 |
 |
National Center for Chronic Disease Prevention and Health Promotion
Home | About the Program | Site Map | Contact Us
|
||||||||
|
|
|
|
|
|
|
||
|
National Center for Chronic Disease Prevention and Health Promotion Home | About the Program | Site Map | Contact Us |
|
|
||||||
|
|
|
Data & TrendsDiabetes Surveillance System1999 Surveillance ReportAppendix
Data Sources:
We determined the incidence and prevalence of diagnosed diabetes in the United States by using data from the National Health Interview Survey (NHIS) of the National Center for Health Statistics (NCHS), CDC. Conducted since 1957, NHIS is an annual household survey of approximately 120,000 civilian, noninstitutionalized United States residents. The survey provides information on the health of the United States population, including information on the prevalence and incidence of disease, the extent of disability, and the utilization of health care services. NHIS has a multistage probability design that has been described elsewhere (1). Each year, a one-sixth subsample of NHIS respondents are asked whether in the past 12 months any family member has had diabetes. If a household member has diabetes, the date of diagnosis is ascertained. In this report, diabetes prevalence was defined as the number of persons with diabetes, regardless of onset. Diabetes incidence was defined as the number of persons who were diagnosed within the past year. Three-year moving averages were used to improve the precision of the incidence and prevalence estimates. Prevalence and incidence estimates were applied to estimates of the U.S. resident population to determine the number of persons with diagnosed diabetes in the United States. State-specific prevalence of diabetes was calculated using data from CDC's Behavioral Risk Factor Surveillance System (BRFSS). The BRFSS is an ongoing, monthly, state-based telephone survey of the noninstitutionalized adult population of states. The survey provides state-specific information on behavioral risk factors and preventive health practices. Respondents were considered to have diabetes if they responded "yes" to the question, "Has a doctor ever told you that you have diabetes?" Women who indicated that they only had diabetes during pregnancy were not considered to have diabetes. A 3-year average was used to improve the precision of the state-specific prevalence estimates. Data Limitations NHIS and the BRFSS underestimate the true prevalence of diabetes. About one-third of persons with diabetes do not know they have it (2). Also, NHIS proxy respondents (i.e., household members responding for absent adult members) are also likely to underreport diabetes. Data Sources:
The National Center for Health Statistics (NCHS) at CDC compiles and codes information on all deaths in the United States and releases annual underlying-cause-of-death and multiple-cause-of-death data tapes. Data on these tapes include decedents' age, race, sex, state of residence, and the underlying cause of death. In addition to these and other variables, the multiple-cause-of-death tapes contain data on up to 20 causes of death (including underlying and contributing causes) for each decedent. Causes of death for each decedent are classified and coded according to the International Classification of Diseases, Ninth Revision (ICD-9). We used these data tapes to extract information on deaths associated with diabetes (ICD-9 code 250) and to examine trends in diabetes as the underlying cause of death and as any listed cause of death. Among deaths for which diabetes was a listed cause, deaths for which the corresponding underlying cause was DKA (ICD-9 code 250.1), stroke (ICD-9 codes 430-434, 436-438), ischemic heart disease (IHD) (ICD-9 codes 410-414), or major CVD (ICD-9 390-448) were also examined. Data Limitations Diabetes is underreported on death certificates. Among persons known to have diabetes, only about 40% have diabetes listed as a cause of death and only 10% have diabetes recorded as the underlying cause of death (3,4). Because only 60% of decedents with diabetes have diabetes recorded as a cause of death, death certificate data cannot be used to examine overall mortality among persons with diabetes. Furthermore, decedents with diabetes recorded as a cause of death are not representative of decedents known to have diabetes (3,4). Although the frequency of recording diabetes on death certificates does not appear to vary by sex, race, or ethnicity, the likelihood of recording diabetes increases with duration of diabetes, decreases with age among those with durations of diabetes of 15 or more years, is higher when comorbidities frequently related to diabetes are also recorded (e.g., ischemic heart, hypertensive, renal, cerebrovascular, and arterial disease), and is higher among those who developed diabetes before age 30 (3). Deaths and death rates may be underestimated for minority populations (5). Data Source:
We used data from NCHS's National Hospital Discharge Survey (NHDS) to estimate diabetes-related hospital discharges. NHDS collects data on hospital discharges from a sample of short-stay, nonfederal hospitals in the United States. Data collected include information on patients' age, race, sex, and length of stay, and on seven diagnoses (one primary and six secondary diagnoses) and four surgical procedures. Methods used for conducting the survey have been described previously (6). Hospital discharges for which diabetes was the primary diagnosis (ICD-9 code 250) and for which diabetes was any listed diagnosis (diabetes-related discharges) were examined. Among discharges with diabetes as a listed diagnosis, discharges for which the primary diagnosis was DKA (ICD-9 code 250.1), stroke (ICD-9 codes 430-434, 436-438), IHD (ICD-9 codes 410-414), or major CVD (ICD-9 390-448) were estimated. We also used NHDS data to examine the incidence of nontraumatic lower extremity amputation. Incident cases were defined as discharges having diabetes as a listed diagnosis and a lower extremity amputation (ICD-9 procedure code of 84.1). Discharges with traumatic amputation diagnosis code (ICD-9 diagnoses codes 895-897) were excluded. Data Limitations Hospitalizations related to diabetes may be underestimated by approximately 40% (7). Underestimation also results from the exclusion of long-term and federal hospitals from the NHDS sample. Race-specific discharges are particularly underestimated because a substantial proportion of discharges are missing racial classification and missing values for race are not imputed (6). Because NHDS samples hospital discharges and not individual persons, NHDS hospital discharge rates for diabetes-related diseases and procedures may not necessarily reflect rates per person; that is, persons who are hospitalized more than once for the same condition may be counted more than once. In 1983, Medicare instituted a prospective payment system that has influenced both hospitalization practices and disease reporting on discharge records. The incidence of nontraumatic lower extremity amputation may be underestimated by data on hospital discharges. Some amputations may be performed in outpatient settings. However, the extent to which outpatient surgery for amputation is occurring is unknown.
