Insuring America's Health: Principles and Recommendations (2004)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

2 Lessons from Previous Reports In this chapter the Committee reviews findings from its five previous reports that document the health and financial repercussions of being uninsured for indi- viduals, families, communities, and the nation.1 The chapter is structured as a series of questions whose answers are fundamental to understanding why it is necessary to redesign our country’s overall approach to health insurance coverage. • The first section examines the scope of uninsurance and current coverage patterns by identifying how many people lack insurance, their basic economic and demographic characteristics, the ways people obtain and lose coverage, current barriers to coverage, and the potential for growth in the uninsured population in the near future. • The second section reports findings on how the lack of coverage affects access to and timely use of appropriate health care services, and adverse health outcomes for children and adults without health insurance. • The third section reviews family and community effects of uninsurance: financial repercussions for family budgets, the extent of uncompensated health care, and potential impacts of large uninsured populations on community access to care and on the economic and physical health of communities. • The fourth section outlines available evidence and projections of the cur- rent cost to the nation of uninsurance by looking at out-of-pocket expenditures of families, the cost of uncompensated care, an estimate of the value of life and health 1This chapter summarizes material presented in earlier reports. Evidence cited in the Committee’s previous reports is updated where newer data are available. 30 

LESSONS FROM PREVIOUS REPORTS 31 lost due to uninsurance, and the potential for offsetting the cost of extended coverage.2 • The fifth section of this chapter looks at how the structure of insurance can affect health care usage and health outcomes, for example, how employment- based insurance, differing eligibility rules for public coverage for persons within the same family, and the cost and availability of individual health insurance policies can result in coverage gaps. • The chapter concludes with a statement of the Committee’s perspective on health insurance in America. UNDERSTANDING THE SCOPE OF UNINSURANCE AND SOURCES OF COVERAGE Americans are often unaware of the characteristics of people who lack health insurance (IOM, 2001a). More than 80 percent of uninsured children and adults under the age of 65 live in working families and about the same percentage are U.S. citizens (Hoffman and Wang, 2003). People may lack coverage regardless of age, education, or state of residence. Nearly two-thirds of all uninsured persons are members of lower income families (earning less than 200 percent of the federal poverty level, or FPL), however. In this section the Committee reviews how many people are uninsured, how coverage is gained or lost, and the pathways and barriers to health coverage. Several alternative economic scenarios are also described, suggesting that without a fundamental change in national policy, the uninsured population is projected to continue growing. How Many People Lack Health Insurance and Who Are They? Estimates of the number of uninsured Americans depend on how uninsurance is measured. Increasingly, the lack of health insurance is understood as a condition for which virtually all Americans are to some extent at risk over the course of their lives, particularly at transitional points such as the age of majority or the loss of student status, rather than as a fixed characteristic of a well-defined segment of the population. Not all people, however, are equally at risk of being uninsured nor are all spells of uninsurance of equal length. During 2002, the Current Population Survey (CPS), conducted annually by the Census Bureau, showed that approximately 43.6 million people in the United States reported being without health insurance coverage for the entire year (Mills and Bhandari, 2003).3 Some analysts believe that the CPS estimate is closer to a 2Italicized terms are defined in the glossary (Appendix B). 3The Committee’s series of reports has relied on the CPS annual estimates of the number of uninsured persons in the United States. See Appendix A for a discussion of the features and limitations of various national surveys measuring insurance status. 

32 INSURING AMERICA’S HEALTH point-in-time count of uninsured Americans than it is of those uninsured for an entire year. In 2000 the CPS added a verification question to improve the accu- racy of its estimates, which reduced the estimate of the full-year uninsured popu- lation, although the CPS still probably overestimates full-year uninsurance. The CPS, however, provides the most consistent data on health insurance coverage over time and is the most widely used source of information on coverage. The Committee has used this historical series as its basic data set throughout the study for these reasons. Other data sets, the Survey of Income and Program Participation (SIPP) and the Medical Expenditure Panel Survey (MEPS), provide more precise information on length of time without coverage. An analysis of MEPS data found that one out of every three Americans under age 65, 80.2 million people, lacked health insur- ance for at least one month during a two-year period, while 23.5 million persons under age 65 were uninsured throughout that period (1996-1997), and 31.6 million were uninsured throughout 1996 (Short, 2001). A recent Congressional Budget Office publication compared SIPP and MEPS, which reported 21 and 31 million persons under age 65, respectively, who were uninsured for the entire year 1998 (CBO, 2003). The latest SIPP reports that, over the 48-month period calendar years 1996 through 1999, the overall median spell without health insurance lasted just under six months (Bhandari and Mills, 2003). Two separate analyses of MEPS (using some combination of data for 1996, 1997, and 1998) found that the average monthly count of uninsured persons for at least a one-year period is 45 million (Short, 2001; Hadley and Holahan, 2003a). People with low family income tend to remain uninsured for longer periods of time than those with incomes above the poverty level (McBride, 1997). Socioeconomic and demographic indicators help characterize those who go without health insurance and identify who is most likely to go without insur- ance. Full-time, full-year employment offers families the best chances of acquiring and keeping health insurance, as does an annual income of greater than 200 percent of FPL.4 While white, non-Hispanic people make up about half of the uninsured, minority group members have a higher risk of being uninsured. Afri- can Americans are nearly twice as likely as non-Hispanic whites to be uninsured, and Latinos are more than three times as likely as non-Hispanic whites to lack 4In 2003, 200 percent of FPL is $36,800 for a family of four (DHHS, 2003). (See Appendix A, Table A.1.) In this report, family income levels are defined as follows: • Low income: an annual income of less than 100 percent of FPL, which is established on a yearly basis for different types of family groups that comprise a given household, for example, one adult, or one adult and two children; • Lower income: an annual income of less than 200 percent of FPL; and • Moderate income: an annual income of between 200 and 400 percent of FPL for a given family group. 

LESSONS FROM PREVIOUS REPORTS 33 coverage. Although most uninsured persons under age 65 are U.S. citizens (79 percent), foreign-born, noncitizen residents are more likely than citizens to be uninsured (Hoffman and Wang, 2003). The uninsured rate declines with the length of time foreign-born persons are in the United States: 42 percent of noncitizens living in the United States for more than 5 years are uninsured, compared with 50 percent of noncitizens in the country for less than 5 years (Hoffman and Wang, 2003). How Do People Obtain and Lose Health Insurance Coverage? Most people choose to enroll in health insurance when it is offered on the job, including lower income workers and young adults who work. About two- thirds of Americans under age 65 are covered through employment-based plans offered at either their job or that of a parent or spouse (Fronstin, 2002).5 Typically employees and their employer share the cost of coverage. Nearly seven percent of Americans under age 65, including some lower income people, purchase their own individual or family policies from the private, nongroup insurance market. About 15 percent are covered by public insurance (primarily Medicaid) (Fronstin, 2002). When parents are insured, whether they are in single- or two-parent families, more than 95 percent of the time all of their children are also covered (IOM, 2002b). Medicare covers nearly all individuals over age 65 (Mills and Bhandari, 2003). Despite the variety of paths to coverage (employment, public programs, individual purchase), 17.2 percent of working-age Americans and children remain uninsured (Mills and Bhandari, 2003). Roughly one-fourth of workers have not been offered coverage by their employer, and half of these remain uninsured (Custer and Ketsche, 2000b). Some with a workplace offer report that they cannot afford the out-of-pocket or employee’s share of the premium. Insurance is becom- ing increasingly expensive for employers of low-wage workers and those with small firms to offer the benefit and for workers to accept coverage when it is offered (Thorpe and Florence, 1999; Chernew et al., 2002; Kaiser/HRET, 2003). For a cohort of adults between ages 21 and 60 that was followed for four years (1996– 2000), a job change or loss of a job was more often the reason for becoming uninsured than was the loss of public coverage. Those experiencing an uninsured spell were more likely than average to be young, African American or Latino, 5Some people report multiple sources of coverage, for example, they may have Medicaid for part of a year and a workplace policy at another time. Therefore, adding together employment-based, indi- vidual, and public insurance coverage rates yield more than the 83.5 percent of the U.S. population with some form of coverage during the year. 

34 INSURING AMERICA’S HEALTH How People Gain Coverage How People Lose Coverage • Get a job where insurance is offered and • Lose a job where insurance was offered, premiums are affordable so employer no longer subsidizes • Purchase insurance on your own, if you premiums qualify and can afford the premiums • Lose Medicaid or SCHIP eligibility once • Marry someone with insurance and if you or your children grow up or if your family out-of-pocket premiums are family’s income increases affordable • Lose your spouse due to separation, • Qualify for Medicaid, SCHIP, or divorce, or death Medicare • Attain the age of 19 or graduate from college and lose eligibility under parents’ plan • Your insurer goes out of business or cancels its contract with you, or your employer denies coverage to you • Be priced out of the market when the cost of premiums increases FIGURE 2.1 Gaining and losing coverage. SOURCE: IOM, 2001a. lower waged, less well educated (high school graduate or less), and from a lower income household (Kuttner and McLaughlin, 2003). Federal reforms, such as the Health Insurance Portability and Accountability Act of 1996, have established limited rights to purchase individual plans for those formerly covered by employ- ment-based insurance but do not regulate the size of the premium that might be charged (Nichols and Blumberg, 1998; IOM, 2001a). People can lose coverage when they change jobs or become unemployed, when life circumstances shift, or when rising premiums make insurance unaffordable (IOM, 2001a); see Figure 2.1. When a worker with employment- based coverage reaches age 65, retires, and qualifies for Medicare, a younger spouse may be left without coverage. When children turn 19 years old, generally the age limit for coverage as a dependent, they must purchase a separate, individual health insurance policy unless they are still in school or become uninsured. While teenagers or those graduating from college may be ready to go to work, they are less likely than their older coworkers to find jobs that include health benefits or to earn enough to purchase insurance independently (Quinn et al., 2000; IOM, 2001a; Collins et al., 2003). Some young, healthy people may choose to take the risk rather than buy coverage. Marriage is associated with job and career choices that lead to an increased likelihood of having employment-based health insurance for the whole family. Becoming separated, divorced, or widowed are other ex- amples of life transitions that can increase the risk that family members will lose their employment-based coverage. 

