Development of a machine learning model to predict mild cognitive impairment using natural language processing in the absence of screening

BACKGROUND

Patients and their loved ones often report symptoms or complaints of cognitive decline that clinicians note in free clinical text, but no structured screening or diagnostic data are recorded. These symptoms/complaints may be signals that predict who will go on to be diagnosed with mild cognitive impairment (MCI) and ultimately develop Alzheimer's Disease or related dementias. Our objective was to develop a natural language processing system and prediction model for identification of MCI from clinical text in the absence of screening or other structured diagnostic information.

METHODS

There were two populations of patients: 1794 participants in the Adult Changes in Thought (ACT) study and 2391 patients in the general population of Kaiser Permanente Washington. All individuals had standardized cognitive assessment scores. We excluded patients with a diagnosis of Alzheimer's Disease, Dementia or use of donepezil. We manually annotated 10,391 clinic notes to train the NLP model. Standard Python code was used to extract phrases from notes and map each phrase to a cognitive functioning concept. Concepts derived from the NLP system were used to predict future MCI. The prediction model was trained on the ACT cohort and 60% of the general population cohort with 40% withheld for validation. We used a least absolute shrinkage and selection operator logistic regression approach (LASSO) to fit a prediction model with MCI as the prediction target. Using the predicted case status from the LASSO model and known MCI from standardized scores, we constructed receiver operating curves to measure model performance.

RESULTS

Chart abstraction identified 42 MCI concepts. Prediction model performance in the validation data set was modest with an area under the curve of 0.67. Setting the cutoff for correct classification at 0.60, the classifier yielded sensitivity of 1.7%, specificity of 99.7%, PPV of 70% and NPV of 70.5% in the validation cohort.

DISCUSSION AND CONCLUSION

Although the sensitivity of the machine learning model was poor, negative predictive value was high, an important characteristic of models used for population-based screening. While an AUC of 0.67 is generally considered moderate performance, it is also comparable to several tests that are widely used in clinical practice.

Prospective Evaluation of Cognitive Health and Related Factors in Elderly at Risk for Developing Alzheimer's Dementia: A Longitudinal Cohort Study

BACKGROUND

The CHARIOT PRO Main study is a prospective, non-interventional study evaluating cognitive trajectories in participants at the preclinical stage of Alzheimer's disease (AD) classified by risk levels for developing mild cognitive impairment due to AD (MCI-AD).

OBJECTIVES

The study aimed to characterize factors and markers influencing cognitive and functional progression among individuals at-risk for developing MCI-AD, and examine data for more precise predictors of cognitive change, particularly in relation to APOE ε4 subgroup.

DESIGN

This single-site study was conducted at the Imperial College London (ICL) in the United Kingdom. Participants 60 to 85 years of age were classified as high, medium (amnestic or non-amnestic) or low risk for developing MCI-AD based on RBANS z-scores. A series of clinical outcome assessments (COAs) on factors influencing baseline cognitive changes were collected in each of the instrument categories of cognition, lifestyle exposure, mood, and sleep. Data collection was planned to occur every 6 months for 48 months, however the median follow-up time was 18.1 months due to early termination of study by the sponsor.

RESULTS

987 participants were screened, among them 690 participants were actively followed-up post baseline, of whom 165 (23.9%) were APOE ε4 carriers; with at least one copy of the allele. The mean age was 68.73 years, 94.6% were white, 57.4% were female, and 34.8% had a Family History of Dementia with a somewhat larger percentage in the APOE ε4 carrier group (42.4%) compared to the non-carrier group (32.4%). Over half of the participants were married and 53% had a Bachelor's or higher degree. Most frequently, safety events typical for this population consisted of upper respiratory tract infection (10.4%), falls (5.2%), hypertension (3.5%) and back pain (3.0%). Conclusion (clinical relevance): AD-related measures collected during the CHARIOT PRO Main study will allow identification and evaluation of AD risk factors and markers associated with cognitive performance from the pre-clinical stage. Evaluating the psycho-biological characteristics of these pre-symptomatic individuals in relation to their natural neurocognitive trajectories will enhance current understanding on determinants of the initial signs of cognitive changes linked to AD.

Psychometric Evaluation of the Neuropsychological Test Battery in Individuals with Normal Cognition, Mild Cognitive Impairment, or Mild to Moderate Alzheimer's Disease: Results from a Longitudinal Study

BACKGROUND

The Neuropsychological Test Battery (NTB) is a combination of widely used clinical neuropsychological tests measuring memory and executive function and was designed to overcome some of the limitations of the traditionally used Alzheimer's disease Assessment Scale - Cognitive subscale (ADAS-Cog). A previously reported account indicated high levels of NTB reliability in patients with mild-to-moderate Alzheimer's disease (AD) and mild cognitive impairment (MCI).

OBJECTIVES

We examined capacity of the Neuropsychological Test Battery (NTB) and its component subtests to measure cognitive change over time. Correlations with other cognitive and functional assessments were also determined. Design, Settings, Participants: This was a multicentre, prospective, non-interventional, longitudinal cohort study involving patients with mild-to-moderate AD (n=196), MCI (n=70), or cognitively normal control participants (NC, n=75).

INTERVENTION

The NTB, as well as other Clinical Outcome Assessments including, ADAS-Cog, other cognitive measures, functional/behavioral questionnaires, health outcome questionnaires, and resource utilization tools were administered.

RESULTS

Mean change from baseline for the NTB composite score and the six individual NTB subtests showed greater reductions in performance over time in the AD and MCI groups, compared with NC group. The ADAS-Cog was found to be more sensitive to change than the NTB in all three populations.

CONCLUSIONS

The NTB showed high correlation with the ADAS-Cog and appears to be a sensitive and reliable assessment tool for measuring cognitive decline in patients with mild-to-moderate AD. However, the ADAS-Cog was found to be more sensitive to change over time in both the AD and MCI populations.

Central Review of Amyloid-Related Imaging Abnormalities in Two Phase III Clinical Trials of Bapineuzumab in Mild-To-Moderate Alzheimer's Disease Patients

BACKGROUND

Amyloid-related imaging abnormalities (ARIA) consist of ARIA-E (with effusion or edema) and ARIA-H (hemosiderin deposits [HDs]).

