Unmet Health Needs of Uninsured Adults in the United States FREE

In 1998, approximately 44 million Americans younger than 65 years—nearly one fifth of the nonelderly population—lacked health insurance, including 33 million adults aged 18 to 64 years.¹ Prior studies have documented that lacking health insurance is associated with important clinical consequences.^2- 4 Uninsured adults generally encounter greater barriers to preventive services and treatment of chronic illnesses than to acute care.⁵ They are more likely than insured adults to report poor health status,^6- 8 delay seeking medical care,⁹ and forgo necessary care for potentially serious symptoms.¹⁰ Uninsured adults receive fewer screening services for cancer and cardiovascular risk factors,^11- 14 present with later-stage diagnoses of cancer,^15- 16 and experience more avoidable hospitalizations.¹⁷ They also face an increased risk of death,¹⁸ particularly when hospitalized¹⁹ or diagnosed as having breast cancer.¹⁵

Although these studies provide compelling evidence of the adverse clinical consequences of being uninsured, most of them examined patterns of care and outcomes prior to 1990, before the rapid growth of managed care. Increasing competitive pressures in the US health care system over the past decade may be eroding access to care for the uninsured by reducing charity care provided by physicians²⁰ and undermining traditional safety-net providers such as community health centers and public hospitals.^21- 24 Several studies have looked at the duration of periods without health insurance^25- 28 or the effect of losing health insurance,^29- 31 but these studies have not distinguished the clinical impact of short-term and long-term periods without insurance. In addition, few studies have assessed the unmet needs of uninsured adults with specific chronic conditions, such as diabetes^32- 33 or hypertension.³⁴

Failing to monitor or effectively treat chronic illnesses such as hypertension in uninsured adults can result in substantial morbidity^35- 36 and may increase the nation's health care costs. These costs are absorbed by clinicians and facilities as free care, passed on to private insurers through cost shifting and higher fees, or paid by taxpayers through higher taxes to finance public hospitals and public insurance programs. The more intensive medical care necessary to treat preventable complications of chronic illnesses may also be costly for the Medicare program when uninsured adults become eligible for this program at age 65 years. Despite such effects, many Americans have perceived being uninsured as a transient phenomenon without significant consequences.^37- 38 Understanding the unmet health needs of uninsured adults could help to overcome such views and bolster federal and state efforts to improve access to care.

We used nationally representative survey data from the Behavioral Risk Factor Surveillance System (BRFSS) for more than 100,000 adults in 1997 and 1998 to compare uninsured and insured adults' access to physicians, especially among adults at increased risk for adverse health outcomes. We also examined rates of clinically indicated preventive services, including cancer screening, cardiovascular risk reduction, and diabetes management. To assess the effect of longer and shorter periods without health insurance, we differentiated adults uninsured 1 year or longer from those uninsured less than 1 year.

The BRFSS is a federally funded survey designed by the Centers for Disease Control and Prevention (CDC) in collaboration with state health departments to monitor health-related behaviors and risk factors in the US population.³⁹ Beginning with 15 states in 1984, the BRFSS has collected data annually through telephone interviews of adults aged 18 years or older residing in households. Since 1994, all 50 states and the District of Columbia have administered the BRFSS survey and submitted survey data to the CDC.

The BRFSS survey instrument has 2 standardized components. First, a core set of questions is asked in all states annually (including health status, health insurance, presence of diabetes, smoking, breast and cervical cancer screening, and human immunodeficiency virus [HIV] risks) or biennially (eg, screening for hypertension, elevated cholesterol, and colorectal cancer, and influenza and pneumococcal vaccinations in 1997; weight control in 1998). Second, individual states may choose each year to include optional modules on topics such as diabetes management or preventive health counseling.

