邢唷��>� ���������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������欹�y� ��鮗bjbj�� v�遻遻篟+�������NN�!�!�!�!�!$�����!�!�!P"��"T�!�.�#LH$^$^$^$M%v�%$�%. . . . . . .$�0�m3r..-�!�%M%M%�%�%..�!�!^$^$�[.U)U)U)�%*�!^$�!^$.U)�%.U)U)�+�+^$���� 匰�!浲����%'"�+�-q.0�.�+�5G'�5�+�5�!�+�%�%U)�%�%�%�%�%....U)�%�%�%�.�%�%�%�%���������������������������������������������������������������������5�%�%�%�%�%�%�%�%�%N n : Appendix. Methodology for Estimating the Effect of Health Reform on Women抯 Insurance Coverage and Breast and Cervical Cancer Screening This technical appendix provides additional details about our methodology for estimating the number of people who are uninsured in 2014 after the implementation of health insurance expansions under the Patient Protection and Affordable Care Act of 2010 (ACA) and for estimating 2014 eligibility for the Situs Slot Gacor抯 National Breast and Cervical Cancer Early Detection Program (NBCCEDP). We focused on low-income women aged 18 to 64 years (the eligible age range for cervical cancer screening) or 40 to 64 (the eligible ages for breast cancer screening). Data The data used in the analysis come from the 2009 American Community Survey (ACS) Public Use Microdata Sample (PUMS) which includes observations on 2,979,656 people living in households in all counties and county equivalents in the 50 US states and the District of Columbia. (The PUMS sample also includes 83,238 people living in group quarters, but we dropped them from our analysis because family income is not available for this population.) These data represent a subset of all responses to the 2009 ACS: the survey as a whole aims to capture 2.5% of each state抯 population, but the PUMS includes only 1% of each state抯 population. Even so, the use of ACS data represents a major increase in the sample size and expected precision of state estimates over prior survey data such as the Current Population Survey. Most existing research on health insurance status draws from other surveys � particularly the Annual Social and Economic Supplement (ASEC) to the Current Population Survey (CPS). The ACS only began including a question about health insurance status in 2008. Despite its limited prior use as a data set for modeling the determinants of health insurance status, the ACS has several advantages over the ASEC-CPS for our purposes. Most importantly, with a sample size roughly 30 times that of the ASEC-CPS and a response rate of 96% (compared to 84% for the CPS), the ACS PUMS is a much larger data set than the ASEC-CPS. For this reason, the Census Bureau recommends using the ACS for studies examining state-level outcomes (1). In addition to its larger sample size, a major advantage of the ACS is that its health insurance status question is relatively easy to interpret. For each member of the household, the question simply asks if the individual currently has insurance through a) a current or former employer or union, b) direct purchase, c) Medicare, d) Medicaid or other government assistance, e) TRICARE or other military, f) the Veterans Administration (VA), or g) Indian Health Service. For our purposes of evaluating insurance status, we treated TRICARE or other military and VA coverage as private insurance and Indian Health Services as uninsured (if the only coverage a person had). This method of asking about health insurance results in transparent data that reflect current insurance status and facilitates the identification of insurance status for respondents� household members. A study for the Census Bureau evaluated the health insurance data from the ACS in comparison to data from the ASEC-CPS and the National Health Interview Survey (NHIS) and found a high level of consistency for the ACS estimates (2). We downloaded the ACS PUMS data from the Census Bureau website and read them into SAS. We transferred files to Stata using StatTransfer and performed all