Mortality Data Documentation
Sources of Data
DeathsThe death certificate is the source document for data included in the mortality query. New Jersey law requires the prompt filing of a death certificate by the proper authority, such as hospital personnel, physicians, medical examiners, and funeral directors, in the event of a death occurring in the state. These certificates are submitted to the office of the State Registrar, where they are recorded and filed permanently. Statistics on deaths of New Jersey residents which occurred in other states are obtained through participation in the national Vital Statistics Cooperative Program (VSCP), which encourages the exchange of information on vital events between the states of occurrence and residence. The mortality data presented in the query system are for New Jersey residents, regardless of where the death occurred.
The mortality data in the query system were generated from data files available at the time of preparation of the back-end dataset. Any data pertaining to a death for which a certificate was filed after that time or relating to corrections or revisions made since the data were processed for the electronic file are not included. Vital events computer files are periodically updated by the New Jersey Department of Health's Bureau of Vital Statistics and Registration (BVS) and Center for Health Statistics (CHS) staff based on correction reports received from local registrars and from data quality control analyses conducted by CHS. The query incorporates data from the most recently updated files.
All of the causes of deaths included in the mortality query are underlying causes, and were coded by the National Center for Health Statistics' SuperMICAR and ACME software in accordance with the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10), adapted for use in the United States. Additional causes of death listed on the certificates, including the immediate and intermediate causes, are not included in the mortality query. The inclusion of all listed causes of death (multiple causes of death) could lead to somewhat different results.
PopulationPopulation estimates used to calculate mortality rates are derived from the most recent vintage of the mid-year estimates provided by the Population Division of the U.S. Bureau of the Census. These estimates are based on the 2000 Census with 31 race categories aggregated to the following race groups: White alone, Black alone, American Indian/ Alaska Native alone, Asian alone, Native Hawaiian/Pacific Islander alone, and Two or more races. Estimates are available for each state and its counties by age, race, Hispanic ethnicity, and gender categories. Race and ethnicity can be combined to form seven race/ethnicity groups, where Hispanics may be of any race, the other six groups do not include Hispanics, and data for White, Black, Asian, Native Hawaiian/Pacific Islander, and American Indian/Alaska Native are counts of persons who only selected one of those races (i.e., "alone").
There are separate queries for state, county, and municipality rates because three different sets of population data are used as denominators in the rate calculations. Population estimates for the state are available by single year of age while estimates for counties are only available in 5-year age groups. Municipality population estimates by age, race/ethnicity, and sex are only available in Census years, therefore age-specific and age-adjusted rates cannot be computed for municipality data. Municipality estimates are provided by the New Jersey Department of Labor and Workforce Development's State Data Center.
Quality of DataThe reporting of deaths is considered to be essentially complete. According to the National Center for Health Statistics (NCHS), more than 99% of deaths are registered. The completeness of reporting by residence is dependent on the effective functioning of the interstate data exchange program for certificates which is fostered and encouraged by NCHS. Research has shown that there is some degree of slippage in receiving information on all deaths of New Jersey residents occurring in other states. However, the number of missing events is thought to be small, relative to the overall number of deaths.
The quality of the data included in the query is a function of the accuracy and completeness of the information recorded on the respective certificates and of the quality control procedures employed in the coding and keying processes. A query program in which the individual(s) responsible for completing the certificate is questioned about missing or conflicting information is carried out by BVS staff. This process is augmented by the data quality control analyses performed by the CHS using all of the NCHS edit criteria.
In order to participate in the national VSCP, states had to achieve an error rate of 2% or less on each certificate item for three consecutive months. The error rates relate to both coding and data entry errors. New Jersey has met the error tolerance requirements for the cooperative program. After satisfying initial requirements, a monthly sample of records is used to determine that the error rate on each death certificate item is approximately 2% for each death certificate item other than the medical cause of death information. Due to the complexity of the coding system, cause of death coding has a 5% error tolerance level set by NCHS.
Changes to the Death Certificate in 2004A new death certificate format was introduced in 2004. Major changes to the certificate include the ability to choose more than one race and Hispanic ethnicity for the decedent, additional place of death and disposition categories, expansion of the pregnancy checkbox items, and the addition of checkboxes for whether the decedent had diabetes and if smoking contributed to death.