The survey instruments and procedures are designed to produce representative samples and ensure consistency of data across states and years. Many questions are derived from other national surveys, including the National Health and Nutrition Examination Survey and the National Health Interview Survey. English and Spanish versions of the 2 standardized components were provided by the CDC to each state. Additional information about BRFSS survey instruments and procedures is available from the CDC.⁴⁰

The BRFSS used random-digit dialing within blocks of telephone numbers to identify a probability sample of all households with telephones in each state. In eligible households, 1 adult aged 18 years or older was randomly selected and interviewed about his/her health and medical care. Adults residing in hospitals, nursing homes, military bases, college dormitories, or prisons were not eligible. In 1997, the number of completed interviews per state ranged from 1505 to 4923 with a median estimated response rate of 62.5%. In 1998, completed interviews ranged from 1452 to 6005, with a median response rate of 59.2%.⁴⁰

Our cohort included adults aged 18 to 64 years from all 50 states and the District of Columbia who completed the 1997 or 1998 BRFSS survey, including 105,764 respondents in 1997 and 117,364 in 1998. We defined uninsured adults as those who reported having no health insurance coverage at the time they were surveyed. We divided this group into the long-term uninsured (those who had been uninsured for ≥1 year) and the short-term uninsured (those who had been uninsured for <1 year), consistent with other research using BRFSS data.²⁸ Respondents who reported having health insurance from any private or public source (including Medicare, Medicaid, and military or veterans' coverage) were classified as insured. We excluded adults older than age 64 years because almost all are eligible for Medicare insurance. We also excluded people who did not report their age (<0.5% in each year) or insurance coverage (<0.3%).

We categorized the sample by numerous demographic variables, including age, sex, race/ethnicity, income, education, employment, and self-reported health status (Table 1). To focus on adults at increased risk for future health problems and early mortality,^41- 42 we identified respondents who reported current smoking, obesity (body mass index [BMI] ≥30 kg/m²), hypertension (diagnosed by a health professional on at least 2 occasions), diabetes mellitus (diagnosed by a health professional, not including gestational diabetes), elevated cholesterol (diagnosed by a health professional), binge drinking (≥5 alcoholic drinks on at least 1 occasion in the past month), or self-perceived risk (medium or high) for HIV infection. Smoking, obesity, diabetes mellitus, and HIV risk were determined from the 1998 BRFSS survey. Hypertension, elevated cholesterol, and binge drinking were assessed in 1997 but not 1998. In California, HIV risk was assessed only for people younger than age 45 years, so we limited analyses of this variable to people aged 18 to 44 years.

We studied 2 types of unmet health needs. The first type, inadequate access to physicians' care, was defined as reporting "a time during the last 12 months when you needed to see a doctor, but could not because of the cost" and not having "visited a doctor for a routine checkup" during the prior 2 years. For the latter measure, we focused on people with clinical risk factors or chronic conditions who would most likely benefit from regular checkups. The second type of unmet health need was failing to receive clinically indicated preventive services, defined as adequate, rather than optimal, care based on the recommended age range and frequency in published national guidelines or evidence-based studies.

These preventive services included self-reported rates of cancer screening with mammography and clinical breast examinations within 2 years for women aged 50 to 64 years,⁴³ Papanicolaou tests within 3 years for women aged 18 to 64 years with an intact uterus,⁴³ fecal occult blood testing with a home kit within 2 years for people aged 50 to 64 years,^44- 46 and sigmoidoscopy within 5 years for people aged 50 to 64 years.^47- 49 Cardiovascular preventive services included hypertension screening within 2 years for all people aged 25 to 64 years,³⁶ cholesterol screening within 5 years for all people aged 45 to 64 years,^43,50- 52 advice about weight loss by a health professional within 1 year for obese individuals aged 18 to 64 years,⁴³ and advice about smoking cessation by a health professional within 1 year for current smokers aged 18 to 64 years.^43,53 Indicators of appropriate preventive services for persons with diabetes aged 18 to 64 years included receiving a glycosylated hemoglobin measurement, foot examination by a health professional, cholesterol measurement, and influenza vaccine within 1 year; dilated eye examination within 2 years; and pneumococcal vaccine at any time in the past.^54- 56 Screening for HIV at any time was evaluated for people aged 18 to 44 years who reported a medium or high risk of HIV infection.⁴³

Information on preventive services was based on 1997 or 1998 data from all 50 states and the District of Columbia except for 4 services. Questions about glycosylated hemoglobin measurements and diabetic eye and foot examinations were asked during 1998 in 37 states, representing about 70% of the US adult population. Questions about smoking cessation counseling were asked during 1997 in 12 states, comprising about 25% of the US adult population.

