Abstract
Objectives:
Examine differences in health care utilization and out-of-pocket (OOP) expenditures among older Mexican adults in 2001 and 2012, and identify individual characteristics associated with utilization and expenditures in both years.
Methods:
Data from the 2001 and 2012 cross-sections of the Mexican Health and Aging Study were utilized. Outcomes included nights spent in the hospital, medical/outpatient procedures, and OOP expenditures with these services. Covariates included demographics and comorbidities. Two-part regression models were used to identify covariates associated with utilization and expenditures in each year.
Results:
The proportion of those who spent at least one night in the hospital or had at least one medical/outpatient visit was higher in 2012 than in 2001, while the proportion of individuals with OOP expenditures decreased between the years. Those with more comorbidities had the highest OOP expenditures in both years.
Discussion:
Although the population paying for health care services OOP was lower in 2012, there is persistent inequality in expenditures across population groups.
Keywords: health care expenditures, health care utilization, Mexico
Introduction
Mexico is experiencing rapid population aging; by 2050, the older adult population above 65 years old will increase 361% compared to 2005 (Jackson, Strauss, & Howe, 2009). Increasing age drives increased health care use and spending. Aging is associated with higher prevalence of chronic diseases and disabilities that can increase the use of health care services (Doubova, Perez-Cuevas, Canning, & Reich, 2015). Those above 60 years of age are more likely to have the highest proportion of acute morbidity, ambulatory care, hospitalizations, and total out-of-pocket (OOP) expenditures in Mexico (Gonzalez-Gonzalez et al., 2011).
The Mexican demographic transition resembles the one U.S. and Europe have experienced, but at a much faster pace (Wong & Palloni,2009). Along with the demographic transition, during the early 2000s Mexico experienced one of the highest health-related disparities on the American continent. Access to health services and life expectancy differed between poor and rich areas of the country and across age groups (Knaul& Frenk, 2005). Disparity in access and health-related outcomes was mainly a result of high administrative complexity and the fragmentation of the Mexican health care system (Knaul & Frenk, 2005).
Prior to 2003, the population had unequal access to health insurance, with nearly half of the population uninsured (Frenk, Gonzalez-Pier, Gomez-Dantes, Lezana, & Knaul, 2006). The unemployed, the informally employed, and those in rural areas were mostly uninsured (Barraza-Llorens, Bertozzi, Gonzalez-Pier, & Gutierrez, 2002; Frenk et al., 2006). Mexico had a mix of private and public insurers but most of the population was insured by public institutions (Frenk et al., 2006). A minority of the population (3%) was insured by private insurance and other small public insurers (Frenk et al., 2006).
Another source of disparity in Mexico was the high OOP expenditures incurred by the population. More than half of the total health care spending in Mexico was OOP (Knaul et al., 2012). The uninsured, those in the lowest income decile, and the older population had limited access to care and also experienced the greatest burden of health care expenditures (Barraza-Llorens et al., 2002; Parker & Wong, 1997).
In the early 2000s, a new health care policy called System of Social Protection in Health was launched to help overcome these structural problems (Frenk et al., 2006). One of the results from this policy was the creation of a public health insurance called Seguro Popular (SP), which was pilot tested in 2002 and implemented nationwide in 2004 (Gakidou et al., 2006). The goal of SP is to decrease the disparities between those with social security benefits and those without them, and decrease the proportion of uninsured in the country (F. M. Knaul et al., 2013). It funds essential health care services to decrease the economic barriers to access health care, and gives financial protection against catastrophic OOP expenditures, that is, spending more than 30% of household income on health care (Frenk et al., 2006; F. M. Knaul et al., 2013). It prioritizes the population with the lowest incomes, and by 2006 the majority of enrollees (40%) were from the two poorest deciles of the population (Gakidou et al., 2006).
Considering the demographic and health care changes in Mexico over the past decade, our objective is to examine health care utilization and OOP expenditures among older Mexican adults within a critical decade, defined here from 2001 to 2012. Moreover, we will identify the population groups more or less likely to use health care and have OOP expenditures in both years. Because the health care reform increased access to care, lowered costs for those with access, and sought to decrease existing disparities, we hypothesize that a greater proportion of elders will utilize services but a smaller proportion will incur OOP expenditures by 2012. We also hypothesize that the oldest and those with comorbidities will be more likely to use health care and to pay for services OOP than their counterparts in both years.
Methodology
Data Source
We utilized two cross-sections of the Mexican Health and Aging Study (MHAS) to represent the decade of structural changes in Mexico. We used Wave 1 in 2001 and Wave 3 in 2012 to examine differences in access to health care, use of health care, and expenditures after a decade of changes in Mexico. The MHAS is a nationally representative longitudinal survey of the population 50 years and older in Mexico. The individuals were sampled in rural and urban areas, and their spouses were also selected independent of age (Wong, Michaels-Obregon, & Palloni, 2017). In 2012, The MHAS added participants born from 1952 to 1962 to refresh the sample. The response rate in 2001 was 91.8% and 88.1% in 2012 (Wong et al., 2015). MHAS data were de-identified and are publicly available (Mexican Health and Aging Study. http://mhasweb.org). The MHAS was approved by the IRB at the University of Texas Medical Branch in the United States, and the Instituto Nacional de Estadística y Geografía in Mexico.