data recoding and calculations in Stata. Because the PUMS data is designed for public use, the Census bureau completes a significant amount of data cleaning and imputation while preparing the data. The Census Bureau edits survey responses using hot-deck and cold-deck allocation to correct for internal inconsistencies and/or item non-response prior to PUMS release. Research Design and Methods Our underlying approach was to first estimate determinants of insurance status in Massachusetts in 2009, based on the ACS data. Massachusetts implemented a state-level health insurance reform in 2006 and many of its features were used in developing the national reform, so it is reasonable to use Massachusetts results to understand effects at the national level. We developed a Massachusetts health insurance model to simulate health reform at the national level, with some adjustments. This approach assumes that individual-level determinants of health insurance for Massachusetts residents in 2009 will be similar to those for all US residents after ACA expansions are implemented in 2014. Rather than run separate models predicting Medicaid and private health insurance coverage, we parsimoniously estimated a model on being insured or uninsured, regardless of source of coverage. It is known that Medicaid coverage influences private health insurance coverage and vice versa. The availability of coverage for breast and cervical cancer screening coverage will be the same whether a person is covered by private insurance or Medicaid under the ACA. We estimated a multivariate logit model of the determinants of health insurance status for adults aged 18 to 64 in Massachusetts in 2009, using the ACS sample weights. (A similar model was developed for men, some of whom are eligible for Situs Slot Gacor抯 colorectal cancer screening program, but those data are not reported in this article. The results of that model and estimates of male uninsured rates are available from the authors.) Our model includes race/ethnicity, marital status, having a child, employment status, industry, income, citizenship status, disability and education. We selected variables based on prior research about the determinants of health insurance coverage (3). To capture information about the type of industry in which people are employed, a known determinant of health insurance status, we created high, average, and low health insurance industry categories. High insurance industries include utilities, finance and insurance, management of companies and enterprises, educational services, and public administration; all have uninsurance rates below 10%. Low insurance industries include agriculture, forestry, fishing, and hunting, construction, administrative support and waste management and remediation, accommodation and food services, and other services (except public administration); each of these has uninsurance rates greater than 25%. While the ACS data do not indicate which individuals were offered employer-sponsored insurance or the level of premiums charged, our industry groupings provide some control for the likelihood of health insurance offers and premium levels, on the basis of the industry in which each individual is employed (if any). Variable selection was based on theoretical reasons and predictive capabilities rather than parsimony; hence, some insignificant variables remain in the model. The results of the multivariate logit model for predicting whether nonelderly women in Massachusetts are uninsured are shown in Table A1. We express the results as odds ratios and 95% confidence intervals, although the results were operationalized as coefficients and predicted probabilities. The following factors are associated with a significantly lower probability of being uninsured: being African American, being below poverty, higher income, working full-time full-year, being married, having a