Chapter 5: Aggregating & Analyzing Performance Improvement Data

Comprehensive vocabulary flashcards covering the concepts, definitions, metrics, and quality improvement tools presented in Chapter 5: Aggregating & Analyzing Performance Improvement Data.

Internal Benchmarking

A performance improvement strategy that compares metrics and processes between departments or teams within the same organization to identify internal best practices using readily accessible, contextually relevant data.

External Benchmarking

A performance improvement strategy that compares an organization's performance against industry peers, competitors, or best-in-class companies to provide a realistic view of industry standards and competitive threats.

Sampling

The practice of selecting a representative subset of patient records or individuals from a larger population to analyze quality, performance, or clinical outcomes efficiently without expending excessive resources.

Check Sheet

A structured, real-time data collection tool used to systematically record, monitor, and analyze clinical or operational processes, thereby reducing fatigue-related medical errors and standardizing workflows.

The 4 W's of Data Collection

The four foundational questions that must be planned prior to recording data on a check sheet: Who will collect the data, What data will be collected, Where will the data be collected, and When will the data be collected.

Qualitative (Categorical) Data

Data that describes attributes, characteristics, or categories rather than numerical measurements, divided into nominal and ordinal subcategories.

Nominal Data

A form of categorical, qualitative data where categories have no inherent order or sequence, such as nationality.

Ordinal Data

A form of categorical, qualitative data that follows a natural or predetermined ranking or order, such as letter test grades.

Quantitative Data

Numerical data that represents measurable quantities, divided into continuous and discrete data types.

Discrete Data

Quantitative data consisting of whole, distinct countable units that cannot be divided into smaller fractional parts, such as counts of cats.

Continuous Data

Quantitative data that can be measured along an infinite continuum and divided into finer fractions or decimal increments, such as distance.

Primary Data

First-hand information gathered directly by a researcher for the first time for a specific performance measurement purpose (e.g., surveys, interviews, experiments); typically expensive and time-consuming but highly accurate and tailored.

Secondary Data

Pre-existing information collected by others for an alternative purpose (e.g., journals, websites, published reports) that is repurposed for performance improvement; faster and cheaper to obtain, but dependent on external data validity.

Absolute Frequency

The exact, direct whole-number count of how many times a particular event, response, or item was observed.

Relative Frequency

The proportion of total observations represented by a specific count, calculated as the absolute frequency divided by the total observations ($\text{Relative Frequency} = \frac{\text{Absolute Frequency}}{\text{Total Observations}}$) and expressed as a decimal, fraction, or percentage.

Data Querying

The structured process of requesting, retrieving, modifying, or deleting specific information from a database to support business intelligence, reporting, and analysis without losing focus.

Data Mining Rabbit Hole

The pitfall occurring when analysts lose sight of the primary business question and become sidetracked by irrelevant data sets, spurious correlations, or excessive secondary research.

Five Components of the Data Lifecycle

The cyclic stages governing information management from inception to obsolescence: capture, process, use, store, and dispose.

Ten Characteristics of Data Quality

The critical attributes that define reliable healthcare data: accuracy, accessibility, comprehensiveness, consistency, currency, definition, granularity, precision, relevancy, and timeliness.

Median

The central value in an ordered numeric dataset, preferred in healthcare analytics because it is resistant to distortion from skewed distributions, extreme outliers, and astronomical values.

Standard Deviation

A statistical metric measuring dispersion around the mean, utilized to establish upper and lower limits for control charts and define acceptable process spread along a bell curve.

Variance (Variation)

In healthcare, any deviation from an established standard of care, clinical pathway, or projected budget, representing a discrepancy between planned and actual outcomes.

Special Cause Variation

A form of process variation not continuously present that arises from specific, out-of-the-ordinary circumstances; it can be attributed to an identifiable source and should be eliminated immediately once detected.

Common Cause Variation

The natural, expected variability inherently present in every outcome of a process due to systemic elements such as design, training, materials, machinery, or working conditions.

Predictive Analytics

The application of statistical techniques, machine-learning algorithms, and historical data patterns to forecast future occurrences and guide organizational decision-making.

Bar Graph

A graphical display featuring separated vertical bars that illustrate counts or frequencies of discrete categories along a horizontal axis, with the vertical axis always originating at $0$.

Histogram

A specialized graphical display showing the continuous distribution and relative frequencies of numerical data divided into intervals (bins) using contiguous, touching bars with no gaps.

Pareto Chart

A quality improvement display combining a bar chart in descending frequency order and a cumulative percentage line graph to prioritize the most critical factors in a dataset.

Pareto Principle (80/20 Rule)

The rule of thumb asserting that roughly $80\%$ of effects or defects stem from $20\%$ of root causes, distinguishing the 'vital few' from the 'trivial many'.

Pie Chart

A circular statistical graphic partitioned into slices representing proportional shares of a whole ($100\%$), most effective when visualizing simple breakdowns of $2\text{--}5$ categories.

Pivot Table

An interactive spreadsheet tool used to rapidly summarize, sort, group, and reorganize voluminous datasets (such as patient claims or diagnosis records) across multiple dimensions.

Run Chart

A chronological line graph that tracks a measure over time against a median centerline to detect systemic process trends, shifts, or runs without requiring advanced statistical calculation.

Shift (Run Chart)

A run chart pattern evidenced by $7$ or more consecutive data points positioned entirely on one side (above or below) of the median line.

Trend (Run Chart)

A run chart pattern evidenced by $7$ or more consecutive data points consistently increasing or decreasing in value.

Astronomical Data Point

A blatantly unusual data point on a run chart that stands dramatically apart from the surrounding baseline pattern, immediately indicating an anomalous occurrence.

Control Chart (Process Behavior Chart)

A statistical graph originally formulated by Walter A. Shewhart in $1931$ that uses calculated upper and lower control limits derived from standard deviation to differentiate between common cause and special cause variation.

Scatter Diagram

A Cartesian coordinate visualization that plots paired numerical measurements (typically a hypothesized cause on the horizontal axis and effect on the vertical axis) across at least $50\text{--}100$ paired samples to determine correlation.

The Checklist Manifesto

A $2009$ book written by Dr. Atul Gawande illustrating how standardized, evidence-based procedural checklists reduce memory lapses, communication failures, surgical infections, and mortality.