OBJECTIVES

To address accurate ascertainment of ARIA identification, a final magnetic resonance imaging (MRI) reading was performed on patients with mild-to-moderate Alzheimer's disease randomized to bapineuzumab IV or placebo during two Phase III trials (APOE ɛ4 allele carriers or noncarriers).

METHODS

Final MRI central review consisted of a systematic sequential locked, adjudicated read in 1,331 APOE ɛ4 noncarriers and 1,121 carriers by independent neuroradiologists. Assessment of ARIA-E, ARIA-H, intracerebral hemorrhages, and age-related white matter changes is described.

RESULTS

In the Final Read, treatment-emergent ARIA-E were identified in 242 patients including 76 additional cases not noted previously in real time. Overall, incidence proportion of ARIA-E was higher in carriers (active 21.2%; placebo 1.1%) than in noncarriers (pooled active 11.3%; placebo 0.6%), and was more often identified in homozygote APOE ɛ4 carriers than heterozygotes (34.5% versus 16.9%). Incidence rate of ARIA-E increased with increased dose in noncarriers. Frequency of ARIA-E first episodes was highest after the first and second bapineuzumab infusion and declined after repeated infusions. Incidence of total HDs <10 mm (cerebral microhemorrhages) was higher in active groups versus placebo.

CONCLUSION

ARIA was detected more often on MRI scans when every scan was reviewed by trained neuroradiologists and results adjudicated. There was increased incidence of ARIA-E in bapineuzumab-treated carriers who had a microhemorrhage at baseline. ARIA-E was a risk factor for incident ARIA-H and late onset ARIA-E was milder radiologically. Age-related white matter changes did not progress during the study.

Dementia prevalence and incidence in a federation of European Electronic Health Record databases: The European Medical Informatics Framework resource

INTRODUCTION

The European Medical Information Framework consortium has assembled electronic health record (EHR) databases for dementia research. We calculated dementia prevalence and incidence in 25 million persons from 2004 to 2012.

METHODS

Six EHR databases (three primary care and three secondary care) from five countries were interrogated. Dementia was ascertained by consensus harmonization of clinical/diagnostic codes. Annual period prevalences and incidences by age and gender were calculated and meta-analyzed.

RESULTS

The six databases contained 138,625 dementia cases. Age-specific prevalences were around 30% of published estimates from community samples and incidences were around 50%. Pooled prevalences had increased from 2004 to 2012 in all age groups but pooled incidences only after age 75 years. Associations with age and gender were stable over time.

DISCUSSION

The European Medical Information Framework initiative supports EHR data on unprecedented number of people with dementia. Age-specific prevalences and incidences mirror estimates from community samples in pattern at levels that are lower but increasing over time.

Validation of an MRI Rating Scale for Amyloid-Related Imaging Abnormalities

INTRODUCTION

Immunotherapeutic agents against amyloid beta (Aβ) are associated with adverse events, including amyloid-related imaging abnormalities with edema and effusion (ARIA-E). Recently, a magnetic resonance imaging (MRI) rating scale was developed for ARIA-E detection and classification. The aim of this study was to validate the use of this rating scale in a larger patient group with multiple raters.

METHODS

MRI scans of 75 patients (29 with known ARIA-E and 46 control subjects) were analyzed by five neuroradiologists with different degrees of expertise, according to the ARIA-E rating scale. For each patient, we included a baseline and a follow-up fluid-attenuated inversion recovery image. Interrater agreement was calculated using intraclass correlation coefficient (ICC).

RESULTS

On average, 4.1% of the ARIA-E cases were missed. We observed a high interrater agreement for scores of sulcal hyperintensity (SH; ICC = .915; 95% CI 85-95) and for the combined scores of the 2 ARIA-E findings, parenchymal hyperintensity (PH) and SH (ICC = .878; 95% CI 79-93). A slightly lower agreement for PH (ICC = .678; 95% CI 51-81) was noted.

CONCLUSION

The ARIA-E rating scale is a simple tool to evaluate the extent of ARIA-E in patients recruited into Aβ-lowering therapeutic trials. It shows high interrater agreement among raters with different degrees of expertise.

Age-Stratified Prevalence of Mild Cognitive Impairment and Dementia in European Populations: A Systematic Review

The prevalence of mild cognitive impairment (MCI) and dementia according to age remain uncertain. We systematically extracted age-stratified estimates of MCI and dementia prevalence reported in European studies published since 1995, and performed meta-analyses for dementia. We identified 10 relevant studies on MCI and 26 studies on dementia. Studies on MCI presented substantial heterogeneity preventing a meta-analysis, with a majority reporting an increase in prevalence at ≥75 years old. Pooled prevalence of dementia rose continuously from 55 years of age, reaching 44.7% (39.8; 49.6) in those ≥95 years of age. Homogenization of MCI criteria, and additional studies in Northern European population would be warranted.

The Impact of Alzheimer's Disease on the Chinese Economy

BACKGROUND

Recent increases in life expectancy may greatly expand future Alzheimer's Disease (AD) burdens. China's demographic profile, aging workforce and predicted increasing burden of AD-related care make its economy vulnerable to AD impacts. Previous economic estimates of AD predominantly focus on health system burdens and omit wider whole-economy effects, potentially underestimating the full economic benefit of effective treatment.

METHODS

AD-related prevalence, morbidity and mortality for 2011-2050 were simulated and were, together with associated caregiver time and costs, imposed on a dynamic Computable General Equilibrium model of the Chinese economy. Both economic and non-economic outcomes were analyzed.

FINDINGS

Simulated Chinese AD prevalence quadrupled during 2011-50 from 6-28 million. The cumulative discounted value of eliminating AD equates to China's 2012 GDP (US$8 trillion), and the annual predicted real value approaches US AD cost-of-illness (COI) estimates, exceeding US$1 trillion by 2050 (2011-prices). Lost labor contributes 62% of macroeconomic impacts. Only 10% derives from informal care, challenging previous COI-estimates of 56%.

INTERPRETATION

Health and macroeconomic models predict an unfolding 2011-2050 Chinese AD epidemic with serious macroeconomic consequences. Significant investment in research and development (medical and non-medical) is warranted and international researchers and national authorities should therefore target development of effective AD treatment and prevention strategies.