LESSONS FROM PREVIOUS REPORTS 35 What Are the Barriers to Insuring People Under the Current System? Health insurance eligibility, enrollment in a plan, and maintenance of enroll- ment depend on many interdependent factors, including the local labor market and health services, state regulatory and program policies, demographics, con- sumer knowledge, and personal choices (see Figure 2.2). Health insurance may or may not be available to an employer depending on its size and type of industry; a firm may or may not offer a policy that its employees perceive to be affordable; an individual applying for a nongroup policy may or may not be healthy enough to qualify or be able to afford the risk-adjusted premium; and public programs may have more or less restrictive eligibility standards. Every state and locality has a particular configuration of characteristics, including its industrial base, regula- tory environment, demographics, and public programs, that ultimately affects the opportunities for coverage and results in more or fewer people having health insurance. Affordability of Premiums As a group, those who lack health insurance most often share the perception that coverage is unaffordable. This is not to say, however, that health insurance becomes “affordable” across the board at a given premium cost or income level (Bundorf and Pauly, 2002; Levy and DeLeire, 2002). Although two-thirds of all people without coverage have incomes below 200 percent of FPL, some individu- als and families with relatively low incomes do take up employer offers of cover- age and some relatively high-income individuals and families forgo coverage (IOM, 2001a; Bundorf and Pauly, 2002). Nonetheless, unaffordability is the top reason uninsured adults give for why they are uninsured (Hoffman and Schlobohm, 2000), as well as the major reason employed persons turn down coverage when their employer offers it (Cooper and Schone, 1997; Thorpe and Florence, 1999; Cutler, 2002). Most uninsured families would not have sufficient funds in their budget to purchase health insurance without a substantial premium subsidy (IOM, 2001a, 2002a). The high cost of premiums is also the most common reason small firms give for not offering health insurance (Kaiser/HRET, 2003). Small employers often receive poorer benefits for premiums comparable to larger firms. Administrative costs and expenses other than benefits are usually 10 percent of premiums for large employers but 20 to 25 percent for small employers (GAO, 2001). Medical Underwriting and Denial of Coverage Medical underwriting practices applied to individual applicants for nongroup coverage are necessarily sensitive to an applicant’s health status, age, family in- come, and geographic area in order to protect the insurer from expected risks 

36 Insurance premium to firm Federal tax policy Employment-based insurance is offered Insurance premium to employee Employer characteristics (firm size, industrial sector) Work status Family income Individual is eligible Individual enrolls Health status Individual nongroup for coverage insurance is offered Public insurance Knowledge of Age; Family is offered eligibility composition Family income Administrative Public policies barriers to enrollment FIGURE 2.2 Factors affecting eligibility and enrollment. NOTE: Bold lines indicate major pathways and decision points. SOURCE: IOM, 2001a. 

LESSONS FROM PREVIOUS REPORTS 37 (Chollet and Kirk, 1998). Medical underwriting directly affects only those who purchase health insurance in the individual market. Nonetheless it is a significant barrier to coverage for some of the people most likely to benefit from coverage because of their relatively high expected use. Medical underwriting is inevitable in a competitive market with small and voluntary risk pools. Whereas young and healthy people face below-average premium prices, an older person or someone in poor health may face prohibitively high premiums. One study of eight nongroup insurance markets found that persons with health problems were quoted a pre- mium price nearly 40 percent higher than were otherwise comparable potential buyers without health problems (Pollitz et al., 2001). In addition, most states allow risk rating by age, making individual policies relatively expensive for older people (Chollet and Kirk, 1998; Blue Cross Blue Shield Association, 2000). Still, because older people highly value having health insurance, they are more likely to pur- chase it in the nongroup or individual market than are younger adults. Some 29 states have established high-risk pools for persons who are uninsurable in the individual market because their poor health puts them at risk for incurring large health care bills. However, there are often waiting lists or closed enrollments as well as high premiums (GAO, 1996; Achman and Chollet, 2001). Eligibility Restrictions for Public Programs Public programs such as Medicaid and the State Children’s Health Insurance Program (SCHIP) provide coverage for specific categories of the poor who tend to be excluded from the employment-based approach to financing health services delivery. The combination of strict eligibility requirements and complex enroll- ment procedures often makes public coverage difficult to obtain and even more difficult to maintain over time. Qualifying to participate in public programs in- volves fulfilling requirements related to income and assets (so-called means test- ing), being a member of a specific group that is eligible for benefits (e.g., pregnant women, minor children, or disabled), and meeting immigration status and resi- dency requirements. Eligibility requirements vary from state to state. Medicaid and SCHIP eligibility and enrollment policies are also subject to fluctuating fiscal conditions for the states that administer these programs, with more expansive eligibility and outreach to potential participants in better eco- nomic and fiscal times and cutbacks in eligibility and enrollment efforts during periods of fiscal stress (Howell et al., 2002; Smith and Ellis, 2002; Nathanson and Ku, 2003). Is the Uninsured Population Growing? Over the past 25 years, growth in the number of uninsured Americans has exceeded the rate of growth in the population under 65 years, despite an increas- ingly tight labor market that expanded employment-based coverage and yielded tax revenues to expand public coverage programs (IOM, 2001a). With the current 

38 INSURING AMERICA’S HEALTH combination of higher unemployment, rapidly rising costs of health care and insurance premiums, and state budget problems, absent major public policy re- forms, the national uninsured rate will rise more rapidly in future years (Chernew et al., 2002; Cutler, 2002). Looking forward, one study of the impact of different economic scenarios estimated that • “Assuming continued economic growth and moderate health care cost inflation, the number of uninsured Americans will rise to more than 48 million in 2009. • In the event of a recession, the number who lack coverage will reach 61 million by 2009. • Rapid economic growth, coupled with rapid health care cost inflation such as characterized the 1980s, would lead to roughly 55 million uninsured in 2009” (Custer and Ketsche, 2000, p. 3). Even without growth in the overall numbers, there is substantial variation in state and local uninsured rates, median durations of uninsured spells of individuals, and sizes and concentrations of uninsured groups within different populations (IOM, 2001a). This varied concentration of uninsured populations means that some adverse effects of uninsurance can be more severe in certain communities than would be expected from the national averaged data discussed in the section above. ASSESSING THE EFFECTS OF HEALTH INSURANCE ON HEALTH-RELATED OUTCOMES Isolating and measuring the independent effect of having or lacking health insurance is an analytic challenge because virtually all studies are observational and many characteristics that vary with health insurance status, including income, education, race and ethnicity, and health behaviors, also affect individual health outcomes. Figure 2.3 outlines the mechanisms by which health insurance influ- ences the amount and kind of health care received and a person’s health outcomes. Coverage facilitates receipt of health care services that can improve personal health. Uninsured people are less likely to have any medical contact and on average have fewer visits for care than people with either public or private coverage. Thus opportunities for detecting the presence of an illness or forestalling the progression of a chronic condition like diabetes are missed. Patients are also less likely to have a regular source of care to coordinate their health care or to have high-quality, evidence-based care. When the uninsured receive care in hospitals, their care management, even for trauma or premature birth, differs from that of insured patients, with uninsured patients receiving less intensive services (Hadley, 2002; IOM, 2002b). This section summarizes the Committee’s findings about the effect of health 

Obtaining Access to Health Care Health-Related Outcomes Individual and Family Level Resources (e.g., health Decision Making Process of Care Intermediate Health Outcomes insurance status, income) Outcomes •Subjective health • Individual, • Presentation of Characteristics about use of illness or • Timeliness of status Needs health services condition diagnosis • Functional/ • Provider (e.g., • Prevention • Severity of disability status differential • Management illness at • Clinical markers management) diagnosis of disease for specific conditions Community Level • Patient-provider communication • Quality, • General and Resources (e.g., health intensity, and disease-specific insurance coverage rates, site of care mortality safety net services) Characteristics Needs Effects on Communities: Health Disparities FIGURE 2.3 A conceptual framework for assessing the effect of health insurance status on health-related outcomes for adults. SOURCE: IOM, 2002a. 39 

