Population Attributable Fraction Estimates Research Articles

Causal attribution of human papillomavirus genotypes to invasive cervical cancer worldwide: a systematic analysis of the global literature

Understanding the proportion of invasive cervical cancer (ICC) caused by different human papillomavirus (HPV) genotypes can inform primary (ie, vaccination) and secondary (ie, screening) prevention efforts that target specific HPV genotypes. However, using the global literature to estimate population attributable fractions (AFs) requires a methodological framework to address HPV genotype-specific causality from aggregated data. We aimed to estimate the proportion of ICC caused by different HPV genotypes at the global, regional, and national level. This systematic review identified studies reporting HPV genotype-specific prevalence in ICC or people with normal cervical cytology. We searched PubMed, Embase, Scopus, and Web of Science up to Feb 29, 2024, using the search terms "cervix" and "HPV", with no language restrictions. Odds ratios (ORs) were estimated by comparing HPV genotype-specific prevalence between HPV-positive ICC and normal cervical cytology with logistic regression models, adjusting for region, year of paper publication, and HPV primer or test. HPV genotypes with a lower bound to the 95% CI of the OR greater than 1·0 were judged as causal to ICC. Corresponding regional genotype-specific AFs were calculated as regional HPV prevalence in ICC multiplied by (1 - [1 / OR]) and were proportionally adjusted to total 100%. Global AFs were calculated from regional AFs weighted by number of regional ICC cases in 2022 (GLOBOCAN). The systematic review identified 1174 studies with 111 902 cases of HPV-positive ICC and 2 755 734 of normal cervical cytology. 17 HPV genotypes were considered causal to ICC, with ORs ranging widely from 48·3 (95% CI 45·7-50·9) for HPV16 to 1·4 (1·2-1·7) for HPV51. HPV16 had the highest global AF (61·7%), followed by HPV18 (15·3%), HPV45 (4·8%), HPV33 (3·8%), HPV58 (3·5%), HPV31 (2·8%), and HPV52 (2·8%). Remaining causal genotypes (HPV35, 59, 39, 56, 51, 68, 73, 26, 69, and 82) had a combined global AF of 5·3%. AFs for HPV16 and 18 and HPV16, 18, 31, 33, 45, 52, and 58 combined were lowest in Africa (71·9% and 92·1%, respectively) and highest in central, western, and southern Asia (83·2% and 95·9%, respectively). HPV35 had a higher AF in Africa (3·6%) than other regions (0·6-1·6%). This study provides a comprehensive global picture of HPV genotype-specific AFs in ICC, before the influence of HPV vaccination. These data can inform HPV genotype-specific vaccination and screening strategies to reduce the burden of ICC. EU Horizon 2020 Research and Innovation Programme.

On the use of the healthy lifestyle index to investigate specific disease outcomes

The healthy lifestyle index (HLI), defined as the unweighted sum of individual lifestyle components, was used to investigate the combined role of lifestyle factors on health-related outcomes. We introduced weighted outcome-specific versions of the HLI, where individual lifestyle components were weighted according to their associations with disease outcomes. Within the European Prospective Investigation into Cancer and Nutrition (EPIC), we examined the association between the standard and the outcome-specific HLIs and the risk of T2D, CVD, cancer, and all-cause premature mortality. Estimates of the hazard ratios (HRs), the Harrell’s C-index and the population attributable fractions (PAFs) were compared. For T2D, the HR for 1-SD increase of the standard and T2D-specific HLI were 0.66 (95% CI: 0.64, 0.67) and 0.43 (0.42, 0.44), respectively, and the C-index were 0.63 (0.62, 0.64) and 0.72 (0.72, 0.73). Similar, yet less pronounced differences in HR and C-index were observed for standard and outcome-specific estimates for cancer, CVD and all-cause mortality. PAF estimates for mortality before age 80 were 57% (55%, 58%) and 33% (32%, 34%) for standard and mortality-specific HLI, respectively. The use of outcome-specific HLI could improve the assessment of the role of lifestyle factors on disease outcomes, thus enhancing the definition of public health recommendations.

Number of teeth and functional disability in community-dwelling older adults.

