Between-subject and within-subject statistical information in dental research

Assessing the immediate impact of Twin-block appliance insertion on adolescents' sleep using a wearable device

OBJECTIVE

This prospective clinical study aimed to evaluate the immediate impact of Twin-block appliance insertion on the sleep of adolescents using a wearable device.

MATERIALS AND METHODS

A total of 24 girls, aged 11-13 years, with Class II division 1 molar relationship, skeletal class 2 malocclusion (ANB ≥5) and overjet measuring ≥5 mm were selected. Exclusion criteria included a history of previous orthodontic treatment, systemic disease, irregular sleep pattern, obstructive sleep apnea, medical history of breathing disorders, or concurrent use of medications. Participants wore a wearable device to measure sleep parameters, including deep sleep, light sleep, minutes awake during sleep, wake-up times, bedtimes and total sleep times. The participants wore the device for 10 days prior to Twin-block insertion and sleep data were collected for another 10 days after insertion.

RESULTS

Following the insertion of the Twin-block appliance, there was a highly statistically significant shift in bedtime and wake-up time to later hours (P < .001). All participants experienced a highly significant delay in bedtime compared to the recommended 10 pm time (P < .001). Additionally, there was a significant increase in the duration of light sleep (P < .05). However, the effect on deep sleep, minutes awake during sleep and sleep duration was not statistically significant. None of the sleep parameters tested showed statistically significant changes between the first 5 days after Twin-block insertion with the subsequent 5 days.

CONCLUSION

The immediate insertion of the Twin-block appliance disrupts sleep onset, wake-up time and light sleep during the specified period of 10 days.

Frequency, location, and morphology of accessory maxillary sinus ostia: a retrospective study using cone beam computed tomography (CBCT)

PURPOSE

To evaluate the reliability of cone beam computed tomography (CBCT) imagining in the assessment of the frequency, location, and morphological characteristics of accessory maxillary ostia (AMOs), and to analyze a potential association with sinus and dentoalveolar pathologies.

METHODS

CBCT scans with bilateral maxillary sinuses that were acquired from September 2016 to September 2018 were initially screened. A total of 160 CBCT scans (320 sinuses) that fulfilled the inclusion criteria were included for further analysis. The presence, location, and morphological characteristics of the AMOs were evaluated in axial, coronal, and sagittal CBCT views. The findings were correlated with age, gender, sinus, and dentoalveolar pathology to assess for potential influencing factors on AMOs.

RESULTS

An AMO was present in 151 (47.2%) of the 320 sinuses. Most of the AMOs were located within the region of the nasal fontanelle or hiatus semilunaris (81.1%) presenting with an ovaloid (48.4%) or a round shape (39.0%). The average length of the AMOs was 2.33 ± 1.42 mm, occupying an area of 3.43 ± 4.51 mm², respectively. Morphological changes of the maxillary sinus mucosa were positively associated with length and area of AMOs. Furthermore, the status of the dentition in the posterior maxilla seemed to be an influencing factor on AMO shape.

CONCLUSIONS

Nearly half of the maxillary sinuses assessed in the present study population had an AMO. Pathologies of the maxillary sinus seem to have an impact on AMOs, which is demonstrated here by morphological changes of the sinus mucosa being associated with an increase in length and area of accessory ostia.

Repeated Measures Regression in Laboratory, Clinical and Environmental Research: Common Misconceptions in the Matter of Different Within- and between-Subject Slopes