child, being disabled, receiving public income (eg, welfare), being a college graduate, being a student, and working in a high-insurance industry. Factors associated with increased uninsurance include being a noncitizen immigrant, working in a low-insurance industry, being over 39 and Hispanic (interaction) and being black and poor (interaction). Table A2 illustrates the performance of the logit model by comparing predicted uninsured rates for subgroups of Massachusetts women to actual uninsured rates for these same groups. The regression coefficients of the Massachusetts equation were applied to each individual in the ACS-PUMS sample and converted into individual-level probabilities of being uninsured for each person in every US state. We recognized that adjustments are needed however, because 1) people in other states might not behave like those in Massachusetts, even after controlling for individual characteristics, 2) institutions and markets (eg, Medicaid agencies, insurance markets) in other states may behave differently, and 3) the Massachusetts reforms are very similar to, but not identical, to those of the ACA (although the final forms of ACA implementation are yet to be determined). Our first adjustment concerned citizenship status, a major determinants of insurance coverage (4). Massachusetts is one of a minority of states providing state-funded Medicaid coverage to recent legal immigrant adults, so noncitizen immigrants have a higher probability of being insured than in states without such coverage. To account for the lower likelihood of insurance coverage among immigrants in less generous states, we doubled the coefficient for noncitizen status in those states not offering Medicaid to recent immigrants. These states were all except California, Connecticut, Delaware, the District of Columbia, Hawaii, Maine, Massachusetts, Minnesota, Nebraska, New York, and Washington. Data on Medicaid policy on immigrants were obtained from the National Immigration Law Center (5). To adjust state estimates to better reflect state-specific characteristics, we ran a logit model on the entire nonelderly female ACS PUMS sample (as opposed to only those in Massachusetts) using the same variables as in the Massachusetts-only model, but also adding dummy variables for each state and the District of Columbia, except Massachusetts. The coefficients for each state dummy variable capture the difference in likelihood of being uninsured due to residing in a given state relative to Massachusetts, after all individual-level characteristics specified in the regression are controlled. These state 揻ixed effects� thus measure variation in state-related insurance outcomes due to omitted variables, such as state health insurance market differences or other sociodemographic factors. Next, we calibrated our model to correspond with other national estimates of the impact of the ACA. Based on estimates of the Congressional Budget Office (CBO)(6) and the Urban Institute (7), we inferred an estimate in the range of 9.6% to 10.0% uninsured for nonelderly adults (both men and women) of all incomes at the national level after reform. We calibrated our estimates to correspond to this target range using a partial adjustment of the results from the state fixed-effects estimates. Because they are derived using pre-health reform data, to use the entire fixed effect coefficient in estimating an individual抯 likelihood of being uninsured would produce estimates that return the state抯 2009 level of uninsurance. Our model assumes that there is a partial adjustment from the state抯 current level of insurance coverage toward the Massachusetts-based estimates. This assumption is based on the facts that health reform increases insurance coverage by creating minimum national standards but still offers states flexibility in policy design and implementation. We estimated the partial adjustment