失智症早期篩檢偵測工具效度及成本效益之統計評估

(1)

國立台灣大學公共衛生學院流行病學與預防醫學研究所

博士論文

Graduate Institute of Epidemiology and Preventative Medicine College of Public Health

National Taiwan University Doctoral Dissertation

失智症早期篩檢偵測工具效度及成本效益之統計評估

Statistical Evaluation of Validity and

Cost-effectiveness of Two Screening Tools for Early Detection of Dementia

盧韻如 Yun-Ru Lu

指導教授 : 劉宏輝博士、陳秀熙博士 Advisor: Horng-Huei Liou, Ph.D.

Hsiu-Hsi Chen, Ph.D.

中華民國104年1月

Jan, 2015

(2)

致謝

豐富的博士生涯將在此告一段落，感謝我的指導教授「陳秀熙教授」及

「劉宏輝教授」帶領我進入學術的殿堂，並將所學事物回饋給社會。

感謝論文指導委員程蘊菁老師、嚴明芳老師、潘信良醫師、許維志醫師所給予寶貴意見，讓本論文臻於至善，並給予我努力的方向。感謝陳立昇老師、范靜媛老師、邱月瑕老師在社區篩檢上給我的幫助，讓我看到陳教授口中的「傳承」

並瞭解做研究應有的態度及方法，其中特別感謝陳立昇老師在技術上指導。

最後，感謝我的家人，謝謝你們一路陪我走過來。

(3)

中文摘要

研究背景

隨著老年人口逐漸增加同時，失智症也有同時增加的趨勢，因此考慮一個符合有效益且有效率的社區型失智症篩檢是非常重要的。在此之前，我們需要一系列的研究，包括流行病學關於失智症早期偵測的認知程度，社區失智症篩檢可行工具的效度，和以社區為基礎的失智症篩檢計畫的經濟評估，及發展出一系列以理論為基礎的統計方法。

研究目的

本論文研究目的有三：(1) 利用一新的估計盛行率/發生率比率為基礎的統計回歸模型，來測量早期偵測，對於失智症發生的平均時間的影響(2)利用一貝氏臨床推理模型，來評估同時合併認知功能檢測(MMSE)及訊息提供問卷(AD8)二種篩檢工具，是否會加社區失智症篩檢診斷的精確度 (3)利用馬可夫決策分析模型，評估社區失智症篩檢中，其不同篩檢模的成本效益分析

研究材料及方法

利用 2000-2003 年期間，全民健保費用申報的大型資料庫，我們分析出 65 歲以上有失智症病人，且符合 ICD-9-CM 編碼為 290, 290.40, 331.0 者，共 99,609 位。另一資料庫來自 2013 年社區主動型篩檢，65 歲以上參與者共 183 位。我們收集關於年紀、性別、及就醫地點等相關訊息。利用貝氏方法估出盛行率與發生率比率，來反映一般民眾對於失智症的認知程度。使用貝氏以盛行率/

發生率比率為基礎的統計回歸模型，來做盛行率與發生率比率的校正，用此反映對失智症的認知程度。第二部分，在 AD8 及 MMSE 的效度評估研究中，我們針對 2013 年台南社區整合型篩檢計畫，50 歲以上民眾，評估 AD8 及 MMSE 二種篩檢工具其單獨使用、或平行檢定、或序列檢定，其效度使否有差異? 利用受試者工作特徵曲線（ROC）來探討 AD8 及 MMSE 在失智症及輕度認知功能缺損的預測功能。

(4)

式的效度。第三部分，架構一馬可夫決策分析模型，來模擬有做失智症社區篩檢的族群其 10 年成本及效益，並比較不同篩檢工具，包括 AD8、MMSE、平行檢定、

序列檢定，與無篩檢的族群的成本及效益分析。我們使用一五階段馬可夫模型來模擬失智症進展，利用已發表的國內外文獻所提供參數(轉移機率、各階段所需醫療費用及照顧費用)。效益部分以品質調整人年命 (QALY)來測量，並含 3%每年折扣率，並估出增加成本效用比 (ICUR)。利用 500 次蒙特卡羅模擬，得到從整體社會觀點畫出的分散成本效益面、及接受曲線，此指標可以顯示在不同付費意願閥值下，可以產生符合成本效益的百分比。

結果

失智症盛行率、發生率在健保資料庫(被動性篩檢組)估計值分別為 2.91%及 1.83%。利用盛行率/發生率比率為基礎的統計回歸模型估計校正區域後，得到，

被動性篩檢中，65 歲以上老人盛行率對發生率的比率，男性從在 65-79 歲族群 1.20(1.15-1,24)到 90 歲以上族群 3,27 (3.13-3.41)。盛行率/發生率比率發現，

65 歲以上老人盛行率對發生率的比率在地理區域北區最高，東區最低。被動性篩檢組，校正年紀、性別、區域對失智症影響後，發現盛行率對發生率的比率 65-79 歲族群 1.45(1.43-1.47) 到 80 歲以上族群 1.64 (1.61-1.66)。參與社區失智症篩檢組 (主動性篩檢組)，發現其盛行率對發生率的比率 65-79 歲族群 4.23(2.68-6.69) 到 80 歲以上族群 4.77 (3.02-7.54)。

AD8 單獨使用對失智症敏感度及特異度分別為 64.71% 及 87.89%，MMSE 單獨使用在校正教育程度的情況，對失智症敏感度及特異度分別為 41.18% 及

84.50% 。兩種篩檢工具合併使用，平行檢定的敏感度及特異度分別為 88.89% 及 70.16% ，序列檢定的敏感度及特異度分別為 50% 及 93.02%。AD8 單獨使用對失智症+輕度認知功能障礙，敏感度及特異度分別為 25.74% 及 90.70%。所有篩檢模式，對於輕度認知功能障礙的敏感度，除了切點切在 26 分以外，其他組模式皆顯示敏感度較差，特異度尚可。利用受試者工作特曲線（ROC）來探討 AD8 及 MMSE 在失智症及輕度認知功能缺損的預測功能。結果發現合併 AD8 和 MMSE (平

(5)

