Family Out-of-Pocket Health Care Burden and Children’s Unmet Needs or Delayed Health Care Pinar Karaca-Mandic, PhD; Sung J. Choi-Yoo, BA; Jinhyung Lee, PhD; Peter Scal, MD, MPH From the Division of Health Policy and Management, School of Public Health, University of Minnesota (Dr Karaca-Mandic and Mr Choi-Yoo), Department of Pediatrics, Medical School, University of Minnesota, Minneapolis, Minn (Dr Scal); and Department of Economics, Sungkyunkwan University, Seoul, Korea (Dr Lee) The authors declare that they have no conflict of interest. Address correspondence to Pinar Karaca-Mandic, PhD, Division of Health Policy and Management, School of Public Health, University of Minnesota, 420 Delaware St SE, MMC 729, Minneapolis, MN 55455 (e-mail: [email protected]). Received for publication April 2, 2013; accepted October 28, 2013.
ABSTRACT OBJECTIVE: To assess the relationship between family members’ out-of-pocket (OOP) health care spending and unmet needs or delayed health care due to cost for children with and without special health care needs (SHCN). METHODS: Data come from the Medical Expenditure Panel Survey, 2002–2009, and include 63,462 observations representing 41,748 unique children. The primary outcome was having any unmet needs/delayed care as a result of the cost of medical care, dental care, or prescription drugs. We also examined having unmet needs/delayed care due to cost for each service separately. Key explanatory variables were OOP spending on the index child and OOP spending on other family members. We estimated multivariate instrumental variable models to adjust the results for potential bias from any unobserved factors that might influence both other family OOP costs and the outcome variable. RESULTS: An increase of other family OOP costs from $500 (50th percentile) to $3000 (90th percentile) was associated
with a higher adjusted rate of any unmet need/delayed care due to cost (1.39% to 5.62%, P < .001, among children without SHCN; 3.17% to 7.87%, P ¼ .01, among those with SHCN). Among children without SHCN, higher OOP costs among other family members were associated with higher levels of unmet needs or delays in medical, prescription drug, and dental care, while among children with SHCN, higher OOP costs among other family members was primarily associated with unmet or delayed dental care. CONCLUSIONS: Programs and policies that reduce the OOP costs of family members other than the child may improve the child’s access to care.
KEYWORDS: access; children’s health care access; children’s unmet needs/delayed needs; family out-of-pocket burden; out of pocket costs; unmet needs ACADEMIC PEDIATRICS 2014;14:101–108
WHAT’S NEW
ber (though the study did not distinguish between the child and adults).11 However, it is unknown whether the OOP health care costs of the child and his or her family members limit that child’s access to and use of needed health care services. A conceptual model of intrafamily resource allocation suggests that when faced with limited budgets, OOP health care costs may financially burden families and deter children’s receipt of needed health care even when the children are insured.12 For example, a child in a family with a sick parent who has high health care OOP spending may have unmet health care needs because the family’s budget constrains additional spending on health care. If children are at risk when their families are burdened by OOP spending, the policy solutions shift toward better coverage for all family members, not just for children. Decades of rising health care costs, together with the rising proportion of family income spent on health care, make it essential to investigate empirically the association of family health care financial burden and children’s receipt of needed health services. In 2001–2002, about 11% of American families spent at least 10% of their family income on OOP costs such as deductibles, coinsurance, and
Children are at risk of unmet needs/delayed health care when a family is experiencing high out-of-pocket burden. This finding shifts solutions to improve child health care access toward better coverage for family members rather than just for children.
OVER THE PAST decade, federal and state health policy efforts in the United States have focused on making sure that children have health insurance; these efforts have been largely successful.1–5 However, many children are underinsured and have health service needs that are not being met even with health insurance.6–8 Taking a family system perspective on health care access and utilization may provide insights into strategies for improving health care access for children. Within a family, parental insurance and access to a usual source of care are positively associated with children’s access to and use of health care services.9,10 Similarly, in families with children, out-of-pocket (OOP) spending for the entire family is shown to be positively associated with delayed or forgone care for at least one family memACADEMIC PEDIATRICS Copyright ª 2014 by Academic Pediatric Association
101
Volume 14, Number 1 January–February 2014
102
KARACA-MANDIC ET AL
copayments.13 From 2001 to 2006, the percentage of families that spent more than 10% of their income on health care increased.14 The recent recession has limited family budgets further.15 We examined OOP spending on children and OOP spending on their family members for the impact of such spending on whether the child has an unmet need or delayed health care due to cost, controlling for socioeconomic status, race, ethnicity, and health insurance. Our primary interest was to understand whether the OOP burden caused by other family members has consequences for meeting the child’s health care needs. Our study spans 2002–2009, which allows us to capture the effects of the recent economic downturn. Forgoing or delaying needed health care could have a long-term adverse effect on children’s health, even for healthy children, but especially for children with special health care needs (SHCN) who require specialized services.16–18 Therefore, we separately considered the impact on children with and without SHCN.