Amyloid-related imaging abnormalities-haemosiderin (ARIA-H) in patients with Alzheimer's disease treated with bapineuzumab: a historical, prospective secondary analysis

BACKGROUND

Amyloid-related imaging abnormalities due to haemosiderin deposition (ARIA-H) occur in patients with mild to moderate dementia due to Alzheimer's disease (AD) and have been reported with increased incidence in clinical trials of amyloid-lowering therapies under development for AD.

OBJECTIVE

Our objective was to explore the relationship between the incidences of ARIA-H during treatment with placebo and different doses of bapineuzumab, a humanised monoclonal antibody directed against amyloid β.

METHODS

Two neuroradiologists independently reviewed 2572 GRE/T2* MRI sequences from 262 participants in two phase two clinical trials of bapineuzumab and an open-label extension study. Readers were blinded to the participant's therapy, APOE ε4 genotype and medical history.

RESULTS

Several risk factors for small ARIA-H <10 mm (microhaemorrhages) were identified: APOE ε4, bapineuzumab treatment, pre-existing small ARIA-H and use of antithrombotics. The HR (95%CI) for incident ARIA-H <10 mm associated with the number of APOE ε4 alleles was 11.9 (3.3 to 42.5) for 2 versus no alleles and 3.5 (1.0 to 12.0) for 1 versus no allele. The HR for bapineuzumab therapy was 3.5 (1.0 to 12.0); for the presence of baseline ARIA-H <10 mm, it was 3.5 (1.6 to 7.8), and for the use of antithrombotic agents it was 2.2 (1.0 to 4.8). The incidence rate for ARIA-H <10 mm was elevated only in the initial 6 months of active treatment and declined after this interval to a rate similar to that observed in the group treated with placebo.

CONCLUSIONS

ARIA-H represents a spectrum of MRI findings due to haemosiderin deposition that appears to be related to impaired vascular integrity. The increased risk for ARIA-H associated with APOE ε4 allele frequency, pre-existing ARIA-H, treatment with bapineuzumab and use of antithrombotic agents provides additional support for this hypothesis of loss of integrity of cerebral vessels due to amyloid burden.

TRIAL REGISTRATION

NCT00112073 and NCT00606476.

Incidence of stroke and seizure in Alzheimer's disease dementia

BACKGROUND

the objective of the study was to estimate and compare the incidence rates of ischaemic and haemorrhagic stroke and seizure among cohorts with and without Alzheimer's disease (AD) dementia.

METHODS

we conducted a retrospective cohort study using electronic medical records (EMRs) from primary care practices that participated in The Health Improvement Network (THIN) in the United Kingdom from 1 January 1990 to 31 July 2009. For each AD-dementia patient, we selected one general population control patient without AD-dementia matched to one AD-dementia patient on year of birth, sex and physician practice.

FINDINGS

the AD-dementia cohorts were 68% female and averaged 80 years of age at the start of follow-up. Populations for analysis included 19,902 AD-dementia and matched non-AD-dementia patients with no history of stroke at baseline in which 790 incident cases of stroke occurred, and similarly, 22,084 AD-dementia and matched patients with no history of seizure at baseline in which 286 cases of seizure occurred. After adjusting for risk factors for each outcome, hazard ratios comparing AD-dementia with non-AD-dementia patients indicated higher rates among AD-dementia patients for stroke (HR = 1.29, 95% CI 1.11, 1.50) and seizure (HR = 5.31, 95% CI 3.97, 7.10). For stroke and seizure, the incidence rate ratios comparing AD-dementia patients with non-AD-dementia controls were greatest for the younger age groups. AD-dementia was observed to be a risk factor for both haemorrhagic stroke and seizures. Increasing age was associated with a decrease in relative risk and an increase in absolute risk.

Understanding the relative importance of preserving functional abilities in Alzheimer's disease in the United States and Germany

PURPOSE

To estimate the relative importance that Alzheimer's disease (AD) caregivers in the United States and Germany place on preserving patients' ability to perform activities of daily living.

METHODS

US and German residents providing care for a person with AD completed an online preference survey. Each respondent completed five best-worst scaling questions. Each question related to five of 10 activities from the Disability Assessment for Dementia scale. Preference weights, indicating the relative importance of preserving the ability to perform these 10 activities for 36 months, were estimated using maximum-difference scaling. A separate model was estimated for each country.

RESULTS

Four hundred and three US and 400 German caregivers completed the survey. In both countries, preserving a patients' ability to use the toilet without accidents was the most important activity and handling money was the least important activity. There were few differences between US and German caregivers in the relative importance across activities.

CONCLUSIONS

Caregivers generally placed greater importance on preserving basic activities of daily living than on preserving instrumental activities of daily living. Understanding differences in the relative importance of functional items in the DAD may contribute to a better understanding of the benefits of different AD treatment and support measures.

Rate of conversion from prodromal Alzheimer's disease to Alzheimer's dementia: a systematic review of the literature

Background

The purpose of this study was to summarize published estimates for conversion from mild cognitive impairment or amnestic mild cognitive impairment to Alzheimer's dementia. We carried out a systematic review of English language publications to identify cohort studies published since January 2006 that reported the risk or rate of conversion.

Summary

Thirty-two cohort studies were identified, of which 14 reported annualized conversion rates (ACRs). Conversions over 1 year ranged from 10.2 to 33.6% (5 studies, median: 19.0%), and over 2 years from 9.8 to 36.3% (7 studies, median: 18.6%). ACRs ranged from 7.5 to 16.5% (7 studies, median: 11.0%) per person-year for studies recruiting from clinics, and from 5.4 to 11.5% (7 studies, median: 7.1%) for community samples.

Key Message

Extensive variation was observed in conversion rates due to the population sampled, diagnostic criteria, and duration, and because many studies did not account for loss to follow-up.

MR imaging features of amyloid-related imaging abnormalities

BACKGROUND AND PURPOSE

AD is one of the few leading causes of death without a disease-modifying drug; however, hopeful agents are in various phases of development. MR imaging abnormalities, collectively referred to as amyloid-related imaging abnormalities, have been reported for several agents that target cerebral Aβ burden. ARIA includes ARIA-E, parenchymal or sulcal hyperintensities on FLAIR indicative of parenchymal edema or sulcal effusions, and ARIA-H, hypointense regions on gradient recalled-echo/T2* indicative of hemosiderin deposition. This report describes imaging characteristics of ARIA-E and ARIA-H identified during studies of bapineuzumab, a humanized monoclonal antibody against Aβ.