Study Population
We selected direct (selected individuals answered the questionnaire) and proxy interviews (another individual answered the questionnaire on behalf of the selected one) for the analysis. Proxy interviews were included as these may be more likely to use health care services and have greater OOP expenditures. For the purposes of this study, the sample of older adults was defined as those 50 years old and older. The initial sample size was 13,679 individuals in 2001 (12,701 direct and 978 proxy interviews) and 14,897 in 2012 (13,654 direct and 1,243 proxy interviews). Individuals missing any study variables were excluded (765 or 5.6% in 2001; and 1,019 or 6.8% in 2012), and the final sample was 12,914 in 2001 and 13,878 in 2012. We applied survey weights in the descriptive analysis to generalize results for the Mexican population 50 years old and older in each year.
Outcomes
We examined two outcomes in this study. The first outcome was utilization of services, measured by self-report of health care services used in the previous year (number of medical/outpatient visits and nights spent in the hospital). We combined medical and outpatient visits into one variable due to the small sample size of outpatient visits. The second outcome was OOP expenditures for health care, measured by the total self-reported annual OOP expenditures associated with these health care services. Missing values for OOP expenditures were imputed by the MHAS research group with a multiple imputation methodology described elsewhere (Wong & Espinoza, 2004; Wong, Orozco-Rocha, Zhang, Michaels-Obregon, & Gonzalez-Gonzalez, 2016). We converted 2001 OOP expenditures in Mexican Pesos (MEX$) to 2012 Mexican Pesos using the Mexican Consumer Price Index, to adjust for inflation and permit comparison between years (Mundi, 2015).
Covariates
We included sociodemographic covariates: sex (men, women); age (50-59, 60-69, ≥70 years old); marital status (single, married/civil union, divorced/separated, widowed); education (0, 1 to 6, ≥7 years); locality of residence (more urban if population ≥100,000, less urban if population<100,000); income tertile (low, medium, high); and primary health care payer (uninsured, Mexican Social Security Institute [IMSS], Institute for Social Security and Services for State Workers [ISSSTE], Seguro Popular in 2012, other/private) as these factors can influence health care utilization and OOP expenditures (Gonzalez-Gonzalez et al., 2011; Shen & McFeeters, 2006; R. Wong & Diaz, 2007).
Additionally, we included self-reported chronic and infectious diseases as morbidities drive health care utilization/expenditures (Lehnert et al., 2011; Paez, Zhao, & Hwang, 2009;Schoenberg, Kim, Edwards, & Fleming, 2007). Older Mexican adults live in a mixed burden of both chronic and infectious diseases (Samper-Ternent, Michaels-Obregon, Wong, & Palloni, 2012) thus it is important to include some infectious diseases to determine how these impact healthcare use and OOP expenditures. Because of the small frequency of some diseases, we created a comorbidity variable to aggregate individuals who reported 0, 1, 2, and 3 or more morbidities. We considered the following chronic diseases: if individuals responded that a doctor or medical personnel had ever diagnosed them with diabetes, cardiovascular disease (hypertension or heart attack), chronic respiratory diseases, stroke, cancer, or arthritis. We also considered two infectious diseases: if individual responded that a doctor or medical personnel had diagnosed them with kidney or liver infection, or lung infection (tuberculosis or pneumonia) in the past two years.
Statistical Analysis
In the univariate analysis, we compared the covariates and outcomes between 2001 and 2012 with a chi-square test for categorical variables and with a t-test for continuous variables. We described the main outcomes as: proportion of individuals who reported nights spent in the hospital or medical/outpatient visits, and proportion of individuals with any OOP expenditures incurred for these health care services. We also calculated the mean number of nights in the hospital, medical/ outpatient visits, and mean total OOP expenditures. Means for health services/OOP expenditures were computed only for those individuals who reported using these services at least once in the previous year (i.e., means were computed excluding zeroes).
In the bivariate analysis, we compared the proportion of individuals with any medical/outpatient visits or nights spent in the hospital versus those with none across covariates with chi-squared tests within each year. We also compared mean total OOP expenditures excluding zero across covariates with a one-way ANOVA.
In the regression analysis, we conducted two-part models for each outcome to identify potential differences in health care use and expenditures by population groups for each year (Belotti, Deb, Manning, & Norton, 2015). The first part of the model was a logistic regression based on the probability of having 0 vs. ≥1 outpatient visits (or 0 vs. ≥1 nights in the hospital, or 0 vs. ≥1 OOP expenditure). Conditional on having use of health care or having OOP expenditures in the first part of the model, the second part modeled the number of non-zero outpatient visits (or non-zero number of nights in the hospital, or non-zero amount of OOP expenditures) with a generalized linear model. We utilized this model because of the highly skewed distribution of outcomes toward zero. All models were adjusted for covariates previously described. We reported the second part of the multivariable regression as adjusted marginal effects, that is, how much individuals use or spend, on average, more or less than their counterparts in the reference group, adjusted by covariates.
Considering the differences in etiology, prognoses, and cost among the diseases included into the comorbidity index and the importance of this variable to the model based on preliminary analysis, we conducted a second set of analyses with a new two-part model for OOP expenditures as the outcome and each of the chronic and infectious diseases as independent variables coded separately as yes/no variables. We identified which conditions were relevant in the model in both years, adjusting for covariates previously described. We also reported the second part of this model as adjusted marginal effects.
All analyses were conducted in Stata 14.0 (College Station, TX). Two-sided significance was considered at alpha=0.05 level of significance.
Results
Descriptive Results
Gender and age distribution were similar for older Mexican adults in 2001 and 2012, but educational levels and insurance status changed greatly in the decade. The proportion of individuals with more than 7 years of formal education almost doubled (18.4% in 2001 and 32.7% in 2012, p<0.001) and the proportion of uninsured shrank two-thirds (45.9% in 2001 to 14.8% in 2012, p<0.001). Furthermore, by 2012, 31.7% of older adults were insured with SP (Table 1).