METHODS DATA We used the 2002–2009 Medical Expenditure Panel Survey (MEPS) Household Component—Full-Year Consolidated Data files and Medical Conditions files. The MEPS is a nationally representative survey of US households; it provides information on health care access and utilization, socioeconomic characteristics, employment, access to care, and related topics. The survey conducts 5 rounds of interviews covering 2 full calendar years and provides up to 2 separate full years of observation for each participant. We used the MEPS definition of a family, which consists of 2 or more persons living together in the same household.19 In MEPS, children with SHCN are identified by the children with SHCN screening instrument.20 OUTCOME VARIABLES We focused on unmet care needs or delayed health care, a commonly used critical indicator for access problems in children.16,17,21,22 The MEPS questionnaire inquires about medical care, dental care, and prescription medications, asking whether a given person was unable to receive care or medication; delayed in receiving either; and reasons for not receiving or delaying care. Responding yes to either of the first 2 questions indicated that the person was unable to receive or was delayed in receiving needed care. Possible reasons for not receiving or delaying care included the following: could not afford care; insurance would not approve/cover/pay for care; doctor refused the individual’s insurance; problems getting to doctor’s office; communications problems as a result of different language; could not get time off work; didn’t know where to go; was refused services; could not get child care; and did not have time.19 Our key outcome measure, a binary indicator for having any unmet needs or delayed care due to cost, was established
ACADEMIC PEDIATRICS
by an affirmative response to any of the questions about unmet needs or delayed care in either medical, dental, or prescription drug care with the reason related to costs. We also considered unmet needs/delayed care due to cost separately for each service. EXPLANATORY VARIABLES Our key explanatory variables were OOP spending on the index child (child OOP) and OOP spending on other family members (other family OOP) defined as the sum of OOP spending on family members other than the index child. In sensitivity analyses, we also calculated the OOP measures as a percentage of family total pretax income spent. All dollar measures were adjusted to reflect 2009 US dollars using the Consumer Price Index. We also controlled for child characteristics, including age, gender, race (white, black, Asian/Native Hawaiian/Pacific Islander, other), ethnicity (Hispanic, non-Hispanic), health insurance status, private dental insurance status, and health status (defined as excellent, very good, good, fair, or poor). The child’s health insurance status was categorized as uninsured all year, privately insured during any part of the year, or having only public insurance during the year. MEPS categorizes children with partial-year public insurance under the last category even if they were uninsured for part of the year. Moreover, we controlled for family characteristics, such as number of family members, federal poverty level (FPL) category, highest family educational attainment, employment status of adult family members, urban status, census region, and health insurance status of adult family members. STATISTICAL MODELS Using linear probability models, we first examined the binary outcome measure of any unmet need or delayed care due to cost as the dependent variable; child OOP, other family OOP, child and family characteristics were independent variables. Because the MEPS follows each child for up to 2 years, there were up to 2 full-year observations for each child. Multi-children families had separate observations for each child. We reported the absolute change in the rate of unmet need/delayed care due to cost per 100 children for each covariate. Then, we summarized the adjusted rates of any unmet needs/delayed care due to cost associated with different points in the OOP distribution (50th, 75th, 90th percentiles) in order to simplify interpretation of the results. Next, we estimated separate multivariate models of unmet/delayed care due to cost by need type: medical, prescription drug, and dental. We reported the absolute change in the rate of unmet need/delayed care due to cost per 100 children associated with the child OOP and other family OOP. In additional analyses, we investigated whether our findings were different during the recent recession (December 2007–June 2009) relative to the period before the recession. We added an indicator variable for the recession period (0 for 2002–2007 and 1 for 2008– 2009) for the interaction term of the child OOP and
ACADEMIC PEDIATRICS
FAMILY OUT-OF-POCKET HEALTH CARE BURDEN
