Article Text
Abstract
Introduction: Psychosocial stress and allostatic load have been postulated as a mechanism explaining socioeconomic inequalities in general and oral health. This study tested whether markers of allostatic load are associated with both ischaemic heart disease and periodontal disease and whether they affect education and income gradients for both conditions.
Methods: Data are from the Third National Health and Nutrition Examination Survey, conducted in the United States from 1988 to 1994. Ischaemic heart disease was determined by the presence of angina or diagnosis of heart attack. Four variables were used for periodontal disease. Individual and aggregate markers of allostatic load were used.
Results: Allostatic load (both aggregate and most individual markers) was associated with higher probabilities of all examined health outcomes. Adjusting for markers of allostatic load attenuated education and income gradients in both ischaemic heart disease and periodontal disease. The relationship between socioeconomic position and the examined health outcomes remained significant.
Conclusion: Indicators of allostatic load were associated with ischaemic heart disease and periodontal disease and had a mediating effect partly explaining the social gradients in both diseases. The results suggest a possible common stress pathway linking socioeconomic position to both conditions.
Statistics from Altmetric.com
A number of chronic conditions are correlated with social status. People at higher positions in the socioeconomic hierarchy live longer and have a lower burden of disease than those of lower status.1 2 These social gradients apply for general as well as oral health.1–6 An important aspect of the gradient across social strata is that the gradient is linked to hierarchy rather than deprivation alone,1 2 highlighting the importance of not only absolute but also relative socioeconomic position.1–5
A range of different plausible mechanisms has been suggested to explain the gradients, including material, behavioural, environmental, and psychosocial. Marmot,2 Brunner and Marmot3 and Brunner7 have presented a model that has focussed in particular on a psychosocial pathway highlighting the effect of stress on the social gradient in health. A probable pathophysiological mechanism of the effect of stress on health is via allostatic load; the ability to adapt successfully to changing environments. When external and internal stress challenges are chronic and frequently beyond the normal ranges of adaptive responses, “wear and tear” on regulatory systems occurs and allostatic load accumulates.8 9 Therefore, allostatic load is a marker of stress, influenced by the number and severity of stressful events and the individual’s ability to cope efficiently with these events.7 10 It is usually measured through different physiological indicators linked to chronic heart disease.11–13
In addition to directly influencing allostatic load, psychosocial stress affects general3 14 and oral health, particularly periodontal disease.15–19 Although it is not surprising that chronic heart disease markers are associated cross-sectionally with ischaemic heart disease, their potential association with periodontal disease may be through their function as indicators of stress. Higher levels of stress indicators, such as C-reactive protein (CRP) and fibrinogen, are associated with worse periodontal health.20–22 Stress could also have an indirect effect on periodontal health via health-related behaviours.7 23
Psychosocial stress has a mediating effect on the social gradients in health, thereby affecting health inequalities.2 3 8 24 Therefore, unsurprisingly, allostatic load is related to the social gradient in general health.7 11 12 As most chronic diseases, including periodontal disease, have common determinants,25 a stress pathway has also been suggested for the social gradient in periodontal disease.19 We postulate a potential allostatic pathway, whereby psychosocial stress may partly explain the social gradients in periodontal disease in the same manner as it does in ischaemic heart disease.7
Whereas studies have investigated the role of allostatic load in the social gradient in general health,7 no study has focused on the assessment of allostatic load in both general and oral health in the same individuals. Such an approach would demonstrate potential similarities in oral and general health determinants. More importantly, none of the studies linking psychosocial stress and periodontal disease have examined the effects of markers of allostatic load on the association between socioeconomic position and periodontitis. Therefore, the objectives of this study on a national representative sample of adults in the United States were to examine the association of both ischaemic heart disease and periodontal disease with allostatic load markers in the same population and to assess the effect of allostatic load markers on the education and income gradients in ischaemic heart disease and periodontitis.
METHODS
Data for this research are from the Third National Health and Nutrition Examination Survey (NHANES III) conducted in 1988–1994.26 NHANES used a stratified multistage probability sampling design representative of the non-institutionalised civilian American population. Data on the adult population aged 17 years and older were used. The survey included a comprehensive medical and dental examination and questionnaire.