行檢定組)的曲線下面積(AUC)為 82.3% (75.1%-89.4%)產生比 AD8

(AUC=73.3% (60.7%-85.9%))單獨使用或 MMSE (AUC=77.4 %( 67.6%-87.3%)單獨使用或系列檢定(AUC= 67.6% (53.4%-81.8%))更好的預測功能。利用貝氏分析方法，可得到平行檢定提高敏感度(97.2%)相較於 MMSE (82.2%)或 AD8 單獨使用 (84.1%)：序列檢定提高特異度度(96.8 %) 相較於 MMSE (86.1%)或 AD8 (77.1%) 單獨使用。關於經濟評估結果如下：只考慮篩檢及醫療支出的直接成本，則成本效用比 (ICUR)在 AD8、MMSE、平行檢定、序列檢定中分別為每一人年為美金 401.4、

457.7、409.8、499.2 元。同時考慮間接成本，在分散成本效益面評估為大約 80

％的模擬值在第四象限（顯性）。由四種篩檢模式中可知，如果政府願意付費閥

值到達美金 20000 元，則可以得到 88-94%成本效益。如果，只考直接成本，評

估為大約 40％的模擬值在第四象限（顯性），當願意付費閥值到達美金 20000 元，

則可以得到 93-99%成本效益。

結論

從預防由失智症造成失能和死亡的臨床方面來看，我們的研究，藉由根據所估計盛行率/發生率比率在社區篩檢模式(主動性偵測) 相較於健保照護體系(被動性偵測) 較高，已證實在一般健保照護體系，失智症的認知程度偏低。合併使用 AD8 和 MMSE 在社區失智症篩檢，可提高工具敏感度。最後，使用 AD8 和 MMSE 在社區失智症篩檢是符合成本效益的：相較於無篩檢組幾乎接近省錢的。最符合成本效益的是合併使用 AD8 和 MMSE 的平行檢定模式。我們的研究結果可應用於健康照護政策評估，及其他有興趣於發展社區失智症篩檢的計畫，進而減少失智症照護的支出。從方法學的角度，我們發展出三種創新方法，包括(1) 估計盛行率/發生率比率為基礎的統計回歸模型。(2) 利用貝氏臨床推理模型來估計不同失智症篩檢工具模式的效度。 (3) 馬可夫決策分析模型，評估社區失智症篩檢中結合 AD8 和 MMSE 的成本效益分析。

(6)

Abstract

Background

As there is an increasing trend in the morbidity of dementia when aging population has been increasing, considering an effective and efficient community based dementia screening programs is of paramount important. Before doing so a series of studies would be required to embrace various aspects including

epidemiological assessment related to awareness of early detection of dementia, the validity of feasible screening tool for community-based screening for dementia, and economic evaluation of community-based screening program with the development of a series of theoretically-sound statistical methods.

Aims

This thesis aimed to (1) quantify the impact of early detection related to

awareness on the average duration of disease based on the measurement of the ratio of prevalence to incidence of dementia with a newly proposed P/I-ratio-based statistical regression model; (2) assess the validity of the accuracy of the early detection of dementia with cognitive test (MMSE) and informant questionnaire (AD8) alone and particularly in the combination applied to a community-based dementia screening with Bayesian clinical reasoning model; and (3) perform cost-effectiveness of community-based dementia screening program with various screening strategies proposed in the second aim with the Markov decision tree model.

Materials and Methods

By using a large-scale, claimed data of the National Health Insurance (NHI) database between 2000 and 2003 in Taiwan, we identified 99,609 patients age over 65 years with dementia (ICD-9-CM code 290, 290.40, and 331.0). The other data source included a total of 183 subjects aged over 65 years participating in an active dementia survey conducted in 2013. Information on age, gender, and geographic areas were also collected. Bayesian P/I-ratio-based statistical regression method was used to estimate the adjusted prevalence/incidence (P/I) ratios of dementia to reflect the awareness of dementia. For the validity of AD8 and MMSE as well as the

combination of the two tools in the parallel and the serial mode, we applied the two screening tools simultaneously in a community-based screening program for dementia to 282 Tainan residents aged over 50 years in 2013. Receiver operating characteristic curves (ROC) were applied to explore the performance of different screen modalities for prediction of MCI or dementia. Bayesian clinical reasoning method was used to estimate the performance of screening modalities in the absence of golden-standard diagnosis.

The Markov decision analysis was conducted to investigate the cost-utility of

(7)

community-based screening of dementia over a 10-year period to compare different screening tools (AD8, MMSE, parallel and serial test of the two) with no screening.

We used a five-state Markov model to simulate the progression of dementia. Disease transition probabilities and costs of different stages were extracted from literatures.

The main outcome measure was cost per quality-adjusted life-year gained with a 3%

annual discount rate. The scattered cost-effectiveness plane (CE plane) and acceptability curve are presented given a 500 Monte Carlo simulated samples.

Results

The prevalence and incidence rate of dementia based on passive survey were estimated as 2.91% and 1.83 %, respectively. The results with the application of Bayesian P/I-ratio-based statistical regression model show the adjusted P/I ratio increased from 1.20 (1.15-1.24) for 70-74 age group to 3.27 (3.13-3.41) for 90+ age group in males. The P/I ratio was the highest in northern area the lowest in eastern area. After controlling for age, gender, and geographic area, the adjusted P/I ratio increased from 1.45 (1.43-1.47) for 65-79 age group to1.64 (1.61-1.66) for 80+ age group through passive detection method (health insurance system). The corresponding figures increased from 4.23 (2.68-6.69) for 65-79 age group to 4.77 (3.02-7.54) for 80+ age group in active community-based survey. The sensitivity and specificity of the sole use of AD8 in dementia screening were 64.71% and 87.89%. The sensitivity and specificity of the sole use of MMSE in dementia with adjustment for education level were 41.18% and 84.50%. The combination of AD8 with MMSE in parallel mode yielded 88.89% of sensitivity and 70.16% of specificity. The combination of AD8 with MMSE in serial mode yielded 50.00% of sensitivity and 93.02% of specificity. The estimates of sensitivity and specificity of using AD8 test alone for MCI plus dementia were 25.74% and 90.70%. All the estimates of sensitivity for all the modes except the MMSE with 26 of cutoff for detecting MCI were poor and the specificity was moderate. By combining prior information derived from the results of previous studies with Bayesian approach, the results show the parallel mode had higher sensitivity (97.2%) than either MMSE (82.2%) or AD8 (84.1%) alone . Besides, the serial test had higher specificity (96.8 %) than AD8 (77.1%) or MMSE (86.1%) alone. ROC curve showed that the combination of MMSE and AD8 in the parallel mode (AUC=82.3% (75.1%-89.4%)) produced a more accurate prediction of dementia than the use of AD8 (AUC=73.3% (60.7%-85.9%)) and MMSE (AUC=77.4

%( 67.6%-87.3%) alone and also the serial mode (AUC= 67.6% (53.4%-81.8%)).