This study determined whether tooth loss was associated with the development of functional disability and estimated the population attributable fraction (PAF) of functional disability due to tooth loss, along with risk factors for functional disability such as physical function and cognitive impairment. The participants were 838 community-dwelling older adults aged ≥70 years living in the Tsurugaya district in Japan in 2003. The exposure variable was the number of remaining teeth (counted by trained dentists). Other variables were age, sex, depressive symptoms, cognitive impairment, educational attainment, physical function and social support. The Cox proportional hazards model was applied to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) of the incidence of functional disability for each risk factor, such as tooth loss. The functional disability PAF due to tooth loss was estimated, and risk factors for functional disability were identified. In total, 619 (73.9%) participants developed functional disability during follow-up. A multivariable model showed that those with <20 teeth (HR, 1.28; 95% CI, 1.08-1.53) were more likely to develop functional disability than those with 20 teeth or more. PAF estimation for functional disability was shown to have decreasing values in the following order: age, female sex, tooth loss and reduced physical function. Tooth loss was associated with the development of functional disability in community-dwelling older Japanese adults. While retaining teeth may be a potential strategy for avoiding functional disability, clinical studies on the effect of dental treatment on preventing functional disability are warranted.

Estimating and displaying population attributable fractions using the R package: graphPAF

Here we introduce graphPAF, a comprehensive R package designed for estimation, inference and display of population attributable fractions (PAF) and impact fractions. In addition to allowing inference for standard population attributable fractions and impact fractions, graphPAF facilitates display of attributable fractions over multiple risk factors using fan-plots and nomograms, calculations of attributable fractions for continuous exposures, inference for attributable fractions appropriate for specific risk factor →\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\rightarrow $$\\end{document} mediator →\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\rightarrow $$\\end{document} outcome pathways (pathway-specific attributable fractions) and Bayesian network-based calculations and inference for joint, sequential and average population attributable fractions in multi-risk factor scenarios. This article can be used as both a guide to the theory of attributable fraction estimation and a tutorial regarding how to use graphPAF in practical examples.

Impact of the use of small-area models on estimation of attributable mortality at a regional level.

The objective of this study is to assess the impact of applying prevalences derived from a small-area model at a regional level on smoking-attributable mortality (SAM). A prevalence-dependent method was used to estimate SAM. Prevalences of tobacco use were derived from a small-area model. SAM and population attributable fraction (PAF) estimates were compared against those calculated by pooling data from three national health surveys conducted in Spain (2011-2014-2017). We calculated the relative changes between the two estimates and assessed the width of the 95% CI of the PAF. Applying surveys-based prevalences, tobacco use was estimated to cause 53 825 (95% CI: 53 182-54 342) deaths in Spain in 2017, a figure 3.8% lower obtained with the small-area model prevalences. The lowest relative change was observed in the Castile-La Mancha region (1.1%) and the highest in Navarre (14.1%). The median relative change between regions was higher for women (26.1%), population aged ≥65 years (6.6%), and cardiometabolic diseases (9.0%). The differences between PAF by cause of death were never greater than 2%. Overall, the differences between estimates of SAM, PAF, and confidence interval width are small when using prevalences from both sources. Having these data available by region will allow decision-makers to implement smoking control measures based on more accurate data.

Population modifiable risk factors associated with neonatal mortality in 35 sub-Saharan Africa countries: analysis of data from demographic and health surveys

Sub-Saharan Africa (SSA) has the highest burden of neonatal mortality in the world. Identifying the most critical modifiable risk factors is imperative for reducing neonatal mortality rates. This study is the first to calculate population-attributable fractions (PAFs) for modifiable risk factors of neonatal mortality in SSA. We analysed the most recent Demographic and Health Surveys data sets from 35 SSA countries conducted between 2010 and 2022. Generalized linear latent and mixed models were used to estimate odds ratios (ORs) along with 95% confidence intervals (CIs). PAFs adjusted for communality were calculated using ORs and prevalence estimates for key modifiable risk factors. Subregional analyses were conducted to examine variations in modifiable risk factors for neonatal mortality across Central, Eastern, Southern, and Western SSA regions. In this study, we included 255,891 live births in the five years before the survey. The highest PAFs of neonatal mortality among singleton children were attributed to delayed initiation of breastfeeding (>1h after birth: PAF=23.88%; 95% CI: 15.91, 24.86), uncleaned cooking fuel (PAF=5.27%; 95% CI: 1.41, 8.73), mother's lacking formal education (PAF=4.34%; 95% CI: 1.15, 6.31), mother's lacking tetanus vaccination (PAF=3.54%; 95% CI: 1.55, 4.92), and infrequent antenatal care (ANC) visits (PAF=2.45; 95% CI: 0.76, 3.63). Together, these five modifiable risk factors were associated with 39.49% (95% CI: 21.13, 48.44) of neonatal deaths among singleton children in SSA. Our subregional analyses revealed some variations in modifiable risk factors for neonatal mortality. Notably, delayed initiation of breastfeeding consistently contributed to the highest PAFs of neonatal mortality across all four regions of SSA: Central, Eastern, Southern, and Western SSA. The PAF estimates in the present study indicate that a considerable proportion of neonatal deaths in SSA are preventable. We identified five modifiable risk factors that accounted for approximately 40% of neonatal deaths in SSA. The findings have policy implications. None.

