Estimation of age of onset and progression of breast cancer by absolute risk dependent on polygenic risk score and other risk factors

BACKGROUND

Genetic, lifestyle, reproductive, and anthropometric factors are associated with the risk of developing breast cancer. However, it is not yet known whether polygenic risk score (PRS) and absolute risk based on a combination of risk factors are associated with the risk of progression of breast cancer. This study aims to estimate the distribution of sojourn time (pre-clinical screen-detectable period) and mammographic sensitivity by absolute breast cancer risk derived from polygenic profile and the other risk factors.

METHODS

The authors used data from a population-based case-control study. Six categories of 10-year absolute risk based on different combinations of risk factors were derived using the Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm. Women were classified into low, medium, and high-risk groups. The authors constructed a continuous-time multistate model. To calculate the sojourn time, they simulated the trajectories of subjects through the disease states.

RESULTS

There was little difference in sojourn time with a large overlap in the 95% confidence interval (CI) between the risk groups across the six risk categories and PRS studied. However, the age of entry into the screen-detectable state varied by risk category, with the mean age of entry of 53.4 years (95% CI, 52.2-54.1) and 57.0 years (95% CI, 55.1-57.7) in the high-risk and low-risk women, respectively.

CONCLUSION

In risk-stratified breast screening, the age at the start of screening, but not necessarily the frequency of screening, should be tailored to a woman's risk level. The optimal risk-stratified screening strategy that would improve the benefit-to-harm balance and the cost-effectiveness of the screening programs needs to be studied.

Impact of Alcohol Consumption on Breast Cancer Incidence and Mortality: The Women's Health Study

Background: Alcohol intake is associated with breast cancer (BC) risk, but estimates of greatest public health relevance have not been quantified in large studies with long duration. Materials and Methods: In this prospective cohort study of 39,811 women (median 25 years follow-up), we examined the association between alcohol consumption and BC incidence and mortality with adjusted hazard ratios (HRs), cubic splines, absolute risks, number needed to harm (NNH), and population-attributable fractions. Results: We documented 2,830 cases of BC, including 237 BC deaths. Each additional alcoholic drink/day was associated with a 10% higher rate (HR = 1.10, 95% confidence intervals [CIs]: 1.04-1.16) of total BC in a linear manner (p = 0.0004). The higher rate was apparent for estrogen receptor (ER)+ (HR = 1.12, 95% CI: 1.06-1.18) but not ER- tumors (HR = 0.95, 95% CI: 0.82-1.10), with a statistically significant difference between these associations (p = 0.03). We constructed models comparing BC incidence among 100,000 women followed for 10 years. Compared to a scenario where all women rarely or never consumed alcohol, we expect 63.79 (95% CI: 58.35-69.24) more cases (NNH = 1,567) had all women consumed alcohol at least monthly and 278.66 (95% CI: 268.70-288.62) more cases (NNH = 358) had all women consumed >1 drink/day. Approximately 4.1% of BC cases were attributable to consumption exceeding one drink/month. Conclusion: Alcohol consumption is associated with a linear dose-response increase in BC incidence even within recommended limits of up to one alcoholic drink/day, at least for ER+ tumors. Our estimates of risk differences, attributable fraction, and NNH quantify the burden that alcohol consumption imposes on women in the general population. ClinicalTrials.gov Identifier: NCT00000479.

EPO rs1617640 A>C is a Protective Factor for Chronic Obstructive Pulmonary Disease: A Case Control Study

BACKGROUND

The occurrence and development of chronic obstructive pulmonary disease (COPD) are regulated by environmental and genetic factors. In hypoxia, Erythropoietin (EPO) satisfies the body's need for oxygen by promoting the production of red blood cells. Hypoxia was proven to be a common physiological condition in COPD progression and associated with many complications. Some studies have found that EPO is involved in the development of COPD. But the mechanism has not been fully proven.

METHODS

We conducted a case-control study enrolled 1095 COPD patients and 1144 healthy controls in Guangdong Province to evaluate the association between EPO polymorphisms (rs1617640 A>C, rs507392 A>G, rs564449 G>T) and COPD susceptibility. 872 participants from southern Gansu Province were recruited to verify the effect of EPO polymorphisms on lung function.

RESULTS

EPO rs1617640 C allele reduced COPD susceptibility in southern Chinese significantly (AC vs. AA: adjusted Odds ratio (OR) = 0.805, 95% CI = 0.669-0.969; AC+CC vs. AA: adjusted OR = 0.822, 95% CI = 0.689-0.980). However, there was no association between rs507392 A>G and rs564449 G>T polymorphisms and COPD susceptibility (p > 0.05). We further observed that the rs1617640 C allele was associated with higher FEV1 and FVC in Guangdong and Gansu populations significantly (both p < 0.05). In brief, the level of FEV1 and FVC increased with the C allele number. We modeled the relative risk for men and women, in which the population-attributable risks chances were 0.449 (0.258-0.641) and 0.262 (0.128-0.396) respectively. In this model, smoking status, coal as fuels, education level, and rs1617640 A>C were finally retained for males, while smoking status, biomass as fuels, and1617640 A>C were retained for females. In the end, using the method developed by Gail and Bruzzi, we fitted a 10-year absolute risk model for southern Chinese with different individual relative risks, which was presented as a table.

CONCLUSIONS

In conclusion, this study found that EPO rs1617640 A>C polymorphism is associated with COPD susceptibility in southern Chinese, and the C allele was associated with better lung function. In addition, it could also be considered a genetic marker associated with environmental factors to predict the absolute 10-year risk of COPD in southern Chinese.

Historical Review of the Use of Relative Risk Statistics in the Portrayal of the Purported Hazards of High LDL Cholesterol and the Benefits of Lipid-Lowering Therapy

The manner in which clinical trial investigators present their findings to healthcare providers and the public can have a substantial influence on their impact. For example, if a heart attack occurs in 2% of those in the placebo group and in 1% of those in the drug-treated group, the benefit to the treated population is only one percentage point better than no treatment. This finding is unlikely to generate much enthusiasm from the study sponsors and in the reporting of the findings to the public. Instead, trial directors can amplify the magnitude of the appearance of the treatment benefit by using the relative risk (RR) value of a 50% reduction of the risk of a heart attack, since one is 50% of two. By using the RR type of data analysis, clinical trial directors can promote the outcome of their trial in their publication and to the media as highly successful while minimizing or disregarding entirely the absolute risk (AR) reduction of only one percentage point. The practice of expressing the RR without the AR has become routinely deployed in the reporting of findings in many different areas of clinical research. We have provided a historical perspective on how this form of data presentation has become commonplace in the reporting of findings from randomized controlled trials (RCTs) on coronary heart disease (CHD) event monitoring and prevention over the past four decades. We assert that the emphasis on RR coupled with insufficient disclosure of AR in the reporting of RCT outcomes has led healthcare providers and the public to overestimate concerns about high cholesterol and to be misled as to the magnitude of the benefits of cholesterol-lowering therapy. The goal of this review is to prompt the scientific community to address this misleading approach to data presentation.

Will Absolute Risk Estimation for Time to Next Screen Work for an Asian Mammography Screening Population?

Personalized breast cancer risk profiling has the potential to promote shared decision-making and improve compliance with routine screening. We assessed the Gail model's performance in predicting the short-term (2- and 5-year) and the long-term (10- and 15-year) absolute risks in 28,234 asymptomatic Asian women. Absolute risks were calculated using different relative risk estimates and Breast cancer incidence and mortality rates (White, Asian-American, or the Singapore Asian population). Using linear models, we tested the association of absolute risk and age at breast cancer occurrence. Model discrimination was moderate (AUC range: 0.580-0.628). Calibration was better for longer-term prediction horizons (E/O_{long-term ranges}: 0.86-1.71; E/O_{short-term ranges}:1.24-3.36). Subgroup analyses show that the model underestimates risk in women with breast cancer family history, positive recall status, and prior breast biopsy, and overestimates risk in underweight women. The Gail model absolute risk does not predict the age of breast cancer occurrence. Breast cancer risk prediction tools performed better with population-specific parameters. Two-year absolute risk estimation is attractive for breast cancer screening programs, but the models tested are not suitable for identifying Asian women at increased risk within this short interval.

Development and validation of a simple prostate cancer risk prediction model based on age, family history, and polygenic risk

BACKGROUND

Accurate prostate cancer risk assessment will enable identification of men who are at increased risk of the disease. Using the UK Biobank population-based cohort, we developed and validated a simple model comprising age, family history, and a polygenic risk score (PRS) to predict 5-year risk of prostate cancer.