40 INSURING AMERICA’S HEALTH insurance on a spectrum of health-related outcomes. These findings were drawn from methodologically sound research studies and natural experiments in which states extended health care coverage.6 Does Lack of Insurance Impede Access to and Use of Health Care? Without coverage, the uninsured—whether children, pregnant women, or other adults—receive fewer services than their insured counterparts or no care at all (IOM, 2002a, b). They are less likely to have any physician visit within a year or establish a “medical home” with a regular source of primary care. They receive, on average, fewer preventive health services, less regular care for management of chronic diseases, and poorer care in the hospital. Lesser Use and Lack of Preventive Health Services for Children and Adults Uninsured children use medical and dental services less frequently and are less likely to get their prescriptions filled than insured children, even after taking into account differences in family income, race and ethnicity, and health status (Newacheck et al., 1998b; McCormick et al., 2001; IOM, 2002b). They are less likely to receive routine, preventive well-child checkups and immunizations. Half of uninsured children have not had a doctor’s visit in the past year, more than twice the rate of privately insured children (McCormick et al., 2001). Adolescents as a group are particularly at risk of not having any physician visits in the past year or a regular source of care. Their need for some kinds of health care services, such as mental health screening and treatment for drinking and other risky behaviors, increases in their late teenage years, yet 17 percent of adolescents ages 15 to 17 are uninsured, the highest rate of all children (McCormick et al., 2001). Over one-quarter (27.4 percent) of adolescents ages 10 to 18 in families earning less than the federal poverty standard are uninsured (Newacheck et al., 1999). Forty-four percent of young adults aged 19-29 are uninsured at least part of the year. Though generally a healthy population, young adults are particularly vulnerable to injuries, HIV, and pregnancy, but when uninsured their regular access to the health system is disrupted (Collins et al., 2003). Uninsured women receive fewer prenatal care services than their insured counterparts and report greater difficulty in obtaining the care they believe they 6The Committee based its findings and conclusions on studies that adjusted for basic social, demo- graphic, and health characteristics. In health outcomes studies, unless otherwise noted, health insur- ance status was measured at baseline and does not reflect duration of coverage or period of uninsurance. Appendix B in Care Without Coverage and Appendix C in Health Insurance Is a Family Matter summarize the studies reviewed by the Committee. 

LESSONS FROM PREVIOUS REPORTS 41 need. In 1996 and 1997, 15 percent of uninsured pregnant women had no prena- tal visit compared with 4 percent of women with private or public coverage (Bernstein, 1999). Other studies find large differences in use between privately insured and uninsured women and smaller differences between uninsured and publicly insured women (IOM, 2002b). Uninsured women in labor are less likely to receive clinically indicated cesarean section deliveries for risks like breech presentation, fetal distress, or failure of labor to progress than are similar women who have coverage (Aron et al., 2000). Uninsured adults are also less likely to receive preventive health services such as mammograms, clinical breast exams, Pap tests, and colorectal screening (Powell-Griner et al., 1999; Ayanian et al., 2000; Breen et al., 2001). Less Access to a Regular Source of Care Maintaining an ongoing relationship with a specific provider who keeps records, manages care, and is available for consultation between visits is a key to high-quality care (O’Connor et al., 1998; IOM, 2001b). When children or adults have health insurance, they are more likely to have a regular provider who watches out for their health. People with insurance are more likely to see the same doctor at their usual source of care, and insurance status is the most important factor in white-Hispanic health differences in access to care (Hargraves and Hadley, 2003). The likelihood that someone who is uninsured lacks a regular source of care has increased substantially over the past several decades (Zuvekas and Weinick, 1999; Weinick et al., 2000). Even if people have health insurance that does not cover preventive services, they are more likely to receive appropriate services than are those without any form of health insurance, partly because they are more likely to have a regular source of care (IOM, 2002a). Uninsured children are more likely to obtain routine and sick care from different sources, such as hospital clinics and health centers, than insured children, who are more likely to be treated in the physician’s office (Holl et al., 1995). Well-child care and a regular care provider are important for monitoring children’s development and detecting potential problems early before they can cause long- term health consequences. Children without insurance were more than three times as likely as children with Medicaid coverage to have no regular source of care (15 percent versus 5 percent), and uninsured adults were more than three times as likely as either privately or publicly insured adults to lack a regular source of care (35 percent versus 11 percent) (Haley and Zuckerman, 2000). Uninsured children with special health needs are particularly disadvantaged because they require considerably more than routine care. One out of every nine children with special needs (roughly the same rate as among all children) remains uninsured (Newacheck et al., 1998a). Uninsured children with special health needs are less likely to have a usual source of care, less likely to have seen a doctor in the previous year, and less likely to get needed medical, mental health, dental or vision care, or prescriptions than are those with insurance. 

42 INSURING AMERICA’S HEALTH Identifying chronic conditions early and providing evidence-based, cost-ef- fective interventions on an ongoing and coordinated basis can improve health outcomes. Yet uninsured adults with chronic conditions are less likely to have a usual source of care than a chronically ill person with coverage (Ayanian et al., 2000; Fish-Parcham, 2001). Nineteen percent of uninsured adults with heart disease and 13 percent with hypertension lack a usual source of care, compared with 8 and 4 percent respectively of their insured counterparts (Fish-Parcham, 2001). Differential Management in Hospital-Based Care Uninsured patients who are hospitalized for any of a number of conditions are more likely to receive fewer services and, when admitted, are more likely to experience substandard care and resultant injury than are insured patients (Burstin et al., 1992). Some of the differences in care and outcomes between insured and uninsured hospital patients may stem from differences in the site of care. For example, a study of maternity patients in San Francisco revealed that privately insured women at a high risk of complications were much more likely to deliver at a hospital with neonatal intensive care facilities than were uninsured women (Phibbs et al., 1993). One statewide study found that uninsured sick newborns average shorter hospital stays and receive fewer inpatient services even after con- trolling for race, ethnicity, diagnoses, and hospital characteristics (Braveman et al., 1991). Uninsured adults can experience treatment differences as well. For example, uninsured women with breast cancer are less likely than privately insured women to receive breast-conserving surgery even when stage of diagnosis is taken into account (Roetzheim et al., 2000a). Uninsured patients with acute myocardial infarction who met expert panel criteria for revascularization were less likely to be transferred to a hospital that performed this procedure: 91 percent of Medicare patients, 82 percent of privately insured patients, 75 percent of Medicaid patients, and just 53 percent of uninsured patients received this indicated surgery (Leape et al., 1999). Does Lack of Insurance Impede Access to High-Quality, Evidence-Based Care? The difference health insurance makes is not just in increasing utilization but also in ensuring appropriate care. Although not all of the care that insured populations use is necessary and appropriate (IOM, 2001b), overall, the care received by those with coverage contributes to health outcomes better than those experienced by otherwise comparable uninsured populations who on average receive many fewer and less appropriate services. People without health insur- ance are at a disadvantage in obtaining high-quality, evidence-based care recom- mended by professional groups. When appropriate care is not obtained, patients’ health is placed at risk, conditions can become more severe, and the effects 

LESSONS FROM PREVIOUS REPORTS 43 can linger. Increased severity frequently demands more intense treatment (e.g., hospitalization). Less Likely to Receive Professionally Recommended Standard of Care Uninsured adults are less likely than those with health insurance to receive preventive and screening services at the frequencies recommended by the U.S. Preventive Services Task Force (Powell-Griner et al., 1999; Ayanian et al., 2000; Breen et al., 2001). Even after adjustments for age, race, education, and regular source of care, uninsured adults are less likely to receive timely screening for breast, cervical, or colorectal cancer. The positive effect of having insurance is more evident with relatively costly preventive services such as mammograms than with less expensive ones such as Pap smears, for example (Zambrana et al., 1999; Cummings et al., 2000). Uninsured adults living with chronic diseases are less likely to receive care for chronic health conditions that meets professionally recommended standards than are those who have health insurance. For example, uninsured adults with diabetes are less likely to receive regular foot or dilated eye exams that are important in the prevention of foot ulcers and blindness (Beckles et al., 1998). Uninsured patients with end-stage renal disease began dialysis at a later stage of their disease and with poorer clinical measures (e.g., more likely to be anemic) (Kausz et al., 2000). Persons diagnosed with a mental illness who lack health insurance are less likely to receive mental health services than are those with any health insurance (Cooper-Patrick et al., 1999; McAlpine and Mechanic, 2000). Furthermore, hav- ing health insurance that covers mental health services increases the likelihood that the care received will be in accordance with professional practice guidelines (Wang et al., 2000). Less Likely to Receive Medications That Are the Standard of Treatment Lack of insurance can interfere with access to appropriate medications for controlling medical conditions. For example, uninsured adults with HIV infection were less likely to receive the highly effective medications shown to improve survival and that have become the standard of treatment (Carpenter et al., 1996, 1998; Goldman et al., 2001). Uninsured adults had less frequent monitoring of blood pressure once they were diagnosed with hypertension and were less likely to stay on recommended drug therapy than insured adults with hypertension (Huttin et al., 2000; Fish-Parcham, 2001). Receiving Care in Less Appropriate Settings The lack of timely screening services and preventive care leads to poor health outcomes because of delayed diagnoses and failure to control treatable conditions. When they finally receive treatment, those without health insurance are more likely to require more expensive services because of deteriorating health. For 

44 INSURING AMERICA’S HEALTH example, uninsured patients are more likely to develop severe uncontrolled hyper- tension requiring emergency admission to the hospital than are insured patients (Shea et al., 1992a, b). Other conditions (e.g., asthma; ear, nose, and throat infections; pneumonia; diabetes) also are best treated early; without timely outpatient care, unnecessary hospitalizations occur. Expansion of SCHIP in Pennsylvania has shown that as the portion of children receiving physician visits increased, emergency room visits decreased (Lave et al., 1998). Expansion of Medicaid coverage also has been shown to reduce potentially avoidable hospitalizations among children (Dafny and Gruber, 2000). What Are the Health Consequences for Individuals and Their Families? Ascertaining whether health insurance improves health outcomes is critical to shaping public policy about health insurance and the financing of health care more generally. Uninsured people are more likely to receive too little medical care and to receive it too late, to be sicker, and to die sooner. Direct measures of health outcomes include self-reported health status, mor- tality, stage of disease at time of diagnosis, and physiologic measures (e.g., con- trolled blood pressure in persons with hypertension). Research studies consistently show that working-age Americans (those between 18 and 65) who do not have health insurance have poorer health and die prematurely. For children, their health is diminished and their long-term development is at risk when they are not covered. Diminished Health-Related Quality of Life Adults in late middle age are more likely to experience declines in function and health status if they lack or lose health insurance coverage (Baker et al., 2001). Changes in health status might include worsening control of blood pressure, decreased ability to walk or climb stairs, or decline of general self-perceived wellness and functioning. The effect of being uninsured on self-reported health measures is greater for lower income persons (Franks et al., 1993b). Another example is the deterioration in health-related quality of life reported for uninsured men with prostate cancer during treatment. They are more likely to have a delayed diagnosis, unlike men with either public or private insurance (Roetzheim et al., 1999; Penson et al., 2001). Developmental Risks for Children For children, using health care services routinely and appropriately is consid- ered a positive health outcome in its own right because well-child care has been demonstrated to be effective in enhancing longer term health and development. 