The Burden of Cardiovascular Disease Attributable to Hypertension in Nigeria: A Modelling Study Using Summary-Level Data.

Globally, cardiovascular disease (CVD) remains the leading cause of mortality and disability, with hypertension being the single most important modifiable risk factor. Hypertension is responsible for about 18% of global deaths from CVD, of which African regions are disproportionately affected, especially sub-Saharan Africa. This study assessed the burden of major CVD subtypes attributable to hypertension in Nigeria. The population attributable fractions (PAF) for myocardial infarction, all strokes, ischaemic stroke and intracerebral haemorrhagic stroke attributable to hypertension in Nigeria were calculated using published results from the INTERHEART and INTERSTROKE studies and prevalence estimates of hypertension in Nigeria. PAF estimates were obtained for age, sex, and geopolitical zones. Overall, hypertension contributed to 13.2% of all myocardial infarctions and 24.6% of all strokes, including 21.6% of all ischaemic strokes and 33.1% of all intracerebral haemorrhagic strokes. Among men aged ≤55 years, the PAF for myocardial infarction ranged from 11.7% (North-West) to 14.6% (South-East), while in older men, it spanned 9.2% (North-West) to 11.9% (South-East). Among women aged ≤65 years, PAF varied from 18.6% (South-South) to 20.8% (South-East and North-Central), and among women aged >65 years, it ranged from 10.4% (South-South) to 12.7% (South-East). Hypertension is a key contributor to the burden of CVD in Nigeria. Understanding the burden of hypertension in the Nigerian population overall and key subgroups is crucial to developing and implementing contextualised health policies to reduce the burden of CVD. Public health interventions and policies centred on hypertension will play a critical role in potentially alleviating the burden of cardiovascular diseases (CVD) in Nigeria.

Long-term exposure to ambient fine particulate matter (PM2.5) and attributable pulmonary tuberculosis notifications in Ningxia Hui Autonomous Region, China: a health impact assessment

IntroductionLong-term exposure to fine particulate matter (≤2.5 µm (PM2.5)) has been associated with pulmonary tuberculosis (TB) notifications or incidence in recent publications. Studies quantifying the relative contribution of long-term PM2.5...

The contribution of metabolic risk factors to cardiovascular mortality in Golestan cohort study: Population attributable fraction estimation

BackgroundCardiovascular diseases (CVDs) are the leading causes of global mortality and disability. Several studies demonstrated that metabolic risk factors increase cardiovascular mortality. The aim of this study is to examine CVDs deaths and population attributable fractions (PAFs) of their metabolic risk factors in Iran. MethodsThis is a study on 8621 participants aged 45–75 years and older, recruited in the repeated measurement phase of the Golestan cohort study (GCS) in northeast of Iran. The Cox proportional hazards model was used to determine the adjusted hazard ratios (HRs). PAFs were calculated to enumerate CVDs mortality avoidable in the population if metabolic risk factors were eliminated. ResultsThe mortality of CVDs was attributable to metabolic factors, including high waist circumference (PAF, 28 %, [95 % CI: 16%–38 %]), high fasting blood sugar (FBS) (20 %, [15%–24 %]), overweight and obesity (19 %, [8%–28 %]), high blood pressure (16 %, [11%–21 %]), high low-density lipoprotein cholesterol (LDL-C) (8 %, [1%–15 %]), and high triglyceride (TG) (7 %, [3%–11 %]). Collectively, these metabolic risk factors accounted for 50 % of CVDs deaths. Females (67 %, [50%–78 %]) had a higher joint PAF of metabolic risk factors compared to males (43 %, [27%–55 %]). ConclusionsThe pattern of CVDs mortality attributable to metabolic risk factors in this study is not the same as similar studies in other parts of the world and previous studies in Iran. It is imperative that CVDs risk factors be specifically evaluated and addressed in various populations due to variety in geographical and temporal patterns in contribution of metabolic risk factors to CVD mortality.

