Jan J. Barendregt

Obesity in adulthood and its consequences for life expectancy: a life-table analysis.

Context Middle-aged adults who are overweight or obese may have shorter life expectancies than normal-weight adults, but how much shorter? Contribution This analysis of data from the Framingham Heart Study from 1948 to 1990 showed that, on average, adults who were obese (body mass index [BMI] 30 kg/m2) at age 40 years lived 6 to 7 years less than their normal-weight counterparts. Adults who were overweight (BMI, 25 to 29.9 kg/m2) and did not smoke lived about 3 years less than normal-weight nonsmokers. Adults who were obese and smoked lived 13 to 14 years less than normal-weight nonsmokers. Cautions Descriptions of lost life expectancy do not necessarily predict length of life that could be gained from obesity prevention or treatment programs. The Editors The increasing prevalence of overweight and obesity, coupled with their associations with death, disability, and disease, has led to their identification as a major, potentially preventable cause of premature morbidity and death (1-9). However, it is difficult to estimate the public health impact of overweight and obesity because of complex interactions with age; smoking; and obesity-related risk factors, such as diabetes, hypertension, and lipid disorders (8, 10-12). The observed relationship between body mass index (BMI) and mortality has been described as J-shaped; mortality increases as a result of underweight, overweight, and obesity. However, preexisting illness and inadequate control of smoking may cause at least part of the increased mortality at very low weight (8). Consequently, there have been no robust estimates of life expectancy lost as a result of obesity. A primary reason is the lack of understanding of probable, healthy, or unhealthy weight trajectories over the life course. Conclusions regarding appropriate weight trajectories between adulthood and older age are complicated by uncertainties about age-appropriate measurements of obesity and the effects of smoking, obesity-associated risk factors for cardiovascular disease, and unintended weight loss (13, 14). We provide an estimate of the effect of obesity and overweight in adulthood on life expectancy, implicitly taking into account the various possible weight trajectories throughout the life course. We take advantage of the cohort follow-up made available by the Framingham Heart Study to analyze the differences in life course for various BMI groups. We make no assumptions about the relationship between BMI and mortality at older ages. Our primary objective was to analyze the reductions in life expectancy associated with overweight and obesity at 40 years of age. Methods Data Source The Framingham Heart Study is a longitudinal study with excellent follow-up on mortality. The original study cohort involved 5209 adults, age 28 through 62 years, residing in Framingham, Massachusetts, between 1948 and 1951 (15). To examine the effect of overweight and obesity in adulthood, we used the data from more than 40 years of follow-up (examinations 1 through 21) on age at death for persons 30 through 49 years of age at baseline (n = 3607). Height and weight were measured at baseline (7, 15). Smoking status at baseline was defined categorically as self-reported current smoker or nonsmoker. No information was available on smoking status before study entry. Information on all three variables was available for 3582 participants (99%). Because the relationship between weight and mortality is affected by underlying disease (8, 14, 16), we excluded participants who had cardiovascular disease (17) at baseline or died within 4 years of follow-up (63 participants, including 50 deaths). Because our analysis focused on the risk for death associated with overweight and obesity, we also excluded 62 underweight persons (BMI < 18.5 kg/m2). The final analyses involved 3457 participants (1550 men and 1907 women). We analyzed the effect of potential confounders on the relationship between obesity and mortality (6, 8, 18). Of the 3457 participants examined, hypertension and diabetes status was available for all participants, and physical activity level was available for 2893 (84%) participants. Total serum cholesterol level was available for 2127 (62%) participants and was therefore not taken into account. We defined hypertension at baseline as either systolic blood pressure of 160 mm Hg or greater or diastolic blood pressure of 95 mm Hg or greater in two repeated measurements. Physical activity (a continuous index derived from hours of activity and rest) was not available until examination 4 (approximately 8 years after baseline). Level of education at baseline was available for 3350 (97%) participants. Potential confounders were analyzed by using only complete cases. BMI Group Classification Body mass index at baseline was calculated as weight in kg/height in m2. We defined three BMI categories based on World Health Organization guidelines (2): group I (normal weight), BMI