Table 1.
Population Characteristics, MHAS 2001 and 2012 Direct Interviews a
| Characteristics | 2001 (n=12,914) |
2012 (n=13,878) |
p-value a |
|---|---|---|---|
| Weighted % | Weighted % | ||
| Sex | |||
| Male | 45.6 | 46.9 | 0.3 |
| Female | 54.4 | 53.1 | |
| Age (in years) | |||
| 50-59 | 45.2 | 46.3 | 0.1 |
| 60-69 | 30.6 | 31.7 | |
| ≥70 | 24.2 | 22.0 | |
| Mean Age in years (SD)b | 62.9 (9.7) | 62.6 (9.5) | 0.2 |
| Marital Status | |||
| Single | 4.5 | 4.8 | <0.001 |
| Married or Civil Union | 66.0 | 70.9 | |
| Divorced or Separated | 9.6 | 9.4 | |
| Widowed | 19.9 | 14.8 | |
| Education (completed years of formal education) | |||
| 0 | 31.3 | 16.7 | <0.001 |
| 1-6 | 50.3 | 50.5 | |
| ≥7 | 18.4 | 32.7 | |
| Primary Payer | |||
| Uninsured | 45.9 | 14.8 | <0.001 |
| IMSS | 36.8 | 34.3 | |
| ISSSTE | 10.3 | 11.4 | |
| Seguro Popular | NA | 31.7 | |
| Other | 6.9 | 7.9 | |
| Locality Size | |||
| Less Urban | 53.5 | 50.6 | 0.02 |
| More Urban | 46.5 | 49.4 | |
| Income | |||
| Low | 36.2 | 33.6 | 0.04 |
| Medium | 34.0 | 34.0 | |
| High | 29.8 | 32.4 | |
Bolding indicates statistically significant difference between the years, chi-square test with p<0.05.
Bolding indicates statistically significant difference between means, two sample t-test with p<0.05.
MHAS: the Mexican Health and Aging Study; SD: standard deviation; IMSS: Mexican Social Security Institute; ISSSTE: Institute for Social Security and Services for State Workers.
The morbidity pattern also changed within the decade (Table 2). Cardiovascular disease rates remained similar (around 38%), whereas the proportion of self-reported diabetes increased from 15.5% in 2001 to 19.7% in 2012 (p<0.001). Prevalence of arthritis/rheumatism was 20.9% in 2001 and decreased to 12.6% in 2012 (<0.001). The report of three or more comorbidities also remained similar between the years (Table 2).
Table 2.
Self-Reported chronic and infectious conditions, MHAS 2001 and 2012 Direct Interviews.
| 2001 | 2012 | pa | |
|---|---|---|---|
| Weighted % | Weighted % | ||
| Has a doctor or medical personnel ever diagnosed you with …? | |||
| Diabetes | |||
| Yes | 15.5 | 19.7 | <0.001 |
| Cancer | |||
| Yes | 1.9 | 1.8 | 0.6 |
| Cardiovascular Disease | |||
| Yes | 38.3 | 38.8 | 0.6 |
| Respiratory Illness | |||
| Yes | 6.0 | 5.2 | 0.2 |
| Stroke | |||
| Yes | 2.5 | 2.0 | 0.2 |
| Arthritis or Rheumatism | |||
| Yes | 20.9 | 12.6 | <0.001 |
| In the last 2 years, has a doctor or medical personnel told you that you have …? | |||
| Kidney or Liver Infection | |||
| Yes | 10.4 | 9.4 | 0.2 |
| Lung Infection (Tuberculosis or Pneumonia) | |||
| Yes | 1.8 | 1.3 | 0.1 |
| Comorbidity Number | |||
| 0 | 42.8 | 42.5 | 0.3 |
| 1 | 32.1 | 33.8 | |
| 2 | 16.7 | 16.3 | |
| 3 or more | 8.4 | 7.4 | |
Bolding indicates statistically significant difference between the years, Chi-square test with p<0.05.
MHAS: the Mexican Health and Aging Study;
The main study outcomes are described in Table 3. The proportion of individuals who spent at least one night in the hospital in the previous year increased from 8.0% in 2001 to 9.6% in 2012 (p=0.02). However, the proportion of those who paid OOP for this service and the average OOP expenditure for hospitalizations did not increase significantly. Similar results were observed for medical/outpatient visits. The use of at least one medical/outpatient visit in the previous year increased (63.9% in 2001 and 70.3% in 2012, p<0.001), and the mean number of medical/outpatient visits excluding zero also increased between the years. On the other hand, significantly fewer people paid for medical/outpatient visits OOP in 2012 (37.2 %) compared to 2001 (53.2%, p<0.001), and the average OOP expenditure for medical/outpatient visits did not differ significantly between years (Table 3).
Table 3.
Health care Utilization and OOP Expenditures among Older Adults, MHAS 2001 and 2012 Direct Interviews.