using a fraction (one-sixth) of the fixed effects coefficient, with the level corresponding to our calibration target. We updated and aged our estimates from 2009 to 2014, using the Census Bureau抯 national level projections of the age and size of the population (8). We adjusted the ACS-PUMS weights by factors associated with the expected change in the population by year of age. After deriving a predicted probability of being uninsured for each individual in the data set, we computed mean predicted probabilities of being uninsured for each state using individual-level weights adjusted to reflect population growth between 2009 and 2014. Finally, we computed state-specific estimates of the number of low-income women aged 18 to 64 where 搇ow income� is defined using each state抯 income criteria for the NBCCEDP program. We also produced estimates of the number of low-income women aged 40 to 64. The following income criteria applied: 185% FPL or below (Oklahoma), 200% FPL or below (Alabama, Arkansas, California, Connecticut, Florida, Georgia, Idaho, Indiana, Missouri, Montana, North Dakota, Ohio, South Carolina, South Dakota, Texas, Virginia, West Virginia), 226% FPL or below (Kansas, Nebraska), 250% FPL or below (all other states). After deriving the number of people in each income group for each state, we used the predicted probabilities of being uninsured derived above to estimate state-by-state uninsured counts and rates for low income women. These counts of uninsured correspond to expected eligibility for the NBCCEDP eligibility groups: cervical cancer screening (women ages 18�64) and breast cancer screening (women ages 40�64). (Note: Situs Slot Gacor is considering increasing the target age for cervical cancer screening to 21�64, based on recent changes in the US Preventive Services Task Force recommended age range.) To derive estimates of the characteristics of the uninsured in 2014, we used the demographic data in the ACS, our 2014 expected population weights, and our predicted probabilities of being uninsured. First, we estimated the expected size of different population subgroups in 2014 using the ACS data and our 2014 weights (eg, Hispanic women, women without a high school degree or GED). We next used the predicted probability of being uninsured (assigned to all people in the ACS sample) to estimate uninsured rates among population subgroups, and subsequently, the number of uninsured in each population subgroup. Using the numbers of uninsured women in population subgroups, we estimated the percent of all uninsured women who were Hispanic, black, white, without a high school degree or GED, etc. Limitations of This Approach Our estimation approach differs significantly from those of the CBO (9) and the Urban Institute (10). Those models were designed as overall health policy simulation tools to estimate national-level effects under a wide range of assumptions for alternative versions of national health reform policies. They are complex simulations based on an amalgam of data sources, particularly the ASEC-CPS and the Medical Expenditure Panel Survey, and entail a large number of models and assumptions about behavioral responses to different policies (11). While these models are extraordinarily useful in estimating budget and other impacts of different policy choices, it has long been recognized that different models, and very slight differences in assumptions made, can have profound differences in the results, so that different models can yield very divergent results (12). Our model used a much stronger sampling base and a simple, transparent set of assumptions. We used the ACS, which is a stronger base for state-level estimates because of its larger sample frame and more straightforward questions about health insurance status. And we assumed that the Massachusetts model can serve as a template for national estimates, with adjustments