Regarding economic evaluation, if only direct cost on screening and medical expenditure were considered, the ICURs for AD8, MMSE, parallel test, and sequential test were $401.4, $457.7, $409.8, and $499.2 per QALY gained,

respectively. The scatted CE plane suggested that around 80% simulated sit in fourth

(8)

quadrant (dominant) when indirect cost was considered. The probability of being cost-effective was 88-94% given willingness-to-pay (WTP) at $20,000 for the four screening scenario. The corresponding figures for being dominant and cost-effective at WTP at $20,000 when only direct cost taking into account were 40% and 93-99%, respectively.

Conclusions

From the practical aspect of prevention of disability and death from dementia, low awareness of dementia has been ascertained in routine health insurance health care system as the P/I ratio of community-based survey (active detection method) was greater than that of health insurance heath care system (passive detection method).

The combination tests of MMSE and AD8 could improve diagnostic accuracy in the community dementia screening. Community-based screening for dementia with AD8 and MMSE is more cost-effective and almost near cost-saving compared with no screening program. The most economic screening strategy is the parallel mode of combining AD8 with MMSE in comparison with other modes.

From the methodological viewpoint, there are three novelties of the

methodological development here, including a P/I-ratio-based regression model, the application of Bayesian model for multiple detection modalities, and the development Markov cycle decision tree model for economic evaluation of population-based screening program with AD8 in combination with MMSE. The empirical data together with the development of theoretically-sound statistical method provides a new insight into how to conduct an effective and efficient community-based screening for dementia.

Key word: Dementia, screening, sensitivity, specificity, cost-effectiveness

(9)

Several population-based studies indicate that 3% to 11% of persons over age 65 years and 25% to 47% of those over 85 suffer from dementia. In 1997, the number of people with Alzheimer's disease in the United States was estimated to be 2.32 million, more than 90% of whom were age 60 years and older.

Alzheimer's disease is considered the 8th leading cause of death in persons over the age of 65 and is 11th overall in the United States. Median survival estimates of people with dementia ranges between 5.0 and 9.3 years after diagnosis. A recent study found the median survival time, adjusting for date of onset, was around 3.3 years ¹. The previous study has addressed the natural history of dementia from mild, through moderate, and to severe AD, taking a long natural course² . The short median survival time leaves much improvement of early detection through awareness and screening to be desired. Dementia causes a high burden of suffering for patients and their families. For patients, it leads to

cognitive and functional deterioration, behavioral complications, increased use of

(16)

health and social services, complicated clinical management of other comorbid conditions, and increased risk for medical complications such as delirium, falls, motor vehicle crashes, incontinence, fractures, and infections. It also causes a heavy burden for society including family members and health caregivers.

In addition to disease burden imposed to society and sufferings resulting from dementia imposed to individual and caregiver, enormous resources, i.e.

annual societal cost of dementia amounting to approximately $100 billion, are also required for both health care and related costs and lost wages for patients and family caregivers. Given an increasing trend in disease burden, sufferings of patient and caregivers, and economic concerns, it is therefore worthwhile to call for a study to embrace these various aspects from epidemiological assessment, screening and early detection, and economic evaluation of population-based screening program.

1.1.2 Implications and usefulness of epidemiological profiles of

Alzheimer’s disease and dementia

Dementia is a syndrome of decline in memory and at least one other cognitive domain such as language, visual-spatial, or executive function

sufficient to interfere with social or occupational functioning in an alert person ³.

(17)

Multiple diseases can cause the syndrome of dementia. The majority of people with dementia have neurodegenerative disease or cerebrovascular ischemia as the underlying cause. Between 60% and 70% of people with the dementia syndrome have Alzheimer's disease; about 20% to 30% have vascular or mixed vascular and Alzheimer's disease causes. A small fraction of people have other causes such as Lowy body dementia, frontal dementia, Parkinson’s disease, hypothyroidism, and vitamin B 12 deficiency ^{4, 5}.

Based on numerous descriptive epidemiological studies, prevalence of dementia shows considerable variations among different countries for persons aged 65 and older, ranging from 1.8-10.5%^{6,7, 8} . The variation of incidence across countries still exists but seems smaller than that of prevalence. Incidence is a fundamental measurement related to the etiology of the disease, whereas prevalence reflects disease burden affected by a constellation of factors including active detection, awareness, and quality of care for patients diagnosed as

dementia. Accordingly, the ratio of prevalence to incidence, indicating the average duration of dementia, can be used for an indicator for patient’s awareness of dementia if both estimates can be provided.

To gain a better understanding of awareness of dementia would be beneficial to patients and also provides an insight into early diagnosis and

(18)

treatment of dementia. Developing an indicator for such a purpose is helpful for the reflection of the extent of awareness in the underlying population.

To sum up, the combined both figures in terms of ratio of prevalence to

incidence ratio (P/I) provide a good indicator for awareness of detecting dementia and health care quality for treating dementia patients. It is postulated that higher P/I ratio was attributed to the earlier diagnosis through active detection based on

community-based active survey.

Few studies have used this indicator for achieving this goal because it requires a large population-based data to estimate prevalence and the continued follow-up over time to estimate incidence in the same study. Moreover, none of study was focused on the development of statistical regression model for quantifying the impact of early detection on the average duration of disease based on the measurement of P /me ratio.

My current empirical finding with the application of the P/I-ratio-based regression model here found a low awareness of our routine health insurance system in the diagnosis of dementia in contrast to active community-based survey. This forms the first part of my thesis.

1.1.3 Screening of dementia

Routine history and physical examinations do not readily diagnose dementia

(19)

during clinic or physician visits. Numerous studies in western countries indicate low identification of dementia by primary care physicians^9-16.More than 50% of patients with dementia have never been diagnosed by a physician^17-19. This argument together with our low P/I ratio derived from routine health insurance system raises the notion of whether effective screening tests should be provided to identify people with dementia at an early stage, thus allowing the possibility of earlier intervention.

