• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of milbankLink to Publisher's site
Milbank Q. Jun 2002; 80(2): 347–379.
PMCID: PMC2690114

4. Using Administrative Data to Study Persons with Disabilities

Administrative data are the by-product of running the health care system: enrolling people in health plans, tracking service utilization, paying claims, and monitoring costs and quality. Although not originally intended for conducting research, administrative data offer significant advantages:

  • They encompass large groups of people, often entire populations (such as all residents of a state), thus enhancing the generalizability of research findings.
  • They represent care practiced throughout the community and are not confined to restricted settings.
  • Longitudinal, person-level, administrative databases can track study subjects over time and across settings of care.
  • Large numbers of cases help hide individual identities; furthermore, researchers generally conduct studies without knowing the specific identities of the research subjects, thus protecting confidentiality.
  • They already exist, are relatively inexpensive to acquire, and are computer readable.

Given these attractions, the 1989 law authorizing the Agency for Health Care Policy and Research (AHCPR) (sec. 6103 of Omnibus Budget Reconciliation Act of 1989, P.L. 101–239) stipulated using large administrative files to examine the “outcomes, effectiveness, and appropriateness” of health care services (sec. 1142(c)). AHCPR's early flagship projects, the Patient Outcomes Research Teams (PORTs), began with administrative data (Clancy and Eisenberg 1997; Lave, Pashos, Anderson, et al. 1994; Mitchell, Bubolz, Paul, et al. 1994). Numerous health services research studies—primarily examining costs and utilization but also exploring outcomes and quality of care—rely exclusively on administrative data (AHCPR 1996, 1997).

Administrative data do have important limitations, however. In particular, their principal clinical insight comes from diagnosis codes with questionable accuracy, completeness, clinical scope, and meaningfulness (Hsia, Ahern, Ritchie, et al. 1992; Hsia, Krushat, Fagan, et al. 1988; Iezzoni 1997a, 1997b; McCarthy, Iezzoni, Davis, et al. 2000; Romano 1993; Romano and Mark 1994). Administrative data also are often cumbersome to handle and several years out of date. These and other problems led the former Office of Technology Assessment (1995, 6) to conclude:

Contrary to the expectations expressed in the legislation establishing AHCPR and the mandates of the PORTs, administrative databases generally have not proved useful in answering questions about the comparative effectiveness of alternative medical treatments. Administrative databases are very useful for descriptive purposes (e.g., exploring variations in treatment patterns), but the practical and theoretical limitations of this research technique usually prevent it from being able to provide credible answers regarding which technologies, among alternatives, work best.

Despite these drawbacks, many health services researchers produce productive and insightful findings using administrative data, especially studies relating to payment policy (e.g., developing risk adjusters for setting capitation rates), expenditures (e.g., payments by type or setting of care for specific populations), and patterns of service utilization (e.g., disparities by race or gender).

This article examines these issues by using administrative data to explore health care experiences and outcomes specifically for persons with disabilities. Depending on the definition of disability, one could argue that many health services researchers have already used administrative data to study people with potentially disabling conditions. Examples include PORTs relating to back pain, schizophrenia, stroke, hip fracture, and studies of Medicaid services used by children receiving Supplemental Security Income. Under other definitions of disability, however, one could argue equally strongly that studying disability using administrative data is impossible, that the data simply do not contain enough information to identify disability. But linking administrative data to other data sources, especially survey information, does offer opportunities for studying disability.

Here, I first consider definitional issues, then discuss the utility of specific administrative data elements for studying people with disabilities, and finally look at linking administrative data with other information sources. This presentation assumes that research using administrative data has three major goals:

  1. Identifying people with disabilities, thus defining the study population.
  2. Examining their health care experiences.
  3. Assessing the outcomes of this care.

Defining Disability

Differing Definitions

Defining disability is complex, with multilayered personal, institutional, and societal ramifications (Williams 2001). Since the 14th century, disability has delineated categories of people meriting assistance—alms, food, shelter. However, “because physical and mental incapacity are conditions that can be feigned for secondary gain …, the concept of disability has always been based on a perceived need to detect deception” (Stone 1984, 23). Separating the deserving from undeserving disabled, therefore, fell to doctors and their theoretically objective medical evidence. The standard “medical model” of disability assumes that individuals “afflicted” with compromising health conditions must adapt their lives and expectations to their limitations.

Over the last half century, various definitions of disability have appeared for various purposes. Some definitions echo the traditional medical model, while others introduce a new concept, that disability results from social and physical environments that do not accommodate persons with differing physical, sensory, cognitive, or emotional attributes. That is, people are not disabled; society is (Barnes, Mercer, and Shakespeare 1999; Charlton 1998; Oliver 1996).

Definitions of disability therefore differ widely, as suggested by

  • The 1990 Americans with Disabilities Act (ADA), P.L. 101–336, sec. 3: “(A) a physical or mental impairment that substantially limits one or more of the major life activities …; (B) a record of such impairment; or (C) being regarded as having such an impairment.”
  • The Social Security Administration (SSA 1998, 2) for determining adult eligibility for Social Security Disability Insurance (SSDI) or Supplemental Security Income (SSI): “the inability to engage in any substantial gainful activity by reason of any medically determinable physical or mental impairment(s) which can be expected to result in death or which has lasted or can be expected to last for a continuous period of not less than 12 months.”
  • The World Health Organization (WHO 1980, 28) for the International Classification of Impairments, Disabilities, and Handicaps (ICIDH): “In the context of health experience, a disability is any restriction or lack (resulting from an impairment) of ability to perform an activity in the manner or within the range considered normal for a human being.”
  • The Institute of Medicine Committee on a National Agenda for the Prevention of Disabilities: “inability or limitation in performing socially defined activities and roles expected of individuals within a social and physical environment” (IOM 1991, 7).

Few people are satisfied with these definitions. Through recent court cases, the U.S. Supreme Court is refining the ADA's definition (e.g., Sutton et al. v. United Air Lines, Inc. and Murphy v. United Parcel Service, Inc., 1999; Toyota Motor Manufacturing Inc. v. Williams, 2002). The SSA is now reviewing aspects of its disability determination provisions for adults (Mathiowetz and Wunderlich 2000). In revising the ICIDH, WHO (1997, 24) initially dropped the word disability, arguing that it causes “misunderstanding between health care professionals and people who experience disablement.” The final version, the International Classification of Functioning, Disability and Health (ICF), reinstates disability but labels it an “umbrella term for impairments, activity limitations or participation restrictions” (WHO 2001, 3). The ICF views “a person's functioning and disability … as a dynamic interaction between health conditions (diseases, disorders, injuries, traumas, etc.) and contextual factors,” including environmental and personal attributes (WHO 2001, 8). The Institute of Medicine Committee on Assessing Rehabilitation Science and Engineering objected that the 1991 (IOM) definition failed to account sufficiently for environmental influences: “Disability is a dependent variable whose value is determined by the relationship between two other variables: the person and the environment” (IOM 1997, 73–4).

Semantic distinctions among disability definitions are sometimes elusive. “It is often difficult to communicate conceptual constructs within the same discipline, let alone across … professional fields, which may account for some of the misinterpretations that have been plaguing this area” (IOM 1991, 321). Nevertheless, a general understanding now defines disability as difficulty conducting daily activities because of health, sensory, cognitive, and emotional conditions interacting with the social and physical environments.

Diversity and Differing Perspectives

Disabilities are diverse in their causes, nature, timing, pace, and societal implications. Some potentially disabling conditions are congenital; others are acquired. Some occur suddenly as a result of injury or accident; other arise slowly, with progressive chronic conditions. Some gradually limit functioning but do not threaten life; others hurry death. Some are visible to outsiders; others are hidden. Some engender stigmatization and blame; others prompt pity and paternalism. Some are seen primarily as “diseases” (e.g., cancer, coronary artery disease, emphysema), although they can become profoundly disabling. As Nancy Mairs (1996, 11) wrote in her book about living with multiple sclerosis, “What I cannot do … is to depict and analyze ‘disability’ as a global subject. The category is simply so broad, and the possible approaches to it so numerous, that all the attempts I've come across at generalizing about it run into difficulty.” The phrase “people with disabilities” is too broad to be meaningful.

