Using tables and graphs to support a presentation is one of the most effective ways to tell a story or communicate an idea, especially to an audience who may not have a strong grasp of how to interpret the data.1 However, merely presenting data using a graph or table does not assure its impact. To stage data in a visually compelling manner, pharmacy must have a clear context of the audience. Moreover, it is important to understand that certain graph types are better able to communicate specific types of data. When pharmacy makes an effort to organize and display data in a deliberate fashion, the resulting presentation can be extremely influential in furthering pharmacy’s objectives within the health system.

The University of North Carolina Medical Center (UNCMC) is a public, not-for-profit, academic medical center with more than 800 beds and a 75-bed community hospital, which serves as an extension of the medical center. In 2013, the UNC department of pharmacy recognized the importance of leveraging data to support decision-making and continued growth. Thus, pharmacy established an embedded Pharmacy Analytics team, which has grown over the years to support reporting and analytics for pharmacy services across the UNC Health Care System.

The Evolving Landscape of Data

As the quantity of health care data increases exponentially, pharmacy leaders are increasingly challenged to use data to drive decision-making and communicate the value of pharmacy services to administration. Given the ever-changing nature of the health care environment, hospitals and health systems must use data not only to promote efficient operations, but also to demonstrate positive patient outcomes. Within the pharmacy department, data should be similarly leveraged to demonstrate financial performance, operational efficiency, and positive impacts on patient care.

When telling a story with pharmacy data, several unique issues must be considered. A major barrier to coalescing pharmacy data is that depending on the setting, the pharmacy may rely on different technology, software, and processes for the same operations, such as purchasing, inventory management, dispensing, and clinical care. Each piece of automation and software used in health care generates data; for this data to bring value, it must be accessed and aggregated.

This task can be complicated by free text documentation, such as notes within the EHR, as the information saved is unstructured and difficult to aggregate. Whenever possible, data collected for tracking operations or productivity should be documented in a discrete format, such as selecting from a list of options, instead of a free-text response. To further confound matters, each pharmacy leader who seeks to examine data to glean information about workload or capacity likely has a different definition of what they would like to measure, which presents additional challenges.

Key Principles in Visualizing Pharmacy Data

When visualizing pharmacy data, there are two key principles to keep in mind: understanding the context of the data and comparing units of equal measure.

Understanding Context

The context of the data being presented should be assessed from the perspective of both the presenter and the audience. Questions from the presenter’s perspective include:

• Do you understand the data in its entirety?
• Can you describe the data sources in detail (eg, identify the discrete field from which the data was accessed, explain what action triggers a time stamp, etc)?

As the owner of the data, the presenter must be able to discuss the visualizations with confidence and mitigate any potential uncertainties prior to dissemination. It may be best to start with a review of all the available data and build an understanding of the context before concentrating on the data that will be presented.

From the audience’s perspective, several questions should be asked:

• Does the visualization make sense?
• Can the visualization stand alone without additional explanation?
• If someone views the data outside of the context of the presentation, will they be able to draw the conclusions you are trying to convey? If not, you may need to add supplemental information, such as captions or labels, to your visualization.

In some situations, adding trend lines to a graph will serve to effectively demonstrate average growth (positive or negative), that would otherwise be unclear. To help the audience visualize the effect of a change over time in a graph with pre- and post- results, a vertical line or arrow can be employed to call attention to the point in time in which the change occurred.

Compare Units of Equal Measure

Beware of aligning disparate items that cannot be compared against each other with any accuracy. Pharmacy is a diverse department with employees engaged in multiple types of work. As such, there is no clear methodology to measure productivity across all pharmacy areas. Leaders should use caution when trying to present multiple metrics that cannot be measured against each other.

Consider the following: Should the number of dispenses from a pharmacy carousel be measured equally to the number of dispenses out of the IV room? Are the same number of technicians and pharmacists required per 100 doses dispensed from each area? If the answers to these questions is no, it would be inaccurate to present this data in side-by-side comparisons.

In addition, note that pharmacists and technicians work in health systems that encompass a variety of hospital types and practice settings; an academic medical center or teaching hospital may have five times the volume compared to a smaller community hospital within the same health system, but seven times the FTEs. There is no value to comparing a standardized metric without taking into account the key differences between entities. For example, larger sites may have a residency program with resident FTEs, or a higher acuity of patients, which may impact metrics.

Visualization Tools

The two most common visualization tools used in presentations are tables and graphs. Understanding when it is appropriate to use each of these is critical to accurately conveying information to administration and other key stakeholders.

