Periodontology 2000, Vol.

58, 2012, 112–120
Printed in Singapore. All rights reserved 

2012 John Wiley & Sons A/S

PERIODONTOLOGY 2000

Analytic epidemiology and
periodontal diseases
BRENDA HEATON & THOMAS DIETRICH

A primary aim of epidemiological research is to
identify the causes of disease. Analytic study designs
are strategies developed to measure disease causation according to fundamental principles of
causation and causal theory. Two key principles of
causation remain at the forefront in critical discussions of analytic designs. One is the temporal
sequence of the exposure and the disease outcome,
and a second is the elimination of alternative explanations (e.g. bias and confounding). In addition to
issues of validity, the importance of the precision of
effect estimates has also received attention.
Previous work has reviewed the analytic epidemiological literature and highlighted issues in the study
of periodontitis etiology other than those related to
study design (2). We review analytic study designs for
periodontal epidemiological research within the
context of validity. Specifically, our aim is to re-frame
the perspective on the strength of evidence in reviews
of the literature, as well as to promote proper design
and conduct of periodontal studies as they relate to
validity. We provide an overview of the design and
conduct of cohort studies, including randomized
designs, case–control and cross-sectional studies,
while also drawing specifically upon the periodontal
research literature.

Theoretical framework for analytic
designs
The goal of any epidemiological study is to accurately
and precisely estimate the effect of a factor on
development of a given outcome. This estimation is
arrived at through a process of study design, conduct
and analysis. The aim of this multi-step process is to
produce a causal contrast by comparing outcomes
across levels of the causal component of interest. The
causal contrast that is aimed for through formal study

112

is best illustrated by the counterfactual ideal, and is
common to both randomized and non-randomized
studies. The counterfactual ideal, discussed in detail
elsewhere in this volume, is an ideal that emphasizes
the properties required for a valid causal contrast at
the outset of any study (27), namely that the unexposed group is comparable to the exposed group in
every other way except for the causal component (the
exposure) under study. Therefore, the manner in
which we design epidemiological studies to estimate
causal effects should mimic the counterfactual ideal.
The counterfactual approach, which is best illustrated by the potential outcomes model, provides a
unifying framework for designing, analyzing and
interpreting epidemiological studies, and assists in
understanding the concerns related to validity (46).
The notion of designing a study with a valid causal
contrast is readily accepted by many investigators
because of their grounding in the paradigm of randomized controlled trials (6). Randomization can be
used as an exposure assignment mechanism that
assists in achieving a valid causal contrast by producing comparability of groups. The strength of this
tool produces a level of comfort in terms of validity
that at times results in restriction of causal inference
to the paradigm of randomized trials by many
investigators (68). Although this paradigm is helpful
in understanding the basic framework of causal
contrasts under the counterfactual ideal, it does not
serve as a supreme model for analytic epidemiology
(52). RubinÕs (67) work in the 1970s, which extended
the counterfactual approach (potential outcomes
model) to allow understanding of causal effects
beyond randomization and randomization-based
inference, initiated development of the unifying
principles described here. This extension emphasizes
that randomization is only one type of exposure
assignment mechanism among many that adhere to
the same principles of causal contrasts. All analytic