The burden of overweight/obesity and weight-related behaviors attributable to food addiction in emerging adults

While food addiction has been positively associated with excess weight and disordered eating behaviors, this has not been examined in representative samples of emerging adults, who are at elevated risk for these outcomes. This study investigated relationships of food addiction with weight outcomes, weight perception, and weight-control behaviors in emerging adults and estimated the population attributable fraction to food addiction. Data from an observational cohort study were collected in seven annual waves from 2010 to 2016. A nationally representative sample of 2785 10th grade students was recruited from schools within each U.S. census region (73% participation) (mean ± SD baseline age = 16.3 ± 0.5years). Wave 7 retention was 81% (n = 2323, 60% female, mean ± SD = 22.6 ± 0.5 years). Outcomes included current BMI, BMI change from baseline – wave 7, increased weight status in wave 7 (increased weight status from baseline-wave 7), perceived overweight, dieting, any weight-control behavior, and extreme weight-control behaviors. Food addiction was measured in wave 7 using the modified Yale Food Addiction Scale. Relative risk of the outcomes associated with food addiction, and population attributable fraction, were estimated using adjusted log-binomial or robust Poisson regression analyses accounting for the complex survey design. Food addiction prevalence was 4.7%. Participants with food addiction were primarily females (91%); food addiction was uncorrelated with other sociodemographics. Food addiction was associated with 48%–167% increased RR for all outcomes, but these were attenuated after adjustment for confounders (31%–64%). The population attributable fraction for food addiction ranged from 2% (high wave 7 BMI) – 5% (extreme weight-control behaviors). Although the population attributable fraction estimates indicate that the public health burden of these outcomes attributable to food addiction may be relatively minor, food addiction may signal the presence of several adverse mental health symptoms.

Abstract 2216: Population attributable fraction of tobacco smoking on cancer incidence in 2020 in the Republic of Korea

Abstract Backgrounds: Cancer is the most common cause of mortality in Korea. According to International Agency for Research on Cancer (IARC), tobacco smoking is the Group I carcinogen, and the National Cancer Center Korea reported in 2009 that 20.9% and 2.3% of total cancer incidence were attributable to tobacco smoking in men and women, respectively. As of 2020, we estimated population attributable fraction (PAF) of cancer incidence due to tobacco smoking. Methods: We organized the Consensus Committee to develop strategy and methodology for PAF estimation. From the consensus, cancer sites for PAF estimation were selected based on the IARC list of classification with sufficient evidence in human. PAF was calculated by using Levin’s formula which includes exposure prevalence and relative risk (RR). Exposure prevalence was collected from the Korea National Health and Nutrition Examination Survey (KNHANES) in 2005, assuming that latent period from risk exposure to cancer development was 15 years. Tobacco smoking was defined as a categorical variable including non-smokers, past smokers, and current smokers. Pooled RRs were estimated from meta-analysis of RRs obtained from Korean Genome and Epidemiology Study, National Health Examination data provided by National Health Insurance Services, Korean Cancer Prevention Study-II, and KNHANES, each of which was linked with Korea Cancer Registry. Cox proportional hazards model were used to estimate RRs, while adjusted by age groups and alcohol consumption. Results: For men, 17.4% of all cancer cases were attributed to smoking, compared to 1.7% for women. Specifically, in men, 57.9% of lung cancer, 36.4% of bladder cancer, 31.9% of esophageal cancer, 39.1% of laryngeal cancer, 25.4% of oral cavity/pharyngeal cancer, 19.7% of liver cancer, and 19.5% of stomach cancer were attributed to tobacco smoking. In women, 51.7% of laryngeal cancer, 15.0% of esophageal cancer, 11.6% of lung cancer, 6.0% of bladder cancer, 5.1% of oral cavity/pharyngeal cancer, 5.0% of cervix cancer, and 4.6% of liver cancer were due to smoking. Conclusion: Tobacco smoking is an important cause of esophageal, laryngeal, and lung cancer in men, and esophageal and laryngeal cancer in women, and comprises a substantial proportion of PAF in cancer incidence. Thus, intervention measures are needed to reduce smoking prevalence for cancer prevention in Korea. Citation Format: Su Jung Byoun, Yoon-Jung Choi, Ye Lin Min, Sue-Kyung Park, Kwang Pil Ko, So_Seul Sung, NaYoung Kim, Keum Ji Jung, Sun Ha Jee, Sung-II Cho, Min Kyung Lim, Sohee Park, Eun Young Park, Hae Dong Woo, Kyu-Won Jung, E Hwa Yun, Byungmi Kim, Jin-Kyoung Oh, Yoonjoo Choi, Bohyun Park, Hong Gwan Seo, Jeong-Soo Im. Population attributable fraction of tobacco smoking on cancer incidence in 2020 in the Republic of Korea [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 2216.