Our analysis had 3 main objectives. First, in demographic and clinical strata we compared the proportions of respondents who were uninsured 1 year or longer, uninsured less than 1 year, and insured. Second, we compared the proportions of these 3 groups that lacked access to care by a physician. Third, we calculated the proportion of each insurance group that had not received clinically indicated preventive services.

To assess the independent effect of lacking health insurance on access to physicians and preventive services, we used multiple logistic regression to control for age, sex, race/ethnicity, census region, employment, education, and income. For all of these variables except income, data were missing for less than 0.5% of respondents, so we excluded these individuals from adjusted analyses. However, income data were missing for 11.7% of the weighted sample in 1997 and 12.8% in 1998. To avoid potential biases due to mishandling of incomplete income data (eg, deleting cases or using indicator variables for missing data) and to make full use of the observed data,^57- 59 we used NORM statistical software to obtain multiple imputed income values for respondents with missing data.⁶⁰

These imputed data sets were analyzed using conventional complete data techniques. We obtained adjusted rates of access measures for each category of insurance status (long-term uninsured, short-term uninsured, and insured) by direct standardization to the demographic characteristics of the full study cohort using the logistic regression model.^61- 62 The BRFSS sampling and poststratification weights were used in fitting the models and calculating adjusted proportions. To derive accurate SEs and tests of statistical significance from these weighted data, we used SUDAAN statistical software^63- 64 and report 2-tailed tests of significance using standard procedures to calculate valid tests with multiply imputed data.⁶⁵

The study population was representative of more than 163 million US adults aged 18 to 64 years residing in households during 1998 and approximately 161 million comparable adults in 1997. In 1998, 9.7% of this population had been uninsured for at least 1 year, including 1.8% for 1 to 2 years, 2.1% for 2 to 5 years, 3.4% for more than 5 years, 2.1% who were never insured, and 0.3% who could not recall when they were last insured. An additional 4.3% of the population was uninsured for less than 1 year, including 2.5% for less than 6 months and 1.8% for 6 to 12 months. In 1997, 9.8% of the cohort had been uninsured for at least 1 year and 4.2% for less than 1 year.

The unadjusted proportion of adults in 1998 who were uninsured for 1 year or longer, uninsured less than a year, or insured are portrayed in Table 1, stratified by demographic and clinical characteristics. Compared with other adults in the cohort, the proportions of uninsured individuals were higher among younger adults, men, blacks, Hispanics, residents of the South and West, those with less education and lower incomes, and those who were self-employed, unemployed, or not in the labor force. Adults whose self-reported health status was good, fair, or poor were 2 to 3 times more likely to have been uninsured for 1 year or longer than those who reported excellent or very good health. Smokers, obese individuals, and binge drinkers were more often uninsured than adults without these risk factors. In contrast, people with self-reported hypertension, diabetes mellitus, and elevated cholesterol were less likely to be uninsured than adults without these conditions.

Table 2 presents the unadjusted proportion of adults in each insurance category who could not see a physician when needed in the past year due to cost, stratified by demographic factors. Nearly two fifths of long-term–uninsured adults and one third of the short-term–uninsured adults reported this problem, compared with only about 1 in 14 insured adults. Among long-term–uninsured adults, cost barriers to seeing a physician were greatest for women, blacks, the unemployed, and those with low incomes.

