Comparison of baseline and repeated measure covariate techniques in the Framingham Heart Study

Prognostic models for short-term annual risk of severe complications and mortality in patients living with type 2 diabetes using a national medical claim database

OBJECTIVE

Prognostic models in patients living with diabetes allow physicians to estimate individual risk based on medical records and biological results. Clinical risk factors are not always all available to evaluate these models so that they may be complemented with models from claims databases. The objective of this study was to develop, validate and compare models predicting the annual risk of severe complications and mortality in patients living with type 2 diabetes (T2D) from a national claims data.

RESEARCH DESIGN AND METHODS

Adult patients with T2D were identified in a national medical claims database through their history of treatments or hospitalizations. Prognostic models were developed using logistic regression (LR), random forest (RF) and neural network (NN) to predict annual risk of outcome: severe cardiovascular (CV) complications, other severe T2D-related complications, and all-cause mortality. Risk factors included demographics, comorbidities, the adjusted Diabetes Severity and Comorbidity Index (aDSCI) and diabetes medications. Model performance was assessed using discrimination (C-statistics), balanced accuracy, sensibility and specificity.

RESULTS

A total of 22,708 patients with T2D were identified, with mean age of 68 years and average duration of T2D of 9.7 years. Age, aDSCI, disease duration, diabetes medications and chronic cardiovascular disease were the most important predictors for all outcomes. Discrimination with C-statistic ranged from 0.715 to 0.786 for severe CV complications, from 0.670 to 0.847 for other severe complications and from 0.814 to 0.860 for all-cause mortality, with RF having consistently the highest discrimination.

CONCLUSION

The proposed models reliably predict severe complications and mortality in patients with T2D, without requiring medical records or biological measures. These predictions could be used by payers to alert primary care providers and high-risk patients living with T2D.

A Scalable Risk Scoring System for COVID-19 Inpatients Based on Consumer-grade Wearables: Statistical Analysis and Model Development

Background

To provide effective care for inpatients with COVID-19, clinical practitioners need systems that monitor patient health and subsequently allow for risk scoring. Existing approaches for risk scoring in patients with COVID-19 focus primarily on intensive care units (ICUs) with specialized medical measurement devices but not on hospital general wards.

Objective

In this paper, we aim to develop a risk score for inpatients with COVID-19 in general wards based on consumer-grade wearables (smartwatches).

Methods

Patients wore consumer-grade wearables to record physiological measurements, such as the heart rate (HR), heart rate variability (HRV), and respiration frequency (RF). Based on Bayesian survival analysis, we validated the association between these measurements and patient outcomes (ie, discharge or ICU admission). To build our risk score, we generated a low-dimensional representation of the physiological features. Subsequently, a pooled ordinal regression with time-dependent covariates inferred the probability of either hospital discharge or ICU admission. We evaluated the predictive performance of our developed system for risk scoring in a single-center, prospective study based on 40 inpatients with COVID-19 in a general ward of a tertiary referral center in Switzerland.

Results

First, Bayesian survival analysis showed that physiological measurements from consumer-grade wearables are significantly associated with patient outcomes (ie, discharge or ICU admission). Second, our risk score achieved a time-dependent area under the receiver operating characteristic curve (AUROC) of 0.73-0.90 based on leave-one-subject-out cross-validation.

Conclusions

Our results demonstrate the effectiveness of consumer-grade wearables for risk scoring in inpatients with COVID-19. Due to their low cost and ease of use, consumer-grade wearables could enable a scalable monitoring system.

Trial Registration

Clinicaltrials.gov NCT04357834; https://www.clinicaltrials.gov/ct2/show/NCT04357834

A Validated Risk Prediction Model for Breast Cancer in US Black Women

PURPOSE

Breast cancer risk prediction models are used to identify high-risk women for early detection, targeted interventions, and enrollment into prevention trials. We sought to develop and evaluate a risk prediction model for breast cancer in US Black women, suitable for use in primary care settings.

METHODS

Breast cancer relative risks and attributable risks were estimated using data from Black women in three US population-based case-control studies (3,468 breast cancer cases; 3,578 controls age 30-69 years) and combined with SEER age- and race-specific incidence rates, with incorporation of competing mortality, to develop an absolute risk model. The model was validated in prospective data among 51,798 participants of the Black Women's Health Study, including 1,515 who developed invasive breast cancer. A second risk prediction model was developed on the basis of estrogen receptor (ER)-specific relative risks and attributable risks. Model performance was assessed by calibration (expected/observed cases) and discriminatory accuracy (C-statistic).

RESULTS

The expected/observed ratio was 1.01 (95% CI, 0.95 to 1.07). Age-adjusted C-statistics were 0.58 (95% CI, 0.56 to 0.59) overall and 0.63 (95% CI, 0.58 to 0.68) among women younger than 40 years. These measures were almost identical in the model based on estrogen receptor-specific relative risks and attributable risks.

CONCLUSION

Discriminatory accuracy of the new model was similar to that of the most frequently used questionnaire-based breast cancer risk prediction models in White women, suggesting that effective risk stratification for Black women is now possible. This model may be especially valuable for risk stratification of young Black women, who are below the ages at which breast cancer screening is typically begun.

A shared frailty model for multivariate longitudinal data on adverse event of radiation therapy

Oral mucositis is an inflammatory adverse event when treating head and neck cancer patients with radiation therapy (RT). The severity of its occurrence is believed to mainly depend on its site and the distribution of a cumulative radiation dose in the mouth area. The motivating study investigating differences in radiosensitivities (mucositis progression) at distinct sites where the severity of mucositis is assessed regularly at eight distinct sites on an ordinal scale results in multivariate longitudinal data and thus poses certain challenges. To deal with the multivariate longitudinal data in this particular setting, we take a time-to-event approach focusing on the first occurrence of severe mucositis at the distinct sites using the fact that the site-specific cumulative radiation dose thought to be the main driver of oral mucositis develops over time. Thereby, we may address multivariate longitudinal processes in a simpler and more compact fashion. In this article, to find out differences in mucositis progression at eight distinct sites we propose a shared frailty model for multivariate parallel processes within individuals. The shared frailty model directly incorporating 'process indicators' as covariates turns out to adequately explain the differences in the parallel processes (here, mucositis progressions at distinct sites) while taking individual effects into account. The parallel result with the one from the previous analysis based on the same data but conducted with an alternative statistical methodology shows adequacy of the proposed approach.

The effect of gestational diabetes mellitus on the risk of asthma in offspring

PURPOSE

To examine the effect of maternal gestational diabetes mellitus on the risk of asthma in the offspring.

METHODS

This cohort study used data from 19,933 children in the National Longitudinal Survey of Children and Youth (NLSCY), 1994/1995-2008/2009, Canada. Children were followed until the first-time report of having health professional-diagnosed asthma (hereafter incident asthma), loss to follow-up, or end of the NLSCY follow-up, whichever occurred first. As a surrogate for Cox proportional hazards regression, pooled logistic regression models, crude and adjusted for potential confounders, were fitted to estimate the effect of gestational diabetes mellitus on the risk of asthma in the offspring.

RESULTS

Among the 19,933 children, 1,178 (5.9%) had mothers with gestational diabetes mellitus. The median duration of follow-up was 4 (interquartile range: 4) years. A total of 1639 children in the cohort had reported incident asthma during the follow-up, and 119 of them had mothers with gestational diabetes mellitus. The adjusted hazard ratio for the association between gestational diabetes mellitus and incident asthma in offspring was 1.25 (95% confidence interval [CI] 1.03, 1.51).

CONCLUSIONS

Our findings suggest that gestational diabetes mellitus increases the risk of asthma in the offspring.

Novel Risk Modeling Approach of Atrial Fibrillation With Restricted Mean Survival Times: Application in the Framingham Heart Study Community-Based Cohort

BACKGROUND

Risk prediction models for atrial fibrillation (AF) do not give information about when AF might develop. Restricted mean survival time (RMST) quantifies risk into the time domain. Our objective was to use RMST to re-express individualized AF risk predictions.

