As a one-time librarian for an environmental engineering firm, this reviewer first encountered the CDC’s somewhat macabre trove of mortality data while trying to ascertain the impact of elevated levels of pollutants on regional health statistics. The CDC’s WONDER databases homepage (http://wonder.cdc.gov/) offers a variety of resources, organized across three tabs: WONDER Systems, Topics and an A-Z Index. However, for the purposes of this review, discussion is limited to the website’s capabilities for downloading mortality data tailored to the needs of the individual user.
From the WONDER databases homepage, users may find links for downloading mortality data under the headings mortality and underlying cause of death. The WONDER databases offer the user considerable latitude in the specificity of the data downloaded, ranging from the very general to the highly detailed. This is initially reflected in the basic options offered on the homepage, which include three options for accessing data for all age groups and causes of mortality – detailed mortality; compressed mortality; multiple cause of death – and a further two options for accessing data of a narrower scope – infant deaths; tuberculosis information.
While similar in many ways, the options in the first group differ in terms of certain capabilities. For example, unlike compressed mortality, detailed mortality allows the user to focus on age ranges as short as one year, specific years and months or even particular days of the week. Multiple cause of death, as its name suggests, permits users to identify the extent to which combinations of two or more specified medical conditions coexisted in study subjects.
There are many ways in which users can define their data requirements. In the detailed mortality option, tabular layout allows grouping of data using up to five different parameters in the categories of location; demographics; year and month; and weekday, autopsy, place of death. Location can take the form of political jurisdictions, such as states and counties, Census Bureau-defined areas and US Department of Health and Human Services (HHS) regions. Parameters for grouping by demographics include age groups, gender, Hispanic origin and race. For the more statistically minded among us, output may include 95 per cent confidence interval and standard error. In addition, because the prevalence of various causes of mortality often varies by age group, data may be age-adjusted to allow more accurate comparisons between groups whose populations have differing age distributions.
As implied by the location options for grouping the tabularized data (described above), users must choose to download data corresponding to either a political jurisdictions (States), a Census Region or an HHS Region. Once a selection has been made in one of these categories, a smaller area (or areas) within that selection can be chosen. For example, a user could choose to specify location in terms of States and then select the state of Florida; a plus (+) sign to the left of the state name indicates that a smaller geographic entity may be specified. Clicking on the open icon would reveal a list of counties within Florida from which the user may select one or more. Upon selecting the desired geographic entity (entities), the user may opt to obtain all data or limit data to areas falling within urbanization categories ranging from NonCore (non-metro) to Large Central Metro.
The Demographics parameter allows users to aggregate data in age ranges of a single year, five years or ten years, as well as infant age groups. gender, Hispanic origin and race are further options for defining data requirements. Year and month permits the user to specify the years for which data should be retrieved. Once a selection has been made for year(s), clicking on the open icon will allow the user to specify months of interest. For weekday, autopsy and place of death, the user has the prerogative of collecting data for specified days of the week. Perhaps, this reflects the observation that certain causes of death are more likely during one time of the week than another. For example, death by violent crime is generally more likely during weekends, whereas heart attacks occur disproportionately on Mondays. The user may also limit data to cases when an autopsy was or was not performed, as well as by place of death (e.g. various types of medical facilities, home, hospices, etc.). Within this section, the user may define the cause(s) of death of interest using the codes for the International Statistical Classification of Diseases and Related Health Problems, which are abbreviated as ICD Codes.
The user may opt to export results to a file, show or not show totals, zero values and suppressed values and indicate the number of decimal places to which results should be shown. In many cases, the options available to the user for downloading compressed mortality and multiple cause of death data are similar. Given their more specialized nature, the infant deaths and tuberculosis data sets have some different defining characteristics. The infant deaths data introduces such considerations as the mother’s educational level and marital status, the birth weight of the child, the child’s gestational age, age at death, the extent to which prenatal care was provided, delivery method and more. The tuberculosis data examines such factors as the deceased’s occupation, whether or not they were born in the USA, where they were residing at the time of their diagnosis, their use of alcohol and drugs, HIV status and type of therapy, among others.
Each of the five data sets has five associated tabs, as follows: request form; results; map; chart; and about. Upon selecting a data set on the homepage, the user is delivered to the last of these, which describes the data set’s contents and limitations on its permitted use. After clicking on the I Agree icon, the user is rerouted to the request form tab, where they specify their desired data (as described above). Once the data request is submitted and fulfilled, the results are shown on the results tab. The remaining two tabs allow the user to view the data in the form of a map or chart.
The WONDER databases contain a remarkable wealth of mortality-related data. In terms of the specificity with which users may define the data in which they are interested, it is certainly structured to meet the needs of researchers with an extensive background in this area of study. This may prove daunting to newer or less experienced users. It is worth noting, however, that although users can go into great detail when defining their requirements, this is an option rather than a requirement.
Users only need to define their data requirements in sufficient detail to meet their particular needs. When indicating locations, time periods and causes of death, the user is, by default, placed in a Browse mode, in which they scroll through their possible selections in a hierarchical format. However, in the event that the user is unfamiliar with the structure of, for example, the International Statistical Classification of Diseases and Related Health Problems (ICD Codes), they may opt instead for a Search mode that allows them to search for possible selections using keywords. In addition, the request form tab is liberally sprinkled with links to finder tool help and, once a selection is made, clicking on a details tab will access a more complete description of the selection.