the recession indicator, and for the interaction term of the other family OOP and the recession indicator. Negative and significant coefficient estimates on the interaction terms would suggest that the association between unmet needs/delayed care and OOP spending was weaker during the recession relative to the period before the recession. It is possible that unobserved factors could influence both a child’s unmet needs and family members’ OOP spending. If these unobserved factors were not accounted for, the results could be biased. For example, parents may have strong preferences for frequent and intensive health care, but this preference is not represented by a specific variable in the MEPS data. Such a preference could lead to a lower likelihood of an unmet need or delayed care for the child, while at the same time the preference would result in a high use of medical care and high OOP spending on other family members. Thus, a strong preference for frequent and intensive health care would result in high family OOP and low unmet needs of the child. In the case where there is a true positive relationship between family OOP spending and the higher likelihood of unmet needs for the child, such relationship would be understated. On the other hand, parents with a strong preference for frequent and intensive health care may be more likely to perceive and report unmet needs in general while having high OOP spending. This could bias the results toward overstating the relationship. Studies that do not specifically address this potential bias may inaccurately report associations. We sought to minimize the potential for bias and to strengthen our claim of causal inference by conducting our analysis using the econometric technique of instrumental variables.23 A valid instrumental variable needs to be correlated with the other family OOP and uncorrelated with unobserved factors that affect the outcome variable itself. We created our instrumental variable as the sum of 3 conditions among other family members: cancer, heart disease, and depression/anxiety disorders. This instrumental variable is valid because family members with these conditions have higher OOP spending and because the prevalence of these conditions among other family members is unlikely to be correlated with unobserved factors (not included as explanatory variables) that directly influence children’s unmet needs/delayed care due to cost. We considered that illness among other family members could be associated with higher unmet need/delayed care for the index child because of a lack of attention on the child. However, the association with unmet needs/delayed care would be as a result of reasons other than cost and thus does not invalidate our instrumental variable. It is possible that these conditions among others in the family result in parent unemployment, reductions in income, and loss of health insurance, and directly influence children’s access to health care due to cost concerns. Therefore, we controlled for employment status of adults in the family, poverty status, and a rich set of health insurance controls both for the children and adults in the family.
We used STATA version 12 to account for the complex sampling design of MEPS and the correlation of observations both within families and over time.24–26
103
RESULTS Our sample included 63,462 observations (for 41,748 unique children) of which 11,235 (for 8337 unique children) were for children with SHCN. Table 1 lists selected characteristics of the children and their families. Among children without SHCN, the average child OOP was $220.30 and other family OOP was $1241.10. Among children with SHCN, the average child OOP and other family OOP were both higher, at $568 and $1663.30, respectively. The average unadjusted rate of any unmet needs/delayed care due to cost among children with SHCN was 5.4%; for those children without SHCN, the unadjusted rate was 2.6%. Table 2 shows the factors associated with unmet needs/ delayed care due to cost based on the instrumental variable analysis. Among children without SHCN, for each additional $1000 in child OOP, the adjusted rate of unmet needs/delayed care due to cost decreased slightly by 0.57% (SE 0.18%). For each additional $1000 in other family OOP, the adjusted rate of unmet need/delayed care due to cost increased by 1.69% (SE 0.41%). For children with SHCN, child OOP spending was associated with a lower rate of unmet need/delayed care due to cost, but this difference was not significant. The adjusted rate of unmet need/ delayed care due to cost increased by 1.89% (SE 0.73%) for each additional $1000 increase in other family OOP. Both for children without SHCN and for children with SHCN, private or public insurance coverage of the child (relative to not having insurance) and private dental insurance were associated with a reduction in any unmet need/ delayed care. Family income above 400% of FPL (relative to 200% to 400% FPL) was associated with a reduction in unmet needs/delayed care. As shown in the model predictions presented in Figure 1, an increase in child OOP from $40 (50th percentile) to $600 (90th percentile) was associated with a modest decrease in the rate of any unmet needs/delayed care from 2.74% to 2.34% (P ¼ .002) for children without SHCN, but the difference was not significant for children with SHCN (P ¼ .34). Model predictions in Figure 2 show that an increase in other family OOP from $500 (50th percentile) to $3000 (90th percentile) was associated with a higher adjusted rate of any unmet needs/delayed care due to cost (1.39% to 5.62%; P < .001 for children without SHCN; 3.17% to 7.87%; P ¼ .01 for children with SHCN). Table 3 reports the association between OOP measures and unmet needs/delayed care due to cost by specific type of care—medical, dental, and prescription drug—while adjusting for all other explanatory variables discussed. Among children without SHCN, higher child OOP was associated with a lower rate of unmet prescription drug and dental need or delayed prescription drug and dental care due to cost, suggesting, as before, that children with
104
KARACA-MANDIC ET AL
ACADEMIC PEDIATRICS