Abstract 2201: Population attributable fraction of alcohol consumption on cancer incidence in 2020 in the Republic of Korea

Abstract Backgrounds: Alcohol consumption is the Group I carcinogen according to International Agency for Research on Cancer (IARC). In 2009, National Cancer Center Korea reported that 3.0% and 0.5% of total cancer incidence were attributable to alcohol consumption in men and women, respectively. Herein, we estimated population attributable fraction (PAF) of cancer incidence due to alcohol consumption as of 2020. Methods: We organized the Consensus Committee to establish strategy and methodology of PAF estimation. Based on the consensus, cancer sites for PAF estimation were selected according to the IARC List of classification with sufficient evidence in human. PAF was calculated by using Levin’s formula which incorporates exposure prevalence and relative risk (RR). Exposure prevalence was obtained from the Korea National Health and Nutrition Examination Survey (KNHANES) in 2005, assuming that latent period between exposure to alcohol consumption and diagnosis of cancer would be 15 years. Alcohol consumption was defined as non-drinking (&amp;lt;0.1g/day), light (0.1~30g/day), moderate (30~60g/day), and heavy (60~588g/day) drinking according to the definition by World Health Organization. We estimated RRs from the cohort data established by Korean Genome and Epidemiology Study, National Health Examination data provided by National Health Insurance Services, Korean Cancer Prevention Study-II, and KNHANES, while each cohort was linked with Korea Cancer Registry. RRs were estimated by using Cox proportional hazards model adjusted by age group and smoking status. RRs were then meta-analyzed to yield pooled RR by using random-effects model. Results: A total of 4.1% and 1.2% of all cancer incidence were attributable to alcohol consumption in men and women. By cancer sites, 67.1% of esophageal cancer, 23.8% of oral cavity/pharyngeal cancer, 15.1% of laryngeal cancer, 12.0% of colorectal cancer, and 7.9% of liver cancer were attributable to alcohol consumption in men. In women, 54.1% of liver cancer, 10.3% of colorectal cancer, 7.8% of oral cavity/pharyngeal cancer, and 3.6% of breast cancer were due to alcohol consumption. Conclusion: PAF of cancer incidence due to alcohol consumption has increased in 2020 compared to PAF in 2009. As alcohol consumption is an important cause of oral cavity/pharyngeal, esophageal, and laryngeal cancer in men, and liver and breast cancer in women, prevention measures are required to reduce alcohol consumption for cancer prevention in Korea. Citation Format: Ye Lin Min, YounJu Park, Su Jung Byoun, Sue_Kyung Park, Kwang Pil Ko, So-Seul Sung, Na Young Kim, Keum Ji Jung, Sun Ha Jee, Sung-Il Cho, Min Kyung Lim, Sohee Park, Eun Young Park, Hae Dong Woo, Kyu-Won Jung, E Hwa Yun, Byungmi Kim, Jin-Kyoung Oh, Yoonjoo Choi, Bohyun Park, Hong Gwan Seo, Yoon-Jung Choi, Jeong-Soo Im. Population attributable fraction of alcohol consumption on cancer incidence in 2020 in the Republic of Korea [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 2201.

Abstract 845: Population attributable fraction of cancer incidence due to infectious agents in 2020 in the Republic of Korea