Cohort studies of periodontal disease etiology are primarily limited to the evaluation of smoking. For example. application of this framework aids in verification of assumptions required to statistically evaluate causal effects (28. Approaches to identifying confounders and evaluating confounding are best understood under the counterfactual framework. precision. the cohort design has proven particularly useful for the study of systemic effects of periodontal disease. individuals are separated according to their level of the exposure variable at baseline and often at subsequent intervals over a specified follow-up period as well (time-varying exposure). Cohort studies Design. There are important conceptual ties between randomized and non-randomized cohorts and between non-randomized cohorts and case–control studies. a process that is paramount to designing analytic strategies (68). Cohort studies are studies in which the incidence of disease outcomes is measured and compared across two or more populations. the Black WomenÕs Health Study (63) and the NursesÕ 113 . an association measure estimating the causal contrast is confounded (or biased due to confounding) if it does not equal the causal contrast in the target population because of imperfect substitution (non-comparability) for the counterfactual experience (48). the most preferred measure of disease frequency for studying causal effects. Cohort studies are inherently ÔprospectiveÕ. Understanding causal contrasts under the counterfactual framework is important to our criticisms of analytic epidemiology as it pertains to validity. two prominent epidemiologists. For general cohorts. because all analytic designs should estimate causal contrasts. in that exposure levels of individuals in the cohort are measured at a time prior to any disease outcomes of interest. For this reason. and person–time is accrued until development (incidence) of the outcome. 61). the Dental Longitudinal Study (36). To paraphrase Maldonado and Greenland (48). diabetic and bacterial exposures (2). Adequate analysis may require the consideration of additional analytic options for effect estimation. and (3) sampling subjects from the target and substitute populations to balance trade-offs among bias. If the assignment mechanism for exposure is not random. In a cohort study with regular follow-up. the reference cohort can be thought of as a ÔconfoundedÕ substitute for the counterfactual experience. The source population enumerated in a cohort study is most often formed on the basis of some common characteristic or experience. an investigator has enormous flexibility with respect to analytic approaches. 35. conduct and analysis The classic cohort study is introduced first in order to illustrate that all analytic designs discussed here can be thought of as nested within an underlying cohort from which design principles emanate (51). different analytic designs can be viewed simply as different ways of (1) choosing a target population that corresponds to the study question. The use of randomization of exposure assignment is only applicable in experimental designs. typically a cohort of those individuals exposed to the factor under study and a cohort of those who are not. possible systemic effects of periodontal disease have recently been evaluated in the Veterans Administration Dental Longitudinal Study (36). (2) choosing a substitute population for the counterfactual experience. confounding is therefore an additionally important consideration in non-randomized studies. coronary heart disease and cognition (8. The cohort design allows study of myriad health effects that stem from a single exposure. study costs and study time. Generally. In case– control studies. particularly confounding and selection bias. non-randomized cohorts are often established without regard to a specific causal contrast. this is manifested as selection bias. This perspective is particularly important for appreciation of the case–control design. Finally. death or loss to follow-up. including outcomes relating to stroke. 60). Validity General. individuals contribute Ôtime at riskÕ to various categories of an exposure or confounder over time. By accumulation of person–time. as assumptions necessary for conventional approaches are often not met in this scenario (23. 37). The accumulation of person–time allows direct calculation of the incidence rate. Examples include the Health Professionals Follow-up Study (31).Analytic epidemiology and periodontal diseases designs should therefore show careful consideration of the assignment mechanism leading to an individualÕs exposure status. which facilitates definitions of confounding that are reflective of the analytic design (28). Individuals are followed over time. The added flexibility comes in large part due to the accumulation of time rather than individuals in the denominator of the rate.

most notably prospective follow-up with accumulation of person–time for the direct calculation of incidence rates. This is of particular concern when substantial loss to follow-up occurs prior to ascertainment of the outcome. whether randomization is employed or not. This quantitative assessment greatly assists in discussion of the possible influence of bias on the effect estimate by determining what unobserved covariates or other biases would have to be like in order to alter conclusions drawn from the effect estimate. on average. these studies have limited application and can only be utilized when the causal contrast of interest can be randomized. as the original match does not extend throughout the person–time available for analysis (25). Of primary concern in a study with longitudinal follow-up is the potential for biased loss to follow-up (18). Additional analytic approaches have been developed to assess the possible influence of remaining bias on the effect estimate (22). competing risks and loss to follow-up. expected to be non-differential. Given the possibility of additional design approaches in non-randomized studies. Some thought must be taken to avoiding differential errors in disease classification but they are often not difficult to avoid. the assignment mechanism) have been measured and can be enumerated so that objective inferences for causal effects can be made. as well as use of quantitative approaches to analyzing the extent of bias. this should. Overall. 27). However. Aside from confounding. similar to randomized studies (19. non-randomized cohort studies should not generally be considered inferior to studies employing randomization. each evaluation of a particular causal contrast within the cohort as the source population will have to give thought to how well the unexposed experience can stand in for the exposed experience in order to guide assessments of validity. Matching unexposed individuals to exposed individuals within categories of a potential confounder does not eliminate confounding by that factor as many would expect. 45). Furthermore. 21. 79). properly conducted cohort studies can promote confidence in accurate estimation of the causal contrast under study. this tool becomes particularly useful. matching at baseline becomes largely inefficient in cohort studies. Dietrich and Garcia (7) performed one of the few applications of basic sensitivity analysis in the periodontal literature in order to evaluate the extent of possible bias due to mis-classification of periodontal disease status in studies of oral–systemic disease associations. The use of matching in any analytic design should be carefully considered. A more commonly known and undertaken approach is matching. Therefore. selection into the cohort is rarely biased.g. Randomized trials can be thought of as a special type of cohort study as they share several design features. Furthermore. In doing so. Therefore. When this happens. . Randomized cohort studies Pre-eminent among analytic study designs has been the randomized trial – a design approach which most closely resembles the counterfactual ideal through use of randomization of exposure. but instead works to increase the statistical efficiency of confounder control in the analysis. concern often remains regarding the possible presence of bias in the effect estimate due to mis-classification. Inadequate justification for matching can lead to bias and losses in efficiency (overmatching) (42). some of which have been applied in an oral health context (72). specifically whether factors leading to an individualÕs exposure status (e.Heaton & Dietrich Health Study (1). This endeavor becomes particularly worthwhile when studying multiple outcomes of one exposure. by extension. methods should be used to estimate the possible bias (22). Among them is bias analysis. propensity scores (62) or instrumental variables (16. Possible approaches to achieving balance prior to any evaluation of an outcome include utilization of sub-classification (6). one can evaluate the extent to which these factors are balanced across exposure levels and identify possible approaches to achieving balance prior to any evaluation of outcomes. selection bias or unmeasured confounding. Given the potential for mis-classification of periodontal disease. particularly confounding. which is always theoretically nested within a cohort. promote confidence in the case–control design. whether enumerated or not (51. and errors in the collection and measurement of exposure information are. to name but a few. the inherent prospective nature of cohort studies prevents many systematic errors that favor exposure or disease. Randomization increases the probability that the unexposed experience can stand in for the exposed experience had they been unexposed (no confounding) (28). with time-varying exposures. more commonly known as sensitivity analysis (20. 114 As classic analytic approaches can only account for measured confounding and random errors. 24. 34). to name but a few.

