WISH Technical Notes and Definitions for Historical Data (pre-2016): Injury Emergency Department Visits and Hospitalization Data
Technical notes
Hospitalizations
The Hospitalization Module provides data on injury-related hospitalizations of Wisconsin residents from 1995 through 2015 based on the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). The ICD-9-CM is used by health care organizations to code for injuries and illnesses. The coding system changed in October 2015 from the 9th Revision (ICD-9-CM) to the 10th Revision (ICD-10-CM). The 9th and 10th revisions are substantially different from each other. Fourth quarter data for 2015, when the transition to the new coding system occurred, are not included in any query module. Because of this, the annual counts and rates for 2015 are estimates based on data from October 2014 through September 2015.
This module includes inpatient stays and not individual patients. This means that patients transferred between hospitals, readmitted to a hospital for further treatment, or otherwise treated in more than one inpatient stay are counted more than once.
The 1995 through September 2015 Injury Hospitalization data subset is limited to stays with an ICD-9-CM principal diagnosis code related to an injury. This means that if a patient was admitted for multiple reasons but the primary reason for admission was NOT an injury, the hospitalization stay will not be included in this dataset. For self-harm hospitalizations, this is particularly important as these stays may often have a principal diagnosis related to mental health and NOT injury. In those situations, these visits would not be included in the injury data subset.
The following ICD-9-CM principal diagnosis codes were included in the hospitalization subset: 800-909.2, 909.4, 909.9-994.9, 995.5-995.59, 995.80-995.85.
Nonfatal query option for hospitalizations
The option to select only hospital stays for individuals who were admitted and subsequently released or transferred to alternative care is available for all Hospitalization queries. We recommend the “Non-fatal hospitalizations ONLY” filter (the default selection) because it prevents counting a hospital stay that resulted in death as both a hospitalization and a death, which may appear to artificially inflate the effects of injury. Selecting “ALL stays” will show the total volume of stays that hospitals experience, including those stays that result in death.
Emergency department (ED) visits
The ED Visits Module provides data on the injury-related ED visits of Wisconsin residents from 2002 through 2015 based on the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). The ICD-9-CM is used by health care organizations to code for injuries and illnesses. The coding system changed in October 2015 from the 9th Revision (ICD-9-CM) to the 10th Revision (ICD-10-CM). The 9th and 10th revisions are substantially different from each other. Fourth quarter data for 2015, when the transition to the new coding system occurred, are not included in any query module. Because of this, the annual counts and rates for 2015 are estimates based on data from October 2014 through September 2015.
This module includes ALL injury-related ED visits: where the patient was treated and released, was admitted as an inpatient in the same acute-care hospital, was transferred to another hospital, or died while in care.
The 2002 through 2015 Injury ED data subset is limited to visits with an ICD-9-CM principal diagnosis code related to an injury or an external-cause code (e-code; described below) related to injury.
The same ICD-9-CM principal diagnosis codes included in the hospitalization subset are included in the ED subset (see Hospital Stays above). In addition, the following e-codes are included: E800-E869, E880-E929, E950-E999.
"Treated and released" query option for ED visits
The option to select only ED visits for individuals who were treated in the ED and subsequently released from care is available for all ED queries. We recommend the "Treated and Released ONLY" filter (the default selection) because it prevents counting a health care visit more than once. Patients transferred to hospitals for inpatient stays or those who died while in the ED are not included. This keeps the effects of injury from appearing artificially high. Selecting “ALL visits” will show the total volume of visits that EDs experience due to injury, including those visits that resulted in a transfer to inpatient hospitals or death in the ED.
Numbers and rates for specific injury types
Numbers and rates for specific injury types are based on diagnosis code and/or e-codes. The data subset of interest, either hospitalization or ED (as described above), is queried to obtain these numbers.
Rank vs specific injury counts and rates
Rank: If the Ranking Measure is selected (Step 2), an ED visit or hospital stay will only count as one type of injury based on the following groupings:
- Drowning
- Cutting/piercing objects
- Falls
- Firearm
- Fire-heat-chemical burns-hot object/scalding
- Machinery
- Motor vehicle
- Natural/environmental factors
- Non-traffic transportation injury
- Overexertion
- Poisoning
- Struck by or against object or person
- Suffocation
The external causes of injury codes are reviewed for the classification into one of these 13 types of injuries. As classification depends on the existence of an e-code, those without an e-code will not be counted in this ranking measure. Each year, there are visits and hospitalizations that are included in the data subsets because of an injury diagnosis code but they do not have an external cause of injury code (e-code). These visits do not get classified in one of the 13 major injury categories above and are not reported in the ranking measure. Additionally, visits classified as "other specified classifiable cause of injury," "other specified cause of injury not elsewhere classifiable," or "unspecified cause of injury" are not included in this ranking option.
Visits missing an e-code or classified as "other specified" or "unspecified" are included in the "ALL Injuries" option in Step 3 of the query module.
For a breakdown of ICD-9-CM codes included in the above 13 categories, please review the U.S. Centers for Disease Control and Prevention (CDC) Matrix of E-code Groupings. Note: the grouping of fire/burn, fire/flame, and hot object/substance are combined in the WISH ranking measure. All nontraffic transportation groupings are also combined in WISH ranking (pedal cyclist, other; pedestrian, other; transport, other).