METHODS AND RESULTS

We included AF-free participants from the Framingham Heart Study community-based cohorts. We predicted new-onset AF over 10-year follow-up according to baseline covariates: age, height, weight, systolic blood pressure, diastolic blood pressure, current smoking, antihypertensive treatment, diabetes mellitus, prevalent heart failure, and prevalent myocardial infarction. First, we fitted a Cox regression model and estimated the 10-year predicted risk of AF. Second, we fitted an RMST model and estimated the predicted mean time free of AF and alive over a time horizon of 10 years. We included 7586 AF-free participants contributing to 11 088 examinations (mean age 61±11 years, 44% were men). During 10-year follow-up, 822 participants developed AF. The Cox and RMST models were in agreement regarding the direction, strength, and statistical significance of associations for all covariates. Low (<5%), intermediate (5%-15%), and high (>15%) 10-year predicted risk of AF corresponded to predicted mean time alive and free of AF of 9.9, 9.6, and 8.8 years, respectively. A 60-year-old woman with a body mass index of 25 kg/m2, no use of hypertension treatment and no history of heart failure had a predicted mean time alive and free of AF of 9.9 years, whereas a 70-year-old man with a body mass index of 30 kg/m2, use of hypertension treatment, and with prevalent heart failure had a predicted mean time alive and free of AF of 7.9 years.

CONCLUSIONS

The RMST can be used to develop risk prediction models to express results in a time scale. RMST may offer a complementary risk communication tool for AF in clinical practice.

Prediagnosis social support, social integration, living status, and colorectal cancer mortality in postmenopausal women from the women's health initiative

BACKGROUND

We evaluated associations between perceived social support, social integration, living alone, and colorectal cancer (CRC) outcomes in postmenopausal women.

METHODS

The study included 1431 women from the Women's Health Initiative who were diagnosed from 1993 through 2017 with stage I through IV CRC and who responded to the Medical Outcomes Study Social Support survey before their CRC diagnosis. We used proportional hazards regression to evaluate associations of social support (tertiles) and types of support, assessed up to 6 years before diagnosis, with overall and CRC-specific mortality. We also assessed associations of social integration and living alone with outcomes also in a subset of 1141 women who had information available on social ties (marital/partner status, community and religious participation) and living situation.

RESULTS

In multivariable analyses, women with low (hazard ratio [HR], 1.52; 95% CI, 1.23-1.88) and moderate (HR, 1.21; 95% CI, 0.98-1.50) perceived social support had significantly higher overall mortality than those with high support (P [continuous] < .001). Similarly, women with low (HR, 1.42; 95% CI, 1.07-1.88) and moderate (HR, 1.28; 95% CI, 0.96-1.70) perceived social support had higher CRC mortality than those with high social support (P [continuous] = .007). Emotional, informational, and tangible support and positive interaction were all significantly associated with outcomes, whereas affection was not. In main-effects analyses, the level of social integration was related to overall mortality (P for trend = .02), but not CRC mortality (P for trend = .25), and living alone was not associated with mortality outcomes. However, both the level of social integration and living alone were related to outcomes in patients with rectal cancer.

CONCLUSIONS

Women with low perceived social support before diagnosis have higher overall and CRC-specific mortality.

Identifying Attrition Phases in Survey Data: Applicability and Assessment Study

BACKGROUND

Although Web-based questionnaires are an efficient, increasingly popular mode of data collection, their utility is often challenged by high participant dropout. Researchers can gain insight into potential causes of high participant dropout by analyzing the dropout patterns.

OBJECTIVE

This study proposed the application of and assessed the use of user-specified and existing hypothesis testing methods in a novel setting-survey dropout data-to identify phases of higher or lower survey dropout.

METHODS

First, we proposed the application of user-specified thresholds to identify abrupt differences in the dropout rate. Second, we proposed the application of 2 existing hypothesis testing methods to detect significant differences in participant dropout. We assessed these methods through a simulation study and through application to a case study, featuring a questionnaire addressing decision-making surrounding cancer screening.

RESULTS

The user-specified method set to a low threshold performed best at accurately detecting phases of high attrition in both the simulation study and test case application, although all proposed methods were too sensitive.

CONCLUSIONS

The user-specified method set to a low threshold correctly identified the attrition phases. Hypothesis testing methods, although sensitive at times, were unable to accurately identify the attrition phases. These results strengthen the case for further development of and research surrounding the science of attrition.

Refining the Association Between Body Mass Index and Atrial Fibrillation: G-Formula and Restricted Mean Survival Times

Background

Previous studies assessing the association between body mass index (BMI) and atrial fibrillation (AF) did not account for time‐varying covariates, which may be affected by previous BMI. We illustrate how the g‐formula can account for time‐varying confounding.

Methods and Results

We included 4392 participants from the Framingham Heart Study who were AF free at ages 45 to 55 years, and followed them for up to 20 years. We estimated hazard ratios (HRs) comparing time‐varying nonobese versus obese with Cox models. We used the g‐formula to compare nonobese versus obese and 10% annual decrease in BMI (until normal weight is reached) versus natural course. We estimated HRs and differences in restricted mean survival times, the mean difference in time alive and AF free. We adjusted for sex, age, and time‐varying risk factors. Cox models indicated that nonobese participants had a decreased rate of AF versus obese participants (HR, 0.83; 95% CI, 0.72–0.97). G‐formula analyses comparing everyone had they been nonobese versus obese yielded stronger associations (HR, 0.73; 95% CI, 0.58–0.91). The restricted mean survival time was 19.22 years had everyone been nonobese and 19.03 years had everyone been obese (difference, 2.25 months; 95% CI, −0.66 to 5.16). When assessing a 10% annual decrease in BMI, the association was weaker (HR 0.96; 95% CI, 0.86–1.08).

Conclusions

Decreased BMI was associated with a lower rate of AF after accounting for time‐varying covariates that depend on previous exposure using the g‐formula, which Cox models cannot accommodate. Absolute measures like the restricted mean survival time difference offer context to relative measures of association.

Natural History of Obesity Subphenotypes: Dynamic Changes Over Two Decades and Prognosis in the Framingham Heart Study

CONTEXT

The natural histories of obesity subphenotypes are incompletely delineated.

OBJECTIVES

To investigate dynamic changes in obesity subphenotypes and associations with outcomes.

DESIGN, SETTING, PARTICIPANTS, AND MEASUREMENTS

Framingham Offspring Cohort participants (n = 4291) who attended the examination cycles 2 (1979 to 1983) to 7 (1998 to 2001), which included 26,508 participant observations. Obesity subphenotypes [metabolically healthy nonobese (MHNO), metabolically healthy obese (MHO), metabolically unhealthy nonobese (MUNO), and metabolically unhealthy obese (MUO)] were ascertained based on metabolic health (<2 Adult Treatment Panel III criteria). The outcomes were subclinical cardiovascular disease (CVD), incident diseases [diabetes, hypertension, chronic kidney disease (CKD), CVD], and all-cause mortality.

RESULTS

At baseline, 4% and 31% of participants exhibited the MHO and MUNO subphenotypes, respectively. Four-year probability of MHO participants becoming MUO was 43% in women and 46% in men. Compared with MHNO, MHO participants had 1.28-fold (95% CI, 0.85 to 1.93) and 1.92-fold (95% CI, 1.38 to 2.68) higher odds of subclinical CVD and coronary artery calcification, respectively; corresponding values for MUNO were 1.95 (1.54 to 2.47) and 1.92 (1.38 to 2.68). During follow-up (median of 14 years), 231 participants developed diabetes, 784 hypertension, 423 CKD, 639 CVD, and 1296 died. Compared with MHNO, MHO conferred higher risks of diabetes [hazard ratio (HR), 4.69; 95% CI, 2.21 to 9.96] and hypertension (HR, 2.21; 95% CI, 1.66 to 2.94). Compared with MUO, MHO conferred lower risks of diabetes (0.21; 0.12 to 0.39), CVD (0.64; 0.43 to 0.95), and CKD (0.44; 0.27 to 0.73), but similar hypertension, cardiovascular mortality, and overall mortality risks.

CONCLUSION

Over time, most MHO participants developed metabolic abnormalities and clinical disease. The MHO subphenotype is a harbinger of future risk.

Body Mass Index and Risks of Incident Ischemic Stroke Subtypes: The Japan Public Health Center-Based Prospective (JPHC) Study

BACKGROUND

The association of body mass index (BMI) with risks of ischemic stroke subtypes have not been established.

METHODS

Cumulative average BMI was calculated using self-reported body weight and height obtained from baseline (Cohort I in 1990, and Cohort II from 1993-1994) and 5- and 10-year questionnaire surveys of Japan Public Health Center-based prospective (JPHC) study. A total of 42,343 men and 46,413 women aged 40-69 years were followed-up for the incidence of lacunar, large-artery occlusive, and cardioembolic strokes. A sub-distribution hazard model was used to estimate sub-distribution hazard ratios (SHRs) and the 95% confidence intervals (CIs).