Approaches to minimizing confounding in a case–control study are the same as for a cohort study. 65. Of all analytic designs. if the underlying cohort or study base is inherently prospective (e. The ability of randomization to create a balanced distribution of covariates across exposure groups is dependent on study size. A well-conducted case–control study is subject to the same limitations as the underlying cohort within which it is nested. 51. Subsequently. cases selected from that study base will also be biased. regardless of the rarity of disease (29. for example. matching controls to cases does not prevent confounding. case–control studies are probably the most misunderstood (64). Foremost among such misconceptions is the view that case– control studies are inferior to cohort studies in terms of producing valid effect estimates (75). if disease is misclassified in the underlying cohort through diagnostic bias or some other means. 55). they still suffer from the same vulnerabilities to bias as a non-randomized cohort design. Additionally. over longitudinal follow-up. Lastly. Case–control studies Case–control studies are studies in which a source population is identified. and that they are therefore inherently ÔretrospectiveÕ (11). 57. Careful consideration of this potential source of bias is particularly important in clinic or hospital-based case–control studies (3). and within which all cases of disease (preferably incident) are identified. Additionally. randomized studies are subject to non-compliance with the study protocol. and can occur equally in cohort and case–control studies. This and other common misconceptions are probably the result of what Fienstein terms the Ôtrohoc fallacyÕ – the fallacious thinking that case–control studies are simply cohort studies performed backwards. do not reap the same benefit from randomization that a large. the possible direction of the bias will have to be evaluated. To reiterate. However. randomization addresses the issue of both measured and unmeasured confounding at the design stage by creating a balance of those factors across exposure groups. a case–control study could be thought of as a cohort study that is simply missing some exposure data at random (78). 77). whether the cohort has been enumerated or not. in large studies. 77). but improves the statistical efficiency of confounder control (10. Most basically. Whether a randomized study can lead to more valid causal inferences than non-randomized studies depends on study size. albeit most likely towards the null value of the effect measure. In addition to the vulnerabilities for biased exposure and disease information that are common to both cohort and case–control studies. Disease or outcome classification may also be biased if subjects or investigators become unblinded over the study period. if non-compliance is biased. 59. Additionally. these vulnerabilities are an issue of study conduct. unjustified matching can lead to losses in study efficiency (26.Analytic epidemiology and periodontal diseases The feature of randomization provides benefit from a design viewpoint beyond that of general cohort studies. Cluster randomized or community trials. not study type. exposure data are derived prior to or independent of the outcome). biased loss to follow-up may occur over the study period. However. a case–control study performed within that source would also be inherently prospective. Although randomized studies enjoy the added benefit of an expected balance of covariates at study baseline. 115 . case–control studies can be performed in such a way that they produce effect estimates that are no different than those obtained from a cohort study conducted within the same study base (40. except for the formal use of randomization. the maintenance of randomization through compliance. blinding and minimized loss to follow-up ⁄ patient drop-out (33). 71. 56. 64). 30). Extensions of this problem include over-emphasized vulnerability to biased exposure information and subject selection. In fact. but effect estimates will still be influenced in the face of noncompliance (exposure mis-classification). The balance that randomization provides can also add greater precision for estimating small effects. In fact. Principles of valid control selection have received widespread attention in the methodology literature (12. As in cohort studies. members of the same source population are sampled in a representative manner to provide an estimate of the exposure distribution (controls). individually randomized trial would. randomization lends meaning to the inferential statistics used to evaluate random error in the effect estimate derived from a study (19). 50. For example.g. concerns relating to validity in a case–control study primarily hinge on selection of controls. despite compliance. Analytic approaches that maintain initial randomization (intention to treat) may address the issue of confounding. 66. A sound understanding of the case–control design reinforces conceptualization of the case–control study as simply a more efficient version of a cohort study performed in the same source population (50. 32.