| 2001a | 2012 | p-valueb | |
|---|---|---|---|
| Weighted | Weighted | ||
| Health Care Utilization | |||
| Nights spent in the hospital/year | |||
| % ≥ 1 | 8.0 | 9.6 | 0.02 |
| % 0 | 92.0 | 90.4 | |
| Mean nights in the hospital excluding 0/year (SE)c | 8.2 (0.8) | 6.3 (0.5) | 0.1 |
| Outpatient or medical visits/year | |||
| % ≥ 1 | 63.9 | 70.3 | <0.001 |
| % 0 | 36.1 | 29.7 | |
| Mean outpatient and medical excluding 0/year (SE)c | 6.2 (0.13) | 6.6 (0.12) | 0.01 |
| Health Care OOP Expenditures | |||
| Hospitalization/ year | |||
| % Paid | 41.6 | 37.0 | 0.3 |
| % Did not pay | 58.4 | 63.0 | |
| Mean MEX$ spent in hospitalization excluding 0/year (SE)c | 17,243 (2,220) |
19,317 (3,331) |
0.6 |
| Outpatient and medical/ year | |||
| % Paid | 53.2 | 37.2 | <0.001 |
| % Did not pay | 46.8 | 62.8 | |
| Mean MEX$ spent in outpatient and medical visits excluding 0/ year (SE)c | 2,582 (149) | 2,646 (298) | 0.8 |
OOP expenditures in 2001 were converted to 2012 Mexican Pesos using the Consumer Price Index
Bolding indicates statistically significant difference between the years, chi-square test with p<0.05
Differences based on two sample t-test. Bolding indicates statistical significance with p<0.05
OOP: out of pocket; MHAS: the Mexican Health and Aging Study; SE: standard error; MEX$: Mexican Pesos.
Bivariate Results
Individuals with the highest proportion of nights spent in the hospital in 2001 were: women, the oldest, the widowed, those living in urban localities, and those with three or more comorbidities (Table 4). These effects were still present in 2012. Those with lower education were significantly more likely than the more educated to spend nights in the hospital in both years.
Table 4.
Bivariate Health Care Utilization and OOP Expenditures.
| 2001 | 2012 | |||||||
|---|---|---|---|---|---|---|---|---|
| ≥1 Nights Spent in the Hospital (%)a |
≥1 Outpatient and Medical Visits (%)a |
Mean Total OOP Expenditure in MEX$ excluding zero (95% CI)b |
≥1 Nights Spent in the Hospital (%)a |
≥1 Outpatient and Medical Visits (%)a |
Mean Total OOP Expenditure in MEX$ excluding zero (95% CI) b |
|||
| Sex | ||||||||
| Male | 8.7 | 56.9 | 4,386.9 (3,647.8-5,126.0) | 9.7 | 67.1 | 4,935.4 (4,080.6-5,790.4) | ||
| Female | 10.8 | 72.0 | 5,008.3 (4,175.2-5,841.4) | 11.9 | 78.8 | 5,251.5 (4,561-5,941.7) | ||
| Age (in years) | ||||||||
| 50-59 | 7.9 | 63.0 | 3,774.2 (3,228.7-4,319.8) | 8.1 | 68.2 | 3,631.0 (2,909.8-4,352.2) | ||
| 60-69 | 10.2 | 66.5 | 3,953.1 (3,438.8-4,467.3) | 11.1 | 75.0 | 5,244.3 (4,331.9-6,156.8) | ||
| ≥70 | 13.2 | 67.6 | 7,656.6 (5,559.7-9,753.6) | 13.9 | 78.0 | 6,760.4 (5,594.8-7,926.0) | ||
| Marital Status | ||||||||
| Single | 8.8 | 61.6 | 4,781.0 (2,498.6-7,063.3) | 9.5 | 64.6 | 4,609.9 (2,319.8-6,899.9) | ||
| Married | 9.0 | 63.8 | 4,709.8 (3,913.2-5,506.5) | 10.4 | 73.0 | 5,078.5 (4,384.8-5,772.3) | ||
| Divorced or Separated | 11.3 | 65.9 | 3,739.4 (2,683.3-4,795.5) | 11.5 | 73.0 | 4,172.4 (2,970.1-5,374.6) | ||
| Widowed | 12.2 | 70.3 | 5,388.4 (4,292.9-6,483.9) | 13.1 | 79.1 | 5,874.9 (4,764.3-6,985.6) | ||
| Education (completed years of formal education) | ||||||||
| 0 | 10.1 | 61.3 | 4,318,1 (3,044.2-5,592.1) | 12.1 | 72.0 | 4,171.9 (3,352.9-4,990.8) | ||
| 1-6 | 10.3 | 65.8 | 4,381.1 (3,665.9-5,096.3) | 11.4 | 74.2 | 4,790.4 (4,139.2-5,441.7) | ||
| ≥7 | 8.5 | 68.1 | 6,434.0 (4,992.9-7,875.2) | 9.6 | 73.8 | 6,092.6 (4,885.7-7,299.5) | ||
| Primary Payer | ||||||||
| Uninsured | 7.1 | 54.2 | 4,225.4 (3,770.7-4,680.1) | 5.2 | 49.4 | 4,393.4 (3,214.6-5,572.3) | ||
| IMSS | 12.0 | 71.5 | 5,795.4 (4,071.8-7,519.1) | 12.3 | 78.5 | 5,291.5 (4,333.1-6,249.9) | ||
| ISSSTE | 11.4 | 73.7 | 4,705.9 (3,663.9-5,747.9) | 13.5 | 79.3 | 6,094.1 (4,536.2-7,652.0) | ||
| Seguro Popular | 9.9 | 73.8 | 4,214.8 (3,456.1-4,973.5) | |||||
| Other | 9.0 | 70.6 | 4,849.5 (2,805.1-6,893.8) | 13.0 | 78.0 | 8,127.1 (5,094.5-11,159.7) | ||
| Locality Size | ||||||||
| Less Urban | 8.9 | 60.9 | 4,531.4 (3,705.5-5,357.3) | 10.0 | 71.3 | 4,739.6 (4,078.2-5,401.1) | ||
| More Urban | 10.5 | 68.1 | 4,976.3 (4,143.9-5,808.5) | 11.6 | 75.3 | 5,408.2 (4,612.8-6,203.6) | ||
| Income | ||||||||
| Low | 9.1 | 61.6 | 4,252.4 (3,552.9-4,951.8) | 10.5 | 67.9 | 4,315.4 (3,497.8-5,133.0) | ||
| Medium | 10.0 | 66.3 | 3,417.1 (2,959.8-3,874.5) | 11.6 | 76.9 | 4,295.7 (3,499.6-5,091.8) | ||
| High | 10.4 | 67.5 | 6,689.7 (5,123.1-8,256.3) | 10.7 | 76.0 | 6,623.1 (5,523.7-7,722.4) | ||
| Comorbidity Number | ||||||||
| 0 | 4.9 | 47.7 | 2,691.0 (2,318.9-3,063.2) | 5.2 | 57.0 | 3,786.3 (2,898.6-4,673.9) | ||
| 1 | 9.3 | 71.6 | 4,340.9 (3,613.1-5,068.8) | 9.6 | 79.3 | 4,733.2 (3,939.5-5,526.9) | ||
| 2 | 14.9 | 80.7 | 5,257.4 (4,328.3-6,186.5) | 16.1 | 86.8 | 6,716.6 (5,339.7-8,093.5) | ||
| 3 or more | 23.8 | 88.2 | 10,493.7 (6,604.1-14,383.2) | 28.9 | 90.9 | 7,521.2 (5,654.2-9,388.3) | ||
Bolding indicates statistically significant difference across variable categories, chi-square test with p<0.05.