Two or more races were selected on only 0.3% of certificates in 2004, so the affect on trend data by race is negligible. The addition of hospice facility as a place of death (3.0%) and donation and removal from state as dispositions (2.5%), on the other hand, had a small but noticeable affect on trends as those categories were previously collapsed into others. Responses to the pregnancy checkbox question affect the underlying cause of death; therefore the expansion of the checkbox items to include deaths up to one year after the end of a pregnancy resulted in a sharp increase in maternal deaths.
Allocation of Data by Residence or OccurrenceFor public health planning and policy determination, the most useful population to study is usually the resident population of an area. In the case of deaths, the existence of resident certificate exchange agreements among the registration areas in the country permits analysis of resident death statistics regardless of where the death occurred. In the query system, the data presented represent deaths to New Jersey residents, regardless of where they occurred. Deaths that occurred in New Jersey to non-NJ residents are not included.
Allocation of deaths by place of residence within the state is sometimes difficult because classification depends on the statement of the usual place of residence provided by the informant at the time the certificate is completed. For a variety of reasons, the information given may be incorrectly recorded. A common source of error is the confusion of mailing address with residence address. Consequently, residence assignment for about 40% of New Jersey municipalites is inaccurate (defined as greater than a 10% over- or undercount). Beginning with the 2004 data year, all death records are run through geocoding software to properly allocate the county and municipality of residence. Prior to 2004, full address information was not available on the electronic death file, therefore geocoding correction was not possible. Subsequently, municipality-level death data are not available for pre-2004 data in the mortality query.
Race, Ethnicity, and NativityPrior to 2004, one race group (White, Black, American Indian/Alaska Native, Chinese, Japanese, Hawaiian, Filipino, Asian Indian, Korean, Samoan, Vietnamese, Guamian, other Asian/Pacific Islander, other race, and an unknown race category) and an ethnicity (Non-Hispanic, Mexican, Puerto Rican, Cuban, Central or South American, other Hispanic, and an unknown ethnicity category) were recorded for each individual for whom a death certificate was filed. Beginning in 2004, two or more races could be selected on death certificates. One or more of the following may be used to describe the race of the decendent: White, Black, American Indian/Alaska Native, Chinese, Japanese, Filipino, Asian Indian, Korean, Vietnamese, other Asian, Native Hawaiian, Samoan, Guamian, other Pacific Islander, other race, and an unknown race category. An ethnicity (Non-Hispanic, Mexican, Puerto Rican, Cuban, Central or South American, other Hispanic, and an unknown ethnicity category) is also recorded for each individual for whom a death certificate is filed. Race and ethnicity can be combined to make a Hispanic "race" group and this is the standard way the New Jersey Department of Health reports death data.
Due to the limitations of the corresponding population data, race/ethnicity designations used in the death query are White, Black, Hispanic, Asian, Native Hawaiian/Other Pacific Islander, American Indian/Alaska Native, Other Single Race, and Two or More Races (for data after 2003), where Hispanics may be of any race and the other race/ethnicity groups do not include Hispanics (but include those with ethnicity not stated). The Hispanic category includes persons of Mexican, Puerto Rican, Cuban, Central/South American, or other Hispanic ethnicity, regardless of race. The Asian category includes persons of Chinese, Japanese, Filipino, Asian Indian, Korean, Vietnamese, and other Asian descent. The Native Hawaiian/Other Pacific Islander category includes persons of Hawaiian, Samoan, Guamian, and other Pacific Islander descent. The Other Single Race category includes persons of any race other than White, Black, Asian, Native Hawaiian/Pacific Islander, and American Indian/Alaska Native. Two or More Races includes anyone for whom more than one race was recorded. When combining race and ethnicity, Hispanic ethnicity takes precedence over whatever race(s) are recorded for the decedent. Other and unknown race as well as unknown age were imputed in the 2000-2003 data files used for the back-end of the mortality query, therefore the query will produce no Other race, unknown race, or unknown age records for pre-2004 data.
