Specht, Dawn M. PhD, MSN, RN, APN
Every day, direct care nurses encounter opportunities for education and research. Research, instead of being seen as intimidating, should be viewed as an opportunity to investigate and improve practice. As you deliver care, be willing to wonder, “Is there a better way to perform this intervention?” “Why do we do this?” “Have patient outcomes been improved?” “Will this education affect practice?”
This article examines nursing research and how to apply it to a clinical scenario. Research can be qualitative (using inductive logic to start with a situation, examine a phenomenon, and then attempt to form a general description) or quantitative (using deductive logic to start with a general rule and apply it to a specific situation). Both forms of research are necessary in nursing because nurses deal with the art and science of caring.
Qualitative research attempts to understand the unique characteristics of a phenomenon. You'd use this type of research if you recognized a clinically significant occurrence but aren't sure of the actual variables and or their relationships.1 For example, you notice that when families participate in the care of ICU patients, they seem happy despite their loved one's critical illness.
In this case, you might consider studying the experience of participating in the care of a critically ill loved one. The approach you might use is phenomenology, which describes the lived experience of the situation of interest. If you think the experience is different for people of different cultural backgrounds, you might use ethnography as the qualitative approach to assess the effect of culture on the situation of interest.
Regardless of the qualitative approach, you'd collect data via interviews, surveys, and observations from family members (the sample) who experienced the phenomenon. The sample size would be considered sufficient when themes emerged from data analysis. A common technique used in data analysis is to have interviews typed word-for-word and coded for repeating themes.
Software is available to assist with theme identification and coding. The qualitative approach lets researchers find truth in the experience. Preconceived notions are set aside as the true data emerge from the experience.
Quantitative research starts with the researcher developing a hypothesis before conducting research. The researcher seeks to test the hypothesis, always assuming the null to be true (more on this shortly). Data can be conceptually described and measured before research begins.
The hypothesis is based on theories and previous research and is tested for its applicability to the current population. For instance, a study about the effect of 20 minutes of exercise a day for 90 days in overweight women may have yielded a decrease in mean arterial BP by 10 mm Hg. You hypothesize that if this exercise intervention worked in women, it should also work in men, so you design a study to test this intervention in men. You'll collect measurable data (in this case, BP) before and after the intervention. Your null hypothesis is that exercising 20 minutes per day for 90 days won't change BP in overweight men.
Designing a study
Now let's consider how you'd design a research study for a specific scenario. You've noticed that patients admitted to your ICU from the ED with indwelling urinary catheters are developing catheter-associated urinary tract infections (CAUTI) at a rate that is greater than patients admitted from the OR. You discuss this with the quality council, who confirm your suspicions. But the data for infection rates can't be linked to patient age, race, or even an immunocompromised status.
You may decide to interview the nurses in the ED for their perceptions of the process of urinary catheterization and see what themes emerge. This study would have a qualitative design.
Alternately, if you've reviewed a study that has directly linked poor hand hygiene to CAUTIs, you may conclude that the ED staff needs to improve hand hygiene. You ask to meet with the ED's shared governance committee to discuss hand hygiene, and discover an opportunity to educate ED nurses about the link between poor hand hygiene and CAUTIs. An educational initiative for all staff is proposed, but how will you measure the effectiveness of the education, and its effect on quality patient care? And what should be taught in the educational initiative?
Does this type of educational initiative lend itself to nursing research? A team is formed to investigate this topic, develop education, and deliver the education. The team finds evidence-based practice guidelines related to hand hygiene and CAUTIs, and a slideshow for healthcare workers.2 Because only RNs insert urinary catheters in the ED, an agreement is made to present the slideshow to these nurses.
Educational effectiveness may be evaluated by assessing nurses' knowledge before and after an educational intervention, usually with a written test. One study evaluated the effectiveness of a training program on nurses' knowledge of hand hygiene.3 They found that education significantly increased hand hygiene knowledge and the self-reported performance of hand hygiene.
Hand hygiene is among the evidence-based educational interventions for CAUTI prevention. Other evidence-based interventions include using sterile technique, removing indwelling urinary catheters as soon as possible, using closed drainage systems, and maintaining unobstructed urinary flow.4 The team decides to assess the knowledge of CAUTI prevention before and after the educational intervention for all nurses who consent to be involved in a study. The knowledge assessment will be one way to assess whether the short-term goal of increasing knowledge has been met. (The study's hypothesis is that the educational intervention will improve nurses' knowledge of CAUTI prevention.)