Table 1. Characteristics of Children and Their Families, 2002–2009† Estimate (SE) for: Characteristic Age, y Female, % No. of adults in family, mean No. of siblings in family, mean No. of other family chronic conditions, mean Family income
ABSTRACT OBJECTIVE: To assess the relationship between family members’ out-of-pocket (OOP) health care spending and unmet needs or delayed health care due to cost for children with and without special health care needs (SHCN). METHODS: Data come from the Medical Expenditure Panel Survey, 2002–2009, and include 63,462 observations representing 41,748 unique children. The primary outcome was having any unmet needs/delayed care as a result of the cost of medical care, dental care, or prescription drugs. We also examined having unmet needs/delayed care due to cost for each service separately. Key explanatory variables were OOP spending on the index child and OOP spending on other family members. We estimated multivariate instrumental variable models to adjust the results for potential bias from any unobserved factors that might influence both other family OOP costs and the outcome variable. RESULTS: An increase of other family OOP costs from $500 (50th percentile) to $3000 (90th percentile) was associated
with a higher adjusted rate of any unmet need/delayed care due to cost (1.39% to 5.62%, P < .001, among children without SHCN; 3.17% to 7.87%, P ¼ .01, among those with SHCN). Among children without SHCN, higher OOP costs among other family members were associated with higher levels of unmet needs or delays in medical, prescription drug, and dental care, while among children with SHCN, higher OOP costs among other family members was primarily associated with unmet or delayed dental care. CONCLUSIONS: Programs and policies that reduce the OOP costs of family members other than the child may improve the child’s access to care.
KEYWORDS: access; children’s health care access; children’s unmet needs/delayed needs; family out-of-pocket burden; out of pocket costs; unmet needs ACADEMIC PEDIATRICS 2014;14:101–108
WHAT’S NEW
ber (though the study did not distinguish between the child and adults).11 However, it is unknown whether the OOP health care costs of the child and his or her family members limit that child’s access to and use of needed health care services. A conceptual model of intrafamily resource allocation suggests that when faced with limited budgets, OOP health care costs may financially burden families and deter children’s receipt of needed health care even when the children are insured.12 For example, a child in a family with a sick parent who has high health care OOP spending may have unmet health care needs because the family’s budget constrains additional spending on health care. If children are at risk when their families are burdened by OOP spending, the policy solutions shift toward better coverage for all family members, not just for children. Decades of rising health care costs, together with the rising proportion of family income spent on health care, make it essential to investigate empirically the association of family health care financial burden and children’s receipt of needed health services. In 2001–2002, about 11% of American families spent at least 10% of their family income on OOP costs such as deductibles, coinsurance, and
Children are at risk of unmet needs/delayed health care when a family is experiencing high out-of-pocket burden. This finding shifts solutions to improve child health care access toward better coverage for family members rather than just for children.
OVER THE PAST decade, federal and state health policy efforts in the United States have focused on making sure that children have health insurance; these efforts have been largely successful.1–5 However, many children are underinsured and have health service needs that are not being met even with health insurance.6–8 Taking a family system perspective on health care access and utilization may provide insights into strategies for improving health care access for children. Within a family, parental insurance and access to a usual source of care are positively associated with children’s access to and use of health care services.9,10 Similarly, in families with children, out-of-pocket (OOP) spending for the entire family is shown to be positively associated with delayed or forgone care for at least one family memACADEMIC PEDIATRICS Copyright ª 2014 by Academic Pediatric Association
101
Volume 14, Number 1 January–February 2014
102
KARACA-MANDIC ET AL
copayments.13 From 2001 to 2006, the percentage of families that spent more than 10% of their income on health care increased.14 The recent recession has limited family budgets further.15 We examined OOP spending on children and OOP spending on their family members for the impact of such spending on whether the child has an unmet need or delayed health care due to cost, controlling for socioeconomic status, race, ethnicity, and health insurance. Our primary interest was to understand whether the OOP burden caused by other family members has consequences for meeting the child’s health care needs. Our study spans 2002–2009, which allows us to capture the effects of the recent economic downturn. Forgoing or delaying needed health care could have a long-term adverse effect on children’s health, even for healthy children, but especially for children with special health care needs (SHCN) who require specialized services.16–18 Therefore, we separately considered the impact on children with and without SHCN.