When using repeated measures linear regression models to make causal inference in laboratory, clinical and environmental research, it is typically assumed that the within-subject association of differences (or changes) in predictor variable values across replicates is the same as the between-subject association of differences in those predictor variable values. However, this is often false. For example, with body weight as the predictor variable and blood cholesterol (which increases with higher body fat) as the outcome: (i) a 10-lb weight increase in the same adult affects more greatly an increase in cholesterol in that adult than does (ii) one adult weighing 10 lbs more than a second indicate higher cholesterol in the heavier adult. A 10-lb weight gain in the first adult more likely reflects a build-up of body fat in that person, while a second person being 10 lbs heavier than the first could be influenced by other factors, such as the second person being taller. Hence, to make causal inferences, different within- and between-subject slopes should be separately modeled. A related misconception commonly made using generalized estimation equations (GEE) and mixed models on repeated measures (i.e., for fitting cross-sectional regression) is that the working correlation structure only influences variance of the parameter estimates. However, only independence working correlation guarantees that the modeled parameters have interpretability. We illustrate this with an example where changing working correlation from independence to equicorrelation qualitatively biases parameters of GEE models and show that this happens because within- and between-subject slopes for the outcomes regressed on the predictor variables differ. We then systematically describe several common mechanisms that cause within- and between-subject slopes to differ: change effects, lag/reverse-lag and spillover causality, shared within-subject measurement bias or confounding, and predictor variable measurement error. The misconceptions we describe should be better publicized. Repeated measures analyses should compare within- and between-subject slopes of predictors and when they do differ, investigate the causal reasons for this.

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.

Frequency, location, and association with dental pathology of mucous retention cysts in the maxillary sinus. A radiographic study using cone beam computed tomography (CBCT)

OBJECTIVES

The purpose of the present study was to evaluate the frequency, locations, and dimensions of mucous retention cysts of the maxillary sinus and analyze potential associated dental pathology.

MATERIALS AND METHODS

A total of 156 cone beam computed tomography (CBCT) scans were included in the analysis, resulting in an evaluation of 310 maxillary sinuses. The presence of mucous retention cysts (MRC) manifesting as dome-shaped radiopacities in the sinus was diagnosed. Their locations were recorded, and dimensions (mm) were measured in coronal and sagittal/axial slices. The patients were grouped into (a) patients/sinuses with MRCs (test), and (b) patients/sinuses with healthy or any other changes (control) for further comparison and evaluation.

RESULTS

There were 40 sinuses (12.9%) with a presence of a total of 56 MRCs. The mean age of involved patients was 29.0 years. The analysis showed that gender, age, sinus side, status of dentition, endodontic status, and periodontal status did not have a significant influence on the presence of MRCs when compared between test and control groups. Age and endodontic status exhibited a significant association with cyst location.

CONCLUSIONS

Most of the sinuses analyzed (79.5%) did not present any MRC, and only 28.6% of the cysts diagnosed were found on the floor of the maxillary sinus. The mean dimension of the MRCs measured 6.28 ± 2.93 mm. No influencing factors on the presence or absence of MRCs were found in the present study.

CLINICAL RELEVANCE

Most MRCs were not located on the floor of maxillary sinus. Future studies should assess their impact on surgical interventions in the sinus.

Caries Experience in Children with and without Molar-Incisor Hypomineralisation: A Case-Control Study

Aim: The aim of this study was to compare the caries experience of children with and without molar-incisor hypomineralisation (MIH). Methods: A case-control study was designed in which 130 children aged between 7 and 13 years with MIH (cases) were matched with 130 children without the condition (controls) according to age, sex, and school. Dental caries and MIH were assessed using the Caries Assessment Spectrum and Treatment (CAST) and European Academy of Paediatric Dentistry (EAPD) criteria, respectively, by three examiners. CAST was converted into DMFT/dmft; the Kruskal-Wallis test was performed to analyse whether dmft/DMFT was influenced by the severity of MIH. Associations between MIH and dental caries were analysed at child and tooth levels: between and within subjects, respectively. To correlate MIH severity and the occurrence of dental caries, the Cochran-Armitage test was used. Results: The mean age of the children was 9.63 ± 1.29 years. The mean dmft for cases was 1.23 ± 1.99 and for controls 1.71 ± 2.22 (p > 0.05). For the DMFT, the mean scores for cases and controls were 0.45 ± 0.90 and 0.07 ± 0.25, respectively (p < 0.001). The between-subject analysis showed no difference in relation to enamel carious lesions; however, the prevalence of dentine carious lesions was significantly higher in children with MIH than in those without the condition. The same pattern was seen for the within-subject analysis. It was observed that the increase in MIH severity resulted in more teeth being affected by dentine carious lesions (p = 0.0003). Conclusion: Children with MIH presented a higher experience of caries in the permanent dentition than those without the condition. MIH was considered a risk factor for caries development.