In addition, being labeled as disabled depends on who is asked. Many people who are born deaf, for example, speak American Sign Language and do not view themselves as disabled, as they can participate fully in a distinct deaf culture (Rockow 2001). According to the 1994–1995 National Health Interview Survey-Disability supplement (NHIS-D), almost 20 percent of manual wheelchair users do not see themselves as disabled, although mainstream society probably does (Iezzoni, McCarthy, Davis et al. 2000). Thus, while outsiders may see a compromised life, people themselves disagree.

Those in what others may perceive to be “poor” health place a relatively high value on their own health since they have adjusted their life styles and expectations to take account of their condition. This may be particularly true of young disabled men and women, since one-quarter of this group of respondents describe their health as “poor” yet value it as “good.”

(Dolan 1996, 559)

Even theoretically “objective” measures of function vary by context. “Capability” indicates what persons “can do” in controlled settings, whereas “performance” assesses what a person “does do” in everyday life. Capability typically exceeds performance (Young, Williams, Yoshida, et al. 1996). Primary care doctors frequently fail to assess accurately patients’ functional abilities (Calkins, Rubenstein, Cleary, et al. 1991; Hoenig 1993; Nelson, Conger, Douglass, et al. 1983).

Implications for Studying Disability Using Administrative Data

Administrative databases do not link information about health conditions with data about performance of daily activities, participation in life situations, and social and physical environmental barriers. Instead, most administrative data report only the diagnostic information attached to specific health care services. Certainly, some diagnoses allow inferences about potential disabilities. But most diagnoses by themselves convey little about their effects on people's daily activities or the impact of their social or physical environments. Administrative data alone reveal nothing about whether people view themselves as disabled.

As noted below, these data do identify people meeting administrative definitions of disability, primarily persons eligible for Medicare and Medicaid through SSDI and SSI, respectively. These are important populations—almost 6.7 million persons in 2000—with significant health and health care concerns. However, findings from SSDI and SSI recipients may not generalize to other people with similar disabling conditions who, for whatever reason, have neither applied for nor are qualified as disabled under Social Security. Although the SSA's judgments about employability putatively use objective medical evidence, the boundaries have become blurred: “The scientific link between [complete] work incapacity and medical condition is a weak one” (U.S. General Accounting Office 1996, 35). The number of applications rises during recessions and falls during times of prosperity (Chirikos 1991, 165), suggesting shifting standards in employment-related disability. In 2000, the SSA processed approximately 1.3 million SSDI applications, though approving only 47 percent (Martin, Chin, and Harrison 2001).

Nevertheless, administrative data permit certain insights that can produce useful research information about people with disabilities. The findings must be interpreted cautiously, however, acknowledging important limitations in identifying study populations.

Administrative Data

Administrative data are what results from overseeing and running the health care system. In some administrative databases, the unit of observation is a specific service such as a hospitalization. Thirty-six states systematically collect information from hospitals about their discharges (Donaldson and Lohr 1994; Elixhauser, Andrews, Ball, et al. 1996). With a few exceptions (e.g., California), individuals do not have unique identification numbers, so multiple hospitalizations for individuals cannot be tracked.

In other databases, mainly those produced while administering health insurance, the unit of observation is the person: individuals are assigned unique identifiers, thereby allowing them to be tracked across care settings covered by the insurer. This article focuses on person-level administrative data produced by both public (e.g., Medicare, Medicaid, Department of Veterans Affairs) and private health insurers. Some health plans primarily or exclusively enroll people with significant functional limitations, although their populations typically represent small and nonrandom subsets of people with disabilities (Master, Dreyfus, Connors, et al. 1996; Riley 2000; Robinson and Karon 2000).

With some exceptions, most health insurance databases contain two types of files: enrollment files, indicating eligibility for the health plan and relevant demographic information; and encounter records, representing individual services or sets of services. For Medicare, the Health Insurance Skeleton Eligibility Write-Off (HISKEW) file contains demographic information updated every quarter on all persons ever enrolled, including date of death. The original reason for eligibility (e.g., disability) also is recorded. Using Health Insurance Claim (HIC) numbers uniquely assigned to each Medicare beneficiary, researchers can link claims or records from health care encounters to track people over time and across care providers.

Most clinical insights from administrative data are derived from claims or encounter records for individual services. Here, I group these files into two types: “standard” and “enriched,” reflecting the clinical content of the data and the setting of the care. Age, residential setting (e.g., nursing home, available on some administrative records), discharge disposition (for hospitalized patients), service-connected condition eligibility for veterans, and dual eligibility for Medicare and Medicaid offer important clues to whether people have functional impairments. In standard administrative data, the most helpful clinical information is

  • Diagnoses coded using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM).
  • Procedures, items, or services coded according to the ICD-9-CM for claims submitted by institutional providers, such as hospitals; the American Medical Association's Current Procedural Terminology (CPT-4), for individual physician's services; the Healthcare Common Procedure Coding System (HCPCS, formerly HCFA's CPCS), for nonphysician services not in CPT-4, including durable medical equipment (DME, e.g., ventilators, supplemental oxygen, wheelchairs); and the National Drug Codes (NDC) codes for prescription drugs.

Standard administrative data are submitted when billing for care (fee-for-service insurance plans) or as an administrative requirement for capitated plans to report all health care encounters (Hornbrook, Goodman, Fishman, et al. 1998; Iezzoni 1997b). Medicare's managed care organizations (MCOs) currently must report encounter records only for hospitalizations; additional reporting requirements remain controversial.

“Enriched” administrative data include additional clinical information collected routinely by regulatory or programmatic mandate. The best-known examples come from Medicare:

Rehabilitation hospitals also collect clinical information. Most use the Functional Independence Measure (FIM, an 18-item scale), while some employ the MDS for Post-Acute Care (MDS-PAC, containing more than 400 items) (Stineman 1997; Stineman, Escarce, Goin, et al. 1994; Stineman, Goin, Tassoni, et al. 1997; Williams, Li, Fries, et al. 1997).

MDS, MDS-PAC, OASIS, and FIM data are collected by clinicians, without incorporating patients’ viewpoints. Only recently have OASIS data begun to be gathered widely. MDS data, though, have been produced for more than a decade and have been used extensively in health services research on nursing home residents (e.g., Fries et al. 1994; Hawes et al. 1997; Swan and Newcomer 2000), as have FIM data from rehabilitation hospitals (see Stineman 1997; Stineman et al. 1994, 1997). Some studies, however, have questioned the quality of MDS data in nursing homes and the logistical burden of MDS-PAC (Medicare Payment Advisory Commission 2001). The multiplicity of data-gathering tools for populations with roughly similar clinical concerns prompted the Medicare Payment Advisory Commission (2001, 94) to recommend creating a single “patient classification system that predicts costs within and across post-acute settings.”

Enriched administrative data represent specific populations that probably contain many persons with disabling conditions. The nursing home data obviously depict persons with serious impairments; for instance, among roughly 2.1 million Medicare beneficiaries living in long-term care facilities in 1997, more than 87 percent had limited mobility (Sharma, Chan, Liu, et al. 2001, 62). The home care population probably contains two subgroups: persons with short-term needs following an acute event who will recover fully and those with chronic conditions who are sufficiently debilitated to remain “homebound” (U.S. General Accounting Office 2000). The following discussion concentrates on standard administrative data.

Enrollment and Eligibility

People enrolled in Medicare or Medicaid through SSDI or SSI, respectively, have fulfilled Social Security's disability criteria.1 Eligibility status is noted in administrative enrollment and most encounter databases. For adults, the SSA defines disability as the inability to perform “substantial gainful employment” because of “an impairment that results from anatomical, physiological, or psychological abnormalities which can be shown by medically acceptable clinical and laboratory diagnostic techniques” (Social Security Administration 1998, 3). As noted earlier, whether these criteria are applied consistently is debatable (Kane 2000; U.S. General Accounting Office 1996). Therefore, although Social Security disability status conveys important information, researchers must use this variable carefully.

Employment is an irrelevant standard for children. Historically, the SSA's definitions of disability for children (leading to SSI eligibility) have been problematic. Until 1990, a restrictive medical definition requiring children to have a disabling physical or mental disorder “per se” limited the number of children participating in SSI (Ettner, Kuhlthau, McLaughlin, et al. 2000). In their 1990 decision for Sullivan v. Zebley, the U.S. Supreme Court required individual functional assessments for children who did not meet the medical definition. Around that time, the SSA expanded its list of qualifying mental impairments for children from four to 11 categories, including attention deficit hyperactivity disorder. The rolls of children receiving SSI have therefore varied widely over time (Ettner et al. 2000; Perrin, Kuhlthau, McLaughlin, et al. 1999), leading to questions about the generalizability of populations identified by this label.