of 18.5 to 24.9 kg/m2; group II (overweight), BMI of 25 to 29.9 kg/m2; and group III (obese), BMI greater than or equal to 30 kg/m2 (including 19 people with BMI > 40 kg/m2). Survival Analysis We used S-Plus 2000 (MathSoft, Inc., Seattle, Washington) for all statistical analyses. Survival curves for each BMI group were compared by using KaplanMeier plots. We assessed the association between BMI group at baseline and mortality over the 40 years of follow-up by using Cox proportional-hazards analysis, with age as the time scale. The effect of BMI was analyzed separately within strata defined by sex and smoking status at baseline. We tested the proportionality of hazards assumption by analysis of the Schoenfeld residuals (19, 20). Statistical significance was set at the 5% level. Life Course Analysis Within each stratum, we estimated age-specific mortality rates for each BMI group by using Poisson regression analysis; age at follow-up and BMI group at baseline were categorical variables. Although the hazard ratios estimated for BMI group from this analysis are equivalent to those estimated from the Cox analyses, Poisson regression also optimizes the hazard associated with each age at follow-up. Life tables were derived for each BMI group, representing populations that were 40 years of age and free of cardiovascular disease at study entry. Conversions between mortality rates and probabilities assumed that within each single age interval, the hazard is constant. The life expectancy at 90 years of age was assumed to be a constant 4.53 for men and 5.05 for women for each BMI group (based on life expectancies of the total Framingham Study sample [21]). The main outcome measure, life expectancy at 40 years of age, was calculated as the mean age at death within a life-table population. Confidence intervals for the life table measures were calculated by using a bootstrap procedure, based on 10 000 replicates. We report the bootstrap biascorrect, adjusted 95% CIs (using the bias-corrected accelerated percentile interval algorithm) (22). Although computationally demanding, the bootstrap procedure is easier than an analytical alternative that includes both the variance of the Poisson model and the variance of the life table. Role of the Funding Source The Framingham Heart Study was conducted and supported by the National Heart, Lung, and Blood Institute (NHLBI) in collaboration with the Framingham Heart Study Investigators. The NHLBI reviewed this article for scientific content and consistency of data interpretation with previous Framingham Heart Study publications; significant comments were incorporated into the text before submission for publication. The NHLBI had no role in the design, conduct, analyses, and reporting of the study or in the decision to submit the manuscript for publication. The Netherlands Heart Foundation and the Netherlands Organization for Scientific Research funded our study. Neither had any role in the design, conduct, analyses and reporting of the study or in the decision to submit the manuscript for publication. Results The characteristics at baseline within the Framingham Study cohort were generally as expected: The probability of death increased with each higher category of BMI group, the relationship between the prevalence of smoking and BMI group was inverse (Table 1), and age generally increased with each higher category of BMI group (7, 8, 23). Although male nonsmokers were a small group and may represent an unusual cohort for that time, they were analyzed in the same way as the other groups. We did this because male nonsmokers had BMI-related risks similar to those of female nonsmokers and to findings in previous studies examining the relationship between BMI and mortality. Table 1. Characteristics of Original Framingham Heart Study Participants, Age 30 to 49 Years at Baseline (19481951) BMI and Survival With participants categorized by BMI at baseline, we used Cox proportional-hazards analysis to determine the relative rate of death over the 40 years of follow-up. We found that sex did not significantly modify the effect of BMI but that smoking status at baseline did, as has been previously described (10, 12, 24, 25). Additional analyses were performed separately for strata defined by sex and smoking status at baseline. The Figure illustrates the empirical survival curves for each BMI group within each of the four strata: female nonsmokers, female smokers, male nonsmokers, and male smokers. The survival disadvantage associated with BMI group II compared with BMI group I is apparently smaller in smokers than in nonsmokers. The hazard ratios for mortality associated with BMI group were generally consistent between strata, although neither male nor female smokers in BMI group II showed an increased mortality risk (Table 2). The proportional hazards assumption seemed appropriate for BMI, both by analysis of the Schoenfeld residuals and by comparison of the Cox- derived hazard ratios for two distinct follow-up periods with approximately