METHODS

Eligible participants were unaffected Caucasian men aged 40-69 years at their baseline assessment who had genotyping data available and had completed 6 or more weeks of follow-up. Family history was the number of affected first-degree relatives: 0, 1, or 2+. We used 264 single-nucleotide polymorphisms (SNPs) of a previously developed 269-SNP PRS and population standardized the PRS to have a mean of 1. Age was categorized into 10-year groups: 40-49, 50-59, and 60-69. In a 70% training data set, we used Cox regression with age as the time axis to model family history, PRS, and age group. The model estimates were used with prostate cancer incidences to derive 5-year risks of prostate cancer. Using 5 years of follow-up in a 30% testing data set, the model was tested in terms of its association per quintile of risk, discrimination, and calibration.

RESULTS

Of the 198 334 eligible participants, 8996 (4.5%) were diagnosed with incident prostate cancer during follow-up and had a mean age of 67.9 (SD = 5.8) years at diagnosis. The best-fitting model included the PRS, family history, 10-year age group, interactions between age and PRS, and age and family history. In the 30% testing data set with follow-up limited to 5 years, the hazard ratio per SD of 5-year risk was 3.058 (95% confidence interval [CI], 2.720-3.438) and the Harrell's C-index was 0.811 (95% CI, 0.800-0.821). Overall, there were 1088 observed and 1159.1 expected prostate cancers, a standardized incidence ratio of 0.939 (95% CI, 0.885-0.996).

CONCLUSIONS

Men at increased risk of prostate cancer could benefit from informed discussions around the risks and benefits of available options for screening for prostate cancer. Although the model was developed in Caucasian men, it can be used with ethnicity-specific polygenic risk and incidence rates for other populations.

A Dynamic Risk Model for Multitype Recurrent Events

Recurrent events can occur more than once in the same individual; such events may be of different types, known as multitype recurrent events. They are very common in longitudinal studies. Often there is a terminating event, after which no further events can occur. The risk of any event, including terminating events such as death or cure, is typically affected by prior events. We propose a flexible joint multitype recurrent-events model that explicitly provides estimates of the change in risk for each event due to subject characteristics, including number and type of prior events and the absolute risk for every event type (terminating and nonterminating), and predicts event-free survival probability over a desired time period. The model is fully parametric, and therefore a standard likelihood function and robust standard errors can be constructed. We illustrate the model with applications to the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (1994-2002) and provide discussion of the results and model features.

Electronic Health Record-Based Absolute Risk Prediction Model for Esophageal Cancer in the Chinese Population: Model Development and External Validation

BACKGROUND

China has the largest burden of esophageal cancer (EC). Prediction models can be used to identify high-risk individuals for intensive lifestyle interventions and endoscopy screening. However, the current prediction models are limited by small sample size and a lack of external validation, and none of them can be embedded into the booming electronic health records (EHRs) in China.

OBJECTIVE

This study aims to develop and validate absolute risk prediction models for EC in the Chinese population. In particular, we assessed whether models that contain only EHR-available predictors performed well.

METHODS

A prospective cohort recruiting 510,145 participants free of cancer from both high EC-risk and low EC-risk areas in China was used to develop EC models. Another prospective cohort of 18,441 participants was used for validation. A flexible parametric model was used to develop a 10-year absolute risk model by considering the competing risks (full model). The full model was then abbreviated by keeping only EHR-available predictors. We internally and externally validated the models by using the area under the receiver operating characteristic curve (AUC) and calibration plots and compared them based on classification measures.

RESULTS

During a median of 11.1 years of follow-up, we observed 2550 EC incident cases. The models consisted of age, sex, regional EC-risk level (high-risk areas: 2 study regions; low-risk areas: 8 regions), education, family history of cancer (simple model), smoking, alcohol use, BMI (intermediate model), physical activity, hot tea consumption, and fresh fruit consumption (full model). The performance was only slightly compromised after the abbreviation. The simple and intermediate models showed good calibration and excellent discriminating ability with AUCs (95% CIs) of 0.822 (0.783-0.861) and 0.830 (0.792-0.867) in the external validation and 0.871 (0.858-0.884) and 0.879 (0.867-0.892) in the internal validation, respectively.

CONCLUSIONS

Three nested 10-year EC absolute risk prediction models for Chinese adults aged 30-79 years were developed and validated, which may be particularly useful for populations in low EC-risk areas. Even the simple model with only 5 predictors available from EHRs had excellent discrimination and good calibration, indicating its potential for broader use in tailored EC prevention. The simple and intermediate models have the potential to be widely used for both primary and secondary prevention of EC.

Physical frailty, adherence to ideal cardiovascular health and risk of cardiovascular disease: a prospective cohort study

BACKGROUND

longitudinal evidence concerning frailty phenotype and the risk of cardiovascular disease (CVD) remained insufficient, and whether CVD preventive strategies exert low CVD risk on frail adults is unclear.

OBJECTIVES

we aimed to prospectively evaluate the association of frailty phenotype, adherence to ideal cardiovascular health (CVH) and their joint associations with the risk of CVD.

METHODS

a total of 314,093 participants from the UK Biobank were included. Frailty phenotype was assessed according to the five criteria of Fried et al.: weight loss, exhaustion, low physical activity, slow gait speed and low grip strength. CVH included four core health behaviours (smoking, physical activity and diet) and three health factors (weight, cholesterol, blood pressure and glycaemic control). The outcome of interest was incident CVD, including coronary heart disease, heart failure and stroke.

RESULTS

compared with the non-frail people whose incident rate of overall CVD was 6.54 per 1,000 person-years, the absolute rate difference per 1,000 person-years was 1.67 (95% confidence interval, CI: 1.33, 2.02) for pre-frail and 5.00 (95% CI: 4.03, 5.97) for frail. The ideal CVH was significantly associated with a lower risk of all CVD outcomes. For the joint association of frailty and CVH level with incident CVD, the highest risk was observed among frailty accompanied by poor CVH with an HR of 2.92 (95% CI: 2.68, 3.18).

CONCLUSIONS

our findings indicate that physical frailty is associated with CVD incidence. Improving CVH was significantly associated with a considerable decrease in CVD risk, and such cardiovascular benefits remain for the frailty population.

A combined clinical and genetic model for predicting risk of ovarian cancer

OBJECTIVE

Women with a family history of ovarian cancer or a pathogenic or likely pathogenic gene variant are at high risk of the disease, but very few women have these risk factors. We assessed whether a combined polygenic and clinical risk score could predict risk of ovarian cancer in population-based women who would otherwise be considered as being at average risk.

METHODS

We used the UK Biobank to conduct a prospective cohort study assessing the performance of 10-year ovarian cancer risks based on a polygenic risk score, a clinical risk score and a combined risk score. We used Cox regression to assess association, Harrell's C-index to assess discrimination and Poisson regression to assess calibration.

RESULTS

The combined risk model performed best and problems with calibration were overcome by recalibrating the model, which then had a hazard ratio per quintile of risk of 1.338 [95% confidence interval (CI), 1.152-1.553], a Harrell's C-index of 0.663 (95% CI, 0.629-0.698) and overall calibration of 1.000 (95% CI, 0.874-1.145). In the refined model with estimates based on the entire dataset, women in the top quintile of 10-year risk were at 1.387 (95% CI, 1.086-1.688) times increased risk, while women in the top quintile of full-lifetime risk were at 1.527 (95% CI, 1.187-1.866) times increased risk compared with the population.

CONCLUSION

Identification of women who are at high risk of ovarian cancer can allow healthcare providers and patients to engage in joint decision-making discussions around the risks and benefits of screening options or risk-reducing surgery.

Lipoprotein(a) and Body Mass Compound the Risk of Calcific Aortic Valve Disease

BACKGROUND

High plasma lipoprotein(a) and high body mass index are both causal risk factors for calcific aortic valve disease.

OBJECTIVES

This study sought to test the hypothesis that risk of calcific aortic valve disease is the highest when both plasma lipoprotein(a) and body mass index are extremely high.

METHODS

From the Copenhagen General Population Study, we used information on 69,988 randomly selected individuals recruited from 2003 to 2015 (median follow-up 7.4 years) to evaluate the association between high lipoprotein(a) and high body mass index with risk of calcific aortic valve disease.

RESULTS

Compared with individuals in the 1st to 49th percentiles for both lipoprotein(a) and body mass index, the multivariable adjusted HRs for calcific aortic valve disease were 1.6 (95% CI: 1.3-1.9) for the 50th to 89th percentiles of both (16% of all individuals) and 3.5 (95% CI: 2.5-5.1) for the 90th to 100th percentiles of both (1.1%) (P for interaction = 0.92). The 10-year absolute risk of calcific aortic valve disease increased with higher lipoprotein(a), body mass index, and age, and was higher in men than in women. For women and men 70-79 years of age with body mass index ≥30.0 kg/m², 10-year absolute risks were 5% and 8% for lipoprotein(a) ≤42 mg/dL (88 nmol/L), 7% and 11% for 42-79 mg/dL (89-169 nmol/L), and 9% and 14% for lipoprotein(a) ≥80 mg/dL (170 nmol/L), respectively.