Most screening tests for dementia can be divided into cognitive tests of patients and functional assessments using both patients and other informants. Cognitive tests, the primary screening approach that researchers have investigated, include the Mini-Mental Status Examination (MMSE). MMSE was developed and considered as screening tools more than 30 years ago since 1975. Since that time it has been widely used in early detection of dementia. However, it requires intensive training to

investigators and is too lengthy for use in general practice. The MMSE often renders signs of detection insensitive, particularly in high education individuals.

Among other available cognitive testing strategies, the Clock-Drawing Test (CDT)²⁰, which can take less than 1 minute to administer, has the best potential for meeting these criteria . The small number of methodologically sound studies regarding other clinically relevant cognitive tests limits our ability to evaluate them adequately.

(20)

Some informant-based functional tests, such as the Functional Activities Questionnaire (FAQ),²¹ the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE),²² and the Instrumental Activities of Daily Living (IADL) Questionnaire,²³ have been used for early diagnosis of dementia. These instruments offer everyday relevance, acceptability by subjects, adaptability to various types of patients, administrative ease, longitudinal perspective, and cross-cultural portability.

The primary limitations of these tests are that not all patients have caregivers and that some functions (e.g., cognition) are not tested.²⁴ Most importantly, few

methodologically sound studies regarding the accuracy of these questionnaires have been completed. In recent years, a brief informant interview with eight-item (AD8) to detect dementia was developed to make the measurement simple and also improve the performance of early detection. However, whether AD8 can be used for

population-based screening for dementia is still unclear because the validity of the accuracy of early diagnosis of dementia has been barely addressed albeit its application to clinical patients has been proven^{25, 26}. In addition, since AD8 and MMSE have played each unique role in detection and diagnosis of dementia how to combine AD8 with MMSE as a set of screening tool for dementia is of great interest to health decision-makers who are involved in population-based screening for

dementia. The comparison of different combinations of screening modalities (parallel,

(21)

serial, AD8 alone, and MMSE alone) based on both AD8 and MMSE is therefore worthy of being investigated. This is the second part of my thesis.

1.1.4 Cost-effectiveness analysis of screening for dementia

As mentioned earlier, the economic and social burden caused by dementia is a grave health care problem in Taiwan. Screening and early detection of dementia is therefore proposed. Besides the evaluation of the validity of screening tools in the second part, it is also indispensable to conduct an economic appraisal of different screening strategies, as indicated in the second part, of community-based screening for dementia as screening often demands enormous costs at initial period but accrues the benefit of reducing enormous cost spent in caring for severe dementia as a result of delayed diagnosis in the absence of an effective community-based screening program²⁷. The implementation of an nationwide opportunistic screening program for dementia can be cost-effective depending on disease severity, treatment effect, costs by disease stage, ages of the participants, and the societal willingness to pay (WTP)²⁸ . Above all things, improving access to more effective therapies The third part of my thesis is therefore focused on cost-effectiveness analysis of community-based screening for dementia by various modalities in combination of AD8 with

(22)

1.2 Aims

By collecting empirical data including a large population-registry health insurance claim data and a community-based dementia screening data, and relevant parameters from literature review, the objectives of thesis in pursuant to the rationales mentioned above are composed of the following three parts.

Part I

The first part was to elucidate the epidemiologic profiles of dementia

including age, gender, area-specific prevalence and the corresponding incidence rate of dementia and prevalence/incidence ratio of dementia in Taiwan and also to develop a P/I-ratio-based statistical regression model to compare the effect size of P/I ratio from active (community-based) survey with that from passive health insurance system to indicate the extent of low awareness (underdiagnosis) of dementia after controlling for age, gender, and geographic variation.

Part II

Through a community-based screening for dementia, the second part is to assess the validity of accuracy of early detection of dementia with different instruments including cognitive test (MMSE), informant questionnaire (AD8) and

(23)

combined test (MMSE and AD8). We also applied a Bayesian estimation method to assess the validity of various combinations of screening modalities with the

incorporation of prior information from literature in conjunction with the likelihood data derived from a community-based study.

Part III

Based on the parameters of the sensitivity and specificity obtained from the second part together with the relevant parameters including the transition probabilities, the efficacy of treatment, and costs on medical aspect and social aspect, a series of probability cost effectiveness analysis of community-based dementia screening programs with various comparisons were conducted from a societal perspective in Taiwan.

(24)

Chapter 2：Literature Review

2.1 Epidemiology of dementia

2.1.1 Prevalence and incidence of dementia

Dementia is one of the most distressing and burdensome mental health problems in the old population. Many studies have been conducted on the prevalence of dementia and its subtypes worldwide. The prevalence of dementia increase steadily with age , roughly double every 5 years ²⁹ .The prevalence of dementia in persons aged 65 and older has been reported to be 3.6% to 10.3% in Western countries, 1.8%to 4.6% in china ^{6, 30}, 3.7% to 6.7% in Japan ⁷,and 9.5% to 10.8% in Korea ⁸ since the mid-1980s.

Several previous studies have shown that the prevalence of dementia in Taiwan is between 1.7% and 4.3% in adults aged 65 and older.³¹^{32, 33} These variations among the reported prevalence may depend on methodological differences such as case finding procedures and the characteristics of the population sample. The diagnostic threshold used to justify a diagnosis of dementia may also contribute considerably to the variations. On the other hand, ethnic differences might exist and they may be due to racial genetic factors, shared cultural practices, or common environmental factors.

Incidence is a fundamental measurement related to the etiology of the disease, whereas prevalence better reflects disease burden and it is useful for the planning of

(25)

the provision of health care services. The incidence of dementia may vary from country to country. Based on the incidence of dementia in different studies, the annual incidence of dementia in the elderly people aged 65 years or older ranged from 12.8 to 20.3 per 1000 person year. 12.8 ‰ of the annual incidence of dementia among those aged 65 and older in Taiwan was found based on 2915 community cohort with one year follow-up study. The incidence in Taiwan is slightly lower than 13.1 per 1000 person years of incidence in European and also lower than 14.6 per 1000 person years in USA.