Adjusted proportions of uninsured and insured adults in clinical risk groups who could not see a physician when needed in the past year due to cost are shown in Table 3. In each of these risk groups, long-term– and short-term–uninsured adults were significantly more likely than insured adults to have experienced cost barriers to needed care in the past year, reflecting the impact of even short-term periods without health insurance. As depicted in Figure 1, 69.1% of long-term–uninsured adults in poor health and 48.8% of those in fair health reported they could not see a physician when needed in the past year due to cost, compared with 51.9% and 42.4% of short-term–uninsured adults and 21.8% and 15.7% of insured adults, respectively.

Table 4 presents the adjusted proportions of uninsured and insured adults who did not have a routine checkup by a physician during the prior 2 years. In all clinical risk groups, long-term–uninsured adults were significantly more likely than insured adults to have lacked a routine checkup. For short-term–uninsured adults, the differences relative to insured adults were smaller and significant only among smokers, obese individuals, and binge drinkers, but not among those with hypertension, diabetes, elevated cholesterol, or HIV risk.

Table 5 shows the adjusted proportions of adults who did not receive appropriate preventive services. Long-term–uninsured adults were significantly more likely than insured adults to have unmet needs for each of these services, except for glycosylated hemoglobin measurements and pneumococcal vaccinations among adults with diabetes and HIV screening among those with self-perceived risk. For clinical services such as breast cancer or hypertension screening, long-term–uninsured adults were 3 to 4 times more likely not to have received these services. Short-term–uninsured adults had intermediate rates of unmet needs for preventive services that were statistically higher than the rates of insured adults for mammography, Papanicolaou tests, and hypertension and cholesterol screening.

Our study provides recent, nationally representative estimates of unmet health needs experienced by uninsured adults—two thirds of whom had been uninsured for 1 year or longer. Nearly half of uninsured adults with annual incomes below $15,000 reported they could not see a physician when needed in the past year due to the cost of care. Uninsured adults with clinically important chronic conditions and health risks were much more likely than insured adults to report this problem, even after adjusting for income and other potential confounders. Alarmingly high proportions of long-term–uninsured adults in poor or fair health reported forgoing needed care, including about two thirds of those in poor health and half of those in fair health. Another national survey has provided examples of the troubling personal narratives embodied in these statistics.⁶⁶ These findings challenge the views of a growing proportion of Americans—from 43% in 1993 to 57% in 1999—who believe that uninsured people are able to get the care they need from physicians and hospitals.^37- 38

Distinctive features of our study included a focus on the unmet health needs of long-term–uninsured adults, particularly those with impaired health or major chronic conditions such as diabetes mellitus, hypertension, and elevated cholesterol. Long-term–uninsured adults with these conditions and other clinical risk factors were much less likely than insured adults to have received routine checkups during which their health risks might have been addressed. In addition, long-term–uninsured adults were less likely to report that they received basic preventive services related to cancer screening and cardiovascular risk reduction, but not HIV screening. The magnitude of these differences was greatest for services such as mammography and cholesterol screening that are often ordered by primary care physicians during routine checkups. For routine checkups and most preventive services, individuals who had been uninsured less than 1 year did not differ substantially from people who were insured, possibly because the short-term uninsured continued to have regular sources of care or had received checkups while they were insured. In contrast, the short-term uninsured were similar to the long-term uninsured in the likelihood they had not seen a physician when needed in the past year due to cost, suggesting that even short-term periods without insurance may cause sizeable numbers of people to forgo needed care.

Our study provides population-based rates of specific clinical services reported by uninsured and insured adults with diabetes mellitus, building on 2 other national studies that examined some of the same clinical services but did not distinguish the short-term and long-term uninsured.^32- 33 Although uninsured adults with diabetes did not differ significantly from insured adults with diabetes in reported rates of glycosylated hemoglobin measurements or pneumococcal vaccinations, long-term–uninsured adults with diabetes were less likely than insured adults with diabetes to have received other basic services, such as eye and foot examinations, cholesterol screening, and influenza vaccinations. These results are particularly concerning because adults with diabetes mellitus face an increased risk of numerous complications, including serious infections, cardiovascular disease, renal failure, and retinopathy, that can be prevented or deferred with effective medical care.⁵⁴ Therefore, improving access to care for uninsured adults with diabetes may help prevent avoidable complications of this serious chronic disease.