One-stage individual participant data meta-analysis models: estimation of treatment-covariate interactions must avoid ecological bias by separating out within-trial and across-trial information

Stratified medicine utilizes individual‐level covariates that are associated with a differential treatment effect, also known as treatment‐covariate interactions. When multiple trials are available, meta‐analysis is used to help detect true treatment‐covariate interactions by combining their data. Meta‐regression of trial‐level information is prone to low power and ecological bias, and therefore, individual participant data (IPD) meta‐analyses are preferable to examine interactions utilizing individual‐level information. However, one‐stage IPD models are often wrongly specified, such that interactions are based on amalgamating within‐ and across‐trial information. We compare, through simulations and an applied example, fixed‐effect and random‐effects models for a one‐stage IPD meta‐analysis of time‐to‐event data where the goal is to estimate a treatment‐covariate interaction. We show that it is crucial to centre patient‐level covariates by their mean value in each trial, in order to separate out within‐trial and across‐trial information. Otherwise, bias and coverage of interaction estimates may be adversely affected, leading to potentially erroneous conclusions driven by ecological bias. We revisit an IPD meta‐analysis of five epilepsy trials and examine age as a treatment effect modifier. The interaction is −0.011 (95% CI: −0.019 to −0.003; p = 0.004), and thus highly significant, when amalgamating within‐trial and across‐trial information. However, when separating within‐trial from across‐trial information, the interaction is −0.007 (95% CI: −0.019 to 0.005; p = 0.22), and thus its magnitude and statistical significance are greatly reduced. We recommend that meta‐analysts should only use within‐trial information to examine individual predictors of treatment effect and that one‐stage IPD models should separate within‐trial from across‐trial information to avoid ecological bias. © 2016 The Authors. Statistics in Medicine published by John Wiley & Sons Ltd.

Daily Sleep Patterns, Sleep Quality, and Sleep Hygiene Among Parent-Child Dyads of Young Children Newly Diagnosed With Juvenile Idiopathic Arthritis and Typically Developing Children

OBJECTIVES

Describe daily sleep patterns, sleep quality, and sleep hygiene in 2-5-year-old children newly diagnosed with juvenile idiopathic arthritis (JIA) and their parents in comparison with typically developing (TD) children and parents.

METHODS

Participants (13 JIA, 16 TD parent-child dyads) wore actigraphs for 10 days. Parents completed sleep diaries and sleep hygiene survey.

RESULTS

Children with JIA had significantly less total sleep time, lower sleep efficiency (SE), and longer naps than TD children. Parents of children with JIA had significantly earlier bedtimes, more wake after sleep onset (WASO) and lower SE than TD parents. Parent-child SE and WASO were interrelated in JIA dyads. Sleep hygiene practices were inconsistent in both groups of children.

CONCLUSIONS

Inadequate amounts of sleep and poor sleep quality were common in parent-child dyads. Early interventions to improve sleep duration and promote sleep hygiene practices may alleviate future sleep problems and improve parent and child well-being.