CONCLUSIONS

Extremely high lipoprotein(a) levels and extremely high body mass index together conferred a 3.5-fold risk of calcific aortic valve disease. Ten-year absolute risk of calcific aortic valve disease by categories of lipoprotein(a) levels, body mass index, age, and sex ranged from 0.4% to 14%.

Should Age-Dependent Absolute Risk Thresholds Be Used for Risk Stratification in Risk-Stratified Breast Cancer Screening?

In risk-stratified cancer screening, multiple risk factors are incorporated into the risk assessment. An individual's estimated absolute cancer risk is linked to risk categories with tailored screening recommendations for each risk category. Absolute risk, expressed as either remaining lifetime risk or shorter-term (five- or ten-year) risk, is estimated from the age at assessment. These risk estimates vary by age; however, some clinical guidelines (e.g., enhanced breast cancer surveillance guidelines) and ongoing personalised breast screening trials, stratify women based on absolute risk thresholds that do not vary by age. We examine an alternative approach in which the risk thresholds used for risk stratification vary by age and consider the implications of using age-independent risk thresholds on risk stratification. We demonstrate that using an age-independent remaining lifetime risk threshold approach could identify high-risk younger women but would miss high-risk older women, whereas an age-independent 5-year or 10-year absolute risk threshold could miss high-risk younger women and classify lower-risk older women as high risk. With risk misclassification, women with an equivalent risk level would be offered a different screening plan. To mitigate these problems, age-dependent absolute risk thresholds should be used to inform risk stratification.

Classifying and communicating risks in prediabetes according to fasting glucose and/or glycated hemoglobin: PREDAPS cohort study

OBJECTIVE

Information about prognostic outcomes can be of great help for people with prediabetes and for physicians in the face of scientific controversy about the cutoff point for defining prediabetes. We aimed to estimate different prognostic outcomes in people with prediabetes.

DESIGN

Prospective cohort of subjects with prediabetes according to American Diabetes Association guidelines.

MAIN OUTCOME MEASURES

The probabilities of diabetes onset versus non-onset, the odds against diabetes onset, and the probability of reverting to normoglycemia according to different prediabetes categories were calculated.

RESULTS

The odds against diabetes onset ranged from 29:1 in individuals with isolated FPG of 100-109 mg/dL to 1:1 in individuals with FPG 110-125 mg/dL plus HbA1c 6.0-6.4%. The probability of reversion to normoglycemia was 31.2% (95% CI 24.0-39.6) in those with isolated FPG 100-109 mg/dL and 6.2% (95% CI 1.4-10.0) in those with FPG 110-125 mg/dL plus HbA1c 6.0-6.4%. Of every 100 participants in the first group, 97 did not develop diabetes and 31 reverted to normoglycemia, while in the second group those figures were 52 and 6.

CONCLUSIONS

Using odds of probabilities and absolute numbers might be useful for people with prediabetes and physicians to share decisions on potential interventions.Key pointsCommunicating knowledge on the course of the disease to make clinical decisions is not always done appropriately.Prediabetes is an example where risk communication is important because the prognosis of subjects with prediabetes is very heterogeneous.Depending on fasting plasma glucose and HbA1c levels, the odds of probabilities against diabetes onset ranged from 29: 1 to 1: 1.Depending on fasting plasma glucose and HbA1c levels, the number of subjects in 100 who revert to normoglycemia ranged from 31 to 6.Using probabilities and number absolutes on the prognosis of prediabetes may be useful for people with prediabetes and physicians to share decisions on potential interventions.

Efficacy of COVID-19 vaccines: Several modes of expression should be presented in scientific publications

Several vaccines are being developed as part of the COVID‐19 pandemic. The results of clinical trials for these vaccines were published with efficacy values of more than 90%, using mainly relative risk (RR). In this paper, we decided to reanalyse the data using the different validated methods of risk expression. Using main publications, absolute risks (AR), AR reduction (ARR), number needed to treat (NNT) were calculated for five COVID‐19 vaccines (tozinameran Comirnaty®, Moderna, Vaxzevria®, Janssen, and Sputnik V vaccines). AR, ARR, NNT, and RR values varied according to COVID‐19 vaccines. The order of the different vaccines was not the same according to the chosen efficacy parameters. This is a further example of the need to express results of clinical trials, using not only RR, but also AR, ARR, and NNT in order to clearly present the clinical interest of drugs.

Assessing Lung Cancer Absolute Risk Trajectory Based on a Polygenic Risk Model

Lung cancer is the leading cause of cancer-related death globally. An improved risk stratification strategy can increase efficiency of low-dose CT (LDCT) screening. Here we assessed whether individual's genetic background has clinical utility for risk stratification in the context of LDCT screening. On the basis of 13,119 patients with lung cancer and 10,008 controls with European ancestry in the International Lung Cancer Consortium, we constructed a polygenic risk score (PRS) via 10-fold cross-validation with regularized penalized regression. The performance of risk model integrating PRS, including calibration and ability to discriminate, was assessed using UK Biobank data (N = 335,931). Absolute risk was estimated on the basis of age-specific lung cancer incidence and all-cause mortality as competing risk. To evaluate its potential clinical utility, the PRS distribution was simulated in the National Lung Screening Trial (N = 50,772 participants). The lung cancer ORs for individuals at the top decile of the PRS distribution versus those at bottom 10% was 2.39 [95% confidence interval (CI) = 1.92-3.00; P = 1.80 × 10-14] in the validation set (P trend = 5.26 × 10-20). The OR per SD of PRS increase was 1.26 (95% CI = 1.20-1.32; P = 9.69 × 10-23) for overall lung cancer risk in the validation set. When considering absolute risks, individuals at different PRS deciles showed differential trajectories of 5-year and cumulative absolute risk. The age reaching the LDCT screening recommendation threshold can vary by 4 to 8 years, depending on the individual's genetic background, smoking status, and family history. Collectively, these results suggest that individual's genetic background may inform the optimal lung cancer LDCT screening strategy. SIGNIFICANCE: Three large-scale datasets reveal that, after accounting for risk factors, an individual's genetics can affect their lung cancer risk trajectory, thus may inform the optimal timing for LDCT screening.

Circulating carotenoids and breast cancer among high-risk individuals

BACKGROUND

Carotenoids represent 1 of few modifiable factors to reduce breast cancer risk. Elucidation of interactions between circulating carotenoids and genetic predispositions or mammographic density (MD) may help inform more effective primary preventive strategies in high-risk populations.

OBJECTIVES

We tested whether women at high risk for breast cancer due to genetic predispositions or high MD would experience meaningful and greater risk reduction from higher circulating levels of carotenoids in a nested case-control study in the Nurses' Health Studies (NHS and NHSII).

METHODS

This study included 1919 cases and 1695 controls in a nested case-control study in the NHS and NHSII. We assessed both multiplicative and additive interactions. RR reductions and 95% CIs were calculated using unconditional logistic regressions, adjusting for matching factors and breast cancer risk factors. Absolute risk reductions (ARR) were calculated based on Surveillance, Epidemiology, and End Results incidence rates.

RESULTS

We showed that compared with women at low genetic risk or low MD, those with higher genetic risk scores or high MD had greater ARRs for breast cancer as circulating carotenoid levels increase (additive P-interaction = 0.05). Among women with a high polygenic risk score, those in the highest quartile of circulating carotenoids had a significant ARR (28.6%; 95% CI, 14.8-42.1%) compared to those in the lowest quartile of carotenoids. For women with a high percentage MD (≥50%), circulating carotenoids were associated with a 37.1% ARR (95% CI, 21.7-52.1%) when comparing the highest to the lowest quartiles of circulating carotenoids.

CONCLUSIONS

The inverse associations between circulating carotenoids and breast cancer risk appeared to be more pronounced in high-risk women, as defined by germline genetic makeup or MD.

Estimating Individualized Absolute Risk for Esophageal Squamous Cell Carcinoma: A Population-Based Study in High-Risk Areas of China

Background

Esophageal squamous cell carcinoma (ESCC) has a high incidence rate and poor prognosis. In this study, we aimed to develop a predictive model to estimate the individualized 5-year absolute risk for ESCC in Chinese populations living in the high-risk areas of China.

Methods

We developed a risk-predicting model based on the epidemiologic data from a population-based case-control study including 244 newly diagnosed ESCC patients and 1,220 healthy controls. Initially, we included easy-to-obtain risk factors to construct the model using the multivariable logistic regression analysis. The area under the ROC curves (AUC) with cross-validation methods was used to evaluate the performance of the model. Combined with local age- and sex-specific ESCC incidence and mortality rates, the model was then used to estimate the absolute risk of developing ESCC within 5 years.

Results

A relative risk model was established that included eight factors: age, sex, tobacco smoking, alcohol drinking, education, and dietary habits (intake of hot food, intake of pickled/salted food, and intake of fresh fruit). The relative risk model had good discrimination [AUC, 0.785; 95% confidence interval (CI), 0.749–0.821]. The estimated 5-year absolute risk of ESCC for individuals varied widely, from 0.0003% to 19.72% in the studied population, depending on the exposure to risk factors.

