台灣地區呼吸器依賴病人之發生率、生活品質、及成本效果研究

國立臺灣大學公共衛生學院職業醫學與工業衛生研究所 博士論文

Institute of Occupational Medicine and Industrial Hygiene College of Public Health

National Taiwan University doctoral dissertation

台灣地區呼吸器依賴病人之發生率、生活品質、及成本 效果研究

Incidence rates, quality of life and cost-effectiveness of patients under prolonged mechanical ventilation in Taiwan

洪美娟

Hung, Mei-Chuan

指導教授：王榮德 博士

Advisor: Wang, Jung-Der, M.D., Sc.D.

中華民國101年3月

March 2012

i

誌謝

完成博士學位一直是我碩班畢業後10 年的心願，首先要把這榮耀獻給上帝，辛勤

指導我的老師王榮德教授，及我親愛的家人。亦感謝在校期間師長在各領域上的 啟發與教導，學長姐的不吝指教，同學的相互打氣，研究團隊們各項支援，包括 國衛院麗光老師與信銘、嘉義基督教醫院陳院長、林主任、范醫師、嚴元鴻醫師 與欣諺、台大醫院蔡甫昌醫師、余忠仁醫師與郭律成醫師、陽明醫學大學施富金 老師，台大神經部鄭建興醫師、中央研究院黃景祥老師、台大心理系姚開屏老師、

台大職能治療系謝清麟老師，及郁瑩、正芬、淑如、建瑋與玉嬋。另外，感謝受 訪醫院、受試者及其家屬接受訪談，還有國衛院及國科會經費的贊助。

一路走來，全職生的我備感壓力，還好有著一份堅定的信仰、老師、家人與朋友 的陪伴及鼓勵，每當我快要放棄的時候，照亮我前面的路。「流淚撒種的必歡呼

收割」(詩 126:5)、 「當將你的事交託耶和華，並倚靠祂，祂就必成全。」(詩 37:5)

iii

摘要 研究目的：

本論文研究之主要目的為：(1)探討呼吸器依賴病人每年發生率 (incidence rates) 、

共病 (co-morbidity) 聚集情形與預期壽命；(2)測量呼吸器依賴病人之生活品質: 病 人、家屬及護理人員的比較；(3)估算呼吸器依賴病人之調整生活品質後的預期壽 命 (quality-adjusted life expectancy, QALE)及調整生活品質後損失的預期壽命；(4) 估算呼吸器依賴病人之共病的遞增成本效用比值 (incremental cost-effectiveness ratio, ICER)。

研究背景：

公衛的核心價值是全體人類的健康，為達成此目標，國家必須有相對的配套措施。

台灣很幸運的1995 年 3 月全民健保實施，人民就醫機會逐漸均等不怕看病，並針

對重大傷病族群予以免部份負擔，包括洗腎、呼吸器使用超過 21 天病人...等，使

得人民不必因病而貧，目前已照顧到超過 99%全體國人的健康。但也因此造成國

家財政的重大負擔。依據中央健保局在2010 年所公佈的重大傷病醫療費用前六大

疾病，長期呼吸器依賴(呼吸器使用超過 21 天)的病人排名第二位，總支出每年約

為新台幣 170 億。為同時顧及照護品質、效率及醫療資源的分配正義，以讓健保

永續經營；我們必須開始思考不同的照護策略，包括先了解此類病人之發生率、

生活品質、存活及成本效果。

長期呼吸器依賴(Prolonged mechanical ventilation or ventilator dependent) 發生背 景：

由於全國加護病床有限，許多長期依賴呼吸器患者佔用急性病房或甚至加護病

房，導致急症患者面臨一床難求。健保局多方考量下，於1998 年提出「改善醫院

急診重症醫療計畫」，訂定「急性呼吸治療病床」及「呼吸照護病床」之設置標準，

2000 年提出「全民健康保險呼吸器依賴患者整合性照護前瞻性支付方式」試辦計 畫迄今這幾年來，此類病人雖然其增加的幅度藉由”整合性傳送服務”(Integrated delivery services)的介入自 2005 年後有延緩的趨勢，有關生命倫理公正性及成本效 果的問題持續存在。這需要由中央健康保險局根據各種病人之存活與生活品質，

與所有涉及此事的參與者(病人及其家屬、醫療照護人員與機構等)之共識，制定一 套合乎醫學倫理之臨床指引來協助解決這個問題。目前國際間尚缺乏探討呼吸器

依 賴 病 人 在 不 同 共 病 下 ， 結 合 存 活 、 生 活 品 質 及 醫 療 花 費 的 成 本 效 果 (cost-effectiveness)實證資料。

研究方法：

本論文研究以「以健保資料和對應之死亡檔登記資料所建立起的 1998-2007 年

50,481 筆呼吸器依賴案例全國代表性樣本長期追蹤資料庫」進行分析此類病人在 不同疾病組合下之發生率及存活壽命；收集台灣北部醫學中心、中南部區域醫院，

及其他呼吸治療醫院，使用EQ-5D 歐盟生活品質問卷測量加護病房、呼吸照護中

心、呼吸照護病房及居家照護之共 142 名呼吸器依賴病人的生活品質。並以嘉義

基督教醫院1998-2007 年間共有 633 名下轉至呼吸照護病房之呼吸器依賴病人，研

究調整生活品質後的預期壽命及調整生活品質後損失的預期壽命。最後，結合 50,481 名病患之生病期間所有的醫療費用，計算其成本效果或生活品質調整後之 人年(quality-adjusted life year, 簡稱 QALY 或健康人年)所負擔的費用。

結果與討論：

過去這10 年來，呼吸器依賴病人成快速成長，每年約達兩萬人以上。其 17-85 歲

累積發生率為0.103 - 0.145，也就是說 17-85 歲的人口族群中，如果沒有死於其他

疾病，約有 1/7 至 1/10 的機會將成為呼吸器依賴病人，應及早採取對策。在預期

壽命方面，退化性神經性疾病、中風或一般外傷之呼吸器依賴病人似乎比癌症或

慢性腎衰竭病人有較長的存活壽命；小於 85 歲的慢性阻塞性肺部疾病(Chronic

Obstructive Pulmonary Disease)相較其他年齡層共病，也有較長的存活(第一章)。

在病人生活品質評估方面，具中等認知功能 (MMSE≧15, Mini-mental status examination)的病人，由主要照顧家屬評值個案之生活品質，似乎較接近個案自評 的結果；喪失認知及溝通能力的病人，由主要照護家屬或護理人員評值病人生活 品質，則無明顯的差異 (第二章)。

平均年齡 76 歲的呼吸器依賴病人之預期壽命為 1.95 年，所損失的預期壽命的為

8.48 年。在調整生活品質後的預期壽命，具中等認知功能，為 0.58 健康人年；約 60%喪失認知及溝通能力，為 0.28-0.29 健康人年。因呼吸器依賴而損失之調整生

活品質後的預期壽命為9.87-10.17 健康人年 ，顯示他們為極需醫療照護之健康弱

勢族群(第三章)。

成本效果分析結果，認知功能差的病患，除肝硬化、一般外傷及小於64 歲的菌血

v

3 倍國內生產毛額(gross domestic product, GDP)；認知功能較佳，慢性腎衰竭、退

化性神經性疾病或超過85 歲的多重共病者亦是如此，似乎已超過世界衛生組織成

本效果参考建議的每個健康人年約1-3 倍 GDP。但在肝硬化、一般外傷及小於 65

歲的共病患者，則少於 1.4 倍 GDP。上述實證資料將可提供病人、家屬、醫療人

員及醫療決策者在面臨病人是否接受長期呼吸插管決策，及資源分配之重要實證 數據，並開啟彼此間的對話平台，最好及早預防走到此地步(第四章)。

結論：

整體而言，長期呼吸器依賴患者中某些特性病人之存活率、生活品質與成本效果 不理想，值得作進一步更深入之分析探討，且在本國社會形成臨床決策之共識，

以促進全民健保資源使用之公平性與效率。所有經呼吸治療救治超過21 天仍需呼

吸器維生者，除繼續進行呼吸器拔除訓練外 (weaning)，針對不可逆之無意識患 者，請醫療團隊似可建議家屬，考慮轉為安寧療護以縮短病人痛苦。未來更應正 視生命教育 (life education)，推廣在身體尚健康時，或疾病早期即簽署免予急救意 願書DNR (Do not resuscitate) ，以對加護及呼吸治療資源作最適當之運用。