2.1.2 Risk factor of dementia

D Dementia is more prevalent in women than in men. This difference is explained by the greater life expectancy and by a high survival rate of women with dementia compared with same age men with dementia. A meta-analysis of European studies by the EURODEM incidence research group showed that women had a greater risk of developing dementia³⁴(odds ratio=1.2). More recently, results from a large

prospective incidence study in the United Kingdom also showed an increased risk for women ³⁵(odd ratio=1.6)^.

Early-onset dementia is more common in individuals with a family history of dementia. The apolipoprotein E (APOE) gene is a risk for Alzheimer’s disease.

(26)

The risk of dementia is higher in heterozygous ε4 allele, and is higher still in those who are homozygous for ε4. A ε2 allele has been linked with a reduced risk of dementia. The effect of APOE depends on a variety of other factors, including age and ethnicity. APOE allele distribution varies across the world. The ε4 is lowest in southern European, the Middle East, and North Africa³⁶. Stroke , Parkinson disease, and poor self-perceived health were all found to be indicator for dementia³⁵.

Vascular risk factors are also commonly associated with an increase in the risk of Alzheimer’s disease³⁷.

Prospective studies have found that moderate intake of alcohol (especially wine) is associated with having of dementia risk.

A systematic review found consistent evidence that diabetes both in midlife and later life is a risk factor for both Alzheimer’s disease and dementia in general³⁸. Effective control of diabetes may reduce this risk. The relationship between single traumatic head injury and dementia id unclear³⁹. A meta-analysis of incidence studies in Europe by the European Community concerted action on the epidemiology and prevention of dementia (EURODEM) group found there was a lacking of increase in dementia risk with a reported history of head trauma⁴⁰.

(27)

2.2 Types of Dementia

Dementias can be classified in a variety of ways and are often grouped by what they have in common, such as what part of the brain is affected, or whether they worsen over time (progressive dementias).Some dementias, such as those caused by a reaction to medications or an infection, are reversible with treatment.

Types of dementias that worsen over time (progressive dementias) include:

2.2.1 Progressive dementias

2.2.1.1 Alzheimer’s disease

In people age 65 and older, Alzheimer's disease is the most common cause of dementia, around 60% in all types of dementia. People generally may develop symptoms after age 60, but some people may have early-onset forms of the disease, often as the result of a defective gene.

Although in most cases the exact cause of Alzheimer's disease isn't known, plaques and tangles are often found in the brains of people with Alzheimer's. Plaques are clumps of a protein called beta-amyloid, and tangles are fibrous tangles made up of tau protein. Certain genetic factors also may make it more likely that people will develop Alzheimer's.

Alzheimer's disease usually progresses slowly over seven to 10 years. Your cognitive

(28)

properly, including parts of your brain that control memory, language, judgment and spatial abilities.

2.2.1.2 Lowy body dementia

Lowy body dementia affects approximately 10 to 22 percent of people with

dementia, making it one of the most common types of dementia. Lowy body dementia becomes more common with age. Lowy bodies are abnormal clumps of protein that have been found in the brains of people with Lowy body dementia, Alzheimer's disease and Parkinson's disease. Lowy body dementia symptoms are similar to symptoms of Alzheimer's disease. Its unique features include fluctuations between confusion and clear thinking (lucidity), visual hallucinations, and tremor and rigidity (Parkinsonism). People with Lowy body dementia often have a condition called rapid eye movement (REM) sleep behavior disorder that involves acting out dreams.

2.2.1.3 Vascular dementia

Vascular dementia, the second most common type of dementia, around 20% in all types of dementia, occurs as a result of brain damage due to reduced or blocked blood flow in blood vessels leading to your brain. Blood vessel problems may be caused by stroke, infection of a heart valve (endocarditis) or other blood vessel

(vascular) conditions. Symptoms usually start suddenly and often occur in people with high blood pressure or people who have had strokes or heart attacks in the past.

(29)

Several different types of vascular dementia exist, and the types have different causes and symptoms. Alzheimer's disease and other dementias also may be present at the same time as this dementia.

2.2.1.4 Frontotemporal dementia

This less common cause of dementia tends to occur at a younger age than does Alzheimer's disease, generally between the ages of 40 and 65.

This is a group of diseases characterized by the breakdown (degeneration) of nerve cells in the frontal and temporal lobes of the brain, the areas generally associated with personality, behavior and language. Signs and symptoms of fronto-temporal dementia can include inappropriate behaviors, language problems, difficulty with thinking and concentration, and movement problems. As with other dementias, the cause isn't known, although in some cases this dementia is related to certain genetic mutations.

2.2.2 Reversed type of dementia

Some causes of dementia or dementia-like symptoms can be reversed. Your doctor may identify and treat these causes:

2.2.2.1 Infections and immune disorders

Dementia can result from fever or other side effects of your body's attempt to fight off an infection. People may develop dementia or thinking difficulties if they

(30)

disease, or conditions that cause a completely compromised immune system, such as leukemia. Conditions such as multiple sclerosis that arise from the body's immune system attacking nerve cells also can cause dementia.

2.2.2.2 Metabolic problems and endocrine abnormalities

People with thyroid problems, too little sugar in the bloodstream (hypoglycemia), too low or too high amounts of sodium or calcium, or an impaired ability to absorb vitamin B-12 may develop dementia or other personality changes.

2.2.2.3 Nutritional deficiencies

Dementia symptoms can occur as a result of not drinking enough liquids (dehydration); not having enough thiamine (vitamin B-1), a condition common in people with chronic alcoholism; and not having enough vitamins B-6 and B-12 in your diet.

2.2.2.4 Reactions to medications

Dementia may occur as a reaction to a single medication or because of an interaction of several medications.

2.2.2.5 Subdural hematomas

Subdural hematomas are caused by bleeding between the surface of the brain and the covering over the brain. They can cause symptoms similar to dementia.

2.2.2.6 Poisoning

(31)

Dementia symptoms can occur as a result of exposure to heavy metals, such as lead, and other poisons, such as pesticides. Dementia symptoms also may occur in some people who have abused alcohol or recreational drugs. Symptoms may disappear after treatment, but in some cases symptoms may still be present after treatment.

2.2.2.7 Brain tumors

Dementia rarely can result from damage caused by a brain tumor.