Bolding indicates statistical significance across variable categories, one-way ANOVA with p<0.05.
OOP: out of pocket; MEX$: Mexican Pesos; IMSS: Mexican Social Security Institute; ISSSTE: Institute for Social Security and Services for State Workers.
Individuals with the highest use of medical/outpatient services in 2001 were: women, the widowed, those insured with ISSSTE, those living in more urban localities, and those with three or more comorbidities (Table 4). These effects were still present in 2012. In 2001, those with more education were significantly more likely to use medical/outpatient visits services, but this difference was not significant by 2012.
Individuals with the highest total OOP expenditures in 2001 were: the oldest, those with ≥7 years of education, and those with three or more comorbidities (Table 4). These effects were still present in 2012. In 2001, OOP expenditures did not differ by primary payer. However, by 2012, those with other insurance and ISSSTE had significantly higher OOP expenditures than their counterparts..
Adjusted Regression Results
Adjusted individual characteristics associated with health care use and OOP expenditures are described in Table 5. For nights spent in the hospital, individuals older than 70 years, those divorced or separated, those insured, and those with at least one morbidity spent, on average, more nights in the hospital than their counterparts in the reference group in both years. On the other hand, women and those with higher education spent fewer nights.
Table 5.
Adjusted Marginal Effects for Health Care Utilization and OOP Expenditures from Two-part Models a
| 2001 | 2012 | |||||
|---|---|---|---|---|---|---|
| Nights Spent in the Hospital dy/dx (SE) |
Outpatient and Medical Visits dy/dx (SE) |
Total Out-of- Pocket Expenditure MEX$ dy/dx (SE) |
Nights Spent in the Hospital dy/dx (SE) |
Outpatient and Medical Visits dy/dx (SE) |
Total Out-of-Pocket Expenditure MEX$ dy/dx (SE) |
|
| Sex (Ref: Male) | ||||||
| Female | −0.06 (0.05) | 0.89 (0.06) | 563.3 (132.1) | −0.19 (0.04) | 1.06 (0.07) | 341.4 (145.5) |
| Age in years (Ref: 50-59) | ||||||
| 60-69 | 0.09 (0.05) | 0.41 (0.07) | 311.8 (130.4) | 0.25 (0.04) | 0.70 (0.07) | 359.9 (136.5) |
| ≥70 | 0.46 (0.07) | 0.84 (0.08) | 1,447.9 (253.0) | 0.29 (0.05) | 1.16 (0.09) | 925.5 (221.9) |
| Marital Status (Ref: Single) | ||||||
| Married | 0.35 (0.07) | 0.37 (0.15) | 494.0 (267.7) | 0.05 (0.09) | 0.85 (0.14) | 464.7 (225.4) |
| Divorced or Separated | 0.67 (0.11) | 0.39 (0.18) | 6.65 (303.8) | 0.25 (0.12) | 1.04 (0.18) | 171.2 (277.9) |
| Widowed | 0.44 (0.08) | 0.09 (0.16) | 102.8 (278.6) | 0.11 (0.10) | 0.69 (0.16) | 366.9 (257.7) |
| Education in completed years of formal education (Ref: <6 years) | ||||||
| 1-6 | −0.23 (0.06) | 0.26 (0.07) | 289.6 (121.5) | 0.02 (0.06) | 0.20 (0.09) | 272.0 (143.4) |
| ≥7 | −0.31 (0.08) | 0.31 (0.10) | 1,729.6 (339.9) | −0.28 (0.06) | 0.01 (0.10) | 956.1 (259.9) |
| Primary Payer (Ref: Uninsured) | ||||||
| IMSS | 0.44 (0.05) | 2.10 (0.07) | −1,184.0 (178.6) | 0.74 (0.04) | 3.70 (0.09) | −1,265.3 (278.5) |
| ISSSTE | 0.32 (0.07) | 1.88 (0.10) | −1.403.8 (202.2) | 0.72 (0.06) | 3.04 (0.11) | −1,039.4 (316.4) |
| Seguro Popular - | − | − | 0.34 (0.04) | 2.27 (0.09) | −779.1 (288.6) | |
| Other | 0.18 (0.09) | 1.52 (0.13) | −1,216.0 (260.1) | 1.04 (0.11) | 2.87 (0.14) | −75.8 (490.3) |
| Locality Size (Ref: Less Urban) | ||||||
| More Urban | 0.16 (0.04) | −0.20 (0.06) | −272.7 (148.4) | −0.02 (0.04) | 0.27 (0.07) | −123.5 (160.3) |
| Income (Ref: Low) | ||||||
| Medium | −0.19 (0.05) | 0.35 (0.07) | −173.2 (131.8) | −0.04 (0.05) | 0.57 (0.08) | −85.7 (142.3) |
| High | 0.13 (0.06) | 0.49 (0.08) | 649.4 (183.8) | 0.01 (0.05) | 0.55 (0.08) | 604.9 (196.9) |
| Comorbidity Count (Ref: 0) | ||||||
| 1 | 0.35 (0.04) | 2.55 (0.07) | 833.1 (116.0) | 0.19 (0.03) | 3.0 (0.07) | 479.7 (132.2) |
| 2 | 0.95 (0.07) | 4.40 (0.10) | 1,435.4 (201.4) | 0.63 (0.05) | 4.76 (0.09) | 1,009.2 (223.1) |
| 3 or more | 1.98 (0.12) | 5.81 (0.13) | 3,708.9 (539.1) | 2.30 (0.13) | 5.93 (0.13) | 1,530.5 (398.1) |
Bolding indicates statistical significance across variable categories with p-value <0.05; dy/dx: Marginal effects compared to reference category. OOP: out of pocket; SE: Standard error; MEX$: Mexican Pesos; IMSS: Mexican Social Security Institute; ISSSTE: Institute for Social Security and Services for State Workers.