關鍵詞: 呼吸照護、預期壽命、發生率、共病、生活品質、調整生活品後的預期 壽命、成本效果

Abstract

Objective: The aims of our study are: (1) to determine the incidence rate (IR), median

survival, life expectancies for different types or co-morbidities of patients undergoing prolonged mechanical ventilation (PMV) ; (2) to explore how much difference on the quality of life (QOL) assessments between patients under PMV and their proxies (family caregivers and nurses); (3) to estimate the quality-adjusted life expectancy (QALE) and the expected lifetime utility loss of different types of patients with PMV (4) and to estimate the incremental cost-effectiveness ratios (ICER) for different types of patients undergoing PMV.

Background: As we know, the core value of public health is health for all and the first

step would be to achieve the goal of health care for all. In Taiwan, we are very fortunate to have developed the National Health Insurance (NHI) that implements a system of universal coverage (over 99% now) for all people of Taiwan. The NHI comprehensively covers various health care services, especially those of catastrophic illnesses, including maintenance hemodialysis for end stage renal disease and chronic ventilator use consecutively for more than 21 days, etc., which results in a tremendous financial burden and threatens the sustainability of the NHI. According to the reimbursement data of NHI in 2010, the prolonged mechanical ventilation (PMV) consumes the second highest in the average health expenditure with an annual total health expenditure of 17 billion NT (New Taiwan) dollars. In order to keep the sustainability of the NHI under good quality, efficiency and equitable distribution of resources at the same time, we began to think of different strategies. One of them is to explore the incidence rate, quality of life, survival, and cost effectiveness of the patients undergone prolonged mechanical ventilation. Although the increasing trend seems slowed down after the pilot

vii

distributive justice on bioethics and cost-effectiveness of such cares still persist and may not be easily resolved without relevant information on the life expectancy and quality of life of different types of patients. To our knowledge, there has been few study that estimates the long term survival, quality of life, lifetime cost, and cost effectiveness together for PMV patients with multiple co-morbidities, which are the major goals of this dissertation.

Material and Methods: With the kind assistance from the National Health Research

Institutes, we have just established a representative national longitudinal data of 50,481 cases who were ventilator dependent between 1998 and 2007. The data were linked with the reimbursement data of the NHI (National Health Insurance) and National Mortality Registration database of Taiwan. We used these databases of PMV to estimate IR, cumulative IR, and survival function, and quantitatively determine the QOL of patients using intensive care unit (ICU), RCC and RCW with the Taiwan version of EQ-5D questionnaire. Moreover, we applied latent class analysis (LCA) to re-classify these patients into several categories and determine their individual survival functions and extrapolated to 300 months in order to provide policy suggestions for proactive prevention. QOL measurements and lifetime survival functions were integrated together to calculate the quality-adjusted life expectancies for different types of PMV patients.

Finally, we combined the data of lifetime health expenditures reimbursed by the NHI, and estimate the cost-effectiveness or incremental costs per quality-adjusted life year (QALY) according to different categories and/or clusters of diagnoses for these patients.

Results: The analysis of 50,481 PMV patients revealed that incidence rates increased as

patients grew older and that cumulative incidence rates (17-85 years old) increased from 0.103 in 1998 to 0.183 in 2004 before stabilising thereafter. The life expectancies of PMV patients suffering from degenerative neurological diseases, stroke, or injuries

tended to be longer than those with chronic renal failure or cancer. Patients with chronic obstructive pulmonary disease survived longer than did those co-morbid with other underlying diseases, especially septicaemia/shock (Chapter 1).

QOL assessments from family caregivers agreed more closely with patients than did those from nurses using EQ-5D evaluations for patients with clear cognition, but either proxy was acceptable for rating PMV patients with poor cognition (Chapter 2).

The average age of subjects was 76 years old. The life expectancy and loss of life expectancy were 1.95 years and 8.48 years, respectively. The QALE of 55 patients with partial cognitive ability and the ability to respond was 0.58 QALY, whereas the QALE of 87 patients with poor consciousness were 0.28 and 0.29 QALY for the EQ-5D measured by family caregivers and nurses, respectively. The loss of QALE for PMV patients was 9.87-10.17 QALY, corresponding to a health gap of 94-97% (Chapter 3).

The ICER for PMV varies greatly depending on different underlying causes and co-morbidities. Among these patients, maintenance treatments for PMV patients with poor cognition or patients more than 85 years of age might be the least cost-effective (Chapter 4).

Conclusion: Theses results of poor prognosis would provide stakeholders evidence for

communication to facilitate clinical decisions. Moreover, the results can also serve as a starting point for a public dialogue on resource allocation of the NHI on critical care, aging and palliative care.

Key Words: respiratory care, life expectancy, incidence rate, co-nobilities, quality of

life, quality-adjusted life expectancy, cost-effectiveness

ix

目錄

口試委員會審定書 ...i

誌謝 ...ii

中文摘要 ...iii

英文摘要 ...vi

目錄 ...ix

圖目錄 ...xi

表目錄 ...xii

論文內文 Overview ...1

Chapter 1 Life expectancies and incidence rates of patients under prolonged mechanical ventilation: A population-based study during 1998-2007 in Taiwan 1.1 Introduction ...5

1.2 Material and methods ...6

1.3 Results ...11

1.4 Discussion ...13

1.5 Conclusions ...16

1.6 References ...17

Chapter 2 Measurement of quality of life using EQ-5D in patients on prolonged mechanical ventilation: comparison of patients, family caregivers and nurses 2.1 Introduction …...27

2.2 Material and methods …...28

2.3 Results …...31

2.4 Discussion …...32

2.5 Conclusions …...35

2.6 Reference …...35

Chapter 3 Estimation of quality-adjusted life expectancy in patients under prolonged mechanical ventilation 3.1 Introduction …...44

3.2 Material and methods …...46

3.3 Results …...49

3.4 Discussion …...51

3.5 Conclusions …...54

3.6 References …...55

Chapter 4 Cost-effectiveness of applying prolonged mechanical ventilation in Taiwan 4.1 Introduction ...65

4.2 Material and methods ...66

4.3 Results ...70

4.4 Discussion ...72

4.5 Conclusions ...75

4.6 References ...75

Chapter 5 Conclusion ...87 附錄

Appendix 1

Hung MH, Yan YH, Fan PS, et al. Measurement of quality of life using EQ-5D in patients on prolonged mechanical ventilation:

comparison of patients, family caregivers and nurses. Qual Life Res 2010; 19:721–7.

Appendix 2

Hung MC, Yan YH, Fan PS, et al. Estimation of quality-adjusted life expectancy in patients under prolonged mechanical ventilation.

Value Health 2011; 14:347-53.

Appendix 3

Hung MC, Lu HM, Chen L, et al. Life expectancies and incidence rates of patients under prolonged mechanical ventilation: A population-based study during 1998-2007 in Taiwan. Crit Care 2011; 15:R107.

