Rates | Age-adjusted Rates | Small Numbers | Behavioral Risk Factor Surveillance System | Race | Cause of Death Coding | Measuring Prenatal Care | Why are the numbers different?

How to Use Alaska Health Profiles Online

Rates

Rates are useful to compare events in population groups of different sizes. The numerator (number of events, such as deaths, births, chronic illnesses, or murders) is divided by the denominator (the entire population at risk) and multiplied by a constant. For example, the number of deaths in Alaska in 2000 (numerator) divided by the total Alaska population in the 2000 census (denominator) multiplied by 100,000 (constant) give a crude death rate of 468 per 100,000.

Crude rates do not adjust for the age of the population. They are helpful (together with the count of events) for showing the burden of disease or death in a population, and also for comparisons between population groups that have similar age distributions.

Death rates (also called mortality rates) are usually expressed as deaths per 100,000 population, for a period such as a year or a multi-year period. Infant mortality and other birth-related measures are expressed as events per 1,000 live births, for a period such as a year or multi-year period.*

Rates based on small numbers of events or small populations may fluctuate from year to year even when there is no "real" change in the likelihood of the event. Comparing confidence intervals can help in evaluating the significance of changes in rate. Combining data from several years makes it possible to calculate rates when the number of events is small.

Alaska Health Profiles Online does not calculate age adjusted rates when the number of events is less than 20. Rates based on such small numbers may not be meaningful.
[back to top]

Age-Adjusted Rates

Age-adjusted rates are calculated so comparisons can be made between populations that have different age distributions. For example, since Alaska has a lower proportion of people over 65 than the United States, it will experience a lower crude death rate. Many diseases, such as cancer, heart disease, and diabetes, are strongly associated with age, and accurate comparisons can be made only with age adjustment.

Death rates in the Health Area Profiles are age adjusted to the standard United States (million) 2000 population. Alaska vital statistics data through 1998 were age adjusted to the standard United States (million) 1940 population. In comparing rates from different years or different sources, always use data age-adjusted to the same standard.

The change to the US 2000 standard age distribution increases most age-adjusted death rates in Alaska. Age-adjusted rates from the Alaska Vital Statistics Annual Reports prior to 1999 cannot be compared to those in reports from 1999 to the present.
[back to top]

Small numbers

Many health indicators for Alaska are based on a small number of events. The 15 leading causes of death in Alaska in 1998, for example, included seven deaths from hypertension, 11 from arteriosclerosis, and 12 from kidney disease. When health events are sub-divided by borough/census area, race, sex, or age, the number that results lacks statistical significance and may inadvertently identify an individual.

Confidentiality issues are likely to arise with small denominators, small numerators, or rare events. Confidentiality is protected by withholding counts less than five for areas smaller than the borough/census area or small populations (such as those defined by age, sex, or race) or by aggregating data over time or over larger geographical areas to produce a larger cell size.

Concerns about the reliability and stability of the data arise with small numbers that represent rare or infrequent events. Twenty events is the usual threshold for reliability for age-adjusted rates. Rates based on fewer than 20 events have relative standard errors of 23 percent or more.

The Health Profiles combine multiple years of data to increase numerator size, or collapse outcome categories, and/or expand the geographic area under consideration. Combining some of the smaller borough/census areas with adjoining areas increases the denominator, the number of events, and the reliability of the rates. Annual rates may be used for major population centers such as Anchorage, while several years of data must be combined to obtain a rate in a sparsely populated region. Rates are not calculated for numerators less than 20. The actual number of events is shown at the borough/census area level.
[back to top]

Behavioral Risk Factor Surveillance System (BRFSS)

Many health indicators are tracked through the Behavioral Risk Factor Surveillance System (BRFSS) system. This data collection process was implemented in Alaska as an ongoing surveillance system in 1991. Alaska adults, age 18 years old and older, are interviewed regarding their health and day-to-day living habits. Households with a telephone are selected by a scientifically designed and conducted random telephone survey. The survey is designed to report population prevalence at a region or state level. Alaska's BRFSS (1998 or later) uses five geographic regions: Anchorage and vicinity, Gulf Coast, Southeast, Rural, and Fairbanks and vicinity.