RESULTS

During a median of 20.0 years of follow-up, we documented 809 and 481 lacunar, 395 and 218 large-artery occlusive, and 568 and 298 cardioembolic strokes in men and women, respectively. After adjustment for baseline age, updated smoking, alcohol consumption, leisure-time physical activity, and histories of hypertension, dyslipidemia, and diabetes mellitus, cumulative average BMI was positively linearly associated with lacunar (trend P = 0.007), large-artery occlusive (trend P = 0.002), and cardioembolic (trend P < 0.001) strokes in men, and with lacunar (trend P < 0.001) and large-artery occlusive (trend P = 0.003) strokes in women. There were approximately two-fold excess risk of cardioembolic stroke in both sexes and of lacunar and large-artery occlusive strokes in women for cumulative average BMI ≥30 kg/m² compared to BMI 23-<25 kg/m².

CONCLUSION

Cumulative average BMI showed a positive linear effect on sub-distribution hazards of lacunar, large-artery occlusive, and cardioembolic strokes in both sexes, except for cardioembolic stroke in women.

Development of predictive risk models for major adverse cardiovascular events among patients with type 2 diabetes mellitus using health insurance claims data

Background

There exist several predictive risk models for cardiovascular disease (CVD), including some developed specifically for patients with type 2 diabetes mellitus (T2DM). However, the models developed for a diabetic population are based on information derived from medical records or laboratory results, which are not typically available to entities like payers or quality of care organizations. The objective of this study is to develop and validate models predicting the risk of cardiovascular events in patients with T2DM based on medical insurance claims data.

Methods

Patients with T2DM aged 50 years or older were identified from the Optum™ Integrated Real World Evidence Electronic Health Records and Claims de-identified database (10/01/2006–09/30/2016). Risk factors were assessed over a 12-month baseline period and cardiovascular events were monitored from the end of the baseline period until end of data availability, continuous enrollment, or death. Risk models were developed using logistic regressions separately for patients with and without prior CVD, and for each outcome: (1) major adverse cardiovascular events (MACE; i.e., non-fatal myocardial infarction, non-fatal stroke, CVD-related death); (2) any MACE, hospitalization for unstable angina, or hospitalization for congestive heart failure; (3) CVD-related death. Models were developed and validated on 70% and 30% of the sample, respectively. Model performance was assessed using C-statistics.

Results

A total of 181,619 patients were identified, including 136,544 (75.2%) without prior CVD and 45,075 (24.8%) with a history of CVD. Age, diabetes-related hospitalizations, prior CVD diagnoses and chronic pulmonary disease were the most important predictors across all models. C-statistics ranged from 0.70 to 0.81, indicating that the models performed well. The additional inclusion of risk factors derived from pharmacy claims (e.g., use of antihypertensive, and use of antihyperglycemic) or from medical records and laboratory measures (e.g., hemoglobin A1c, urine albumin to creatinine ratio) only marginally improved the performance of the models.

Conclusion

The claims-based models developed could reliably predict the risk of cardiovascular events in T2DM patients, without requiring pharmacy claims or laboratory measures. These models could be relevant for providers and payers and help implement approaches to prevent cardiovascular events in high-risk diabetic patients.

Electronic supplementary material

The online version of this article (10.1186/s12933-018-0759-z) contains supplementary material, which is available to authorized users.

Trajectories in Leisure-Time Physical Activity and Risk of Stroke in Women in the California Teachers Study

BACKGROUND AND PURPOSE

Whether changes in leisure-time physical activity (LTPA) over time are associated with lower risk of stroke is not well established. We examined the association between changes in self-reported LTPA 10 years apart, with risk of incident stroke in the CTS (California Teachers Study). We hypothesized that the risk of stroke would be lowest among those who remained active.

METHODS

Sixty-one thousand two hundred and fifty-six CTS participants reported LTPA at 2 intensity levels (moderate and strenuous activity) at 2 time points (baseline 1995-96; 10-year follow-up 2005-2006). LTPA at each intensity level was categorized based on American Heart Association (AHA) recommendations (moderate, >150 minutes/week; strenuous, >75 minutes/week). Changes in LTPA were summarized as follows: (1) not meeting recommendations at both time points; (2) meeting recommendations only at follow-up; (3) meeting recommendations only at baseline; and (4) meeting recommendations at both time points. Incident strokes were identified through California state hospitalization records. Using multivariable Cox models, we examined the associations between changes in LTPA with incident stroke.

RESULTS

Nine hundred and eighty-seven women were diagnosed with stroke who completed both questionnaires. Meeting AHA recommendations at both the time points was associated with a lower risk of all stroke (adjusted hazard ratio, 0.84; 95% confidence interval, 0.72-0.98). The protective effects for stroke were driven by meeting AHA recommendations for moderate activity and largely observed for ischemic strokes (adjusted hazard ratio, 0.70; 95% confidence interval, 0.55-0.88).

CONCLUSIONS

Meeting AHA recommendations for moderate activity had a protective effect for reducing ischemic stroke risk. Participants who met AHA recommendations at baseline but not at follow-up, however, were not afforded reduced stroke risk.

Oral glucocorticoid use and osteonecrosis in children and adults with chronic inflammatory diseases: a population-based cohort study

OBJECTIVES

We studied oral glucocorticoids and osteonecrosis, a rare but serious bone disease, in individuals with various chronic inflammatory diseases. We hypothesised that we would find stronger associations in adults versus children and in people with autoimmune diseases.

DESIGN

Retrospective cohort study.

SETTING

Population-representative data (1994-2013) from general practices in the UK.

PARTICIPANTS

Children and adults diagnosed with asthma; inflammatory bowel disease; juvenile, psoriatic or rheumatoid arthritis; psoriasis; or systemic lupus.

EXPOSURES

Oral glucocorticoid patterns.

PRIMARY AND SECONDARY OUTCOME MEASURES

Diagnosed osteonecrosis (primary) and osteonecrosis plus clinical features (eg, symptoms, pain medication, surgical repair) (secondary). Discrete time failure models estimated the adjusted hazard ratio (aHR) of incident osteonecrosis following oral glucocorticoid exposure. Hypothesis testing was one sided (with corresponding 90% CI) since glucocorticoids were unlikely protective.

RESULTS

After adjusting for demographic, disease-related and health utilisation factors, glucocorticoid exposure was associated with osteonecrosis in adults (ages 18-49, aHR 2.1 (90% CI 1.5 to 2.9); ages ≥50, aHR 1.3 (90% CI 1.01 to 1.7)). However, low-dose glucocorticoids, corresponding to average doses <7.5 mg prednisolone daily and maximum doses <30 mg daily, were not associated with osteonecrosis in adults. Furthermore, even at high glucocorticoid doses, there was no evidence of increased osteonecrosis among glucocorticoid-exposed children (p=0.04 for interaction by age) (any glucocorticoid exposure, ages 2-9: aHR 1.1 (90% CI 0.7 to 1.7); ages 10-17: aHR 0.6 (90% CI 0.3 to 1.6)). Arthritis, inflammatory bowel disease and lupus were independently associated with osteonecrosis, but there was a similar dose relationship between glucocorticoids and osteonecrosis among adults with low-risk and high-risk diseases.

CONCLUSIONS

Glucocorticoid use was clearly associated with osteonecrosis in a dose-related fashion in adults, especially young adults, but this risk was not detectable in children. The absolute risk of glucocorticoid-associated osteonecrosis in the general paediatric population and in adults taking low glucocorticoid doses is at most extremely small.

A comparison of time dependent Cox regression, pooled logistic regression and cross sectional pooling with simulations and an application to the Framingham Heart Study

Background

Typical survival studies follow individuals to an event and measure explanatory variables for that event, sometimes repeatedly over the course of follow up. The Cox regression model has been used widely in the analyses of time to diagnosis or death from disease. The associations between the survival outcome and time dependent measures may be biased unless they are modeled appropriately.