The global burden of low back pain: estimates from the Global Burden of Disease 2010 study

Objective To estimate the global burden of low back pain (LBP). Methods LBP was defined as pain in the area on the posterior aspect of the body from the lower margin of the twelfth ribs to the lower glutaeal folds with or without pain referred into one or both lower limbs that lasts for at least one day. Systematic reviews were performed of the prevalence, incidence, remission, duration, and mortality risk of LBP. Four levels of severity were identified for LBP with and without leg pain, each with their own disability weights. The disability weights were applied to prevalence values to derive the overall disability of LBP expressed as years lived with disability (YLDs). As there is no mortality from LBP, YLDs are the same as disability-adjusted life years (DALYs). Results Out of all 291 conditions studied in the Global Burden of Disease 2010 Study, LBP ranked highest in terms of disability (YLDs), and sixth in terms of overall burden (DALYs). The global point prevalence of LBP was 9.4% (95% CI 9.0 to 9.8). DALYs increased from 58.2 million (M) (95% CI 39.9M to 78.1M) in 1990 to 83.0M (95% CI 56.6M to 111.9M) in 2010. Prevalence and burden increased with age. Conclusions LBP causes more global disability than any other condition. With the ageing population, there is an urgent need for further research to better understand LBP across different settings.

Cost-Effectiveness of Interventions to Promote Physical Activity: A Modelling Study

Linda Cobiac and colleagues model the costs and health outcomes associated with interventions to improve physical activity in the population, and identify specific interventions that are likely to be cost-saving.

The health care costs of smoking

BACKGROUND Although smoking cessation is desirable from a public health perspective, its consequences with respect to health care costs are still debated. Smokers have more disease than nonsmokers, but nonsmokers live longer and can incur more health costs at advanced ages. We analyzed health care costs for smokers and nonsmokers and estimated the economic consequences of smoking cessation. METHODS We used three life tables to examine the effect of smoking on health care costs - one for a mixed population of smokers and nonsmokers, one for a population of smokers, and one for a population of nonsmokers. We also used a dynamic method to estimate the effects of smoking cessation on health care costs over time. RESULTS Health care costs for smokers at a given age are as much as 40 percent higher than those for nonsmokers, but in a population in which no one smoked the costs would be 7 percent higher among men and 4 percent higher among women than the costs in the current mixed population of smokers and nonsmokers. If all smokers quit, health care costs would be lower at first, but after 15 years they would become higher than at present. In the long term, complete smoking cessation would produce a net increase in health care costs, but it could still be seen as economically favorable under reasonable assumptions of discount rate and evaluation period. CONCLUSIONS If people stopped smoking, there would be a savings in health care costs, but only in the short term. Eventually, smoking cessation would lead to increased health care costs.

Estimating clinical morbidity due to ischemic heart disease and congestive heart failure : the future rise of heart failure

OBJECTIVES Many developed countries have seen declining mortality rates for heart disease, together with an alleged decline in incidence and a seemingly paradoxical increase in health care demands. This paper presents a model for forecasting the plausible evolution of heart disease morbidity. METHODS The simulation model combines data from different sources. It generates acute coronary event and mortality rates from published data on incidences, recurrences, and lethalities of different heart disease conditions and interventions. Forecasts are based on plausible scenarios for declining incidence and increasing survival. RESULTS Mortality is postponed more than incidence. Prevalence rates of morbidity will decrease among the young and middle-aged but increase among the elderly. As the milder disease states act as risk factors for the more severe states, effects will culminate in the most severe disease states with a disproportionate increase in older people. CONCLUSIONS Increasing health care needs in the face of declining mortality rates are no contradiction, but reflect a tradeoff of mortality for morbidity. The aging of the population will accentuate this morbidity increase.

Lifetime prevalence estimates of major depression: an indirect estimation method and a quantification of recall bias.

The measurement of lifetime prevalence of depression in cross-sectional surveys is biased by recall problems. We estimated it indirectly for two countries using modelling, and quantified the underestimation in the empirical estimate for one. A microsimulation model was used to generate population-based epidemiological measures of depression. We fitted the model to 1-and 12-month prevalence data from the Netherlands Mental Health Survey and Incidence Study (NEMESIS) and the Australian Adult Mental Health and Wellbeing Survey. The lowest proportion of cases ever having an episode in their life is 30% of men and 40% of women, for both countries. This corresponds to a lifetime prevalence of 20 and 30%, respectively, in a cross-sectional setting (aged 15–65). The NEMESIS data were 38% lower than these estimates. We conclude that modelling enabled us to estimate lifetime prevalence of depression indirectly. This method is useful in the absence of direct measurement, but also showed that direct estimates are underestimated by recall bias and by the cross-sectional setting.

By how much would limiting TV food advertising reduce childhood obesity

Background: There is evidence suggesting that food advertising causes childhood obesity. The strength of this effect is unclear. To inform decisions on whether to restrict advertising opportunities, we estimate how much of the childhood obesity prevalence is attributable to food advertising on television (TV). Methods: We constructed a mathematical simulation model to estimate the potential effects of reducing the exposure of 6- to 12-year-old US children to TV advertising for food on the prevalence of overweight and obesity. Model input was based on body measurements from NHANES 2003–04, the CDC-2000 cut-offs for weight categories, and literature that relates advertising to consumption levels and consumption to body mass. In an additional analysis we use a Delphi study to obtain experts’ estimates of the effect of advertising on consumption. Results: Based on literature findings, the model predicts that reducing the exposure to zero would decrease the average BMI by 0.38 kg/m−2 and lower the prevalence of obesity from 17.8 to 15.2% (95% uncertainty interval 14.8–15.6) for boys and from 15.9% to 13.5% (13.1–13.8) for girls. When estimates are based on expert opinion, these values are 11.0% (7.7–14.0) and 9.9% (7.2–12.4), respectively. Conclusion: This study suggests that from one in seven up to one in three obese children in the USA might not have been obese in the absence of advertising for unhealthy food on TV. Limiting the exposure of children to marketing of energy-dense food could be part of a broader effort to make childrens diets healthier.