Meta-analysis of a continuous outcome combining individual patient data and aggregate data: a method based on simulated individual patient data

When some trials provide individual patient data (IPD) and the others provide only aggregate data (AD), meta-analysis methods for combining IPD and AD are required. We propose a method that reconstructs the missing IPD for AD trials by a Bayesian sampling procedure and then applies an IPD meta-analysis model to the mixture of simulated IPD and collected IPD. The method is applicable when a treatment effect can be assumed fixed across trials. We focus on situations of a single continuous outcome and covariate and aim to estimate treatment-covariate interactions separated into within-trial and across-trial effect. An illustration with hypertension data which has similar mean covariates across trials indicates that the method substantially reduces mean square error of the pooled within-trial interaction estimate in comparison with existing approaches. A simulation study supposing there exists one IPD trial and nine AD trials suggests that the method has suitable type I error rate and approximately zero bias as long as the available IPD contains at least 10% of total patients, where the average gain in mean square error is up to about 40%. However, the method is currently restricted by the fixed effect assumption, and extension to random effects to allow heterogeneity is required.

Sleep measures predict next-day symptoms in women with irritable bowel syndrome

STUDY OBJECTIVES

Patients with irritable bowel syndrome (IBS) often report sleep disturbances. Previously, we have shown that self-reported sleep difficulties predicted exacerbations of next-day IBS symptoms, mood disturbance, and fatigue. The purpose of this study was to explore whether objectively measured sleep using actigraphy, as well as self-report, predicts next-day symptoms in women with IBS and to explore whether or not symptoms also predict self-report and objective sleep.

METHODS

Women aged 18-45 years with IBS were community-recruited (n = 24, mean age = 32 ± 8 years). Participants completed sleep and IBS symptom diaries for one menstrual cycle and wore Actiwatch-64 actigraphs for 7 days at home. Statistical analyses used generalized estimating equation (GEE) models.

RESULTS

Poorer self-reported sleep quality significantly (p < 0.05) predicted higher next-day abdominal pain, anxiety, and fatigue, but was not significant for gastrointestinal (GI) symptoms or depressed mood. Actigraphic sleep efficiency (SEF) significantly predicted worsening next-day anxiety and fatigue, but not abdominal pain, GI symptoms, or depressed mood. On temporally reversed analyses, none of the symptoms significantly predicted subsequent sleep, except that GI symptoms significantly predicted higher actigraphic sleep efficiency.

CONCLUSION

This small exploratory study supports previous findings that self-reported sleep disturbance predicted exacerbation of next-day symptoms in women with IBS and extends this relationship using an objective sleep measure. The study adds further evidence that sleep quality predicts subsequent IBS symptoms, but not the converse. The findings from this small study support the importance of additional longitudinal research to further understand the relationships between sleep and IBS.

Relationship between patterns of alcohol consumption and gastrointestinal symptoms among patients with irritable bowel syndrome

OBJECTIVES

Heavy alcohol intake may exacerbate gastrointestinal (GI) symptoms in adults with irritable bowel syndrome (IBS); however, the role of alcohol in IBS is unclear. We investigated prospective associations between daily patterns of alcohol intake and next day's GI symptoms using daily diaries.

METHODS

In an observational study of women aged 18-48 years with IBS and healthy controls, participants recorded daily GI symptoms, alcohol intake, caffeine intake, and cigarette smoking for ≈ 1 month. GI symptoms included abdominal pain, abdominal bloating, intestinal gas, diarrhea, constipation, nausea, stomach pain, heartburn, and indigestion. Binge drinking was defined as 4+ alcohol-containing drinks/day.

RESULTS

Patterns of alcohol intake did not differ between IBS patients and controls. Although patterns of drinking were associated with GI symptoms among women with IBS, this was not the case with the healthy controls. The strongest associations for IBS patients were between binge drinking and the next day's GI symptoms (e.g., diarrhea, P=0.006; nausea, P=0.01; stomach pain, P=0.009; and indigestion, P=0.004), whereas moderate and light drinking either were not associated or weakly associated with GI symptoms. Associations between alcohol intake and GI symptoms were stronger for women with IBS-diarrhea than for IBS-constipation or IBS-mixed. Effects of binge drinking on GI symptoms were strongest when comparing between individuals (rather than within individuals).

