Download presentation
Presentation is loading. Please wait.
Published byScott Young Modified over 8 years ago
1
Assessing SES differences in life expectancy: Issues in using longitudinal data Elsie Pamuk, Kim Lochner, Nat Schenker, Van Parsons, Ellen Kramarow National Center for Health Statistics Centers for Disease Control and Prevention
2
Why is this important? Socioeconomic disparities Life expectancy Longitudinal data
3
Socioeconomic disparities Focus of health policy “Inequalities in income and education underlie many health disparities in the United States.” Healthy People 2010: Understanding and Improving Health
4
Life Expectancy Useful (and intuitive) measure for summarizing mortality rates across all ages Derived from a life table
5
Cohort life table for Swedish women born in 1890 Ageqxqx axax lxlx dxdx LxLx TxTx exex 00.0960.3100000960193663567792756.78 1-40.0721.5903996484345432558426461.77 5-90.0262.2839152184413539523883262.43 10-140.0192.5817311563404776482529359.04 15-190.0232.5801681807396315442051755.14 20-240.0262.7783612013387089402420151.35 25-290.0382.7763482924375058363711247.64 30-340.0222.4734241579362953326205444.43 35-390.0232.5718451632355189289910240.35 40-440.0242.4702131685346705254391236.23 45-490.0292.5685281972337737219720832.06 50-540.0342.6665562290327160185947127.94 55-590.0482.6642663117313786153231023.84 60-640.0682.7611494131296041121852519.93 65-690.1062.657018605227072392248416.18 70-740.1672.650965851323467165176112.79 75-790.2662.642453112811851584170919.82 80-840.3732.531172116211267552319337.44 85-890.5362.31955010487698751051785.38 90-940.6952.29063630327651353033.90 95-990.8432.027612327681276522.77 100-1040.9351.74344068128391.93 105-10911.028 0.98 110+..0000.
6
Period Life Table for the United Kingdom, 1990 AgePopdeathsMxqxqx axax dxdx lxlx LxLx TxTx exex <12,533200.00790.00780.178410000099294719869171.99 1-411,13010.00010.00040.53699216396793709939771.55 5-915,51920.00010.00060.56499180495742670260467.58 10-1416,40940.00020.00120.512199116495280620686262.62 15-1916,13390.00060.00280.527698996494289571158257.70 20-2421,482100.00050.00230.523098720493026521729352.85 25-2915,997220.00140.00690.567598490490765472426747.97 30-3416,026350.00220.01090.5106297816486422423350143.28 35-3919,800340.00170.00860.582796753481698374708038.73 40-4416,076390.00240.01210.5115795926476739326538134.04 45-4913,404590.00440.02180.5206394770468690278864229.43 50-5413,0271080.00830.04060.5376592706454120231995325.02 55-5910,0511360.01350.06540.5582088942430157186583220.98 60-6410,2201760.01720.08260.5686283121398451143567517.27 65-699,1903200.03480.16020.51221476259350763103722413.60 70-747,4274450.05990.26060.5166876404627851068646110.72 75-795,2314140.07910.33040.515645473581976794079518.61 80-852,8843550.12310.47060.514925317131212532102726.63 85+1,8403470.18861.00000.516788 89019 5.30
7
Longitudinal data Allows the calculation of life expectancies for groups defined by survey characteristics Eliminates numerator/denominator inconsistencies
8
Limits the age range & number of education groups that can be compared “This result tends to validate a specific concern about U.S. mortality estimates calculated from death certificate data. NCHS typically publishes U.S. mortality rates by education level for ages 25–64 because of concerns about the accuracy of death certificate education information at older ages ….”
