The Application of Multilevel Modeling in the Analysis of Longitudinal Periodontal Data –Part II: Changes in Disease Levels over Time

Endpoints of active periodontal therapy

Aim

Position paper on endpoints of active periodontal therapy for designing treatment guidelines. The question was as follows: How are, for an individual patient, commonly applied periodontal probing measures—recorded after active periodontal therapy—related to (a) stability of clinical attachment level, (b) tooth survival, (c) need for re‐treatment or (d) oral health‐related quality of life.

Methods

A literature search was conducted in Ovid MEDLINE(R) and Epub Ahead of Print, In‐Process & Other Non‐Indexed Citations and Daily <1946 to 07 June 2019>.

Results

A total of 94 papers were retrieved. From the literature search, it was found that periodontitis patients with a low proportion of deep residual pockets after active periodontal therapy are more likely to have stability of clinical attachment level over a follow‐up time of ≥1 year. Other supporting literature confirms this finding and additionally reports, at the patient level, that probing pocket depths ≥6 mm and bleeding on probing scores ≥30% are risks for tooth loss. There is lack of evidence that periodontal probing measures after completion of active periodontal treatment are tangible to the patient.

Conclusions

Based on literature and biological plausibility, it is reasonable to state that periodontitis patients with a low proportion of residual periodontal pockets and little inflammation are more likely to have stability of clinical attachment levels and less tooth loss over time. Guidelines for periodontal therapy should take into consideration (a) long‐term tangible patient outcomes, (b) that shallow pockets (≤4 mm) without bleeding on probing in patients with <30% bleeding sites are the best guarantee for the patient for stability of his/her periodontal attachment, (c) patient heterogeneity and patient changes in immune response over time, and (d) that treatment strategies include lifestyle changes of the patient. Long‐term large population‐based and practice‐based studies on the efficacy of periodontal therapies including both clinical and patient‐reported outcomes (PROs) need to be initiated, which include the understanding that periodontitis is a complex disease with variation of inflammatory responses due to environment, (epi)genetics, lifestyle and ageing. Involving people living with periodontitis as co‐researchers in the design of these studies would also help to improve their relevance.

Fine-tuning multilevel modeling of risk factors associated with nonsurgical periodontal treatment outcome

This retrospective study evaluated the influence of known risk factors on nonsurgical periodontal treatment (NSPT) response using a pocket depth fine-tuning multilevel linear model (MLM). Overall, 37 patients (24 males and 13 females) with moderate-to-severe chronic periodontitis underwent NSPT. Follow-up visits at 3, 6, and 12 months included measurements of several clinical periodontal parameters. Data were sourced from a previously reported database. In a total of 1416 initially affected sites (baseline PD ≥ 4 mm) on 536 teeth, probing depth (PD) and clinical attachment loss (CAL) reductions after NSPT were evaluated against known risk factors at 3 hierarchical levels (patient, tooth, and site). For each post-treatment follow-up, the variance component models fitted to evaluate the 3-level variance of PD and CAL decrease revealed that all levels contributed significantly to the overall variance (p < 0.001). Patients who underwent NSPT and were continually monitored had curative results. All 3 hierarchical levels included risk factors influencing the degree of PD and CAL reduction. Specifically, the type of tooth, surfaces involved, and tooth mobility site-level risk factors had the strongest impact on these reductions and were highly relevant for the success of NSPT.

Long-term assessment of periodontal disease progression after surgical or non-surgical treatment: a systematic review

The primary aim of this systematic review was to assess the evidence on periodontal disease progression after treatment in patients receiving supportive periodontal therapy (SPT) and to identify predictors of clinical attachment level (CAL) loss. A protocol was developed to answer the following focused question: In adult patients treated for periodontitis, what is the disease progression in terms of CAL loss after surgical or non-surgical treatment? Randomized controlled clinical trials, prospective cohort studies, and longitudinal observational human studies with a minimum of 5 years of follow-up after surgical or non-surgical treatment that reported CAL and probing depth changes were selected. Seventeen publications reporting data from 14 investigations were included. Data from 964 patients with a follow-up range of 5-15 years was evaluated. When the CAL at the latest follow-up was compared to the CAL after active periodontal therapy, 10 of the included studies reported an overall mean CAL loss of ≤0.5 mm, 3 studies reported a mean CAL loss of 0.5-1 mm, and 4 studies reported a mean CAL loss of >1 mm. Based on 7 publications, the percentage of sites showing a CAL loss of ≥2 mm varied from 3% to 20%, and a high percentage of sites with CAL loss was associated with poor oral hygiene, smoking, and poor compliance with SPT. The outcomes after periodontal therapy remained stable over time. Disease progression occurred in a reduced number of sites and patients, mostly associated with poor oral hygiene, poor compliance with SPT, and smoking.