g. (5) recently conducted a study of the effect of diabetes on periodontal disease status. It should be noted. they must be selected in such a way that they are fully representative of that population. the observed odds ratio will therefore estimate the rate ratio (15). 55). if controls are selected from among all subjects at baseline or the Ôat riskÕ population. individuals may be selected as controls who later develop disease during follow-up. The continued incorrect use of case ⁄ control terminology and inappropriate conduct of studies perpetuates this problem. the risk or rate ratio) (40. cohort studies are particularly useful for studying multiple outcomes of a single exposure and case–control studies are particularly useful for studying multiple exposures of a single outcome. Periodontal disease status was then determined for the enrolled diabetic and non-diabetic subjects. If cases of disease are selected from existing cases (prevalent). we aim to measure the effect of an exposure on disease by comparing the disease incidence among those with the exposure level or condition of interest to that among those with the referent level or condition. Note that in both of the previous scenarios. classic misconceptions about the case–control design have been perpetuated in the periodontal literature (39). however.e. they are often in direct opposition from a utility standpoint. 80). The bulk of purported case–control studies of periodontal disease etiology evaluate gene polymorphisms. In this case. the observed estimate of effect will be the true odds ratio. It is only under this scenario that the rare disease assumption becomes meaningful for approximation of the risk ratio. latency periods are long. To do so. It is important to recall that. this understanding assists in determining what exposure distribution the control group represents. When investigators focus on the selection of ÔhealthyÕ controls. This particular approach is sometimes referred to as a case–cohort design. the Ôtrohoc fallacyÕ refers to the fallacious view that case–control studies are simply cohort studies performed backwards. This mistake is often committed when an investigator is most familiar with studies performed in a controlled environment and is therefore accustomed to the term ÔcontrolÕ referring to unexposed subjects (9). A convenience sample of studies designed to evaluate the effect of pro-inflammatory cytokines (e. and therefore the estimate of effect is the prevalence odds ratio. and the observed odds ratio will therefore approximate the risk ratio with no need for a rare disease assumption (29). hospital-based. Similarly. The sampling of controls is responsible for the gains in efficiency over a cohort study. Although case–control studies and cohort studies are similar from a validity standpoint. To briefly illustrate. The study investigators inappropriately termed their study a case–control study despite the fact that case–control design principles were not employed. etc. Case– control studies may be the only useful alternative to a cohort study when diseases are rare. Similar errors are evident in a more recent study of metabolic control and oral health (4). cohort studies are the only practical approach when the exposure of inter- 116 est is sufficiently rare. hence the name. This remains regardless of the source of controls. The study consisted of selecting diabetic ÔcaseÕ and non-diabetic ÔcontrolÕ subjects to serve as the exposed and unexposed populations. the observed odds ratio will be the prevalence odds ratio. and. Understanding that case–control studies are always theoretically nested within a cohort is useful in avoiding bias in control selection. A comprehensive review of the literature on this topic was recently published (44). If controls are selected from among non-diseased subjects at the theoretical end of follow-up. If controls are selected from among the underlying cohort at the time a case develops (risk-set ⁄ density sampling). Assuming that the underlying cohort is fixed and cases represent incident disease. All studies described here evaluated prevalent periodontal disease.g. i.Heaton & Dietrich 79. by extension. Campus et al. they will represent the distribution of exposure among the total population at risk at study baseline (denominator of the risk). (80). interleukin-1) on periodontitis is used to highlight four additional common mistakes in the conduct of case–control studies. with any analytic study. the underlying fallacy is the . By extension. population-based. Healthy controls As stated previously. and ⁄ or a study base has not previously been enumerated. The most basic and minimally consequential problem is the use of case ⁄ control terminology to refer to exposed and unexposed subjects. Furthermore. what the observed effect estimate will approximate (e. that case–control studies of genetic exposures have some additional unique considerations (38). Common problems with study conduct Despite the wealth of literature to the contrary. The sole purpose of controls is to represent the exposure distribution in the study base containing the cases. the controls will represent the distribution of exposed person–time (denominator of the rate).