METHODS DATA We used the 2002–2009 Medical Expenditure Panel Survey (MEPS) Household Component—Full-Year Consolidated Data files and Medical Conditions files. The MEPS is a nationally representative survey of US households; it provides information on health care access and utilization, socioeconomic characteristics, employment, access to care, and related topics. The survey conducts 5 rounds of interviews covering 2 full calendar years and provides up to 2 separate full years of observation for each participant. We used the MEPS definition of a family, which consists of 2 or more persons living together in the same household.19 In MEPS, children with SHCN are identified by the children with SHCN screening instrument.20 OUTCOME VARIABLES We focused on unmet care needs or delayed health care, a commonly used critical indicator for access problems in children.16,17,21,22 The MEPS questionnaire inquires about medical care, dental care, and prescription medications, asking whether a given person was unable to receive care or medication; delayed in receiving either; and reasons for not receiving or delaying care. Responding yes to either of the first 2 questions indicated that the person was unable to receive or was delayed in receiving needed care. Possible reasons for not receiving or delaying care included the following: could not afford care; insurance would not approve/cover/pay for care; doctor refused the individual’s insurance; problems getting to doctor’s office; communications problems as a result of different language; could not get time off work; didn’t know where to go; was refused services; could not get child care; and did not have time.19 Our key outcome measure, a binary indicator for having any unmet needs or delayed care due to cost, was established
ACADEMIC PEDIATRICS
by an affirmative response to any of the questions about unmet needs or delayed care in either medical, dental, or prescription drug care with the reason related to costs. We also considered unmet needs/delayed care due to cost separately for each service. EXPLANATORY VARIABLES Our key explanatory variables were OOP spending on the index child (child OOP) and OOP spending on other family members (other family OOP) defined as the sum of OOP spending on family members other than the index child. In sensitivity analyses, we also calculated the OOP measures as a percentage of family total pretax income spent. All dollar measures were adjusted to reflect 2009 US dollars using the Consumer Price Index. We also controlled for child characteristics, including age, gender, race (white, black, Asian/Native Hawaiian/Pacific Islander, other), ethnicity (Hispanic, non-Hispanic), health insurance status, private dental insurance status, and health status (defined as excellent, very good, good, fair, or poor). The child’s health insurance status was categorized as uninsured all year, privately insured during any part of the year, or having only public insurance during the year. MEPS categorizes children with partial-year public insurance under the last category even if they were uninsured for part of the year. Moreover, we controlled for family characteristics, such as number of family members, federal poverty level (FPL) category, highest family educational attainment, employment status of adult family members, urban status, census region, and health insurance status of adult family members. STATISTICAL MODELS Using linear probability models, we first examined the binary outcome measure of any unmet need or delayed care due to cost as the dependent variable; child OOP, other family OOP, child and family characteristics were independent variables. Because the MEPS follows each child for up to 2 years, there were up to 2 full-year observations for each child. Multi-children families had separate observations for each child. We reported the absolute change in the rate of unmet need/delayed care due to cost per 100 children for each covariate. Then, we summarized the adjusted rates of any unmet needs/delayed care due to cost associated with different points in the OOP distribution (50th, 75th, 90th percentiles) in order to simplify interpretation of the results. Next, we estimated separate multivariate models of unmet/delayed care due to cost by need type: medical, prescription drug, and dental. We reported the absolute change in the rate of unmet need/delayed care due to cost per 100 children associated with the child OOP and other family OOP. In additional analyses, we investigated whether our findings were different during the recent recession (December 2007–June 2009) relative to the period before the recession. We added an indicator variable for the recession period (0 for 2002–2007 and 1 for 2008– 2009) for the interaction term of the child OOP and
ACADEMIC PEDIATRICS
FAMILY OUT-OF-POCKET HEALTH CARE BURDEN