MATERIALS AND METHODS

Two neuroradiologists with knowledge of imaging changes reflective of ARIA reviewed MR imaging scans from 210 bapineuzumab-treated patients derived from 3 phase 2 studies. Each central reader interpreted the studies independently, and discrepancies were resolved by consensus. The inter-reader κ was 0.76, with 94% agreement between neuroradiologists regarding the presence or absence of ARIA-E in individual patients.

RESULTS

Thirty-six patients were identified with incident ARIA-E (17.1%, 36/210) and 26 with incident ARIA-H (12.4%, 26/210); of those with incident ARIA-H, 24 had incident microhemorrhages and 2 had incident large superficial hemosiderin deposits.

CONCLUSIONS

In 49% of cases of ARIA-E, there was the associated appearance of ARIA-H. In treated patients without ARIA-E, the risk for incident blood products was 4%. This association between ARIA-E and ARIA-H may suggest a common pathophysiologic mechanism. Familiarity with ARIA should permit radiologists and clinicians to recognize and communicate ARIA findings more reliably for optimal patient management.

An MRI rating scale for amyloid-related imaging abnormalities with edema or effusion

BACKGROUND AND PURPOSE

Immune therapy against amyloid-β appears to be a promising target in Alzheimer disease. However, a dose-related risk for ARIA on FLAIR images thought to represent parenchymal vasogenic edema or sulcal effusion (termed "ARIA-E"), has been observed in clinical trials. To assess the intensity of ARIA-E presentation, an MR imaging scale that is both reproducible and easily implemented would assist in monitoring and evaluating this adverse event.

MATERIALS AND METHODS

On the basis of a review of existing cases from a phase II bapineuzumab study, a scale was constructed with a 6-point score for the 6 regions on each side of the brain (range, 0-60). Scores would be obtained for both parenchymal and sulcal hyperintensities and frequently co-occurring gyral swelling. Inter-rater reliability between 2 neuroradiologists was evaluated in 20 patients, 10 with known ARIA-E and 10 without, by using the intraclass correlation coefficient.

RESULTS

The 2 raters had excellent agreement in the identification of ARIA-E cases. A high inter-rater agreement was observed for scores of parenchymal hyperintensity (ICC = 0.83; 95% CI, 48-96) and sulcal hyperintensity (ICC = 0.89; 95% CI, 63-97) and for the combined scores of the 2 ARIA-E findings (ICC = 0.89; 95% CI, 62-97). Gyral swelling scores were observed to have lower inter-rater agreement (ICC = 0.54; 95% CI, -0.06-0.86).

CONCLUSIONS

The proposed rating scale provides a reliable and easily implemented instrument to grade ARIA-E imaging findings. We currently do not recommend including swelling.

Mild cognitive impairment: disparity of incidence and prevalence estimates

BACKGROUND

The purpose of conducting this study was to identify areas of concordance and sources of variation for the published rates of prevalence and incidence associated with various definitions for mild cognitive impairment (MCI).

METHODS

The study used systematic review of studies published in English since 1984. Studies were identified by searching MEDLINE and EMBASE databases. Population-based observational studies of incidence or prevalence of MCI and related terms were eligible for inclusion.

RESULTS

A total of 3,705 citations were identified, and 42 were accepted for inclusion; 35 included data on prevalence and 13 on incidence. The following four terms predominated: age-associated memory impairment (AAMI); cognitive impairment no dementia (CIND); MCI; and amnestic MCI (aMCI). Within each term, the operational definition varied. Substantial variation was observed for both incidence (MCI: 21.5-71.3; aMCI: 8.5-25.9 per 1,000 person-years) and prevalence of each definition of cognitive impairment (AAMI 3.6%-38.4%; CIND 5.1%-35.9%; MCI 3%-42%; aMCI 0.5%-31.9%). CIND and MCI showed increasing prevalence among older age groups, whereas age-specific rates of aMCI were lower and without any apparent age relationship.

CONCLUSIONS

Prevalence and incidence estimates associated with MCI vary greatly both between definitions and within a definition across the 42 publications. These wide differences pose a significant challenge to our understanding of the social burden of this disease. Enhancement and standardization of operational definitions of the subtypes of cognitive impairment could improve estimates of disease burden and provide a mechanism to assist in the identification of individuals at risk for future Alzheimer's disease and other dementias.

Amyloid-related imaging abnormalities in patients with Alzheimer's disease treated with bapineuzumab: a retrospective analysis

BACKGROUND

Amyloid-related imaging abnormalities (ARIA) have been reported in patients with Alzheimer's disease treated with bapineuzumab, a humanised monoclonal antibody against amyloid β. ARIA include MRI signal abnormalities suggestive of vasogenic oedema and sulcal effusions (ARIA-E) and microhaemorrhages and haemosiderin deposits (ARIA-H). Our aim was to investigate the incidence of ARIA during treatment with bapineuzumab, and evaluate associated risk factors.

METHODS

Two neuroradiologists independently reviewed 2572 fluid-attenuated inversion recovery (FLAIR) MRI scans from 262 participants in two phase 2 studies of bapineuzumab and an open-label extension study. Readers were masked to the patient's treatment, APOE ɛ4 genotype, medical history, and demographics. Patients were included in risk analyses if they had no evidence of ARIA-E in their pre-treatment MRI, had received bapineuzumab, and had at least one MRI scan after treatment. We used Kaplan-Meier survival analysis to examine the distribution of incident ARIA-E from the start of bapineuzumab treatment and proportional hazards regression models to assess risk factors associated with ARIA.

FINDINGS

210 patients were included in the risk analyses. 36 patients (17%) developed ARIA-E during treatment with bapineuzumab; 15 of these ARIA-E cases (42%) had not been detected previously. 28 of these patients (78%) did not report associated symptoms. Adverse events, reported in eight symptomatic patients, included headache, confusion, and neuropsychiatric and gastrointestinal symptoms. Incident ARIA-H occurred in 17 of the patients with ARIA-E (47%), compared with seven of 177 (4%) patients without ARIA-E. 13 of the 15 patients in whom ARIA were detected in our study received additional treatment infusions while ARIA-E were present, without any associated symptoms. Occurrence of ARIA-E increased with bapineuzumab dose (hazard ratio [HR] 2·24 per 1 mg/kg increase in dose, 95% CI 1·40-3·62; p=0·0008) and presence of APOE ɛ4 alleles (HR 2·55 per allele, 95% CI 1·57-4·12; p=0·0001).