Specific injuries: Specific injury categories listed in Step 3 of the injury queries do NOT include all possible injury types or causes. If you are interested in a total count or rate of all injuries, you can select the first option in Step 3: "ALL injuries." For specific injury types and causes, review the rest of the list and select the option that suits your needs. The program will limit how many you can select for any one query because the options are NOT mutually exclusive and a visit may be included in one or more of the injury types listed. For example, a motor vehicle injury may also be included as a traumatic brain injury.
Based on ICD-9-CM coding, the first valid e-code is reviewed and determines how the ED visit or hospitalization will be classified. How this impacts the counts can be seen in the following example: If an ED visit has a first valid e-code of self-inflicted poisoning (ICD-9-CM E950, for example) and a second valid e-code of a fall, this visit will only be counted as a poisoning and not as a fall. In addition, the first valid e-code also indicated the manner (self-inflicted), so it will be included as a “self-harm injury” as well.
The ICD-9-CM codes included for each injury cause and manner in Step 3 are detailed in the CDC Matrix of E-code Groupings with a few caveats: 1) unintentional fire-related injury is based only on the “fire/flame” grouping; 2) drowning injuries include the e-codes in the matrix and visits with a diagnosis code of 994.1 (drowning and nonfatal submersion); 3) hip fractures include visits with a diagnosis code of 820 (fracture of neck of femur); and 4) traumatic brain injuries (TBI) include a list of diagnosis codes.
Definitions
External cause of injury codes (e-codes)
The external underlying cause of injury is the way in which the person sustained the injury; how the person was injured; or the process by which the injury occurred. External causes for injuries are coded from the medical record by hospital staff according to the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM).
Rates
Injury hospitalization and ED visit rates
The injury hospitalization and ED rates are calculated by dividing the number of injury-related visits per year by the population. It is usually expressed as the number of injuries per 100,000 residents. The rate may refer to a specific group, or to injury visits due to a specific cause, or to all injury visits in the entire population. The rate may be adjusted for the age composition of the group (age-adjusted) or it may be the observed rate (unadjusted).
Unadjusted rates
This is the observed rate and is displayed as the number of visits per 100,000 residents. We calculate this rate by dividing the number of injury-related visits by the total population of interest and then multiply by 100,000. We recommend that you use this rate when describing injury among your population of interest (for instance, the rate of unintentional falls in Rock County). This rate should not be used to compare different populations or time periods to each other as we have not adjusted these rates to account for differences in age composition across populations or time periods.
Age-Specific Rates
If you want to compare rates across groups or populations for a specific age group (for example, children 1-4 years old by race), comparing observed age-specific rates may be your best option. If this is what you want to do, select Unadjusted Rate in Step 2 and also select your one age group of interest from Age Group 1 (Step 6). An age-specific rate is calculated by dividing the total number of events for the specific age group of interest by the total population of that age group. This calculation is the same as an observed rate for the age group selected.
Alternatively, if you want to combine multiple age groups (from either Age Group option in Step 6) and compare across populations, geography, or time, we recommend selecting age-adjusted rather than age-specific rates, as age distributions can vary greatly.
Age-adjusted rates
This is a standardized rate. We recommend that you select age-adjusted rates when making comparisons between two populations (for example, Southern Region injury rates compared to Northern Region injury rates), or two different time periods (for example, 2010 injury rates compared to 2015). Age-adjustment accounts for differences in age composition across populations and time. We use the Standard U.S. Population for Year 2000 for age-adjustment.
We also recommend selecting age-adjusted rates if you want to compare across geography or time for a wide age range (a range larger than a single age category listed in the age groups in Step 6). For instance, if you want to compare the rate of injury for women of child-bearing age (15-44) over time, select age-adjusted rate and in Step 6 (Patient Characteristics), and select the multiple age categories that combine to match the 15-44 range of interest.
The age groups used to create the age-adjusted rates are those in Age Group 1 of Step 6.
Race and ethnicity
Race and ethnicity data are available for all years for hospitalization stays. For ED visits, these data are available only for 2013 and after due to a significant proportion of missing race and ethnicity information in earlier years.
The population estimates used as denominators for the injury hospitalization rates in WISH are based on the bridged race estimates provided by the U.S. Census Bureau and the National Center for Health Statistics (NCHS). The estimates have been controlled so they sum to the annual estimates published by the Office of Health Informatics.
Producing the bridged race estimates was necessary because race categories in Census 2000 differed from those used in previous years. Specifically, data on race from Census 2000 were not directly comparable to data from previous years due, in large part, to giving respondents the option to report more than one race.
As a result, NCHS and the Census Bureau produced bridged race estimates that allow calculation of rates by race and ethnicity across years. These estimates distribute (or "bridge") the "more than one race" and "some other race" populations into one of four major race groups (American Indian/Alaska Native, Asian/Pacific Islander, Black, and White) and two ethnicity groups (Hispanic/Latino, non-Hispanic/Latino).
NCHS and the Census Bureau have produced this set of bridged race estimates extending back to the 1990 Census.