WK 3 DIS, Data

QI study performed at a hospital in NY with data from 2017 to 2019, focused on the incidence of SSI in C section patients (CB births) with a hospital-wide perioperative bundle implemented.

2,875 total CBs (N)

Of those, there were 1,086 in the period of time described as “prebundle” and you see from the table that there is a mean age and standard deviation are listed. There are also mean scores for gestational age (GA) with a standard deviation. Both age and GA are measured as continuous data, so an average makes good sense.

Now, sometimes the median is chosen as a better alternative to the mean. Often, that is because the mean is subject to outliers that can distort the distribution. A median score might be more useful. The range is used instead of the SD, b’c it is a measure of variability that is appropriate for the median. The SD shows deviation from the mean. So if the data are subject to broad variability with outliers, the median and range are better measures of central tendency and dispersion that the M and SD. The median is often used for ordinal level data that uses ranks.

This study is provided for your review and you can take a look and get a better idea of the variables ROM and Parity.

For now let’s take a look at ethnicity. This is a categorial variable that is essentially nominal level data. Note that for ethnicity you see counts and percentages. For example, you can see that there were 677 Asian patients in the prebundle that represents 62.3% of the total. You can easily do the math: 677+75+261+73= 1,086 CBs in the prebundle group. Of these, 677 were of Asian descent. So (677/1086) x 100 = 62.3%

7

Hypothesis testing: Part I

The null hypothesis states that there is no relationship or association between variables (the independent and the dependent)

The alternate hypothesis states that there is a relationship or association between variables (the independent and the dependent)

Hypotheses can be directional or nondirectional

The null hypothesis states that there is no relationship or association between variables (the independent and the dependent).

Here is an example from Beydoun et al. (2022)

There is no association between perioperative prophylaxis practices and surgical site infections (SSIs) in patients undergoing vascularized reconstruction of the upper aerodigestive tract (UADT)

The dependent variable is the outcome variable, the SSIs. The independent variable is the one that gets manipulated. In this case that would be the perioperative prophylaxis. That includes both topical prophylaxis and antibiotic use.

The alternate hypothesis states that there is a relationship or association between variables (the independent and the dependent)

There is an association between perioperative topical antisepsis and surgical site infections (SSIs) in patients undergoing vascularized reconstruction of the upper aerodigestive tract (UADT)

Now, alternate hypothesis can be directional or non directional. In this study, the researchers had participation from 12 academic medical centers over 11 months with 554 patients and tracked in the type of prophylaxis used (topical and/or antibiotics). They found a decreased risk of SSI with the use of topical and antibiotic prophylaxis.

Most research uses both descriptive and inferential statistics. Published research studies will not usually have an actual hypothesis. Sometimes a research question is posed, but you can always determine the hypothesis from the published work. Inferential statistics come in many, many types: both parametric (based on probability theory) and nonparametric, which is typically used in situations where the assumptions of the parametric test are not met. These include categorical data like dichotomous data which has only two choices. Ordinal data has a ranking within it; nonparametric inferential testing could also be useful.

Now you can usually determine the type of inferential used from a published work, whether parametric or nonparametric by the presence of a p value.

8

Standardized infection rate (SIR)

A summary measure used to track hospital acquired infections (HAIs)

The SIR has a risk adjustment component that allows for comparisons at the national, state, or local level.

The SIR compares to a prediction (the NHSN baseline) taking into account several risk factors

A SIR of > 1.0 indicates more HAI than predicted; a SIR of < 1.0 indicates fewer HAIs than predicted.

Let’s take a look at Hicks, T (2024) from the Walden repository.

Hicks implemented an educational process with a revised CLABSI prevention bundle and designed a pre and postintervention comparison to determine if the outcome (SIR) changed after the bundle was implemented and to determine whether or not the process indicators (like, documentation of dressing change, saline flush, etc.) were improved pre to post. He found a statistically significant improvement in compliance with the bundle and four consecutive quarters with a zero incidence of CLABSI.

9

Atlantic Health System Morristown Medical Center