Further. In doing so. In most studies. controls should be healthy. 82). many investigators confuse the case–control design with a cross-sectional design because of inappropriate use of case ⁄ control terminology (41. 76. Incorrect use of case ⁄ control terminology in a cross-sectional design Finally. the effect of exposure on disease occurrence is not assessed. regardless of variations on the design. All identified cases from the underlying source were recruited for participation. In contrast to those studies described above. 81. subjects were then classified as periodontal disease cases and non-cases. 43. However. however. The emphasis on ÔhealthyÕ controls violates principles of valid control selection and validity is compromised (57). depending on their conduct. the criteria for selection of controls should simply be that they (1) did not meet the case criterion. Studies using this approach identified a control group who were deemed ÔhealthyÕ by several systemic criteria that were not applied to the case group (17. Samples for all enrolled subjects were evaluated for genetic markers. Case ⁄ control criteria The counterfactual basis for study design should remind investigators that the causal contrast of interest is always that of exposed individuals versus unexposed individuals. More importantly. The case ⁄ control terminology was incorrectly applied to cases and non-cases. In most cases. case–control studies and cross-sectional studies may not differ much in terms of validity. 53. Furthermore. 69). and (2) were members of the same population as the cases. 54. One control for every case was then randomly selected from among the remaining non-cases and recruited for participation in the study. 70. several studies compared exposure frequencies among cases and controls (17. All three studies were conducted in the same source population: participants in the National Survey of Adult Oral Health in Australia (74). 73). 58. Great care should be taken to ensure that the comparison of interest is between individuals who were highly exposed and those who are truly unexposed – a task that is most often undertaken in cohort studies. and DNA samples were collected for exposure ascertainment (IL-1). particularly if case–control studies use prevalent cases and the temporal association between exposure and disease can be established. cross-sectional studies cannot provide effect estimates that approximate the risk or rate ratio. Instead of comparing case ⁄ control ratios across levels of the exposure. Cross-sectional studies.Analytic epidemiology and periodontal diseases unjustified reasoning that if cases are diseased. and analysis of the whole source population for several exposures would be financially prohibitive. can never distinguish between incidence and natural history of disease. 73. 47. i. 53. the strength of the evidence obtained from cross-sectional studies has to be considered in context. in reality. cross-sectional studies are often inferior to the cohort and case–control designs due to issues of temporality and dependence on prevalent cases. Instead of the control criterion simply being non-cases of periodontal disease – in other words. the principle that controls should be comparable to the population that gave rise to the cases was violated. 70. the effect on the odds ratio in this situation is unpredictable. a sample of three recent studies on this topic represents adequately conducted case–control studies that also highlight the utility of the case–control design (13. While the test of statistical significance is unharmed. cross-sectional studies cannot distinguish between 117 . a convenient cross-section of a clinic population was identified. 54. Great detail was often provided regarding control criteria when. 14. analysis of gingival crevicular fluid samples is expensive. 82). Cross-sectional studies From a validity standpoint. 58. the Ôtrohoc fallacyÕ can also manifest in the analysis of a case–control study. 81). the same inclusion and exclusion criteria at study outset. individuals not meeting the case definition – more stringent criteria were placed on controls in several studies to ensure that they were truly non-diseased or periodontally ÔhealthyÕ (43. Although the cross-sectional design may be considered inferior from a validity standpoint due to its inherent limitations for establishing temporality and dependence on prevalent disease. Comparing exposure frequencies among cases and controls In addition to backwards thinking in the conduct of a case–control study. despite misconceptions perpetuated in the literature (39). 49.e. The conduct of these studies also highlights the utility of case–control studies. 76. Another manifestation of the Ôtrohoc fallacyÕ occurs when investigators are instead concerned with the contrast between cases and controls (56). First. 49. the same case–control study population was used for all three studies to evaluate several biomarkers and even a behavioral exposure.

