Department heads in healthcare and senior living are shown charts constantly: census trends, staffing hours, call volumes, incident counts, survey results. Most of us were never taught how to read them critically. A chart can look authoritative while hiding as much as it shows, and decisions made from a misread chart can be costly.
You do not need a statistics degree. A few habits will help you ask better questions and avoid being misled, whether by a vendor, a colleague or your own enthusiasm.
Before looking at the lines or bars, read the labels.
A chart with an axis that begins well above zero can make a small change look dramatic. If falls rise from 8 to 9 in a month, a truncated axis can make it look like a surge. Bar charts in particular should normally start at zero, since the bar's height is meant to represent the amount. Line charts sometimes reasonably zoom in, but you should know when they do.
Notice whether the steps are even and whether the units are clear. Are we counting residents, incidents, hours or percentages?
A chart showing the last three months may hide a longer pattern. A dip that looks alarming may be normal seasonal variation.
A number means little without something to compare it to. Ask, "Compared to what?"
Last month, last year or the same period last year?
A target or standard?
Another facility of similar size?
Be careful with comparisons across different sizes. Ten falls in a 120-bed building and ten in a 40-bed building are not the same. Look for rates, such as falls per 1,000 resident days, instead of raw counts when sizes differ.
An average can be accurate and still misleading.
Outliers pull averages. A few extremely long call-light response times can raise the average, while most responses are fine. Or the reverse, where an average looks fine and hides a handful of very long waits that matter to residents.
Averages hide variation. Two units can have the same average with very different experiences.
Median is the middle value, and it is often better when a few extreme values exist.
Ask for the range, the median or a breakdown by shift, unit or day of week.
With small counts, a change of one or two events can swing a percentage dramatically. Going from one pressure injury to two is a 100 percent increase, which sounds dramatic but may be chance. Ask how many events sit behind a percentage.
Two lines moving together do not prove that one caused the other. Staff turnover and falls may both rise for several reasons. Treat a pattern as a clue and ask what else could explain it.
Ask what the chart does not show. Were any facilities or periods excluded? Did the definition of the measure change midway? Were data missing?
Charts with two vertical axes can create an illusion of relationship.
Three-dimensional effects distort sizes.
Cherry-picked time frames can make a trend look better or worse.
Colors can imply good or bad without evidence.
Where does the axis start?
Compared to what?
How many events are behind this?
Is this an average, and what is the spread?
Has the definition changed?
What would change my mind?
UnityCare IT helps organizations bring their data together and present it clearly, with honest scales and sensible comparisons. Good dashboards make good conversations possible, and a little data literacy across your leadership team makes them better still.
An outsourced IT department with proactive maintenance and one number to call.
Call or text: 405-285-3845
New customers: start@unitycareit.com
Existing customers: support@unitycareit.com
Address: UnityCare Technologies, 2524 N Broadway Ste 554, PMB 947974, Edmond, Oklahoma 73034-4172