Appendix 4

Dynamic changes and lifetime scores of quality of life measured by WHOQOL-BREF for patients with different types of ischemic

xi

圖目錄 Figure 1.1

Flow chart of the selection process used for the study cohort Figure 2.1

Recruitment process for study participants Figure 3.1

The life expectancy (namely, area under the dotted line) and loss of life expectancy (namely, shadowed area between the two curves) in years for 633 patients under PMV (prolonged mechanical ventilation) after extrapolation to 300 months

Figure 3.2

Quality adjusted survival for patients undergoing PMV (prolonged mechanical ventilation) after adjustment of survival function (N=633) with the utility values of quality of life measured with EQ-5D.

Figure 4.1

Flow chart of the computation process for cost-per-QALY (quality-adjusted life year) and cost-per-expected life

Figure 4.2

Quality adjusted survival for patients undergoing PMV (prolonged mechanical ventilation) with partial cognition after adjustment for survival function (N=50,481) with the utility values of quality of life measured with EQ-5D (N=55). The result of QALE (quality-adjusted life expectancy) of an average patient was 0.98 QALY by summing the areas under the quality-adjusted survival curve.

Figure 4.3

Curves of average monthly healthcare expenditures adjusted by survival probability plotted along time after beginning PMV (prolonged mechanical ventilation), stratified by cancer, injuries, and COPD (chronic obstructive pulmonary disease) cases with age less than 64 years old. The areas under these curves were the lifetime costs paid by the National Health Insurance.

Figure 4.4

Plot of association between the conventional ICER (or, incremental cost-per- QALY [quality-adjusted life year] gained) and the cost-per-expected life for PMV patients with different specific illnesses [○] and multiple co-morbidities stratified by age groups (<65[●], 65-74[■], 75-84[▼], and >84[◆] years old)

表目錄 Table 1.1

Age-specific incidence rates (per 100,000 person-years), and CIR* of patients under prolonged mechanical ventilation

Table 1.2.

Demographic characteristics and survival of patients undergoing prolonged mechanical ventilation stratified by different underlying diseases

Table 1.3.

Clusters of different co-morbidities categorized by latent class analysis in patients with prolonged mechanical ventilation

Table 2.1.

Characteristics of study subjects and proxies (family caregivers and nurses). Values are expressed as means (standard deviations) or percentages.

Table 2.2.

Frequency distribution of agreement among patients with fair to good cognition (mini-mental status examination score >15) for each dimension of the EQ-5D.

Table 2.3.

Frequency distribution of agreement among patients with poor cognition and their proxies for each dimension in the EQ-5D.

Table 2.4.

Differences and reliabilities of EQ-5D index measured in patients on PMV (prolonged mechanical ventilation) and their proxies, according to value systems established in the United Kingdom (U.K.) and United States (U.S.).

Table 3.1.

Demographic and clinical characteristics of patients under prolonged mechanical ventilation (PMV) in a cohort group to obtain the survival function and a cross-sectional sample for measurements of quality of life.

Table 3.2.

Quality-adjusted life expectancy (QALE, in years) and expected lifetime loss of utility for patients under PMV (prolonged mechanical ventilation) based on different value systems of EQ-5D and patient’s cognition for sensitivity analysis

Table 4.1.

xiii

undergoing prolonged mechanical ventilation in Taiwan, stratified by different underlying diseases, with sensitivity analysis of quality-adjusted life expectancy (QALE) under different states of cognition

Table 4.2.

Cost-per-QALY (quality-adjusted life year) and cost-per-expected life for patients undergoing prolonged mechanical ventilation in Taiwan, stratified by different co-morbidities and categorized by latent class analysis stratified by age, with sensitivity analysis of quality-adjusted life expectancy (QALE) under different states of cognition.

Overview

As we know, the core value of public health is health for all and the first step would be to achieve the goal of health care for all. In Taiwan, we are very fortunate to have developed the National Health Insurance (NHI) that implements a system of universal coverage (over 99% now) for all people of Taiwan. The NHI comprehensively covers various health care services, especially those of catastrophic illnesses, including maintenance hemodialysis for end stage renal disease and chronic ventilator use consecutively for more than 21 days, etc., which results in a tremendous financial burden and threatens the sustainability of the NHI. According to the reimbursement data of NHI in 2010, the prolonged mechanical ventilation (PMV) consumes the second highest in the average health expenditure with an annual total health expenditure of 17 billion NT (New Taiwan) dollars. In order to keep the sustainability of the NHI under good quality, efficiency and equitable distribution of resources at the same time, we began to think of different strategies. One of them is to explore the incidence rate, quality of life, and survival of the patient undergone prolonged mechanical ventilation, and cost effectiveness of such a healthcare service.

In 1998, the Bureau of NHI drafted a prospective payment program to encourage integrated care for mechanically ventilated patients, which was implemented in July 2000. After several revisions, this program ultimately covered four types of mechanical ventilator care: ICU care (acute stage, <21 days), respiratory care center (RCC, a sub-acute stage for weaning training of up to 42 days), respiratory care ward (RCW, a chronic stage or long-term care), and home care service (stable stage, in which the patient is cared for directly by family caregivers).

The number of patients who require PMV is rapidly increasing worldwide,

- 2 -

availability and effectiveness of this new technology. The fact that many patients require continued respiratory care after being transferred into a rehabilitation facility creates a tremendous financial burden. Furthermore, there is often a gap between families’

unreasonably optimistic expectations and clinicians’ professional judgment. This gap frequently results in difficulties arriving at consensus clinical decision-making. In many cases, these challenges are not easily resolved. These issues are exacerbated by the lack of evidence regarding expected survival times for different subgroups of patients, especially for those suffering from multiple co-morbidities.

The financial impact is particularly heavy for countries with a national health insurance system with universal coverage of their citizens and is aggravated by the innovation of new technologies and an aging population. Most of these countries have adopted economic analyses to improve overall cost-effectiveness and contain costs.

Evaluation of the cost per QALY (quality-adjusted life year) gained from different healthcare services will ensure the most cost-effective policy decision for health care.

Similar to what is the case in western countries, discrepancies frequently exist in Taiwan between a family’s initial expectations and their physician’s professional judgment. These discrepancies impair communication among patients, their families, and health care workers for clinical decision-making before and throughout the course of installing mechanical ventilation. Thus, there is a need to estimate the incidence rates and life expectancies for PMV patients with various diagnoses. Accurate prognoses are essential to propose and establish a sustainable national policy and to facilitate communication among different stakeholders. To examine the above issues, we collected a random sample from the national database and compared age-specific incidence rates, cumulative incidence rates (CIRs), median survival, and life expectancies of PMV patients stratified according to their underlying diseases.

To our knowledge, there has been few study that estimates the long term survival, quality of life, lifetime cost, and cost effectiveness together for PMV patients with multiple co-morbidities, which are the major goals of this dissertation. The aims of our study are: (1) to determine the incidence rate (IR), median survival, life expectancies for different types or co-morbidities of patients undergoing prolonged mechanical ventilation (PMV) ; (2) to explore how much difference on the quality of life (QOL) assessments between patients under PMV and their proxies (family caregivers and nurses); (3) to estimate the quality-adjusted life expectancy (QALE) and the expected lifetime utility loss of different types of patients with PMV (4) and to estimate the incremental cost-effectiveness ratios (ICER) for different types of patients undergoing PMV.

To achieve these aims, together we have already performed the following four studies:

1. Life expectancies and incidence rates of patients under prolonged mechanical ventilation: A population-based study during 1998-2007 in Taiwan (Chapter 1)

2. Measurement of quality of life using EQ-5D in patients on prolonged mechanical ventilation: comparison of patients, family caregivers and nurses (Chapter 2)

3. Estimation of quality-adjusted life expectancy in patients under prolonged mechanical ventilation (Chapter 3)

4. Cost-effectiveness of applying prolonged mechanical ventilation in Taiwan (Chapter 4)

- 4 -

Life expectancies and incidence rates of patients under prolonged mechanical ventilation: A population-based study during 1998-2007 in Taiwan

(Chapter 1) Abstract

Introduction: This study examined the median survival, life expectancies, and cumulative incidence rate (CIR) of patients undergoing prolonged mechanical ventilation (PMV) stratified by different underlying diseases.

