Education
Visualizing surgical quality data with treemaps

https://doi.org/10.1016/j.jss.2014.03.046Get rights and content

Abstract

Background

Treemaps are space-constrained visualizations for displaying hierarchical data structures using nested rectangles. The visualization allows large amounts of data to be examined in one display. The objective of this research was to examine the effects of using treemap visualizations to help surgeons assess surgical quality data from the American College of Surgeons created the National Surgical Quality Improvement Program database in a quick and timely manner.

Study design

A controlled human subjects experiment was conducted to assess the ability of individuals to make quick and accurate judgments on surgery data by visualizing a treemap, with data hierarchically displayed by surgeon group, surgeon, and patient. Participants were given 20 task questions to complete involving examining the treemap and comparing surgeons' patients based on outcomes (dead or alive) and length of stay days. The outcomes measured were error (incorrect or correct) and task completion time.

Results

120 participants completed 20 task questions for a total of 2400 responses. The main effects of layout and node size were found to be significant for absolute error, P < 0.0505 and P < 0.0185, respectively. The average judgment time to complete a task was 24 s with an accuracy rate of approximately 68%.

Conclusions

This study served as a proof of concept to determine if treemaps could be beneficial in assessing surgical data retrospectively by allowing surgeons and healthcare administrators to make quick visual judgments. The study found that factors about the layout design affect judgment performance. Future research is needed to examine whether implementing the treemap within a dashboard system will improve on judgment accuracy for surgical quality questions.

Introduction

Quality in healthcare has come to the forefront in recent years as an essential concept in providing adequate and efficient healthcare for patients. The Institute of Medicine has defined quality of healthcare as “the degree to which health services for individuals and populations increase the likelihood of desired health outcomes and are consistent with current professional knowledge” [1]. An important aspect of measuring quality in healthcare is directly attributed to healthcare data. For healthcare organizations, data are central to both effective healthcare and to financial survival. Data about the effectiveness of treatment, accuracy of diagnoses, and practices of healthcare providers are crucial to maintain and improve healthcare delivery. As such, the American College of Surgeons created the National Surgical Quality Improvement Program database to help with quality data reviews.

The program uses a prospective validated database to quantify 30-d risk-adjusted surgical outcomes to measure and improve the quality of surgical care, which allows valid comparison of outcomes among hospitals [2]. ACS NSQIP collects data on 136 variables, including preoperative risk factors, intraoperative variables, and outcomes for patients undergoing major surgical procedures in both the inpatient and outpatient setting. Hospitals use the NSQIP database for continuing education, quality improvement, and research. The semiannual summaries, provided by the ACS, assist in targeting problematic surgical outcomes. High quality data, however, only satisfies its intended use in decision-making and planning if it can be effectively monitored and interpreted.

Although the NSQIP database is very effective, it is challenging to review vast amounts of quality data. Surgeons can be presented with individual, sectional, divisional, and departmental level data, so it may be difficult to quickly determine where the priority clinical issues exist. The current output for the database is in the form of traditional summary charts that often have six pages of information. Clinicians have limited time to spend reviewing quality data, therefore they could use some means to visualize the entire dataset to quickly get an overview of what is going on, with access to details as needed, without having to query the system repeatedly, similar to having a dashboard system that can quickly relay key information. Data visualization can meet the need of optimizing the clinicians' ability to summarize, synthesize, and prioritize the large amounts of data available to them.

Specifically, treemap visualizations are a potential beneficial way of visualizing NSQIP data. This type of visualization allows for large-scale hierarchical data to be displayed on one screen. Treemaps are space-constrained visualizations that display multidimensional data as sets of nested rectangles with areas proportional to a specified dimension of the data [3]. The space-filling approach is characterized by subdividing a window into parts representing the branches and leaves of the tree [4]. The area of these parts is often related to some attribute such as size, which can be aggregated. This approach gives a better overview of the entire hierarchy, especially for the attribute that is mapped to the area. Treemaps have been found to be very effective in various domains from finance to air traffic flow management in analyzing large hierarchal datasets [5], [6] but there have been few examples to illustrate whether the effectiveness of treemap visualizations can translate to assessing data in the healthcare domain.

Section snippets

Background

The goal of this research was to map data from the NSQIP database to treemaps to test the validity of using this visualization to analyze data quickly and efficiently. There are several healthcare quality questions that surgeons and health personnel use the NSQIP data to answer that would translate to being visualized with treemaps. Table 1 gives an example of these use case questions for surgery data. The questions range from simple monitoring and search questions to more intensive analysis

Participants

One hundred twenty undergraduate and graduate students of the University of Virginia volunteered for the study (n = 56 males, n = 64 females). The average age for participants was 23.6 (standard deviation [SD] = 5.02). A protocol for this study was submitted and approved by the Institutional Review Board for Social and Behavioral Sciences at the University of Virginia (# 2012-0422-00).

Apparatus

Each participant performed the experiment using a password-protected web-based custom application on a computer

Results

Overall, the average completion time per question was M = 24 s, (SD = 17.7 s). The main effects and the error percentages for correct responses and average completion for the different independent variables are shown in Table 3, Table 4, respectively.

From Table 4 it is implied that the rate of error depends mainly on the layout (grouped versus traditional) and the size of the individual patient's rectangles (varied to represent LOS versus not varied). The main effect of layout was found to be

Discussion

The aim of this study was to explore using treemap visualizations to help surgeons assess NSQIP quality data. This visualization can potentially serve as an alternative to the current way of looking at NSQIP data by paging through multiple pages of reports. The study found that factors about the layout design do affect judgment performance. One would typically think that mapping LOS to the size of the rectangles would be a logical approach when mapping data characteristics onto the features of

Acknowledgment

Author contributions: A.L.H., S.A.G., and F.E.T. contributed to the study conception and design. A.L.H. did the acquisition of data. A.L.H., S.A.G., and F.E.T. did the analysis and interpretation. A.L.H., S.A.G., and F.E.T. did the drafting of the article. A.L.H., S.A.G., and F.E.T. did the critical revision of the article.

The research described was supported in part by Grant Number T15LM009462 from the National Library of Medicine. The content is solely the responsibility of the authors and

References (12)

  • Crossing the quality chasm: a new health system for the 21st century

    (2001)
  • American College of Surgeons. ACS national surgical quality improvement program. Available at:...
  • Shneiderman B. Tree visualization with tree-maps: 2-d space-filling approach. ACM TOG 1991;...
  • T. Asahi et al.

    Using treemaps to visualize the analytic hierarchy process. Technical report CAR-TR-719

    (1994)
  • Smartmoney map of the market. Available at: http://www.smartmoney.com/marketmap. Accessed November...
  • W. Jiening et al.

    Visualizing air traffic flow management alert information using squarified treemaps

    IEEE Computer Graphics, Imaging And Visualization

    (2009)
There are more references available in the full text version of this article.

Cited by (0)

View full text