Conclusions

Our model based on readily identifiable risk factors showed good discriminative accuracy and strong robustness. And it could be applied to identify individuals with a higher risk of developing ESCC in the Chinese population, who might benefit from further targeted screening to prevent esophageal cancer.

The Development of Empirically Derived Australian Low-Risk Gambling Limits

This study derived a set of Australian low-risk gambling limits and explored the relative and absolute risk associated with exceeding these limits. Secondary analysis of population-representative Tasmanian and Australian Capital Territory (ACT) cross-sectional (11,597 respondents) and longitudinal studies (2027 respondents) was conducted. Balancing sensitivity and specificity, the limits were: gambling frequency of 20–30 times per year; gambling expenditure of AUD $380–$615 per year (USD $240–$388 per year); gambling expenditure comprising 0.83–1.68% of gross personal income; and two types of gambling activities per year. All limits, except number of activities, predicted subsequent harm, with limits related to gambling expenditure consistently the best-performing. Exceeding the limits generally conferred a higher degree of relative and absolute risk, with gamblers exceeding the limits being 3–20 times more likely to experience harm than those who do not, and having a 5–17% risk of experiencing harm. Only 7–12% of gamblers exceeding the limits actually experienced harm. Gambling consumption lower than the limits also conferred a considerable amount of harm. Using a relative risk method, this study derived similar limits from disparate Australian states and territories. These limits can serve as working guidelines for the consideration of researchers, clinicians, and policy makers, but need to be subject to further rigorous empirical investigation.

Clinical context for cancer risk of immunosuppressive agents used in dermatology

Dermatologic care of inflammatory skin conditions has been transformed over recent decades through the use of small molecules disease-modifying anti-rheumatic drugs and targeted biologic therapies. Alongside the tremendous benefit of these agents, concerns remain regarding possible side effects, particularly cancer risk. To improve guidance and counseling of patients with skin diseases who are considering treatment with such agents, this article reviews available information on the risk of malignancies in patients treated with these agents. When possible, this article adds clinical context to risk through a number needed to harm that estimates the number of patients a provider would need to treat with a given agent in 1 year to cause a single adverse outcome over time.

Inclusion of a gene-environment interaction between alcohol consumption and the aldehyde dehydrogenase 2 genotype in a risk prediction model for upper aerodigestive tract cancer in Japanese men

The well-known gene-environment interaction between alcohol consumption and aldehyde dehydrogenase 2 (ALDH2) genotype in upper aerodigestive tract cancer risk may improve our ability to identify high-risk subjects. Here, we developed and validated risk prediction models for this cancer in Japanese men and evaluated whether adding the gene-environment interaction to the model improved the predictive performance. We developed two case-cohort datasets in the Japan Public Health Center-based Prospective Study: one from subjects in the baseline survey for model development (108 cases and 4049 subcohort subjects) and the second from subjects in the 5-year follow-up survey for model validation (31 cases and 1527 subcohort subjects). We developed an environmental model including age, smoking status, and alcohol consumption, and a gene-environment interaction model including age, smoking status, and the combination of alcohol consumption and the ALDH2 genotype. We found a statistically significant gene-environment interaction for alcohol consumption and the ALDH2 genotype. The c-index for the gene-environment interaction model (0.71) was slightly higher than that for the environmental model (0.67). The values of integrated discrimination improvement and net reclassification improvement for the gene-environment interaction model were also slightly higher than those for the environmental model. Goodness-of-fit tests suggested that the models were well calibrated. Results from external model validation by the 5-year follow-up survey were consistent with those from the model development by the baseline survey. The addition of a gene-environment interaction to a lifestyle-based model might improve the performance to estimate the probability of developing upper aerodigestive tract cancer for Japanese men.

Outcome measures reported in abstracts of randomized controlled trials in leading clinical journals: A bibliometric study

BACKGROUNDS

The CONSORT for Abstracts checklist published in 2008 recommends that authors report effect size for their studies. Meanwhile, the FDA strongly recommends reporting both ratio and difference measures. However, the measures of effect used in recent clinical trial reports remain unknown. This study is aimed to reveal trends regarding the measures of effect of interventions described in abstracts of recent randomized controlled trials (RCTs) in leading journals.

METHODS

A bibliometric analysis of data was obtained by electronic searches. Human RCTs published in 2016 in the following five journals were searched using PubMed: Annals of Internal Medicine, British Medical Journal, Journal of American Medical Association, The Lancet, and New England Journal of Medicine. Main outcome is numbers of studies reporting each measure in their abstracts.

RESULTS

Among abstracts of 334 articles, measures most frequently used were relative risk alone (n = 169), followed by absolute risk alone (n = 92), and raw data alone (n = 58). Reporting of the following measures was relatively limited: both ratio and difference measures (n = 8), raw data with ratio measures (n = 5), and raw data with difference measures (n = 2). None of the studies reported raw data with both ratio and difference measures. Only 15 articles described multiple measures of effect in their abstracts.

CONCLUSIONS

More than half of the RCT abstracts published in the five leading journals in 2016 reported risk ratio alone to indicate effect size. Even abstracts in the five leading journals did not adhere fully to the CONSORT for Abstracts or FDA recommendations.

Risk Prediction Models for Head and Neck Cancer in the US Population From the INHANCE Consortium

Head and neck cancer (HNC) risk prediction models based on risk factor profiles have not yet been developed. We took advantage of the large database of the International Head and Neck Cancer Epidemiology (INHANCE) Consortium, including 14 US studies from 1981-2010, to develop HNC risk prediction models. Seventy percent of the data were used to develop the risk prediction models; the remaining 30% were used to validate the models. We used competing-risk models to calculate absolute risks. The predictors included age, sex, education, race/ethnicity, alcohol drinking intensity, cigarette smoking duration and intensity, and/or family history of HNC. The 20-year absolute risk of HNC was 7.61% for a 60-year-old woman who smoked more than 20 cigarettes per day for over 20 years, consumed 3 or more alcoholic drinks per day, was a high school graduate, had a family history of HNC, and was non-Hispanic white. The 20-year risk for men with a similar profile was 6.85%. The absolute risks of oropharyngeal and hypopharyngeal cancers were generally lower than those of oral cavity and laryngeal cancers. Statistics for the area under the receiver operating characteristic curve (AUC) were 0.70 or higher, except for oropharyngeal cancer in men. This HNC risk prediction model may be useful in promoting healthier behaviors such as smoking cessation or in aiding persons with a family history of HNC to evaluate their risks.

Designing clinical trials with (restricted) mean survival time endpoint: Practical considerations

BACKGROUND/AIMS

The difference in mean survival time, which quantifies the treatment effect in terms most meaningful to patients and retains its interpretability regardless of the shape of the survival distribution or the proportionality of the treatment effect, is an alternative endpoint that could be used more often as the primary endpoint to design clinical trials. The underuse of this endpoint is due to investigators' lack of familiarity with the test comparing the mean survival times and the lack of tools to facilitate trial design with this endpoint. The aim of this article is to provide investigators with insights and software to design trials with restricted mean survival time as the primary endpoint.

METHODS

A closed-form formula for the asymptotic power of the test of restricted mean survival time difference is presented. The effects of design parameters on power were evaluated for the mean survival time test and log-rank test. An R package which calculates the power or the sample size for user-specified parameter values and provides power plots for each design parameter is provided. The R package also calculates the probability that the restricted mean survival time is estimable for user-defined trial designs.

RESULTS

Under proportional hazards and late differences in survival, the power of the mean survival time test can approach that of the log-rank test if the restriction time is late. Under early differences, the power of the restricted mean survival time test is higher than that of the log-rank test. Duration of accrual and follow-up have little influence on the power of the restricted mean survival time test. The choice of restriction time, on the other hand, has a large impact on power. Because the power depends on the interplay among the design factors, plotting the relationship between each design parameter and power allows the users to select the designs most appropriate for their trial. When modification is necessary to ensure the difference in restricted mean survival time is estimable, the three available modifications all perform adequately in the scenarios studied.

CONCLUSION

The restricted mean survival time is a survival endpoint that is meaningful to investigators and to patients and at the same time requires less restrictive assumptions. The biggest challenge with this endpoint is selection of the restriction time. We recommend selecting a restriction time that is clinically relevant to the disease and the clinical setting of the trial of interest. The practical considerations and the R package provided in this work are readily available tools that researchers can use to design trials with restricted mean survival time as the primary endpoint.

Mortality risk comparing walking pace to handgrip strength and a healthy lifestyle: A UK Biobank study

AIMS

Brisk walking and a greater muscle strength have been associated with a longer life; whether these associations are influenced by other lifestyle behaviours, however, is less well known.