2.2.2.8 Anoxia

This condition, also called hypoxia, occurs when organ tissues aren't getting enough oxygen. Anoxia may occur due to severe asthma, heart attack, carbon monoxide poisoning or other causes.

2.2.2.9 Heart and lung problems

Brain can't survive without oxygen. Dementia symptoms may occur in people with chronic lung problems or a heart condition that deprives the brain of the oxygen it needs.

2.2.2.10 Normal-pressure hydrocephalus

Sometimes people have normal-pressure hydrocephalus, a condition caused by enlarged ventricles in the brain. This condition can cause walking problems, urinary difficulty and memory loss. Shunt surgery, which delivers cerebrospinal fluid from

(32)

the head to the abdomen or heart, may help these symptoms.

2.3 Common cause of dementia

Alzheimer’s disease accounts for most cases of dementia in North America (50–

85%)^{4, 5}, with an additional 10–20% attributed to vascular (“multi-infarct”) dementia.

The relative importance of vascular dementias is higher in populations where

hypertension and stroke are more common (Asians, African Americans, persons over 85) ^41-43.Other important causes of dementia include alcoholism, Parkinson’s disease, metabolic disorders (vitamin B12 deficiency, hypothyroidism), central nervous

system infections (e.g., HIV, neurosyphilis), intracranial lesions, and other illnesses

4, 44.

2.4 Diagnosis of dementia and MCI

According to the National Institute on Aging and the Alzheimer’s Association(NIAA), criteria for all-cause dementia and for AD dementia in 2011⁴⁵,dementia is diagnosed when there are cognitive or behavioral

(neuropsychiatric) symptoms that : 1. Interfere with the ability to function at work or at usual activities; and 2. Represent a decline from previous levels of functioning and performing; and 3. Are not explained by delirium or major psychiatric disorder; 4.

Cognitive impairment is detected and diagnosed through a combination of (1)

(33)

history-taking from the patient and a knowledgeable informant and (2)an objective cognitive assessment, either a “bedside” mental status examination or

neuropsychological testing. 5. The cognitive or behavioral impairment involves a minimum of two of the following domains: (a) impaired ability to acquire and remember new information. (b) Impaired reasoning and handling of complex tasks, poor judgment. (c) Impaired visuospatial abilities––symptoms include: inability to recognize faces or common objects or to find objects in direct view despite good acuity, inability to operate simple implements, or orient clothing to the body.(d) Impaired language functions (speaking, reading, writing).(e) Changes in personality, behavior, or comportment–symptoms.

MCI was often diagnosed, based on the criteria recommended by the NIA-AA⁴⁶,

as below : (1) Cognitive concern reflecting a change in cognition reported by patient or informant or clinician (2) Objective evidence of Impairment in one or more cognitive domains, typically including memory (3)Preservation of independence in functional abilities (4) Not demented

(34)

2.5 Early detection of dementia

Most persons with dementia remain undiagnosed by their primary care physicians.

Prospective longitudinal studies demonstrate serious deficiencies in the healthcare system’s ability to recognize dementia. Most dementia remains unrecognized in the primary care setting. Persons with mild dementia are more likely to go unrecognized by physicians and family (over 90%) than persons with moderate to severe dementia (over 70%); however, those with early disease are best treated with available

medications^47-49. Family members often under-recognize cognitive decline in elders (over 50%). Many elderly live alone and have limited contact with distant relatives.

Under-recognition of dementia is a serious, unsolved healthcare problem despite multiple expert panels that have discussed recommendations on dementia screening.

2.5.1 Instruments of neuropsychological measurements in dementia

screening

2.5.1.1 Cognitive test: Mini-Mental Status Examination (MMSE)

The Mini-Mental Status Examination (MMSE) is the best-studied instrument for screening for cognitive impairment. The Agency for Health Care Policy and Research (AHCPR) supported a systematic review and meta-analysis of studies (published primarily before 1994) that evaluated the MMSE for screening.⁵⁰

(35)

The AHCPR panel used mean effect size as the measure of effectiveness, as described by Hasselblad and Hedges.⁵¹The mean effect size for discrimination between patients with and without dementia was 1.78. This effect size corresponds to an equivalent sensitivity and specificity of 84% and a sensitivity of approximately 75%, for a fixed specificity of 90%.Studies from 1994 to 2001 have had 2 usual orientations when evaluating the MMSE: primary investigations into its validity when adjusting for either cultural or educational factors (or both) and secondary investigations that compare the performance of newer screening tools to that of the MMSE. Table 2 compares the finding of 5 MMSE studies.⁵²Excluding the Wilder et al. study

(evaluating specificity levels for 90% sensitivity), the MMSE sensitivity (71% to 92%) and specificity (77 % to 96%) fell into a moderate range and the percentage of falsely classified individuals (false negatives and false positives as a percentage of the total number of tested individuals) ranged from 4% to 18%.The primary factors

determining the rate of false diagnoses are likely to be related to cut-off values and the overall percentage of individuals with dementia in each study. Folstein et al., in 1975, documented that the MMSE is a reliable instrument.⁵³ Two decades later, McDowell et al. provided additional reliability data that confirmed the earlier findings.⁵²

The accuracy of the MMSE depends upon a person’s age and educational level:

using an arbitrary cut-point may potentially lead to more false-positives among older

(36)

people with lower educational levels, and more false-negatives among younger people with higher educational levels.

2.5.1.2 The informant-based tests：AD8 Questionnaires

The AD8 is a brief informant-based questionnaire developed by Washington University in St Louis. It is a screening tool with 8 questions to reliably differentiate no demented from demented individuals even at the very mild stage.^{26, 54} The AD8 is sensitive to the earliest signs of cognitive change as reported by an informant. The AD8 is highly correlated with gold standard, the CDR, as well as performance on brief objective measures such as the MMSE.

The AD8 test the subjective cognitive abilities in memory, temporal orientation, judgment, and function. The score on AD8 was range from 0 to 8. It is sensitive to the earliest signs of cognitive change as reported by an informant²⁶. The cut points of AD8 for distinguishing dementia cases was greater than 2. The sensitivity of AD8 for dementia ranges from 68% to 95.9%, and the specificity ranges from 78.1% to 90%^{25, 26, 55}

2.5.2 Accuracy and reliability of early detection instruments in dementia

The goals of any screening test are to separate people with a high probability of having the disease from those with a low probability and to presumptively identify

(37)

unrecognized disease. Diagnostic confirmation is generally required. An effective screening test should be inexpensive, and its characteristics should include reliability, sensitivity, specificity, social acceptability, safety, and brevity.