Modern epidemiology. Int J Aging Hum Dev 1972: 3: 125–137. Campus G. Vokonas PS. 30. J Chronic Dis 1985: 38: 551–552. Rosner B. Fitzsimmons TR. Gregio AMT. Greenland S. Circulation 2008: 117: 1668–1674. Uzzau S. Slade GD. References 1. Borrell LN. 27. 23. The epidemiologic trohoc. 9. exchangeability. the ablative risk ratio. 13. Greenland S. Greenland S. Glymour MM. 16. Palesch Y. Azevedo-Alanis LR. Greenland S. 20. 34. Local and systemic biomarkers in gingival crevicular fluid increase odds of periodontitis. Garcia RI. An introduction to instrumental variables for epidemiologists. Sanders JJ. Feinstein AR. 7. Philadelphia: Lipponcott Williams & Wilkins. Greenland S. Considerations in determining matching criteria and stratum sizes for case–control studies. The case–control study: valid selection of subjects. Bittencourt MS. Robins JM. Int J Epidemiol 1996: 25: 1107–1116. Greenland S. Slade GD. 118 15. Belanger CF. Greenland S. Stampfer M. Epidemiology 1990: 1: 421–429. 8. In: Rothman KJ. Methodology and preliminary findings. Aust Dent J 2009: 54: 115–122. Galbraith GM. 33. 25. The effects of sensitivity and specificity of case selection on validity. J Epidemiol Community Health 2004: 58: 265–8271. XX. 19. Hernan MA. Gore EA. Greenland S. Krall EA. Cochran WG. and power in hospital-based case–control studies. Instruments for causal inference: an epidemiologistÕs dream? Epidemiology 2006: 17: 360–372. Philadelphia: Lippincott Williams & Wilkins. when causal variables under study are immutable characteristics. Greenland S. sample size. J R Statist Soc A 1965: 128: 234–266. 11. Greenland S. Hill AB. Hennessy S. 2008: 451–455. J Clin Periodontol 1998: 25: 781–785. Berlin JA. Krall Kaye EA. Feinstein AR. Periodontitis and incidence of cerebrovascular disease in men. Flanders WD. Ann Neurol 2009: 66: 505–512. precision. Garcia RI. 10. 6. 24. where temporal sequence is not an issue (2). 26. Association between metabolic control and oral health in adolescents with type 1 diabetes mellitus. J Periodontol 2005: 76: 418–425. 29. 2008: 345–380. Clinical biostatistics. Randomization. Robins JM. Thomas DC. Kapur KK. Dupont WD. Am J Epidemiol 1990: 131: 151–159. Bilker WB. statistics. 3. editors. Biomarkers of periodontal inflammation in the Australian adult population. However. A study of the aetiology of carcinoma of the lung. Machado MAN. Associations between periodontal disease and systemic disease: evaluating the strength of the evidence. Thomas DC. Modeling longitudinal data. Morgenstern H. Jimenez M. editors. Greenland S. Grobbee DE. Matching and efficiency in cohort studies. Int J Epidemiol 2000: 29: 722–729. N Engl J Med 1990: 323: 1026–1032. Crosssectional studies have been utilized to a great extent to study possible predictors of periodontal disease status. In: Rothman KJ. Giovannucci E. BMJ 1952: 2: 1271–1286. Bartold PM. 2. Bias analysis. Modern epidemiology. Dietrich T. Coffee. Power calculations for matched case–control studies. Speizer FE. 35. Lash TL. 28. Age-dependent associations between chronic periodontitis ⁄ edentulism and risk of coronary heart disease. Garcia R. 5. Greenland S. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2010: 109: e51–e56. A definition of causal effect for epidemiological research. and ÔretrospectiveÕ research. Willett W. Am J Epidemiol 1977: 106: 184–187. Hennekens CH. such as gender. Colditz G. Clin Pharmacol Ther 1973: 14: 291–307. 18. Diabetes and periodontal disease: a case–control study. Biometrics 1988: 44: 1157–1168. Lash TL. 36. Factors influencing the optimal control-to-case ratio in matched case– control studies. caffeine. Salem A. Savitz DA. Rimm EB. J Periodontol 2005: 76: 2175–2184. editors. Bartold PM. J Clin Periodontol 2010: 37: 30–36. 32. and cardiovascular disease in men. Morgenstern H. Strom BL. Doll R. 22. Vokonas PS. Am J Epidemiol 1986: 124: 869–883. Greenland S. Morgenstern H. Analytical epidemiology of periodontitis. Am J Epidemiol 1999: 149: 195–197. Modern epidemiology.Heaton & Dietrich temporal cause and effect. Identifiability. Stat Sci 1999: 14: 29–46. 17. The nursesÕ health study. race ⁄ ethnicity and genetic polymorphisms. Papapanou PN. Am J Epidemiol 1990: 132: 181–192. Busato IMS. Philadelphia: Lippincott Williams & Wilkins. Interleukin-1b+3953 allele 2: association with disease status in adult periodontitis. Response and follow-up bias in cohort studies. Robins JM. Stat Med 1991: 10: 739–747. Basic methods for sensitivity analysis of biases. Pearl J. Int J Epidemiol 1986: 15: 413–419. Int J Epidemiol 1981: 10: 389–392. Fitzsimmons TR. 21. Thomas DC. 2008: 202–204. and causal inference. The rare-disease assumption revisited. Confounding and collapsibility in causal inference. Sensitivity analysis. Jimenez M. . Greenland S. Lash TL. Greenland S. The planning of observational studies of human populations. Instrumental variables. Brenner H. Am J Epidemiol 1982: 116: 547–553. Baldoni E. and Bayesian uncertainty assessment. Tonolo G. Sanders AE. Risk Anal 2001: 21: 579–583. such as genetic determinants of disease. Greenland S. Dietrich T. Pandey JP. Chambers SP. On the need for the rare disease assumption in case–control studies. Greenland S. Lash TL. 14. the crosssectional design can be a powerful alternative to other designs to generate hypotheses that can later be tested using designs that can distinguish between the incidence and natural history of disease. A critique of Ôestimators of relative risk for case–control studiesÕ. 4. The veterans administration longitudinal study of oral health. In: Rothman KJ. 12. and epidemiological confounding. Analytic methods for two-stage case–control studies and other stratified designs. Greenland S. J Clin Periodontol 2005: 6: 132–158. Sanders AE. Dietrich T. Monte Carlo risk analysis. 31. Hernan MA. Am J Nurs 1978: 78: 1039–1040.