CONCLUSIONS

Our findings indicate that IBS symptoms differ according to the pattern of alcohol intake among IBS patients, suggesting that the pattern of drinking may in part explain the inconsistent findings between alcohol and IBS symptoms.

Hierarchical modeling of agreement

Kappa-like agreement indexes are often used to assess the agreement among examiners on a categorical scale. They have the particularity of correcting the level of agreement for the effect of chance. In the present paper, we first define two agreement indexes belonging to this family in a hierarchical context. In particular, we consider the cases of a random and fixed set of examiners. Then, we develop a method to evaluate the influence of factors on these indexes. Agreement indexes are directly related to a set of covariates through a hierarchical model. We obtain the posterior distribution of the model parameters in a Bayesian framework. We apply the proposed approach on dental data and compare it with the generalized estimating equations approach.

Informative cluster sizes for subcluster-level covariates and weighted generalized estimating equations

Williamson, Datta, and Satten's (2003, Biometrics 59, 36-42) cluster-weighted generalized estimating equations (CWGEEs) are effective in adjusting for bias due to informative cluster sizes for cluster-level covariates. We show that CWGEE may not perform well, however, for covariates that can take different values within a cluster if the numbers of observations at each covariate level are informative. On the other hand, inverse probability of treatment weighting accounts for informative treatment propensity but not for informative cluster size. Motivated by evaluating the effect of a binary exposure in presence of such types of informativeness, we propose several weighted GEE estimators, with weights related to the size of a cluster as well as the distribution of the binary exposure within the cluster. Choice of the weights depends on the population of interest and the nature of the exposure. Through simulation studies, we demonstrate the superior performance of the new estimators compared to existing estimators such as from GEE, CWGEE, and inverse probability of treatment-weighted GEE. We demonstrate the use of our method using an example examining covariate effects on the risk of dental caries among small children.

Meta-analysis of a binary outcome using individual participant data and aggregate data

In this paper, we develop meta-analysis models that synthesize a binary outcome from health-care studies while accounting for participant-level covariates. In particular, we show how to synthesize the observed event-risk across studies while accounting for the within-study association between participant-level covariates and individual event probability. The models are adapted for situations where studies provide individual participant data (IPD), or a mixture of IPD and aggregate data. We show that the availability of IPD is crucial in at least some studies; this allows one to model potentially complex within-study associations and separate them from across-study associations, so as to account for potential ecological bias and study-level confounding. The models can produce pertinent population-level and individual-level results, such as the pooled event-risk and the covariate-specific event probability for an individual. Application is made to 14 studies of traumatic brain injury, where IPD are available for four studies and the six-month mortality risk is synthesized in relation to individual age. The results show that as individual age increases the probability of six-month mortality also increases; further, the models reveal clear evidence of ecological bias, with the mean age in each study additionally influencing an individual's mortality probability. Copyright © 2010 John Wiley & Sons, Ltd.

Association between enamel hypoplasia and dental caries in primary second molars: a cohort study

The purpose of this study was to assess the longitudinal relationships between enamel hypoplasia and caries experience of primary second molars. The study sample was 491 subjects who received dental examinations at both age 5 and 9 by the calibrated examiners. Four primary second molars (n = 1,892) were scored for the presence of enamel hypoplasia for each participant. Caries presence and number of decayed and filled surfaces (dfs) were determined at age 5 and 9. The relationships between enamel hypoplasia and caries experience were assessed. Among primary second molars, 3.9% of children and 1.7% of primary second molars had enamel hypoplasia. At age 5, 36.8% of children with hypoplasia had caries, while 16.9% of children without enamel hypoplasia had caries. At age 9, the corresponding numbers were 52.6% for children with hypoplasia and 34.5% for children without hypoplasia, respectively. At the tooth level, for age 5, 28.1% of teeth with hypoplasia had caries (mean dfs = 0.40), and 7.6% of teeth without hypoplasia had caries (mean dfs = 0.11). At age 9, the corresponding numbers were 41.9% (mean dfs = 0.76) for teeth with hypoplasia and 18.3% (mean dfs = 0.34) for teeth without hypoplasia. In multivariable logistic regression analyses, teeth of subjects with enamel hypoplasia had a significantly higher risk for caries at age 5 and 9 after controlling for other risk factors. Enamel hypoplasia appears to be a significant risk factor for caries and should be considered in caries risk assessment.

