Chapter 4 - illustrate data summaries

CHAPTER SUMMARY

After data surveillance is complete it is time to publish your findings. How do you do this? No one is going to read a long spread sheet! You need to use small tables, line graphs, epidemic curve, population pyramid graphs, frequency polygons, etc. This chapter introduced these different ways of showing your powerful data so readers can...well.. feel your data's power.

CHAPTER CONTENT

Review Data Type. Nominal data is qualitative data where the order of categories does not make a difference. For example male and female data. Ordinal data is also qualitative and here order does matter to an extent. For example, low medium and high, or stage 1, 2, 3, and 4. Discrete data is quantitative data in the form of integers. For example, number of ill cases. Continuous data is quantitative data that can have any numerical value. For example, time of symptom onset.

Table Creation: how to make your table stand alone.  First, include person, place, and time followed by the table number in the title of your table. You can't read a book by it's cover but you should be able to understand a table and graph by theirs. Second, label units in column and row headings. Third, sometimes you should make a column or row for totals, like when your table includes percents. Fourth, provide disclaimers for any missing data and excluded trials rather in your table or in a footnote. Fifth, define all codes and abbreviations with in a footnote. Sixth, beneath your table provide the source of your data and mention rounding error if totals add up to less than or more than 100%.

The Facts About Tables. First, the smaller your table the more time readers spend looking at it. Second, columns are compared to other columns for example data on males and data on females should be written in columns if you want them compared. Third, by using percents in a table you are showing the relative burden of illness or other outcome.

Types of Tables. One Variable Tables describe occurrence based on one variable such as age, gender, etc. Two Variable Tables describe occurrence based on two variables. Three Variable Tables describe occurrence based on three variables. Both Two and Three Variable Tables  are also contingency tables which means the table is being used to expose the association between variables. A type of Two Variable contingency table is a Two-by-Two which has two variables are compared that each have two categories. An example can better explain Two-by-Two tables. Imagine Number of people and Exposure. Number of people is a variable that has the category with the disease and without the disease. Exposure also has two categories, with exposure or without exposure. Composite Tables create a variable called 'characteristics', under this variable many other variables can be written like, age (with intervals), gender, demographics, risk factors, etc. Even though many variables are present, no association between variables are being made. Because of this Composite Tables operate like a bunch of One Variable Tables crammed into one table. Usually data tables are pre-made in what is called a Table Shell. These include titles, headings, and categories, really all the information is there except the numbers. Also many Table Shells include more variables than will seen in the final table. This is done in order to simply the table after the correct variable associations are discovered.

Strategies For Grouping Data. Try making groups of equal sizes. Or make groups based on mean, standard deviation and range. Another way is to make groups based on the range divided by the number of desired groups. Remember, class intervals should be mutually exclusive and exhaustive. This means that a person should only fit into one interval and that intervals should include every case.

Graphs: display numerical data in visual form. For graph creation follow the same rules as went over under the heading "Table Creation: how to make your table stand alone". In addition to those rules limit the number of lines you use. Every line you draw should maximize it's meaning for the space it takes up. Remember that frequency is usually put on the vertical axis and categories on the horizontal axis. When making a graph, make the horizontal axis longer than the vertical axis, this creates a landscape picture and looks nicer.

Types of Graphs. Arithmetic-Scale-Line Graphs are a simple graph with an x and y-axis, and points connected by lines. This graph is used to show long series of data, to compare several series of data, and to show rates over time. Semilogarithmic Line Graphs are very cool! They are used when the data has changed from a very high number to a very low number. This graph using a logarithm scale/order of magnitudes for the y-axis. Imagine the y-axis labeled as 0, 1, 10, 100, 1000. This is used to show the change in rate. Histograms are very common but are still powerful. For example an Epidemic Curve Histogram looks like this:

When drawing an epidemic curve use time intervals between 1/3 and 1/8 the incubation period. If you are comparing genders use a Population Pyramid like this:

Frequency Polygon Graphs uses a histogram graph and creates a point in the middle of each bar. Each point is then connected creating a line graph. Imagine turning the epidemic curve histogram into a frequency polygon. It would look pretty cool. A Cumulative Frequency Graph plot on the y-axis, cumulative data like the data in the form of percentages and make it easy for readers to see the mean, medium, etc. Survival Curves are cumulative frequency graphs but start with 100% instead of 0%.

Not as Common Graphs: Scatter Diagrams, Bar Charts where the length of text labels determine if the bars will be vertical or horizontal (longer texts) and discrete data is shown where histograms show continuous. Grouped Bar Charts show multiple variables, often subgroups of variables. Stacked Bar Charts better show the comparative value of the first variable. 100% Bar Chart is a stacked Bar chart where one axis is in percent and all bars go to 100%, this better helps you compare subgroups. Deviation Bar Chart shows both negative and positive values as the center of the graph there is a line at '0'. Pie Charts are great for comparing groups to each other and to the overall sum of the groups. Pie Charts should label values of segments, start at 12 o'clock, and should be ordered from largest to smallest slices, other categories can be put last. Often multiple pie charts will be used instead of a 100% bar chart. Dot Plots and Box and Whisker Plots are used to show a continuous variable over a categorical variable. Whisker plots show quartiles where the medium, 2nd, and 3rd quartile are used to create a box and whiskers display the extent of range. These plots are used to compare skewness (seen when the medium is not centered inside the box) between variables and to compare the inner 50% data between variables. Forest Plots are cool. You use them to compare results from different studies showing the same results. The plot shows a line for the confidence interval and a point for the point-estimate. If points line up then studies agree with each other. Also a vertical line is drawn through point 0 showing that any point that is not on this line is showing significance (that the variable has been found to make a difference). Phylogenetic Tree shows the genetic linage of organisms involved in an outbreak. Decision Trees map out choices, outcomes and probabilities. Under each outcome is the outcomes probability which ranges from 0 to 1.

Maps. Spot Maps show dots for every reported case. This is good because it does not show population. See Atlas of United States Mortality, Aids, An Historical Geography of Human Viral Disease. Chloropleth Maps which are area maps use different shades to show different values.

A geographic information system is a computer system for the input, editing, storage, retrieval, analysis, synthesis, and output of location-based information.(22) In public health, GIS may use geographic distribution of cases or risk factors, health service availability or utilization, presence of insect vectors, environmental factors, and other location-based variables. GIS can be particularly effective when layers of information or different types of information about place are combined to identify or clarify geographic relationships.