METHODS

Information on usual walking pace (self-defined as slow, steady/average, or brisk), dynamometer-assessed handgrip strength, lifestyle behaviours (physical activity, TV viewing, diet, alcohol intake, sleep and smoking) and body mass index was collected at baseline in 450,888 UK Biobank study participants. We estimated 10-year standardised survival for individual and combined lifestyle behaviours and body mass index across levels of walking pace and handgrip strength.

RESULTS

Over a median follow-up of 7.0 years, 3808 (1.6%) deaths in women and 6783 (3.2%) in men occurred. Brisk walkers had a survival advantage over slow walkers, irrespective of the degree of engagement in other lifestyle behaviours, except for smoking. Estimated 10-year survival was higher in brisk walkers who otherwise engaged in an unhealthy lifestyle compared to slow walkers who engaged in an otherwise healthy lifestyle: 97.1% (95% confidence interval: 96.9-97.3) vs 95.0% (94.6-95.4) in women; 94.8% (94.7-95.0) vs 93.7% (93.3-94.2) in men. Body mass index modified the association between walking pace and survival in men, with the largest survival benefits of brisk walking observed in underweight participants. Compared to walking pace, for handgrip strength there was more overlap in 10-year survival across lifestyle behaviours.

CONCLUSION

Except for smoking, brisk walkers with an otherwise unhealthy lifestyle have a lower mortality risk than slow walkers with an otherwise healthy lifestyle.

A Multi-Center Competing Risks Model and Its Absolute Risk Calculation Approach

In the competing risks frame, the cause-specific hazard model (CSHM) can be used to test the effects of some covariates on one particular cause of failure. Sometimes, however, the observed covariates cannot explain the large proportion of variation in the time-to-event data coming from different areas such as in a multi-center clinical trial or a multi-center cohort study. In this study, a multi-center competing risks model (MCCRM) is proposed to deal with multi-center survival data, then this model is compared with the CSHM by simulation. A center parameter is set in the MCCRM to solve the spatial heterogeneity problem caused by the latent factors, hence eliminating the need to develop different models for each area. Additionally, the effects of the exposure factors in the MCCRM are kept consistent for each individual, regardless of the area they inhabit. Therefore, the coefficient of the MCCRM model can be easily explained using the scenario of each model for each area. Moreover, the calculating approach of the absolute risk is given. Based on a simulation study, we show that the estimate of coefficients of the MCCRM is unbiased and precise, and the area under the curve (AUC) is larger than that of the CSHM when the heterogeneity cannot be ignored. Furthermore, the disparity of the AUC increases progressively as the standard deviation of the center parameter (SDCP) rises. In order to test the calibration, the expected number (E) of strokes is calculated and then compared with the corresponding observed number (O). The result is promising, so the SDCP can be used to select the most appropriate model. When the SDCP is less than 0.1, the performance of the MCCRM and CSHM is analogous, but when the SDCP is equal to or greater than 0.1, the performance of the MCCRM is significantly superior to the CSHM. This suggests that the MCCRM should be selected as the appropriate model.

Estimation of Relative and Absolute Risks in a Competing-Risks Setting Using a Nested Case-Control Study Design: Example From the ProMort Study

In this paper, we describe the Prognostic Factors for Mortality in Prostate Cancer (ProMort) study and use it to demonstrate how the weighted likelihood method can be used in nested case-control studies to estimate both relative and absolute risks in the competing-risks setting. ProMort is a case-control study nested within the National Prostate Cancer Register (NPCR) of Sweden, comprising 1,710 men diagnosed with low- or intermediate-risk prostate cancer between 1998 and 2011 who died from prostate cancer (cases) and 1,710 matched controls. Cause-specific hazard ratios and cumulative incidence functions (CIFs) for prostate cancer death were estimated in ProMort using weighted flexible parametric models and compared with the corresponding estimates from the NPCR cohort. We further drew 1,500 random nested case-control subsamples of the NPCR cohort and quantified the bias in the hazard ratio and CIF estimates. Finally, we compared the ProMort estimates with those obtained by augmenting competing-risks cases and by augmenting both competing-risks cases and controls. The hazard ratios for prostate cancer death estimated in ProMort were comparable to those in the NPCR. The hazard ratios for dying from other causes were biased, which introduced bias in the CIFs estimated in the competing-risks setting. When augmenting both competing-risks cases and controls, the bias was reduced.

Primary Absolute Cardiovascular Disease Risk and Prevention in Relation to Psychological Distress in the Australian Population: A Nationally Representative Cross-Sectional Study

People who experience psychological distress have an elevated risk of incident cardiovascular disease (CVD). However, the extent to which traditional CVD prevention strategies could be used to reduce the CVD burden in this group is unclear because population-level data on CVD risk profiles and appropriate management of risk in relation to distress are currently not available. The aim of this study was to use nationally representative data to quantify variation in CVD risk and appropriate management of risk according to level of psychological distress in the Australian population. Data were from 2,618 participants aged 45-74 years without prior CVD who participated in the 2011-12 Australian Health Survey, a cross-sectional and nationally representative study of Australian adults. Age-and sex-adjusted prevalence of 5-year absolute risk of primary CVD (low <10%, moderate 10-15%, or high >15%), CVD risk factors, blood-pressure, and cholesterol assessments, and appropriate treatment (combined blood pressure- and lipid-lowering medication) if at high primary risk, were estimated. Prevalence ratios (PR) quantified variation in these outcomes in relation to low (Kessler-10 score: 10-<12), mild (12-<16), moderate (16-<22) and high (22-50) psychological distress, after adjusting for sociodemographic characteristics. The prevalence of high absolute risk of primary CVD for low, mild, moderate and high distress was 10.9, 12.3, 11.4, and 18.6%, respectively, and was significantly higher among participants with high compared to low distress (adjusted PR:1.62, 95%CI:1.04-2.52). The prevalence of CVD risk factors was generally higher in those with higher psychological distress. Blood pressure and cholesterol assessments were reported by the majority of participants (>85%) but treatment of high absolute risk was low (<30%), and neither were related to psychological distress. Our findings confirm the importance of recognizing people who experience psychological distress as a high risk group and suggest that at least part of the excess burden of primary CVD events among people with high psychological distress could be reduced with an absolute risk approach to assessment and improved management of high primary CVD risk.

Regression analysis in an illness-death model with interval-censored data: A pseudo-value approach

Pseudo-values provide a method to perform regression analysis for complex quantities with right-censored data. A further complication, interval-censored data, appears when events such as dementia are studied in an epidemiological cohort. We propose an extension of the pseudo-value approach for interval-censored data based on a semi-parametric estimator computed using penalised likelihood and splines. This estimator takes interval-censoring and competing risks into account in an illness-death model. We apply the pseudo-value approach to three mean value parameters of interest in studies of dementia: the probability of staying alive and non-demented, the restricted mean survival time without dementia and the absolute risk of dementia. Simulation studies are conducted to examine properties of pseudo-values based on this semi-parametric estimator. The method is applied to the French cohort PAQUID, which included more than 3,000 non-demented subjects, followed for dementia for more than 25 years.

Thrombotic and Infectious Risks of Parenteral Nutrition in Hospitalized Pediatric Inflammatory Bowel Disease

BACKGROUND

Malnutrition is common in inflammatory bowel disease (IBD), requiring timely and sufficient nutritional supplementation. In patients hospitalized for active disease, symptoms and/or altered intestinal function hinder enteral nutrition feasibility. In this scenario, parenteral nutrition (PN) is used. We aimed (1) to assess the frequency of PN use between 1997 and 2012 among hospitalized pediatric patients with IBD, (2) to determine the risk of in-hospital thrombus and infection associated with PN, and (3) to identify predictors of thrombus and infection in pediatric IBD hospitalizations utilizing PN.

METHODS

We performed a cross-sectional analysis of pediatric patients hospitalized between 1997 and 2012. We used the Kids' Inpatient Database (KID) to identify pediatric patients (≤18 years of age) with Crohn's disease (CD) or ulcerative colitis (UC), PN exposure, and primary outcomes including thrombus and infection. We used multivariable regression to identify risk factors for outcomes of interest.

RESULTS

Parenteral nutrition was utilized in 3732 (12%) of 30,914 IBD hospitalizations. Three percent of PN patients experienced a thrombotic complication, and 5.5% experienced an infectious complication. Multivariate analysis showed PN as an independent risk factor for thrombus (odds ratio [OR], 4.3; 95% confidence interval [CI], 3.2-5.6) and infection (OR, 3.8; 95% CI, 3.1-4.6). Surgery was an independent risk factor for thrombus (OR, 2.0; 95% CI, 1.4-2.7) and infection (OR, 2.5; 95% CI, 2.0-3.1) in hospitalizations exposed to PN.

CONCLUSIONS

Hospitalized pediatric IBD patients, particularly surgical, receiving PN are at increased risk for thrombosis and infection. Clinicians must balance these risks with the benefits of PN.