Site-level progression of periodontal disease during a follow-up period

Periodontal disease is assessed and its progression is determined via observations on a site-by-site basis. Periodontal data are complex and structured in multiple levels; thus, applying a summary statistical approach (i.e., the mean) for site-level evaluations results in loss of information. Previous studies have shown the availability of mixed effects modeling. However, clinically beneficial information on the progression of periodontal disease during the follow-up period is not available.

We conducted a multicenter prospective cohort study. Using mixed effects modeling, we analyzed 18,834 sites distributed on 3,139 teeth in 124 patients, and data were collected 5 times over a 24-month follow-up period. The change in the clinical attachment level (CAL) was used as the outcome variable. The CAL at baseline was an important determinant of the CAL changes, which varied widely according to the tooth surface. The salivary levels of periodontal pathogens, such as Porphyromonas gingivalis and Aggregatibacter actinomycetemcomitans, were affected by CAL progression. “Linear”- and “burst”-type patterns of CAL progression occurred simultaneously within the same patient. More than half of the teeth that presented burst-type progression sites also presented linear-type progression sites, and most of the progressions were of the linear type. Maxillary premolars and anterior teeth tended to show burst-type progression. The parameters identified in this study may guide practitioners in determining the type and extent of treatment needed at the site and patient levels. In addition, these results show that prior hypotheses concerning "burst" and "linear" theories are not valid.

Risk factors of chronic periodontitis on healing response: a multilevel modelling analysis

Background

Chronic periodontitis is a multifactorial polygenetic disease with an increasing number of associated factors that have been identified over recent decades. Longitudinal epidemiologic studies have demonstrated that the risk factors were related to the progression of the disease. A traditional multivariate regression model was used to find risk factors associated with chronic periodontitis. However, the approach requirement of standard statistical procedures demands individual independence. Multilevel modelling (MLM) data analysis has widely been used in recent years, regarding thorough hierarchical structuring of the data, decomposing the error terms into different levels, and providing a new analytic method and framework for solving this problem. The purpose of our study is to investigate the relationship of clinical periodontal index and the risk factors in chronic periodontitis through MLM analysis and to identify high-risk individuals in the clinical setting.

Methods

Fifty-four patients with moderate to severe periodontitis were included. They were treated by means of non-surgical periodontal therapy, and then made follow-up visits regularly at 3, 6, and 12 months after therapy. Each patient answered a questionnaire survey and underwent measurement of clinical periodontal parameters.

Results

Compared with baseline, probing depth (PD) and clinical attachment loss (CAL) improved significantly after non-surgical periodontal therapy with regular follow-up visits at 3, 6, and 12 months after therapy. The null model and variance component models with no independent variables included were initially obtained to investigate the variance of the PD and CAL reductions across all three levels, and they showed a statistically significant difference (P < 0.001), thus establishing that MLM data analysis was necessary. Site-level had effects on PD and CAL reduction; those variables could explain 77–78% of PD reduction and 70–80% of CAL reduction at 3, 6, and 12 months. Other levels only explain 20–30% of PD and CAL reductions. Site-level had the greatest effect on PD and CAL reduction.

Conclusions

Non-surgical periodontal therapy with regular follow-up visits had a remarkable curative effect. All three levels had a substantial influence on the reduction of PD and CAL. Site-level had the largest effect on PD and CAL reductions.

Electronic supplementary material

The online version of this article (10.1186/s12911-017-0533-2) contains supplementary material, which is available to authorized users.

Factors influencing the outcome of non-surgical periodontal treatment: a multilevel approach

AIM

To investigate, by means of multilevel analysis, factors that may affect the short-term clinical outcome of non-surgical periodontal treatment.

MATERIALS AND METHODS

Forty-one patients randomly assigned to two protocols of non-surgical therapy were included. The impact of different covariates on the probability of "pocket closure" [i.e. probing pocket depth (PPD)<or=4 mm] was explored using a logistic multilevel model. The impact on the final PPD was explored using a continuous multilevel model.

RESULTS

The logistic model revealed a significant impact of smoking (p<0.001), presence of plaque at the site (p<0.001) and location of the pocket at a multi-rooted tooth (p<0.001). The model explained 44% of the total variability. Of the unexplained variance, 19% was attributed to inter-patient variability. The continuous model revealed the same factors to be significant and an additional significant impact of interactions between the covariates. The R(2) was 0.50 and the random slopes model revealed an increase in the variability of the final pocket depth with an increase in the initial PPD.

CONCLUSION

Smoking habits, plaque at site level and tooth type were significant factors in determining the short-term clinical outcome of non-surgical periodontal treatment.

Area deprivation and oral health in Scottish adults: a multilevel study

OBJECTIVES

To explore the association between area deprivation and adult oral health.

METHODS

Multilevel regression analysis of data taken from the 1998 Adult Dental Health Survey in the UK comprising 632 participants living in 346 households located in 31 postcode sectors in Scotland. Three oral health outcomes were investigated: number of sound (including restored) teeth; has one or more unsound teeth; has periodontal pocketing 4 mm or more in one or more teeth.