The U.S. baseline values represent the national median using the number of states collecting the same information through their BRFSS systems, and are age-adjusted to the standard million population. Alaska BRFSS data is not age-adjusted because of the added complexity of this step for within-state analysis. Trends can be observed between the national and state data without age adjustment.

In Alaska, Community Health and Emergency Medical Services (CHEMS) administers BRFSS. More information on the program as well as survey results are available through CHEMS.

National and state BRFSS results are available through the Centers for Disease Control and Prevention.

[back to top]

Race

Explanation of the Bridge Series for Census 2000 Comparable Race Categories to 1990 Major Groups. Excerpted and adapted from Greg Williams, Race and Ethnicity in Alaska, Alaska Economic Trends, October 2001.

The Census Bureau changed the method of reporting race in accordance with the guidelines of the U.S. Office of Management and Budget (OMB) in conducting the 2000 census. All race and ethnicity was self-reported by the respondent and represented each individual's interpretation of the choices presented. In addition to the race and ethnic categories recognized by OMB, the census allowed people to define themselves as "some other race" and to write in their race. The Census Bureau has so far provided race and ethnicity tabulations in the following forms for 2000 data:
1. One race alone or two or more races.
2. Race alone or in combination.
3. 63 race categories.

Racial data from the 2000 census, therefore, cannot be compared directly with previous census results. A National Academy of Sciences panel studied a series of possible ways to combine the 2000 race data to produce race data comparable to the 1990 and earlier definitions of race. These tabulations are referred to as "bridge" estimates, because they allow comparison of two sets of incompatible data.

One way to combine the new race data to create tabulations that are comparable to earlier data is the "equal proportion or equal fractions" method. The principle of "equal proportion" involves weighting the multi-race responses on the assumption that they are equal shares of each race. For example, the category of "Alaska Native and White" would be weighted 0.5 Alaska Native and 0.5 White. After all the multi-races are proportionately weighted, the race fractions are summed and rounded to the nearest whole person to obtain the estimated number of persons equivalent to the single race responses of earlier censuses. This procedure has been used at the "place" level (small geographic areas) so that the aggregate figures for census areas or boroughs, or the state as a whole, will be consistent with the place-specific figures.

The equal proportion method is used in the racial breakdown for the "Population" section of the Health Area Profiles. Equal proportion figures are not yet available for the United States.
[back to top]

Cause of Death Coding

Causes of death are classified by the Tenth Revision International Classification of Disease (ICD-10). In 1999 the Tenth Revision replaced the Ninth Revision (ICD-9) that had been in use for the previous 20 years. The change from ICD-9 to ICD-10 results in discontinuities between selected causes of death by introducing new causes of death titles and their corresponding cause of death codes.

Caution is necessary in comparing mortality rates before 1999 with current rates. Nationally, only 7 of the 15 leading causes of death titles using ICD-9 remain the same under ICD-10 coding. The break in comparability results from changes in category titles, changes in structure and content of the classification, and from changes in the coding rules used to select the underlying cause of death. Mortality statistics are generally based on underlying cause of death, which is defined as "the disease or injury which initiated the train of morbid events leading directly to death, or the circumstances of the accident or violence which produced the fatal injury." The process for identifying the leading causes of death is published in the National Center for Health Statistics Instruction Manual, Part 9, ICD-10 Cause of Death Lists for Tabulating Mortality Statistics, Effective 1999.