INTERPRETATION

ARIA consist of a spectrum of imaging findings with variable clinical correlates, and some patients with ARIA-E remain asymptomatic even if treatment is continued. The increased risk of ARIA among APOE ɛ4 carriers, its association with high bapineuzumab dose, and its timecourse in relation to dosing suggest an association between ARIA and alterations in vascular amyloid burden.

FUNDING

Elan Corporation, Janssen Alzheimer Immunotherapy, Wyeth Pharmaceuticals, and Pfizer.

Prevalence of apolipoprotein E4 genotype and homozygotes (APOE e4/4) among patients diagnosed with Alzheimer's disease: a systematic review and meta-analysis

BACKGROUND

Population allele frequencies of apolipoprotein E (APOE) vary by geographic region. The purpose of this study is to summarize and evaluate published estimates for the prevalence of APOE e4 carrier status among the population diagnosed with Alzheimer's disease (AD) by geographic region and country.

METHODS

A systematic review of English-language publications from January 1, 1985, through May 31, 2010, was conducted. Studies reporting APOE e4 status for patients diagnosed with AD were included in the analysis; trials and autopsies were excluded. APOE e4 data were pooled, and prevalence and 95% confidence intervals (CIs) were calculated.

RESULTS

Pooled estimates for APOE e4 carrier prevalence data were derived from 142 independent samples: 48.7% (95% CI: 46.5-51.0), and from 73 samples for e4/4 (homozygotes): 9.6% (95% CI: 8.4-10.8). The highest estimates were in Northern Europe: 61.3% (95% CI: 55.9-66.7), e4/4 prevalence: 14.1% (95% CI: 12.2-16.0). The lowest estimates were in Asia and Southern Europe. Substantial heterogeneity of these prevalence estimates was observed.

CONCLUSIONS

APOE e4 genotype prevalence varies among AD patients by region and within each country. Further exploration is warranted to better understand the substantial heterogeneity of these prevalence estimates.

Apolipoprotein E ε4 prevalence in Alzheimer's disease patients varies across global populations: a systematic literature review and meta-analysis

BACKGROUND

The ε4 allele of apolipoprotein E (APOE) is associated with Alzheimer's disease (AD). However, attributable risk due to APOE4 varies by region and by race/ethnicity.

METHODS

A literature review and meta-analysis were conducted to estimate the prevalence of APOE4 by geographic area among AD patients.

RESULTS

Although estimates varied significantly by study design and case definition, AD patients recruited in Asian and southern European/Mediterranean communities seemed to have significantly lower E4 carrier status estimates (37 and 43%) than those recruited in North America (58%) or northern Europe (64%; all: p < 0.05).

CONCLUSIONS

APOE4 genotype frequency varies among AD patients in regional patterns similar to that of the general population. Study level differences may also contribute to the heterogeneity of published estimates of APOE4 in AD cases.

Hip fracture risk and subsequent mortality among Alzheimer's disease patients in the United Kingdom, 1988-2007

BACKGROUND

hip fractures result in a significant burden to the patient, their caregivers and the health care system. Patients with Alzheimer's disease (AD) have a higher incidence of hip fracture compared with other older people without AD, although it is not clear if AD is an independent risk factor for hip fracture.

METHODS

a retrospective cohort study was conducted using anonymised electronic medical records from primary care practices in the United Kingdom. Proportional hazards regression modelling with adjustment for potential confounders was used to evaluate AD as an independent risk factor for predicting hip fractures.

RESULTS

the incidence of hip fracture among patients with and without AD was 17.4 (95% CI, 15.7-19.2) and 6.6 (95% CI, 5.8-7.6) per 1,000 person years, respectively. Patients with AD had a hazard that was 3.2 (95% CI, 2.4-4.2) times that of non-AD patients after controlling for potential confounders. AD patients who experienced a hip fracture also had an increased mortality rate compared with non-AD patients who experienced a hip fracture (hazard ratio = 1.5; 95% CI, 1.1-1.9).

CONCLUSION

patients with AD and their caregivers should be advised on how to prevent hip fractures and more attention should be given to AD patients who are undergoing rehabilitation following a hip fracture.

Web-based application to project the burden of Alzheimer's disease

BACKGROUND

Health care planning and research would benefit from tools that enable researchers to project the future burden of Alzheimer's disease (AD) and evaluate the effect of potential interventions.

METHODS

We created a web-based application of the AD prevalence model developed by Brookmeyer et al (Am J Public Health 1998;88:1337-42; Alzheimers Dement 2007;3:186-91). The user defines the disease parameters and any interventions that may either reduce risk or slow disease progression. We expanded the parameters to include the cost and weights for disability-adjusted life years.

APPLICATION

The secure, web-based application generates detailed AD projections for each calendar year to 2050, and allows users to create personal accounts for them to save, retrieve, and modify the input parameters. The flexibility of the application is illustrated with a forecast for the state of Maryland, USA.

CONCLUSIONS

The application generates AD burden projections, costs, and disability-adjusted life years, along with changes associated with potential interventions.

Adverse events associated with common therapy regimens for moderate-to-severe Crohn's disease

OBJECTIVES

We sought to determine whether treatment with steroids, immunosuppressives (ISs), and anti-tumor necrosis factor (TNF) agents is associated with an increased risk of adverse events in patients with Crohn's disease (CD).

METHODS

This study analyzed claims from patients with CD and controls without CD from the United States with private insurance (2002-2005). Patients were classified by treatment with steroids, ISs, anti-TNF agents, combinations of two or three, and none of these medications. Follow-up adverse events in patients with CD and controls were compared across different treatment categories and are presented as hazard ratios (HRs) and 95% confidence intervals (CIs). Within the CD patients, a subset analysis examined the relationship between therapies and outcomes.