Researchers and practitioners in this clinical area have traditionally divided screening tests into cognitive tests and functional assessment. Most screening tests have been evaluated in studies with small sample sizes, and the populations of patients on whom screening instruments have been tested have varied greatly, making it difficult to determine the overall performance of screening tests for dementia. The best evidence is available for a cognitive test—the Mini-Mental Status Examination

(MMSE)—from studies in primary care settings that used standardized diagnostic instruments (e.g., the DSM-IV) as a “gold standard.” Depending upon the cut point used for an abnormal test, the sensitivity of MMSE for dementia ranges from 71% to 92%, and the specificity ranges from 56% to 96%^{52, 56}.The predictive value of a positive test, in a population with 10% prevalence of dementia, may range from 15%

to 72%.⁵⁷ A drawback of MMSE is that its accuracy depends upon age, education, and ethnicity of the individual; it is most accurate for whites with at least a high school education. Other cognitive screening tests, such as the Short Portable Mental Status Questionnaire, Clock Drawing Test, Modified MMSE, Mini-Cog, Hopkins Verbal Learning Test, and the 7-minute screen are promising, but have not

(38)

been adequately evaluated in primary care settings.

Some informant-based functional tests, such as the Functional Activities Questionnaire (FAQ), the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE), and the Instrumental Activities of Daily Living (IADL) Questionnaire, have also been tested. ⁵⁸ The sensitivity and specificity of FAQ is reported to be 90%. The functional test instruments offer the advantages of “everyday relevance,” acceptability by subjects, adaptability to various types of patients,

administrative ease, longitudinal perspective, and cross-cultural portability. The primary limitations of these tests are that not all patients have caregivers and that some functions (e.g., cognition) are not tested. Most important, few methodologically sound studies regarding the accuracy of these questionnaires in primary care settings have been completed. Testing for genetic mutations may eventually prove useful in screening individuals at risk for Alzheimer’s disease. There are, however, limited population-based data regarding the absolute risk of dementia among individuals having a positive genetic test. Thus the potential benefits and harms of testing for an individual patient are uncertain. Finally, the ethical issues in genetic testing for dementia are unresolved.

(39)

2.5.3 The benefit of early detection in dementia

Persons who screen positive are referred to their local physician for follow-up along with the results of the testing. Previous studies demonstrate that over 60% of individuals with positive screens seek follow-up care.^{59, 60} Studies show that 10-20%

of individuals will score positive during a routine screening. The percentage of expected positive screening depends upon the age of the screening population, the location, and multiple other variables.

Early identification of at-risk patients provides multiple benefits to the individual, the family, and society. For the affected individual, identification of early stage

dementia allows early aggressive use of available treatments. Early stage patients can be offered support groups to diminish the psychological impact of the disorder.

Moreover, the total medical care for this individual can be adjusted to meet the needs of a cognitively impaired patient. Issues such as patient education, self-medication, compliance, and hospital care can be adjusted to meet the needs of a mildly demented person who is at risk for common complications such as delirium and depression. The early identification of dementia supports individual patient rights and

self-determination. Most mildly impaired patients are capable of charting the future course of their care and making substantial decisions on issues like end-of-life care, resuscitation, disposition of wealth, etc. Informing at-risk patients about abnormal

(40)

screening does not produce hardship or harm to the patient or family caregiver.⁶¹ About one-third of elders live by themselves and these individuals are at risk for accidents, injuries, exploitation, and other adverse outcomes. Early identification allows safeguards and home assistance to assure continued maximization of home placement. Family caregivers derive multiple benefits from early identification. Early identification may reduce the burden of later life decision-making on issues like resuscitation, disposition of wealth, etc. as families can solicit the opinion of the patient while still competent.

Screening and early identification may benefit society by protecting individuals and reducing costs of healthcare. Unrecognized dementia can increase the likelihood of avoidable complications such as delirium, adverse drug reactions, noncompliance, etc.

These complications can reduce the autonomy of the patient. Enhancing compliance and protecting demented patients has obvious financial benefits to the healthcare system. Adverse outcomes from screening programs are rarely reported by available literature or experienced by community providers. Published studies on screening for community-based elders demonstrate effectiveness and acceptance ⁶². Screening programs detect possible impairment in 10-20% of screened individuals^60. Patient and family satisfaction has been reported as high based on published studies and

experience by AFA membership.

(41)

The benefits to citizens are clear. Patients can receive available therapy when identified and diagnosed. The healthcare management can be adjusted to incorporate treatment strategies that accommodate a person with cognitive impairment.

Home-based support systems can be adjusted to maximize home placement for this person. Safeguards can be taken to prevent avoidable complications such as delirium during hospitalization. In persons with dementia, advanced directives can be

discussed that incorporate the wishes of the individuals and reduces the burden of surrogate decision making for the family. Available treatments for Alzheimer’s disease and other forms of dementia are most helpful in the early stages of illness. Early identification allows optimal therapy with available and emerging medications.

For persons with a normal screen, this intervention provides a valuable opportunity to promote cognitive wellness and successful aging. A simple, direct, cognitive wellness message can be presented to these individuals that may reduce their likelihood for developing dementia at a later age. The emotional boost from a normal dementia screen can be used as an opportunity to discuss basic, preventive interventions such as compliance with anti-hypertensive, responsible drinking, intellectual stimulation and other recommendations that may further protect a patient’s cognitive function ⁶³. Presently, there is no national policy on dementia screening. Despite the acceptable accuracy of screens as well as the availability of medications for early stage disease,

(42)

there is no public health policy on assessing for dementia. The present Medicare screening and prevention program does not include cognitive function. Local

organizations are left to create their own programs without assistance or guidance. A national system of dementia screening will require several years for development and implementation. A flexible array of services and instruments will be required. A policy executed today would only be fully available in the field several years from now.