the recession indicator, and for the interaction term of the other family OOP and the recession indicator. Negative and significant coefficient estimates on the interaction terms would suggest that the association between unmet needs/delayed care and OOP spending was weaker during the recession relative to the period before the recession. It is possible that unobserved factors could influence both a child’s unmet needs and family members’ OOP spending. If these unobserved factors were not accounted for, the results could be biased. For example, parents may have strong preferences for frequent and intensive health care, but this preference is not represented by a specific variable in the MEPS data. Such a preference could lead to a lower likelihood of an unmet need or delayed care for the child, while at the same time the preference would result in a high use of medical care and high OOP spending on other family members. Thus, a strong preference for frequent and intensive health care would result in high family OOP and low unmet needs of the child. In the case where there is a true positive relationship between family OOP spending and the higher likelihood of unmet needs for the child, such relationship would be understated. On the other hand, parents with a strong preference for frequent and intensive health care may be more likely to perceive and report unmet needs in general while having high OOP spending. This could bias the results toward overstating the relationship. Studies that do not specifically address this potential bias may inaccurately report associations. We sought to minimize the potential for bias and to strengthen our claim of causal inference by conducting our analysis using the econometric technique of instrumental variables.23 A valid instrumental variable needs to be correlated with the other family OOP and uncorrelated with unobserved factors that affect the outcome variable itself. We created our instrumental variable as the sum of 3 conditions among other family members: cancer, heart disease, and depression/anxiety disorders. This instrumental variable is valid because family members with these conditions have higher OOP spending and because the prevalence of these conditions among other family members is unlikely to be correlated with unobserved factors (not included as explanatory variables) that directly influence children’s unmet needs/delayed care due to cost. We considered that illness among other family members could be associated with higher unmet need/delayed care for the index child because of a lack of attention on the child. However, the association with unmet needs/delayed care would be as a result of reasons other than cost and thus does not invalidate our instrumental variable. It is possible that these conditions among others in the family result in parent unemployment, reductions in income, and loss of health insurance, and directly influence children’s access to health care due to cost concerns. Therefore, we controlled for employment status of adults in the family, poverty status, and a rich set of health insurance controls both for the children and adults in the family.
We used STATA version 12 to account for the complex sampling design of MEPS and the correlation of observations both within families and over time.24–26
103
RESULTS Our sample included 63,462 observations (for 41,748 unique children) of which 11,235 (for 8337 unique children) were for children with SHCN. Table 1 lists selected characteristics of the children and their families. Among children without SHCN, the average child OOP was $220.30 and other family OOP was $1241.10. Among children with SHCN, the average child OOP and other family OOP were both higher, at $568 and $1663.30, respectively. The average unadjusted rate of any unmet needs/delayed care due to cost among children with SHCN was 5.4%; for those children without SHCN, the unadjusted rate was 2.6%. Table 2 shows the factors associated with unmet needs/ delayed care due to cost based on the instrumental variable analysis. Among children without SHCN, for each additional $1000 in child OOP, the adjusted rate of unmet needs/delayed care due to cost decreased slightly by 0.57% (SE 0.18%). For each additional $1000 in other family OOP, the adjusted rate of unmet need/delayed care due to cost increased by 1.69% (SE 0.41%). For children with SHCN, child OOP spending was associated with a lower rate of unmet need/delayed care due to cost, but this difference was not significant. The adjusted rate of unmet need/ delayed care due to cost increased by 1.89% (SE 0.73%) for each additional $1000 increase in other family OOP. Both for children without SHCN and for children with SHCN, private or public insurance coverage of the child (relative to not having insurance) and private dental insurance were associated with a reduction in any unmet need/ delayed care. Family income above 400% of FPL (relative to 200% to 400% FPL) was associated with a reduction in unmet needs/delayed care. As shown in the model predictions presented in Figure 1, an increase in child OOP from $40 (50th percentile) to $600 (90th percentile) was associated with a modest decrease in the rate of any unmet needs/delayed care from 2.74% to 2.34% (P ¼ .002) for children without SHCN, but the difference was not significant for children with SHCN (P ¼ .34). Model predictions in Figure 2 show that an increase in other family OOP from $500 (50th percentile) to $3000 (90th percentile) was associated with a higher adjusted rate of any unmet needs/delayed care due to cost (1.39% to 5.62%; P < .001 for children without SHCN; 3.17% to 7.87%; P ¼ .01 for children with SHCN). Table 3 reports the association between OOP measures and unmet needs/delayed care due to cost by specific type of care—medical, dental, and prescription drug—while adjusting for all other explanatory variables discussed. Among children without SHCN, higher child OOP was associated with a lower rate of unmet prescription drug and dental need or delayed prescription drug and dental care due to cost, suggesting, as before, that children with
104
KARACA-MANDIC ET AL
ACADEMIC PEDIATRICS
Table 1. Characteristics of Children and Their Families, 2002–2009† Estimate (SE) for: Characteristic Age, y Female, % No. of adults in family, mean No. of siblings in family, mean No. of other family chronic conditions, mean Family income