Data Sources:
These three surveys are conducted by NCHS, CDC and provide data on different aspects of the use of health care services among persons with diabetes. For a description of the NHIS see the previous section of Prevalence and Incidence. We used NHIS to estimate the number and rate of physician contacts and physician visits among persons with diabetes. Because these estimates include all contacts and visits regardless of purpose or reason, they reflect health service use among persons with diabetes, rather than use because of diabetes. The National Ambulatory Care Survey (NAMCS) is a national cross-sectional survey conducted annually that provides data on office visits made in the United States by ambulatory patients to nonfederally employed physicians who are principally engaged in office practice (excludes anesthesiology, pathology, and radiology). The basic sampling unit is the physician-patient encounter or visit. Data collected include physician's practice characteristics, patient and visit characteristics, and the physician's diagnoses (one principal and up to 2 additional diagnoses). A diabetes visit was defined as an ICD-9 diagnosis code of 250 listed as any one of three diagnoses. In contrast to the NHIS, NAMCS data are collected from physicians, rather than the patients, telephone contacts are excluded, and diabetes had to be one of three diagnoses listed by the physician as associated with the ambulatory care visit. Methods used for conducting the survey have been described previously (8). The National Hospital Ambulatory Medical Care Survey (NHAMCS) is a national cross-sectional survey conducted annually that provides data on visits made in the United States. to the emergency departments and outpatient departments of nonfederal, short-stay, and general hospitals. Clinics that specialize in radiology, laboratory sciences, physical rehabilitation, or other ancillary services were excluded from the survey. Data are collected on patient and visit characteristics, and the physician's diagnoses (one principal and up to 2 additional diagnoses). A diabetes visit was defined as an ICD-9 diagnosis code of 250 listed in any of the three physician diagnosis fields. Details of the survey methodology have been described previously (9). Data Limitations All three surveys may underestimate the use of health care services by persons with diabetes, since one-third of all persons with diabetes do not know that they have it (2). Physician contacts and visits as measured by NHIS reflect health service use among persons with diabetes rather than use due to diabetes. Therefore, these estimates overestimate health services used because of diabetes. Both NAMCS and NHAMCS may underreport diabetes-related visits. Because persons with diabetes often have comorbid conditions, physicians may not always list diabetes as one of the three diagnoses that can be recorded. However, the extent of this underreporting is unknown. Because NAMCS and NHAMCS sample patient visits and not individuals persons, rates developed from this data are not necessarily rates per person. For example, persons who are seen in an outpatient clinic 2 or more times during the 4-week reporting period of NHAMCS would be counted more than once. In some cases, a portion of emergency department, hospital outpatient, and ambulatory care may be provided by telephone or by nonphysician providers. These encounters are not included in the NAMCS and NHAMCS. Data Source:
The U.S. Renal Data System (USRDS) is a surveillance system for end-stage renal disease (ESRD). The system is funded by the National Institute of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health. The Health Care Financing Administration (HCFA) supplies most of the data used by the USRDS and provides expertise on data management. We used USRDS data to examine the incidence and prevalence of ESRD attributed to diabetes (ESRD-DM) because more than 90% of ESRD treatment in the United States is reimbursed by HCFA's Medicare program. We defined ESRD-DM incidence as cases initiating treatment for ESRD (dialysis or kidney transplantation) and having diabetes listed as the primary cause of renal failure. Data Limitations The data are for persons receiving ESRD treatment as reported to HCFA and do not include patients who die of ESRD before receiving treatment and those who are not reported to HCFA. Ascertainment of incident cases was incomplete because Medicare reimburses about 90% of ESRD treatment (10,11). Because the incidence of ESRD attributed to diabetes was defined in terms of initiation of ESRD treatment, changes in incidence may have been due to changes in factors other than disease incidence. These include changes in reporting diabetes as the primary cause of renal failure and changes in the use of treatment. The latter may be influenced by changes in treatment availability and in the definition of treatment eligibility (12). The count of new ESRD patients for 1993 was lower than expected from prior trends and persisted even with the usual updating done by the USRDS and HCFA (12). According to the USRDS, a compensatory overcount of new patients appeared to occur in 1994, and presumably, some of these patients were truly incident in 1993. Why this occurred is not fully understood. In addition, a new Medical Evidence form introduced in 1995 and required for all new dialysis patients is a source of duplicate records as non-Medicare patients become Medicare-entitled (11). Data Source:
We derived indicators of disability among persons with diabetes from the NHIS (see the brief description of this data source in the previous section on Prevalence and Incidence). Two of the major indicators of disability used in the NHIS are activity limitation and activity restriction. Activity limitation reflects a long-term reduction in activity resulting from one or more chronic diseases or impairments. Reduction in activity is measured in terms of activities normal for a person's age-sex group: "ordinary play" for children aged <5 years, "going to school" for children aged 5-17 years, "working at a job or business" or "keeping house" for persons aged 18-69 years, and independent performance of basic life activities (e.g., bathing, eating, shopping) for persons aged greater than or equal to 70 years. Persons can be categorized as being (1) unable to perform their major activity, (2) able to perform their major activity but limited in the kind or amount of this activity, (3) not limited in major activity but limited in other activities, and (4) not limited in activity. This analysis examined persons limited in activity (categories 1-3), limited in major activity (categories 1-2), and unable to perform major activity (category 1). Three-year moving averages were used to improve precision of the estimates. The other major indicator of disability used in NHIS is activity restriction. This indicator refers to a reduction in activity caused by either short-term or long-term conditions. Activity restriction is measured as school-loss days (for children aged 5-17 years), work-loss days (for the currently employed aged 18-69 years), cut-down days (days in which persons reduce or cut down on the things they usually do), and bed days (inpatient hospital days or days in which a person stayed in bed for more than half a day because of illness or injury). The total number of restricted activity days is the total number of days that a person experiences at least one of the previously described types of days. Because of small sample sizes, this analysis only presents data in 3-year moving averages on total restricted activity days and bed days. Data Limitations Although NHIS provides a stable source of annual estimates of disability, it does not sample the institutionalized United States population, which accounts for a significant proportion of all disability. Therefore, estimates of disability derived from NHIS underestimate the total amount of disability associated with diabetes. In 1982, NHIS changed the way it measured disability indicators. For this reason, analysis of these indicators began with 1983 data. Also see limitations mentioned under Prevalence and Incidence Data. Data Sources:
We used population estimates to calculate rates. We also used estimates of the diabetic population (derived by applying NHIS prevalence rates to population estimates) to calculate rates. Data Limitations For limitations of NHIS, see the previous discussion of limitations under Prevalence and Incidence of Diabetes. Data Source:
The prevalence of diabetes-related preventive care practices in the United States was determined by using data from the Behavioral Risk Factor Surveillance System (BRFSS). An ongoing, monthly, state-based telephone survey of the noninstitutionalized adult population in each state, the BRFSS provides state-specific information on behavioral risk factors for disease and on preventive health practices. Respondents were considered to have diabetes if they responded "yes" to the question, "Has a doctor ever told you that you have diabetes?" Women who indicated that they only had diabetes during pregnancy were not considered to have diabetes. Among persons with diabetes, only persons who were asked the questions from the BRFSS diabetes module were included in the analysis. Responses to the following questions were used to determine the prevalence of diabetes-related preventive care practices among persons with diabetes: "When was the last time you had an eye exam in which the pupils were dilated?" "About how many times in the last year has a health professional checked your feet for any sores or irritations?" "About how often do you check your blood for glucose or sugar?" "About how many times in the last year has a doctor, nurse, or other health professional checked you for glycosylated hemoglobin or hemoglobin 'A one C'?" "During the past 12 months, have you had a flu shot?" "Have you ever had a pneumonia vaccination?" Only persons who reported having seen a physician within the past year were asked if they had their feet examined, and only patients who had seen a physician within the past year and heard of the term "glycosylated hemoglobin" or "hemoglobin 'A one C' " were asked how many times had they had their glycosylated hemoglobin checked. We assumed that persons who were not asked the questions did not receive the services. The standard population used for age-adjusted estimates was the 1994 BRFSS diabetic population. With the exception of receiving the influenza and pneumococcal vaccinations, a 3-year average for the prevalence of all preventive care practices was used to improve the precision of the estimates. Each 3-year estimate is composed of at least 2 years of data. The vaccination questions are included on the survey every other year, so 3-year averages were not used, limiting the ability to stratify by age or sex within state. Data Limitations Persons residing in nursing homes and in households without telephones are not included in this survey; therefore, these results cannot be generalized to those segments of the population. All data in the BRFSS are obtained by self-report and are subject to recall bias or may be underreported or overreported. Self-report of diabetes and self-report of socio-demographic characteristics are highly accurate (13,14). Self-report of a dilated-eye examination and influenza vaccine have been shown to have high accuracy as well (15-17). Self-report of a glycosylated hemoglobin measurement has been shown to have a high sensitivity and low specificity (18). Further investigation of the reliability and validity of self-report of foot examination, pneumococcal vaccination, and self-monitoring of blood glucose is needed.
|
Diabetes & Me