Scientists and researchers are trained to accept treatment strategies that incorporate evidence-based practices. Although, this conceptual model is the gold standard, this strategy has significant limitations that are rarely emphasized by the scientific community. Mass scale public health interventions are tested over a multi-decade period. Researchers are generally preoccupied with conclusive scientific data and the promotion of research. In contrast, public systems must use a pragmatic approach, i.e.,

“best possible solution”. To date, national policy has been dominated by expert opinions provided by clinical and basic science researchers. The failure of the “magic bullet” approach warrants an alternative strategy that incorporates interventions to limit the impact of this public health problem. No professional organization contends that undiagnosed, unprepared, uninformed patients with dementia are preferable to individuals with accurate diagnoses and appropriate, early interventions.

2.6 Cost-effectiveness analysis of dementia

(43)

The essential part for the cost-effectiveness analysis is to understand the disease course. As dementia involves different disease states and death, the multi-state Markov model can be used to depict the disease course. In this section, we also

reviewed literatures for the disease burden of dementia and the evidence regarding the efficacy of a memory screening test. Finally, the empirical findings of

cost-effectiveness analysis of dementia screening from two studies were described.

2.6.1 Multi-state Markov models

Multi-state Markov models are often used to depict the evolution of disease progress over time.

Let {X t, t > = 0} (data realization) be a discrete random variable with state space = {1, 2….. m.} representing a stochastic process of disease. In clinical practice, data realization involves a series of successive observations where disease would progress to severe stages and/or an ‘absorbing’ state, often death, and may regress to mild stages. The possible states of disease and/or death constitute the state space,.

For example, one could develop a five-state Markov model to define the dementia-related status in a community with Xt of state space = {0, 1, 2, 3, 4}, where

(44)

 0=normal,

 1=mild dementia,

 2=moderate dementia,

 3=severe dementia, and

 4=death.

Suppose one subject has the disease history of

Normal, Normal, Mild, Mild, Mild, Severe, Death

at seven distinct time (t=0, 1, 6). The joint probability of disease history for this subject can be expressed as

0, 0, 1, 1, 1, 3, 4 (1)

Due to sequential order by time of the observation, equation (1) can be written as

4| 3, 1, 1, 1, 0, 0

3| 1, 1, 1, 0, 0

1| 1, 1, 0, 0

1| 1, 0, 0 1| 0, 0

0| 0 0

(2)

(45)

The Markov assumption suggests that given the knowledge of a present state, say Xt, the outcome in the future (Xt+1, Xt+2, …) is no longer dependent on the past (X0, X1, …, Xt-1) ⁶⁴. Equation (2) can then be simplified as

4| 3 3| 1 1| 1

1| 1 1| 0 0| 0

0

(3)

For each Xt we denote the absolute probability by

And for each pair of random variables, X and X,  <, the conditional probability is

| _,

with the conditions that∑ _∈ 1, and ∑ _∈ _, 1.

Thus, equation (3) can be written as

(4)

(46)

Namely, a Markov model can be completely determined by the initial absolute probability distribution ( ) and the transition probabilities ( _, ).

A Markov model is called homogeneous with respect to time if the transition probabilities ( _, ) are independent of time, and is non-homogeneous with respect to time if the transition probabilities ( _, ) are function of time,. For the ease to the presentation, the transition probabilities can be arranged in the form of a square matrix. For example, the probabilities of all possible transitions for the abovementioned 5-state dementia example can be described in

00 01 02 03 04

10 11 12 13 14

20 21 22 23 24

30 31 32 33 34

0 1 2 3 4 0 (Normal)

1 (Mild) 2 (Moderate) 3 (Severe)

4 (Death) 0 0 0 0 1

P P P P P

 

 

  

 

 

P (5)

The final row in P of (0 0 0 0 1) is for the absorbing state, death. Transition probabilities can be modified according to the clinical plausibility. For example, if dementia is impossible to reverse to normal (dementia-free), then =0 for k=1, 2, 3, and 4. If severe state of dementia is believed not to recover, then =0 for k=0, 1, and 2.

The Markov model can also be classified as discrete-time and

continuous-time Markov model if one concerned the one-step transition probabilities

given data realizations are in regular time intervals or instantaneous transition rate

(47)

where data realizations could be given in regular or irregular intervals, respectively.

For the former, parameters of interests are and _, , whereas the latter is described by and instantaneous transition rates between states, and

∆

ν lim_∆→ ^, ^{∆ ,} ^∆

∆ (6)

Whereν ∑ ν .

For the continuous time Markov model, the transition probability matrix, P, is a function of instantaneous transition rates, which can also be expressed in matrix form.

Take again the abovementioned 5-state dementia as an example, in which we do not allow regression from any state of dementia to normal, and regression from severe dementia to any earlier states, and treat death as an absorbing state. Its transition intensity matrix Q can be expressed as

(7) The derivation of the forward Kolmogorov differential equation gives the

solution of transition probability matrix in a time interval, t, P(t) as a function of Q subject to P(0)=I, where I denotes a unit matrix ⁶⁴⁶⁵as

失智症早期篩檢偵測工具效度及成本效益之統計評估

國立台灣大學公共衛生學院流行病學與預防醫學研究所

博士論文

失智症早期篩檢偵測工具效度及成本效益之統計評估

Statistical Evaluation of Validity and

Cost-effectiveness of Two Screening Tools for Early Detection of Dementia

盧韻如 Yun-Ru Lu

指導教授 : 劉宏輝博士、陳秀熙博士 Advisor: Horng-Huei Liou, Ph.D.

Hsiu-Hsi Chen, Ph.D.

中華民國104年1月

Jan, 2015

Contents

Table contents

Figure contents

Chapter 1：Introduction

Chapter 2：Literature Review

失智症早期篩檢偵測工具效度及成本效益之統計評估

國立台灣大學公共衛生學院 流行病學與預防醫學研究所

博士論文

失智症早期篩檢偵測工具效度及成本效益之統計評估

Statistical Evaluation of Validity and

Cost-effectiveness of Two Screening Tools for Early Detection of Dementia

盧韻如 Yun-Ru Lu

指導教授 : 劉宏輝博士、陳秀熙博士 Advisor: Horng-Huei Liou, Ph.D.

Hsiu-Hsi Chen, Ph.D.

中華民國104年1月

Jan, 2015

Contents

Table contents

Figure contents

Chapter 1：Introduction

Chapter 2：Literature Review

國立台灣大學公共衛生學院流行病學與預防醫學研究所