Our results are generally consistent with earlier reports that assessed the impact of insurance status on a narrower range of preventive services. Rates of breast and cervical cancer screening have increased substantially among uninsured women over the past 2 decades,^11- 12 perhaps in part because the National Breast and Cervical Cancer Early Detection Program has funded and promoted free screening services for uninsured, low-income women in all 50 states since 1991.⁶⁷ However, the absolute differences in screening rates between uninsured and insured women have remained remarkably consistent when our findings are compared with earlier studies.^11- 12 Furthermore, another report based on BRFSS data found that age-adjusted mammography rates increased by about 6% for both insured and uninsured women between 1991-1992 and 1996-1997, but the absolute difference between these groups remained approximately 25%.⁶⁸ Lower rates of cancer screening among uninsured adults may be the principal reason why they are diagnosed at later, less curable stages of breast and colorectal cancer than insured adults.^15- 16

Although patchwork programs can fill gaps in specific services such as cancer screening for uninsured adults, national health goals represented in the Healthy People 2010 objectives are unlikely to be met without more vigorous efforts to extend affordable health insurance to the uninsured for a wide range of basic medical services.⁶⁹ Fourteen states have taken steps to expand health insurance coverage for low-income adults.⁷⁰ Evidence from 3 states (Minnesota, Oregon, and Tennessee) suggests these programs have helped reduce both the prevalence and duration of periods without insurance for low-income adults.⁷¹ Comparable efforts to improve access by other states and the federal government could target long-term–uninsured adults with low incomes or chronic health conditions and the approximately 2 million uninsured adults aged 55 to 64 years who are approaching eligibility for Medicare, as others have suggested.^37,72- 75 Our findings suggest that the health benefits of extending insurance to these groups could be substantial, but the costs borne by uninsured people must be low enough to encourage broad participation, especially for those who are poor or near poor.⁷⁶

Our study had some limitations. Although we focused on clinically important access measures that would be expected to yield better health outcomes and are supported by national guidelines, we did not analyze clinical outcomes or the effectiveness of specific treatments. Our study was a cross-sectional analysis, so we could not directly examine the effects on the uninsured of the growth of managed care or other temporal changes in the health care system. In addition, the cross-sectional nature of BRFSS data did not allow us to link the timing of preventive services to specific periods in which individuals were uninsured. Our measures were based on self-reported data that were not independently verified. The reliability of such self-reports is very good or excellent for most variables included in our study.^77- 80 Studies of the validity of self-reported data indicate that telephone surveys closely approximate the prevalence of diabetes and smoking obtained from clinical testing, but may underestimate rates of obesity, hypertension, and elevated cholesterol,⁸¹ and overestimate rates of mammography.⁸²

We have no reason to expect substantial differences in the accuracy of health data reported by uninsured and insured adults, particularly after we controlled for numerous demographic confounders and used rigorous multiple-imputation methods to minimize potential bias related to missing income data. However, our study may underestimate the unmet health needs of uninsured adults if they are less likely to be aware of important health conditions due to inadequate screening. Even when uninsured adults with low incomes obtain care, they report that the quality and continuity of care is substantially worse than reported by insured low-income adults.^31,74 The study cohort also did not include adults living in households without telephones, who represent about 5% of the US population and are more likely to be poor, black, Hispanic, and residents of the South.³⁹ Because these groups are more often uninsured than other groups of adults, our estimates of uninsured adults with deficits in basic care probably would increase if people without telephones had been included.

The federal and state governments have begun to extend affordable health insurance to uninsured children in the United States by expanding Medicaid and launching the Children's Health Insurance Program.^3,83 In contrast, the unmet health needs of 33 million uninsured adults continue to fester in the health care system without a cohesive political response by the federal government or most states. Concerted and collaborative action by policymakers and health care professionals will be required to address these persistent needs.