Section 6.3: Statistical Analysis Fundamentals in Healthcare

Statistical Analysis Fundamentals in Healthcare

Healthcare quality professionals must analyze data to identify trends, measure improvement, and make evidence-based decisions. Understanding descriptive and inferential statistics is crucial for passing the CPHQ exam and leading successful quality initiatives. This section covers key statistical concepts, formulas, and clinical applications.

1. Descriptive Statistics in Healthcare

Descriptive statistics summarize, organize, and describe the characteristics of a specific dataset. They do not draw conclusions beyond the data analyzed.

Measures of Central Tendency

Measures of central tendency identify the single value that represents the center or typical case in a dataset.

• Mean (Arithmetic Average): Calculated by adding all data points and dividing by the total number of points ($N$).
  • Characteristics: Highly sensitive to extreme outliers. For example, in a sample of five patient hospital charges—$1,000, $1,200, $1,300, $1,500, and $25,000—the mean is $6,000. This value is distorted by the single outlier and does not reflect the typical patient charge.
• Median (Middle Value): The midpoint of the dataset when arranged in numerical order. If the sample size is even, the median is the average of the two middle numbers.
  • Characteristics: Highly resistant to outliers and skewed data. In skewed distributions typical in healthcare—such as length of stay (LOS), patient wait times, or costs—the median is the preferred measure of central tendency.
• Mode (Most Frequent Value): The value that occurs with the greatest frequency.
  • Characteristics: Useful primarily for categorical data (e.g., identifying the most common primary diagnosis in an outpatient clinic).

Measures of Dispersion (Spread)

Measures of dispersion describe how spread out the data points are from the center.

• Range: The difference between the highest and lowest values in a dataset. While simple to calculate, it only considers the two most extreme values, making it highly sensitive to outliers.
• Standard Deviation (SD): Measures the average distance of each data point from the mean. A low standard deviation indicates that data points are clustered tightly around the mean; a high standard deviation indicates that the data is spread widely.
• The Empirical Rule (Normal Distribution): In a symmetric, bell-shaped distribution (a normal distribution):
  • 68.27% of all data points fall within $\pm 1$ standard deviation of the mean.
  • 95.45% of all data points fall within $\pm 2$ standard deviations of the mean.
  • 99.73% of all data points fall within $\pm 3$ standard deviations of the mean.

---

2. Inferential Statistics & Hypothesis Testing

Inferential statistics allow quality professionals to make generalizations, predictions, and decisions about a larger population based on a representative sample.

Hypothesis Testing

Hypothesis testing evaluates two mutually exclusive statements about a population:

• Null Hypothesis ($H_0$): The assumption that there is no true difference, effect, or relationship between the compared groups (e.g., 'The new fall-prevention protocol has no effect on patient fall rates').
• Alternative Hypothesis ($H_1$): The statement that there is a statistically significant difference or effect (e.g., 'The new protocol reduces patient fall rates').

Type I and Type II Errors

When conducting a hypothesis test, quality teams can commit two types of statistical errors:

• Type I Error (Alpha, $\alpha$): Rejecting the null hypothesis when it is actually true. This is a false positive (e.g., concluding that a new infection-control protocol is effective when the reduction in infections was actually due to random chance). The probability of committing a Type I error is set by the significance level (usually $\alpha = 0.05$, or a 5% chance).
• Type II Error (Beta, $eta$): Failing to reject the null hypothesis when it is actually false. This is a false negative (e.g., concluding that a new protocol has no effect when it actually does reduce infections).
• Statistical Power ($1 - \beta$): The probability of correctly rejecting the null hypothesis (detecting a true effect when one exists). Power is increased by having a larger sample size.

---

3. p-values and Confidence Intervals

To interpret the results of statistical tests, quality professionals analyze p-values and confidence intervals.

The p-value

The p-value represents the probability that the observed difference between groups occurred solely due to random variation under the assumption that the null hypothesis is true.

• p < 0.05: The result is considered statistically significant. The null hypothesis is rejected, and there is less than a 5% probability that the difference was due to random chance.
• CPHQ Exam Tip: Do not confuse statistical significance with clinical or practical significance. With a large enough sample size, a statistical test can detect a tiny difference (e.g., reducing clinic wait time by 10 seconds) that has no practical or clinical impact on patient care.

Confidence Intervals (CI)

A confidence interval is a range of values that is likely to contain the true population parameter with a specified level of confidence (typically 95%). For example, a 95% confidence interval for a catheter-associated urinary tract infection (CAUTI) rate of 2.1 per 1,000 catheter days might be [1.5, 2.7], meaning we are 95% confident the true population rate lies within that range.

• Interpreting Overlap: When comparing two independent groups, if their 95% confidence intervals do not overlap, the difference between the groups is statistically significant at the 0.05 level. If they overlap significantly, the difference is likely not statistically significant.
• Evaluating Ratios: For metrics reported as ratios (e.g., Relative Risk [RR] or Odds Ratio [OR]): if the confidence interval includes the value 1.0, the finding is NOT statistically significant. A ratio of 1.0 indicates equal risk between groups.
• Evaluating Differences: For metrics comparing the difference between two means: if the confidence interval includes 0.0, the difference is NOT statistically significant. A difference of 0.0 indicates that the means are equal.