Chavers LS. Causal inference from complex longitudinal data. Lash TL. Interleukin-1 genetic association with periodontitis in clinical practice. Spencer AJ. 1997: 69–117. Rubin DB. J Dent Res 2001: 80: 1695–1699. J Periodontol 2010: 81: 62–69. Lopez NJ. 72. 65. Joseph R. Roberts-Thomson KF. J Clin Periodontol 2000: 27: 682– 689. Neiderud AM. Khoury MJ. 2009. Kupper LL. Latent variable modeling and applications to causality. Robins JM. Methodological aspects of epidemiological studies of periodontal diseases. Howard G. Millikan R. 42. 54. Applying quantitative bias analysis to epidemiologic data. J Clin Periodontol 2001: 28: 389–396. Controls who experienced hypothetical causal intermediates should not be excluded from case–control studies. The Ôcase–controlÕ study: valid selection of subjects. Pearce N. Little RJ. Rubin DB. 48. -511. Farre MA. Papapanou PN. Borzani I. McDevitt MJ. Association of single nucleotide gene polymorphism at interleukin-1b +3954. Laine ML. Fink AK. 44. Kaye EK. Wilson TG Jr. design. Miettinen OS. Winkel EG. Scapoli C. 57. Am J Epidemiol 2008: 168: 1073–1081. 1985. Straus SE. Banerjee M. In: Rothman KJ. Rothman KJ. Estimating causal effects of treatments in randomized and nonrandomized studies. Loos BG. J Am Geriatr Soc 2010: 58: 713–718. Kleinbaum DG. Newman MG. In: Berkane M. Taylor JJ. Int J Epidemiol 2002: 31: 422–429. Tooth loss and periodontal disease predict poor cognitive function in older men. 69. Rubin DB. 2005. Baba N. Duff GW. Wang HY. The black womenÕs health study: a follow-up study for causes and preventions of illness. Modern epidemiology. Kingman A. Annu Rev Public Health 2000: 21: 121–145. Association of interleukin-1 gene polymorphisms with early-onset periodontitis. inflammatory biomarkers in gingival crevicular fluid and periodontitis.Analytic epidemiology and periodontal diseases 37. Shelton BJ. Palmer JR. J Educ Psychol 1974: 66: 688–701. 63. Guida L. Valid selection of subjects in case–control studies. 40. Guarnelli ME. and -31 in chronic periodontitis and aggressive periodontitis in Dravidian ethnicity. Poole C. Polymorphisms of the interleukin-1 gene family. Crane A. Hennig BJ. J Dent Res 2010: 89: 457–461. van Dijk LJ. 51. Parkhill JM. design trumps analysis. Maldonado G. oral microbial pathogens. For objective causal inference. Modern epidemiology. Glasziou P. 46. and background knowledge in etiologic inference. A case–control study. Estimating causal effects. J Clin Epidemiol 1989: 42: 491–496. Karon JM. Greenland S. Ann Appl Stat 2008: 2: 808–840. Lash TL (editors). Gwinn M. di Giovine FS. 119 . Case–control studies. Sanders AE. The central role of the propensity score in observational studies for causal effects. What does the odds ratio estimate in a case– control study? Int J Epidemiol 1993: 22: 1189–1192. 45. The interleukin-1 genotype as a severity factor in adult periodontal disease. Lu Z. Schlesselman JJ. 74. 62. Pirk FW. Scott P. Richardson WS. Valenzuela CY. Greenland S. Causal effects in clinical and epidemiological studies via potential outcomes: concepts and analytical approaches. 75. Valencia A. 60. Philadelphia: Churchill-Livingstone. Annunziata M. 66. Gonzalez G. Greenland S. J Clin Periodontol 1997: 24: 72–77. Rothman KJ. Physical activity. 50. 49. Vijayan NN. Interleukin-1 gene polymorphism and periodontal status. Modern epidemiology. 59. Rothman KJ. Theoretical epidemiology. Heasman PA. Kornman KS. di Giovine FS. Association of interleukin-1 polymorphisms with aggressive periodontitis. Fitzsimmons TR. Jara L. Control selection. Lopez NJ. 55. Philadelphia: Lipponcott Williams & Wilkins. Gilbert GH. 73. Shete AR. Comparing longitudinal binary outcomes in an observational oral health study. 2008: 111–127. Spiro A III. Ham AJ. Trombelli L. J Chronic Dis 1985: 38: 549–550. 39. editors. Knol MJ. Duff GW. Evidence-based medicine: how to practice and teach it. 2008: 179–181. 64. Biometrics 1981: 37: 271–291. Vandenbroucke JP. Stat Med 2003: 22: 2057– 2070. Greenland S. Am J Epidemiol 1999: 150: 547–551. 68. IL-1 gene cluster is not linked to aggressive periodontitis. Lewis DK. New York: Wiley. Lash TL (editors). Lash TL (editors). 58. Gene polymorphisms in chronic periodontitis. Bradshaw P. Genetic and molecular epidemiology. Dahlen G. Garcia RI. 43. Robins J. 2008: 564–579. Fox MP. Timms J. J Chronic Dis 1985: 38: 543–548. J Periodontol 2004: 75: 1509–1515. editor. Crusius JB. Crielaard W. Newman MG. Overmatching. New York: Springer. Vandenbroucke JP. J Periodontol 2005: 76: 234–243. Albandar JM. Knobelman C. Slade GD. Poole C. 52. 47. Exposure opportunity in case–control studies. Mamolini E. Biometrika 1983: 70: 41–55. Quappe L. Association of interleukin1 polymorphisms with periodontal disease. Kornman KS. Periodontol 2000 2002: 29: 11–30. Am J Epidemiol 1986: 123: 352–358. Adams-Campbell L. Egger M. Pena AS. Chapple IL. Robins JM. Matching in epidemiologic studies: validity and efficiency considerations. Rubin DB. van Winkelhoff AJ. Rosenbaum PR. Int J Dent 2010: 2010: 324719. Wang HY. The validity of case–control studies with nonrandom selection of controls. Lash TL. Morgenstern H. Pike M. Epidemiology 1990: 1: 273–284. Miettinen OS. Laine ML. The clinical trial as a paradigm for epidemiologic research. Srinivas L. J Periodontol 2000: 71: 156–163. Greenland S. Slade GD. New York: Springer Verlag. 71. What do case–control studies estimate? Survey of methods and assumptions in published case–control research. 70. Epidemiology 2001: 12: 313–320. Philadelphia: Lippincott Williams & Wilkins. 56. Bartold PM. J Clin Periodontol 2009: 36: 388–395. Rosenberg L. AustraliaÕs dental generations: the National Survey of Adult Oral Health 2004–2006. 67. 2008: 115–116. Haynes RB. Dietrich T. Miettinen OS. Philadelphia: Lippoincott Williams & Wilkins. 53. 41. J Am Med Womens Assoc 1995: 50: 56–58. and smoking in adult periodontitis. 38. Data. Jara L. 61. 2007. Canberra: Australian Institute of Health and Welfare. Higginbottom FL. Philadelphia: Lippincott Williams & Wilkins. Sandros J. Modern epidemiology.

