetics in our statin-using patients when compared to non-users. Also, the proportion of current smokers was higher in statin-using than in non-using patients of this study. Both diabetes and smoking are predisposing factors for deterioration of periodontal tissue 15. In spite of this, the positive effect on periodontal health observed in the statin using group was clear and statistically significant.

Low socio-economic status has been reported as a predisposing factor for chronic periodontitis 30. CVD is associated with low socio-economic status 31. Statin users are CVD risk patients. Thus, it is unlikely that statin medication selects patients with high socio-economic status and that this could affect the results.

The inflammatory potential of dentition is proportional to the number of teeth. All of our subjects had at least 15 remaining teeth, thus having at least 50% of the total possible dental inflammation sites. The small difference in the number of teeth between the groups did not affect the results. Statins had a substantial positive effect on local periodontal inflammatory injury, evident also when adjusted for the number of teeth.

Untreated periodontal pockets may serve as a source of persistent medium-grade inflammation with systemic consequences although the inflammation is milder and more local than in, for instance, rheumatoid arthritis. It is noteworthy that the prevalence of rheumatoid arthritis in the general population is about 1%, whereas the prevalence of gingival inflammation exceeds 50% 13. Deep periodontal pockets liberate more inflammatory mediators and cytokines than shallow pockets, thus contributing more to the inflammatory burden. We used periodontal probing pocket depth (PPD) values to describe periodontal inflammatory injury. The number of tooth surfaces with increased PPD values reflects the total inflammatory burden caused by the local periodontal inflammation. The use of PPD instead of loss of clinical tooth attachment (PPD + gingival recession) is well reasoned in this context because measurements of clinical attachment loss do not reflect the inflammatory area of soft tissues surrounding the teeth.

The Periodontal Inflammatory Burden Index (PIBI) developed for this study takes into account the enlargement of inflamed subgingival surface area in the deeper periodontal pockets. This is achieved by weighting the number of advanced periodontal lesions with a coefficient 2 (see Equation 3). The actual value of the coefficient is probably greater than our estimate and remains to be elucidated in future studies.

Previous reports suggest that periodontal inflammation has systemic effects and can be a burden on remote tissues 193233. Chronic periodontitis causes deterioration of the diabetic condition 34, oral infections increase obstetric complications 35, and severe periodontitis is associated with systemic inflammation and a dysmetabolic status 19. Reduced number of teeth has been shown to be a corollary of development of CVD 3637 and an independent risk factor of aortic valve sclerosis 38. The inflammatory component could be an important link between periodontal diseases and development of CVD 15161718323940. Interestingly, although atorvastatin is a more potent cholesterol lowering compound than simvastatin [41], in our study both simvastatin and atorvastatin showed an equal association with low values of periodontal inflammatory burden index PIBI.