Ratios of comparability between ICD-9 and ICD-10 have been studied by the dual classification of mortality records at the national level. A comparability ratio of 1.00 indicates that the same number of deaths was assigned to a particular cause or combination of causes whether ICD-9 or ICD-10 was used. This does not necessarily indicate that the cause was unaffected by changes in classification and coding procedures but merely that there was no net change in the number assigned. A ratio less than 1.00 indicates that fewer deaths are assigned to the cause of death under ICD-10 than with ICD-9. A ratio greater than 1.00 results from an increase in deaths assigned to a cause in ICD-10 compared with the comparable ICD-9 cause. Technical reports for the National Vital Statistics system are available throught the National Center for Health Statisitics. The National Vital Statistics Report Volume 49, No. 2, May 18, 2001 (Comparability of Cause of Death Between ICD-9 and ICD-10: Preliminary Estimates) provided an initial estimate of comparability.

The change to ICD-10 increases the number of deaths attributed to Alzheimer's disease, septicemia, unintentional injuries, Sudden Infant Death Syndrome (SIDS), and several other common causes of death. The number of deaths attributed to heart disease, asthma, pneumonia, and congenital anomalies decreases under ICD-10.

The Profiles give death rates for 1990-1998 with the ICD-9 codes. Rates for 1999 and later are presented separately with the ICD-10 codes. In the future, mortality rates for the decade may be adjusted for the change in coding by using comparability ratios.

See also: International Classification of Diseases Codes Used in Health Area Profiles
[back to top]

Measuring Prenatal Care:
The Adequacy of Prenatal Care Utilization (APNCU) Index

The APNCU index, also called the Kotelchuck Index, uses two crucial elements obtained from birth certificate data-when prenatal care began (initiation) and the number of prenatal visits from when prenatal care began until delivery (received services). The APNCU index classifies the adequacy of initiation as follows: pregnancy months 1 and 2, months 3 and 4, months 5 and 6, and months 7 to 9 (Figure 17). To classify the adequacy of received services, the number of prenatal visits is compared to the expected number of visits for the period between when care began and the delivery date. The expected number of visits is based on the American College of Obstetricians and Gynecologists prenatal care standards for uncomplicated pregnancies and is adjusted for the gestational when care began and for the gestational age at delivery.

A ratio of observed to expected visits is calculated and grouped into four categories-Inadequate (received less than 50% of expected visits), Intermediate (50%-79%), Adequate (80%-109%), and Adequate Plus (110% or more). The final APNCU index measure combines these two dimensions into a single summary score. The profiles define adequate prenatal care as a score of 80% or greater on the APNCU Index, or the sum of the Adequate and Adequate Plus categories.

The APNCU Index does not measure the quality of prenatal care. It also depends on the accuracy of the patient or health care provider's recall of the timing of the first visit and the number of subsequent visits. The APNCU Index uses recommendations for low-risk pregnancies, and may not measure the adequacy of care for high-risk women. The APNCU Index is preferable to other indices because it includes a category for women who receive more than the recommended amount of care (adequate plus, or intensive utilization).
[back to top]

Why are the numbers different?

National health statistics for Alaska, such as those from the National Center for Health Statistics available through the CDC, usually differ from statistics presented by state agencies such as the Division of Public Health. National statistics use U.S. Census data as the denominator, while the Alaska Bureau of Vital Statistics uses the Department of Labor population estimates. See "General information about population estimates and projections" at the Department of Labor and Workforce Development website.

Birth certificates and death certificates that are filed late, referred from other states, or amended are included in Alaska statistics but may not be reflected in national data. Rounding of numbers may cause minor discrepancies.

Definitions of events vary between state and federal agencies. For example, states report the number of substantiated reports of child abuse and neglect to the National Child Abuse and Neglect Data Systems (NCANDS). The definition of "substantiated", however, varies from state to state and the results should be compared with caution.

Many discrepancies can be traced to the use of different standard populations for age adjustment, the difference between crude and age-adjusted rates, and changes in coding and definitions such as the International Classification of Diseases system.

Division of Public Health publications may include figures that differ from those used in earlier publications. Updates to Health Status in Alaska: 2000 Edition, for example, will include revised mortality rates for the leading causes of death 1990-1998. The same data will be age adjusted to the U.S. 2000 population instead of the U.S. 1940 population. This adjustment will increase most death rates. Rates may also be recalculated to make them consistent with coding changes made in 1999.
[back to top]