Methods: According to the National Health Insurance Research Database (NHIRD) of Taiwan, there were 8,906,406 individuals who obtained respiratory care during the period from 1997-2007. A random sample of this population was performed, and subjects who had continuously undergone mechanical ventilation for more than 21 days were enrolled in the current study. Annual incidence rates and CIR were calculated.

After stratifying the patients according to their specific diagnoses, latent class analysis (LCA) was performed to categorize PMV patients with multiple co-morbidities into several groups. The life expectancies of different groups were estimated using a semi-parametric method with a hazard function based on the vital statistics of Taiwan.

Results: The analysis of 50,481 PMV patients revealed that incidence rates increased as patients grew older and that cumulative incidence rates (17-85 years old) increased from 0.103 in 1998 to 0.183 in 2004 before stabilising thereafter. The life expectancies of PMV patients suffering from degenerative neurological diseases, stroke, or injuries tended to be longer than those with chronic renal failure or cancer. Patients with chronic obstructive pulmonary disease survived longer than did those co-morbid with other underlying diseases, especially septicaemia/shock.

Conclusions: PMV provides a direct means to treat respiratory tract diseases and to sustain respiration in individuals suffering from degenerative neurological diseases, and

individuals with either of these types of conditions respond better to PMV than do those with other co-morbidities. Future research is required to determine the cost-effectiveness of this treatment paradigm.

1.1 Introduction

The number of patients who require prolonged mechanical ventilation (PMV) is rapidly increasing worldwide, apparently due to aging, a greater number of co-morbidities, and the increasing availability and effectiveness of this new technology [1-3]. The fact that many patients require continued respiratory care after being transferred into a rehabilitation facility creates a tremendous financial burden [3-4]. Furthermore, there is often a gap between families’ unreasonably optimistic expectations and clinicians’

professional judgment. This gap frequently results in difficulties arriving at consensus clinical decision-making [5]. In many cases, these challenges are not easily resolved.

These issues are exacerbated by the lack of evidence regarding expected survival times for different subgroups of patients, especially for those suffering from multiple co-morbidities.

The National Health Insurance (NHI) of Taiwan has implemented a system of comprehensive coverage for various health care services, including maintenance hemodialysis and chronic respiratory care. The NHI was first established in 1995 and has been extended to cover over 99% of the citizens of Taiwan [6-7]. In 1998, the Bureau of the NHI drafted a prospective payment program to encourage integrated care for mechanically ventilated patients, which was implemented in July, 2000 [8]. After several revisions, this program ultimately covered mechanical ventilator care in the following settings: intensive care units (ICU, acute stage, <21 days), respiratory care centres (RCC, a sub-acute stage for weaning training, up to 42 days), respiratory care

- 6 -

during which the patient is cared for directly by family caregivers). The rising number of patient-days for mechanical ventilation usage during 1997-2004 increased the financial burden of the NHI [9]. Similar to what is the case in Western countries [5], discrepancies frequently exist in Taiwan between a family’s initial expectations and their physician’s professional judgment. These discrepancies impair communication among patients, their families, and health care workers for clinical decision-making before and throughout the course of installing mechanical ventilation. Thus, there is a need to estimate the incidence rates and life expectancies for PMV patients with various diagnoses. Accurate prognoses are essential to propose and establish a sustainable national policy and to facilitate communication among different stakeholders. To examine the above issues, we collected a random sample from the national database and compared age-specific incidence rates, cumulative incidence rates (CIRs), median survival, and life expectancies of PMV patients stratified according to their underlying diseases.

1.2 Materials and methods

Study population, datasets, and calculation of age-specific and cumulative incidence rates

This study was approved by the Institutional Review Board of the National Taiwan University Hospital, which also waived the requirement for obtaining informed consent because the study was conducted on a secondary database with encrypted identification numbers. The reimbursement data file obtained from the NHI of Taiwan was transformed into a research database by the National Health Research Institutes (in Chunan, Taiwan) [10]. The identification numbers of all individuals in the reimbursement data file were encrypted to protect their privacy. These files contained detailed demographic data (including birth date and sex) and information regarding the

health care services provided for each patient, including all payments for outpatient visits, hospitalizations, prescriptions, diagnoses and intervention procedures. The data for each inpatient hospitalization included up to five diagnoses, which were coded according to the International Classification of Diseases, Ninth Revision (ICD-9) and the date of each prescription or procedure. In total, 8,906,406 individuals had undergone invasive or non-invasive respiratory care at least once during the period from 1997-2007.

This number corresponds to approximately 29.4% of the entire insured population.

Because the government has established guidelines stating that no more than 10% of all data can be drawn for research, we applied for a random sample of these patients with a 3.4:1 ratio and enrolled subjects who had undergone mechanical ventilation for more than 21 days.

According to the definition of PMV in Taiwan[8], we included patients over the age of 17 who had undergone either invasive or non-invasive mechanical ventilation, with negative or positive pressure ventilators for at least 21 consecutive days in the ICU or the RCC. To ensure that all of the patients were incident cases, we excluded all prevalent cases found in 1997 and began the collection in 1998, as illustrated in Figure 1.1. The calendar year- and age-specific incidence rates (IRi) were determined by taking the number of new cases of PMV patients in that stratum, multiplying by the sampling factor of 3.4, and then dividing the resulting value by the number of individuals within the specific stratum obtained from the census of the Ministry of the Interior in Taiwan [11]. The CIR (cumulative incidence rate) formula was calculated as follows [12]:

CIR=1-exp [-Σi (IRi) (△ti )], where IRi represents the age-specific incidence rate and

△ti indicates the range of each age stratum. We calculated the CIR17-85, which estimates the likelihood that an average person in Taiwan would require PMV assuming that he or

- 8 -

Categorization of patients for estimation of life expectancies

All recruited subjects were followed until the end of 2007 to determine whether they were alive, deceased, or censored. Because patients who had undergone PMV usually suffered from a combination of multiple co-morbidities and 5 major diagnoses can be retrieved for each PMV patient prior to and closest to the 1^st day, we developed a strategy to identify different homogeneous groups to make more-accurate estimates of life expectancy. First, we excluded specific categories with extremely low frequencies, including HIV (human immunodeficiency virus) infection (n=27) and complications during pregnancy/childbirth/perinatal period (n=240). Second, people with major diagnoses that may cause premature mortality were stratified and analysed separately, including cancer, end-stage renal disease, liver cirrhosis, multiple sclerosis or degenerative neurological diseases, Parkinson’s disease and injury or poisoning , as each group shares major common characteristics that predict mortality. Third, because the vast majority of patients suffered from a combination of multiple chronic diseases that may cause premature mortality, such as diabetes mellitus, coronary and/or heart failure, hypertension, respiratory system and/or urinary tract infections, acute renal failure, and septicaemia with and without shock, etc., we conducted latent class analysis for these cases to categorize them into clusters or specific homogeneous groups for estimations of life expectancies. During this process, we grouped several closely related diagnoses together and converted their original ICD-9 codes into Clinical Classifications Software (CCS) codes [13] so that sufficient numbers could be obtained for survival analysis. Following the above direction, the ICD-9 codes of septicaemia included 0031, 0202, 0223, 0362, all subcategories of 038, and 7907 (bacteraemia);

those of shock included all subcategories under code 785.

Statistical analysis

Binary and categorical variables were summarized using frequency counts and percentages. Continuous variables that were distributed normally are presented as means.