Quantitative health impact assessment: current practice and future directions

Study objective: To assess what methods are used in quantitative health impact assessment (HIA), and to identify areas for future research and development. Design: HIA reports were assessed for (1) methods used to quantify effects of policy on determinants of health (exposure impact assessment) and (2) methods used to quantify health outcomes resulting from changes in exposure to determinants (outcome assessment). Main results: Of 98 prospective HIA studies, 17 reported quantitative estimates of change in exposure to determinants, and 16 gave quantified health outcomes. Eleven (categories of) determinants were quantified up to the level of health outcomes. Methods for exposure impact assessment were: estimation on the basis of routine data and measurements, and various kinds of modelling of traffic related and environmental factors, supplemented with experts’ estimates and author’s assumptions. Some studies used estimates from other documents pertaining to the policy. For the calculation of health outcomes, variants of epidemiological and toxicological risk assessment were used, in some cases in mathematical models. Conclusions: Quantification is comparatively rare in HIA. Methods are available in the areas of environmental health and, to a lesser extent, traffic accidents, infectious diseases, and behavioural factors. The methods are diverse and their reliability and validity are uncertain. Research and development in the following areas could benefit quantitative HIA: methods to quantify the effect of socioeconomic and behavioural determinants; user friendly simulation models; the use of summary measures of public health, expert opinion and scenario building; and empirical research into validity and reliability.

Estimating remission from untreated major depression: a systematic review and meta-analysis

BACKGROUND Few studies have examined spontaneous remission from major depression. This study investigated the proportion of prevalent cases of untreated major depression that will remit without treatment in a year, and whether remission rates vary by disorder severity. METHOD Wait-list controlled trials and observational cohort studies published up to 2010 with data describing remission from untreated depression at ≤ 2-year follow-up were identified. Remission was defined as rescinded diagnoses or below threshold scores on standardized symptom measures. Nineteen studies were included in a regression model predicting the probability of 12-month remission from untreated depression, using logit transformed remission proportion as the dependent variable. Covariates included age, gender, study type and diagnostic measure. RESULTS Wait-listed compared to primary-care samples, studies with longer follow-up duration and older adult compared to adult samples were associated with lower probability of remission. Child and adolescent samples were associated with higher probability of remission. Based on adult samples recruited from primary-care settings, the model estimated that 23% of prevalent cases of untreated depression will remit within 3 months, 32% within 6 months and 53% within 12 months. CONCLUSIONS It is undesirable to expect 100% treatment coverage for depression, given many will remit before access to services is feasible. Data were drawn from consenting wait-list and primary-care samples, which potentially over-represented mild-to-moderate cases of depression. Considering reported rates of spontaneous remission, a short untreated period seems defensible for this subpopulation, where judged appropriate by the clinician. Conclusions may not apply to individuals with more severe depression.

Preventing fatal diseases increases healthcare costs: cause elimination life table approach

Abstract Objectives: To examine whether elimination of fatal diseases will increase healthcare costs. Design: Mortality data from vital statistics combined with healthcare spending in a cause elimination life table. Costs were allocated to specific diseases through the various healthcare registers. Setting and subjects: The population of the Netherlands, 1988. Main outcome measures: Healthcare costs of a synthetic life table cohort, expressed as life time expected costs. Results: The life time expected healthcare costs for 1988 in the Netherlands were £56 600 for men and £80 900 for women. Elimination of fatal diseases—such as coronary heart disease, cancer, or chronic obstructive lung disease—increases health- care costs. Major savings will be achieved only by elimination of non-fatal disease—such as musculoskeletal diseases and mental disorders. Conclusion: The aim of prevention is to spare people from avoidable misery and death not to save money on the healthcare system. In countries with low mortality, elimination of fatal diseases by successful prevention increases healthcare spending because of the medical expenses during added life years. Key messages In countries with low mortality prevention of fatal diseases adds life years predominantly to old age, when disabling conditions are prevalent If fatal diseases are eliminated, the medical costs of life extension at old age will generally be higher than the costs prevented. Prevention of disabling conditions, particularly mental disorders and musculoskeletal conditions, might both lower healthcare costs and improve public health The aim of prevention is to save people from preventable morbidity and mortality not to save money For the time being, prevention of disability should have the highest priority for future research