Abstract Purpose: Earlier in 2009, National Cancer Center Korea estimated population attributable fraction (PAF) of cancer incidence due to infectious agents as 24.5% in men and 15.4% in women. This study aimed to investigate the proportion of cancer incidence attributable to infectious agents in the Korean population as of 2020. Materials and Methods: We organized the Consensus Committee to establish strategy and methodology of PAF estimation. Based on the consensus, infectious agents and cancer sites for PAF estimation were selected according to the IARC List of classification with sufficient evidence in human. As a result, infectious agents included hepatitis B and C virus (HBV and HCV), Helicobacter pylori (H. pylori), human papilloma virus (HPV), Clonorchis sinensis (C.sinensis), Epstein Barr virus, and human immunodeficiency virus, and cancer sites included hepatocellular carcinoma (HCC), cholangiocarcinoma (CCC), gastric cancer, pancreatic cancer, cervical cancer, and non-Hodgkin lymphoma. The PAF was calculated by using Levin’s formula including exposure prevalence and relative risks (RRs). Exposure prevalence data were collected from literature review of studies in Korean population in 2005 or earlier, assuming at least 15 years of latent period from risk exposure to cancer development. We performed systematic review for cohort studies and case-control studies which investigated RRs or ORs of cancer due to infectious agents in Korean population. Pooled RRs were estimated by meta-analysis of RRs or ORs by using random-effects model. Results: The fractions of all cancer incidence attributable to infectious agents were estimated to be 13.5% and 7.1% in men and women, respectively. The reduction in PAFs of cancer incidence due to infection in 2020 compared to 2009 is considered due to decrease in exposure prevalence. The major findings regarding infectious agents with large PAFs are as the following. We found that the fractions of HCC cases attributable to HBV infection were 47.7% and 38.4% for cancer incidence in men and women, respectively. The estimated PAFs of gastric cancer due to H. pylori were 40.4% in men and 38.2% in women. C. sinensis was related to 9.8% of CCC cases in men and 5.2% in women. A total of 77.7% of cervical cancer incidence was attributed to HPV infection. Conclusion: Although the total PAF due to infection has decreased over the last decade, infection still contributes a substantial fraction to cancer incidence in Korea. These recent estimates of cancer burden due to infection are important for planning and evaluating the national cancer prevention programs. Citation Format: Kyung-Shin Lee, Hwa Jin Lee, Yoon-Jung Choi, Moran Ki, Aesun Shin, Jong-Hun Kim, Sue-Kyung Park, So-Seul Sung, Sung Ji Moon, Hong Gwan Seo, Jin-Kyoung Oh, Jeong-Soo Im. Population attributable fraction of cancer incidence due to infectious agents in 2020 in the Republic of Korea [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 845.

Attributable risk of suicide for populations in Australia.

Each year approximately 3,000 Australians die by suicide. We estimated the population attributable risk for identified target populations to provide evidence on how much of the overall burden of suicide in the Australian population is experienced by each of them. We identified 17 demographic and clinical target populations at risk of suicide and calculated the population attributable fraction (PAF) using a single or pooled suicide risk and the proportional representation of each target population within Australia. Large PAF estimates were found for men (52%, 95% confidence interval (CI) 51%-53%), people bereaved by suicide (35%, 95% CI 14%-64%), people with a mental health or behavioural condition (33%, 95%CI 17%-48%), people with a chronic physical condition (27%, 95%CI 18%-35%), adults aged 25-64 years (13%, 95%CI 12%-14%), LGB populations (9%, 95%CI 6%-13%), offenders (9%, 95%CI 8%-10%), and people employed in blue collar occupations (8%, 95%CI 4%-12%). The PAF is limited by assumptions, namely, that risk factors are independent, and that the relationship between risk factors and outcomes are unidirectional and constant through time. Considerable reductions in the overall suicide rate in Australia may occur if risk factors are addressed in identified populations with large PAF estimates. These estimates should be considered as an adjunct to other important inputs into suicide prevention policy priorities.

Bias assessment and correction for Levin’s population attributable fraction in the presence of confounding

In 1953, Morton Levin introduced a simple approach to estimating population attributable fractions (PAF) depending only on risk factor prevalence and relative risk. This formula and its extensions are still in widespread use today, particularly to estimate PAF in populations where individual data is unavailable. Unfortunately, Levin’s approach is known to be asymptotically biased for the PAF when the risk factor-disease relationship is confounded even if relative risks that are correctly adjusted for confounding are used in the estimator. Here we describe a simple re-expression of Miettinen’s estimand that depends on the causal relative risk, the unadjusted relative risk and the population risk factor prevalence. While this re-expression is not new, it has been underappreciated in the literature, and the associated estimator may be useful in estimating PAF in populations when individual data is unavailable provided estimated adjusted and unadjusted relative risks can be transported to the population of interest. Using the re-expressed estimand, we develop novel analytic formulae for the relative and absolute asymptotic bias in Levin’s formula, solidifying earlier work by Darrow and Steenland that used simulations to investigate this bias. We extend all results to settings with non-binary valued risk factors and continuous exposures and discuss the utility of these results in estimating PAF in practice.

Associations between BMIin youth and site-specific cancer in men-A cohort study with register linkage.

This study examined BMI in young men and incident site-specific cancer to estimate population attributable fractions due to BMI based on projected obesity prevalence. A population-based cohort study with measured height and weight at age 18. Cox regression models assessed linear associations for BMI and included age, year, and site of conscription as well as parental level of education as covariates. Primary analyses were performed in 1,489,115 men, of whom 78,217 subsequently developed cancer during a mean follow-up of 31 years. BMI was linearly associated with risk of developing all 18 site-specific cancers assessed (malignant melanoma; leukemia; myeloma; Hodgkin lymphoma; non-Hodgkin lymphoma; and cancer in the lungs, head and neck, central nervous system, thyroid, esophagus, stomach, pancreas, liver and gallbladder, colon, rectum, kidney, and bladder), in some instances evident at BMI levels usually defined as normal (20-25 kg/m2 ). Higher BMI was associated with lower risk of prostate cancer. The highest hazard ratios and population attributable fractions were seen for some gastrointestinal cancers. This study reports linear associations between BMI at age 18 and subsequent site-specific cancers, calling for rapid action to stem the obesity epidemic and to prepare the health care system for steep increases in cancer cases.