as we describe. To try to fit the national estimates, we � like the Urban Institute � calibrated our results to correspond approximately with the CBOs, so that everyone is working with similar national assumptions. Like any forecasting model, our model is subject to flaws because we assumed that future behaviors and policies can be predicted based on behaviors and policies of the past. As noted earlier, our estimates correspond only to the non-institutionalized population in the 50 states and the District of Columbia. Although we made adjustments for national population growth and aging, we assumed that state patterns are similar to those projected for the nation. Our models also assumed that other characteristics in 2014, such as employment, income, and education levels, are similar to those of 2009. In 2009, the country was in a recession and unemployment levels were relatively high. If the economy strengthens by 2014, as we hope it will, uninsurance rates might be lower than estimated. On the other hand, if health care costs rise faster than anticipated (resulting a decreased propensity of individuals and firms to obtain health insurance), then uninsurance rates might be higher than estimated. Finally, if the ACA is not implemented in a fashion consistent with the expectations of the CBO, then the number of uninsured people may vary. While these limitations introduce significant uncertainty into our estimates, the sheer magnitude of the changes reported in our analyses imply that major changes are in store for the number and distribution of uninsured women in the United States after the implementation of health reform. References 1. US Census Bureau. About health insurance. Washington (DC). HYPERLINK "http://www.census.gov/hhes/www/hlthins/about/index.html"http://www.census.gov/hhes/www/hlthins/about/index.html. Accessed April 26, 2011. 2. Turner J, Boudreaux M, Lynch V. A preliminary evaluation of health insurance coverage in the 2009 American Community Survey. Washington (DC): Health Policy Center, Urban Institute; 2009. 3. Fronstin P. Sources of health insurance and characteristics of the uninsured: analysis of the March 2010 Current Population Survey. EBRI Issue Brief No. 347. Washington (DC): Employee Benefit Research Institute; 2010. 4. Ku L, Matani S. Left out: immigrants� access to health care and insurance. Health Aff (Millwood) 2001;20(1):247�56. 5. Table: medical assistance programs for immigrants in various states. HYPERLINK "http://www.nilc.org/pubs/guideupdates/med-services-for-imms-in-states-2010-07-28.pdf"http://www.nilc.org/pubs/guideupdates/med-services-for-imms-in-states-2010-07-28.pdf. Accessed May 23, 2011. 6. Congressional Budget Office. Final cost estimate, March 20, 2010. In: Selected CBO publications related to health care legislation, 2009�2010. Washington (DC): Congressional Budget Office; 2010. p. 3. 7. Buettgens M, Holahan J, Carrol C. Health reform across the states: increased insurance coverage and federal spending on the exchanges and Medicaid. Washington (DC): Urban Institute; 2011. 8. U.S. Census Bureau. US population projections. HYPERLINK "http://www.census.gov/population/www/projections/summarytables.html"http://www.census.gov/population/www/projections/summarytables.html. Accessed April 26, 2011. 9. Congressional Budget Office. CBO抯 health simulation model: a technical description. Washington (DC): Congressional Budget Office; 2007. 10. The Urban Institute抯 health microsimulation capabilities. Washington (DC): Urban Institute; 2010.  HYPERLINK "http://www.urban.org/uploadedpdf/412154-Health-Microsimulation-Capabilities.pdf" http://www.urban.org/uploadedpdf/412154-Health-Microsimulation-Capabilities.pdf. Accessed August 29, 2012. 11. Bilheimer L, Reischauer R. Confessions of the estimators: numbers and health reform. Health Aff (Millwood) 1995;14(1):37�55. 