Methods

In this paper we explore the Time Dependent Cox Regression Model (TDCM), which quantifies the effect of repeated measures of covariates in the analysis of time to event data. This model is commonly used in biomedical research but sometimes does not explicitly adjust for the times at which time dependent explanatory variables are measured. This approach can yield different estimates of association compared to a model that adjusts for these times. In order to address the question of how different these estimates are from a statistical perspective, we compare the TDCM to Pooled Logistic Regression (PLR) and Cross Sectional Pooling (CSP), considering models that adjust and do not adjust for time in PLR and CSP.

Results

In a series of simulations we found that time adjusted CSP provided identical results to the TDCM while the PLR showed larger parameter estimates compared to the time adjusted CSP and the TDCM in scenarios with high event rates. We also observed upwardly biased estimates in the unadjusted CSP and unadjusted PLR methods. The time adjusted PLR had a positive bias in the time dependent Age effect with reduced bias when the event rate is low. The PLR methods showed a negative bias in the Sex effect, a subject level covariate, when compared to the other methods. The Cox models yielded reliable estimates for the Sex effect in all scenarios considered.

Conclusions

We conclude that survival analyses that explicitly account in the statistical model for the times at which time dependent covariates are measured provide more reliable estimates compared to unadjusted analyses. We present results from the Framingham Heart Study in which lipid measurements and myocardial infarction data events were collected over a period of 26 years.

Electronic supplementary material

The online version of this article (doi:10.1186/s12874-016-0248-6) contains supplementary material, which is available to authorized users.

Repeat Cardiovascular Risk Assessment after Four Years: Is There Improvement in Risk Prediction?

Background

Framingham risk equations are widely used to predict cardiovascular disease based on health information from a single time point. Little is known regarding use of information from repeat risk assessments and temporal change in estimated cardiovascular risk for prediction of future cardiovascular events. This study was aimed to compare the discrimination and risk reclassification of approaches using estimated cardiovascular risk at single and repeat risk assessments

Methods

Using data on 12,197 individuals enrolled in EPIC-Norfolk cohort, with 12 years of follow-up, we examined rates of cardiovascular events by levels of estimated absolute risk (Framingham risk score) at the first and second health examination four years later. We calculated the area under the receiver operating characteristic curve (aROC) and risk reclassification, comparing approaches using information from single and repeat risk assessments (i.e., estimated risk at different time points).

Results

The mean Framingham risk score increased from 15.5% to 17.5% over a mean of 3.7 years from the first to second health examination. Individuals with high estimated risk (≥20%) at both health examinations had considerably higher rates of cardiovascular events than those who remained in the lowest risk category (<10%) in both health examinations (34.0 [95%CI 31.7–36.6] and 2.7 [2.2–3.3] per 1,000 person-years respectively). Using information from the most up-to-date risk assessment resulted in a small non-significant change in risk classification over the previous risk assessment (net reclassification improvement of -4.8%, p>0.05). Using information from both risk assessments slightly improved discrimination compared to information from a single risk assessment (aROC 0.76 and 0.75 respectively, p<0.001).

Conclusions

Using information from repeat risk assessments over a period of four years modestly improved prediction, compared to using data from a single risk assessment. However, this approach did not improve risk classification.

Low-density-lipoprotein cholesterol concentrations and risk of incident diabetes: epidemiological and genetic insights from the Framingham Heart Study

AIMS/HYPOTHESIS

Statins and niacin (nicotinic acid) reduce circulating LDL-cholesterol (LDL-C) levels by different mechanisms. Yet, both increase the risk of diabetes mellitus. Our objective was to relate blood LDL-C concentrations and a genetic risk score (GRS) for LDL-C to the risk of incident diabetes in individuals not treated with lipid-modifying therapy.

METHODS

We evaluated participants of the Framingham Heart Study who attended any of Offspring cohort examination cycles 3-8 and Third Generation cohort examination cycle 1 (N =14,120 person-observations, 6,011 unique individuals; mean age 50 ± 11 years, 56% women), who were not treated with lipid-modifying or antihypertensive medications and who were free from cardiovascular disease at baseline. Incident diabetes was assessed at the next examination.

RESULTS

The GRS was significantly associated with LDL-C concentrations (sex- and age-adjusted estimated influence 0.24, p < 0.0001). On follow-up (mean 4.5 ± 1.5 years), 312 individuals (2.2%) developed new-onset diabetes. In multivariable models, a higher LDL-C concentration was associated with lower risk of diabetes (OR per SD increment 0.81, 95% CI 0.70, 0.93, p = 0.004). The GRS was associated with incident diabetes in a similar direction and of comparable magnitude (OR per SD increment 0.85, 95% CI 0.76, 0.96, p = 0.009).

CONCLUSIONS/INTERPRETATION

Among individuals not treated with lipid-modifying therapy low LDL-C concentrations were associated with increased diabetes risk. These observations may contribute to our understanding of why lipid-lowering treatment may cause diabetes in some individuals. Additional studies are warranted to elucidate the molecular mechanisms underlying our observations.

Enhancing Breast Cancer Recurrence Algorithms Through Selective Use of Medical Record Data

BACKGROUND

The utility of data-based algorithms in research has been questioned because of errors in identification of cancer recurrences. We adapted previously published breast cancer recurrence algorithms, selectively using medical record (MR) data to improve classification.

METHODS

We evaluated second breast cancer event (SBCE) and recurrence-specific algorithms previously published by Chubak and colleagues in 1535 women from the Life After Cancer Epidemiology (LACE) and 225 women from the Women's Health Initiative cohorts and compared classification statistics to published values. We also sought to improve classification with minimal MR examination. We selected pairs of algorithms-one with high sensitivity/high positive predictive value (PPV) and another with high specificity/high PPV-using MR information to resolve discrepancies between algorithms, properly classifying events based on review; we called this "triangulation." Finally, in LACE, we compared associations between breast cancer survival risk factors and recurrence using MR data, single Chubak algorithms, and triangulation.

RESULTS

The SBCE algorithms performed well in identifying SBCE and recurrences. Recurrence-specific algorithms performed more poorly than published except for the high-specificity/high-PPV algorithm, which performed well. The triangulation method (sensitivity = 81.3%, specificity = 99.7%, PPV = 98.1%, NPV = 96.5%) improved recurrence classification over two single algorithms (sensitivity = 57.1%, specificity = 95.5%, PPV = 71.3%, NPV = 91.9%; and sensitivity = 74.6%, specificity = 97.3%, PPV = 84.7%, NPV = 95.1%), with 10.6% MR review. Triangulation performed well in survival risk factor analyses vs analyses using MR-identified recurrences.

CONCLUSIONS

Use of multiple recurrence algorithms in administrative data, in combination with selective examination of MR data, may improve recurrence data quality and reduce research costs.

Five-Year Risk of Mechanical Ventilation in Community-Dwelling Adults: The Framingham-Intermountain Anticipating Life Support Study

OBJECTIVES

To develop a quantitative tool for identifying outpatients most likely to require life support with mechanical ventilation within 5 years.

DESIGN

Retrospective cohort study.

SETTING

Framingham Heart Study (FHS) 1991 to 2009 and Intermountain Healthcare clinics 2008 to 2013.

PARTICIPANTS

FHS participants (n = 3,666; mean age 74; 58% female) in a derivation cohort and Intermountain Healthcare outpatients aged 65 and older (n = 88,302; mean age 73, 57% female) in an external validation cohort.

MEASUREMENTS

Information on demographic characteristics and comorbidities collected during FHS examinations to derive a 5-year risk score for receiving mechanical ventilation in an intensive care unit, with external validation using administrative data from outpatients seen at Intermountain Healthcare. A sensitivity analysis investigating model performance for a composite outcome of mechanical ventilation or death was performed.

RESULTS

Eighty (2%) FHS participants were mechanically ventilated within 5 years after a FHS examination. Age, sex, diabetes mellitus, hypertension, atrial fibrillation, alcohol use, chronic pulmonary disease, and hospitalization within the prior year predicted need for mechanical ventilation within 5 years (c-statistic = 0.74, 95% confidence interval (CI) = 0.68-0.80). One thousand seven hundred twenty-five (2%) Intermountain Healthcare outpatients underwent mechanical ventilation. The validation model c-statistic was 0.67 (95% CI = 0.66-0.68). Approximately 1% of individuals identified as low risk and 5% to 12% identified as high risk required mechanical ventilation within 5 years. Sensitivity analysis demonstrated a c-statistic of 0.75 (95% CI = 0.75-0.75) for risk prediction of a composite outcome of mechanical ventilation or death.

