IHP 525 Module 2 Descriptive Statistics Paper Example

Describing the Baseline: Summary Statistics for Bayview's Community Blood Pressure Program

[Student Name]

Southern New Hampshire University

IHP 525: Biostatistics

Module Two Paper

[Instructor Name]

[Date]

Describing the Baseline: Summary Statistics for Bayview's Community Blood Pressure Program

Before testing whether Bayview County's community blood pressure program worked, the analyst must describe who enrolled and what their data look like. Good description shows readers the shape and spread of the data, reveals problems such as outliers and missing values and supports the choice of later analyses. This paper describes baseline data for 240 program participants, explains the choice of each summary statistic, calculates a confidence interval for mean systolic pressure, argues for keeping blood pressure continuous and examines missing follow-up data.

Matching Statistics to Data Types

The type and shape of each variable determine how to summarize it. Categorical variables, such as sex or race, are summarized with counts and percentages. Continuous variables that are roughly symmetric, such as systolic blood pressure in this sample, are summarized by their average and SD, which describe the center and the typical distance from it. Continuous variables that are skewed, such as the number of medications or body mass index with a long upper tail, are better summarized with the median and interquartile range, because a few extreme values can pull the mean away from where most observations lie.

Participant Characteristics

The 240 participants had a mean age of 54.3 years, with a standard deviation of 9.8 years. Of these, 142, or 59.2%, were men and 98, or 40.8%, were women; 168, or 70.0%, identified as Black, 46, or 19.2%, as Hispanic and 26, or 10.8%, as white or another group. Body mass index was right-skewed, with a median of 30.8 and an interquartile range of 27.4 to 35.1. Participants took a median of one blood pressure medication, with an interquartile range of zero to two; 58, or 24.2%, took none.

A comparison group of 260 clinic patients will be described with the same statistics in a later milestone, so that differences between groups at baseline can be seen before any outcomes are compared.

Table 1. Baseline Characteristics (n = 240)

Note. Data are illustrative for the composite program.

Systolic Blood Pressure

Baseline systolic pressure had a mean of 152.4 mm Hg and a standard deviation of 14.1 mm Hg; the median was 150 mm Hg, close to the mean, and a histogram showed a roughly symmetric distribution with a slight upper tail. About 95% of participants had values within roughly two standard deviations of the mean, between about 124 and 181 mm Hg. Two values above 200 mm Hg were checked against the original records; both were real and were kept, since removing true extreme values would misrepresent the population.

A box plot by recruitment site showed similar distributions across the six barbershops and four churches, suggesting that participants recruited in different settings started from comparable pressure levels and that site need not be treated separately in the main comparison.

Precision of the Mean

Gardner and Altman (1986) argued that researchers should report confidence intervals, which convey the span of plausible values, rather than relying on p-values alone. The same principle applies to descriptive estimates. To get the standard error of mean systolic pressure, take the SD of 14.1 and divide by 15.5, the square root of 240; the result is 0.91 mm Hg. A 95% confidence interval is the mean plus or minus 1.96 times the standard error: 152.4 plus or minus 1.78, or 150.6 to 154.2 mm Hg. The narrow interval reflects the large sample; it describes the precision of the estimated average, not the spread of individual values, which the standard deviation describes.

Keeping Blood Pressure Continuous

Program reports often summarize blood pressure only as the percentage of people with controlled pressure, below 140/90. That measure is meaningful to clinicians, but Altman and Royston (2006) showed that splitting a continuous variable into two groups throws away information, reduces statistical power, treats people just above and just below the cutoff as completely different and can hide the true shape of relationships. They estimated that dichotomizing can waste the equivalent of a substantial share of the data. For Bayview, a participant whose pressure falls from 170 to 145 mm Hg improves greatly but still counts as uncontrolled. The evaluation will therefore report mean systolic pressure and mean change as primary measures, with the control rate as a secondary, clinically familiar measure.

Missing Data

Of the 240 participants, 218 had a six-month blood pressure measurement; 22, or 9.2%, were missing. Sterne et al. (2009) explained that analyzing only people with complete data can bias results if those with missing values differ systematically, for example if people who were not improving stopped attending. They described multiple imputation, which fills in missing values several times based on other information and combines the results, as a useful approach when missingness can be explained by variables that were recorded, while warning about common pitfalls such as omitting the outcome from the imputation model. Participants with missing follow-up were younger, with a mean age of 47.9 years, and had slightly higher baseline pressure, 156.8 mm Hg, suggesting they differ from completers and that complete-case analysis could be biased.

Implications for Analysis

The description supports several analytic choices. Because systolic pressure is roughly symmetric, a t-test comparing mean change between groups is reasonable. Because missing participants differ, the main analysis will use multiple imputation, with complete-case results as a sensitivity check. Because body mass index is skewed, it will be summarized with medians and, if used as a covariate, kept continuous. Because the sample is predominantly Black and male, results may not apply equally to other groups.

Conclusion

Careful description revealed a sample of middle-aged adults with high baseline pressure, symmetric blood pressure data suitable for mean comparisons, skewed variables that need medians and missing follow-up that could bias a simple analysis. Keeping blood pressure continuous, reporting intervals and planning for missing data will make the evaluation more accurate and more honest.

References

Altman, D. G., & Royston, P. (2006). The cost of dichotomising continuous variables. BMJ, 332(7549), 1080. https://doi.org/10.1136/bmj.332.7549.1080

Gardner, M. J., & Altman, D. G. (1986). Confidence intervals rather than P values: Estimation rather than hypothesis testing. BMJ, 292(6522), 746-750. https://doi.org/10.1136/bmj.292.6522.746

Sterne, J. A. C., White, I. R., Carlin, J. B., Spratt, M., Royston, P., Kenward, M. G., Wood, A. M., & Carpenter, J. R. (2009). Multiple imputation for missing data in epidemiological and clinical research: Potential and pitfalls. BMJ, 338, Article b2393. https://doi.org/10.1136/bmj.b2393