RESULTS

A total of 22,310 patients with CD (8,581 longitudinal cohort cases) and 111,550 controls were identified. Compared with the controls, CD patients had higher rate ratios for all pre-specified events. Within the CD patient population subgroup, monotherapy with steroids, ISs, or anti-TNF agents was associated with an increased risk of tuberculosis (TB) (HR 2.7; 95% CI, 1.0-7.3), candidiasis (HR 2.7; 95% CI, 1.8-4.0), herpes zoster (HR 1.7; 95% CI, 1.0-2.7), sepsis (HR 1.3; 95% CI, 1.1-1.5), demyelinating conditions (HR 3.2; 95% CI, 1.5-6.9), and cervical dysplasia (HR 1.5; 95% CI, 1.2-2.0) as compared with patients not receiving these medications. The use of two or three of these medications further increased these risks: TB (HR 7.4; 95% CI, 2.1-26.3), candidiasis (HR 3.8; 95% CI, 2.0-7.6), herpes zoster (HR 3.7; 95% CI, 1.8-7.5), sepsis (HR 1.6; 95% CI, 1.2-2.1), and cervical dysplasia (HR 1.8; 95% CI, 1.1-3.0).

CONCLUSIONS

Treatment with steroids, ISs, or anti-TNF agents singly and in combination in patients with CD is associated with increased risks of infection, demyelinating disorders, and cervical dysplasia.

Burden of Alzheimer's disease and association with negative health outcomes

OBJECTIVE

To examine the association of Alzheimer's disease (AD) with common chronic conditions, acute care events, and risk of hospitalization.

STUDY DESIGN

Retrospective matched cohort analysis.

METHODS

Community-dwelling subjects with a diagnosis of and/or medication for AD were matched to subjects without AD based on age, sex, and geographic region. Administrative claims from commercially insured health plans for medical and pharmacy services provided from January 1, 2000, to March 31, 2006 (inclusive) were analyzed. The Deyo Charlson Index (DCI) was used to assess the number of chronic conditions. The outcomes of interest were risk of fractures and hospitalization.

RESULTS

Among 5396 persons with AD and a matched cohort of 5396 persons without the condition, subjects with AD were more likely to have a diagnosis for any of the DCI components, had a higher rate of fractures (17.7% vs 7.9%, P <.00) and other urgent medical events (eg, pneumonia 14.0% vs 6.3%, P <.00), and were more likely to be hospitalized (odds ratio = 1.7; 95% confidence interval = 1.5, 1.9). There were significant differences in the medication use between the 2 groups, with the use of psychotics/tranquilizers 9-fold higher among persons with AD.

CONCLUSION

Persons with AD have higher odds of experiencing a fracture, being hospitalized, and requiring other acute care medical services than those without AD. The disease also is associated with a higher prevalence of common chronic conditions.

Correlates of osteoprotegerin levels in women and men

Osteoprotegerin (OPG) is a potent antiresorptive molecule that binds the final effector for osteoclastogenesis, receptor activator of NF-kappaB ligand (RANK-L). OPG production is regulated by a number of cytokines and hormones, including sex steroids, but there are few data on age and gender effects on circulating serum OPG levels, as well as possible relationships between OPG levels and bone turnover markers or bone mineral density (BMD). Thus, we measured serum OPG levels in an age-stratified, random sample of men (n = 346 age range, 23-90 years) and women (n = 304; age range 21-93 years) and related them to sex steroid levels, bone turnover markers and BMD. Serum OPG levels increased with age in both men (R = 0.39, p < 0.001) and women (R = 0.18, p < 0.01). Premenopausal women had higher OPG levels than men under age 50 years (171 +/- 6 pg/ml vs 134 +/- 6 pg/ml, respectively, p < 0.001), whereas serum OPG levels were no different in postmenopausal women compared with men = 50 years (195 +/- 7 pg/ml vs 188 +/- 7 pg/ml, respectively, p = 0.179). OPG levels correlated inversely with serum bioavailable testosterone levels in men = 50 years (R = -0.27, p < 0.001), but no associations were present with either estrogen or testosterone levels in the women. In the men, there was a trend for OPG levels to be associated positively with bone resorption markers and inversely with BMD. Collectively, the gender difference in OPG levels suggests that sex steroids may regulate OPG production in vivo, as has been found in vitro. Moreover, OPG production may also rise with increases in bone turnover, probably as a homeostatic mechanism to limit bone loss. Further studies directly testing these hypotheses should provide additional insights into the potential role of OPG in bone loss related to aging and sex steroid deficiency.

Does arsenic exposure increase the risk for circulatory disease?

Studies of residents in communities with high endemic concentrations of arsenic in drinking water suggest a deleterious effect on the circulatory system; however, studies among workers with high occupational exposures generally have shown either no or weak associations. This discrepancy could be a result of the healthy worker effect, including the healthy hire component and the healthy worker survivor effect (HWSE). Therefore, the authors conducted analyses of arsenic exposure in relation to circulatory disease mortality among 2,802 Tacoma, Washington, smelter workers by using 1) internal comparisons to control for the healthy hire effect and 2) the lagging method, adjustment for employment status, and the G-null test to control for the HWSE. Both lagging and adjustment for work status increased circulatory mortality rate ratios at all exposure levels, as compared with a baseline Poisson model. This excess mortality was limited to cardiovascular disease; no excess was observed for cerebrovascular disease. G-null analyses suggested no adverse effect, but power was very limited for this analysis. Overall, these results may indicate that the HWSE obscures an effect of arsenic on circulatory disease. Since cardiovascular deaths constitute about one-third of total mortality, small rate ratios translate into large numbers of excess deaths and, if causal, could be of wide public health significance. Further studies of arsenic exposure and cardiovascular disease are needed, and those conducted in occupational cohorts must control for the HWSE.

The impact of asthma and asthma-like illness in Seattle school children

The impact of asthma and asthma-like illness was measured in a population of 5-9-year-old Seattle public school children. Child health information was obtained from a survey of 1665 parents of first and second grade students to assess medical services use and impaired physical functioning among diagnosed asthmatics and those with current wheezing, defined as wheezing in the past 12 months without a diagnosis of asthma, relative to an asymptomatic population with neither condition. Relative to the asymptomatic population, the prevalence of respiratory-related activity limitation, and perception of poorer child health was larger among diagnosed asthmatics than children with current wheezing. However, the prevalence of sleep disturbances, school absences, medical services use, and parental concern over their child's health was similar for both the asthmatic and wheezing groups relative to the asymptomatic group. Also, in both symptomatic groups, a history of moderate or severe wheezing was associated with an increased prevalence of respiratory-related sleep disturbances and activity limitation. The similarity between the impact of diagnosed asthma and undiagnosed asthma-like illness suggests that the overall social and economic burden of asthma may be higher than previously estimated.