Are comparisons of consumer satisfaction with providers biased by nonresponse or case-mix differences?

OBJECTIVE

This study examined how consumer satisfaction ratings differ between mental health care providers to determine whether comparison of ratings is biased by differences in survey response rates or consumer characteristics.

METHODS

Consumer satisfaction surveys mailed by a mixed-model prepaid health plan were examined. Survey data were linked to computerized records regarding consumers' demographic (age, sex, and type of insurance coverage) and clinical (primary diagnosis and initial versus return visit) characteristics. Statistical models examined probabilities of returning the survey (N=8,025 returned surveys) and of giving an excellent satisfaction rating. Variability was separated into within-provider effects and between-provider effects.

RESULTS

The overall response rate was 33.8%, and 49.9% of responders reported excellent satisfaction. Neither response rate nor satisfaction rating was related to primary diagnosis. Within the practices of individual providers, response rate and receiving an excellent rating were significantly associated with female sex, older age, longer enrollment in the health plan, and making a return visit. Analyses of between-provider effects, however, found that only a higher proportion of return visitors was significantly associated with higher response rates and higher satisfaction ratings.

CONCLUSIONS

There was little evidence that differences in response rate or in consumers served biased comparison of satisfaction ratings between mental health providers. Bias might be greater in a setting with more heterogeneous consumers or providers. Returning consumers gave higher ratings than first-time visitors, and analyses of satisfaction ratings may need to account for this difference. Extremely high or low ratings should be interpreted cautiously, especially for providers with a small number of surveys.

Evaluating the effect of change on change: a different viewpoint

RATIONALE

When a causal variable and its presumed effect are measured at two time points in a cohort study, most researchers prefer to fit some type of a change model. Many of them believe that such an analysis is superior to a cross-sectional analysis 'because change models estimate the effect of change on change', which sounds epistemologically stronger than 'estimating a cross-sectional association'.

METHODS

In this paper I trace two commonly used regression models of change to their cross-sectional origin and describe these models from the perspectives of time-stable confounders, effect modification, and causal diagrams. In addition, I cite three viewpoints from the statistical literature.

RESULTS

The so-called change models do not estimate anything conceptually different from cross-sectional models. A change model is superior to a cross-sectional model mainly because it corresponds to a self-matched design. Statistical viewpoints markedly differ about the appropriate parameterization and interpretation of such data.

CONCLUSION

Contrary to prevailing thought, a model of changes between two time points does not estimate any special causal idea called 'longitudinal effect'. The main advantage of regressing 'change on change' is complete control of time-stable confounders, a key concern in observational studies. Many analysts fail to realize that that important advantage is usually lost when they fit a random effects model.

Meta-analysis of continuous outcomes combining individual patient data and aggregate data

Meta-analysis of individual patient data (IPD) is the gold-standard for synthesizing evidence across clinical studies. However, for some studies IPD may not be available and only aggregate data (AD), such as a treatment effect estimate and its standard error, may be obtained. In this situation, methods for combining IPD and AD are important to utilize all the available evidence. In this paper, we develop and assess a range of statistical methods for combining IPD and AD in meta-analysis of continuous outcomes from randomized controlled trials. The methods take either a one-step or a two-step approach. The latter is simple, with IPD reduced to AD so that standard AD meta-analysis techniques can be employed. The one-step approach is more complex but offers a flexible framework to include both patient-level and trial-level parameters. It uses a dummy variable to distinguish IPD trials from AD trials and to constrain which parameters the AD trials estimate. We show that this is important when assessing how patient-level covariates modify treatment effect, as aggregate-level relationships across trials are subject to ecological bias and confounding. We thus develop models to separate within-trial and across-trials treatment-covariate interactions; this ensures that only IPD trials estimate the former, whilst both IPD and AD trials estimate the latter in addition to the pooled treatment effect and any between-study heterogeneity. Extension to multiple correlated outcomes is also considered. Ten IPD trials in hypertension, with blood pressure the continuous outcome of interest, are used to assess the models and identify the benefits of utilizing AD alongside IPD.