Depending on one's analytic goals, using SSDI and SSI to identify people with disabilities may be sufficient. But researchers may also want to know the underlying condition that caused the person to qualify as disabled. The SSA retains information about the major medical reason for disability determinations and disseminates aggregated information about this. For example, in 2000, the SSA awarded benefits to 610,700 disabled workers. The most common reason was musculoskeletal problems (25 percent), followed by mental disorders (24 percent), circulatory conditions (12 percent), cancer (10 percent), and disorders involving the nervous system or sensory organs (8 percent) (Martin, Chin, and Harrison 2001). Data on individuals’ medical conditions are not available to researchers outside the SSA.

Using the diagnoses on Medicare or Medicaid claims offers reasonable inferences about the underlying medical condition, but this approach has limitations. Children qualifying for SSI because of mental retardation, for example, rarely generate claims specifically for mental retardation–related services (Perrin, Kuhlthau, Ettner, et al. 1998b). Analyses using Medicaid claims to identify children with mental retardation, therefore, probably underrepresent this population. Nevertheless, children rarely have multiple, comorbid conditions; rather, they typically have single diagnoses, unlike adults who often have many coexisting illnesses. For adults, selecting the disabling condition from among the many disease codes may be difficult. In contrast, the diagnoses coded for children are more likely to represent their underlying disabling disease. Many studies have used Medicaid data to examine patterns of diagnoses and service use for children receiving SSI (Adams, Ellwood, and Pine 1989; Ettner et al. 2000; Kuhlthau, Perrin, Ettner, et al. 1998; Perrin, Ettner, McLaughlin, et al. 1998a; Perrin et al. 1998b, 1999).

Another issue is the delays in Medicare eligibility: adults qualifying for SSDI do not receive Medicare coverage until two years after first payment of SSDI cash benefits. (The one exception is amyotrophic lateral sclerosis [ALS] patients, who since 2001 have immediately qualified for Medicare upon receiving SSDI, because many ALS patients die or become significantly impaired during the two-year wait.) Medicare claims files do not list the services that people obtain during this waiting period. Because some people may defer services during these two years (e.g., because they cannot afford care), their use of the services immediately upon qualifying for Medicare may be unusually high. Medicare beneficiaries age 65 years and older who entered Medicare at younger ages because of disability have 43 percent higher costs than otherwise comparable beneficiaries (Pope, Ellis, Ash, et al. 2000a, 106). Analyses of elderly Medicare beneficiaries should therefore consider their “originally disabled” status.

Claims or Encounter Records

To produce claims or encounter records—and their associated diagnosis and procedure codes—people must first have health insurance. A nationwide survey of persons age 16 and older with and without various disabilities found that both groups had comparable insurance rates, 90 and 89 percent, respectively (Harris Interactive 2000, 54). Thus, roughly 10 percent of the population, regardless of disability, are not represented in person-level administrative databases.

The services must be covered by a person's insurance in order for claims to be filed. However, many important disability-related services are not covered, especially by private health plans and Medicare. Disabled Medicare beneficiaries who also were not eligible for Medicaid spent an average of $1,532 out-of-pocket for health care services in 1995, an amount that increased to $2,175 for persons with two or more limitations in their daily living activities (Foote and Hogan 2001, 247). In addition, many private health insurance policies exclude for six months to a year any services for previously existing conditions (Pelka 1997, 147), and private plans generally set annual limits on mental health services (Gitterman, Strum, and Scheffler 2001) and durable medical equipment (Pelka 1997). Physical, speech-language, and occupational therapy are typically restricted to restoring persons to baseline functioning, rather than providing long-term maintenance or preventing functional declines. “Indemnity, preferred provider, and HMO [health maintenance organization] insurance products combine variations in cost sharing in myriad ways with variation in coverage, including or excluding physical therapy, rehabilitation, mental health, … and durable medical [equipment]” (Robinson 1999, 54).

Medicare explicitly excludes coverage for “any services that are not reasonable and necessary for the … diagnosis or treatment of illness or injury or to improve the functioning of a malformed body member” (42 C.F.R. sec. 411.15[k]). This criterion of medical necessity specifically limits reimbursement for assistive devices: “In many cases, assistive technologies instrumental to maintaining an independent lifestyle and often essential to preventing secondary conditions do not satisfy the criteria on the Medicare screening list for durable medical equipment” (IOM 1991, 257). For example, Medicare does not cover glasses or hearing aids (Cassel, Besdine, and Siegel 1999), and it restricts other DME coverage by setting of use: Part B “pays for the rental or purchase of durable medical equipment” only “if the equipment is used in the patient's home or in an institution that is used as a home” (42 C.F.R. sec. 410.38[a]). If someone needs a wheelchair only for going outside, for example, Medicare will not cover it.

Twenty-eight percent of insured people with disabilities reported that they had special needs (for particular therapies, equipment, medications) that were not covered by their health plan, compared with 7 percent of those without disabilities, and 40 percent with very severe disabilities had special needs that were not covered by insurance (Harris Interactive 2000, 56, 57). Nineteen percent of disabled persons reported that they needed medical care during the last year but did not get it, compared with 6 percent of nondisabled persons (Harris Interactive 2000, 60). Thirty-five percent of disabled people attributed these failures to a lack of insurance coverage.

Thus, the services represented by claims or encounter records do not reflect fully the health needs of people with disabilities. To interpret the information from administrative databases, researchers must appreciate the implications of health insurance coverage policies. People needing uncovered assistive devices sometimes get them through private sources, such as disease advocacy associations. Children with disabilities, in particular, may obtain health-related services and equipment through a complex maze of state-sponsored programs. Services “carved out” from plans (e.g., mental health) are generally missing from claims files. Obviously, services obtained from outside sources are not included in health insurance databases.

Diagnosis Codes

ICD-9-CM contains more than 15,000 codes, including many for conditions that are technically not “diseases.” Table 1 lists examples of codes representing physical functional impairments. Numerous other codes depict low vision and deafness, mental retardation, cognitive impairments, mental illnesses, medical diseases, and other conditions underlying disability. Creative combinations of ICD-9-CM codes produce suggestive clinical stories like the following:

TABLE 1
Examples of ICD-9-CM Codes Representing Physical Functional Impairments

A person with multiple sclerosis (code 340) has weakness of the legs (code 344.9, “paralysis unspecified”), which is a “condition influencing their health status” (code V49.2, “motor problems with limbs”). The person uses a wheelchair (code V53.8), has an inaccessible home (V60.1, “inadequate housing”), and is unemployed (V62.0).

Health services research studies have used diagnosis codes from claims, especially for chronic and disabling conditions, to identify persons for whom the cost of care is likely to be high in the future (Ash, Ellis, Pope, et al. 2000; Ellis, Pope, Iezzoni, et al. 1996; Iezzoni, Ayanian, Bates, et al. 1998; Kronick, Zhou, and Dreyfus 1995; Kronick, Dreyfus, Lee, et al. 1996; Kronick, Gilmer, Dreyfus, et al. 2000; Pope, Ellis, Ash, et al. 2000b). These methods are now used by Medicare and several Medicaid programs to adjust capitation payments (Greenwald 2000; Iezzoni et al. 1998; Kronick, Gilmer, Dreyfus, et al. 2000; Pope et al. 2000b).

By definition, however, diagnosis codes in administrative databases are not meant to tell stories but to generate reimbursement. Physicians (or their office staffs) list codes for outpatient office visits, and hospital medical record departments assign discharge diagnoses for hospitalizations. Extensive guidelines govern the assignment of codes; for example, discharge diagnoses may include only those conditions affecting the current hospitalization (Brown 1989). Nonetheless, numerous studies document significant problems with the accuracy, completeness, clinical scope, and meaningfulness of diagnosis codes (Hsia et al. 1988, 1992; Iezzoni 1997a, 1997b; McCarthy et al. 2000; Romano 1993; Romano and Mark 1994). In particular, V codes, a supplementary classification of factors influencing health status and contact with health services, are coded infrequently. Insurers rarely pay based on V codes; exceptions include V581, maintenance chemotherapy, and V580, radiotherapy session, which must accompany hospital claims for these services.