Garibog˘lu S. Silverman DT. I. 120 80. Orima K. Selection of controls in case–control studies. Gou E. Otlu H. 77. Yu interleukin-1b. Hiller KA. J Chronic Dis 1983: 36: 685–697. Shimada Y. 78. Epidemiology 1995: 7: 144–150. Selection of controls in case–control studies. Thomas DC. Silverman DT. J Clin Periodontol 2007: 34: 823–827. The relative efficiencies of matched and independent sample designs for case–control studies. 79. 82. Types of controls. Christgau M. Schmitz G. Kobayashi T. II. Interleukin-11. ¨ cel OO. Moder D. Berker E. J Clin Periodontol 2002: 29: 882–888. The case–control study as data missing by design: estimating risk differences. Am J Epidemiol 1992: 135: 1029–1041. Schmalz G. Endo M. Principles. Kaminski WE. Yamazaki K. . interleukin-12 and the pathogenesis of inflammatory periodontal diseases. Wacholder S. Wacholder S. Aslanidis C. Wagner J. McLaughlin JK. Yoshie H. Tai H. Prevalence of OPG and IL-1 gene polymorphisms in chronic periodontitis. 81. Wacholder S. J Clin Periodontol 2008: 35: 365–370.Heaton & Dietrich 76. Greenland S. Mandel JS. Mandel JS. Am J Epidemiol 1992: 135: 1019–1028. McLaughlin JK. Association of interleukin-1 receptor antagonist gene polymorphisms with early onset periodontitis in Japanese.

. or email articles for individual use. download. users may print. However.Copyright of Periodontology 2000 is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission.