•Introduce self.
•Link this talk to the two previous data talks…
…We’ve discussed the importance of data to an IP Program…
…We’ve identified sources of community-based injury data…and…
…We’ve discussed planning considerations for collecting data.
•In this session, we’ll address the issue…Once you have data, what do you do with it?
•Session will include discussion & a data analysis exercise
Instructor Note: This lecture is estimated at 45 minutes plus a 30 minute exercise.
Instructor Note: The exercise requires students to make calculations with a calculator. Coordinate with other instructors to ensure 10-12 calculators will be available for student use.
Self explanatory.
•We associate data analysis with a number of epidemiology terms and concepts, including this list.
•First four concepts/terms are review
•Risk, although not a new term, previous courses/sessions have not discussed “risk” in detail.
•Following slides will discuss each
In review…
•Read definition of Epidemiology.
•In simple terms, we use epidemiology in IP to help a community better describe trends and patterns of injury in their community.
•Specifically, we use epidemiology and injury data to determine the who, where, what, when, why, and how of injury.
Another basic concept in epidemiology and injury prevention is…read slide.
Trigger discussion:
•Headline of shark attacks compared to graph of leading causes of unintentional injury death.
•In fact, there were only 4 shark-attack fatalities during this time; compared to over 25,000 MVCs, 12,000 poisonings, 10,000 falls, and almost 3,000 each of drowning and suffocations.
•Key point: if Florida based all of its IP efforts on the “problem” of shark attacks, they would be fail to address the leading causes of injury in their state.
•Key point: Emotion, the media, and politics can and will influence an IP Program; but utilize data to keep the leading causes of injury for your community in perspective.
•Can you think of other “headlines” that might not be representative of the community’s injury problem?
Hint: school violence, gun violence, gang violence.
Another basic concept in review is…read slide.
Example Epi-curve slide:
•Graph shows the distribution of crashes along a 10 mile section of HWY70 on the San Carlos Apache Reservation.
•Horizontal axis represents the milepost location of each crash and the vertical axis represents the number of crashes. This graph illustrates that crashes along this stretch of highway are not randomly distributed but instead there is a particular pattern that identifies a high incidence at one location, MP 272.5.
•Our role as injury prevention practitioners is to identify the etiologic (causative) factors for crashes at this location and determine what interventions are warranted to reduce the incidence of crashes.
Ask students: Any thoughts as to why one section of highway might have a dramatically difference incidence of crashes than an other along that same highway (and relative same traffic volume)? In this case, the major contributing factors were:
•Influx of homes/population/traffic at MP272.5 over a period of time
•Road design did not keep up with nearby development and population growth
•Slow moving local traffic in conflict with fast moving through traffic
•Need for turning lanes
•Vision obstructions (overgrowth and embankments)
More review…
•A primary purpose of injury data is to describe the magnitude of the injury problem for a given population.
•The two general categories of injury data, based on the severity of the injury, are referred to as Mortality and Morbidity.
Mortality Data
•Mortality data refer to the those injuries resulting in death.
•Typically, mortality data are easier to obtain because death records (death certificates) are maintained in every state and statistics are aggregated at both state and local levels. Late in the course we’ll take a look at one national injury mortality database: CDC’s WISQARS.
•While mortality data provide a good indication of the magnitude of the public health problem of injuries, they are only the tip of the iceberg – representing only a fraction of injuries in a given population.
Morbidity Data
•Morbidity data refer to those non-fatal injuries resulting in hospitalization, emergency department/ambulatory care, and etc.
•There are fewer systematic state and national data collection systems for injury morbidity data, and therefore these data are more difficult to obtain.
•Since the leading causes nonfatal and fatal injuries are often very different, IP programs are
encouraged to collect morbidity data in order to have a more complete description of the magnitude of the injury problem.
(Source: Injury Prevention: Meeting the Challenge, pg. 36-37)
Lastly…in review…
•The text, A Dictionary of Epidemiology, provides this basic definition for risk: Read Slide.
•AKA: the odds, chance, or likelihood an event – such as an injury – might occur.
•For example, we can all agree than if you drink and drive your odds of crashing are greater…and if you do not wear a seat belt when in that crash it is likely you will suffer an injury.
Related to our common use of the term risk is Risk Factor. There are several definitions for risk factor, including: Read Slide.
Discussion
•(Definition 1) Factors such as age, sex, race, and family history, which are not subject to change, are often major determinants of risk. This is particularly true of chronic diseases such as heart disease and diabetes.
•(Definition 2) Some risk factors can be altered.
(Brief class discussion.)
•What injury related risk factors can be altered to reduce injury? Hints: seat belt use, no drinking and driving, personal protective equipment, etc.
What’s the greater risk, driving a rural road or a busy city freeway?
It depends…
•Exposure must be considered when assessing the risk of an event occurring. Variables used to describe exposure include: the population exposed (i.e, the number of people in your community, tribe, or state) and the volume of traffic on a given roadway.
•Environmental factors (lighting, guard rails, shoulders, law enforcement presence, access to healthcare, etc)
For example (1) when comparing injury risks between populations, we must consider not only the number of injuries that occur in each community but the number of people or population size in each community.
For example (2) when comparing injury risks between two roadways, we must consider not only the number of crashes that occur on each road but how busy (the traffic volume) each road is.
The number of events or injuries relative to exposure help us determine injury rates. Rates are direct indicators of risk. The higher an injury rate…the higher the injury risk.
We’ll discuss rates in more detail later in this session.
•There are numerous methods to analyze data.
•The level of analysis will depend on what you are trying to determine from the data (e.g., descriptive study, grants, evaluation) and your technical abilities to analyze the data.
•While it is not the intent of this course to create a room full of statisticians, there are some basic statistics that we can utilize to better describe and understand injury data.
•The basic statistics we’ll review are: Read Slide.
Simply presenting the numeric value of a data variable is one data analysis method. This method is very common and the information is easy to understand.
For example….(Consider use of flip chart)…
•One community reported: There were 60 severe injuries (10 fatalities and 50 hospitalizations) due to assault related injuries during 1991-1993.
•Certainly, this information is important to the community. At the very least, it tells you that 10
members of the community were homicide victims and 50 suffered an injury requiring hospitalization.
With some simple arithmetic, you can use this information to report approximately 3 homicides per year occur in the community and for every 1 homicide there are 5 assaults resulting in hospitalization.
•Unfortunately, because of variations in exposure (e.g. population size) the numeric value of one variable cannot be compared to that of the same variable in a different population.
•For example…The occurrence of 10 homicides in a small community is public health problem of a completely different magnitude that the occurrence of 10 homicides in a large city.
•As a result, numeric values do not indicate risk.
Definition:
•A proportional distribution, is the percent (i.e., proportion) of the total number of events in a data set which occurred in each of the categories (or subgroups) of that set. Source: Principles of
Epidemiology-Statistical Measures used in Epidemiology, CDC Self Study Course Manual 3.
•Like numeric values, percentages are commonly used, easy to understand, and simple to calculate.
•For any given data set, the sum of all values must equal 100%
•Use flip chart to review calculation method. Review with students the terms numerator and denominator.
•Use flip chart to calculate occupant restraint use from an observational survey. N=227; Yes=68 (30%); No=159 (70%)
•When data sets are small, percentages can be misleading. N=20; Yes=6 (30%); No=14 (70%)
•Also like numeric values, percentages are not a measure of risk.
•Trigger phrase: “…comparing apples to oranges…”
•This phrase is applicable when comparing the number of injury deaths in different communities without considering population size; or comparing the number of crashes on different roadways without considering traffic volume.
•To adjust for variations in exposure, like population size or traffic volume, rates are calculated.
•Define rate:
An expression of the frequency with which an event occurs in a defined population over a specific period of time and converted to a whole number by multiplying to some power of 10 (usually 10,000 or 100,000).