Ischaemic heart disease was determined on participants having angina according to the WHO Rose questionnaire27 or reporting a diagnosis of heart attack. Periodontal measures were assessed on randomly assigned half-mouths, one upper and one lower quadrant using the National Institute of Dental Research protocol.28 Three variables, previously used in NHANES studies29 30 were created to indicate the extent of periodontal disease. They were extent of periodontal pockets of 4 mm or greater, extent of loss of periodontal attachment of 3 mm or greater and extent of gingival bleeding. They were calculated as the ratio of tooth sites with the periodontal characteristic to the total number of tooth sites examined. In addition, a binary variable on periodontitis31 was also used (individuals having at least one gingival bleeding site and one site with loss of periodontal attachment of 3 mm or greater). The use of these four periodontal variables, indicating the severity and presence of different manifestations of periodontitis, ensures that the results are not coincidental and limits bias resulting from using a single variable measuring one aspect of periodontitis.
Different markers of allostatic load have been proposed.11–13 We selected seven of those markers as they have been reported to be associated with periodontitis.20–22 32 They are: central obesity (waist circumference >102 cm for men and >88 cm for women), high blood pressure (systolic ⩾130 mm Hg, or diastolic ⩾85 mm Hg), hypertriglyceridemia (triglycerides ⩾150 mg/dl), low high-density lipoprotein (HDL)-cholesterol (<40 mg/dl for men and <50 mg/dl for women), high plasma glucose (⩾110 mg/dl), CRP and fibrinogen. CRP and fibrinogen were used both as continuous and dichotomous (⩾10 mg/l for CRP and ⩾3.25 g/l for fibrinogen) variables. These cut-off points were used as indicators of disease level in other studies.22 32 33 In addition, a clustered variable was created including the seven aforementioned dichotomous indicators, thus providing a continuous aggregate indicator of allostasis (zero to seven).
The poverty : income ratio and years of education were used as indicators of socioeconomic position. The poverty : income ratio was computed as the ratio between family income and the poverty threshold in the calendar year of the interview to account for inflation during the six years of the survey.26 Education was categorised according to years spent in education: less than 12 years of education, 12 years (reflecting high school completion), and more than 12 years (tertiary education). Both education and income are considered adequate measures of socioeconomic position in US populations.2 34 35 Other covariates adjusted for in the regression models included ethnicity, age, sex, medical insurance, dental insurance, and smoking status (current smoker, non-smoker, non-respondent).
Data analysis
Analysis was conducted using survey command in Stata (Stata Corp, College Station, Texas, USA). Final sampling weights were used throughout. First, the associations of allostatic load markers (seven individual markers and the aggregate clustered variable) with ischaemic heart disease and periodontal disease were assessed, both unadjusted and adjusted for the effect of the covariates mentioned above.
Then we assessed the effect of adjusting for allostasis on the social gradients in general and oral health. For this, we first controlled the associations between socioeconomic position and health for the effect of the aggregate allostatic load variable. Second, we constructed regression models whereby the association between each health outcome with each social measure (education and income) was adjusted for covariates other than allostatic load. For ischaemic heart disease logistic regression was used controlling for income, education, ethnicity, sex, age, smoking and medical insurance. Linear regression models were constructed for each of the periodontal disease extent outcomes and logistic regression for the dichotomous periodontal variable, adjusting for the same variables but replacing medical insurance by dental insurance. Then the aggregate variable of allostasis was additionally introduced into the models to assess its adjusted effect on the variation in ischaemic heart disease and periodontal disease. Analysis for each outcome was conducted for the whole available sample for that outcome.