12. Nichols L. Perspectives: numerical estimates and the policy debate. Health Aff (Millwood) 1995;14(1):56�9. Table A1. Logit Results for Determinants of Female Uninsured Status in Massachusetts, Ages 18 to 64, 2009 American Community Surveya VariableOdds Ratio (95% Confidence Interval)Age1.02 (0.97�1.08)Age squared1.00 (1.00�1.00)Hispanic1.02 (0.70�1.47)Non-Hispanic African American0.44 (0.20�0.98)Other race1.21 (0.87�1.68)Noncitizen2.31 (1.77�3.01)Income as % of poverty0.67 (0.61�0.73)Income below poverty 0.52 (0.37�0.71)Works full-time full-year0.51 (0.38�0.70)Works but not full-time full-year0.79 (0.60�1.04)Self employed0.79 (0.51�1.23)Not in the labor force0.89 (0.64�1.26)Married0.59 (0.46�0.76)Has child under age 7 y at home0.48 (0.35�0.66)Has child aged 7 y to17 y at home0.56 (0.43�0.72)Disabled0.65 (0.44�0.96)Receives public income0.20 (0.12�0.35)Graduated from college0.75 (0.58�0.97)High school dropout1.18 (0.89�1.58)Current student0.48 (0.34�0.68)Works in high-insurance industry0.63 (0.42�0.93)Works in low-insurance industry1.53 (1.20�1.95)Over age 39/Hispanic interaction1.73 (1.05�2.88)Black/income interaction1.35 (1.06�1.73)Black/federal poverty level interaction1.94 (0.71�5.30)a N = 20,548; model �225 = 625.8 Table A2. Comparison of Actual Uninsured Rates to Predicted Uninsured Rates for Massachusetts Female Subgroups, 2009 PopulationUninsured Actual RatePredicted Uninsured RateWomen aged 18 640.04310.0431Hispanics aged 18 640.09310.0931Blacks aged 18-640.06240.0624 Over age 390.03170.0324Hispanic over age390.09620.0962Black over age 390.05070.0514 Income <201% poverty, aged18-640.08770.0936Hispanic, income <201% poverty, 18�640.11870.1139Black, income <201% poverty, aged 18�640.07030.0805Income <201% poverty, over age 390.07800.0782Hispanic, income <201% poverty, over age 390.12110.1187Black, income <201% poverty, over age 390.06930.0727      PAGE \* MERGEFORMAT 1 &��������� - l � � � * + , - 4 6 K O | � � � � � s � � � � � � � � � �   3 a o p � � � � � �  1 2 3 8 W  VW��黩赍彷彷豳豳豳召奄嵬召唾召召嵘梁颜讯投讯讯讯讯讯讯讯秃埠埠埠餐胀翰烧赏h�(h�� h7}�h齱�h��h齱�5�h齱�h��h齓�hv%}hEUh�3�h7}� h%Dk5�h%Dkh%Dk5� hv%}5�h��h2y}5�F�� � �1=Y��#'�)�,0�4�5�6A9�;�>�>4B|DJ������������������������� d�ゐgd3_� d�ゐgd�h d�ゐgdw\� d�ゐgd���� ��Yu������+8uy������25MP~���(45+-./01@x}������'CX�����<=Ygpsvx��e���豇豇豇痿祠祠祠祠痂彷彷彷徼麴麴麴豇嬴音焖裘扛扛盔� hP h7}�h7}�h��h齱�5� h齱�h齱� hJ�h齱�hw\�h36K h麿�h3_�h3_�hv%}h2i�h齱�h��h�(G������-O]mu�������j�������/0��������=Id�������$%&����������� &(~€���繇繇繇骓繇忭酎繇艮谵洲廪肄阝掴尬摁礅硎糁羝羝翕硪卖眙碇眙眙眙眸绝豇壶�hw\�h�(h齱�h2i�h漺�hG@h36Kh謌�h�1�h�hhv%} h@ �h7}� hP h7}�h7}�h��hEUI������Mo����  + 2 � � � !!!!5!$"3"p"x"�"�"�"�"�"�"##A#T#X#w#z#�#�#�#�#�#�#�#7$N$c$v$�$�$%:%|%�%�%�%�%F&e&�&�&'''麴祓翳翡綮翡綮羿綮糗翡綮綮剀仂谚淹衍阳茗禅溧气气栖凄缕帏绝舴 hP h齱�h�3�hA1 h鬡�h齱�h�� hP h7}�hs�hv%}h��h2i�h謌�h�(hEUh7}�h齱�h漺�H'J'T'V'_'�'�'�'�'�'�'�'((-(.(](v(�(�(�(�(�(�()) )�)�)�)�)�)�)�)�)�)�)�)"*$*%*&*'*(*j*�*�*�* + +++"+0+D+U+o+w+x+�+�+�+�+�+�,�,�,�,�,�,�,5-6-7-8-聃睃觞殄揄挹挹挹磙谵陧挹挹挹挹褛挹逯遐遛谵亿尬椅椅椅掊挹奘禄驶叔皱掊肄 hz?h9L�hz?h9L�>*h9L�h��hC�hw\�h7}� hP h7}�h��h�(hA1h謌� hP h齱�h齱�J8-E-S-W-e-o-v-�-�-+.3.7.C.D.[.y.�.�.�.�.-/;/=/I/J/W/X/^/e/s/�/�/�/�/�/�/�/�/00 0000"0$0:0H0g0|0�0�0�0�0�0�0�0�0�0�0�0�0�0�0�0�0�0�01111.1S1X1Y1_1a1m1n1�1�1�1�1�1�12顸觞觞觞蹴觞觞觞觞觞蹰觞觞觞觞睃觞蹴礤觞礤睃狨眭聃聃聃冱眭眭聃耥躐hC�h$bGh/`hw\�h `�h��hA1 hP h7}�h7}�V2�2�2�2�2^3n3o34454@4A4D4E4Q4v4y4�4�4�4�4�4�4�4�4a5c5d5e5f5g5p5r5�586B6C6a6�6�6�6�6�6�6�6�6�6!73777;7=7�7�7�7�7�7�7�7�7�7�7�7�7�7 8 8�8�8@9A9�9�9�9�9::`:z:聃聃睐貔轾殄踽轳轳轳蓠橘轳轾踬僬颜颜僬僬驼颜颜遒橘樯橥刨捧捧捧h=ph獸�h^,hA1h `�h3_�h/`h$bGh��h7}�h謌�h��hC� hP h7}�Nz:�:�:�:�:�:�:�:�:�:�:�:�:0;1;6;U;�;�;�;�;�;�;"<�c<��<��<��<�====�>�>�>????/?2?3?4?5?6?7?�?�?�?�?