ICD-9-CM codes present special challenges for identifying disability. First, codes for individual medical diagnoses reveal little about the severity or extent of the condition or the pace of its progression. Code 340, for example, indicates multiple sclerosis but says nothing about neurological signs or symptoms or whether the disease is in remission or advancing rapidly. Certainly, additional (secondary) diagnoses supply more details about neurological impairments (e.g., optic neuritis, “ paralysis”), but physicians rarely code more than one diagnosis for outpatient visits.

Second, incomplete coding is especially problematic for people with chronic conditions. Good, albeit circumstantial, evidence suggests that hospitals undercode chronic conditions, such as diabetes mellitus and hypertension, for acutely ill patients (Iezzoni, Foley, Daley, et al. 1992; Jencks, Williams, and Kay 1988). Similarly, physicians often do not list chronic, disabling conditions on outpatient claims. Of Medicare beneficiaries coded with either an inpatient or an outpatient diagnosis of dementia in 1994, only 59 percent had dementia coded in 1995; for patients coded with paraplegia or quadriplegia in 1994, only 52 percent had these codes in 1995 (Medicare Payment Advisory Commission 1998, 17). Using Medicaid data from seven states, the percentages of people coded with specific diagnoses in the second year who had the code in the first year were 80 percent for schizophrenia, 68 percent for diabetes, 58 percent for multiple sclerosis, 57 percent for quadriplegia, 43 percent for ischemic heart disease, and 34 percent for cystic fibrosis (Kronick et al. 2000, 60). None of these conditions disappears, so their absence in the subsequent year only highlights the incompleteness of the coding.

Third, unless a condition is actively treated, physicians and hospitals are unlikely to code it. Although codes for low vision, deafness, and hard of hearing exist, they are rarely listed on claims for services unrelated to the eyes or ears. Mental retardation is infrequently coded for adults or children, perhaps because few health interventions directly target this condition (Kronick et al. 2000; Ettner et al. 2000). Sometimes doctors intentionally withhold potentially stigmatizing diagnoses (e.g., mental health disorders) when other conditions can legitimately be listed.

Fourth, coding is predicated on documentation in medical records, but information on functional status is notoriously missing from medical records (Bierman 2001). One study of inpatient records found inadequate documentation of elderly patients’ functional deficits not only in physicians’ notes but also in medical record entries by nurses and physical and occupational therapists (Bogardus, Towle, Williams, et al. 2001). Outpatient records are even less likely to document functional impairments.

Finally, ICD-9-CM diagnosis codes do not adequately capture functional status. For example, code 344.9, paralysis unspecified, is used for conditions from complete paralysis to generalized weakness. No codes indicate performance of daily living activities or even the rudimentary physical functioning metrics asked in the NHIS-D (e.g., whether the respondent can climb 10 stairs without resting or walk three city blocks). Without valid information about physical, sensory, cognitive, or emotional functioning, one cannot accurately identify people with potentially disabling conditions or track meaningful outcomes of care. Although the ICF, promulgated by WHO (2001), classifies functional abilities and social and environmental contexts, this coding scheme is not used routinely for administrative data collection in the United States.

Procedure Codes

Some procedures imply debility or potentially disabling physical impairments such as amputation of a limb, hip arthroplasty or knee replacement, transplantation of a major organ, insertion of a tracheostomy tube (for long-term mechanical ventilation), or placement of a gastrostomy tube (for feeding). Certain treatments like chemotherapy or radiation therapy also suggest potentially disabling conditions. People receiving such interventions often generate high health care costs, reflecting considerable need for services (Pope et al. 2000a).

Because health care providers are paid for specific procedures, the accuracy of the coding and the completeness of procedures are generally better than for diagnoses (Fisher, Whaley, Krushat, et al. 1992; Iezzoni 1997b). The coding of procedures is especially good for high-risk or costly procedures such as chemotherapy and mechanical ventilation (Romano and Luft 1992). Medicare, Medicaid, and private insurers generally pay without hesitation for acute care services, although proof of medical necessity or appropriateness is sometimes required (e.g., for transplantation, services considered experimental). Because of high costs, patients are unlikely to obtain such services without insurance coverage. Therefore, administrative databases should represent virtually all procedures or treatments obtained within a given time period.

Relying on procedure codes to identify people with disabilities raises important questions, however. First, people receiving these procedures are a nonrandom subset of all people with disabling conditions. Second, even for potentially life-prolonging interventions (e.g., gastrostomy tube placement), significant variations in their use are likely because of differing preferences for care, practice styles, and availability of services across patients, physicians, institutions, and geographic regions. Third, people may have received the service before the time period covered by the administrative database. Although certain V codes (ICD-9-CM diagnoses) indicate some earlier procedures (e.g., “transplant status”), pertinent V codes are not always listed. Fourth, restrictive payment policies limiting physical, speech-language, or occupational therapy mean only that certain people may receive these services, so that drawing inferences about disability from their presence or absence is risky. Finally, the presence of procedure codes reveals little about outcomes, for instance, whether the intervention improved functioning. A knee replacement, for example, may enable a patient to become mobile and hence not disabled.

Durable Medical Equipment

Durable medical equipment (DME) frequently compensates for or ameliorates functional limitations and can prolong life. Examples include supplemental oxygen, mechanical ventilation devices, dialysis supplies and equipment, parenteral and enteral nutrition supplies, limb prostheses, wheelchairs, walkers, and hospital beds. For Medicare, the most common single category of DME is wheelchairs, obtained by 2.7 percent of beneficiaries in 1996 (Pope et al. 2000a, 6-4). The use of such equipment is the quintessential visible symbol of disability, at least from a societal perspective.

Identifying people with disabilities through DME claims, however, raises questions. Most important, as noted earlier, restrictive coverage policies mean that paid claims for DME (especially mobility aids) represent only a fraction of need. Medicare and private insurers often reimburse certain DME only in certain circumstances (e.g., homebound patients). In addition, as for procedure codes, DME use varies significantly based on differing personal preferences and practice styles. People may have obtained DME at an earlier time, or V codes (e.g., indicating wheelchair use) are listed sporadically. Although claims for DME offer useful information, they identify only some of the people with potentially disabling conditions.

Outpatient Pharmacy Claims

Administrative databases from Medicaid, veterans’ affairs, and private insurers offering prescription drug benefits list outpatient pharmacy claims. Medicaid's computerized pharmacy data file contains information on all prescriptions dispensed by pharmacies, including the recipient's identification number, prescription date, specific drug (e.g., National Drug Code number), quantity, prescribing physician, and reimbursement information. Data on drug prescriptions in Medicaid files are reasonably reliable (Bright, Avorn, and Everitt 1989), although pharmacy data represent only those prescriptions that were actually filled (instead of prescriptions written but never obtained by the patient). Obviously, administrative files do not report compliance with drug regimens (e.g., whether patients took the medications as prescribed). In addition, the quality of pharmacy claims information not required for reimbursement, such as outpatient diagnoses indicating the need for a drug, may not be reliable (Ray and Griffin 1989)

Pharmacy data help identify people with specific conditions like diabetes, bipolar disorder, asthma, and acquired immunodeficiency syndrome (Fishman and Shay 1999; Lamers 1999). They suggest the severity of medical conditions; for example, people receiving insulin presumably have more demanding diabetes mellitus than do those using oral hypoglycemic agents. Certain drugs suggest the pace of the illness or the course of a disease; for instance, people with multiple sclerosis prescribed interferon-beta probably have relapsing-remitting rather than secondary progressive disease. Pharmacy data do, though, have the same problems as procedure and DME codes. That is, only certain subgroups of people are identified; the varying practice patterns compromise the generalizability of findings; and those prescriptions not covered by insurance plans are obviously not listed.

Linking Administrative Files with Other Data Sets

Merging administrative data with other data sources can greatly enhance their analytic utility (U.S. General Accounting Office 2001), including that for disability-related research. Most such merges involve Medicare data, because of their huge size and nationwide scope. To examine Medicare costs for newly entitled disabled persons, for example, analysts at the SSA and the Health Care Financing Administration (HCFA, renamed the Centers for Medicare & Medicaid Services, or CMS, in 2001) merged the SSA's Master Beneficiary Record (records of entitlement and cash payments for all persons who ever received Social Security benefits), the Continuous Disability History Sample (a 20 percent sample of SSA disability determinations), and Medicare claims files (Bye, Riley, and Lubitz 1987). Using this merged file, they examined long-term Medicare costs for disabled beneficiaries (Bye, Riley, and Lubitz 1987) and the potential costs to Medicare of eliminating the two-year waiting period between SSDI cash benefits and Medicare eligibility (Bye, Dykacz, Hennessey, et al. 1991; Bye and Riley 1989).