CONCLUSION

A simple risk score using clinical examination data or administrative data may be used to predict 5-year risk of mechanical ventilation or death. Further study is necessary to determine whether use of a risk score enhances advance care planning or improves quality of care of older adults.

Association of Smoke-Free Laws With Lower Percentages of New and Current Smokers Among Adolescents and Young Adults: An 11-Year Longitudinal Study

IMPORTANCE

Smoke-free laws are associated with a lower prevalence of smoking.

OBJECTIVE

To quantify the effect of 100% smoke-free laws on the smoking behavior of adolescents and young adults in a longitudinal analysis.

DESIGN, SETTING, AND PARTICIPANTS

Pooled logistic regression and zero-inflated negative binomial regression analysis of participants in the National Longitudinal Survey of Youth 1997 (data from 1997 to 2007), with complete data on initiation of smoking (n = 4098) and number of days respondents reported smoking in the past 30 days (n = 3913).

EXPOSURES

Laws for 100% smoke-free workplaces, laws for 100% smoke-free bars, and state cigarette taxes.

MAIN OUTCOMES AND MEASURES

Smoking initiation (first report of smoking cigarette), current (for 30 days) smoking, and number of days respondents reported smoking in the past 30 days among current smokers.

RESULTS

Laws for 100% smoke-free workplaces, but not bars, were associated with significantly lower odds of initiating smoking (odds ratio, 0.66 [95% CI, 0.44-0.99]). Laws for 100% smoke-free bars were associated with lower odds of being a current smoker (odds ratio, 0.80 [95% CI, 0.71-0.90]) and fewer days of smoking (incidence rate ratio, 0.85 [95% CI, 0.80-0.90]) among current smokers. Taxes were associated with a lower percentage of new smokers but not current smokers among adolescents and young adults. The effect of smoke-free workplace laws on smoking initiation is equivalent to a $1.57 (in 2007 dollars) tax increase. Smoke-free bar laws are associated with lower rates of current smoking, as well as a decrease in the number of days reported smoking among current smokers.

CONCLUSIONS AND RELEVANCE

Smoke-free laws are an important tobacco control tool. They not only protect bystanders from secondhand smoke but also contribute to less smoking among adolescents and young adults.

Improvements in long-term survival after spinal cord injury?

OBJECTIVE

To investigate whether there have been improvements in long-term survival after spinal cord injury in recent decades.

DESIGN

Survival analysis using time-varying covariates. The outcome variable was survival or mortality, and the explanatory variables were age, sex, level and grade of injury, and calendar year. The data were analyzed using the logistic regression model, Poisson regression model with comparison to the general population, and the computation of standardized mortality ratios for various groups.

SETTING

National Spinal Cord Injury Model Systems facilities.

PARTICIPANTS

Persons (N=31,531) who survived 2 years postinjury, were older than 10 years, and who did not require ventilator support. These persons contributed 484,979 person-years of data, with 8536 deaths over the 1973 to 2012 study period.

INTERVENTIONS

Not applicable.

MAIN OUTCOME MEASURES

Survival; survival relative to the general population; life expectancy.

RESULTS

After adjustment for age, sex, race, etiology of injury, time since injury, and level and grade of injury, mortality in persons with spinal cord injury was higher in the 2005 to 2012 period than in 1990 to 2004 or 1980 to 1989, the odds ratios for these 3 periods were .857, .826, and .802 as compared with the 1970 to 1979 reference period.

CONCLUSIONS

There was no evidence of improvement. Long-term survival has not changed over the past 30 years.

A joint test for progression and survival with interval-censored data from a cancer clinical trial

Clinical trials often assess efficacy by comparing treatments on the basis of two or more event-time outcomes. In the case of cancer clinical trials, progression-free survival (PFS), which is the minimum of the time from randomization to progression or to death, summarizes the comparison of treatments on the hazards for disease progression and mortality. However, the analysis of PFS does not utilize all the information we have on patients in the trial. First, if both progression and death times are recorded, then information on death time is ignored in the PFS analysis. Second, disease progression is monitored at regular clinic visits, and progression time is recorded as the first visit at which evidence of progression is detected. However, many patients miss or have irregular visits (resulting in interval-censored data) and sometimes die of the cancer before progression was recorded. In this case, the previous progression-free time could provide additional information on the treatment efficacy. The aim of this paper is to propose a method for comparing treatments that could more fully utilize the data on progression and death. We develop a test for treatment effect based on of the joint distribution of progression and survival. The issue of interval censoring is handled using the very simple and intuitive approach of the Conditional Expected Score Test (CEST). We focus on the application of these methods in cancer research.

Predictors of nicotine dependence symptoms among never-smoking adolescents: a longitudinal analysis from the Nicotine Dependence in Teens Study

BACKGROUND

Recent cross-sectional studies suggest some adolescents who have never smoked cigarettes experience nicotine dependence (ND) symptoms and that exposure to second-hand smoke, social exposure to smoking, and alcohol use are plausible correlates. The aim of this study was to replicate and extend these findings by investigating possible predictors of ND symptoms longitudinally.

METHOD

Participants included 847 secondary school students who had never smoked cigarettes enrolled in the Nicotine Dependence in Teens Study. Adolescents completed self-report questionnaires measuring smoking status, ND symptoms, and risk factors for ND in smokers (i.e., socio-demographic indicators, social exposure to smoking, psychosocial indicators, and substance use) in 20 survey cycles from 7 to 11th grade. Generalized estimating equations, which account for repeated measures within individuals, were used to test the predictors of ND symptoms.

RESULTS

Consistent with previous research, 7.8% of never-smokers across all cycles endorsed at least one ND symptom. Younger age (p ≤ .001), country of birth (p ≤ .05), peer smoking (p ≤ .001), teacher smoking (p ≤ .05), depression (p ≤ .05), stress (p ≤ .001), lower self-esteem (p ≤ .05), impulsivity (p ≤ .05), and alcohol use (p ≤ .001) predicted greater ND symptoms in multivariable modeling.

CONCLUSIONS

Replicating previous cross-sectional findings, peer smoking and alcohol use predicted ND symptoms among never-smoking adolescents. Extending these findings, previous predictors only observed among ever-smokers, including socio-demographic and psychosocial indicators, also predicted ND symptoms. This longitudinal investigation demonstrated the temporal relation of the predictors preceding ND symptoms. Future research should consider longer prospective studies with younger children to capture early onset of ND symptoms and with longer follow-up to detect eventual smoking uptake.

High- and low-fat dairy intake, recurrence, and mortality after breast cancer diagnosis

BACKGROUND

Dietary fat in dairy is a source of estrogenic hormones and may be related to worse breast cancer survival. We evaluated associations between high- and low-fat dairy intake, recurrence, and mortality after breast cancer diagnosis.

METHODS

We included 1893 women from the Life After Cancer Epidemiology study diagnosed with early-stage invasive breast cancer from 1997 to 2000, who completed the Fred Hutchinson Cancer Research Center Food Frequency Questionnaire after diagnosis. A total of 349 women had a recurrence and 372 died during a median follow-up of 11.8 years, with 189 deaths from breast cancer. We used delayed entry Cox proportional hazards regression to evaluate associations between categories of the cumulative average of dairy fat at baseline and at follow-up 5 to 6 years later and subsequent outcomes. Tests of statistical significance were two-sided.

RESULTS

In multivariable-adjusted analyses, overall dairy intake was unrelated to breast cancer-specific outcomes, although it was positively related to overall mortality. Low-fat dairy intake was unrelated to recurrence or survival. However, high-fat dairy intake was positively associated with outcomes. Compared with the reference (0 to <0.5 servings/day), those consuming larger amounts of high-fat dairy had higher breast cancer mortality (0.5 to <1.0 servings/day: hazard ratio [HR] = 1.20, 95% confidence interval [CI] = 0.82 to 1.77; and ≥1.0 servings/day: HR = 1.49, 95% CI = 1.00 to 2.24, P trend = .05), higher all-cause mortality (P trend < .001), and higher non-breast cancer mortality (P trend = .007); the relationship with breast cancer recurrence was positive but not statistically significant. The higher risk appeared consistent across different types of high-fat dairy products.

CONCLUSIONS

Intake of high-fat dairy, but not low-fat dairy, was related to a higher risk of mortality after breast cancer diagnosis.