What is already known on this subject
Allostatic load is associated with heart disease
Allostatic load as a marker of stress mediates the social gradients in general health
What this study adds
Allostatic load is associated in the same way with periodontal disease and ischaemic heart disease
Allostatic load partly explains socioeconomic gradients in periodontal disease and ischaemic heart disease
Policy implications
The similarities of the pathways for the social gradients in oral and general health support the integration of oral and general health promotion strategies
RESULTS
Table 1 shows the association between the health outcomes and the seven indicators of allostasis, as well as the aggregate clustered variable of allostatic load. All seven markers and the aggregate variable were significantly associated with a higher probability of ischaemic heart disease, both in the unadjusted models and after adjusting for the effect of education, poverty : income ratio, age, sex, ethnicity, medical insurance and smoking. In relation to the dichotomous periodontal disease variable, people with higher levels of allostasis indicators had higher odds ratios for periodontal disease. The unadjusted and adjusted associations were significant for the aggregate variable, as well as for the biological markers of allostasis, with the exception of CRP and fibrinogen.
There was a significantly greater extent of gingival bleeding with all markers of allostasis and the aggregated allostasis variable in the unadjusted and adjusted models, with the exception of fibrinogen in the adjusted model. Similarly, a greater extent of loss of periodontal attachment was significantly associated with all markers of allostatic load in the unadjusted models and with fibrinogen, low HDL-cholesterol, glucose and the aggregated allostasis variable in the adjusted models. Finally, CRP, fibrinogen, glucose and the aggregate allostasis variable had a significant association with the extent of pocket depth in the unadjusted and adjusted models (table 1).
The crude effect of allostatic load on explaining the social gradient in ischaemic heart disease and periodontal disease is shown by the attenuation of the odds ratios for the education gradients of the aforementioned variable, after adjusting for the aggregate allostasis variable (fig 1). Participants in the middle and lowest education groups were 1.29 (95% CI 0.97 to 1.72) and 2.17 (95% CI 1.66 to 2.86), respectively, times more likely to have ischaemic heart disease than those in the highest education group. After adjusting only for the aggregated allostatic load variable, these odds ratios attenuated to 1.09 (95% CI 0.81 to 1.45) and 1.72 (95% CI 1.31 to 2.26), respectively. As for periodontal disease, individuals in the middle and lowest education groups, respectively, were significantly 1.47 (95% CI 1.23 to 1.76) and 2.38 (95% CI 1.96 to 2.89) times more likely to have periodontitis than those in the highest education group. Adjusting for allostatic load attenuated these odds ratios to 1.35 (95% CI 1.12 to 2.64) and 2.11 (95% CI 1.73 to 2.58), respectively (fig 1).
A similar crude attenuating effect of allostatic load was also shown in relation to the poverty : income ratio gradients of both ischaemic heart disease and periodontal disease. Participants with a higher income were significantly less likely to have ischaemic heart disease (OR 0.81; 95% CI 0.75 to 0.88) and periodontitis (OR 0.83; 95% CI 0.79 to 0.87) than those at the next lower income level. After adjusting just for allostasis, the respective effects were 0.84 (95% CI 0.78 to 0.91) and 0.84 (95% CI 0.81 to 0.89), respectively. Similar results were also observed for the education and poverty : income ratio gradients of periodontal extent variables, in which inclusion of the aggregate allostasis variable in the regression models resulted in the attenuation of regression coefficients.
The adjusted effects of allostatic load on the education and income gradients in ischaemic heart disease and periodontal diseases are shown in tables 2 and 3. Regression models adjusting for other relevant confounders but not for allostatic load were compared with the models that also controlled for the effects of the allostatic load. Generally, the social gradients attenuated for all health outcomes after additionally adjusting for allostasis, but the attenuation was modest for the two adjusted models pertaining to periodontitis and ischaemic heart disease (table 2). The probabilities of the greater extent of gingival bleeding, loss of periodontal attachment and pocket depth were significantly higher as education and income decreased. These probabilities attenuated after additionally adjusting for allostasis, but remained significant for all three indicators of the extent of periodontal disease (table 3).
DISCUSSION
This study examined the effects of allostatic markers on ischaemic heart disease and periodontal disease and assessed the possible stress pathways linking socioeconomic position to these two conditions. This is the first study to examine the relationship between allostatic load and oral health. Furthermore, no previous study has compared the effect of allostatic load on general and oral health in the same population.