@@鏎闌隌霡+B.B/B0B7BBB`BnBzB旴朆〣狟7C8CECFC朇礐轈逤鵆栲栲栲翳滂噼湄孕绦匀绦鹪鹪鹪刑性腥性性酿孕责孕孕孕贼�h8 �hw\�h/`h3_�h麿�h3_�h麿�5�hhj�hLcah獸�h��h��hA1hC�h=pM鵆鶦DDD:D荄菵翬FF_F`F禙誇JHiH貶鳫↖獻釯颕JJ&J(J)J*J;JRJdJeJ↗㎎郕酛釰麶齁�J K"K/KV?VEVFV]VaVbVgVhVnVoVyV~V哣奦媀怴慥梀榁琕癡盫禫稸絍綱蜼襐覸豓賄遃郪鞻頥WWW W WWWW W3W7W8W箐箐萏咎咎咎潴漭摊箐箐萏潴漭叹摊箐箐萏潴潴漭摊箐箐萏咎潴漭叹摊�h"+`h�9�B*CJaJphhU�B*CJ\乤Jph h"+`h[WB*CJ\乤Jph h)�h[Wh"+`h[WB*CJaJphhU�B*CJaJphE鯱鱑V*V|sg $$Ifa$gdU� $IfgdU��kdp$$If杔���0���$f f t��0������� %6��������������4�4� lB�a�ytU�*V+V4VEV|sg $$Ifa$gdU� $Ifgd&��kd $$If杔���0���$f f t��0������� %6��������������4�4� lB�a�ytU�EVFV]VnV|sg $$Ifa$gdU� $Ifgd&��kd� $$If杔���0���$f f t��0������� %6��������������4�4� lB�a�ytU�nVoV哣梀|sg $$Ifa$gdU� $Ifgd&��kd8 $$If杔���0���$f f t��0������� %6��������������4�4� lB�a�ytU�梀榁琕絍|sg $$Ifa$gdU� $Ifgd&��kd� $$If杔���0���$f f t��0������� %6��������������4�4� lB�a�ytU�絍綱蜼遃|sg $$Ifa$gdU� $Ifgd&��kdh $$If杔���0���$f f t��0������� %6��������������4�4� lB�a�ytU�遃郪WW|sg $$Ifa$gdU� $Ifgd&��kd $$If杔���0���$f f t��0������� %6��������������4�4� lB�a�ytU�WW3WDW|sg $$Ifa$gdU� $Ifgd&��kd� $$If杔���0���$f f t��0������� %6��������������4�4� lB�a�ytU�8W=W>WDWEWIWNWfWjWkWpWqWwWxW慦昗朩沇淲網薟蟇蠾誛諻躓軼轜遅閃霿XXXXXX.X痄疠叹摊漯漯萏痄疠咎痄痄疠Б懢懢m懢dhU�5丆JaJ#hU�hU�B*CJH*\乤Jph#hU�hU�B*CJH*\乤Jph h"+`hU�B*CJ\乤Jph hU�H*hU�hU�H*h"+`h�9�B*CJaJphhU�B*CJ\乤Jph h"+`h[WB*CJ\乤Jph h)�h[WhU�B*CJaJphh"+`h[WB*CJaJph'DWEWfWwW|sg $$Ifa$gdU� $Ifgd&��kd0 $$If杔���0���$f f t��0������� %6��������������4�4� lB�a�ytU�wWxW慦|sg $$Ifa$gdU� $Ifgd&��kd� $$If杔���0���$f f t��0������� %6��������������4�4� lB�a�ytU�薟躓|sg $$Ifa$gdU� $IfgdU��kd`$$If杔���0���$f f t��0������� %6��������������4�4� lB�a�ytU�躓軼XXYYJY|Y|wwwnbb $$Ifa$gd&� $Ifgd&�gd謌��kd�$$If杔���0���$f f t��0������� %6��������������4�4� lB�a�ytU�.XDXYYY|Y~Y奩擸榊歒燳綴釿鑉Z ZZZ$Z%Z&Z*Z.Z1Z@ZNZQZTZcZnZrZuZ刏匷孼抁橺沍碯絑腪蘘箐依癄弫弫彔亸爮亸爮亸亸爮亸爮亸爮p弍弫彔弍� h&�h� �B*CJ\乤JphhU�B*CJ\乤Jph h&�h謌�B*CJ\乤Jphh&�h謌�B*CJaJph h&�h謌�5丅*CJaJph#h&�h謌�5丅*CJ\乤Jph#h謌�h謌�5丅*CJ\乤Jphh謌�5丅*CJ\乤Jphh謌�h謌�5丆JaJ,|Y~Y燳甕糦^UII $$Ifa$gdU� $IfgdU��kd�$$If杔�� 諪���$�� R t��0�������6�� ���� ���� ���� ���4�4� la�p�������yt粅�糦綴鑉鯵ZocTT$d�$Ifa$gdU� d�$Ifgd&��kdF$$If杔�斷�諪���$�� R t��0�������6�� ���� ���� ���� ���4�4� la�ytU�ZZZZ#ZocTT$d�$Ifa$gdU� d�$Ifgd&��kd�$$If杔�斷�諪���$�� R t��0�������6�� ���� ���� ���� ���4�4� la�ytU�#Z$Z1Z8Z?ZocTT$d�$Ifa$gdU� d�$IfgdU��kdp$$If杔�斷�諪���$�� R t��0�������6�� ���� ���� ���� ���4�4� la�ytU�?Z@ZTZ[ZbZocTT$d�$Ifa$gdU� d�$Ifgd&��kd$$If杔�斷�諪���$�� R t��0�������6�� ���� ���� ���� ���4�4� la�ytU�bZcZuZ|Z僙ocTT$d�$Ifa$gdU� d�$Ifgd&��kd�$$If杔�斷�諪���$�� R t��0�������6�� ���� ���� ���� ���4�4� la�ytU�僙刏琙砕ocTT$d�$Ifa$gdU� d�$Ifgd&��kd/$$If杔�斷�諪���$�� R t��0�������6�� ���� ���� ���� ���4�4� la�ytU�砕碯踆鈀閆ocTT$d�$Ifa$gdU� d�$IfgdU��kd�$$If杔�斷�諪���$�� R t��0�������6�� ���� ���� ���� ���4�4� la�ytU�蘘覼譠賈踆闦[[ [[[["[6[=[A[D[S[\[c[k[r[y[}[€[廩玔瞇禰筟萚蒣蔥蘙蚚蟍衃襕覽誟諿韀頪颷餥骩鬧鮗镛修溜扌扌蘖镛修赁镛镛修溜扌蘖翰疅敓嚐儺�h�@f  G^Title�,P�O伣m$-@�B* CJ4KHOJQJaJ4mHph6]sHtHV䁖�1V  G^ Title Char)@�B* CJ4KHOJPJQJ^JaJ4ph6]&& G1�pTOC 1D@D G1�pTOC 2d^勡CJOJQJaJD@D G1�pTOC 3d^劯CJOJQJaJF"@F  G^Captionと5B*CJ\aJphO伣b﨩b  G^p TOC Heading$$む@& a$ B*CJOJQJ\^JaJph6_�D䁖��D  G^ No Spacing_HaJmH sH tH P��P  G^ Colorful List - Accent 1^勑:U`��: 7}�0 Hyperlink>*B*^Jph�D+@�D 7}�0 Endnote TextPJmHsHtHF䁖��F 7}�0Endnote Text Char CJPJaJB*`��B 7}�Endnote ReferenceH*^J*W`��* 7}�`Strong5乗乺﨩�r 7}�� Light List1�:V �0��j�.@�€�佒傊j�.�€�佒傊j�.�€�佒傊j�囍 ��4�4( f�d���f�d���5吺5\吺5\吺5\吺5B*\ph���>@> "7}�0 Footnote Text!CJaJ<���!<� !7}�0Footnote Text CharD&`�1D 7}�0Footnote ReferenceH*^J�﨩�C� 麿��Light Shading1G:V$�0����j�;@佒��������傊��������冎��������勚��������囍 �览�j�;佒��������傊��������冎��������勚��������囍 �览�j諨�€�佒��������傊��������冎��������勚��������j諨�€�佒��������傊��������冎��������勚���������4�4($f�d���f�d���5B*ph吺5\吺5\吺5\吺5\0䁖�Q0 麿�0 cit-title3^J@@b@ '��0Header &�H�$ mHsHtH6䁖�q6 &��0 Header CharCJaJ@ @�@ )��0Footer (�H�$ mHsHtH6䁖��6 (��0 Footer CharCJaJP橜�P +��0 Balloon Text*CJOJQJaJmHsHtHN䁖��N *��0Balloon Text CharCJOJQJ^JaJB'`��B ��0Comment ReferenceCJaJ<�@�<� .��0 Comment Text-CJaJ:���: -��0Comment Text CharLj@��L 0��0Comment Subject/5乗乵HsHtHF䁖�F /��0Comment Subject Char5乗丗V`�F 3_�0FollowedHyperlink >*B* ph€€j��#j [W� Table Grid7:V2�0������2^䁖�1^ 9L�bib_organization&CJehfH�q� ��r��N䁖�AN 9L�bib_year&CJehfH�q� ���r��4䁖�Q4 za�bib_urlCJehfH€PK!