Evaluation of 2 breast cancer risk models in a benign breast disease cohort

BACKGROUND

More than 1.5 million women per year have a benign breast biopsy resulting in concern about their future breast cancer (BC) risk. This study examined the performance of 2 BC risk models that integrate clinical and histologic findings in this population.

METHODS

The BC risk at 5 and 10 years was estimated with the Breast Cancer Surveillance Consortium (BCSC) and Benign Breast Disease to Breast Cancer (BBD-BC) models for women diagnosed with benign breast disease (BBD) at the Mayo Clinic from 1997 to 2001. Women with BBD were eligible for the BBD-BC model, but the BCSC model also required a screening mammogram. Calibration and discrimination were assessed.

RESULTS

Fifty-six cases of BC were diagnosed among the 2142 women with BBD (median age, 50 years) within 5 years (118 were diagnosed within 10 years). The BBD-BC model had slightly better calibration at 5 years (0.89; 95% confidence interval [CI], 0.71-1.21) versus 10 years (0.81; 95% CI, 0.70-1.00) but similar discrimination in the 2 time periods: 0.68 (95% CI, 0.60-0.75) and 0.66 (95% CI, 0.60-0.71), respectively. In contrast, among the 1089 women with screening mammograms (98 cases of BC within 10 years), the BCSC model had better calibration (0.94; 95% CI, 0.85-1.43) and discrimination (0.63; 95% CI, 0.56-0.71) at 10 years versus 5 years (calibration, 1.31; 95% CI, 0.94-2.25; discrimination, 0.59; 95% CI, 0.46-0.71) where discrimination was not different from chance.

CONCLUSIONS

The BCSC and BBD-BC models were validated in the Mayo BBD cohort, although their performance differed by 5-year risk versus 10-year risk. Further enhancement of these models is needed to provide accurate BC risk estimates for women with BBD.

Proportion and Characteristics of US Adults Who May Be Eligible From Additional Blood Pressure Lowering Based on Absolute Risk

BACKGROUND

The Systolic Blood Pressure Intervention Trial (SPRINT) and the Heart Outcomes Prevention Evaluation 3 (HOPE-3) trial demonstrated the merits of blood pressure (BP) lowering to reduce cardiovascular events in intermediate to high cardiovascular risk adults. However, the population impact of an absolute risk-based strategy for BP lowering remains unclear.

METHODS

We examined 3 treatment thresholds using the National Health and Nutrition Examination Survey. First, the JNC8 guideline was used to determine treatment goals. Second, adults with a systolic BP (SBP) of 130 mm Hg and high cardiovascular risk (based on eligibility for SPRINT) were considered eligible for additional BP lowering. Finally, we combined the treatment threshold for high-risk adults with an SBP treatment threshold of 140 mm Hg for intermediate-risk adults that met the eligibility criteria for HOPE-3.

RESULTS

Under the JNC8 guideline, 78.0% of adults ≥50 years were at target while 22.0% were eligible for additional BP lowering. If an SBP treatment threshold of 130 mm Hg was used for adults at high cardiovascular risk, 31.1% would be eligible for additional BP lowering (an increase of 8.1 million). If an SBP threshold of 140 mm Hg was additionally used for adults at intermediate risk, 31.4% of adults would be eligible for BP lowering (an increase of 8.3 million). The proportion of adults eligible for BP lowering with established coronary artery disease decreased with the risk-based strategies.

CONCLUSION

An absolute risk treatment strategy would modestly increase the proportion of adults eligible for BP lowering.

Comparisons of risk prediction methods using nested case-control data

Using both simulated and real datasets, we compared two approaches for estimating absolute risk from nested case-control (NCC) data and demonstrated the feasibility of using the NCC design for estimating absolute risk. In contrast to previously published results, we successfully demonstrated not only that data from a matched NCC study can be used to unbiasedly estimate absolute risk but also that matched studies give better statistical efficiency and classify subjects into more appropriate risk categories. Our result has implications for studies that aim to develop or validate risk prediction models. In addition to the traditional full cohort study and case-cohort study, researchers designing these studies now have the option of performing a NCC study with huge potential savings in cost and resources. Detailed explanations on how to obtain the absolute risk estimates under the proposed approach are given. Copyright © 2016 John Wiley & Sons, Ltd.

Individualized Absolute Risk Calculations for Persons with Multiple Chronic Conditions: Embracing Heterogeneity, Causality, and Competing Events

Approximately 75% of adults over the age of 65 years are affected by two or more chronic medical conditions. We provide a conceptual justification for individualized absolute risk calculators for competing patient-centered outcomes (PCO) (i.e. outcomes deemed important by patients) and patient reported outcomes (PRO) (i.e. outcomes patients report instead of physiologic test results). The absolute risk of an outcome is the probability that a person receiving a given treatment will experience that outcome within a pre-defined interval of time, during which they are simultaneously at risk for other competing outcomes. This allows for determination of the likelihood of a given outcome with and without a treatment. We posit that there are heterogeneity of treatment effects among patients with multiple chronic conditions (MCC) largely depends on those coexisting conditions. We outline the development of an individualized absolute risk calculator for competing outcomes using propensity score methods that strengthen causal inference for specific treatments. Innovations include the key concept that any given outcome may or may not concur with any other outcome and that these competing outcomes do not necessarily preclude other outcomes. Patient characteristics and MCC will be the primary explanatory factors used in estimating the heterogeneity of treatment effects on PCO and PRO. This innovative method may have wide-spread application for determining individualized absolute risk calculations for competing outcomes. Knowing the probabilities of outcomes in absolute terms may help the burgeoning population of patients with MCC who face complex treatment decisions.

Men and women--similar but not identical: insights into LDL-lowering therapy in women from the Cholesterol Treatment Trialists Collaboration

Evaluation of: Fulcher et al. Efficacy and safety of LDL-lowering therapy among men and women: meta-analysis of individual data from 174,000 participants in 27 randomised trials. Lancet 385(9976), 1397-1405 (2015). A recent publication has explored the role of gender in determining the benefit from statins. Using data on 174,000 patients (including 46,000 women) collected up to 2010, a meta-analysis was performed using individual patient data, separately analyzing results for men and women and adjusting for baseline risk and nongender related risk factors. Over a median duration of follow-up of 4.9 years, statins reduced the risk of major vascular events by 21% for each mmol/l reduction of LDL cholesterol (relative risk: 0.79; 95% CI: 0.77-0.81; p < 0.001), reducing risk by 22% for men and 16% for women. There was no significant overall heterogeneity for the benefit achieved in men versus that achieved in women after adjusting for baseline risk. Baseline risk substantially affected the absolute number of events prevented, but did not affect the proportional benefit attributed to the use of statins. Total mortality was similarly and significantly reduced in men (10%) and women (9%). This study adds to existing literature in confirming that statins have demonstrable benefit in men and women.

Invited Commentary: Clinical Utility of Prediction Models for Rare Outcomes--The Example of Pancreatic Cancer

Translating relative risk estimates into absolute risks is important in evaluating the potential clinical and public health relevance of etiologic discoveries. Predicting high absolute risk is challenging, particularly for rare endpoints such as pancreatic cancer. Recent efforts to develop risk prediction models for pancreatic cancer have found moderate risk levels for very small parts of the population. A new approach in which clinical symptoms and medication use are evaluated in addition to information on risk factors is presented by Risch et al. in this issue of the Journal (Am J Epidemiol. 2015;182(1):26-34). The authors estimated absolute risks based on the relative risks obtained from their case-control study. Their absolute risk estimates were higher than those from previous approaches but remained restricted to a very small proportion of the general population. In the present commentary, we address issues of absolute risk stratification (particularly for rare diseases), specific analytic methods, and how actionable information will differ based on the disease and possible intervention. We suggest that moving from cancer-specific models to broader models used to predict risk for multiple outcomes can make risk prediction for rare diseases more effective. When considering translational goals, it is important to estimate absolute risk at the early stages of etiologic research. The results can be sobering but allow focusing on the most promising goals.

Invited commentary: How big is that interaction (in my community)--and in which direction?

In an accompanying article, Turner et al. (Am J Epidemiol. 2014;180(12):1145-1149) compare the joint effects of smoking and air pollution to make inferences about the reduction in lung cancer mortality achieved when reducing each exposure separately and when reducing both together. In this commentary, we use first principles to quantify the difference between the risk or mortality reduction obtained from reducing each of 2 exposures together and the sum of the risk differences obtained from reducing the 2 exposures separately. Metrics of the impact of joint effects or comparisons of joint effects presented in units of absolute risk, such as Rothman's I, can provide more meaningful quantitative measures of public health impact than unitless metrics (e.g., ratios) and standardized metrics (e.g., the population attributable fraction) of potential interventions for reducing smoking and air pollution exposure. In particular, the venerable attributable community risk metric can provide an estimate of the community impact of such interventions in units of absolute risk. A spreadsheet we provide demonstrates the calculation of the various metrics for hypothetical data similar to those reported by Turner et al. Using algebra, graphics, and examples, we show that positive interaction, or synergy, on the additive scale implies that the impact on risk reduction from a program that applies both interventions will be lesser than the sum of the impacts of the separate interventions.