Merging SSA disability and Medicare files required careful interdepartmental negotiations, primarily to ensure the privacy and security of the resulting database. With detailed information on the SSA's disability determination and their Medicare claims, even files stripped of specific identifiers (name, Social Security number) could present privacy risks. Merged SSA and Medicare databases are generally not released to outside investigators.

Several other merged databases are available to funded researchers who meet specified data security and usage requirements. One prominent example was merging Medicare claims with the National Cancer Institute's (NCI) Surveillance, Epidemiology, and End Results (SEER) Program data (Potosky, Riley, Lubitz, et al. 1993). The SEER Program gathers information from 11 population-based cancer registries and three supplemental registries covering about 14 percent of the U.S. population and containing information about more than 2.5 million cancer cases (National Cancer Institute 2001). Investigators have used these merged files to examine many cancer-related questions (Virnig and McBean 2001), including those about Medicare beneficiaries under age 65. One study found that young, disabled women are diagnosed with breast cancer at later stages than are nondisabled women of similar ages (Roetzheim and Chirikos 2002). These data do not indicate functional status, such as the Karnofsky score (a global functional status measure often used for cancer patients; see Karnofsky, Abelmann, Craver, et al. 1948). Rather, SEER focuses exclusively on cancer stage and treatment information.

Merging two Medicare sources—the Medicare Current Beneficiary Survey (MCBS) and Medicare's National Claims History file—has created a valuable database for studying people with disabilities. The MCBS is an ongoing, longitudinal survey of roughly 12,000 Medicare beneficiaries, with an oversampling of persons under age 65 and those over age 85 (Adler 1994, 1995). MCBS respondents typically remain impaneled for four years, with in-person interviews conducted three times each year. Surveyors also track respondents when they move, so as to minimize the loss of follow-up information.

Two types of MCBS surveys (based on residence in communities or institutions) solicit detailed information about physical and sensory functioning, health conditions, satisfaction with and access to care, services reimbursed outside Medicare (including out-of-pocket expenses), and numerous other topics. Merging MCBS survey responses with Medicare claims facilitates analyses of people with disabilities, such as examining screening and preventive services (Chan, Doctor, McLehose, et al. 1999), satisfaction with care (Adler 1995; Hermann, Ettner, and Dorwart 1998; Rosenbach 1995), access to care (Foote and Hogan 2001; Rosenbach 1995), and out-of-pocket expenditures according to ability to perform routine daily activities (Foote and Hogan 2001).

Several programmatic data collections survey people enrolled in health plans. The Health of Seniors survey, for example, evaluates the performance of Medicare MCOs by looking at the percentage of senior Medicare risk-plan members whose self-reported health status improves, stays the same, or gets worse. The changes are measured over two years and include mental and physical components (National Committee for Quality Assurance 1996). Some private MCOs ask Medicare enrollees about their health risks and physical functioning in order to identify people who might benefit from case management (Boult, Rassen, Rassen, et al. 2000; Boult, Boult, Pirie, et al. 1994; Pacala, Boult, and Boult 1995; Pacala, Boult, Reed, et al. 1997). Although potentially valuable, programmatic survey data have not yet been widely used by health services researchers (Hornbrook et al. 1998). Because many MCOs routinely generate little information about health care encounters or services, merging these survey files with the available administrative data may be a moot issue for these plans.

Surveys using national sampling have been merged with Medicare data, notably the Longitudinal Survey of Aging (LSOA) and the National Long-Term Care Survey, both of which contain extensive functional status information. Before these files can be linked, the survey respondents must give their consent. Although typically 80 percent do, the consent rates vary across surveys (Lillard and Farmer 1997, 694). Linkage problems arise, mainly with the accuracy and uniqueness of the Medicare identification numbers supplied by the respondents. These linked survey files have been used extensively to study patterns of service use, especially by elderly persons with functional deficits (Culler, Callahan, and Wolinsky 1995; Manton, Corder, and Stallard 1993; Mor, Wilcox, Rakowsky, et al. 1994). Plans are afoot to merge NHIS responses with Medicare claims files, to facilitate wide-ranging analyses of service utilization associated with personal attributes, self-reported functional status, and health-related behaviors captured by the NHIS. However, to protect confidentiality, the content of these files and access to these data may be carefully regulated.

Conclusions

Administrative data offer significant advantages for health services research, including studies of people with disabilities. These data are especially useful for large-scale analyses of costs and service utilization, studies with important health policy applications. Administrative data by themselves, however, typically offer little information about physical, sensory, cognitive, or emotional functioning or about physical and social environments. This lack of information makes it difficult for analysts to identify people with disabilities and also their health care outcomes, such as changes in physical or social functioning. Using administrative categories of disability (e.g., SSDI or SSI), along with Medicare or Medicaid data, can provide insights into an important subgroup of people with disabilities. The best analytic opportunities come from administrative data merged with other information sources, especially survey databases containing self-reported functional and health status.

Although diagnoses and procedures are routinely coded on claims and encounter records, functional status is not, for several reasons. As noted earlier, a classification system does exist to code function abilities, as well as social and environmental barriers: the ICF created by WHO (2001). For reasons that are not clear, however, the ICF has attracted little attention in the United States. Historically, health insurance has emphasized the coverage of acute care services deemed “medically necessary” (Fox 1989, 1993), and diagnosis and procedure codes typically capture this acute care perspective better than functional status codes do. Therefore, insurers seek diagnostic and procedural information about routine administrative transactions (e.g., to justify payment). Nowadays, adding data to administrative records, such as enrollment files or claims, demands a compelling “business case” that the element is important, relevant, feasible, valid, and cost effective (National Committee on Vital and Health Statistics 2001). Provisions of the 1996 Health Insurance Portability and Accountability Act (HIPAA, P.L. 104–191) mandate standardized formats for all administrative transactions within the U.S. health care system. Inserting codes for functional status into administrative records now requires multiple layers of approval, culminating in federal regulation: a tall and almost insurmountable hurdle.

In the mid-1980s, HCFA staff briefly considered adding a modified version of the Karnofsky score (a global functional status measure) to Medicare physicians’ claims, reasoning that it would routinely provide invaluable functional status information. However, they soon abandoned the idea. How could they persuade physicians, already reluctant to spend time coding diagnoses, to code Karnofsky scores? Medical records poorly document patients’ functional status, even of persons with impairments (Bierman 2001; Bogardus et al. 2001). How, then, could HCFA ensure the accuracy of newly mandated information? Furthermore, for whatever reasons, physicians could manipulate Karnofsky scoring, making their patients appear more or less impaired. How could this manipulation be detected, given the inherent subjectivity at the margins of the Karnofsky assessment? How much would it cost to add Karnofsky score coding slots to long-established administrative records and their associated computerized processing systems? A well-intentioned effort to gather functional status information foundered on intransigent practical considerations.

To overcome several of these problems, some people have suggested routinely asking insurance plan members to complete questionnaires about their health and functional status. Such initiatives could produce large databases of self-reported functioning ripe for outcomes research. Many plan members may happily answer these questions openly and honestly, pleased that somebody cares enough to ask how they are doing. But what about those people with disabilities or potentially stigmatizing conditions who worry about having health insurance? Everybody knows about uninsured Americans—estimated at 42 million (15 percent of the population) in 1999 (IOM 2001, 23). People know about health plans that leave Medicare, setting their members adrift; indeed, 43 percent of dropped members now worry about paying their health care bills (Laschober, Neumann, Kitchman, et al. 1999, 155). Even though workers typically rely on their employers for health insurance, more than half of uninsured people with a disability are employed (Meyer and Zeller 1999, 11). Clearly, some employers avoid hiring disabled workers, fearing higher health insurance premiums (Batavia 2000).

Without universal health insurance, many people might fear that they will be denied services or future insurance coverage. Accordingly, despite assurances that their functional status information will be kept confidential, these fears could affect their responses and lead them to report better functioning than they really have. Although these fears may appear irrational to well-motivated data gatherers wanting to do good (e.g., by collecting relevant data), these feelings reflect a lifetime of confronting disenfranchisement and discrimination. People thus may answer in the way that they think will be least harmful to them. Here, as in the Karnofsky scale example, fundamental problems in the U.S. health care system discourage the idea of routinely gathering meaningful data about disability.