Three nativity categories are used in the query: native-born, Puerto Rico-born, and foreign-born. Native-born includes those born in the 50 states or Washington, D.C. Foreign-born includes those born anywhere other than the 50 states, D.C., and Puerto Rico. Foreign-born, therefore, includes other U.S. territories such as American Samoa, Guam, and U.S. Virgin Islands because all territories are not coded separately in the electronic death file. Foreign-born also includes those born abroad to American parents because death certificates do not have an item to distinguish those cases. Numbers of deaths to New Jersey residents born in territories other than Puerto Rico or born abroad to American parents are extremely small.
Cause of Death RankingsRankable causes of death in the death query are based on 50 distinct causes of death derived from the NCHS List of 113 Selected Causes of Death.
New Jersey also uses a list of 31 cause groups and two residual categories modified from the 50 Cause List that groups causes of death that rarely occur among New Jersey residents. This list may also be used for ranking causes of death with the caveat that Other and Unspecified Infectious or Parasitic Diseases; Symptoms, Signs and Abnormal Clinical and Laboratory Findings, NEC; All Other Diseases (Residual); and All Other External Causes are not eligible to be ranked.
Leading cause of death ranking should be done based on numbers of deaths not rates since ranking by age-adjusted rates may produce different results.
RatesThe presentation of death rates facilitates comparisons between political subdivisions with populations of different sizes or between subgroups of a population. Crude death rates are calculated by dividing the number of deaths of the residents of an area or demographic subgroup by the resident population of that area or subgroup and are usually expression per 100,000 residents. Deaths are limited to those that occur within a specific time period, usually a year, and the population is, in general, the mid-year estimate of the resident population of the area.
In order to compare death experiences among various ages and races/ethnicities or between the sexes, death rates may be computed for demographic subgroups of the population. These are referred to as age-, race/ethnicity-, or sex-specific rates and are calculated by dividing the number of deaths within a subgroup by the population in the subgroup. Death rates from specific causes may also be calculated, with the numerator consisting of the deaths from the particular cause in an area and the denominator comprised of the population at risk of the disease or condition. For example, the population at risk for prostate cancer is males only, so the denominator would consist of males, not the entire population.
The numbers of deaths in an area are directly related to the demographic characteristics of the area's population. In comparing rates over time or among geographic areas, it is helpful to eliminate the effects of the differences in the populations' demographic characteristics on the comparison. This can be accomplished through adjustments of the rates for the particular characteristics of interest. The most common type of adjustment of rates is for age. Direct adjustment of death rates involves application of existing rates (age-, race/ethnicity-, or sex-specific) to a standard population to arrive at the theoretical number of events that would occur in the standard population, at the rates prevailing in the actual population. These events are then divided by the total number of persons in the standard population to arrive at an adjusted rate. Adjusted rates are index numbers and cannot be compared to crude or other actual rates. The use of adjusted rates is limited to comparison with other adjusted rates, based on the same standard population. The standard population used in the death query is the United States 2000 standard million, derived from the projection of counts from the 2000 decennial census. Mortality data prior to 1999 used the US 1940 standard million for age-adjustment, although the 2000 standard may be used for pre-1999 data. Use caution when comparing age-adjusted rates to make sure they are using the same standard.
The definition of rates used in the query system are in the Definitions of Public Health Terms and Acronyms. It should be noted that alternative forms exist for some of these statistics. Some other states and the federal government may employ different formulae for the computation of selected rates.
The lack of county-level estimates by single year of age means that age-adjustment must be done using 10 age groups (Under 5, 5-14, 15-24, ..., 75-84, 85+) rather than the standard 11 age groups (Under 1, 1-4, 5-14, 15-24, ..., 75-84, 85+) used to calculate age-adjusted rates at the state level. This may result in minor differences, usually no more than ± 0.1, when age-adjusted rates are produced for the state (i.e., Total) using the county query page instead of the state query page.
Small NumbersCaution should be exercised in the interpretation of rates based on small numbers. Chance variations in the number of deaths occurring in sparsely populated areas can cause rates to fluctuate widely over time. In accordance with NCHS standards, rates based on fewer than 20 deaths or fewer than 20 persons in the population are considered unreliable for analysis purposes. Therefore, these rates are not displayed and are indicated by ** in the appropriate cell. For purposes of analyzing mortality rates for small areas, calculation of three- or five-year average rates and other statistical methodologies for analyzing small numbers may provide more meaningful measures.