LESSONS FROM PREVIOUS REPORTS 45 Health conditions that are readily treatable and that could affect a child’s long- term development and life chances if untreated are more likely to go undetected when children are not insured. Conditions such as asthma, iron deficiency anemia, and middle-ear infections, if left untreated or improperly controlled, can affect mental development and school performance, language development, and hearing (IOM, 2002b). For example, there is an increased likelihood of mild or moderate mental retardation associated with iron deficiency anemia; approximately 9 per- cent of toddlers, 9 to 11 percent of adolescent girls, and 11 percent of women of childbearing age are iron deficient (Looker et al., 1997). Although long-term studies linking insurance status to these conditions and later life outcomes have not been conducted, uninsured children are at greater risk of such undetected condi- tions because of their lack of routine care. Increased Risk for Adverse Events For the five disease conditions that the Committee studied (diabetes, cardio- vascular disease, end-stage renal disease, HIV infection, and mental illness), unin- sured adults have consistently worse clinical outcomes than do insured patients. For example, uncontrolled blood glucose levels, which put diabetics at increased risk of hospitalization and additional complications (e.g., heart and kidney disease) and disability (e.g., amputations and blindness), are more frequent for uninsured adults (IOM, 2002a). In addition, uninsured pregnant women are more likely to have adverse maternal outcomes, such as pregnancy-related hypertension and placental abruption, than privately insured women. Improved maternal outcomes, how- ever, may require enhanced services such as counseling and other enabling services in addition to health insurance (Weis, 1992; Haas et al., 1993; IOM, 2002b). Decreased Life Expectancy for Newborns and Children Uninsured newborns are more likely than insured newborns to have poorer health outcomes such as low birthweight, which is a risk factor for developmental problems. Uninsured babies are also more likely to die prematurely. Measures across entire geographic populations, however, yield mixed evidence of improve- ment in health outcomes as a result of increased health insurance rates (IOM, 2002b). State experiments in Washington and California have found better birth outcomes in terms of birthweight and prematurity when additional services such as targeted case management, psychosocial and nutritional counseling, and other services were made available (Homan and Korenbrot, 1998; Salganicoff and Wyn, 1999). Insurance coverage as a result of Medicaid eligibility extensions in the 1980s and 1990s was associated with reductions in child mortality after the first year of life (Currie and Gruber, 1996a). At the same time, insurance coverage without enhanced prenatal services has not always yielded improvements in neo- natal outcomes (Baldwin et al., 1998; Marquis and Long, 2002). 

46 INSURING AMERICA’S HEALTH Pediatric trauma patients with private insurance have less risk of in-hospital mortality than the uninsured, even after adjusting for injury severity (Li and Davis, 2001). Uninsured children are estimated to be 40 percent less likely to receive medical attention for serious injuries than insured children, regardless of their race or ethnicity (Overpeck et al., 1997). Decreased Life Expectancy Among Adults The Committee has estimated that uninsured adults have age-specific mortal- ity rates approximately 25 percent higher than those of privately insured adults, based upon its extensive review of health outcomes studies and as estimated in longitudinal studies of overall mortality that adjust for multiple sociodemographic and health-related characteristics (Franks et al., 1993a; Sorlie et al., 1994; Hadley, 2002; IOM, 2002a). The Committee estimated that, for the year 2000, an esti- mated 18,000 excess deaths among adults between ages 25 and 64 could be attributed to lack of coverage (IOM, 2002a). The 18,000 excess deaths annually associated with uninsurance are comparable in magnitude to the 17,500 estimated deaths from diabetes and 19,000 deaths from cerebrovascular disease (stroke) in 2001 among the population under age 65. Also, there were 14,100 HIV deaths and 19,700 homicides among the total population that year. Uninsured people who have cancer, HIV, or cardiovascular disease or who are victims of car crashes have worse mortality experience than do insured people with these conditions. Uninsured cancer patients are more likely to die prema- turely than persons with insurance largely because of delayed diagnosis. This evidence comes from population-based research using area or statewide cancer registries for breast, cervical, colorectal, and prostate cancer and melanoma (IOM, 2002a). For example, uninsured women with breast cancer have a risk of dying that is between 30 and 50 percent higher than the risk for women with private health insurance (Ayanian et al., 1993; Lee-Feldstein et al., 2000; Roetzheim et al., 2000a), and uninsured patients with colorectal cancer are about 50 percent more likely to die than are patients with private coverage, even when the cancer is diagnosed at similar stages (Roetzheim et al., 2000b). Uninsured adults are at greater risk of premature death, reflecting the fact that they receive fewer screening services for serious illnesses such as cancer and less intensive and effective treatment for acute conditions such as traumatic injury and heart attacks. In one statewide study of hospitalized car crash victims, uninsured patients were found to receive less care and had a 37 percent higher mortality rate than did privately insured patients (Doyle, 2001). Having health insurance of any kind has been found to reduce mortality in HIV-infected adults by 71 to 85 percent over a 6-month period, with the greater reduction found for a more recent time period during which effective drug therapies were in more widespread use (Goldman et al., 2001). Uninsured patients with acute cardiovascular disease are less likely to be admitted to a hospital that performs angiography or revascularization procedures, are less likely to receive these diagnostic and treat- 

LESSONS FROM PREVIOUS REPORTS 47 ment procedures, and are more likely to die in the short term (IOM, 2002a). Health insurance reduces the disparity in receipt of these services for women relative to men and for members of racial and ethnic minority groups (Carlisle et al., 1997; Daumit et al., 1999, 2000). What Are the Health Effects of Discontinuous Coverage? Health insurance is most likely to improve health outcomes if coverage is continuous rather than intermittent and if it links people to appropriate health care. Adults with intermittent or no health insurance coverage experience greater declines in health status over time than do adults with continuous coverage. Being uninsured for relatively short periods (1 to 4 years) appears to result in a decrease in general health status (Lurie et al., 1984, 1986; Baker et al., 2001). If Americans were to become insured on a continuous basis, their health would be expected to improve. The survival benefits derived from insurance coverage, however, can be achieved in full only when health insurance is acquired well before the development of advanced disease. The problem of later diagnosis and higher mortality among uninsured women with breast cancer, for example, cannot be solved by insuring women once their disease is diagnosed (Perkins et al., 2001). Discontinuity affects access to health care and outcomes. For example, 25 percent of adults with diabetes went without a checkup within the past two years when they were uninsured for a year or more, compared to 7 percent of diabetics who were uninsured for less than a year and 5 percent of diabetics with health insurance (Beckles et al., 1998; Ayanian et al., 2000). Lacking health insurance for longer periods increases the risk of uncontrolled blood sugar levels in persons with diabetes, which, over time, put them at risk for additional chronic disease and disability. Similarly, discontinuity of insurance disrupts therapeutic relationships for persons with hypertension and worsens blood pressure control (Carlisle et al., 1997; Daumit et al., 1999, 2000). Children with gaps in health insurance coverage have less access to health services than do those with continuous coverage. For example, after expansion of SCHIP in New York, immunization rates rose, with the greatest increase among previously uninsured children and those who had a gap in coverage longer than six months (Rodewald et al., 1997). The improved immunization rate followed the shift in provision of immunizations from public health departments to health care providers whose coordination of care for the children served as a medical home for their patients. Generally, patients with private insurance have the best outcomes, while the uninsured have the highest proportion of late-stage diagnoses, with intermediate outcomes for Medicaid enrollees. However, some studies find that Medicaid enrollees’ outcomes are more comparable to the uninsured. The latter may stem in part from the frequent transitions to uninsured status that many Medicaid enrollees experience and the resulting interruptions likely to occur in the use of health 

48 INSURING AMERICA’S HEALTH services (Perkins et al., 2001; IOM, 2002a). Coverage is particularly episodic for lower income women. Medicaid enrollment periods for single women tend to be short and may depend on pregnancy status. More than half maintain enrollment for less than a year, not quite one-third last more than two years, and just 15 percent remain enrolled for five years (Short and Freedman, 1998). These medical consequences of coverage gaps will be discussed in the context of the structure of health insurance later in this chapter. FAMILY AND COMMUNITY EFFECTS OF UNINSURANCE Health insurance serves multiple constituencies and distinct purposes. For individuals and families, insurance is one means to promote health and plan for, if not prevent, exceptional health care costs. Providers of health care benefit from insurance as a reliable source of payment. Employers offer health benefits to attract workers and retain a satisfied and productive workforce. Communities benefit from a healthier population and potentially more stable health care institutions. Uninsurance can have effects that extend beyond uninsured persons and their families to community health care institutions, providers, and even the insured population. For example, a hospital outpatient department that serves an increas- ing number of uninsured patients without commensurate increases in financial support for uncompensated care may discontinue costly services, affecting access for everyone in the community. Although study of community-level effects of uninsurance is relatively recent, and the Committee has outlined additional needed research in this area, the evidence that is available justifies the immediate adoption of policies to address the lack of health insurance (IOM, 2003a). What Are the Financial Repercussions of Uninsurance for Families? Families with at least one uninsured member are predominantly lower in- come families. To be able to pay for rent, food, and other necessities, poor families have to decide between paying insurance premiums or forgoing coverage and paying out of pocket for health care services when necessary (Levy and DeLiere, 2002). Figure 2.4 shows how the cost of a full premium for an average employ- ment-based benefit for family coverage compares with family income. High Out-of-Pocket Expenses Despite Doing Without Care Because of concerns about the associated costs, uninsured families are parsi- monious in their use of health services. On average, families with some or all members uninsured are more likely to report using no health services. Those that use some services use fewer than do families with all members covered by private insurance (IOM, 2002b). Thus, it is not surprising that they spend on average less 

LESSONS FROM PREVIOUS REPORTS 49 49.3% Average Premium of Employment-based Coverage, If No Employer Subsidy ($9,068) 24.6% 16.4% 100% FPL 200% FPL 300% FPL ($18,400) ($36,800) ($55,200) FIGURE 2.4 Share of a four-person family’s income compared to premium costs to purchase family coverage in 2003. SOURCES: Kaiser/HRET, 2003; DHHS, 2003. on health care in absolute dollars. However, out-of-pocket health expenditures are a higher portion of family income than they are in insured families. In families with some members covered, out-of-pocket expenses relative to income still exceed that of fully insured families. Even in the healthiest of families, if one member has an accident or a costly hospital stay, the resulting medical bills can affect the economic stability of the whole family. Uninsured patients are likely to be charged more than those with insurance because they do not have a large insurer to negotiate a discount (Miller, 2003). The cost to a patient of hospital admission in 1999 for simple pneumonia in an adult might be $100 to $3,434 for a fee-for-service plan, but $9,812 for a person with no insurance (IOM, 2001a). As the time without health insurance lengthens, families continue to gamble with their financial security as well as their health. 

50 INSURING AMERICA’S HEALTH Even brief periods without coverage expose families to the financial risk of incur- ring extraordinarily high medical expenses for an incident like a car crash or treatment for cancer. Potential for Bankruptcy How do families cope with the burden of medical bills? Some families delay payment and may be dunned by collections agents. Families with uninsured members have fewer assets than do fully insured families and are less likely to have the capacity to borrow to cover major unexpected health care costs. More than half of all working-age adults uninsured currently or in the recent past reported difficulties paying medical bills, compared with less than a quarter of insured adults (Duchon et al., 2001). Of those with severe bill problems, two-thirds reported borrowing from family or a friend and a quarter needed a loan or mortgage on their home in order to pay (Duchon et al., 2001). Some families resort to declaring bankruptcy and put their future credit rating in jeopardy. Medical bills are a factor in nearly half of all bankruptcy filings, but it is not known whether bankruptcy is more likely for uninsured families than for those with coverage (Jacoby et al., 2000). Who Pays for the Care of Uninsured Persons? When a family is uninsured and cannot pay all its medical bills, the financial burden falls on the providers of services and on the broader community, through a variety of public and private mechanisms. Persons who are uninsured for the full year pay 35 percent, on average, of the overall cost of medical services they receive (Hadley and Holahan, 2003a; IOM, 2003a). The amount paid out of pocket varies by type of service. For example, uninsured individuals pay for nearly all (88 percent) of their prescription medications but for only 7 percent on average of hospital expenses they incur (AHRQ, 2001). Supports for care to uninsured patients come from federal, state, and local programs and grants; from organized philanthropy; and from the donated services and uncompensated care absorbed by providers. There is no uniform public responsibility to subsidize or pay for the care delivered to uninsured persons. The public sector financed between 75 and 85 percent of $35 billion in uncompensated care estimated to have been rendered to uninsured persons in 2001 (Hadley and Holahan, 2003a). The size of the lower income population (families with incomes less than 200 percent of FPL) in an area determines the relative need for public support for health care of all kinds (includ- ing both insurance and direct services) (Holahan and Pohl, 2003). The financing that exists is spread unevenly and varies by locality, not necessarily matching local needs for care (IOM, 2003a). The financial and organizational relationships be- tween safety net arrangements and mainstream health care imply that unreimbursed 

LESSONS FROM PREVIOUS REPORTS 51 expenditures for health services delivered to uninsured persons are borne by both public and private payers and by federal taxpayers as well as state and local ones. Private payers (e.g., employers, insurers) and private-sector health care pro- viders (e.g., nonprofit hospitals, physician practices) are widely assumed to cross- subsidize the costs of care for uninsured patients. Over the past 25 years, public policies to control health care costs and increasingly competitive health care markets have constrained payment rates. As a result, opportunities for private cross-subsidy of uncompensated care have eroded. The existence of such subsidies is difficult to document. Most analysts believe the opportunities for such cost shifting declined during the past decade (IOM, 2003a). Over the same period, the proportion of physicians who provided charity care (whether to uninsured or in- sured persons) declined, a decrease that is explained as a response to reduced provider revenues (Reed et al., 2001; Cunningham et al., 1999). Does the Size of the Uninsured Population Matter for Access to Care in a Community? The effects of uninsurance on communities have been felt most strongly in communities with large or growing uninsured populations, particularly in central urban neighborhoods and in rural areas, and in parts of the health care system such as public hospitals that serve many uninsured persons (IOM, 2003a). Programs to provide and pay for uninsured care are often stretched to their resource limits, with existing dollars outstripped by the perceived health needs of the population (Lewin and Altman, 2000; Felt-Lisk et al., 2001; IOM, 2003a). Health care providers who treat a large number of uninsured patients are likely to accumulate uncompensated costs that may impair their ability to continue delivering care. The Committee’s exploratory analyses of uninsurance effects on the availability of hospital services and hospital financial margins could not capture and do not reflect differences in the strategic behavior of hospitals in response to uninsured popula- tions, differences that affect the relative shares of uninsured patients at different facilities in the same community. Not all hospitals, even those in communities with high uninsurance rates, serve a high or growing number of uninsured pa- tients, and hospitals that do serve large numbers have varied experiences, depend- ing on corporate resources and the public supports and subsidies available to them (Gaskin, 1999; Catholic Hospital Association, 2002). A high uninsured rate in a community can result in reduced access to clinic- based primary care, specialty health services, and hospital-based care, particularly emergency medical services and trauma care (IOM, 2003a). It may also result in reduced availability of other primary and preventive care services and the loss of community health care providers, including the closure of hospitals. Access to health services and consequent benefits may be compromised for others within the community beyond those who lack coverage. Even for healthy community mem- bers, having more advanced medical services and resources available has real value. 

52 INSURING AMERICA’S HEALTH Reduced Availability of Clinic-Based Primary Care Services Serving a high or increasing number of uninsured persons reduces a commu- nity health center’s capacity to provide ambulatory care to all of its clients, insured as well as uninsured. The growing number of uninsured patients in community health centers, public health clinics, and other safety-net facilities may be due to referrals from other providers trying to reduce their uncompensated care burden (Hawkins and Rosenbaum, 1998). Community clinics face financial pressures with reduced income from Medicaid patients (especially due to managed care contracting), reduced public subsidies, and an increased portion of their popula- tion that is uninsured (Lewin and Altman, 2000). The loss of ambulatory care clinic services is particularly damaging to vulnerable populations because these clinics serve a higher proportion of low-income and minority group members under age 65 and have safety-net missions (e.g., serving farmworkers, homeless persons) and patients with specific high-risk diagnoses (e.g., tuberculosis, sub- stance abuse). A longitudinal study of 588 community health centers between 1996 and 1999 found that reductions in the scope of services and capacity to provide primary care were concentrated among health centers with a sizable uninsured population or a recent rapid increase in the number of uninsured clients that lessened the centers’ ability to provide primary care (McAlearney, 2002). Reduced Availability of Hospital-Based Care Many hospitals operate with financial constraints that leave little room for cross-subsidizing the costs of uninsured patients. In urban areas the adverse finan- cial effects on hospitals are more likely to have an impact on low-income residents (insured and uninsured) who have few options for care other than their local public or private safety-net hospitals. Inner city hospitals face pressure to cut their costs by cutting services. As a result, hospitals in urban areas with high uninsured rates have less total inpatient capacity, offer fewer services for vulnerable popula- tions (such as AIDS care), and are less likely to offer trauma and burn care (Gaskin and Needleman, 2003; IOM, 2003a). Hospitals in neighboring areas without so many uninsured may continue to operate as usual (Draper et al., 2001a). In rural areas, there is less opportunity to segment the market, and the adverse effects of high uninsured rates on hospital operating margins are likely to affect the availabil- ity of services for all community residents (Sutton et al., 2001). Hospitals in rural areas with higher uninsured rates have fewer intensive care unit beds, offer fewer psychiatric inpatient services, and are less likely to offer high-technology services such as radiation therapy (IOM, 2003a; Needleman and Gaskin, 2003). Emergency departments (EDs) and trauma centers are key examples of how market pressures and public policies interact in ways that create incentives for hospitals to reduce their exposure to financial losses associated with an “open door” policy to serve all comers without regard to their ability to pay. Rising use rates by the uninsured can worsen ED overcrowding and the financial status of ED 

LESSONS FROM PREVIOUS REPORTS 53 operations. A recent survey of 1,501 hospitals finds that 62 percent report their EDs to be at or over capacity, often due to the inability to move seriously ill or injured patients into inpatient beds (Lewin Group, 2002). A survey of urban hospitals by the General Accounting Office finds that many are ill equipped to handle a large number of patients, such as the number who would be seen with a bioterrorism incident (GAO, 2003). By limiting the number of inpatient beds available, hospital administrators are better able to control admissions. When a backup is severe, patients may be diverted to other facilities. In the mean time, the ED is unavailable to the insured and uninsured alike. Members of medically underserved groups are particularly likely to suffer reduced access because they have fewer options to obtain primary care outside EDs (Felt-Lisk et al., 2001; Weinick et al., 2002). Financial stresses for hospitals have also increased the difficulty of recruiting medical specialists to serve on on-call panels for emergency rooms as required under the federal Emergency Medical Treatment and Labor Act (EMTALA). For example, in Phoenix, AZ, which had a 17 percent uninsured rate in 2001, spe- cialty physicians such as orthopedists and neurosurgeons ended their affiliations with emergency departments because of the expected lack of compensation due to the high number of uninsured patients (Draper et al., 2001b). Similarly, a high proportion of uninsured patients places significant financial stress on regional trauma centers. Trauma center patients are already more likely to be uninsured than all hospital patients (Eastman et al., 1991). Hospitals may decline to open a trauma center or may decide to close one in response to this financial stress. The closure of a regional trauma center puts the health of everyone in a community at risk, whether insured or uninsured. Difficulty Obtaining Referrals for Specialty Health Services Relatively high uninsured rates are also associated with the decreased ability of primary care providers to obtain specialty referrals for patients who are mem- bers of medically underserved groups. For example, insured and uninsured rural residents already experience reduced access to specialty care. Increasing numbers of uninsured patients prompt more providers to leave the community because their practices are not financially viable (Ormond et al., 2000). A study of primary care providers in 20 community health centers in 10 states revealed difficulties for practitioners in high uninsured-rate areas in obtaining specialty referrals for all of their patients, not just the uninsured (Fairbrother et al., 2002). Urban safety-net hospitals and academic health centers also have been affected by increased num- bers of uninsured patients. Under financial pressure, some facilities have cut out specialty services with poor rates of reimbursement, such as burn units, pediatric and neonatal intensive care units, and HIV/AIDS care (Gaskin, 1999; Common- wealth Fund, 2001). 

54 INSURING AMERICA’S HEALTH Does the Size of the Uninsured Population Affect a Community’s Economic and Fiscal Health? Although it is clear that the economic vitality of a community affects insur- ance coverage rates, the Committee also hypothesizes the converse—that health insurance coverage rates can affect the economic fortunes of a community, prima- rily through financial impacts on health services providers and institutions (IOM, 2003a). A high uninsured rate likely reflects fewer employers offering coverage and a lower wage labor force. The tenuous financial viability of local health providers due to uninsurance additionally affects both service delivery and jobs. Each of these puts financial stress on communities that try to maintain services through higher local prices for health care services and higher local taxes. Alter- nately, cutting local support can result in more uninsured in the area and reduced service availability for all community residents. During economic downturns, when growth in uninsured populations in- creases the demand for uncompensated care, state and local reductions in health care spending may reduce the flow of federal dollars (such as Medicaid matching funds) into a community. Local community institutions and programs tend to serve as providers of last resort. The dispersion and concentration of uninsured populations across regions, states, counties, and cities and across urban, suburban, and rural areas vary widely (IOM, 2003a, Appendix B). While the national average uninsured rate for Ameri- cans under 65 years is 17.2 percent, the rates for the states range from 8.8 percent in Minnesota to a high of 28.4 percent in Texas (Mills and Bhandari, 2003). Within a single state there can be considerable variation. Among the 67 counties in Florida, the uninsured rate varies from 12 percent to 30 percent uninsured, with many counties significantly different from the state’s overall rate of 16.8 percent (Lazarus et al., 2000).7 This variable concentration implies that the adverse effects of uninsurance will be more or less severe locally than national or even state-level uninsured rates would suggest. Compared with the tax burden of federally or federal-and-state financed health insurance (Medicare or Medicaid and SCHIP), the costs of uncompensated care for those without coverage fall more heavily on local health care providers and taxpayers (IOM, 2003a). Communities that are disproportionately affected by uninsurance, for example, those with a preponderance of lower-wage and service- sector jobs, likewise have a smaller tax base with which to address the health care needs of uninsured residents. State-level efforts can make a significant difference in 7The estimates of uninsurance for Florida and its counties were devised using methods and sample sizes different from the Census Bureau’s Current Population Survey, which is the basis for the discussion of the range of uninsured rates among the states. Because of these differences, the estimates given for Florida, and for Colorado and Minnesota in the example that follows, are not comparable to CPS estimates. See Appendix B in A Shared Destiny for more details (IOM, 2003a). 

LESSONS FROM PREVIOUS REPORTS 55 the rate of uninsurance. In two states with comparable portions of residents lacking employment-based coverage (22 percent for Colorado and 18 percent for Minnesota), the state of Minnesota enrolled 45 percent of persons without em- ployment-based coverage in public insurance compared to Colorado, which en- rolled 19 percent. As a result, the uninsured rate is 10 percent in Minnesota, compared with 17 percent in Colorado (Holahan, 2002). Does the Size of the Uninsured Population Affect the Physical Health of a Community? The sheer number of uninsured persons in an area adds to the community burden of disease and disability because uninsured persons are more likely than their privately insured counterparts to have poorer health, to be at greater risk for some communicable diseases, and to draw on public health resources (IOM, 2003a). Worse Health Status and More Preventable Hospitalizations Geographic differences in self-reported health status among the states corre- late with state uninsured rates (Holahan, 2002). For urban, suburban, and non- metropolitan communities across the country, uninsured rates also correlate with the health status reported by residents. Community uninsurance rates converge with a number of other factors that affect access to care and health status, such as the proportion of the population that is lower income and the proportion that consists of racial and ethnic minorities (Shi, 2000, 2001; IOM, 2001a, 2002a). Although these geographic disparities in health status are certainly not entirely attributable to the lack of coverage, the uncompensated care demands on health care providers and population health impacts are greater in communities with high uninsured rates (IOM, 2003a). Potentially avoidable hospitalizations (sometimes called ambulatory-care-sen- sitive conditions) serve as an indicator of adequate access to primary and regular care. Uninsured patients are more likely to experience avoidable hospitalizations than are privately insured patients when measured as the proportion of all hospi- talizations (Pappas et al., 1997). Hospitalization rates for potentially avoidable hospitalizations are higher in communities that include greater proportions of lower income and uninsured residents, indicating both access problems and greater severity of illness (IOM, 2003a). Increased Risk for Communicable Diseases Areas with relatively high uninsured rates are likely to have a greater burden of vaccine-preventable and communicable diseases and disability. For example, underimmunization increases the vulnerability of entire communities to outbreaks of diseases such as measles, pertussis (whooping cough), flu, pneumonia, and other 

56 INSURING AMERICA’S HEALTH diseases (IOM, 2000). Childhood and adult immunization levels are positively correlated with having either public or private health insurance. When uninsured children received coverage under SCHIP in New York state, statewide immuni- zation rates for young children increased (Rodewald et al., 1997). Increased Burden on Public Health Resources Competing demands on state and local health departments as providers of last resort and as guardians of public health can adversely affect their ability to perform both functions adequately (GAO, 2003; IOM, 2003a,c). Public health functions include disease and immunization surveillance, community-based health educa- tion and behavioral interventions, emergency preparedness, and environmental health. The need for population-based public health services is expected to be greater now; at the same time there are growing demands on health departments to provide or pay for safety-net services for the uninsured. A recent survey finds that more than one-quarter of local health departments serve as the only safety-net provider in their community (Keane et al., 2001). Budgets for population-based public health activities that benefit all members of the community frequently are squeezed by these demands. CURRENT COSTS AND SOCIETAL COSTS ATTRIBUTABLE TO UNINSURANCE As a society, we directly invest in the health of those who have health insurance through tax subsidies for private insurance and publicly sponsored cov- erage. We also spend substantial public resources for direct health care services for those who lack coverage, yet the uninsured continue to have worse health out- comes. By estimating the health services costs now incurred by the tens of millions of uninsured Americans and some of the incremental costs and benefits across society of extending coverage, we provide an economic baseline against which health insurance reform strategies can be measured. What Are the Health Services Costs Now Borne by the Uninsured Themselves? As described earlier in the chapter, the uninsured use fewer health services yet have higher average out-of-pocket health expenses as a proportion of family income when they do use services. When uninsured people use health care services, they are often charged substantially more than are insured patients, whose insurance company has negotiated discounts (Lagnado, 2003). Among families with no members insured during the entire year and incomes below the poverty level, more than one in four had out-of-pocket expenses that exceeded 5 percent of income in 1996 (Taylor et al., 2001b). Families with no insurance for any of its members for the full year were nearly twice as likely to exceed the 5 percent 

LESSONS FROM PREVIOUS REPORTS 57 TABLE 2.1 Out-of-Pocket Expenses as Percentage of Family Income, by Insurance Coverage and Duration, Non-Medicare Families, 1996a Families Exceeding Threshold (percent) 5% or More 10% or More of Income of Income All members insured for entire year 8.8 3.0 Some members uninsured and/or some period without health insurance 10.7 4.6 All members uninsured for entire year 15.4 8.0 All families 10.0 4.0 aThese out-of-pocket expenses cover medical services; they do not include insurance premiums. SOURCE: Medical Expenditure Panel Survey 1996 data across families of all sizes in Merlis, 2001. threshold of out-of-pocket expenditures as were fully insured families in 1996 (Merlis, 2001; see Table 2.1). Paying medical bills can have profound, long-lasting economic and social effects on uninsured families (IOM, 2002b). If the uninsured were to gain coverage, the change in their out-of-pocket costs would depend on both the scope of benefits and cost-sharing requirements. What Are the Costs of Uncompensated Care Used by the Uninsured? The best estimate of the value of uncompensated health care services provided to persons who lack health insurance for some or all of a year is roughly $35 billion annually, about 2.8 percent of total national spending for personal health care services (Hadley and Holahan, 2003b). This estimate includes the value of free hospital, physician, and clinic services that the uninsured use annually, adjusted to reflect spending in 2001. About two-thirds of this uncompensated care ($23.6 billion) is public subsi- dies to hospitals, paid through federal Medicaid and Medicare disproportionate share hospital (DSH) adjustments and other financing mechanisms, state Medicaid DSH payments, and other state and local appropriations for the support of hospital services and operating costs. The public also supports a variety of governmental grant and direct care programs such as Community Health Centers, National Health Service Corps, Department of Veterans Affairs, and local health depart- ments, amounting to about $7 billion annually. Donated physician time and 

58 INSURING AMERICA’S HEALTH services account for the remaining $5 billion of uncompensated care (Hadley and Holahan, 2003b). If the uninsured were to gain coverage, some of the current spending for their health care would be redistributed among payers, for example, from physicians, who now provide uncompensated care, to taxpayers. Hospital bad debt and char- ity care caseloads would decline. As mentioned previously, the public sector already finances between 75 and 85 percent of the uncompensated care burden for uninsured individuals. Depending on the particular plan design, some of that amount could be reallocated to health insurance. What Is the Cost of the Worse Health and Shorter Lives of Uninsured Americans? In Hidden Costs, the Committee adapted an analytic strategy similar to that used by government agencies to estimate the benefits of life-saving and health- improving safety and environmental interventions in order to assess the economic losses borne by uninsured individuals as a result of their poorer health outcomes relative to those of insured counterparts. This analysis, commissioned by the Committee from economist Elizabeth Vigdor, is the first modeling exercise of its kind to evaluate the health benefits of coverage and is included in its entirety as Appendix B of Hidden Costs (2003b). Vigdor assigned an economic value to health by imputing a monetary value to a healthy life year and calculated the average difference in the present value in money terms of expected years of life in particu- lar states of health between otherwise similar insured and uninsured populations. This difference constitutes an estimate of the economic value lost as a result of the current level of uninsurance within the U.S. population. Conversely, it is an estimate of the economic value of the better health outcomes that could be realized if the entire population had continuous health insurance coverage. Based on this analysis, the Committee estimates that, in the aggregate, the diminished health and shorter life spans of Americans who lack health insurance are worth between $65 and $130 billion for each year spent without health insurance (IOM, 2003b). This estimate does not include spillover losses to families and society as a whole of the poorer health of the uninsured population. It does account for the value of those experiencing poorer health, including individual losses in work effort and developmental losses due to poor health in children. If the uninsured were to gain coverage comparable to that of the currently insured population, this $65–$130 billion in “health capital” would be an economic benefit rather than a cost. The Committee’s calculation of the economic value of improved health and longer life is likely an underestimate of the actual health benefits of continuous health insurance, in part because it only includes the effects up to age 65. Addi- tional positive effects on health and longevity after age 65 also would be likely if health insurance were continuous before this age. In addition, there could be 

LESSONS FROM PREVIOUS REPORTS 59 savings to the Medicare program. As it is now, there is likely to be pent-up demand for services at age 65 among those previously uninsured (IOM, 2003b). Are Additional Costs Associated with Uninsurance? Although the Committee could not develop specific dollar estimates, other public programs such as Social Security Disability Insurance and the criminal justice system likely have higher budgetary costs than they would if the U.S. population in its entirety had health insurance up to age 65 (IOM, 2003b). Disability insurance claims could decrease with the health and functional status improvements that health insurance accords. Increased productivity in the workplace could also accompany increased population coverage rates. Studies have shown that productivity is lost on the job when workers have particular illnesses. Separate studies suggest that workers’ health status can improve as a result of having coverage. However, the effects of coverage on workplace productivity have not been studied systematically or in any controlled fashion. As already discussed above, persons with either private or public insurance are more likely than those without any coverage to receive appropriate treatment for mental health problems. More than 3 million adults have serious mental illness that can involve psychosis and aberrant behavior; 20 percent of these adults who do not reside in institutions lack health insurance. Between 600,000 and 700,000 persons with severe mental illness are jailed each year. Access to effective treat- ment prior to incarceration would be expected to reduce criminal justice ex- penses. Ironically, once people with serious mental illness have contact with the criminal justice system, they have an increased chance of obtaining access to specialty mental health services (McAlpine and Mechanic, 2000). What Would the Additional Services Cost That the Uninsured Would Be Expected to Use if They Gained Coverage? Closing the access gap for the uninsured would mean increased utilization of services. As described earlier, the uninsured population is less likely to use any kind of health service within a given year, and on average the uninsured person uses one-half to two-thirds of the volume and value of services that the privately insured person uses (IOM, 2003b). The Committee reviewed several sets of estimates of the value of the additional health services that would be provided to the uninsured once they became insured. Estimates of the additional costs of health services that the population that now lacks insurance could be expected to use if they gained coverage range from $34 billion to $69 billion a year in 2001 dollars. This range reflects the difference between the average per capita expendi- ture within public insurance programs (primarily Medicaid) and that for popula- tions with private health insurance (Hadley and Holahan, 2003a). This range of estimated costs amounts to between 2.8 and 5.6 percent of national spending for 

60 INSURING AMERICA’S HEALTH personal health care services in 2001, equivalent to roughly half of the 8.7 percent increase in personal health care spending between 2000 and 2001 (IOM, 2003b). These estimates do not reflect the costs of any particular plan to provide health insurance to the uninsured, nor do they include any costs of establishing a minimum benefit package that would affect the currently insured population. The cost range of $34 billion to $69 billion, which encompasses the results of three independent analyses, assumes no structural changes in the systems of health services delivery or finance, scope of benefits, or provider payment from those that currently operate in the public and private sectors (2003b). The defined benefit package for covering the uninsured would influence the nature of benefits offered to the currently insured. If it were a relatively generous package and there were no other structural changes, the costs of additional services would be greater than the $69 billion because of greater use by the currently insured as well as new utilization by the previously uninsured. Ultimately, the full cost of any reform will depend on the specific features of the approach taken, an estimate of which is beyond the scope of this report. In its previous report, the Committee assessed the individual health and longevity benefits of continuous health insurance coverage for the uninsured population relative to the costs of providing this population with the kind and amount of health care used by comparable insured populations (IOM, 2003b). The Committee concluded that the economic value that would be gained in terms of better health outcomes among those now uninsured would likely exceed the incremental resource costs of providing the uninsured with the level of services now used, on average, by demographically similar people with either public or private coverage. In addition, the Committee’s estimate of providing this “in- sured” level of health care services to those who now lack coverage compares favorably with other societal investments in improving health and extending life (IOM, 2003b). EFFECTS OF THE STRUCTURE OF INSURANCE The current amalgam of service arrangements and the mix of public and private insurance sources were not designed as an integrated system; rather, they have resulted from the aggregation, over time, of initiatives and developments in both the private and public sectors. A variety of factors related to the terms and structure of insurance affect eligibility for and affordability of insurance. Current insurance mechanisms and programs do not match the needs of all persons over time. Life-course and employment transitions, in particular, result in gaps in coverage (Kuttner and McLaughlin, 2003). Coverage Gaps Associated with Employment-Based Coverage Nearly two-thirds of all firms offer health benefits to their employees (66 percent), with offer rates ranging from 55 percent for small firms employing 3 to 9 workers to 98 percent for firms employing 200 or more workers 