Latent class analysis

To determine the underlying causes that were more likely to lead to PMV, we applied latent class analysis (LCA) to group separate co-morbidity diagnoses into no more than 10 clusters of in-patients who had undergone PMV. Because pneumonia and respiratory failure are the most common reasons for mechanical ventilation, these conditions were not included in this model. This analysis resulted in an LCA model consisting of 32 broad diagnosis categories that included chronic diseases that had been previously classified into 260 categories by CCS. LCA assumes that responses are conditionally independent within classes after accounting for class membership [14]. In other words, LCA allows for the grouping of the PMV patients into several relatively homogeneous clusters of diagnosis patterns. In constructing the model, each cluster or class was named after the major disease (i.e., with the highest prevalence or likelihood) present within each age strata. Akaike Information Criteria (AIC) was used to assess the goodness of fit of the model [15]. Lower AIC statistics were considered to indicate a better statistical fit of the model to the data. If any single category exhibited a prevalence approaching 100% for a given condition across different age groups, then we assumed that these conditions could be re-classified into groups with specific diseases, and life expectancy estimations were conducted separately. Throughout this process, we found that only stroke could be further separated from the groups of multiple co-morbidities, and thus, the life expectancy estimation for stroke patients was performed independently. SAS statistical software (version 9.1) and R statistical

- 10 -

Estimation of life expectancy

Each new patient who fulfilled the definition of PMV was followed beginning on the 1^st day of PMV treatment and continuing until he/she was deceased or censored on December 31^st of 2007. The median survival, or the time at which only half of the patients within a given category were still alive, was estimated by the Kaplan-Meier method. In general, most patients did not survive longer than 1-3 years, although some patients did exhibit a longer survival time. All patients survived the initial 21 days of treatment by mechanical ventilation, and the survival times reported here exclusively refer to survival duration thereafter. Thus, the lifetime survival of PMV patients (up to 300 months when excluding those older than 85 years) was obtained using a linear extrapolation of a logit-transformed curve of the survival ratio between the PMV and an age- and gender-matched reference population generated by the Monte Carlo method from the life table of the general population of Taiwan. The detailed method and mathematical proof assuming a constant excess hazard have been described in our previous reports [16-20]. To facilitate the computation, we used ISQoL, a software program that was built based on the R statistical package for lifetime expectancy estimation and 300-month extrapolation (excluding those older than 85 years) and can be downloaded for free [21].

Validation of the extrapolation method for survival functions

Empirical PMV data from the National Health Research Institutes provided us with an opportunity to validate the actual performance of our semi-parametric method of estimation. Thus, we selected sub-cohorts of patients beginning on the 1^st day that they received PMV between 1998 and 2001. We assumed that these cohorts were only followed until the end of 2001 and then extrapolated these results to the end of 2007.

We compared our predictions with the Kaplan-Meier estimates of the direct follow-ups

from 1998-2007. Assuming that the Kaplan-Meier estimates are the gold standard, we calculated the relative biases for sub-cohorts stratified by different underlying diseases and co-morbidities [22]. The relative biases were computed to compare the differences in values between the Kaplan-Meier estimates and the Monte Carlo extrapolation method.

1.3 Results

Basic characteristics of the PMV Cohort

A total of 50,481 new patients with PMV were included during the study period (40%

female, mean age of 72±14.5 years, median survival 0.37 years, and overall life expectancy of 2.68 years). If we counted only the primary diagnosis (out of a maximum of five diagnoses) for each patient, the top five primary diagnoses were acute respiratory failure (15%), pneumonia (12%), intra-cerebral haemorrhage (5%), septicaemia (3%) and chronic airways obstruction (2%). The tracheotomy rate was 60.1%, which reflects the ethnic Chinese cultural tradition that typically avoids additional traumatic wounds if a patient is expected to pass away soon.

Trends of age-specific incidence rates and CIR over time

After NHI began to reimburse long term usage of mechanical ventilation to relieve the congested intensive care ward in 1998, the incidence rate started to rise and showed an increased trend with older age (Table 1.1). In the groups aged 65-74, 75-84, and older than 85 years, increased incidence rates of 76%, 88%, and 119%, respectively, were observed from 1998 to 2004, followed by a slight drop after 2005. The CIR (17-85 years) increased from 0.103 in 1998 to 0.183 in 2004 and then decreased to 0.145 in 2007.

Life expectancies of PMV patients with specific underlying diseases

- 12 -

summarized in Table 1.2. Although median survival of most categories were less than one year, many patients showed life expectancies longer than 2-3 years, indicating that some patients survived relatively long periods of time. The median survival and life expectancies of PMV patients with degenerative neurological disease, stroke, or injuries were generally longer than those with chronic renal failure or cancer. When a patient contracted both cancer and chronic renal failure, the median survival durations and life expectancies were the shortest. Patients with stroke were initially included in the latent class analysis because of the presence of multiple co-morbidities, but a distinctive category of 100% prevalence of stroke consistently appeared across different age strata.

Thus, we separated this group and estimated the associated life expectancies for different age strata, as summarized in Table 1.2.

Life expectancies of age-specific clusters in PMV patients with multiple co-morbidities

Among the 23,697 PMV patients with multiple co-morbidities, the latent class model usually yielded 3-4 clusters, including heart diseases, septicaemia/shock, chronic obstructive pulmonary diseases and/or others (e.g., urinary tract infections), as summarized in Table 1.3. Diabetes mellitus seemed to be the most frequent co-morbid disease among all different clusters because the prevalence rates were all above 14.5%.

The life expectancy and median survival of PMV patients with COPD (chronic obstructive pulmonary disease) were generally longer than those of other clustered groups, especially those with septicaemia/shock. This trend continued until the age of 85, after which PMV patients with different underlying co-morbidities seem to show similar outcomes.

Validation results of the extrapolation method

The results obtained to validate our semi-parametric method show that the relative biases were all below 20%. Among them, the relative biases of most PMV patients with

a specific diagnosis ranged between 0.9% and 5.5%. Stroke patients were an exception and usually suffered from other co-morbidities. Patients with a combination of different diseases (or clusters) appeared less likely to fulfil the assumption of a constant excess hazard completely and resulted in greater relative biases, perhaps because they represent a relatively heterogeneous patient population. Nonetheless, the absolute differences between our estimates and those obtained using the Kaplan-Meier method were all below 0.25 life-years, except for the 65-74-year-old multiple co-morbidity categories, which showed an absolute difference of 0.39 life-years.

1.4 Discussion

To our knowledge, this is the first study to analyze a nationally representative PMV dataset to estimate the incidence rates, CIR, and life expectancies stratified by age and different clusters of diagnoses. Our findings showed that new cases of PMV increased significantly from 9,296 to 21,818 between 1998 and 2004. The age-specific incidence rates increased as people grow older, a result that is consistent with previous reports from scholars in the U.S. and Canada [1-2, 4, 23]. However, the highest age-specific incidence rate of PMV was observed in patients older than 85 years in Taiwan, and this rate was approximately 4 to 5 times higher than those reported in the U.S. [24]. We attempted to quantify the lifetime risk of PMV by calculating the CIR17-85, which increased from 0.103 to 0.145 between 1998 and 2007 (Table 1.1). This finding implies that an adult person in Taiwan who lives until the age of 85 has 10-15% chance of requiring PMV. Given the resource-intensiveness of PMV, this issue requires special attention. When the Bureau of National Health Insurance of Taiwan began to audit the quality of the integrated respiratory care system in 2003, including the rates of successful weaning, readmission and nosocomial infection, the incidence of PMV

- 14 -

In the past, there has been a general lack of data regarding the life expectancies associated with different diagnoses for patients undergoing PMV. This has made it difficult for stakeholders to reach consensus clinical decisions regarding optimal treatment strategies. The issue becomes even more complicated when payment is provided via NHI or a third party. It is understandable that the patient and his/her family always expect successful weaning and good recovery, even after more than 21 days of continuous mechanical ventilation or PMV. However, according to our previous study, most patients undergoing PMV survive only approximately 1.5 to 2 years, and approximately 62% of them suffer from cognitive impairments and poor quality of life.