Estimating Risk with Time-to-Event Data: An Application to the Women's Health Initiative

Accurate and individualized risk prediction is critical for population control of chronic diseases such as cancer and cardiovascular disease. Large cohort studies provide valuable resources for building risk prediction models, as the risk factors are collected at the baseline and subjects are followed over time until disease occurrence or termination of the study. However, for rare diseases the baseline risk may not be estimated reliably based on cohort data only, due to sparse events. In this paper, we propose to make use of external information to improve efficiency for estimating time-dependent absolute risk. We derive the relationship between external disease incidence rates and the baseline risk, and incorporate the external disease incidence information into estimation of absolute risks, while allowing for potential difference of disease incidence rates between cohort and external sources. The asymptotic properties, namely, uniform consistency and weak convergence, of the proposed estimators are established. Simulation results show that the proposed estimator for absolute risk is more efficient than that based on the Breslow estimator, which does not utilize external disease incidence rates. A large cohort study, the Women's Health Initiative Observational Study, is used to illustrate the proposed method.

Communicating Relative Risk Changes with Baseline Risk: Presentation Format and Numeracy Matter

BACKGROUND

Treatment benefits and harms are often communicated as relative risk reductions and increases, which are frequently misunderstood by doctors and patients. One suggestion for improving understanding of such risk information is to also communicate the baseline risk. We investigated 1) whether the presentation format of the baseline risk influences understanding of relative risk changes and 2) the mediating role of people's numeracy skills.

METHOD

We presented laypeople (N = 1234) with a hypothetical scenario about a treatment that decreased (Experiments 1a, 2a) or increased (Experiments 1b, 2b) the risk of heart disease. Baseline risk was provided as a percentage or a frequency. In a forced-choice paradigm, the participants' task was to judge the risk in the treatment group given the relative risk reduction (or increase) and the baseline risk. Numeracy was assessed using the Lipkus 11-item scale.

RESULTS

Communicating baseline risk in a frequency format facilitated correct understanding of a treatment's benefits and harms, whereas a percentage format often impeded understanding. For example, many participants misinterpreted a relative risk reduction as referring to an absolute risk reduction. Participants with higher numeracy generally performed better than those with lower numeracy, but all participants benefitted from a frequency format. Limitations are that we used a hypothetical medical scenario and a nonrepresentative sample.

CONCLUSIONS

Presenting baseline risk in a frequency format improves understanding of relative risk information, whereas a percentage format is likely to lead to misunderstandings. People's numeracy skills play an important role in correctly understanding medical information. Overall, communicating treatment benefits and harms in the form of relative risk changes remains problematic, even when the baseline risk is explicitly provided.

Assessing the goodness of fit of personal risk models

We describe a flexible family of tests for evaluating the goodness of fit (calibration) of a pre-specified personal risk model to the outcomes observed in a longitudinal cohort. Such evaluation involves using the risk model to assign each subject an absolute risk of developing the outcome within a given time from cohort entry and comparing subjects' assigned risks with their observed outcomes. This comparison involves several issues. For example, subjects followed only for part of the risk period have unknown outcomes. Moreover, existing tests do not reveal the reasons for poor model fit when it occurs, which can reflect misspecification of the model's hazards for the competing risks of outcome development and death. To address these issues, we extend the model-specified hazards for outcome and death, and use score statistics to test the null hypothesis that the extensions are unnecessary. Simulated cohort data applied to risk models whose outcome and mortality hazards agreed and disagreed with those generating the data show that the tests are sensitive to poor model fit, provide insight into the reasons for poor fit, and accommodate a wide range of model misspecification. We illustrate the methods by examining the calibration of two breast cancer risk models as applied to a cohort of participants in the Breast Cancer Family Registry. The methods can be implemented using the Risk Model Assessment Program, an R package freely available at http://stanford.edu/~ggong/rmap/.

Population-based absolute risk estimation with survey data

Absolute risk is the probability that a cause-specific event occurs in a given time interval in the presence of competing events. We present methods to estimate population-based absolute risk from a complex survey cohort that can accommodate multiple exposure-specific competing risks. The hazard function for each event type consists of an individualized relative risk multiplied by a baseline hazard function, which is modeled nonparametrically or parametrically with a piecewise exponential model. An influence method is used to derive a Taylor-linearized variance estimate for the absolute risk estimates. We introduce novel measures of the cause-specific influences that can guide modeling choices for the competing event components of the model. To illustrate our methodology, we build and validate cause-specific absolute risk models for cardiovascular and cancer deaths using data from the National Health and Nutrition Examination Survey. Our applications demonstrate the usefulness of survey-based risk prediction models for predicting health outcomes and quantifying the potential impact of disease prevention programs at the population level.

A regression model for risk difference estimation in population-based case-control studies clarifies gender differences in lung cancer risk of smokers and never smokers

BACKGROUND

Additive risk models are necessary for understanding the joint effects of exposures on individual and population disease risk. Yet technical challenges have limited the consideration of additive risk models in case-control studies.

METHODS

Using a flexible risk regression model that allows additive and multiplicative components to estimate absolute risks and risk differences, we report a new analysis of data from the population-based case-control Environment And Genetics in Lung cancer Etiology study, conducted in Northern Italy between 2002-2005. The analysis provides estimates of the gender-specific absolute risk (cumulative risk) for non-smoking- and smoking-associated lung cancer, adjusted for demographic, occupational, and smoking history variables.

RESULTS

In the multiple-variable lexpit regression, the adjusted 3-year absolute risk of lung cancer in never smokers was 4.6 per 100,000 persons higher in women than men. However, the absolute increase in 3-year risk of lung cancer for every 10 additional pack-years smoked was less for women than men, 13.6 versus 52.9 per 100,000 persons.

CONCLUSIONS

In a Northern Italian population, the absolute risk of lung cancer among never smokers is higher in women than men but among smokers is lower in women than men. Lexpit regression is a novel approach to additive-multiplicative risk modeling that can contribute to clearer interpretation of population-based case-control studies.

Cumulative incidence of cancer after solid organ transplantation

BACKGROUND

Solid organ transplantation recipients have elevated cancer incidence. Estimates of absolute cancer risk after transplantation can inform prevention and screening.

METHODS

The Transplant Cancer Match Study links the US transplantation registry with 14 state/regional cancer registries. The authors used nonparametric competing risk methods to estimate the cumulative incidence of cancer after transplantation for 2 periods (1987-1999 and 2000-2008). For recipients from 2000 to 2008, the 5-year cumulative incidence, stratified by organ, sex, and age at transplantation, was estimated for 6 preventable or screen-detectable cancers. For comparison, the 5-year cumulative incidence was calculated for the same cancers in the general population at representative ages using Surveillance, Epidemiology, and End Results data.

RESULTS

Among 164,156 recipients, 8520 incident cancers were identified. The absolute cancer risk was slightly higher for recipients during the period from 2000 to 2008 than during the period from 1987 to 1999 (5-year cumulative incidence: 4.4% vs. 4.2%; P = .006); this difference arose from the decreasing risk of competing events (5-year cumulative incidence of death, graft failure, or retransplantation: 26.6% vs. 31.9%; P < .001). From 2000 to 2008, the 5-year cumulative incidence of non-Hodgkin lymphoma was highest at extremes of age, especially in thoracic organ recipients (ages 0-34 years: range, 1.74%-3.28%; aged >50 years; range, 0.36%-2.22%). For recipients aged >50 years, the 5-year cumulative incidence was higher for colorectal cancer (range, 0.33%-1.94%) than for the general population at the recommended screening age (aged 50 years: range, 0.25%-0.33%). For recipients aged >50 years, the 5-year cumulative incidence was high for lung cancer among thoracic organ recipients (range, 1.16%-3.87%) and for kidney cancer among kidney recipients (range, 0.53%-0.84%). The 5-year cumulative incidence for prostate cancer and breast cancer was similar or lower in transplantation recipients than at the recommended ages of screening in the general population.

CONCLUSIONS

Subgroups of transplantation recipients have a high absolute risk of some cancers and may benefit from targeted prevention or screening.

Absolute risk regression for competing risks: interpretation, link functions, and prediction

In survival analysis with competing risks, the transformation model allows different functions between the outcome and explanatory variables. However, the model's prediction accuracy and the interpretation of parameters may be sensitive to the choice of link function. We review the practical implications of different link functions for regression of the absolute risk (or cumulative incidence) of an event. Specifically, we consider models in which the regression coefficients β have the following interpretation: The probability of dying from cause D during the next t years changes with a factor exp(β) for a one unit change of the corresponding predictor variable, given fixed values for the other predictor variables. The models have a direct interpretation for the predictive ability of the risk factors. We propose some tools to justify the models in comparison with traditional approaches that combine a series of cause-specific Cox regression models or use the Fine-Gray model. We illustrate the methods with the use of bone marrow transplant data.

