The Effect of the 921 Chi-Chi Earthquake on the Mortality Risk of the Middle-aged and Elderly

Transcription

1 The Effect of the 921 Chi-Chi Earthquake on the Mortality Risk of the Middle-aged and Elderly Authors: Dr. Chih-Chia Chen Prof. Jack C. Yue Prof. Wei-Der Tsai Time: 2018/11/13

2 Natural Disasters Earthquakes, typhoons and natural disasters often cause comprehensive damage and destruction to human life and health. Survivors of natural disasters, especially the middle-aged and the elderly, are most vulnerable after natural disasters (for bearing physical and mental damages). Higher medical demand and higher mortality risk.

3 Natural Disaster -- Mortality & Morbidity Higher mortality risk closer to epic center e.g. Liang et al. (2001), Kario et al. (1998) Higher death risk due to cardiovascular disease e.g. Armenian et al. (1998), Kario et al. (1998) Infectious diseases, Urinary tract infection, Respiratory illnesses, Gastroenteritis, and Kidney diseases are common after natural disasters e.g. Kario et al. (1998), Armenian et al. (1998), Phifer et al. (1988), Ligon (2006), Lala et al. (2006) Mental health (Emotional trauma, Suicide) e.g. Kar (2006), Kowalski and Kalayjian (2001)

4 Aftermath of Natural Disasters Most past work addressed the immediate impact of disasters on the medical demand and morality risk of survivors. Few literature focused on the aftermath of natural disasters to the morality risk of survivors. In this study, we use the National Health Insurance database to explore the mortality risk of the 921 Chi-Chi Earthquake s survivors. Aged 50 and beyond in disaster counties

5 About 921 Chi-Chi Earthquake

6

7 Disaster Area of 921 Chi-Chi Earthquake The disaster area is located in the central Taiwan, mostly Nan-tou and Tai-chung counties. County Nan-tou Deaths (+ Missing) 886 (36.7%) Injuries 678 (47.1%) Building Destroyed 23,127 (52.2%) Building Damaged 16,792 (40.3%) Tai-chung 1, ,861 12,341 (47.8%) (28.5%) (38.0%) (29.6%) Chang-hua ,048 3,054 (1.2%) (0.8%) (2.4%) (7.3%) Others ,302 9,446 (14.4%) (23.7%) (7.5%) (22.7%) Total 2,415 1,441 44,338 41,633 Source:

8 Medical Usage Study about 921 Earthquake About the medical usage and disease risk for the disaster people: Disaster people have higher medical usage than the non-disaster people, e.g., Huang (2000) and Lee et al. (2000); Increase in out-patient visit about diabetes, high blood pressure (hypertension), and cardiovascular disease, e.g., Lin (2002).

9 Mortality Study about 921 Earthquake About the mortality risk after the quake: The elderly and the kids have higher mortality risk if closer to the epic center, e.g., Liang et al. (2001); Disaster people have higher suicide rate, e.g., Chao et al. (2003); People with mental disorder, in-patient visit, and lower income before the quake have higher mortality risk one month after 921 quake, e.g., Chao et al. (2004).

10 Study Goal Quite a lot of studies focused on the impact of 921 quake, but most of them are interested in the short term effect and not in the elderly. We will explore the time needed for the disaster people (elderly) to go back to the normal. Explore the mortality risk with respect to different causes 5 years after the 921 quake; Compare the people in disaster counties to those non-disaster counties nearby.

11 About the Study Objects Taiwan has started the national health insurance (NHI) in 1995, and more than 98% population are covered (excluding oversea workers). We use NHI database ( ) Disaster counties (Nan-tou & Tai-chung) vs. Non-disaster county (Chang-hua) as control; People aged 50 and over (middle-aged & elderly); Evaluate the impact of 921 quake for disaster vs. non-disaster areas (& people).

12 Residential Information Official statistics is based on the registered population, not permanent residents, and there exists discrepancies, e.g. Chen and Liu (2002), Yan and Yue (2010). Residential records decided by NHI database: Upper respiratory tract infection can decide 70.1% township information, e.g., Lin et al. (2011); Insured address and permanent resident address are the same for most farmers & no-income people, e.g., Lin et al. (2011), Lian (2011).

13 Identification of Location We adapt the idea of Lin et al. (2011) and use minimal lesion diseases (e.g., flu, dental visit) to decide the permanent resident. (Note: Insured address is used if cannot be decided.) It is not likely for the middle-aged and elderly to travel distance for minimal lesion diseases. Most objects are near or retired, and insured as farmers or no-income. Their permanent addresses are the same as insured addresses in general.

14 Elderly Population of 3 Central Taiwan Counties (Census, Population Register, NHI) (unit: 1,000) 中部三縣 內政部與主計處中 NHI Database 中部三縣樣本數 Population Register 1998 內政部人口統計 ( 民國 87 年 ) Population Register 1999 內政部人口統計 ( 民國 88 年 ) 內政部人口統計 Population Register ( 民國 年 ) 主計處 89 年人口普查 2000 Census 人數 ( 萬人 ) ~54 60~64 70~74 80~84 Age 五齡組

15 About Taiwan s 3 Central Counties Population of the middle-aged and elderly in 3 counties is million, is about the same as that from Department of Interior (0.667 million). According to 2002 survey (Directorate General of Budget, Accounting and Statistics), about 10% of population aged 50 and over are not at their registered addresses. Registration without residence & residence without registration both are about 10% (e.g., Hong, 1995; Chang, 2003).

16 Identification of Earthquake Victims Taiwan government issued earthquake card to earthquake victims if there are deaths/injures in their family or their home is destroyed/damaged. The card was effective for about one year, until Sept. 30, 2009, and people with the card enjoy the medical benefits such as Waiver of copayment. NHI database added the records Waiver of copayment for the earthquake card holders, and we can judge the earthquake victims from the records.

17 Death Records Death records in NHI database are not complete and not reliable. For example, databases of in-patient visits (TRAN_CODE== 4 ) and catastrophic illness (STOP_REASON== M ) have death records but the number of deaths is only 32.64%. We need to apply other information, such as outpatient visits and adding/cancelling policies records to identify if one is still alive.

18 Types of Insured Status We use the insured records to judge if one is actively insured, and there are 3 possibilities: TYPE1: Under Insured (521,456 persons, 79.58%); TYPE2: Not Insured (72,051 persons, 11.00%); TYPE3: Un-determined (61,741 persons, 9.42%). The insured status can be used to judge if one is still alive, and it is highly correlated to the data accuracy.

19 TYPE1:Under Insured (521,456 persons, 79.58%) Example 1 Adding Example 2 Adding Cancel Adding Adding Cancel Adding

20 TYPE2:Not Insured (72,051 persons, 11.00%) Example 1 Example 2 Adding Cancel Adding Cancel Adding Cancel

21 TYPE3:Un-determined(61,741 persons, 9.42%) Example 1 Adding Example 2 Adding Cancel Adding Cancel Adding Adding Cancel

22 We use death note and cancellation to verify if one is still alive. It seems these two variables provide useful information. Summary of death note and cancellation Insured 3 Counties Death Note Death Note + Cancel Type persons % persons % persons % Active 521, Cancel 72, , , Undermined 61, , Total 655, , ,

23 Additional Information for Identifying Death We can use the medical records to enhance the data quality of death records, since more medical visits before death (Lee et al., 2011). 5 possible types of death: D1: Death Note; D2: Not Insured + Heavy medical visits 2 months before cancelling insurance; D3: Not Insured + Catastrophic illness patient; D4: Not Insured + In-patient visit 6 months or last medical visit is emergency call before cancelling insurance; D5: Not Insured + No medical visits for two years.

24 Ratio Death Records (Our Method vs. MOI) 中部三縣不同死亡樣本擷取條件下的死亡人數占內政部死亡人數的比例 ( 民國 死亡占率 MOI D1+D2+D3+D4+D5 D1+D2+D3+D4 D1+D2+D3 內政部死亡率 D1+D2+D3+D4+D5 D1+D2+D3+D4 D1+D2+D3 D1+D2 D1 D ~54 歲 55~59 歲 60~64 歲 65~69 歲 70~74 歲 75~79 歲 80~84 歲 85 歲以 Age (D1+D2+D3+D4)/MOI 96.90% ; D5/MOI 99.71% 五齡組

25 Annual # of Deaths (Our Method vs. MOI) 1998 民國 87 年 1999 民國 88 年 死亡人數 Our 中部三縣死亡人數 method MOI 內政部死亡人數 records 死亡人數 Our 中部三縣死亡人數 method MOI 內政部死亡人數 records ~54 55~59 60~64 65~69 70~74 75~79 80~84 85 以上 Age 五齡組 50~54 55~59 60~64 65~69 70~74 75~79 80~84 85 以 Age 五齡組 2000 民國 89 年 2001 民國 90 年 死亡人數 Our 中部三縣死亡人數 method MOI 內政部死亡人數 records 死亡人數 Our 中部三縣死亡人數 method MOI 內政部死亡人數 records ~54 55~59 60~64 65~69 70~74 75~79 80~84 85 以上 Age 五齡組 50~54 55~59 60~64 65~69 70~74 75~79 80~84 85 以 Age 五齡組

26 Study Period of Mortality Risk Study period: 1998Q1~2002Q4 (2+ years before after 921 Earthquake, 1999/9/21). Age-specific mortality rate, not crude death rate, is used to compare mortality risk. In addition, we consider Direct Method Adjustment (or ADR, Age-adjusted Death Rate), where j s x n mx n Px ADR = s P P s s and & n P x are the populations of whole nation & ages j x~x+n, and n m x is the central death rate of ages x~x+n for population j.

27 Mortality Risk: Logistic Regression Two major variables (dummy variables) Disaster area and earthquake victim Other personal variables Age, (Age) 2 Medical Record (CCI_scores) Income (Insured income) Occupation

28 Disaster vs. Non-disaster (a) Areas 災區與非災區死亡率 Disaster Area + (b) Quake Victim* 死亡率 (1/1000) Disaster 災區 ( 南投縣與台中縣 area ) Non-disaster 非災區 ( 彰化縣 ) area 死亡率 (1/1000) Earthquake 災區災民 victim Disaster 災區非災民 & Non-victim Non-disaster 非災區民眾 area Q1 87Q198Q3 87Q399Q1 88Q199Q3 88Q3 00Q1 89Q100Q3 89Q3 01Q1 90Q101Q3 90Q302Q1 91Q1 02Q3 91Q3 Time 民國 87 年第一季至民國 91 年第四季 99Q488Q4 00Q1 89Q1 00Q2 89Q200Q3 89Q300Q4 89Q4 01Q1 90Q101Q2 90Q2 01Q3 90Q3 01Q4 90Q4 02Q1 91Q1 01Q2 91Q2 01Q3 91Q3 01Q4 91Q4 Time 震災後 ( 民國 88 年第四季 ) 至民國 91 年第四季 Mortality Rates after ADR (a) 全國 Mortality Rates after ADR (b) 全國 死亡率 (1/1000) 2.5 死亡率 (1/1000) Q1 87Q1 98Q3 87Q3 99Q1 88Q1 99Q3 88Q3 00Q1 89Q1 00Q3 89Q3 01Q1 90Q1 01Q3 90Q3 02Q1 91Q1 02Q3 91Q3 99Q488Q4 00Q1 89Q1 00Q2 89Q200Q3 89Q300Q4 89Q4 01Q1 90Q101Q2 90Q2 01Q3 90Q3 01Q4 90Q4 02Q1 91Q1 01Q2 91Q2 01Q3 91Q3 01Q4 91Q4 Time 民國 87 年第一季至民國 91 年第四季 Time 震災後 ( 民國 88 年第四季 ) 至民國 91 年第四季

29 Cardio-vascular* (a) Mortality 心血管疾病患者之 Digestive System* (b) Mortality 消化系統疾病 死亡率 (1/1000) 4 3 死亡率 (1/1000) 4 3 罹病者 Victim, 且為災區災民 + Patient 罹病者 Disaster,, 且為災區非災民 Non-victim+Patient 罹病者 Non-disaster, 且為彰化縣民眾 area + Patient 未罹病者之中部三縣民眾 Not Patient Q488Q4 00Q1 89Q1 00Q2 89Q2 00Q3 89Q3 00Q4 89Q4 01Q1 90Q1 01Q2 90Q2 01Q3 90Q3 01Q4 90Q4 02Q1 91Q1 01Q2 91Q2 01Q3 91Q3 01Q4 91Q4 99Q488Q4 00Q1 89Q1 00Q2 89Q200Q3 89Q300Q4 89Q4 01Q1 90Q101Q2 90Q2 01Q3 90Q3 01Q4 90Q4 02Q1 91Q1 01Q2 91Q2 01Q3 91Q3 01Q4 91Q4 Time 震災後至民國 91 年第四季 Time 震災後至民國 91 年第四季 Mortality Rates after ADR (a) 以 Mortality Rates after ADR (b) 4 4 死亡率 (1/1000) 3 死亡率 (1/1000) Q488Q4 00Q189Q1 00Q2 89Q200Q3 89Q300Q4 89Q401Q1 90Q101Q2 90Q201Q3 90Q3 01Q4 90Q4 02Q1 91Q1 01Q2 91Q2 01Q3 91Q3 01Q4 91Q4 99Q488Q4 00Q1 89Q100Q2 89Q200Q3 89Q3 00Q4 89Q4 01Q1 90Q101Q2 90Q2 01Q3 90Q3 01Q4 90Q4 02Q1 91Q1 01Q2 91Q2 01Q3 91Q3 01Q4 91Q4 Time 震災後至民國 91 年第四季 Time 震災後至民國 91 年第四季

30 Cancer Mortality (c) 呼吸系統疾病患 Respiratory System (d) Mortality 惡性腫瘤患者之死 死亡率 (1/1000) 死亡率 (1/1000) 罹病者 Victim, 且為災區災民 + Patient 罹病者 Disaster,, 且為災區非災民 Non-victim+Patient 罹病者 Non-disaster, 且為非災區民眾 area + Patient 未罹病者之中部三縣民眾 Not Patient 99Q488Q4 00Q1 89Q100Q2 89Q200Q3 89Q3 00Q4 89Q4 01Q1 90Q1 01Q2 90Q2 01Q3 90Q3 01Q4 90Q4 02Q1 91Q1 01Q2 91Q2 01Q3 91Q3 01Q4 91Q4 88Q4 89Q1 89Q2 89Q3 89Q4 90Q1 90Q2 90Q3 90Q4 91Q1 91Q2 91Q3 91Q4 99Q4 00Q1 00Q2 00Q3 00Q4 01Q1 01Q2 01Q3 01Q4 02Q1 01Q2 01Q3 01Q4 Time 震災後至民國 91 年第四季 Time 震災後至民國 91 年第四季 Mortality Rates after ADR (c) 以 Mortality Rates after ADR (d) 以 死亡率 (1/1000) 死亡率 (1/1000) 99Q4 88Q4 00Q189Q1 00Q2 89Q2 00Q3 89Q300Q4 89Q401Q1 90Q101Q2 90Q201Q3 90Q301Q4 90Q4 02Q1 91Q101Q2 91Q2 01Q3 91Q301Q4 91Q4 88Q4 89Q1 89Q2 89Q3 89Q4 90Q1 90Q2 90Q3 90Q4 91Q1 91Q2 91Q3 91Q4 99Q4 00Q1 00Q2 00Q3 00Q4 01Q1 01Q2 01Q3 01Q4 02Q1 01Q2 01Q3 01Q4 Time 震災後至民國 91 年第四季 Time 震災後至民國 91 年第四季

31 Odds Ratio Table 5. Mortality Risk (Logistic Regression) Disaster Area All Causes All Causes Cardiovascular 921quake Victim Digestive System Cancer Respiratory System 1999Q Q Q *** 1.154*** 1.156** 1.265*** Q ** Q Q ** ** Q

32 Mortality Risk (Disaster + Victim) Ratio 非災區 災區領有震災卡的災民 Disaster area + Victim ( 所有樣本 ) Mortality Ratio 死亡率比值 Upper 死亡率比值信賴區間上界 Bound Lower 死亡率比值信賴區間下界 Bound Non-disaster area Ratio 死亡率比值死亡率比值信賴區間上界 Upper Bound 死亡率比值信賴區間下界 Lower Bound 非災區 災區其他非災民 Disaster area + Non-victim Mortality Ratio Non-disaster area ( 所有樣本 ) Age Age

33 Conclusions No differences in mortality risk between disaster and non-disaster areas before 921 earthquake; Quake victims have higher mortality risk after; Earthquake s influence continues for a while (e.g., quake victims have higher risk in 2000Q3) ; The middle-aged and elderly have higher death risk due to cardiovascular disease and digestive system disease for the earthquake victims.

34 Study Limitations The mortality risk of 1999Q4 is likely to be underestimated, since the issue time of earthquake card is Oct (Q4 covers only Nov. & Dec.). Control group (Chang-hua county) also has quake victims and 2,241 middle-aged & elderly with quake cards (0.86%). Include death records with NHI database The earthquake effect on the mortality risk for a longer period?

35 35