Accounting for these factors results in an overall quality-adjusted life expectancy of only 0.3-0.4 to 0.6-0.7 quality-adjusted life years (QALYs) [25-26]. Thus, this study further provided crucial estimates of the median survival and life expectancies of patients undergoing PMV with different diagnoses or co-morbidities, as summarized in Tables 1.2 and 1.3. Table 1.2 shows that the life expectancies were shortest for PMV patients with chronic renal failure and cancer or any condition co-morbid with them, followed by Parkinson’s disease and stroke. In contrast, the life expectancies for degenerative neurological diseases, liver cirrhosis, injuries and poisonings were greater than 3.6 years. When stratified by age categories, the median survival and life expectancies for PMV patients older than 85 years were below 4.6 months and 21.8 months, respectively, which were also observed for all of the different types of co-morbidities (Tables 1.2 and 1.3). The above figures call into question the cost-effectiveness of current policies and should be considered by policy makers and the public in discussions regarding the bioethics of PMV care, especially given the limited resources of the NHI in Taiwan. Although more and more countries have tried to implement the principle of universal coverage in their national health insurance plans

[27], our results provide data highlighting the needed evidence for developing strategies of sustainable management.

Although previous studies have shown similar characteristics of multiple co-morbidities in PMV patients, these reports did not stratify patients into special clusters [1, 9, 24, 28].

The LCA showed that the underlying co-morbidities associated with PMV could be largely classified into the major categories of heart diseases, septicaemia/shock, and chronic obstructive pulmonary disease based on the high prevalence of each cluster.

Overall, LCA indicates that the life expectancies generally decreased with older age. In particular, we found that approximately 50% of the PMV patients with a combination diagnosis of septicaemia and shock usually survived for less than 4 months, and their life expectancies were usually shorter than those determined for the other clusters within the same age stratum. The generally longer survival time of PMV patients with COPD corroborated the hypothesis that the establishment of mechanical ventilation provides more direct access for clinicians to solve problems coming from the respiratory tract, while patients with other underlying diseases may not be improved significantly unless their underlying disorders were also resolved. This advantage disappeared in individuals over 85 years of age because a high proportion of these COPD patients also suffered from other major diseases, including urinary tract infection (29%) and other respiratory diseases (26%), as shown in Table 1.3.

Our study has several limitations. First, the database did not contain any information regarding the severity and/or actual clinical data of the PMV patients. Thus, we were unable to further stratify these patients. However, because they were all under PMV care for more than 21 days, all of the patients were associated with extremely severe conditions, which resulted in a very short life expectancy and suggested that 10 years of

- 16 -

must fulfil all of the reimbursement regulations of the NHI, it is possible that some diagnoses are over-represented because they were more easily reimbursed. However, the National Health Insurance of Taiwan has offered a list of 30 major categories of catastrophic illnesses that are exempt from partial co-payments, and each has its specific diagnostic criteria to prevent any abuse [29]. For example, all types of malignant neoplasm do not require co-payments, and evidence of histopathology and/or cytology is generally required for diagnoses of cancer. A diagnosis of end stage renal disease requires documentation of chronic kidney disease with an irreversible creatinine level of

more than 8 mg/dL, or creatinine level more than 6 mg/dL with diabetes mellitus as a co-morbid condition [30]. Thus, we have strict criteria for almost all the major diagnoses listed in Table 1.2. The potential selection bias for the common diseases listed in Table 1.3 is probably minimal because the 43 broad categories were collapsed from the 260 categories of CCS codes [13], and latent class analysis ensured that each category was as homogeneous as possible.

1.5 Conclusions

The number of PMV patients in Taiwan has increased during the last decade. Patients with different underlying diseases showed different median survival and life expectancies. The establishment of mechanical ventilation directly targets to solve problems of respiratory tract and provides sustainable ventilation, which may improve the survival of patients with COPD or degenerative neurological diseases more than those with other underlying causes such as septicaemia/shock, heart failure, cancer, or end stage renal disease. However, the advantages of PMV seem to decrease for the elderly, especially those over age 85. The results also call for further evaluation of the cost-effectiveness and bioethics of such care in Taiwan and highlight the need for early planning of resource allocation in any system of health insurance with universal

coverage.

Key messages

z The number of new patients undergoing treatment of PMV (prolonged mechanical ventilation) has increased rapidly during the last decade in Taiwan.

z The life expectancies of PMV patients with degenerative neurological diseases, stroke, or injuries/poisoning as their primary co-morbidity seemed to survive longer than those with chronic renal failure or cancer, or a co-morbidity with them.

z Among PMV patients with multiple co-morbidities, those with COPD as the major underlying co-morbidity seem to survive longer than patients with other co-morbidities, perhaps because this treatment specifically targets the respiratory tract, which is compromised in COPD. The benefits of PMV decrease for the elderly, especially those with an age of over 85.

1.6 References

1. Needham DM, Bronskill SE, Sibbald WJ, Pronovost PJ, Laupacis A. Mechanical ventilation in Ontario, 1992-2000: incidence, survival, and hospital bed utilization of noncardiac surgery adult patients. Crit Care Med 2004; 32:1504-1509.

2. Carson SS, Cox CE, Holmes GM, Howard A, Carey TS. The changing epidemiology of mechanical ventilation: a population-based study. J Intensive Care Med 2006; 21:173-182.

3. Carson, S.S. Outcomes of prolonged mechanical ventilation. Curr Opin Crit Care 2006; 12(5), 405-411.

4. Cox CE, Carson SS, Holmes GM, Howard A, Carey TS. Increase in tracheotomy for prolonged mechanical ventilation in North Carolina, 1993-2002. Crit Care Med 2004; 32 (11), 2219-2226.

- 18 -

Carson SS, Tulsky JA. Expectations and outcomes of prolonged mechanical ventilation. Crit Care Med 2009; 37:2888-2894.

6. Department of Statistics, Ministry of the Interior: Number of Population. Available online at: http://www.moi.gov.tw/stat/english/monthly.asp.

7. Bureau of National Health Insurance: Insurance Coverage. Available online at:

http://www.nhi.gov.tw/english/index.asp.

8. Bureau of National Health Insurance: Prospective Payment System of Ventilator Dependents Managed Care Demonstration. Available online at:

http://www.nhi.gov.tw/nhi/index.aspx. (In Chinese).

9. Cheng SH, Jan IS, Liu PC. The soaring mechanic ventilator utilization under a universal health insurance in Taiwan. Health Policy 2008; 86:288-294.

10. National Health Research Institute: National Health Research Database. Available online at: http://w3.nhri.org.tw/nhird//en/Background.html.

11. Ministry of the Interior in Taiwan: Census. Available online at: http:/

http://sowf.moi.gov.tw/stat/year/y02-01.xls. (In Chinese).

12. Wang JD: Basic measurements in epidemiological research. In basic principles and practical applications in epidemiological research. Edited by Wang, JD. Singapore:

World Scientific Press; 2002: 135-136.

13. Agency for Healthcare Research and Quality: Clinical Classification Software (CCS) 2010. Available online at: http://www.ahrq.gov/data/hcup/datahcup.htm.

14. McCutcheon AL: Basic concepts and procedures in single and multiple group latent class analysis. In applied latent class analysis. Edited by Jacques A. Hagenaars, Allan L. McCutcheon. New York: Cambridge University Press, 2002.

15. Akaike H. Factor analysis and the AIC. Psychometrika 1987; 52:317-332.

16. Hwang JS, Wang JD. Monte Carlo estimation of extrapolation of quality-adjusted

survival for follow-up studies. Stat Med 1999; 18:1627-1640.

17. Chu PC, Hwang JS, Wang JD, Chang YY. Estimation of the financial burden to the National Health Insurance for patients with major cancers in Taiwan. J Formos Med Assoc 2008; 107:54-63.

18. Fang CT, Chang YY, Hsu HM, Twu SJ, Chen KT, Lin CC, Huang LY, Chen MY, Hwang JS, Wang JD, Chuang CY. Life expectancy of patients with newly-diagnosed HIV infection in the era of highly active antiretroviral therapy.

QJM 2007; 100(2):97-105.

19. Hwang JS, Wang JD. Integrating health profile with survival for quality of life assessment. Qual Life Res 2004; 13:1-10

20. Hwang JS, Tsauo JY, Wang JD. Estimation of expected quality adjusted survival by cross sectional survey. Stat Med 1996; 15:93-102.

21. Software: ISQoL package. Available online at:

http://www.stat.sinica.edu.tw/jshwang.

22. Chu PC, Wang JD, Hwang JS, Chang YY. Estimation of life expectancy and the expected years of life lost in patients with major cancers: extrapolation of survival curves under high-censored rates. Value Health 2008; 11:1102-1109.

23. Needham DM, Bronskill SE, Calinawan JR, Sibbald WJ, Pronovost PJ, Laupacis A.

Projected incidence of mechanical ventilation in Ontario to 2026: preparing for the aging baby boomers. Crit Care Med 2005; 33:574-579.

24. Zilberberg MD, de Wit M, Pirone JR, Shorr AF. Growth in adult prolonged acute mechanical ventilation: implications for healthcare delivery. Crit Care Med 2008;

36:1451-1455.

25. Hung MC, Yan YH, Fan PS , Lin MS, Chen CR, Kuo LC, Yu CJ , Yao G, Hsieh CL,

- 20 -

mechanical ventilation: comparison of patients, family caregivers and nurses. Qual Life Res 2010; 19:721-727.

26. Hung MC, Yan YH, Fan PS, Lin MS, Chen CR, Kuo LC, Yu CJ, Wang JD.

Estimation of quality-adjusted life expectancy in patients under prolonged mechanical ventilation. Value Health 2011; 14 (2): 347-353.

27. Jennings CC. Hayes KJ. Health insurance reform and the tensions of Federalism. N Engl J Med 2010; 362(24):2244-2246

28. Esteban A, Anzueto A, Frutos F, Alía I, Brochard L, Stewart TE, Benito S, Epstein SK. Characteristics and outcomes in adult patients receiving mechanical ventilation:

a 28-day international study. JAMA 2002; 287:345-355.

29. National Health Research Institute: Catastrophic Illness Patients. Available online at:

http://www.nhi.gov.tw/english/.

30. Lai MN, Lai JN, Chen PC, Hsieh SC, Hu FC, Wang JD. Risks of Kidney Failure Associated With Consumption of Herbal Products Containing Mu Tong or Fangchi:

A Population-Based Case-Control Study. Am J Kidney Dis 2010; 55: 507-518.

22

Table 1.1 Age-specific incidence rates (per 100,000 person-years), and CIR* of patients under prolonged mechanical ventilation

Age group, years 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

No. of new cases 9,296 12,651 12,913 15,660 17,731 19,737 21,818 21,692 20,414 19,723

17-34 5.1 6.1 5.5 5.8 6.1 8.4 9.9 9.2 9.4 9.9

35-44 10.1 10.8 11.8 14.5 14.4 15.4 19.2 17.4 19.8 18.1 45-54 21.8 33.3 29.3 32.4 37.2 42.0 40.5 43.5 39.5 39.7 55-64 78.2 101.1 92.2 102.5 111.6 118.9 129.6 119.8 101.7 93.0 65-74 224.0 296.5 284.5 329.2 361.9 379.8 393.9 369.2 331.8 306.4 75-84 622.0 817.2 814.6 967.2 1,045.5 1,072.3 1,166.4 1,036.0 1,004.6 909.2

≧85 1,182.0 1,536.0 1,702.6 2,064.1 2,253.0 2,563.9 2,584.0 2,554.0 2,161.0 2,046.0 CIR (17-85 yrs) 0.103 0.133 0.132 0.153 0.165 0.173 0.183 0.170 0.159 0.145

*CIR: cumulative incidence rates (aged 17-85 years old)

Table 1.2 Demographic characteristics and survival of patients undergoing prolonged mechanical ventilation stratified by different underlying diseases

Number of cases Mean age (SD)*

Female (%)

Median survival (yrs)

Life expectancy (yrs) (SE)*

Cases with single specific disease 21,316 69 (15) 38 0.35 3.40 (0.09)

Cancer 5,367 70 (14) 33 0.17 1.51 (0.13)

Chronic renal failure 2,032 73 (12) 51 0.78 1.36 (0.16)

Liver cirrhosis 1,478 65 (17) 35 0.19 3.59 (0.33)

Multiple sclerosis or degenerative nervous system conditions

378 65 (17) 39 0.89 4.05 (0.64)

Parkinson's disease 341 79 (7) 36 0.85 2.06 (0.30)

Stroke 6,765 70 (13) 42 0.72 3.38 (0.15)

<64 yrs 1,955 53 (9) 35 1.65 5.21 (0.39)

65-74 yrs 1,818 70 (3) 43 0.77 2.98 (0.17)

75-84 yrs 2,176 79 (3) 44 0.56 2.09 (0.13)

>85 yrs 816 88 (3) 54 0.39 1.68 (0.13)

Intracranial and/or spinal cord injury or poisoning 4,955 65 (19) 34 1.06 6.27 (0.24)

<64 yrs 1,949 45 (14) 26 6.20 10.20 (0.49)

65-74 yrs 1,116 70 (3) 39 0.82 3.77 (0.22)

75-84 yrs 13,66 79 (3) 35 0.47 2.67 (0.19)

24

>85 yrs 524 88 (3) 48 0.33 1.82 (0.13)

Cases with more than two specific diseases 4,772 68 (15) 39 0.32 2.96 (0.13)

Cancer and Chronic renal failure 165 71 (11) 44 0.14 1.21 (0.45)

Cancer and others 1,609 70 (14) 35 0.19 1.88 (0.22)

Chronic renal failure and others 743 70 (13) 50 0.21 1.71 (0.28)

* SD: standard deviation; SE: standard error of the mean

Table 1.3 Clusters of different co-morbidities categorized by latent class analysis in patients with prolonged mechanical ventilation

<64 yrs (n= 3,520) 65-74 yrs (n= 5,397) 75-84 yrs (n= 9,747) >85 yrs (n= 5,033)

Class 1 Class 2 Class 3 Class 4 Class 1 Class 2 Class 3 Class 1 Class 2 Class 3 Class 4 Class 1 Class 2 Class 3 Heart

diseases SP/

shock

UTI/

SP

COPD/

others

Heart diseases

SP/

shock

COPD/

others

Heart diseases

SP/

shock

COPD/

others

Respiratory diseases

Heart diseases

SP/

shock

COPD/

others Number of cases 616 919 197 1,788 1,074 1,824 2,499 1,404 2,856 4,142 1,345 870 1,359 2,804 Prevalence of

co-morbidity (%)

Septicaemia 11.3 62.9 72.2 6.7 10.6 66.1 5.5 11.8 72.8 11.0 7.2 10.2 79.1 11.7 Diabetes mellitus 35.5 26.8 26.1 19.7 39.7 26.1 26.5 27.1 20.5 22.6 17.2 18.6 14.5 14.6 Hypertension 17.9 4.9 9.8 8.3 20.8 6.8 13.9 16.1 4.8 17.3 12.5 15.4 3.3 15.2 AMI/Coronary

atherosclerosis

39.9 3.9 2.3 2.4 43.9 4.6 5.2 42.8 5.5 6.6 4.0 42.8 5.6 4.0

COPD 6.5 3.2 0.5 17.9 11.0 11.0 33.9 18.2 12.4 39.6 29.6 24.9 12.7 33.0 Other respiratory

diseases

21.8 23.7 13.8 30.1 22.3 18.1 26.7 20.9 17.2 0 100 24.6 19.4 25.9

Acute renal failure 12.2 20.6 9.5 4.8 12.7 16.4 4.5 12.1 16.7 4.2 5.3 8.7 15.2 5.6