Tables

As a communication tool, tables provide a simple, traditional method of visualizing data and they serve an important role in presenting certain kinds of information. Tables are read; the data can be read across rows, down columns, or to compare values. They are useful to communicate data to a broad audience who will be looking for their row of interest, as well as to represent multiple units of measure.1 Tables are often used for housing large quantities of information—eg, multiple rows or columns that would be confusing or impractical in the form of line and bar graphs.

In situations where tables are the most practical visualization tool, and the presenter would like to highlight specific trends or points of intensity, it is useful to utilize heat maps, a type of table that leverages color saturation in order to draw attention to points of interest (eg, high values).1 For example, UNCMC leverages a heat map to demonstrate the total number of dispenses by hour of the day, at each pharmacy location (see TABLE). Due to the large number of pharmacies, a bar or line graph would be visually confusing. The heat map format provides the information of interest, with the color differences allowing the end users to see at a glance the busiest hours of the day for their teams, as well as the busiest dispensing pharmacies.

Click here to view a larger version of this Table

Because tables are meant to be read, they are less ideal for live presentations. It is challenging for audiences to listen to a speaker while reading slide content. Therefore, other forms are preferred for these scenarios.

Graphs

As opposed to tables, graphs are visual; they are viewed, not read. As such, graphs can communicate information more quickly than a table. While many graph types exist, the majority of data can be well-represented with the two most common types: line graphs and bar graphs.1 In general, extravagant graph types should be avoided, as they can unnecessarily add complexity. Graphs present data in summary format and reduce data complexity when leveraged appropriately.

• Line Graphs. Line graphs are effective for plotting continuous variables. In a line graph, the individual data points are connected, which is not the case for categorical data. Most commonly, the continuous data is a unit of time. Therefore, to demonstrate a change over time, line graphs are most effective. Line graphs also are useful to demonstrate a logical progression of values (eg, trends in pharmacy data).1,2

  These graphs can represent a single series of points or multiple series. The number of orders, charges, or prescriptions filled by month for the fiscal year to date are examples of situations in which lines graphs are effective. As noted earlier, applying a trend line over the line graph can be useful to illustrate the overall directionality of the data.

  An example of a line graph is included in GRAPH 1.

• Bar Graphs. In comparison, bar graphs provide more versatility than line graphs. Audiences can compare the end of bar charts to make inferences: For example, what are the largest, the smallest, and the incremental differences?

  While line graphs can be used to show trends over time, bar graphs are more useful in situations demonstrating large-scale changes. Bar graphs are used to compare specific values and are representative of different groups or categories. Stacked bars can be used to show the proportional contributions that make up the total of each bar (ie, the top of each bar represents the accumulated total for that category). When viewing bar graphs, the audience is comparing the relative end points; it is important to always have a zero baseline when using a bar graph.2,3

  An example of a useful bar graph is one demonstrating the number of verifications compared by patient locations. Stacked bars can be used to demonstrate number of verifications by order priority (eg, STAT, routine).

  An example of a bar graph is included in GRAPH 2.

Click here to view a larger version of this Graph

Mary-Haston Leary, PharmD, MS, BCPS is the clinical manager, analytics, outcomes, and impact, at University of North Carolina (UNC) Health Care and assistant professor of clinical education with the UNC Eshelman School of Pharmacy. She received her PharmD from the University of Mississippi School of Pharmacy, and completed her PGY1/PGY2/MS with UNC Hospitals and the UNC Eshelman School of Pharmacy. Mary-Haston’s professional interests include practice advancement through analytics and data-driven decision-making, as well as practice-based outcomes research.

Evan W. Colmenares, PharmD, is the lead pharmacist, analytics, outcomes, and impact, at UNC Health Care, and a PhD candidate in the UNC Eshelman School of Pharmacy division of pharmaceutical outcomes and policy. He received his PharmD from the UNC Eshelman School of Pharmacy. Evan’s professional interests include pharmaceutical outcomes research, innovative pharmacy practice models, and data-driven clinical decision-making.

References

1. Knaflic CN. Storytelling with Data: a Data Visualization Guide for Professionals. Hoboken, NJ: Wiley; 2015.
2. When and how to use a line graph. Lynda. www.lynda.com/Design-tutorials/when-how-use-line-graph/585239/641580-4.html. Accessed June 12, 2019.
3. Chart Types. IBM. www.ibm.com/support/knowledgecenter/en/SSVJ22_10.2.2/com.ibm.swg.ba.cognos.dsk_ug.10.2.2.doc/c_dsk_charts_types.html. Accessed June 12, 2019.