Personalized estimates of breast cancer risk in clinical practice and public health

This paper defines absolute risk and some of its properties, and presents applications in breast cancer counseling and prevention. For counseling, estimates of absolute risk give useful perspective and can be used in management decisions that require weighing risks and benefits, such as whether or not to take tamoxifen to prevent breast cancer. Absolute risk models are also useful in designing intervention trials to prevent breast cancer and in assessing the potential reductions in absolute risk of disease that might result from reducing exposures that are associated with breast cancer. In these applications, it is important that the risk model be well calibrated, namely that it accurately predicts the numbers of women who will develop breast cancer in various subsets of the population. Absolute risk models are also needed to implement a 'high risk' prevention strategy that identifies a high-risk subset of the population and focuses intervention efforts on that subset. The limitations of the high-risk strategy are discussed, including the need for risk models with high discriminatory accuracy, and the need for less toxic interventions that can reduce the threshold of risk above which the intervention provides a net benefit. I also discuss the potential use of risk models in allocating prevention resources under cost constraints. High discriminatory accuracy of the risk model, in addition to good calibration, is desirable in this application, and the risk assessment should not be expensive in comparison with the intervention.

Inherited genetic markers discovered to date are able to identify a significant number of men at considerably elevated risk for prostate cancer

BACKGROUND

Prostate cancer (PCa) risk-associated single-nucleotide polymorphisms (SNPs) are continuously being discovered. Their ability to identify men at high risk and the impact of increasing numbers of SNPs on predictive performance are not well understood.

METHODS

Absolute risk for PCa was estimated in a population-based case-control study in Sweden (2,899 cases and 1,722 controls) using family history and three sets of sequentially discovered PCa risk-associated SNPs. Their performance in predicting PCa was assessed by positive predictive values (PPV) and sensitivity.

RESULTS

SNPs and family history were able to differentiate individual risk for PCa and identify men at higher risk; ∼18% and ∼8% of men in the study had 20-year (55-74 years) absolute risks that were twofold (0.24) or threefold (0.36) greater than the population median risk (0.12), respectively. When predictive performances were compared at absolute risk cutoffs of 0.12, 0.24, or 0.36, PPV increased considerably (∼20%, ∼30%, and ∼37%, respectively) while sensitivity decreased considerably (∼55%, ∼20%, and ∼10%, respectively). In contrast, when increasing numbers of SNPs (5, 11, and 28 SNPs) were used in risk prediction, PPV approached a constant value while sensitivity increased steadily.

CONCLUSIONS

SNPs discovered to date are suitable for risk prediction while additional SNPs discovered in the future may identify more subjects at higher risk. Men identified as high risk by SNP-based testing may be targeted for PCa screening or chemoprevention. The clinical impact on improving the effectiveness of these interventions can be and should be assessed.

Evaluating health risk models

Interest in targeted disease prevention has stimulated development of models that assign risks to individuals, using their personal covariates. We need to evaluate these models and quantify the gains achieved by expanding a model to include additional covariates. This paper reviews several performance measures and shows how they are related. Examples are used to show that appropriate performance criteria for a risk model depend upon how the model is used. Application of the performance measures to risk models for hypothetical populations and for US women at risk of breast cancer illustrate two additional points. First, model performance is constrained by the distribution of risk-determining covariates in the population. This complicates the comparison of two models when applied to populations with different covariate distributions. Second, all summary performance measures obscure model features of relevance to its utility for the application at hand, such as performance in specific subgroups of the population. In particular, the precision gained by adding covariates to a model can be small overall, but large in certain subgroups. We propose new ways to identify these subgroups and to quantify how much they gain by measuring the additional covariates. Those with largest gains could be targeted for cost-efficient covariate assessment.

Logistic regression in estimates of femoral neck fracture by fall

The latest methods in estimating the probability (absolute risk) of osteoporotic fractures include several logistic regression models, based on qualitative risk factors plus bone mineral density (BMD), and the probability estimate of fracture in the future. The Slovak logistic regression model, in contrast to other models, is created from quantitative variables of the proximal femur (in International System of Units) and estimates the probability of fracture by fall.

OBJECTIVES

The first objective of this study was to order selected independent variables according to the intensity of their influence (statistical significance) upon the occurrence of values of the dependent variable: femur strength index (FSI). The second objective was to determine, using logistic regression, whether the odds of FSI acquiring a pathological value (femoral neck fracture by fall) increased or declined if the value of the variables (T-score total hip, BMI, alpha angle, theta angle and HAL) were raised by one unit.

PATIENTS AND METHODS

Bone densitometer measurements using dual energy X-ray absorptiometry (DXA), (Prodigy, Primo, GE, USA) of the left proximal femur were obtained from 3 216 East Slovak women with primary or secondary osteoporosis or osteopenia, aged 20-89 years (mean age 58.9; 95% CI: -58.42; 59.38). The following variables were measured: FSI, T-score total hip BMD, body mass index (BMI), as were the geometrical variables of proximal femur alpha angle (α angle), theta angle (θ angle), and hip axis length (HAL).

STATISTICAL ANALYSIS

Logistic regression was used to measure the influence of the independent variables (T-score total hip, alpha angle, theta angle, HAL, BMI) upon the dependent variable (FSI).

RESULTS

The order of independent variables according to the intensity of their influence (greatest to least) upon the occurrence of values of the dependent FSI variable was found to be: BMI, theta angle, T-score total hip, alpha angle, and HAL. An increase of one unit of an independent variable was shown, with statistical significance, to either raise or decrease the odds of the dependent FSI variable. Specific findings were as follows: an increase by 1° of the α angle escalated the probability of FSI acquiring a pathological value by 1111 times; an increase by 1° of the θ angle was found to boost these odds 1231 times; an increase by 1 mm of the HAL was found to increase these odds by 1043 times; an increase by 1.0 kg/m(2) of the BMI raised the odds 1302 times; an increase by +1 standard deviation of the value of the T-score total hip subsequently decreased these odds 198 times.

CONCLUSION

The equation of the Slovak regression model makes it possible in praxis to determine the probability or absolute risk of femoral neck fracture by fall at those densitometrical workplaces without a program for measuring the FSI variable.

Applying the Lorenz curve to disease risk to optimize health benefits under cost constraints

This paper shows how the Lorenz curve can be used, together with models of disease risk, to allocate scarce resources so as to optimize a health benefit. Consider the example of breast cancer mortality. If there were sufficient resources to provide all women with mammograms, a certain maximal number of lives could be saved. Suppose, however, that only a fraction of that amount of money is available for prevention activities. Suppose that a questionnaire could be given to assess a woman's risk of dying of breast cancer. Depending on the amount of money available, on the ratio of the cost of a questionnaire to the cost of a mammogram, and on the Lorenz curve of the distribution of risks of breast cancer mortality, I calculate the proportion of women who should be given questionnaires, the proportion of women given the questionnaires who should be given mammograms because they have high risks, and the proportion of women not given questionnaires who should be assigned to receive mammograms at random so as to maximize the number of lives saved.

Effects of information framing on the intentions of family physicians to prescribe long-term hormone replacement therapy

OBJECTIVE

To determine whether the way in which information on benefits and harms of long-term hormone replacement therapy (HRT) is presented influences family physicians' intentions to prescribe this treatment.

DESIGN

Family physicians were randomized to receive information on treatment outcomes expressed in relative terms, or as the number needing to be treated (NNT) with HRT to prevent or cause an event. A control group received no information.

SETTING

Primary care.

PARTICIPANTS

Family physicians practicing in the Hunter Valley, New South Wales, Australia.

INTERVENTION

Estimates of the impact of long-term HRT on risk of coronary events, hip fractures, and breast cancer were summarized as relative (proportional) decreases or increases in risk, or as NNT.

MEASUREMENTS AND MAIN RESULTS

Intention to prescribe HRT for seven hypothetical patients was measured on Likert scales. Of 389 family physicians working in the Hunter Valley, 243 completed the baseline survey and 215 participated in the randomized trial. Baseline intention to prescribe varied across patients-it was highest in the presence of risk factors for hip fracture, but coexisting risk factors for breast cancer had a strong negative influence. Overall, a larger proportion of subjects receiving information expressed as NNT had reduced intentions, and a smaller proportion had increased intentions to prescribe HRT than those receiving the information expressed in relative terms, or the control group. However, the differences were small and only reached statistical significance for three hypothetical patients. Framing effects were minimal when the hypothetical patient had coexisting risk factors for breast cancer.

CONCLUSIONS

Information framing had some effect on family physicians' intentions to prescribe HRT, but the effects were smaller than those previously reported, and they were modified by the presence of serious potential adverse treatment effects.