Using a flip chart, write the basic formula for a rate (see slide 22) and describe the following:
Components of a rate are:
…numerator (the number of events in a specific time period)
…denominator (generally the population exposed; sometimes related to other expression of exposure, like traffic volume)
…and a power of ten
•You’ve seen rates expressed in many different ways.
•Here’s a simple table that expresses the injury rate for three different communities.
•Which community has the greater rate? In which community would a person have a greater probability (or risk) of being injured?
•Let’s take a look at the basic calculation of a rate…
•Describe calculation.
•Identify sources of denominator (census, tribal enrollment, user population, traffic volume estimates)
•Remind students that (K) is usually expressed as 10,000 or 100,000. If necessary, relate this piece of the calculation to percentages with a statement like…just as we multiply by 100 in calculating
percentages, we multiple by 10,000 or 100,000 in calculating rates.
•Refer to the “(same time period)” piece of the calculation. Indicate that a common mistake people make when calculating a rate for a multi-year period is that they forget the population should be the combined population for each year. (Consider using a flip chart to elaborate with an example.) For example, the injury death rate for a community for the 3-year period 1990-1992 is calculated as: # of cases in 1990 + # of cases in 1991 + # of cases in 1992 divided by the population in 1990 + the population in 1991 + the population in 1992) times 100,000.
•Review calculation.
•There are different types of rates.
•Crude rate: based on the actual number of events in a total population over a given period of time.
Example: Injury Death Rate for the whole community.
•Specific rate: based on the actual number of events in a subgroup of a population over a given period of time.
Example: Injury Death Rate for specific age-groups in a community.
•Adjusted rate: rates constructed to permit fair comparison between groups differing in some important characteristic.
Example: Adjusted rates for miscoding of Indian race
Example: Age Adjusted rates for variations in age among different populations (Florida adjusted b/c so many retirees; AI/AN adjusted b/c such a young pop)
Some general considerations for rates:
•The numerate should be accurate…that’s the number of your cases.
•The denominator is typically estimated (e.g, population).
•The denominator isn’t always population based. Instead it is some other indicator of exposure, such as vehicle miles or worker hours.
•Rates are primarily used to compare different groups (like communities) or different subgroups (like age groups within a community).
•Rates indicate the probability (or risk) of an event (like injury) occurring.
Years of Potential Life Lost or YPLL is another method of data analysis.
You all are aware that “…the burden of injury falls disproportionately on the young.”
•Comparing the total number of injury deaths with deaths from other causes (e.g., cancer, heart disease) can be misleading.
•It is important to consider how the deaths of so many young affect the future.
•The effect of this premature mortality is reflected in the measurement of YPLL.
•YPLL measures the potential life lost for persons between ages 1 and 65 at the time of death.
•The calculation is simple: 65 – age at death.
•For example: For a person killed in a car crash at age 25, the YPLL is 40; A person who dies of cancer at 60, the YPLL is 5.
•In 1985 the YPLL for injury in the US was 3,476,752. In comparison the YPLL for cancer was
1,813,245 and for heart disease was 1,600,265. More potential years of life were lost due to injury than due to cancer and heart disease combined.
•Note: WISQARS allows for YPLL calculation.
Self explanatory except for last bullet.
“Utilize available resources” refers to use of Epi Info (free), data sources (WISQARS), and technical expertise of local IHS IP Practitioners (Service Unit, District, Area); and experts (Statisticians, Epidemiologists, etc).
•Total exercise time: 30 minutes
•Break into groups of 2-3.
•Hand out Rate exercise worksheet (or have them turn to it if in the course book).
•Provide 20 minutes for students to work through the calculations. Instructors should float around the class and look for students having problems with mathematical calculations. Provide assistance as necessary.
•Review the answers, making sure you discuss the difference in interpreting proportions and rates; and the importance of relating the number of injuries in a given population (age-group in this example) to the total population.