This design is referred to as a pretest and posttest intervention. The next question for the committee to decide is if everyone will complete the education and then be posttested or if one group will serve as a control (and won't receive the education). The control group will, however, complete both the pretest and the posttest. Some researchers will have the control group complete an alternative assignment like a stress reduction lecture while the experimental group receives the true education.
The team decides that everyone will receive the education because there's no feasible way to prevent the other nurses from finding out the education from their peers. The tests will be scored to produce a numerical value and to examine changes in scores from pretest to posttest. The statistical method used to assess the data depends in part on the type of data; in this case, ratio data represented by numbers.
The research team will discuss the intent of any intervention, the process for the intervention, the assessment of subjects, the time frame, and the type of statistical analysis. All staff will take the pretest before the educational sessions, which will be taught by the same person, in the same room, at various times of the day and week. After each educational session, staff will take the posttest.
The posttest should be administered immediately, to prevent extrinsic factors (such as nurses discussing the educational intervention) from affecting the test results. Study replication is the true long-term evaluation of the results; the strongest studies are those that can be easily replicated and yield similar results.
The educational intervention will be a lecture available in the CDC's CAUTI toolbox. The team develops a 10-item knowledge test to use as a pretest and posttest.
The knowledge test is evaluated by outside content experts, who find that it has good content validity. Using a newly developed knowledge test might limit the generalizability of the study, but because no standardized test was found in the literature, one had to be developed. This fact will need to be included when the team shares the data with their peers.
The committee has decided to use a quantitative pretest posttest design, and will be able to compare scores from before and after the educational intervention to see if they change based on the educational intervention. The statistic of interest will compare mean score on the pretest with mean score on the posttest. To protect nurses' anonymity but still be able to match pretest and posttest scores, each nurse is assigned a coding number.
The dependent variable (what we expect to change based on the intervention) in this design is test scores. Nurses' knowledge about CAUTI before the educational intervention should correlate with knowledge after the intervention. In other words, nurses who are well educated on CAUTI should have higher pretest and posttest scores. Nurses who aren't well educated on CAUTI should have better posttest scores than pretest scores. High scores before the educational intervention can affect mean changes and make the data appear statistically insignificant (as in the case of a nurse who scores a 90 on the pretest and 94 on the posttest). The larger the improvement in scores (for example, a nurse who scores 70 on the pretest and 90 on the posttest), the more likely statistical significance will be found. The best educational interventions should help all subjects improve regardless of their knowledge level before the intervention.
The paired t-test statistical analysis is used when one group is tested twice; this method takes into account that the chance difference between matched subjects (in this case, a pretest and posttest by the same nurse) won't be as great as the difference between unrelated subjects (for example, pretests and posttests from different nurses).5
Once the type of statistical analysis is determined, the number of nurses required for the study may be set, using the t statistic chart found in many statistical manuals.
Testing the test
The knowledge test is administered to a group of ED nurses from another facility (and therefore not involved in the study), and is found to have an internal consistency reliability (or alpha) of 0.80. Internal consistency reliability evaluates each test question in relationship to the other test questions; for example, if nurses who know the answer to question 1 should also know the answer to question 2. To be considered valuable in an experiment, a test must have a reliability score of 0.70 or higher.
If we accept that the null hypothesis is false (that is, that the educational intervention won't improve nurses' knowledge of CAUTI prevention), the risk of harm to the study subjects (nurses) is low, so the level of significance is set at 0.05. The t statistic is paired, and we expect scores to improve with education (one direction of change), so we'll use a one-tailed design. (A two-tailed design would be used if the researcher expects scores to go up and down, or in two directions.)
After all the educational sessions have been completed and pretest and posttest scores tabulated, the researchers enter the scores into a computer program. The t-test compares the mean of each scoring group (pretest and posttest) and evaluates it for mathematical change based on each pairing. The analysis ultimately evaluates whether the change is due to chance or the educational intervention; a P value of 0.05 means that the results are statistically significant, and the scores changed because of the educational intervention, not chance.
Properly planning the data analysis and statistical techniques before the study can improve the study's strength, or power, and the probability of rejecting a false null hypothesis. Through well-planned nursing research, you can obtain findings that can help you and your facility improve nursing practice.
© 2013 Lippincott Williams & Wilkins, Inc.