Similar results were observed for medical/outpatient visits. Women, those above 60 years old, those married or divorced, those with 1-6 years of education, those insured, those of medium or high income, and those with at least one comorbidity had, on average, more medical/outpatient visits than their counterparts in the reference group in both years.
Total OOP expenditures also differed by individual characteristics (Table 5). Women, those above 60 years old, those with ≥7 years of education, those of high income, and those with at least one comorbidity spent, on average, more on OOP expenditures in both years. Conversely, those with any insurance spent, on average, less in OOP expenditures compared to the uninsured in both years.
Those with three or more comorbidities (compared to those with none) had the greatest difference in health care use and OOP expenditures across all population groups. They spent on average 1.98 more nights in the hospital in 2001 and 2.30 more nights in 2012; they had 5.81 more medical/outpatient visits in 2001 and 5.93 more visits in 2012; and they spent, on average, MEX$3,708.9 more OOP in 2001 and MEX$1,530.5 in 2012 (Table 5).
The average OOP expenditures according to specific chronic and infectious conditions (Figure 1) were highest in both years for those who reported having cancer. Those with diabetes and respiratory illnesses in 2001 had significantly higher OOP expenditures than those without each condition. However, this difference was not significant by 2012 (Figure 1).
Figure 1. Health Care Out of Pocket (OOP) Expenditures by Type of Chronic or Infectious Disease.
a Adjusted marginal effects of OOP expenditures in Mexican Pesos (MEX$) for each chronic and infectious disease based on the second part of the multivariable regression for each year. Also adjusted by sex, age, marital status, education, insurance, locality size, and income. (*) indicates that those with the disease spent significantly more OOP than those without the disease in each year at p-value <0.05.
Discussion
Our results showed large differences in the health care use and OOP expenditures of older Mexican adults in the decade between 2001 and 2012. First, as hypothesized, more individuals spent nights in the hospital and used medical/outpatient services by 2012, although fewer incurred OOP expenditures for these services. Second, also as hypothesized, the oldest of the old were more likely to use services and pay for services OOP, as were those with more comorbidities. Only insurance was a protective factor against OOP expenditures in both years.
The greater access to care and decrease in OOP expenditures in 2012 compared to 2001 could be a result of greater insurance coverage, as shown by the reduction of uninsured from 45.9% in 2001 to 14.8% in 2012 (p<0.001). Approximately one third of the older adult population in our sample is now insured with the new SP. Researchers have observed that SP facilitated the inclusion of populations previously excluded from the public health insurance system (Doubova et al., 2015; Knaul et al., 2012; Salinas, 2015).
The rise in health care utilization and the lower proportion of OOP expenditures in 2012 can also be associated with the health care reform. SP facilitated health care access, especially for primary care and diagnosis (Parker, Saenz, & Wong, 2018). Compared to those previously uninsured, those insured with SP were more likely to see a doctor after receiving SP (Parker et al., 2018). By 2012, SP had 53 million enrollees (Knaul et al., 2012) and catastrophic expenditures fell 54% nationwide (Galarraga, Sosa-Rubi, Salinas-Rodriguez, & Sesma-Vazquez, 2010). The OOP health expenditures among those now insured with SP also fell 22% (Knaul et al., 2012).
Higher education in the 2012 population may also be associated with the increase in insured, the lower proportion of OOP expenditures and the greater access to care in 2012. Those more educated are more likely to have a formal job with access to health insurance, have greater health literacy, visit the doctor, have more positive health behaviors, among other mechanisms to access and pay for health care (Zimmerman, Woolf, & Haley, 2015).
Although the proportion of the population paying for health care services OOP decreased for medical/outpatient visits and did not increase for nights spent in the hospital, our results showed a persistent inequality in OOP expenditures across population groups. The differences in OOP expenditures across groups observed in 2001 were still present in 2012, despite an overall decreasing trend. This disparity was especially the case for sicker adults compared to their healthier counterparts. Those with three or more comorbidities had the highest health care use and OOP expenditures in both years. However, by 2012, they appeared to be paying less OOP. A dose-response relationship between multiple chronic conditions and OOP expenditures was also observed among older adults in a study of U.S. older adults. These results also indicate that OOP expenditures may differ depending upon the multiple combinations of morbidities presented by individual patients (Schoenberg et al., 2007).