9
Issues arising from using longitudinal data to estimate life expectancies
10
Data quality
11
How Records are Linked 11 NCHS Records SSN Name DoB Sex State of Birth Race State of Residence Marital Status Administrative Records Name DoB Sex State of Birth Race State of Residence Marital Status Non matchesPotential matches Scoring system, clerical review True matchesNon matches Linked Data File NCHS Records SSN Name DoB Sex State of Birth Race State of Residence Marital Status Administrative Records SSN Name DoB Sex State of Birth Race State of Residence Marital Status Non matchesPotential matches Scoring system, clerical review True matchesNon matches Linked Data File
12
Probabilistic Matching Procedure Missing identifying information from survey respondent ineligible for matching Ineligibility not random across groups
13
Percent of survey participants ineligible for NDI match: NHIS 1986-2004 survey years
14
Addressing insufficient information for matching: Ineligibility-adjusted weights Reweighting of matched respondents to be representative of civilian, non-institutionalized population Exclusion of problem groups No separate analysis of Hispanics
15
Generating appropriate measures of sampling variability
16
Person-year calculations for the denominators of age-specific mortality rates Hypothetical participants in a longitudinal study 1995-2000 with follow-up through 2003 Interviewed in 1997 at age 65, died at age 71 Age 65Age 66Age 67Age 68Age 69Age 70Age 71 1/6 1 year I year1 year 1/6 Date of interview Date of death 1-Jan-971-Jan-981-Jan-991-Jan-001-Jan-011-Jan-021-Jan-03 End of follow-up Interviewed in 2000 at age 68, no record of death Age 65Age 66Age 67Age 68Age 69Age 70Age 71 1/3 yrI year1 year1/3 yr Date of interview 1-Jan-971-Jan-981-Jan-991-Jan-001-Jan-011-Jan-021-Jan-03 End of follow-up
17
Life Table for men with less than a high school education NHIS 2000-2004 with mortality follow-up through 12/31/2006 Person- Ageyearsdeaths nMxnMx qxqx lxlx dxdx LxLx TxTx exex 25-29178338611770.00070.0033100000329499177472655647.27 30-34280129744500.00160.007999671789496382422737942.41 35-39328422478690.00240.0119988821178491467373099837.73 40-444090508221750.00540.0267977052613481991323953133.16 45-493810275277760.00730.0358950923404466949275754029.00 50-543262289411040.01260.0611916885600444439229059124.98 55-593303044651280.01970.0940860888088410218184615221.44 60-643448930687580.01990.0949779997406371482143593418.41 65-6936269321152630.03180.14727059410392326989106445215.08 70-7435207091585320.04500.2024602021218327055373746312.25 75-7931285412270890.07260.307248019147512032204669109.72 80-8422708252095920.09230.375033268124751351562636907.93 85-8912277951815660.14790.53982079411225759071285356.18 90+5431311240320.22841.00009569 41901526285.50 All data weighted using eligibility adjusted sample weights; closing value for the life table taken from 2000 vital statistics
18
Obtaining standard errors for life expectancy derived from longitudinal data Ideally, should take into account: Correlation within age-groups resulting from survey sampling design Correlation across age-groups resulting from respondents contributing to more than one age group
19
Case study of the sensitivity of life expectancy standard errors to study and sample design: Compared standard errors derived by Chiang method (traditional) Balanced Repeated Replication Hybrid methods: BRR & Chiang Taylor Series (SUDAAN proc RATIO) & Chiang
20
Comparison of standard errors Standard error as estimated by: E25Chiang Chiang / SUDAAN Chiang/ BRR BRR TOTAL53.130.0470.054 0.061 - Men50.670.0680.075 0.080 - non-Hispanic White51.280.0760.084 0.092 - < HS education46.670.2240.2420.2430.235 -Women55.400.0620.071 0.074 -non-Hispanic Black51.900.1840.2070.2080.210 - > HS education54.550.3660.4280.4270.435
21
Study Conclusions Traditional method (Chiang) consistently underestimate standard errors If balanced repeated replication procedure is impractical, Taylor Series (SUDAAN proc Ratio)/Chiang hybrid can yield reasonably accurate results for finer subgroups
22
Exclusion of the institutionalized population
23
Life expectancy at age 25 by sex and education level: NHIS/NDI linked mortality files, 1990-96* & 2000-06* MenWomen 1990-96*2000 -06*1990-96*2000 -06* Educatione2595% CIe25 e25 e25 <HS46.1 45.4 -48.247.346.4-48.252.351.6-53.152.151.1-53.2 HS/GED50.3 49.8 -52.251.651.0-52.256.456.0-56.857.757.1-58.2 Some college51.2 50.5 -53.352.651.9-53.357.857.2-58.558.958.3-59.6 College Grad+54.4 53.8 -57.556.856.1-57.558.858.0-59.561.160.3-61.9 Difference: College grad+ - <HS8.39.56.59.0 * 1990-94 with follow-up through 1996; 2000-04 with follow-up through 2006
24
Examination of the effect of excluding the facility dwelling elderly: Used MCBS data for facility dwelling beneficiaries for 1992-96/98 and 2000- 2004/06 Calculated death rates by sex and education level for ages 70-89 and combined with NHIS/NDI rates
25
Life expectancy at age 25 by sex and education level: NHIS/NDI linked mortality files, 1990-96 & 2000-06 combined with MCBS files, 1992-98 & 2000-06 MenWomen NHIS/NDI+ MCBS/NDINHIS/NDI+ MCBS/NDI Education 1990- 96* 2000- 06 1990- 96* 2000- 06 1990- 96* 2000- 06 1990- 96* 2000- 06 <HS46.147.3 45.446.2 52.352.1 51.149.9 HS/GED50.351.6 49.550.8 56.457.7 55.255.6 Some college51.252.6 51.151.7 57.858.9 56.856.6 College Grad+54.456.8 54.255.8 58.861.1 58.559.1 Difference: College grad+ - <HS 8.39.5 8.89.6 6.59.0 7.49.2 *NHIS/NDI 1990-94/96 combined with MCBS 1992-96/98
26
Using longitudinal data to examine SES differences in life expectancy Adds to our ability to routinely monitor SES differences in mortality Brings with it several methodological challenges
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.