Association between blood pressure and mortality in a Spanish cohort of persons aged 65 years or over: a dynamic model

INTRODUCTION AND OBJECTIVES

Few studies have used time-dependent correction to analyze the relationship between blood pressure and all-cause mortality, and to our knowledge none has been performed in older people from the Mediterranean area. This study aimed to estimate the relationship between baseline blood pressure and blood pressure as a time-dependent covariate with the risk of all-cause mortality in a population cohort of persons aged 65 or older in Spain.

METHODS

Data were taken from the population-based study "Aging in Leganés" with 17 years of follow-up, launched in 1993 in a random sample (n=1560) of persons aged ≥65 years. Mortality was assessed in 2010. Cox proportional hazards models were fitted to examine the effects on mortality of blood pressure at baseline and of blood pressure as a time-dependent covariate.

RESULTS

The lowest mortality was observed at baseline systolic blood pressure of 136 mmHg and time-dependent covariate value of 147 mmHg. The highest risk of mortality for time-dependent covariates occurred with systolic blood pressure <115 mmHg and >93 mmHg and diastolic blood pressure <80 mmHg. Diastolic blood pressure over 85 mmHg did not increase the risk of death.

CONCLUSIONS

Based on the dynamic association between blood pressure and mortality, a U-shaped relationship was found for systolic blood pressure and a negative relationship for diastolic blood pressure and all-cause mortality. The lowest mortality corresponded to a systolic blood pressure level slightly over the diagnostic hypertension value and suggests that a value of 140 mmHg is not adequate as a diagnostic and therapeutic threshold in an elderly population.

Social networks, social support, and burden in relationships, and mortality after breast cancer diagnosis in the Life After Breast Cancer Epidemiology (LACE) study

Larger social networks have been associated with lower breast cancer mortality. The authors evaluated how levels of social support and burden influenced this association. We included 2,264 women from the Life After Cancer Epidemiology study who were diagnosed with early-stage, invasive breast cancer between 1997 and 2000, and provided data on social networks (spouse or intimate partner, religious/social ties, volunteering, time socializing with friends, and number of first-degree female relatives), social support, and caregiving. 401 died during a median follow-up of 10.8 years follow-up with 215 from breast cancer. We used delayed entry Cox proportional hazards regression to evaluate associations. In multivariate-adjusted analyses, social isolation was unrelated to recurrence or breast cancer-specific mortality. However, socially isolated women had higher all-cause mortality (HR = 1.34, 95 % CI: 1.03-1.73) and mortality from other causes (HR = 1.79, 95 % CI: 1.19-2.68). Levels of social support and burden modified associations. Among those with low, but not high, levels of social support from friends and family, lack of religious/social participation (HR = 1.58, 95 % CI: 1.07-2.36, p = 0.02, p interaction = 0.01) and lack of volunteering (HR = 1.78, 95 % CI: 1.15-2.77, p = 0.01, p interaction = 0.01) predicted higher all-cause mortality. In cross-classification analyses, only women with both small networks and low levels of support (HR = 1.61, 95 % CI: 1.10-2.38) had a significantly higher risk of mortality than women with large networks and high levels of support; women with small networks and high levels of support had no higher risk of mortality (HR = 1.13, 95 % CI: 0.74-1.72). Social networks were also more important for caregivers versus noncaregivers. Larger social networks predicted better prognosis after breast cancer, but associations depended on the quality and burden of family relationships.

A review of the reporting and handling of missing data in cohort studies with repeated assessment of exposure measures

BACKGROUND

Retaining participants in cohort studies with multiple follow-up waves is difficult. Commonly, researchers are faced with the problem of missing data, which may introduce biased results as well as a loss of statistical power and precision. The STROBE guidelines von Elm et al. (Lancet, 370:1453-1457, 2007); Vandenbroucke et al. (PLoS Med, 4:e297, 2007) and the guidelines proposed by Sterne et al. (BMJ, 338:b2393, 2009) recommend that cohort studies report on the amount of missing data, the reasons for non-participation and non-response, and the method used to handle missing data in the analyses. We have conducted a review of publications from cohort studies in order to document the reporting of missing data for exposure measures and to describe the statistical methods used to account for the missing data.

METHODS

A systematic search of English language papers published from January 2000 to December 2009 was carried out in PubMed. Prospective cohort studies with a sample size greater than 1,000 that analysed data using repeated measures of exposure were included.

RESULTS

Among the 82 papers meeting the inclusion criteria, only 35 (43%) reported the amount of missing data according to the suggested guidelines. Sixty-eight papers (83%) described how they dealt with missing data in the analysis. Most of the papers excluded participants with missing data and performed a complete-case analysis (n=54, 66%). Other papers used more sophisticated methods including multiple imputation (n=5) or fully Bayesian modeling (n=1). Methods known to produce biased results were also used, for example, Last Observation Carried Forward (n=7), the missing indicator method (n=1), and mean value substitution (n=3). For the remaining 14 papers, the method used to handle missing data in the analysis was not stated.

CONCLUSIONS

This review highlights the inconsistent reporting of missing data in cohort studies and the continuing use of inappropriate methods to handle missing data in the analysis. Epidemiological journals should invoke the STROBE guidelines as a framework for authors so that the amount of missing data and how this was accounted for in the analysis is transparent in the reporting of cohort studies.

Social networks, social support and burden in relationships, and mortality after breast cancer diagnosis

Though larger social networks are associated with reduced breast cancer mortality, there is a need to clarify how both social support and social burden influence this association. We included 4,530 women from the Women's Health Initiative who were diagnosed with breast cancer between 1993 and 2009, and provided data on social networks (spouse or intimate partner, religious ties, club ties, and number of first-degree relatives) before diagnosis. Of those, 354 died during follow-up, with 190 from breast cancer. We used Cox proportional hazards regression to evaluate associations of social network members with risk of post-diagnosis mortality, further evaluating associations by social support and social burden (caregiving, social strain). In multivariate-adjusted analyses, among women with high but not low social support, being married was related to lower all-cause mortality. By contrast, among women with high but not low social burden, those with a higher number of first-degree relatives, including siblings, parents, and children, had higher all-cause and breast cancer mortality (among caregivers: 0-3 relatives (ref), 4-5 relatives, HR = 1.47 (95% CI: 0.62-3.52), 6-9 relatives, HR = 2.08 (95% CI: 0.89-4.86), 10+ relatives, HR = 3.55 (95% CI: 1.35-9.33), P-continuous = 0.02, P-interaction = 0.008). The association by social strain was similar though it was not modified by level of social support. Other social network members were unrelated to mortality. Social relationships may have both adverse and beneficial influences on breast cancer survival. Clarifying these depends on understanding the context of women's relationships.

Food pattern analysis over time: unhealthful eating trajectories predict obesity

BACKGROUND

Analysis of dietary patterns is prominent in nutrition literatures, yet few studies have taken advantage of multiple repeated measurements to understand the nature of individual-level changes over time in food choice, or the relation between these changes and body mass index (BMI).

OBJECTIVE

To investigate changes in eating patterns at the individual level across three exam periods, and to prospectively examine the relation of eating trajectories to BMI at the cohort level.

DESIGN

The study included 3418 participants at baseline. Clinically measured BMI and dietary intake were assessed during three exam periods between 1991 and 2001 using a validated food frequency questionnaire. An individual's eating trajectory across exam periods was analyzed using sequence analysis, and then used to estimate outcomes of continuous BMI and categorical obesity status. Ordinary least squares regression models with robust standard errors were adjusted for socio-economic and demographic confounders, baseline BMI and baseline eating.

RESULTS

A total of 66.2% (n=1614) of participants change their diet pattern during the study period, 33.8% (n=823) remain stable. After accounting for potential confounders, an unhealthful trajectory is significantly associated with a 0.42 kg m(-2) increase in BMI (confidence interval (CI): 0.1, 0.7). Those with an unhealthful trajectory are 1.79 times more likely to be overweight (relative risk ratio, 95% CI: 1.1, 2.8) and 2.4 times more likely to be obese (relative risk ratio, 95% CI: 1.3, 4.4). Moreover, a number of specific diet transitions between exams are predictive of weight gain or loss.

CONCLUSION

Contextualizing an individual's current eating behaviors with an eye towards diet history may be an important boon in the reduction of obesity. Although it may not be realistic for many people to shift from the least to most healthful diet, results from this study suggest that consistent movement in an overall healthier direction is associated with less weight gain.