Acknowledgments

This manuscript was funded by the Agency for Healthcare Research and Quality.

Endnote

1Depeding on the state, other categories of People with disabilities qualify for medicaid, such as People certified as blind. Following so-called 209(b) options, states can tighten medicaid eligibility requirements beyond the standard SSI disability or means tests (Tanenbaum 1989). States may also liberalize medicaid eligibility under the Social security Act amendments encouraging SSI recipients to return to work.

References

  • Adams EK, Ellwood MR, Pine PL. Utilization and Expenditures under Medicaid for Supplemental Security Income Disabled. Health Care Financing Review. 1989;11(1):1–24. [PubMed]
  • Adler GS. A Profile of the Medicare Current Beneficiary Survey. Health Care Financing Review. 1994;15(4):153–63. [PubMed]
  • Adler GS. Medicare Beneficiaries Rate Their Medical Care: New Data from the MCBS. Health Care Financing Review. 1995;16(4):175–87. [PubMed]
  • Agency for Health Care Policy and Research (AHCPR) PORT & PORT-II Abstracts. Rockville, Md.: Agency for Health Care Policy and Research; 1996. AHCPR Pub. no. 97-R017., November.
  • Agency for Health Care Policy and Research (AHCPR) Medical Treatment Effectiveness Program. Research Projects 1989 through 1996. Rockville, Md.: Agency for Health Care Policy and Research; 1997. AHCPR Pub. no. 97-R047. March.
  • Ash AS, Ellis RP, Pope GC, et al. Using Diagnoses to Describe Populations and Predict Costs. Health Care Financing Review. 2000;21:7–28. [PubMed]
  • Barnes C, Mercer G, Shakespeare T. Exploring Disability. A Sociological Introduction. Cambridge: Polity Press; 1999.
  • Batavia AI. Ten Years Later: The ADA and the Future of Disability Policy. In: Francis LP, Silvers A, editors. Americans with Disabilities: Exploring Implications of the Law for Individuals and Institutions. New York: Routledge; 2000. pp. 283–92.
  • Bierman AS. Functional Status: The Sixth Vital Sign. Journal of General Internal Medicine. 2001;16:785–6. [PMC free article] [PubMed]
  • Bogardus ST, Towle V, Williams CS, Desai MM, Inouye SK. What Does the Medical Record Reveal about Functional Status? A Comparison of Medical Record and Interview Data. Journal of General Internal Medicine. 2001;16:728–36. 10.1046/j.1525-1497.2001.00625.x. [PMC free article] [PubMed]
  • Boult C, Rassen J, Rassen A, Moore RJ, Robison S. The Effect of Case Management on the Costs of Health Care for Enrollees in Medicare Plus Choice Plans: A Randomized Trial. Journal of the American Geriatrics Society. 2000;48:996–1001. [PubMed]
  • Boult L, Boult C, Pirie P, Pacala JT. Test-Retest Reliability of a Questionnaire That Identifies Elders at Risk for Hospital Admission. Journal of the American Geriatrics Society. 1994;42:707–11. [PubMed]
  • Bright RA, Avorn J, Everitt DE. Medicaid Data as a Resource for Epidemiologic Studies: Strengths and Limitations. Journal of the American Medical Association. 1989;42:937–45. [PubMed]
  • Brown F. ICD-9-CM Coding Handbook, without Answers. Chicago: American Hospital Publishing; 1989.
  • Bye BV, Dykacz JM, Hennessey JC, Riley GF. Medicare Costs prior to Retirement for Disabled-Worker Beneficiaries. Social Security Bulletin. 1991;54:2–23. [PubMed]
  • Bye BV, Riley GF. Eliminating the Medicare Waiting Period for Social Security Disabled-Worker Beneficiaries. Social Security Bulletin. 1989;52:2–15. [PubMed]
  • Bye BV, Riley GF, Lubitz J. Medicare Utilization by Disabled-Worker Beneficiaries: A Longitudinal Analysis. Social Security Bulletin. 1987;50:13–28. [PubMed]
  • Calkins DR, Rubenstein LV, Cleary PD, et al. Failure of Physicians to Recognize Functional Disability in Ambulatory Patients. Annals of Internal Medicine. 1991;114:451–4. [PubMed]
  • Cassel CK, Besdine RW, Siegel LC. Restructuring Medicare for the Next Century: What Will Beneficiaries Really Need? Health Affairs. 1999;18:118–31. [PubMed]
  • Chan L, Doctor JN, MacLehose RF, et al. Do Medicare Patients with Disabilities Receive Preventive Services? A Population-Based Study. Archives of Physical Medicine and Rehabilitation. 1999;80:642–6. [PubMed]
  • Charlton JI. Nothing about Us without Us: Disability Oppression and Empowerment. Berkeley and Los Angeles: University of California Press; 1998.
  • Chirikos TN. The Economics of Employment. In: West J, editor. The Americans with Disabilities Act: From Policy to Practice. New York: Milbank Memorial Fund; 1991. pp. 150–79. edited by.
  • Clancy CM, Eisenberg JM. Outcomes Research at the Agency for Health Care Policy and Research. Disease Management and Clinical Outcomes. 1997;1(3):72–80. 10.1016/S1088-3371(98)80003-X.
  • Culler SD, Callahan CM, Wolinsky FD. Predicting Hospital Costs among Older Decedents over Time. Medical Care. 1995;33:1089–105. [PubMed]
  • Dolan P. The Effect of Experience of Illness on Health State Valuations. Journal of Clinical Epidemiology. 1996;49:551–64. [PubMed]
  • Donaldson MS, Lohr KN. Health Data in the Information Age: Use, Disclosure, and Privacy. Washington, D.C.: National Academy Press; 1994.
  • Elixhauser A, Andrews R, Ball J, Coffey R, Johantgen M, Purcell P. Data Needs in an Era of Health Reform, 359. Washington, D.C.: U.S. Department of Health and Human Services; 1996. Lessons for National Standards on Health Data: State Practices in Building Databases on Hospital Inpatient Stays. Proceedings of the 25th Public Health Conference on Records and Statistics and the National Committee on Vital and Health Statistics 45th Anniversary Symposium.
  • Ellis RP, Pope GC, Iezzoni LI, et al. Diagnosis-Based Risk Adjustment for Medicare Capitation Payments. Health Care Financing Review. 1996;17:101–28. [PubMed]
  • Ettner SL, Kuhlthau K, McLaughlin TJ, Perrin JM, Gortmaker SL. Impact of Expanding SSI on Medicaid Expenditures of Disabled Children. Health Care Financing Review. 2000;21:185–201. [PubMed]
  • Fisher ES, Whaley FS, Krushat WM, et al. The Accuracy of Medicare's Hospital Claims Data: Progress Has Been Made, but Problems Remain. American Journal of Public Health. 1992;82:243–8. [PMC free article] [PubMed]
  • Fishman PA, Shay DK. Development and Estimation of a Pediatric Chronic Disease Score Using Automated Pharmacy Data. Medical Care. 1999;37:874–83. 10.1097/00005650-199909000-00004. [PubMed]
  • Foote SM, Hogan C. Disability Profile and Health Care Costs of Medicare Beneficiaries under Age Sixty-Five. Health Affairs. 2001;20(6):242–53. [PubMed]
  • Fries BE, Schneider DP, Foley WJ, Gavazzi M, Burke R, Cornelius E. Refining a Case-Mix Measure for Nursing Homes: Resource Utilization Groups (RUGS-III) Medical Care. 1994;32:668–85. [PubMed]
  • Gitterman DP, Strum R, Scheffler RM. Toward Full Mental Health Parity and Beyond. Health Affairs. 2001;20(4):68–76. [PubMed]
  • Greenwald LM. Medicare Risk-Adjusted Capitation Payments: From Research to Implementation. Health Care Financing Review. 2000;21:1–5. [PubMed]
  • Harris Interactive, Inc. 2000 N. O. D./Harris Survey of Americans with Disabilities. New York: Conducted for the National Organization on Disability; 2000.
  • Hawes C, Mor V, Phillips CD, et al. The OBRA-87 Nursing Home Regulations and Implementation of the Resident Assessment Instrument: Effects on Process Quality. Journal of the American Geriatrics Society. 1997;45:977–85. [PubMed]