檗�[Content_Types].xml瑧薔�0E鱄鼉�-J湶@%閭菐洽|廊�$韶钵UL襎B� l,�3鳛;鉹�得槣B+$�G]ミ7O侪V墎$�┇最� !)O赹齬虲$駓@摪磔�/瓂H*�橊劥�)戅祶鬟粖譛Db俙}"譹蹕擩讞枻肵^�)I`n蘀�紛p)�杵li筕[]�1M<斷绒彥O蠵擊6r�=瘔抸纆b營g吜u崘S賓b嘱€O缽嗕掘肦罝郢櫉反qu 痝嫎Z岸串o~俸lAp發x妏T0�+[}`j纂鹾絲A�帮儲V�2虵蒳朄鰍瀡分�5\|夻蕼汰Nвle瞂�ds趈cs倭惻7琊嵨f坊赅 肉W琊�+唻7爤唁`�陲g� 葮稠J�雷j|唫h(�驞-姷咩� dX�﹊J曝�(钼x$(� �:隶;渌�!� I_蠺到S 1w犍鳢�?E��?勉?ZB为m渼錟/魁煜��?瀪篁�誼Y�'奎鼀5襣&螊/燑鲮蓩�>籊餗丟e鴲艱�眢3Vq%'#q��域娡$�8翚K秊�敉)f檞9:牡�澹 x}r�x墘�渨⒇顁�:\TZaG�*檡8I耲鎎R祈c|X呕�强絀 u3KG駈D1�NIB襰鼆� 眍R曦u楘侹>V�.EL+M2�#'歠嫸i ~橵� l硔u8z�篐� �*�鏄�:�(W�鈽� ~J攘T鴈\O*餿HG絸HY垫�}KN吡P�*菥甩眿�T鸭�9/#辐A7聁Z��$*c?��韖U咤n嗚w�N蝴%幓O�穒鑸4 =3炯N�績)cbJ u�4峦(Tn酸� 7斒_?異鹠-贈皗U逄鰤B�w�襷消╪湐鞘�"Z x娬J躗氺p;嫇熟 劐辿/�<�P;檸,)''K蠶踜5棝騫邛苝N喦8疜軬b耬摨� 鸖撡d洯\17 阷鮝�>ОSR!�枒 3晠K4'+�r蘻Q TT3I辈琉疘vt]K芻猥渤K#趘�5+D�眽厍鄜獱O@%\w槉燺鄋N[跮古9K候崢临q桃g錠炆n R!儁+�篣蕁�;�*&�/H時�蟃邀 �>��>\ 宼Υ=.T摹 �觖S; Z鄜�!ㄠ傏�銹��9gi槾咰ぺ�!�# B鰻,欒;匵=刍,I�2UW�9$l╧嗒捋=Aj挄�;顊朅79鍇s*Y摈�;浱爺[嘙C撣�県f华]o栫{oY=1k�yV骋V惺隐5E8鏥k+譁扑蚛8疴计0X4D)�!!��?*|f縱� u洒《"鴛A谸T_矋��帬q矁6�4)k诂u襐�7� 顃 �'尛%;嬁蟟膦9s�9箈懫�,熵趲-45x鰀娐�8?�菢蟙�/Y|t�� &LIL饾J`琛& �-G硉�/��PK! 褠煻'theme/theme/_rels/themeManager.xml.rels剰M �0匃倃oo雍�&輬协�勪5 6?$Q祉 �,.嘺緳i粭澤c2�1h�:闀q毩m胳嶡RN壻;d癭値o7�g慘(M&$R(.1榬'J摐袏T鶂�8V�"&A然蠬鱱}狇�|�$絙{�朠�除8塯/]As賲(⑵锑#洩L蔥汉倪��PK-!檗�[Content_Types].xmlPK-!ブх�6 0_rels/.relsPK-!ky���theme/theme/themeManager.xmlPK-!0軨)���theme/theme/theme1.xmlPK-! 褠煻'� theme/theme/_rels/themeManager.xml.relsPK]� �鍾�����;����� *********-���'8-2z:鵆籏郟裇烼閁8W.X蘘鮗.012345679:;BKTYbJ鱏 T+TFTvT揟癟賂U-UaU乁猆腢鯱*VEVnV梀絍遃WDWwW躓|Y糦Z#Z?ZbZ僙砕閆![R[嶽荹誟鮗/8<�=>?@ACDEFGHIJLMNOPQRSUVWXZ[\]^_`acdefghdB˙郆WE窫 F銰4HxH嘔錓5J鍾X�晫X�晫X�晫X�€ $&-!�晙餖�# � 喠帕@����€€€�饞��0�( � �養 �S ���� ?��� _Hlt335058148 _Hlt335058149 _Hlt335058877 _Hlt335058878菶菶颕颕鏡@@@@蒃蒃餓餓鏡6?� � � � 8:��� � �4�4,B2B綜艭鞤餌鯢�FG GGGAIPIVJ_J睯礘"K%K罧菿篟糝絉縍繰翿肦臨芌銻鏡筊篟芌酭鏡&B;BBD\D鍰EFGF蒅銰擧碒嘔6J蒏鳮麷,L5LMM.MPM侻橫臡錗鱉N郚OO3O薕�O癙罰萈蜳蠵錚霵騊鬚 Q QQQ(Q)QKQLQRQTQlQmQ淨漄襋観 R RRR;R4OS<�WW<�5!>肠颁>搕>骋蔼�蔼/皑础�6贵.镑骋$产骋�J�DJ�0H�za�漺�麿�謌�s�&}�8 ��.�U��9���齓��3�3_���0�I-��3��� `�&�-F��"��1���G%�C��.�6d�{j���G1�I�篟糝�@€疨疨疨疨@劺{(馩馪鍾�@�X��@��Unknown������������ G��*郃x� �Times New Roman5�€Symbol3.� �*郈x� �Arial7���@�Cambria7.����@ �Calibri5.� �.醄`�)�Tahoma?=� �*郈x� �Courier New;�€WingdingsA���$B�Cambria Math"q�鹦�h浬 G浬 GX bF *悥X bF *�!�����亖0怰怰2僎�HP � !?����������������������2y}2!xx� ���}Methodology for Estimating the Effect of Health Reform on Women s Insurance Coverage and Breast and Cervical Cancer ScreeningRPreventing Chronic Disease, breast and cervical cancer screening, health insurance(cancer, health insurance, women's health Alice Levy#Immoor, Kristen (CDC/OD/OADC) (CTR) �鄥燆鵒h珣+'迟0l��t��� � ( 4 @LT\d�€Methodology for Estimating the Effect of Health Reform on Women抯 Insurance Coverage and Breast and Cervical Cancer ScreeningTPreventing Chronic Disease, breast and cervical cancer screening, health insurance Alice Levy,cancer, health insurance, women's health Normal.dotm$Immoor, Kristen (Situs Slot Gacor/OD/OADC) (CTR)2Microsoft Office Word@@*Bq!浲@*Bq!浲 X bF�胀諟.摋+,D胀諟.摋+,�€ hp���� ���� � b� George Washington University�*怰 ~Methodology for Estimating the Effect of Health Reform on Women抯 Insurance Coverage and Breast and Cervical Cancer Screening Title�(QYy _PID_HLINKS Language�A5s Phttp://www.urban.org/uploadedpdf/412154-Health-Microsimulation-Capabilities.pdf4.Dhttp://www.census.gov/population/www/projections/summarytables.htmlKUhttp://www.nilc.org/pubs/guideupdates/med-services-for-imms-in-states-2010-07-28.pdfr*8http://www.census.gov/hhes/www/hlthins/about/index.htmlEnglish  !"#$%&'()*+,-./0123456789:;<�=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghi���klmnopqrstu���wxyz{|}~€���������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������Root Entry�������� �F 匰�!浲�Data ������������jB1Table��������v�5WordDocument ����v�SummaryInformation(�������������DocumentSummaryInformation8���������MsoDataStore��������@濭�!浲 匰�!浲�LXH4������FG�X�W�NYO�==2��������@濭�!浲 匰�!浲Item ���� �����Properties������������UCompObj���� r��������������� ��� ����������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������� � ���� �F Microsoft Word 97-2003 Document MSWordDocWord.Document.8�9瞦