Associations of APOE gene polymorphisms with bone mineral density and fracture risk: a meta-analysis

To determine the association of the Apolipoprotein E (APOE) E4 gene polymorphism with bone mineral density (BMD) and fractures we conducted a meta-analysis of 17 reports. Despite lower trochanteric and lumbar BMD in APOE4 carriers, there is insufficient evidence to support a consistent association of APOE with bone health.

INTRODUCTION

APOE has been studied for its potential role in osteoporosis risk. It is hypothesized that genetic variation at APOE locus, known as E2, E3, and E4, may modulate BMD through its effects on lipoproteins and vitamin K transport. The purpose of this study was to determine the association of the APOE-E4 gene polymorphism with bone-related phenotypes.

METHODS

We conducted a meta-analysis that combined newly analyzed individual data from two community-based cohorts, the Framingham Offspring Study (N = 1,495) and the vitamin K clinical trial (N = 377), with 15 other eligible published reports. Bone phenotypes included BMD measurements of the hip (total hip and trochanteric and femoral neck sites) and lumbar spine (from the L2 to L4 vertebrae) and prevalence or incidence of vertebral, hip, and other fractures.

RESULTS

In sex-pooled analyses, APOE4 carriers had a 0.018 g/cm(2) lower weighted mean trochanteric BMD than non carriers (p = 0.0002) with no evidence for between-study heterogeneity. A significant association was also detected with lumbar spine BMD (p = 0.006); however, inter-study heterogeneity was observed. Associations with lumbar spine and trochanteric BMD were observed predominantly in women and became less significant in meta-regression (p = 0.055 and 0.01, respectively). There were no consistent associations of APOE4 genotype with BMD at other skeletal sites or with fracture risk.

CONCLUSIONS

Based on these findings, there is insufficient evidence to support a strong and consistent association of the APOE genotype with BMD and fracture incidence.

Cardiovascular outcomes in framingham participants with diabetes: the importance of blood pressure

We designed this study to explore to what extent the excess risk of cardiovascular events in diabetic individuals is attributable to hypertension. We retrospectively analyzed prospectively collected data from the Framingham original and offspring cohorts. Of the 1145 Framingham subjects newly diagnosed with diabetes mellitus who did not have a previous history of cardiovascular events, 663 (58%) had hypertension at the time that diabetes mellitus was diagnosed. During 4154 person-years of follow-up, 125 died, and 204 experienced a cardiovascular event. Framingham participants with hypertension at the time of diabetes mellitus diagnosis exhibited higher rates of all-cause mortality (32 versus 20 per 1000 person-years; P<0.001) and cardiovascular events (52 versus 31 per 1000 person-years; P<0.001) compared with normotensive subjects with diabetes mellitus. After adjustment for demographic and clinical covariates, hypertension was associated with a 72% increase in the risk of all-cause death and a 57% increase in the risk of any cardiovascular event in individuals with diabetes mellitus. The population-attributable risk from hypertension in individuals with diabetes mellitus was 30% for all-cause death and 25% for any cardiovascular event (increasing to 44% and 41%, respectively, if the 110 normotensive subjects who developed hypertension during follow-up were excluded from the analysis). In comparison, after adjustment for concurrent hypertension, the population-attributable risk from diabetes mellitus in Framingham subjects was 7% for all-cause mortality and 9% for any cardiovascular disease event. Although diabetes mellitus is associated with increased risks of death and cardiovascular events in Framingham subjects, much of this excess risk is attributable to coexistent hypertension.

Life-course socioeconomic position and type 2 diabetes mellitus: The Framingham Offspring Study

Evidence is lacking on whether the duration and timing of low socioeconomic position (SEP) across a person's life course may be associated with incidence of type 2 diabetes mellitus (T2D). The authors' objectives were to investigate associations between cumulative SEP and the incidence of T2D in the Framingham Offspring Study (n = 1,893; 52% women; mean baseline age = 34 years). Pooled logistic regression analyses demonstrated that age-adjusted cumulative SEP was associated with T2D in women (for low vs. high cumulative SEP, odds ratio (OR) = 1.92, 95% confidence interval (CI): 1.08, 3.42). Age-adjusted analyses for young-adulthood SEP (7.85 for ≤12 vs. >16 years of education, OR = 2.84, 95% CI: 1.03), active professional life SEP (for laborer vs. professional/executive/supervisory/technical occupations, OR = 2.40, 95% CI: 1.05, 5.47), and social-mobility frameworks (for declining life-course SEP, OR = 2.99, 95% CI: 1.39, 6.44; for stable low vs. stable high life-course SEP, OR = 1.85, 95% CI: 1.02, 3.35) all demonstrated associations between low SEP and T2D incidence in women. No association was observed between childhood SEP and T2D in women for father's education (some high school or less vs. any postsecondary education, OR = 1.26, 95% CI: 0.72, 2.22). In men, there was little evidence of associations between life-course SEP and T2D incidence. These findings suggest that cumulative SEP is inversely associated with incidence of T2D in women, and that this association may be primarily due to the women's educational levels and occupations.

A test for the relationship between a time-varying marker and both recovery and progression with missing data

For clinical studies of chronic diseases, patients are followed to determine whether treatment results in either improvement or decline in their clinical status. During periodic exams, laboratory specimens are collected that are believed to reflect both the progression and improvement in the disease status. Often patients miss visits and return with a changed disease status, resulting in interval-censored laboratory markers and indicators of disease status. The goal of this paper is to propose a single test that would evaluate the relationship between a longitudinal marker and clinical progression or recovery when missing visits result in interval-censored covariate and outcome data. We apply our test to evaluate the relationship between treatment compliance and progression and remission of renal disease in diabetic patients.

A score test for association of a longitudinal marker and an event with missing data

Often clinical studies periodically record information on disease progression as well as results from laboratory studies that are believed to reflect the progressing stages of the disease. A primary aim of such a study is to determine the relationship between the lab measurements and a disease progression. If there were no missing or censored data, these analyses would be straightforward. However, often patients miss visits, and return after their disease has progressed. In this case, not only is their progression time interval censored, but their lab test series is also incomplete. In this article, we propose a simple test for the association between a longitudinal marker and an event time from incomplete data. We derive the test using a very intuitive technique of calculating the expected complete data score conditional on the observed incomplete data (conditional expected score test, CEST). The problem was motivated by data from an observational study of patients with diabetes.

Diet quality indices and postmenopausal breast cancer survival

Research on diet in breast cancer survival has been focused on single nutrients or foods, particularly dietary fat, fruits, vegetables, fiber, and alcohol. We hypothesized that diet quality indices decrease the risk of total and non-breast-cancer-related deaths in women diagnosed with breast cancer. We evaluated 4 dietary quality scores: Alternate Healthy Eating Index (AHEI), Diet Quality Index-Revised (DQIR), Recommended Food Score (RFS), and the alternate Mediterranean Diet Score (aMED), among 2,729 women from the Nurses' Health Study with invasive Stage 1-3 breast cancer diagnosed between 1978 and 1998 with follow-up through 2004. In multivariate adjusted analyses, no association was found between diet quality indices and either total or non-breast-cancer-related deaths. However, a higher aMED score was associated with a lower risk of non-breast-cancer death in women with low physical activity; the RR comparing the highest to lowest tertile was 0.39 (95% CI, 0.20-0.75, P trend = 0.0004). Our results suggest that a higher-quality diet after breast cancer diagnosis does not considerably change the risk of death from breast cancer. However, healthy dietary choices may be important because women are at risk of death from non-breast-cancer-related causes affected by diet.

Impact of blood pressure levels on different types of stroke: the Hisayama study

OBJECTIVE

Clinical uncertainty remains whether the blood pressure classification and risk stratifications recommended by the Japanese Society of Hypertension Guidelines for the Management of Hypertension (JSH 2009) are useful in predicting the risks of stroke and its subtypes in the general Japanese population.

METHODS

A total of 1621 stroke-free residents of a Japanese community aged at least 40 years were followed up for 32 years. Outcomes were total and cause-specific stroke (lacunar infarction, atherothrombotic infarction, cardioembolic infarction, cerebral haemorrhage and subarachnoid haemorrhage). Incidence was calculated by the pooling of repeated observations method.