  • Health Care Financing Administration (U.S. Department of Health and Human Services) Medicare and Medicaid Programs; Revision of Conditions of Participation for Home Health Agencies and Use of Outcome Assessment Information Set (OASIS); Proposed Rules. Federal Register. 1997;62:11004–64. March 10,42 CFR pt. 484.
  • Hermann RC, Ettner SL, Dorwart RA. The Influence of Psychiatric Disorders on Patients’ Ratings of Satisfaction with Health Care. Medical Care. 1998;36:720–7. 10.1097/00005650-199805000-00011. [PubMed]
  • Hoenig H. Educating Primary Care Physicians in Geriatric Rehabilitation. Clinics in Geriatric Medicine. 1993;9:883–93. [PubMed]
  • Hornbrook MC, Goodman MJ, Fishman PA, Meenan RT, O'Keefe-Rosetti M, Bachman DJ. Building Health Plan Databases to Risk Adjust Outcomes and Payments. International Journal on Quality in Health Care. 1998;10:531–8. [PubMed]
  • Hsia DC, Ahern CA, Ritchie BP, Moscoe LM, Krushat WM. Medicare Reimbursement Accuracy under the Prospective Payment System, 1985 to 1988. Journal of the American Medical Association. 1992;268:896–9. [PubMed]
  • Hsia DC, Krushat WM, Fagan AB, Tebbutt JA, Kusserow RP. Accuracy of Diagnostic Coding for Medicare Patients under the Prospective-Payment System. New England Journal of Medicine. 1988;318:352–55. [PubMed]
  • Iezzoni LI. Assessing Quality Using Administrative Data. Annals of Internal Medicine. 1997a;127:666–74. [PubMed]
  • Iezzoni LI. Data Sources and Implications: Administrative Databases. In: Iezzoni LI, editor. Risk Adjustment for Measuring Healthcare Outcomes. Chicago: Health Administration Press; 1997b. pp. 169–242.
  • Iezzoni LI, Ayanian JZ, Bates DW, Burstin H. Paying More Fairly for Medicare Capitated Care. New England Journal of Medicine. 1998;339:1933–8. [PubMed]
  • Iezzoni LI, Foley SM, Daley J, Hughes J, Fisher ES, Heern T. Comorbidities, Complications, and Coding Bias: Does the Number of Diagnosis Codes Matter in Predicting In-Hospital Mortality? Journal of the American Medical Association. 1992;267:2197–2203. [PubMed]
  • Iezzoni LI, McCarthy EP, Davis RB, Siebens H. Mobility Problems and Perceptions of Disability by Self- and Proxy-Respondents. Medical Care. 2000;38:1051–7. [PubMed]
  • Institute of Medicine (IOM) Disability in America: Toward a National Agenda for Prevention. Washington, D.C.: National Academy Press; 1991.
  • Institute of Medicine (IOM) Enabling America: Assessing the Role of Rehabilitation Sciences and Engineering. Washington, D.C.: National Academy Press; 1997.
  • Institute of Medicine (IOM) Coverage Matters: Insurance and Health Care. Washington, D.C.: National Academy Press; 2001. Committee on the Consequences of Uninsurance.
  • Jencks SF, Williams DK, Kay TL. Assessing Hospital-Associated Deaths from Discharge Data: The Role of Length of Stay and Comorbidities. Journal of the American Medical Association. 1988;260:2240–6. [PubMed]
  • Kane RL. Choosing and Using an Assessment Tool. In: Kane RL, Kane RA, editors. Assessing Older Persons: Measures, Meaning, and Practical Applications. New York: Oxford University Press; 2000. pp. 1–13.
  • Karnofsky DA, Abelmann WH, Craver LF, Burchenal JH. The Use of the Nitrogen Mustards in the Palliative Treatment of Carcinoma. Cancer. 1948;1:634–8.
  • Kronick R, Dreyfus T, Lee L, Zhou Z. Diagnostic Risk Adjustment for Medicaid: The Disability Payment System. Health Care Financing Review. 1996;17:7–33. [PubMed]
  • Kronick R, Gilmer T, Dreyfus T, Lee L. Improving Health-Based Payment for Medicaid Beneficiaries: CDPS. Health Care Financing Review. 2000;21:29–64. [PubMed]
  • Kronick R, Zhou Z, Dreyfus T. Making Risk Adjustment Work for Everyone. Inquiry. 1995;32:41–55. [PubMed]
  • Kuhlthau K, Perrin JM, Ettner SL, McLaughlin TJ, Gortmaker SL. High-Expenditure Children with Supplemental Security Income. Pediatrics. 1998;102:610–5. [PubMed]
  • Lamers LM. Pharmacy Cost Groups: A Risk-Adjuster for Capitation Payment Based on the Use of Prescription Drugs. Medical Care. 1999;37:824–30. 10.1097/00005650-199908000-00012. [PubMed]
  • Laschober MA, Neuman P, Kitchman MS, Meyer L, Langwell KM. Medicare HMO Withdrawals: What Happens to Beneficiaries. Health Affairs. 1999;18(6):150–7. [PubMed]
  • Lave JR, Pashos CL, Anderson GF, et al. Costing Medical Care: Using Medicare Administrative Data. Medical Care. 1994;32:JS77–JS89. [PubMed]
  • Lillard LA, Farmer MM. Linking Medicare and National Survey Data. Annals of Internal Medicine. 1997;127(2) 8:691–5. [PubMed]
  • Mairs N. Waist-High in the World. Boston: Beacon Press; 1996.
  • Manton KG, Corder L, Stallard E. Changes in the Use of Personal Assistance and Special Equipment, from 1982 to 1989: Results from the 1982 and 1989 NLTCS. Gerontologist. 1993;33(2):168–76. [PubMed]
  • Martin L, Chin C, Harrison CA. Annual Statistical Report on the Social Security Disability Insurance Program, 2000. 2001. [accessed December 28, 2001]. Available at: http://www.ssa.gov/statistical/di_asr/2000/index.html.
  • Master R, Dreyfus T, Connors S, Tobias C, Kronick R. The Community Medical Alliance: An Integrated System of Care in Greater Boston for People with Severe Disability and AIDS. Managed Care Quarterly. 1996;4:26–37. [PubMed]
  • Mathiowetz N, Wunderlich GS, editors. Survey Measurement of Work Disability: Summary of a Workshop. Washington, D.C.: National Academy Press; 2000.
  • McCarthy EP, Iezzoni LI, Davis RB, et al. Does Clinical Evidence Support ICD-9-CM Diagnosis Coding of Complications? Medical Care. 2000;38:868–76. 10.1097/00005650-200008000-00010. [PubMed]
  • Medicare Payment Advisory Commission. Report to the Congress: Medicare Payment Policy. Analytical Papers. Vol. 2. Washington, D.C.: Medicare Payment Advisory Commission; 1998. March.
  • Medicare Payment Advisory Commission. Report to the Congress: Medicare Payment Policy. Washington, D.C.: Medicare Payment Advisory Commission; 2001. March.
  • Meyer JA, Zeller PJ. Profiles of Disability: Employment and Health Coverage. Washington, D.C.: Henry J. Kaiser Family Foundation; 1999.
  • Mitchell JB, Bubolz T, Paul JE, et al. Using Medicare Claims for Outcomes Research. Medical Care. 1994;32:JS38–JS51. [PubMed]
  • Mor V, Wilcox V, Rakowski W, Hiris J. Functional Transitions among the Elderly: Patterns, Predictions, and Related Hospital Use. American Journal of Public Health. 1994;84:1274–80. [PMC free article] [PubMed]
  • Morris JN, Murphy K, Nonemaker S. Long Term Care Resident Assessment Instrument User's Manual. Baltimore: Health Care Financing Administration; 1995. , October Version 2.0.
  • National Cancer Institute. Linking SEER and Medicare Data. 2001. [accessed February 10, 2001]. Available at: http://www-dccps.ims.nci.nih.gov/ARP/SEERMed/information.html.
  • National Committee for Quality Assurance. HEDIS 3.0 (Health Plan Employer Data Information Set) Washington, D.C.: 1996.
  • National Committee on Vital and Health Statistics. Classifying and Reporting Functional Status. Hyattsville, Md.: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Health Statistics; 2001. , June.
  • Nelson E, Conger B, Douglass R, et al. Functional Health Status Levels of Primary Care Patients. Journal of the American Medical Association. 1983;249:3331–8. [PubMed]
  • Office of Technology Assessment. Bringing Health Care Online: The Role of Information Technologies. Washington, D.C.: U.S. Government Printing Office; 1995. OTA-ITC-624.
  • Oliver M. Understanding Disability: From Theory to Practice. New York: St. Martin's Press; 1996.