Our additional results showed that the overall decreasing trend in OOP expenditures was not observed for all diseases. In 2001, cancer, cardiovascular disease, and diabetes were large drivers of OOP expenditures. By 2012, OOP expenditures decreased for all diseases (diabetes and respiratory illness even lost significance), except stroke which had a small increase.
High OOP expenditures for these diseases may be related to the limited health care coverage for specific conditions highly prevalent in the older population in Mexico, and a large focus on infant and obstetrical health services. For those above 18 years old in 2012, the only conditions covered under the protection against catastrophic OOP expenditures in SP were: certain cancers (uterus, breast, prostate, testicular, and lymphoma non- Hodgkin’s), HIV/AIDS, bone marrow or cornea transplant, and cataract surgery (Comision Nacional de Protecction Social en Salud, 2012).
Parker and colleagues (2018) showed that SP did not have a great impact on disease treatment, even though many medications are available for those insured (Parker et al., 2018). Diabetes in one of the only chronic conditions that shows improvement in disease management and treatment among SP enrollees (Rivera-Hernandez, Rahman, Mor, & Galarraga, 2016; Sosa-Rubi, Galarraga, & Lopez-Ridaura, 2009). This may explain the fact that diabetes was the only chronic condition without significant increase in OOP expenditures in 2012.
Over the last 15 years, changes in the Mexican health care system were in the direction of universal health care coverage (Frenk et al., 2006; Knaul & Frenk, 2005). However, further steps to achieve this goal include the health care response to chronic conditions of an aging population. The main focus of health care systems in developing countries such as Mexico has been acute diseases. Chronic conditions have often been considered as multiple acute events and have received disease-specific focus. However, the rapid epidemiological transition should promote a change in the structure of the health care system to one with a person-centered approach (Knaul, Bhadelia, Atun, & Frenk, 2015).
This study is limited by a self-report measure of health care utilization and OOP expenditures. To minimize bias, the values were probed by the interviewer, divided into specific boundaries and missing data was imputed by the MHAS research group with a multiple imputation methodology. Another limitation is that the 2001 and 2012 cross-sections of adults aged 50 and older are not completely independent, although treated as two separate cross-sections in this study.
In conclusion, the 10 years of structural changes in Mexico appear to have positively impacted older Mexican adults, reducing the proportion of uninsured and increasing access to health care while decreasing the proportion of payers with OOP expenditures. However, there are persistent inequalities to address, such as how to provide financial protection to the sickest population with different disease patterns. This group needs specific interventions from the health care system in Mexico to offer accessible health care at an affordable cost.
Acknowledgments
Funding
The Mexican Health and Aging Study is supported by the National Institute of Health/National Institute on Aging in the United States [grant number R01AG018016], and the Instituto Nacional de Estadística y Geografía (INEGI) in Mexico.
Footnotes
Conflict of Interests
The authors declare that there is no conflict of interest.
References
- Barraza-Llorens M, Bertozzi S, Gonzalez-Pier E, & Gutierrez JP (2002). Addressing inequity in health and health care in Mexico. Health Aff (Millwood), 21(3), 47–56. [DOI] [PubMed] [Google Scholar]
- Belotti F, Deb P, Manning W, & Norton E (2015). twopm: Two-part models. The Stata Journal, 15(1), 3–20. [Google Scholar]
- Comisión Nacional de Protección Social en Salud (2012). Catálogo Universal De Servicios De Salud 2012 (Causes). Retrieved from http://www.salud.df.gob.mx/portal/seguro_popular/index/pdf/causes2012.pdf.
- Doubova SV, Perez-Cuevas R, Canning D, & Reich MR (2015). Access to healthcare and financial risk protection for older adults in Mexico: secondary data analysis of a national survey. BMJ Open, 5(7), e007877. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Frenk J, Gonzalez-Pier E, Gomez-Dantes O, Lezana MA, & Knaul FM (2006). Comprehensive reform to improve health system performance in Mexico. Lancet, 368(9546), 1524–1534. [DOI] [PubMed] [Google Scholar]
- Gakidou E, Lozano R, Gonzalez-Pier E, Abbott-Klafter J, Barofsky JT, Bryson-Cahn C, . . . Murray CJ (2006). Assessing the effect of the 2001-06 Mexican health reform: an interim report card. Lancet, 368(9550), 1920–1935. [DOI] [PubMed] [Google Scholar]
- Galarraga O, Sosa-Rubi SG, Salinas-Rodriguez A, & Sesma-Vazquez S (2010). Health insurance for the poor: impact on catastrophic and out-of-pocket health expenditures in Mexico. Eur J Health Econ, 11(5), 437–447. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Gonzalez-Gonzalez C, Sanchez-Garcia S, Juarez-Cedillo T, Rosas-Carrasco O, Gutierrez-Robledo LM, & Garcia-Pena C (2011). Health care utilization in the elderly Mexican population: expenditures and determinants. BMC Public Health, 11, 192. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Jackson R, Strauss R, & Howe N (2009). Latin America’s aging challenge: Global Aging Iniciative. Center for strategic and international studies. [Google Scholar]