Indoor risk factors for asthma and wheezing among Seattle school children

Indoor risk factors for physician-diagnosed asthma and wheezing in the past 12 months without previous asthma diagnosis were assessed in a survey of parents of 5-9-year-old Seattle primary school students. Among the 925 respondents, 106 (11%) reported a physician diagnosis of asthma, 66 (7%) had wheezing without diagnosis, and 753 (82%) were asymptomatic. After adjusting for age, sex, gender, ethnicity, medical history, socioeconomic status (SES) and parental asthma status, an increased risk of physician diagnosis of asthma was associated with household water damage, the presence of one or more household tobacco smokers, and at least occasional environmental tobacco smoke (ETS) exposure. Similarly, an increased risk of wheezing in the past 12 months among children without diagnosed asthma was associated with household water damage, presence of one or more household tobacco smokers, and occasional or more frequent ETS exposure. No increased risk of either condition was associated with gas, wood, or kerosene stove use, household mold, basement water, or wall/window dampness. Similarities in the indoor risk factors patterns between diagnosed asthma and wheezing without diagnosis suggested a similar etiology of these two conditions. The slightly higher association between ETS and asthma may indicate that parents of diagnosed asthmatics were more conscious of ETS, and were more likely to prohibit household smoking by resident smokers. Future research is needed to quantify which aspects of household water damage are related to respiratory illness.

A cost of illness study of allergic rhinitis in the United States

BACKGROUND

Allergic rhinitis is a common condition, but the burden of this condition on the national economy is not well understood.

OBJECTIVE

The purpose of this study was to estimate the national direct and indirect costs of allergic rhinitis.

METHODS

Data from the National Medical Expenditure Survey were used to provide estimates of resource utilization, medical expenditures, and lost productivity. With the complex survey design, variance estimates were used to construct confidence intervals for cost estimates of resource utilization and lost productivity.

RESULTS

It is estimated that approximately 39 million persons in the United States experienced allergic rhinitis in 1987. However, only 12.3% (4.8 million) sought medical treatment for allergic rhinitis. The total estimated cost of the condition, in 1994 dollars, was $1.23 billion (95% confidence interval, $846 million to $1.62 billion). Direct medical expenses accounted for 94% of total costs. Allergic rhinitis results in approximately 811,000 missed workdays, 824,000 missed school days, and 4,230,000 reduced activity days.

CONCLUSION

Allergic rhinitis clearly creates a burden in terms of the number of persons affected, total expenditures, and lost productivity. It also appears that a relatively large proportion of persons with allergic rhinitis were not seeking medical treatment.

Controlling the healthy worker survivor effect: an example of arsenic exposure and respiratory cancer

OBJECTIVE

This investigation sought to examine whether methods proposed to control the healthy worker survivor effect would influence the shape or magnitude of the dose-response curve for respiratory cancer induced by arsenic.

METHODS

Results from an unadjusted analysis are compared with results obtained by applying four different methods for control of the healthy worker survivor effect to data on arsenic exposure and respiratory cancer. The four methods are: exposure lag, adjustment for work status, cohort restriction, and the G null test.

RESULTS

Cohort restriction gave erratic results depending upon the minimum years of follow up used. Exposure lag substantially increased the rate ratios and a non-linear shape (decreasing slope) compared with an unlagged analysis. Adjusting for work status (currently employed upsilon retired or otherwise not employed) yielded slightly higher rate ratios than an unadjusted analysis, with an overall shape similar to the baseline analysis. Results from the G null test procedure of Robins (1986), although not directly comparable with the baseline analysis, did show an adverse effect of exposure that seemed to reach a maximum when exposure was lagged between 10 and 20 years.

CONCLUSIONS

All results confirm an adverse effect of arsenic exposure on respiratory cancer. In these data, it seems that the healthy worker survivor effect was not strong enough to mask the strong effect of arsenic exposure on respiratory cancer. Nevertheless, several methods show a stronger association between arsenic exposure and respiratory cancer after adjustment for the healthy worker survivor effect, suggesting that for weaker causal associations, studies not controlling for this source of bias will have low power to detect results. Although the G methods are theoretically the most unbiased, further work elucidating the validity of the assumptions underlying lagging, adjustment for work status, and the G methods are needed before clear recommendations can be made.

Controlling for time-since-hire in occupational studies using internal comparisons and cumulative exposure

Employees within an occupational cohort may demonstrate a more favorable mortality experience while maintaining employment than those who leave employment. At the same time, they may experience an apparent decline in health with time-since-hire. The time-since-hire effect may occur independently of exposure but may nevertheless result in groups categorized by cumulative exposure that are not comparable. Controlling for time-since-hire appears to solve this problem. To quantify the empirical bias in estimates of exposure effect due to confounding from time-since-hire, we analyzed two occupational cohorts using Poisson regression with and without adjustment for time-since-hire or time-since-start-of-follow-up. In a cohort exposed to airborne arsenic, a strong dose-response relation with respiratory cancer mortality had been established. In a cohort exposed to external, penetrating ionizing radiation, a weak and controversial dose-response relation had been reported. The parameter estimates relating exposure to disease from the models that explicitly adjusted for time-since-hire or time-since-start-of-follow-up are within 10% of the estimates from models that did not. It appears, from this empirical analysis of two datasets, that occupational studies may not need to adjust explicitly for such time-related factors as time-since-hire or time-since-start-of-follow-up if these are implicitly controlled through other variables in the model.

US asthma mortality: 1941 to 1989

BACKGROUND

Asthma mortality in the United States has nearly doubled in the past 10 years. An examination of long-term trends in United States asthma mortality places the current mortality rates in a historical perspective, identifies high-risk groups for interventions, and may provide clues to the etiology of asthma mortality.

METHODS

Asthma deaths for the population aged 5 to 34 years were abstracted from United States vital statistics reports for the period 1941 through 1989. Race-specific and sex-specific mortality rates were age-adjusted to the 1989 estimate of the US population.