  • Pacala JT, Boult C, Boult L. Predictive Validity of a Questionnaire That Identifies Older Persons at Risk for Hospital Admission. Journal of the American Geriatrics Society. 1995;43:374–7. [PubMed]
  • Pacala JT, Boult C, Reed RL, Aliberti E. Predictive Validity of the Pra Instrument among Older Recipients of Managed Care. Journal of the American Geriatrics Society. 1997;45:614–7. [PubMed]
  • Pelka F. The ABC-CLIO Companion to the Disability Rights Movement. Santa Barbara, Calif.: ABC-CLIO; 1997.
  • Perrin JM, Ettner SL, McLaughlin TJ, Gortmaker SL, Bloom SR, Kuhlthau K. State Variations in Supplemental Security Income Enrollment for Children and Adolescents. American Journal of Public Health. 1998a;88:928–31. [PMC free article] [PubMed]
  • Perrin JM, Kuhlthau K, Ettner SL, McLaughlin TJ, Gortmaker SL. Previous Medicaid Status of Children Newly Enrolled in Supplemental Security Income. Health Care Financing Review. 1998b;19:117–27.
  • Perrin JM, Kuhlthau K, McLaughlin TJ, Ettner SL, Gortmaker SL. Changing Patterns of Conditions among Children Receiving Supplemental Security Income Disability Benefits. Archives of Pediatric and Adolescent Medicine. 1999;153:80–4. [PubMed]
  • Pope GC, Ellis RP, Ash AS, et al. Diagnostic Cost Group Hierarchical Condition Category Models for Medicare Risk Adjustment. Final Report. Waltham, Mass.: Health Economics Research; 2000a. Prepared for the Health Care Financing Administration under Contract no. 500-95-048.
  • Pope GC, Ellis RP, Ash AS, et al. Principal Inpatient Diagnostic Cost Group Model for Medicare Risk Adjustment. Health Care Financing Review. 2000b;21:93–118. [PubMed]
  • Potosky AL, Riley GF, Lubitz JD, Mentnech RM, Kessler LG. Potential for Cancer Related Health Services Research Using a Linked Medicare-Tumor Registry Database. Medical Care. 1993;31:732–48. [PubMed]
  • Ray WA, Griffin MR. Use of Medicaid Data for Pharmacoepidemiology. American Journal of Epidemiology. 1989;129:837–49. [PubMed]
  • Riley GF. Risk Adjustment for Health Plans Disproportionately Enrolling Frail Medicare Beneficiaries. Health Care Financing Review. 2000;21:135–48. [PubMed]
  • Robinson JC. The Corporate Practice of Medicine: Competition and Innovation in Health Care. Berkeley and Los Angeles: University of California Press; 1999.
  • Robinson JC, Karon SL. Modeling Medicare Costs of PACE Populations. Health Care Financing Review. 2000;21:149–70. [PubMed]
  • Rockow K. Working with Deaf and Hard of Hearing Patients. A Guide for Medical Professionals. Allston, Mass.: D.E.A.F., Inc.; 2001.
  • Roetzheim RG, Chirikos TN. Breast Cancer Detection and Outcomes in a Disability Beneficiary Population. Journal of Health Care for the Poor and Underserved. 2002 in press. [PubMed]
  • Romano PS. Can Administrative Data Be Used to Compare the Quality of Health Care? Medical Care Review. 1993;50:451–77. [PubMed]
  • Romano PS, Luft HS. Getting the Most out of Messy Data: Problems and Approaches for Dealing with Large Administrative Data Sets. In: Grady ML, Schwartz HA, editors. Medical Effectiveness Research Data Methods. Rockville, Md.: Agency for Health Care Policy and Research; 1992. pp. 57–75. AHCPR Pub. no. 92-0056.
  • Romano PS, Mark DH. Bias in the Coding of Hospital Discharge Data and Its Implications for Quality Assessment. Medical Care. 1994;32:81–90. [PubMed]
  • Rosenbach ML. Access and Satisfaction within the Disabled Medicare Population. Health Care Financing Review. 1995;17(2):147–67. [PubMed]
  • Schneider DP, Fries BE, Foley WJ, Desmond M, Gormley WJ. Health Care Financing Review. 1988. Case Mix for Nursing Home Payment: Resource Utilization Groups, Version II. , spec. no. 39-52. [PubMed]
  • Sharma R, Chan S, Liu H, Ginsberg C. Health and Health Care of the Medicare Population. Data from the 1997 Medicare Current Beneficiary Survey. Rockville, Md.: Westat.; 2001.
  • Shaughnessy PW, Crisler KS, Schlenker RE, Arnold AG. Outcomes across the Care Continuum: Home Health Care. Medical Care. 1997;35:NS115–NS123. 10.1097/00005650-199711001-00013. [PubMed]
  • Shaughnessy PW, Schlenker RE, Hittle DF. Case Mix of Home Health Patients under Capitated and Fee-for-Service Payment. Health Services Research. 1995;30:79–113. [PMC free article] [PubMed]
  • Social Security Administration (SSA), Office of Disability. Washington, D.C.: 1998. Disability Evaluation under Social Security. SSA Publ. no. 64-039.
  • Stineman MG. Measuring Casemix, Severity, and Complexity in Geriatric Patients Undergoing Rehabilitation. Medical Care. 1997;35:JS90–JS105. 10.1097/00005650-199706001-00018. [PubMed]
  • Stineman MG, Escarce JJ, Goin JE, Hamilton BB, Granger CV, Williams SV. A Case-Mix Classification System for Medical Rehabilitation. Medical Care. 1994;32:366–79. [PubMed]
  • Stineman MG, Goin JE, Tassoni CJ, Granger CV, Williams SV. Classifying Rehabilitation Inpatients by Expected Functional Gain. Medical Care. 1997;35:963–73. [PubMed]
  • Stone DA. The Disabled State. Philadelphia: Temple University Press; 1984.
  • Swan J, Newcomer R. Residential Care Supply, Nursing Home Licensing, and Case Mix in Four States. Health Care Financing Review. 2000;21:203–29. [PubMed]
  • Tanenbaum S. Medicaid and Disability: The Unlikely Entitlement. Milbank Quarterly. 1989;67(2) supp. 2:288–310. [PubMed]
  • U.S. General Accounting Office. SSA Disability. Program Redesign Necessary to Encourage Return to Work. Washington, D.C.: 1996. GAO/HEHS-96-62.
  • U.S. General Accounting Office. Medicare Home Health Care: Prospective Payment System Could Reverse Recent Declines in Spending. Washington, D.C.: 2000. , September GAO/HEHS-00-176.
  • U.S. General Accounting Office. Record Linkage and Privacy. Issues in Creating New Federal Research and Statistical Information. Washington, D.C.: 2001. GAO-01-126SP.
  • Virnig BA, McBean M. Administrative Data for Public Health Surveillance and Planning. Annual Review of Public Health. 2001;22:213–29. 10.1146/annurev.publhealth.22.1.213. [PubMed]
  • Williams BC, Li Y, Fries BE, Warren RL. Predicting Patient Scores between the Functional Independence Measure and the Minimum Data Set: Development and Performance of a FIM-MDS “Crosswalk.” Archives of Physical Rehabilitation. 1997;78:48–54. [PubMed]
  • Williams G. Theorizing Disability. In: Albrecht GL, Seelman KD, Bury M, editors. Handbook of Disability Studies. Thousands Oaks, Calif.: Sage Publications; 2001. pp. 123–44.
  • World Health Organization (WHO) International Classification of Impairments, Disabilities, and Handicaps. Geneva: 1980.
  • World Health Organization (WHO) ICIDH-2: International Classification of Impairments, Activities and Participation. A Manual of Dimensions of Disablement and Functioning. Beta-1 Draft for Field Testing. Geneva: 1997.
  • World Health Organization (WHO) International Classification of Functioning, Disability and Health. Geneva: 2001.
  • Young NL, Williams JI, Yoshida KK, Bombardier C, Wright JG. The Context of Measuring Disability: Does It Matter Whether Capability or Performance Is Measured? Journal of Clinical Epidemiology. 1996;49:1097–101. [PubMed]

Articles from The Milbank Quarterly are provided here courtesy of Milbank Memorial Fund
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

  • PubMed
    PubMed
    PubMed citations for these articles

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...