RESULTS

In the most deprived areas, individuals had a mean of 4.6 fewer sound teeth than those in the least deprived areas. The difference in normalised number of sound teeth between least and most deprived areas was -0.707 (95% CI -1.164, -0.250), P = 0.024. After adjusting for age, sex, qualification status, head of household social class and household income, the estimated difference was -0.238 (-0.591 to 0.115) (P = 0.164). Area deprivation was not associated with having one or more unsound teeth or periodontal pocketing 4 mm or more in one or more teeth.

CONCLUSIONS

There is a univariate association of area deprivation with the number of sound teeth. This association is largely explained by household and individual level socioeconomic variables. A small area deprivation effect cannot be excluded in these data. The findings challenge current understanding of the relationship between area deprivation and oral health. Further multilevel research exploring the relationship between area deprivation and oral health is required using a larger sample and a prospective longitudinal design.

The application of multilevel modelling to dental caries data

Clinical studies of dental caries experience generate multiple outcome data for each participant, with information collected for each individual tooth surface. This paper investigates multilevel modelling as a method of analysis for dental caries data, allowing for full use of the data collected at surface level. Data from a clinical trial of a caries preventive agent in adolescents are modelled. The effect of tooth position within the mouth on the development of dental caries is investigated, with the results showing the importance of differentiating between the upper and lower arches, when modelling the probabilities of caries developing on teeth. Calculation of the intracluster correlation using the threshold model is suggested for use in multilevel logistic regression modelling of caries data. This model, which assumes that a dichotomous outcome is based on an underlying continuous variable with a threshold point where the outcome changes from zero to one, is identified to be appropriate for the analysis of caries which is a continuous process, but is only identified as present in a clinical trial when it has reached a certain level of severity.

Description of a System Designed to Assist Primary Dental Care Clinicians in Decision-Making with Regard to Specialist Periodontal Referrals and Report of Two Clinical Audits using the System

Introduction

Identifying which patients are at risk of significant periodontal breakdown and may need specialist care is a process fraught with problems. This paper summarises factors that should be considered by referring clinicians, describes a system to assist them, and presents the results of two sequential audits carried out in general dental practice using the system.

Aims, Materials and Methods

The author aims to present the system and the results of two sequential audits, carried out in the same eight general dental practices by 23 general dental practitioners (GDPs) after training from a specialist in periodontics. The system is based on the use of the Basic Periodontal Examination (BPE), a constant pressure periodontal probe, and a computer screen that prompts the clinician who is performing periodontal screening. A series of risk factors, including smoking, presence of relevant systemic disease and age, is used to modulate the score. Patients are then classified into one of three complexity groups. Those in the Complexity 3 group are deemed to be potentially at risk of significant periodontal breakdown and in need of referral for specialist advice and/or care, in accordance with the published British Society of Periodontology guidelines. In each audit a random sample of the records of ten dentate patients over the age of 20 years, who attended for recall or new examinations by each of the 23 GDPs over a calendar month, were assessed to see whether or not BPE scores had been recorded and whether or not those classified as Complexity 3 (and in need of referral) had been correctly identified and referred.

Results

In both audits the total random sample of patient records was 225 (ten for 22 dentists and five for the remaining dentist who only worked for a short time during the months of the audit). In the first audit cycle 139 (62%) of the random sample of records included a BPE score and 11 (8%) of the patients concerned were identified as Complexity 3 and referred for specialist advice and/or treatment. All 11 accepted the referral and when examined by the specialist in periodontics were confirmed as Complexity 3. In the second audit cycle, which took place seven months later, 179 (80%) of the random sample of patient records included a BPE score and 14 patients were assessed by their GDPs as being Complexity 3. All 14 accepted a referral and were deemed to be Complexity 3 when they were subsequently examined by the specialist in periodontics.

Conclusion

Although the full potential of the system was not investigated during the two audits, these preliminary results suggest that it may assist referring clinicians in identifying which patients to refer and encourage clinicians to perform periodontal screening.

Collinearity in linear regression is a serious problem in oral health research

The aim of this article is to encourage good practice in the statistical analysis of dental research data. Our objective is to highlight the statistical problems of collinearity and multicollinearity. These are among the most common statistical pitfalls in oral health research when exploring the relationship between clinical variables using multiple regression analysis. We hope that this article will show why these problems arise and how they can be avoided and overcome. Examples from the periodontal literature will be used to illustrate how collinearity and multicollinearity can seriously distort the model development process as a result of the phenomenon of mathematical coupling. Knowledge of these problems can help to eliminate misleading results and improve any subsequent interpretations. Regression analyses are useful tools in oral health research when their limitations are recognized. However, care is required in planning and it is worthwhile seeking statistical advice when formulating the study's research questions.