We examined seven markers of allostatic load, measuring their individual and collective effects on health in a national sample of adult Americans. Some of the allostatic markers used here have also been used in previous studies on the association between coronary artery disease and oral health or have been reported at higher levels among people with periodontal diseases.20–22 32 None of the aforementioned studies, however, used the range of different periodontal disease measures employed in the current analysis. Neither did they address the markers of stress as indicators of allostatic load, nor assess their collective effects on periodontal diseases.
The aggregate allostatic load variable, indicating clustering of the different allostatic markers, was significantly associated with all examined health outcomes, both as a crude effect in unadjusted models and also after controlling for the effects of a number of covariates, such as age, ethnicity, income, education, medical/dental insurance and smoking. The findings clearly indicate the association of allostatic load with the presence and severity of both ischaemic heart disease and periodontal disease. Furthermore, the similarity of the statistical relationship may indicate that stress affects both chronic conditions through similar pathways.
When the aggregate clustered variable was substituted by each of the seven different markers of allostatic load, similar results were found. The separate analyses showed that higher levels of each of the different allostatic load indicators were significantly associated with greater levels of ischaemic heart disease and periodontal disease in most unadjusted and adjusted models, with the extensive adjustment being responsible for the loss of statistical significance in some of those relationships. These results confirm findings from other studies suggesting an effect of stress on heart disease7 11 14 and periodontitis.15–19
The associations between indicators of allostasis and general and oral health outcomes is not coincidental or a result of statistical confounding. These indicators are also markers of stress and it is partly in that capacity that they are associated with health outcomes, as stress has been linked to periodontitis and ischaemic heart disease.7 19 For example, there is no reasonable biological assumption that central obesity should be associated with periodontitis. In addition, some markers of allostasis are directly associated with specific health outcomes, which may explain differences in their associations with different heath measures. For example, apart from being a marker of stress, CRP is also a marker of inflammation and showed stronger association with gingival bleeding than with loss of attachment.
This study also examined the effects of adjusting for allostasis on the social gradients in ischaemic heart disease and periodontal diseases. The crude effect of allostasis indicated a decrease in the education and poverty : income ratio gradients for almost all health outcomes investigated. Even after taking into account a number of other relevant confounders, the additional inclusion of allostatic markers in the regression analyses led to a further attenuation of social gradients for most health outcomes examined. This suggests a role for allostatic load in partly explaining general and oral health inequalities and supports the theories of stress pathways linking the gradients to ischaemic heart disease and periodontitis.7 19 The comprehensive adjustment in the two models pertaining to ischaemic heart disease and periodontal disease, however, led to only modest changes in education and income gradients after adjusting for allostatic load. The similarity of the effects of allostatic load on the social gradients in ischaemic heart disease and periodontal disease supports the theories about the commonality of pathways between general and oral health determinants.25
Despite this effect of allostatic load on the social gradient, the relationship between socioeconomic position and both general and oral health remained significant for most of the outcomes, implying the contribution of other factors and mechanisms. Other researchers have argued that the determinants of the social gradients are complex.2 36 37 The exploration of the stress pathway between socioeconomic status and health status carried out in this study only partly explains the social gradients in health.
As a result of the cross-sectional nature of the study, our results cannot support a causal pathway. Other limitations of this analysis relate to the lack of other indicators of allostatic load in NHANES III, such as cortisol, norepinephrine, epinephrine, adrenaline and dehydroepiandrosterone. In addition, the cut-off points for the categorisation of allostatic markers used here were derived from previous studies on periodontal diseases20–22 but were different from those used in other general health studies.10 13 Future research using longitudinal data needs to address these issues further.
This analysis explored the individual and clustered effects of the markers of allostasis on different indicators of periodontal diseases and ischaemic heart disease. Higher levels of allostatic load were observed in individuals with higher levels of ischaemic heart disease and periodontal disease. Markers of allostatic load had a similar effect in both ischaemic heart disease and periodontitis and explained part of the education and income gradients in the aforementioned chronic diseases. This indicates the probable existence of a common pathway for both these conditions.
REFERENCES
Footnotes
Competing interests: None declared.
Linked Articles
- In this issue