Impact of Alcohol Consumption on Multiple Sclerosis Using Model-based Standardization and Misclassification Adjustment Via Probabilistic Bias Analysis.

The etiology of multiple sclerosis (MS) is still not well-demonstrated, and assessment of some risk factors like alcohol consumption has problems like confounding and measurement bias. To determine the causal effect of alcohol consumption on MS after adjusting for alcohol consumption misclassification bias and confounders. In a population-based incident case-control study, 547 patients with MS and 1057 healthy people were recruited. A minimally sufficient adjustment set of confounders was derived using the causal directed acyclic graph. The probabilistic bias analysis method (PBAM) using beta, logit-logistic, and triangular probability distributions for sensitivity/specificity to adjust for misclassification bias in self-reporting alcohol consumption and model-based standardization (MBS) to estimate the causal effect of alcohol consumption were used. Population attributable fraction (PAF) estimates with 95% Monte Carlo sensitivity analysis (MCSA) intervals were calculated using PBAM and MBS analysis. Bootstrap was used to deal with random errors. The adjusted risk ratio (95% MCSA interval) from the probabilistic bias analysis and MBS between alcohol consumption and MS using the three distribution was in the range of 1.93 (1.07 to 4.07) to 2.02 (1.15 to 4.69). The risk difference (RD) in all three scenarios was 0.0001 (0.0000 to 0.0005) and PAF was in the range of 0.15 (0.010 to 0.50) to 0.17 (0.001 to 0.47). After adjusting for measurement bias, confounding, and random error alcohol consumption had a positive causal effect on the incidence of MS.

Potentially modifiable dementia risk factors in all Australians and within population groups: an analysis using cross-sectional survey data

Dementia is the second leading cause of disease burden in Australia. We aimed to calculate the population attributable fractions (PAFs) of dementia attributable to 11 of 12 previously identified potentially modifiable health and social risk factors (less education, hearing loss, hypertension, obesity, smoking, depression, social isolation, physical inactivity, diabetes, alcohol excess, air pollution, and traumatic brain injury), for Australians overall and three population groups (First Nations, and those of European and Asian ancestry). We calculated the prevalence of dementia risk factors (excluding traumatic brain injury) and PAFs, adjusted for communality, from the cross-sectional National Aboriginal and Torres Strait Islander Health Survey (2018-19), National Aboriginal and Torres Strait Islander Social Survey (2014-15), National Health Survey (2017-18), and General Social Survey (2014) conducted by the Australian Bureau of Statistics. We conducted sensitivity analyses using proxy estimates for traumatic brain injury (12th known risk factor) for which national data were not available. A large proportion (38·2%, 95% CI 37·2-39·2) of dementia in Australia was theoretically attributable to the 11 risk factors; 44·9% (43·1-46·7) for First Nations Australians, 36·4% (34·8-38·1) for European ancestry, and 33·6% (30·1-37·2) for Asian ancestry. Including traumatic brain injury increased the PAF to 40·6% (39·6-41·6) for all Australians. Physical inactivity (8·3%, 7·5-9·2), hearing loss (7·0%, 6·4-7·6), and obesity (6·6%, 6·0-7·3) accounted for approximately half of the total PAF estimates across Australia, and for all three population groups. Our PAF estimates indicate a substantial proportion of dementia in Australia is potentially preventable, which is broadly consistent with global trends and results from other countries. The highest potential for dementia prevention was among First Nations Australians, reflecting the enduring effect of upstream social, political, environmental, and economic disadvantage, leading to greater life-course exposure to dementia risk factors. Although there were common dementia risk factors across different population groups, prevention strategies should be informed by community consultation and be culturally and linguistically appropriate. Australian National Health and Medical Research Council and University College London Hospitals' National Institute for Health Research (NIHR) Biomedical Research Centre, and North Thames NIHR Applied Research Collaboration.