RESULTS

The age-adjusted incidence of total stroke rose progressively with higher blood pressure levels in both sexes (both P for trend <0.0001). A similar pattern was observed for lacunar infarction in both sexes and for cerebral haemorrhage in men: the differences were significant between optimal blood pressure and grades 1-3 hypertension (all P < 0.05). The age-adjusted incidence of atherothrombotic infarction in either sex and that of cardioembolic infarction and subarachnoid haemorrhage in women significantly increased in grade 3 hypertension (all P < 0.05). These associations remained substantially unchanged even after adjustment for other risk factors. In regard to risk stratification, the age-adjusted incidence of stroke significantly increased with the level of risk in both sexes.

CONCLUSION

Our findings suggest that the blood pressure classification and risk stratifications recommended by the JSH 2009 guidelines are useful in predicting the risk of stroke in a general Japanese population, but the magnitude and patterns of the impact of blood pressure categories are different among stroke subtypes.

Initiation of anaemia management in patients with chronic kidney disease not on dialysis in the Veterans Health Administration

BACKGROUND

Erythropoiesis-stimulating agents (ESAs) are frequently used to treat anaemia of chronic kidney disease (CKD) in the dialysis setting; however, few data are available regarding factors influencing initiation of ESAs and other therapies in non-dialysis patients.

METHODS

A retrospective cohort study of Veterans Health Administration data from 2003 to 2005 for 89 585 patients identified as having CKD and anaemia based on two outpatient estimated glomerular filtration rates <60 ml/min/1.73 m(2) and at least one outpatient haemoglobin (Hb) <11 g/dL. Hb levels, patient demographics, clinical and provider characteristics and procedures predicted ESA treatment initiation over 1 year of follow-up. Multivariable logistic and pooled logistic survival models identified predictors of ESA initiation.

RESULTS

Overall, 6381 subjects (7.1%) initiated ESAs within 1 year of the index Hb; initiation was more common (8.6%) for patients with Hb <10 g/dL. Iron therapy use varied by initial Hb levels (27.6% to 52.4%) as did transfusions (12.5% to 42.8%); each was more common at lower Hb levels. Hbs rose to above 11 g/dL for 25-50% of patients in the absence of any treatment or by transfusion/iron therapy. Factors predicting time to ESA initiation included: nephrologist [odds ratio (OR = 2.3)] or haematologist care (OR = 2.2) and iron therapy (OR = 1.6). Transfusions increased for patients with increasing follow-up time.

CONCLUSION

Iron therapy is more common than ESA treatment in patients with CKD and Hbs <11 g/dL in the VA. Correction of anaemia in the absence of any ESA treatment was common at higher Hbs levels, but much less so when Hb levels fell below 10 g/dL.

Measurement error correction for the cumulative average model in the survival analysis of nutritional data: application to Nurses' Health Study

The use of the cumulative average model to investigate the association between disease incidence and repeated measurements of exposures in medical follow-up studies can be dated back to the 1960s (Kahn and Dawber, J Chron Dis 19:611-620, 1966). This model takes advantage of all prior data and thus should provide a statistically more powerful test of disease-exposure associations. Measurement error in covariates is common for medical follow-up studies. Many methods have been proposed to correct for measurement error. To the best of our knowledge, no methods have been proposed yet to correct for measurement error in the cumulative average model. In this article, we propose a regression calibration approach to correct relative risk estimates for measurement error. The approach is illustrated with data from the Nurses' Health Study relating incident breast cancer between 1980 and 2002 to time-dependent measures of calorie-adjusted saturated fat intake, controlling for total caloric intake, alcohol intake, and baseline age.

Living healthier for longer: comparative effects of three heart-healthy behaviors on life expectancy with and without cardiovascular disease

Background

Non-smoking, having a normal weight and increased levels of physical activity are perhaps the three key factors for preventing cardiovascular disease (CVD). However, the relative effects of these factors on healthy longevity have not been well described. We aimed to calculate and compare the effects of non-smoking, normal weight and physical activity in middle-aged populations on life expectancy with and without cardiovascular disease.

Methods

Using multi-state life tables and data from the Framingham Heart Study (n = 4634) we calculated the effects of three heart healthy behaviours among populations aged 50 years and over on life expectancy with and without cardiovascular disease. For the life table calculations, we used hazard ratios for 3 transitions (No CVD to CVD, no CVD to death, and CVD to death) by health behaviour category, and adjusted for age, sex, and potential confounders.

Results

High levels of physical activity, never smoking (men), and normal weight were each associated with 20-40% lower risks of developing CVD as compared to low physical activity, current smoking and obesity, respectively. Never smoking and high levels of physical activity reduced the risks of dying in those with and without a history of CVD, but normal weight did not. Never-smoking was associated with the largest gains in total life expectancy (4.3 years, men, 4.1 years, women) and CVD-free life expectancy (3.8 and 3.4 years, respectively). High levels of physical activity and normal weight were associated with lesser gains in total life expectancy (3.5 years, men and 3.4 years, women, and 1.3 years, men and 1.0 year women, respectively), and slightly lesser gains in CVD-free life expectancy (3.0 years, men and 3.1 years, women, and 3.1 years men and 2.9 years women, respectively). Normal weight was the only behaviour associated with a reduction in the number of years lived with CVD (1.8 years, men and 1.9 years, women).

Conclusions

Achieving high levels of physical activity, normal weight, and never smoking, are effective ways to prevent cardiovascular disease and to extend total life expectancy and the number of years lived free of CVD. Increasing the prevalence of normal weight could further reduce the time spent with CVD in the population.

Chronic musculoskeletal pain and the occurrence of falls in an older population

CONTEXT

Chronic pain is a major contributor to disability in older adults; however, the potential role of chronic pain as a risk factor for falls is poorly understood.

OBJECTIVE

To determine whether chronic musculoskeletal pain is associated with an increased occurrence of falls in a cohort of community-living older adults.

DESIGN, SETTING, AND PARTICIPANTS

The Maintenance of Balance, Independent Living, Intellect, and Zest in the Elderly (MOBILIZE) Boston Study is a population-based longitudinal study of falls involving 749 adults aged 70 years and older. Participants were enrolled from September 2005 through January 2008.

MAIN OUTCOME MEASURE

Participants recorded falls on monthly calendar postcards mailed to the study center during an 18-month period.

RESULTS

There were 1029 falls reported during the follow-up. A report of 2 or more locations of musculoskeletal pain at baseline was associated with greater occurrence of falls. The age-adjusted rates of falls per person-year were 1.18 (95% confidence interval [CI], 1.13-1.23) for the 300 participants with 2 or more sites of joint pain, 0.90 (95% CI, 0.87-0.92) for the 181 participants with single-site pain, and 0.78 (95% CI, 0.74-0.81) for the 267 participants with no joint pain. Similarly, more severe or disabling pain at baseline was associated with higher fall rates (P < .05). The association persisted after adjusting for multiple confounders and fall risk factors. The greatest risk for falls was observed in persons who had 2 or more pain sites (adjusted rate ratio [RR], 1.53; 95% CI, 1.17-1.99), and those in the highest tertiles of pain severity (adjusted RR, 1.53; 95% CI, 1.12-2.08) and pain interference with activities (adjusted RR, 1.53; 95%CI, 1.15-2.05), compared with their peers with no pain or those in the lowest tertiles of pain scores.

CONCLUSIONS

Chronic pain measured according to number of locations, severity, or pain interference with daily activities was associated with greater risk of falls in older adults.

Determinants of first puff and daily cigarette smoking in adolescents

Few prospective studies of smoking initiation have investigated a wide range of time-varying and invariant predictor variables at the individual and contextual levels concurrently. In this study (1999-2005), 877 Canadian students (mean age = 12.7 years) who had never smoked at baseline completed self-report questionnaires on cigarette smoking and 32 predictor variables in 20 survey cycles during secondary school. Height and weight were measured in survey cycles 1, 12, and 19. School administrators completed questionnaires on school tobacco control policies/activities, and trained observers collected data on access to tobacco products in commercial establishments near schools. Younger age, single-parent family status, smoking by parents, siblings, friends, and school staff, stress, impulsivity, lower self-esteem, feeling a need to smoke, not doing well at school, susceptibility to tobacco advertising, alcohol use, use of other tobacco products, and attending a smoking-tolerant school were independent determinants of smoking initiation. Independent determinants of daily smoking onset among initiators of nondaily smoking included smoking by siblings and friends, feeling a need to smoke, susceptibility to tobacco advertising, use of other tobacco products, and self-perceived mental and physical addiction. Adolescent tobacco control programs should address multiple individual and contextual-level risk factors. Strategies that address nicotine dependence symptoms are also needed for adolescents who have already initiated smoking.