UTI 5.3 0 100 14.8 5.7 25.1 21.1 11.7 28.0 24.3 21.1 16.1 30.6 29.0

Shock 11.2 39.6 33.1 2.7 10.3 39.1 2.7 8.9 38.5 4.9 3.7 7.3 41.6 5.0

Heart failure 42.2 3.2 2.6 2.4 37.0 5.8 4.8 50.1 7.2 7.7 7.4 50.5 8.3 7.7

Median survival 0.80 0.34 0.88 1.64 0.39 0.23 0.55 0.29 0.21 0.95 0.38 0.32 0.20 0.35

26 (yrs)

Life expectancy (yrs) (SE)

5.09 (0.60)

4.51 (0.49)

4.82 (1.49)

5.25 (0.37)

2.55 (0.21)

2.14 (0.13)

2.56 (0.13)

1.86 (0.23)

1.66 (0.10)

2.12 (0.07)

2.18 (0.17)

1.48 (0.12)

1.12 (0.07)

1.52 (0.05) COPD: chronic obstructive pulmonary disease; UTI: urinary tract infections; AMI: acute myocardial infarction; SP: septicaemia; SE: standard error of the

mean

Measurement of quality of life using EQ-5D in patients on prolonged mechanical ventilation: comparison of patients, family caregivers and nurses

(Chapter 2) Abstract

Purpose: This study reports how QOL (quality of life) assessments differ between patients on prolonged mechanical ventilation (PMV) and their proxies (family caregivers and nurses).

Methods: We enrolled consecutive subjects on PMV for more than 21 days from five institutions. We conducted QOL assessments using the Taiwanese version of the EQ-5D in face-to-face interviews. Direct caregivers (family members and nurses) also completed the EQ-5D from the patient’s point of view.

Results: For 55 of the 142 enrolled patients who were able to assess their QOL, we recruited 44 patient-family caregiver pairs, 53 patient-nurse pairs, and 42 family caregiver-nurse pairs. There were 81 family caregiver-nurse pairs out of 87 patients with poor cognition. The agreement between patient-family caregiver pairs was generally higher than that of patient-nurse pairs. As the proportions of exact agreement between family caregivers and nurses for patients with poor cognition were 98-99% for observable dimensions of mobility, self-care, and usual activities, they lead to a minimal difference in the final values.

Conclusions: QOL assessments from family caregivers agreed more closely with patients than did those from nurses using EQ-5D evaluations for patients with clear cognition, but either proxy was acceptable for rating PMV patients with poor cognition.

2.1 Introduction

Patients who require prolonged mechanical ventilation (PMV) are rapidly increasing in number, as the improved quality of care in ICUs (intensive care units) has

28

resulted in long-term survival for many patients [1-2]. However, these patients often require continued respiratory care after transfer to a rehabilitation facility, skilled nursing facility or home care, creating financial burdens for the insurance system and/or the patients’ families [1, 3]. Several studies have reported [1, 4-6], however, that survivors often feel their quality of life (QOL) to be acceptable.

The National Health Insurance (NHI) of Taiwan uses a system of comprehensive coverage for various health care services, including maintenance hemodialysis and chronic respiratory care. The NHI was first established in 1995 and has been extended to cover over 99% of Taiwanese citizens [7-8]. In 1998, the Bureau of NHI drafted a prospective payment program to encourage integrated care for mechanically ventilated patients, which was implemented in July 2000 [9]. After several revisions, this program ultimately covered four types of mechanical ventilator care: ICU care (acute stage, <21 days), respiratory care center (RCC, a sub-acute stage for weaning training of up to 42 days), respiratory care ward (RCW, a chronic stage or long-term care), and home care service (stable stage, in which the patient is cared for directly by family caregivers).

Unlike patients on PMV in other countries [10-14], many patients in Taiwan suffer from concomitant cognitive impairments that preclude direct assessment of their subjective QOL, making evaluation by proxy unavoidable in many cases. The objective of this study was to evaluate the utility assessment of QOL in patients on PMV and to compare patients’ own QOL assessments with those made on their behalf by proxies (family caregivers and professional nurses).

2.2 Material and methods Subjects and Methods

We recruited subjects from five institutions in northern and southern Taiwan and consecutively enrolled current PMV subjects who had already been on mechanical

ventilation for at least for 21 days at various levels of care (ICU, RCC and RCW): 1 medical center, 1 regional hospital, and 3 small local hospitals were included. The study began after approval was obtained from the Institutional Review Boards (IRBs) of the National Taiwan University Hospital and the Chia-Yi Christian Hospital. The three local hospitals also approved the study after reviewing the approval documents of the two major IRBs above. For subjects with basic cognition who were able to communicate (including through clear body language) and who scored at least 15 on the mini-mental status examination (MMSE) [15], we conducted QOL measurements using the EQ-5D in direct, face-to-face interviews. The primary caregivers (family members and nurses) were also asked to independently complete the EQ-5D questionnaire from the patient’s point of view, i.e., each proxy rated how he or she thought the patient would rate his or her own QOL on the day of the interview. All the interviews with the patient and his/her proxy were required to be finished within three days, and we also made sure that the individual patient’s clinical condition was stable by verifying with the medical records. The proxy’s results were directly compared with the patient’s own rating, which was considered to be the gold standard.

Quality of life measured by EQ-5D

The utility value of the QOL for patients on PMV was estimated using the EuroQol five-dimensional (EQ-5D) questionnaire, a generic preference-based instrument. The five dimensions assessed by the EQ-5D are mobility, self-care, usual activities, pain/discomfort and anxiety/depression, with three levels of severity (no problems, some/moderate problems and severe/extreme problems). This provides a utility value that ranges from 0 to 1 based upon the five-dimensional health state classification, in which 0 represents the worst health status and 1 represents perfect health [16-17]. The EQ-5D instrument is a valid and reliable tool for measuring health status that has been

30

extensively used in many countries as an outcome measure, including for critically ill patients [18-20]. It was also recently applied in Taiwan with good validity and moderate reliability [21]. We applied the value systems of the U.S. and the U.K. as a comparison [17, 22]. There is general agreement that patients are the best raters of their QOL;

however, when a patient’s mental status is too poor, family caregivers and nurses may be considered as proxies [11-13]. QOL assessment by EQ-5D was conducted by three research assistants who were first trained in the standard operating procedure. For each patient and his/her proxy, the researcher recorded characteristics including age, gender, education, marital status, relationship to the patient, whether or not the proxy lived with the patient, frequency of caring for the patient, location (ICU or RCC or RCW), Glasgow Coma Scale (GCS), presence/absence of tracheotomy and duration of mechanical ventilation, and history of depression. The duration-to-date for each measurement was defined as beginning on the firstday after PMV and ending on the date of interview by our researcher.

Statistical Analysis

Binary and categorical variables were summarized using frequency counts and percentages. Continuous variables are presented as means if normally distributed. The agreement between QOL scores from patients, family caregivers and nurses for each dimension was determined for each pair by calculating weighted kappa scores [23]. To evaluate any systematic tendency for proxy respondents to overestimate or underestimate QOL, we computed the patient-proxy and proxy-proxy mean differences in the final utility value of the EQ-5D. The mean difference between pair-scores and values were tested with the rank sum and paired-t tests to detect any significant difference when the sample size is small. We calculated the effect size by dividing the mean patient-proxy difference by the standard deviation of the patient responses. The