LESSONS FROM PREVIOUS REPORTS 61 (Kaiser/HRET, 2003). Most employees take up the offer of coverage or obtain coverage through a family member’s workplace health insurance plan. The link between health insurance and employment, however, creates many opportunities for loss of coverage. Job loss and retirement increase the risk of losing coverage. Work choices may be constrained for those with private coverage by the need to obtain and maintain health benefits with the current job (sometimes referred to as job lock). Even people who receive public insurance coverage may be limited in their job choices because of means testing for public benefits (IOM, 2002b). Many of the uninsured, however, are not eligible for the plan offered where they work or work in settings that do not offer any plan. During the mid-1990s, only 55 percent of workers who earned less than $7 per hour were offered employment-based insurance compared with 96 percent of workers whose hourly rate was above $15 per hour (Cooper and Schone, 1997). This can result in inequities in coverage even if total family income is the same. A family having a single wage earner with a salary of $50,000 is more likely to have access to health insurance than is a family of two wage earners, each of whom earns an annual salary of $25,000.8 So while working improves the chances of coverage, even members of families with two full-time wage earners have an 8 percent chance of being uninsured (Hoffman and Wang, 2003). The prime economic force behind the declining portion of Americans cov- ered by employment-based insurance is the gap between workers’ purchasing power and increases in health services costs and costs of purchasing insurance (Cooper and Schone, 1997; Holahan and Kim, 2000; Cutler, 2002). In constant 1998 dollars, the cost of employment-based insurance increased 260 percent be- tween 1977 and 1998 and the employee’s share of insurance premiums increased 350 percent (Gabel, 1999). During the same period, median household incomes increased in real terms by 17 percent (U.S. Census Bureau, 2000). Health care cost increases exceeded growth in the general economy due to factors that included technology changes and increased use of services per capita, including prescription drugs (Glied, 2003). A recent econometric analysis by David Cutler (2002) con- cluded that virtually all of the decline in employee take-up rates between 1988 and 2001 (for full-time male workers, from 94 to 90 percent take-up of offers of coverage) could be attributed to increases in the employee share of premiums over this period. 8The distinction between the income of a “health insurance unit,” i.e., family members who qualify for coverage together, and that of a household is important to the interpretation of coverage trends. For example, some have concluded that the increase of 1.4 million uninsured persons between 2000 and 2001 in the CPS represents an influx of middle-income persons, because about 800,000 of them lived in households with yearly incomes of at least $75,000 (Mills, 2002). Analyzing these estimates in terms of health insurance units, however, and accounting for changes in household composition supports a different conclusion: an increasing number of people with lower incomes joined households that earned over $75,000 annually, and almost all of the newly uninsured in 2001 (1.3 million people) were in families that earned less than twice the federal poverty level (Holahan, et al., 2003b). 

62 INSURING AMERICA’S HEALTH Coverage Gaps Due to Differing Eligibility Rules for Members of the Same Family Achieving coverage of the entire family can prove difficult (IOM, 2002b). Private insurers often restrict the definition of family to a traditional family struc- ture. This mismatch between insurers’ eligibility criteria and a functional family unit affects coverage. Most publicly financed health insurance programs are even more restrictive because they provide coverage for individuals rather than for families. Lower income parents are more likely to lack coverage than are their children because public programs provide coverage for children up to higher family income levels than they do for adults. Public programs also tend to have more generous family income limits for younger children than older ones, with the result that uninsurance rates are higher among older children (Hoffman and Wang, 2003). Furthermore, simplification of eligibility rules and enrollment processes can reduce barriers to coverage. More than half of the 7.8 million children uninsured in 2002 were eligible for Medicaid or SCHIP coverage (Kenney et al., 2003). Parents’ decisions on whether, when, and from whom to seek care for their children may be influenced by their own experiences with and knowledge of the health system (IOM, 2002b). When states have expanded Medicaid programs broadly to include low-income parents as well as their children, the enrollment of eligible children has increased more than it has in states without the broader parental coverage (Ku and Broaddus, 2000). Gaps Because of the Cost of Coverage The national average total annual premium for a family policy in an employ- ment-based group exceeded $9,000 in 2003, with workers themselves picking up, on average, 27 percent of the cost of family coverage (Kaiser/HRET, 2003). In firms with at least 35 percent low-waged workers, employees pay a greater portion of the premium, typically 36 percent of the premium—an extra $68 per month compared with the national average (Kaiser/HRET, 2003). For a worker earning $20,000 per year, roughly $10 per hour, the employee’s premium share for family coverage would take more than 16 percent of his or her income before taxes. When it is available, individual coverage is often a stop-gap measure for adult children who lose their coverage as dependents before they can obtain job-based coverage and for retirees under the age of 65 before they become eligible for Medicare. A study comparing premiums for individual and group insurance plans with comparable benefits (comprehensive, preferred provider organization) in 17 health insurance markets found that for young adults (27 years old), the median premium for individually purchased (nongroup) coverage is roughly comparable to the premium for group coverage, especially for men, but that the median 

LESSONS FROM PREVIOUS REPORTS 63 premium for individually purchased coverage for 55-year-olds is more than twice the premium for group coverage (Gabel et al., 2002). Underwriting assumptions account for some of the differences in premiums that employers and employees face (IOM, 2001a). The price of an insurance premium that the insurer offers to a firm reflects a number of considerations, including firm size, whether it is unionized, the employment sector, and any risk or experience rating of the employees. Employers have sought ways to economize by increasing employee premiums and cost sharing, dropping coverage for retirees, and restricting benefit packages. Gaps in Service Coverage and Their Effects Lack of coverage affects the availability of care across the spectrum of preven- tive health services, chronic disease care, medications, mental health, acute care, emergency room treatment, and hospital care. When health insurance includes preventive and screening services, prescription drugs, and mental health care as well as acute and diagnostic care, it is more strongly associated with the receipt of appropriate care than when insurance does not have these features (IOM, 2002a). Generally, insurance benefits are less likely to include preventive and screen- ing services than physician visits for acute care or diagnostic tests for symptomatic conditions. A positive and statistically significant “dose-response” relationship has been found between the extent of coverage for preventive services and their receipt (Faulkner and Schauffler, 1997). Yet as long as people have some form of health insurance, even if it does not cover preventive services, they are more likely to receive appropriate services, partly because they are more likely to have a regular source of care or a primary provider (IOM, 2002a). Adults with mental health coverage are more likely to receive mental health services from both general medical and specialty mental health providers and to receive care consistent with clinical practice guidelines than are those without any health insurance or with insurance that does not cover mental health conditions. Receipt of appropriate care has been associated with improved functional out- comes for depression and anxiety disorders (Sturm and Wells, 1995; Wang et al., 2000). Studies also show that uninsured adults with severe mental illnesses receive less appropriate care or medications and experience delays in receiving services until they gain insurance coverage (Rabinowitz et al., 1998, 2001; McAlpine and Mechanic, 2000). Lack of insurance also can impede access to necessary prescription drugs. For example, persons without health insurance have been shown to wait an average of 4 months longer than privately insured patients to receive newer drug therapies for HIV (Shapiro et al., 1999). Only 43 percent of uninsured children have their prescriptions filled compared with 61 percent of privately insured and 56 percent of publicly insured children (McCormick et al., 2001). 

64 INSURING AMERICA’S HEALTH SUMMARY Health insurance is one of the most common, flexible, and reliable means used to gain access to health care. In the United States, health insurance is a voluntary matter, yet some people do not have the choice of coverage and many find it unaffordable. Because so many common events can precipitate the loss of insurance, the chance of being uninsured over the course of a lifetime is substan- tial. There is no guarantee for most people under the age of 65 that they will be eligible for or able to afford to purchase or retain health insurance. Reviewing the Committee’s work to date allows us to draw some general conclusions important for designing effective strategies for extending coverage to everyone: • Currently, health insurance coverage is not universal and it is not continu- ous, resulting in gaps in coverage that put people’s health and finances at risk. Although coverage is needed throughout the course of life, persons can become uninsured regardless of age and family circumstances, with the notable exception of Medicare for those over age 65. • Efforts to fill coverage gaps need to be affordable to individuals, employers, and the public budgets of government agencies that purchase insurance. Most uninsured families would need a substantial premium subsidy in order to purchase health insurance. • Health insurance is important as a stable and efficiently targeted revenue source for health care service providers. Local communities with disproportionate populations without coverage are unable to shoulder the burden alone. • Lack of coverage affects access to care across a spectrum of health care services. A high uninsured rate in a community can also affect availability of primary through tertiary care for both uninsured and insured community resi- dents. • Lack of coverage affects the amount and adequacy of care delivered and ultimately health outcomes. Having coverage for preventive health services, chronic disease care, medications, mental health, acute and diagnostic care, and hospital care promotes improved access to these services and improved health outcomes. Coverage increases the likelihood of receipt of cost-effective, evi- dence-based services in the appropriate settings (e.g., avoiding expensive crisis care for chronic conditions such as hypertension or asthma with regular use of appro- priate ambulatory care). Increased coverage is especially likely to improve the health of people who are in the poorest health and who are most disadvantaged in terms of access to care and thus would likely reduce health disparities among racial and ethnic groups. Broad-based health insurance strategies across the entire unin- sured population would be more likely to produce the desired health benefits than would “rescue” programs aimed only at the seriously ill or those continuing to piece together categorical coverage. 

LESSONS FROM PREVIOUS REPORTS 65 • Public dollars are not as well targeted to achieve improvements in health across the population as they could be if everyone had insurance-based financing for health care services. • Current health insurance arrangements are complex and inefficient. • When insurance becomes available, the uninsured will use more health services. Their increased use would be a positive change; the services, quality, and continuity of care that those without coverage do not get accounts for their poorer health outcomes compared with otherwise similar insured persons. • Federal or shared federal-state health insurance programs distribute the burden of financing health care more broadly among taxpayers than the costs of uncompensated care, which fall more heavily on local communities with concen- trations of uninsured residents. Insurance-based financing could alleviate some of the financial demands on communities disproportionately affected by uninsurance.