The Burden of Respiratory Disease from Formaldehyde, Damp and Mould in English Housing

Quantifying the burden of disease from exposure to poor indoor air pollution can support policy development. In England, there is current regulatory and public attention on the health implications of residential exposure to formaldehyde, damp and mould. However, there is scarce information on these health impacts at the population scale. As such, we assessed the burden of key respiratory diseases from residential formaldehyde, damp and/or mould for the English population aged 0–14 and 15–49. We obtained data on the percentage of dwellings affected by damp and/or mould from the English Housing Survey and estimated the distribution of residential formaldehyde concentrations (annual average (μg/m3)) by pooling data from monitoring studies conducted in England. Exposures were combined with epidemiological relationships and national health data to estimate Population Attributable Fractions (PAFs), disease incidence, and Disability Adjusted Life Years (DALYs) lost associated with residential formaldehyde or damp and/or mould exposure in England. We made estimates for the year 2019 but also looked back several years in time. Exposure to formaldehyde was associated with approximately 4000 new cases of childhood asthma (~800 DALYs lost) in 2019, though the estimates were sensitive to the placement of the lower exposure threshold. Exposure to damp and/or mould was associated with approximately 5000 new cases of asthma (~2200 DALYs) and approximately 8500 lower respiratory infections (~600 DALYs) among children and adults in 2019, though the PAFs were unequally distributed across dwellings based on income and ethnicity. Alternative data sources suggest that the percentage of dwellings affected by damp and/or mould may even be higher, resulting in a possible 3–8-fold greater number of cases and DALYs. Our assessment emphasizes a potential respiratory health burden in England associated with residential formaldehyde as well as damp and/or mould, further highlighting the public health importance of good indoor air quality and good quality housing.

Contribution of Preterm Birth to Mortality Among Neonates With Birth Defects

ABSTRACT Birth defects are an important cause of death among infants who die within their first year. Among other complications associated with birth defects, mothers are at higher risk for preterm delivery, which is also associated with infant mortality. Some previous research has shown that in a population of infants with birth defects, those born at term are 3 to 4 times more likely to survive as their preterm peers. This study aimed to clarify the relationship between birth defects, infant mortality, and preterm birth using the population attributable fraction (PAF), examining the mortality risk attributable to preterm birth in infants with birth defects. Data for this study came from the Texas Birth Defects Registry, which includes live births, still births, and pregnancy terminations with a structural birth defect or chromosomal abnormality. The sample included live-born infants with 1 or more major birth defects between January 1, 1999 and December 31, 2014. Exclusion criteria were infants with a chromosomal abnormality or diagnosed syndrome, infants missing data on gestational age or with a gestational age less than 24 weeks, and anencephaly. The final sample included 169,148 infants; of these, 40,872 were preterm (&lt;37 weeks), 128,276 were term, and 2715 died as neonates. Distributions of maternal and infant characteristics were significantly different between the 2 groups (P &lt; 0.01), with more mothers of preterm infants experiencing diabetes, hypertension, multiple gestation pregnancies, or cesarean delivery. In addition, more preterm infants had multiple birth defects or congenital heart defects than their term peers. Analysis of the data included estimates of PAF for infant mortality risk attributed to 31 different birth defects, with results ranging from 12.5% to 71.9%. Highest PAF estimates were observed for anotia or microtia (71.9%; CI, 41.1–86.6), hypospadias (69.4%; CI, 53.9–79.7), talipes equinovarus (69.4%; CI, 60.0–76.6), and stenosis or atresia of the large intestine, rectum, or anal canal (69.1%; CI, 56.3–78.1). On the other hand, lowest PAF estimates included hypoplastic left heart syndrome (12.5%; CI, 8.7–16.1), coarctation of the aorta (19.7%; CI, 9.8–28.4), diaphragmatic hernia (23.8%; CI, 19.2–28.2), and holoprosencephaly (26.0%; CI, 14.5–35.9). For each defect associated with a higher PAF, a large percentage of infant deaths were those born preterm. Infants with defects that had a high PAF estimate were 11.4–17.8 times more likely to die if born preterm than term, and those with a low PAF estimate were only 1.9–2.3 times more likely to die if born preterm as opposed to full term. Overall, analysis showed that 51.7% of neonatal mortality in infants with birth defects is attributable to preterm birth. Stratified analyses performed to reduce confounders showed similar results. The relationship between birth defects and other factors contributing to preterm delivery and infant mortality is complex, and thus this study is unable to fully characterize these relationships. The associations revealed in this analysis indicate that preterm birth is a significant contributor to infant mortality in infants with birth defects, and further research can more fully characterize this association by attempting interventions both to mitigate preterm birth and to reduce infant mortality after preterm birth. Further research might also focus on particular types of birth defects and how rates of preterm birth might be reduced in specific cases of either infant or maternal risks.