RESULTS

Among the population aged 5 to 34 years, three distinct periods of asthma mortality were observed: 1941 to 1964, 1965 to 1977, and 1978 to 1989. From 1941 through 1964, nonwhites exhibited a gradual increase in asthma mortality rates; in contrast, whites showed no change in mortality rates. From 1965 through 1977, a marked decline in mortality rates was observed among both nonwhites and whites. From 1978 through 1989, asthma mortality increased with a near doubling in the mortality rates in both nonwhites and whites. Overall, nonwhites had mortality rates 4 times those of whites, with nonwhite males aged 15 to 34 at the greatest risk of death due to asthma. Age and sex differences in asthma mortality are also apparent. The population aged 15 to 34 years generally had mortality rates greater than the population aged less than 15 years. For the population aged 5 to 14 years, regardless of calendar year, females had lower mortality rates than males. Among the population aged 15 to 34 years, females experienced twice the mortality rates of males prior to 1965; however, by the 1980s this relationship no longer existed.

CONCLUSIONS

The asthma mortality rate for nonwhites was 4-fold higher than for whites, although the reported prevalence of asthma is less than 2-fold greater. The population aged 15 to 34 years had higher mortality rates than the population aged 5 to 14 years despite exhibiting a lower prevalence of asthma. There are also gender differences in asthma mortality. These differences may be attributable to differences in asthma severity, or differences in disease management or reflect actual differences in mortality. These findings suggest that the search for clues to understand the increase in asthma mortality from 1978 to 1989 should include an attempt to understand why asthma mortality declined in the preceding decade from 1967 to 1977 as these may not be completely unrelated trends.

The evolving concept of the healthy worker survivor effect

The "healthy worker survivor effect" describes a continuing selection process such that those who remain employed tend to be healthier than those who leave employment. In an analysis of exposure-response patterns in an occupational study, the healthy worker survivor effect generally attenuates an adverse effect of exposure. In practical terms, such attenuation will be more problematic when evaluating subtle rather than strong associations. The use of an internal referent does not guarantee elimination of this effect, since by definition, it manifests within an occupational cohort. Although documented over 100 years ago, there is little consensus regarding the most appropriate method to control for the healthy worker survivor effect. Four methods have been proposed for its control: (1) restriction of the cohort to survivors of a fixed number of years of follow-up, (2) lagging the exposure to exclude recent exposure incurred by those who remained on the job, (3) adjusting for employment status as a confounder, and (4) treating the healthy worker survivor effect simultaneously as an intermediate and confounding variable by means of the G-null test or its extension, G-estimation analysis, using structurally nested failure time models. This paper reviews the concept of the healthy worker survivor effect and the four methods to control for it.

Natural history of prostatism: urinary flow rates in a community-based study

Urinary flow rates were measured in a randomly selected community sample of more than 2,000 men 40 to 79 years old with no history of prostate surgery, prostate cancer or certain other diseases known to interfere with normal voiding. Peak urinary flow rates decreased from a median of 20.3 ml. per second in men 40 to 44 years old to 11.5 ml. per second for men 75 to 79 years old, while voided volumes decreased from a median of 355.5 to 222.5 ml. for the same age ranges. Peak flow rates of less than 10 ml. per second were found in 6% of the men aged 40 to 44 years, increasing to 35% among men aged 75 to 79 years. Urological standards for peak urinary flow rate should be based on community data, and should account for age and voided volume. Our study may serve as the starting point for the development of community-based flow rate normal ranges. Nomograms are given to permit estimating flow rate percentiles as a function of age and voided volume.

Natural history of benign prostatic hyperplasia and risk of prostatectomy. The Baltimore Longitudinal Study of Aging

The natural history of prostatism (clinically diagnosed benign prostatic hyperplasia) is examined based on symptom questionnaires and digital rectal examinations administered periodically to 1,057 men followed prospectively for up to thirty years in the Baltimore Longitudinal Study of Aging (BLSA). Benign prostatic hyperplasia (BPH) was clinically diagnosed in 527 men, 110 had a prostatectomy for BPH, and in 21 prostate cancer developed. Among men aged sixty or older with prostatic enlargement and obstructive symptoms, the twenty-year probability of surgery was 39 percent; for men aged fifty to fifty-nine years this probability was 24 percent; and for men aged forty to forty-nine years, the probability was 13 percent. The age-specific prevalence of clinically diagnosed BPH agreed closely at all ages with the age-specific autopsy prevalence of pathologically defined BPH from a published international compilation of 5 independent autopsy studies involving 1,075 prostates.

Cumulative prevalence of prostatism matches the autopsy prevalence of benign prostatic hyperplasia

The age-specific cumulative prevalence of clinically diagnosed prostatism among 1,057 generally healthy men followed for up to 30 years in the Baltimore Longitudinal Study of Aging was compared with the age-specific autopsy prevalence of pathologically defined benign prostatic hyperplasia (BPH) from a published compilation of five independent autopsy studies involving 1,075 prostates. Clinical prevalences based on medical histories and digital rectal examinations showed excellent agreement with autopsy prevalences at all ages. Despite the well-known lack of correspondence between prostatism and objective urologic evidence of BPH in individual patients, these results suggest that the proportion of the male population with clinically recognizable prostatism by a given age is about the same as the proportion with pathological evidence of BPH.

Symptoms and signs of prostatism as risk factors for prostatectomy

Symptom questionnaires and physical examinations administered periodically to 1,057 men followed prospectively for up to 30 years in the Baltimore Longitudinal Study of Aging (BLSA) were analyzed to determine which symptoms of prostatism were predictive of subsequent prostatectomy for benign prostatic hyperplasia (BPH). Change in the size and force of the urinary stream (P = 0.0001) and a sensation of incomplete emptying (P = 0.0005) were the only symptoms positively predictive in a multivariate analysis. Prostate enlargement by rectal palpation was an independent risk factor for prostatectomy (P less than 0.03) and was associated with hesitancy (P less than 0.02) and a sensation of incomplete emptying (P less than 0.02). Of the men with all three risk factors, 37% (41/112) eventually had a prostatectomy, in contrast to 8% of the remaining 945 men. This is the first study to document the extent to which urologic symptoms and a digital rectal examination predict subsequent prostatectomy for BPH.