Modeling multivariate binary responses with multiple levels of nesting based on alternating logistic regressions: an application to caries aggregation

Clustered binary responses are commonly encountered in dental research. Data analysis may include modeling both the marginal response probabilities (i.e., risk) and the dependence structure between pairs of responses (i.e., aggregation). While second-order generalized estimating equations (GEE2) is a well-known approach for such data, alternating logistic regressions (ALR) is a computationally efficient alternative method, especially for large clusters. We illustrate ALR with an application to caries aggregation using a dataset with 3 levels of nesting: tooth surfaces within an interproximal (IP) region, IP regions within a jaw, and jaws within a subject. Caries lesions appear to aggregate strongly within subjects with a spatially distributed risk. The minimum within-IP-region odds ratio (OR) was 2.25 (95% confidence interval 1.15, 4.41), and the within-IP-region ORs were always greater than the between-IP-region ORs. ALR is a convenient and useful regression technique for explicit modeling of the dependence structure, and may be applicable to other dental research problems involving clustered or nested responses.

The Application of Multilevel Modeling in the Analysis of Longitudinal Periodontal Data – Part I: Absolute Levels of Disease

BACKGROUND

Statistical analyses of periodontal data that average site measurements to subject mean values are unable to explore the site-specific nature of periodontal diseases. Multilevel modeling (MLM) overcomes this, taking hierarchical structure into account. MLM was used to investigate longitudinal relationships between the outcomes of lifetime cumulative attachment loss (LCAL) and probing depth (PD) in relation to potential risk factors for periodontal disease progression.

METHODS

One hundred males (mean age 17 years) received a comprehensive periodontal examination at baseline and at 12 and 30 months. The resulting data were analyzed in two stages. In stage one (reported here), the absolute levels of disease were analyzed in relation to potential risk factors; in stage two (reported in a second paper), changes in disease patterns over time were analyzed in relation to the same risk factors. Each approach yielded substantially different insights.

RESULTS

For absolute levels of disease, subject-level risk factors (covariates) had limited prediction for LCAL/PD throughout the 30-month observation period. Tooth position demonstrated a near linear relationship for both outcomes, with disease increasing from anterior to posterior teeth. Sites with subgingival calculus and bleeding on probing demonstrated more LCAL and PD, and supragingival calculus had an apparently protective effect. Covariates had more "explanatory power" for the variation in PD than for the variation in LCAL, suggesting that LCAL and PD might be generally associated with a different profile of covariates.

CONCLUSION

This study provides, for a relatively young cohort, considerable insights into the factors associated with early-life periodontal disease and its progression at all levels of the natural hierarchy of sites within teeth within subjects.

Separation of individual-level and cluster-level covariate effects in regression analysis of correlated data

The focus of this paper is regression analysis of clustered data. Although the presence of intracluster correlation (the tendency for items within a cluster to respond alike) is typically viewed as an obstacle to good inference, the complex structure of clustered data offers significant analytic advantages over independent data. One key advantage is the ability to separate effects at the individual (or item-specific) level and the group (or cluster-specific) level. We review different approaches for the separation of individual-level and cluster-level effects on response, their appropriate interpretation and give recommendations for model fitting based on the intent of the data analyst. Unlike many earlier papers on this topic, we place particular emphasis on the interpretation of the cluster-level covariate effect. The main ideas of the paper are highlighted in an analysis of the relationship between birth weight and IQ using sibling data from a large birth cohort study.