- Knaul FM, Gonzalez-Pier E, Gomez-Dantes O, Garcia-Junco D, Arreola-Ornelas H, Barraza-Llorens M, . . . Frenk J (2012). The quest for universal health coverage: achieving social protection for all in Mexico. Lancet, 380(9849), 1259–1279. [DOI] [PubMed] [Google Scholar]
- Knaul FM, Bhadelia A, Atun R, & Frenk J (2015). Achieving Effective Universal Health Coverage And Diagonal Approaches To Care For Chronic Illnesses. Health Aff (Millwood), 34(9), 1544–5208. [DOI] [PubMed] [Google Scholar]
- Knaul FM, & Frenk J (2005). Health insurance in Mexico: achieving universal coverage through structural reform. Health Aff (Millwood), 24(6), 1467–1476. [DOI] [PubMed] [Google Scholar]
- Knaul FM, González-Pier E, Gómez-Dantés O, García-Junco D, Arreola-Ornelas H, Barraza-Lloréns M, . . . Frenk J (2013). Hacia la cobertura universal en salud: protección social para todos en México. Salud Publica Mex, 55, 207–235. [DOI] [PubMed] [Google Scholar]
- Lehnert T, Heider D, Leicht H, Heinrich S, Corrieri S, Luppa M, . . . Konig, H. H. (2011). Review: health care utilization and costs of elderly persons with multiple chronic conditions. Med Care Res Rev, 68(4), 387–420. [DOI] [PubMed] [Google Scholar]
- Health Mexican and Study Aging. Official website. Retrieved from http://mhasweb.org
- Mundi I (2015). Mexico - Consumer price index. Retrieved from http://www.indexmundi.com/facts/mexico/consumer-price-index
- Paez KA, Zhao L, & Hwang W (2009). Rising out-of-pocket spending for chronic conditions: a ten-year trend. Health Aff (Millwood), 28(1), 15–25. [DOI] [PubMed] [Google Scholar]
- Parker SW, Saenz J, & Wong R (2018). Health Insurance and the Aging: Evidence From the Seguro Popular Program in Mexico. Demography, 55(1), 361–386. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Parker SW, & Wong R (1997). Household income and health care expenditures in Mexico. Health Policy, 40(3), 237–255. [DOI] [PubMed] [Google Scholar]
- Rivera-Hernandez M, Rahman M, Mor V, & Galarraga O (2016). The Impact of Social Health Insurance on Diabetes and Hypertension Process Indicators among Older Adults in Mexico. Health Serv Res, 51(4), 1323–1346. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Salinas JJ (2015). Preventive health screening utilization in older Mexicans before and after healthcare reform. Salud Publica Mex, 57(0 1), S70–78. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Samper-Ternent R, Michaels-Obregon A, Wong R, & Palloni A (2012). Older Adults Under a Mixed Regime of Infectious and Chronic Diseases. Salud Publica Mex, 54(5), 487–495. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Schoenberg NE, Kim H, Edwards W, & Fleming ST (2007). Burden of common multiple-morbidity constellations on out-of-pocket medical expenditures among older adults. Gerontologist, 47(4), 423–437. [DOI] [PubMed] [Google Scholar]
- Shen YC, & McFeeters J (2006). Out-of-pocket health spending between low-and higher-income populations: who is at risk of having high expenses and high burdens? Med Care, 44(3), 200–209. [DOI] [PubMed] [Google Scholar]
- Sosa-Rubi SG, Galarraga O, & Lopez-Ridaura R (2009). Diabetes treatment and control: the effect of public health insurance for the poor in Mexico. Bull World Health Organ, 87(7), 512–519. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wong, & Palloni, A. (2009). Aging in Mexico and Latin America In Uhlenberg P (Ed.), International Handbook of Population Aging (pp. 231–252). Dordrecht: Springer Netherlands. [Google Scholar]
- Wong R, & Diaz JJ (2007). Health care utilization among older Mexicans: health and socioeconomic inequalities. Salud Publica Mex, 49 Suppl 4, S505–514. [DOI] [PubMed] [Google Scholar]
- Wong R, & Espinoza M (2004). Imputation of Non-Response on Economic Variables in the Mexican Health and Aging Study (MHAS/ENASEM) 2001. Project Report. Retrieved from http://mhasweb.org/Resources/DOCUMENTS/2001/Imputation_of_Non-Reponse_on_Economic_Variables_in_the_MHAS-ENASEM_2001.pdf [Google Scholar]
- Wong R, Michaels-Obregon A, & Palloni A (2017). Cohort Profile: The Mexican Health and Aging Study (MHAS). Int J Epidemiol, 46(2), e2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wong R, Michaels-Obregón A, Palloni A, Gutiérrez-Robledo LM, González-González C, López-Ortega M, . . . Mendoza-Alvarado, L. R. (2015). Progression of aging in Mexico: the Mexican Health and Aging Study (MHAS) 2012. Salud Publica Mex, 57(0 1), S79–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wong R, Orozco-Rocha K, Zhang D, Michaels-Obregon A, & Gonzalez-Gonzalez C (2016). Imputation of Non-Response on Economic Variables in the Mexican Health and Aging Study (MHAS/ENASEM) 2012. Project Report. Retrieved from http://mhasweb.org/Resources/DOCUMENTS/2012/Imputations/Imputation_of_Non_Reponse_on_Economic_Variables_in_the_MHAS_ENASEM_2012.pdf
- Zimmerman E, Woolf S, & Haley A (2015). Understanding the Relationship Between Education and Health: A Review of the Evidence and an Examination of Community Perspectives. Section V. Emerging Tools for Studying Population Health. Population Health: Behavioral and Social Science Insights. Rockville, MD: Agency for Healthcare Research and Quality. Retrieved from https://www.ahrq.gov/professionals/education/curriculum-tools/population-health/zimmerman.html [Google Scholar]