NURS 500 Week 6 Research Proposal Part 4, Measurement, Analysis and Outcomes Example

Caregiver Teach-Back Before Discharge and Tube-Related Emergency Visits After a New Gastrostomy, Part 4: Measurement, Analysis, and Outcomes

The final part of the proposal explains how the variables defined in Part 1 will be measured, how the data collected under the procedures in Part 3 will be analyzed to answer the three research questions, and what outcomes are expected.

Measurement

Levels of Measurement

Each variable's level of measurement determines the tests that can be used. Education type, the independent variable, has two unordered categories, usual teaching and teach-back, and is therefore nominal. The main outcome, whether the patient returned to an emergency department for a tube problem in the first month, is also nominal, with only two possible values. Caregiver preparedness is measured on a summed Likert scale whose mean score is treated as interval data, a common and defensible practice when a scale has several items and a roughly symmetric distribution (Gray & Grove, 2021). Covariates include patient age, caregiver age and distance from home to the hospital in miles, all ratio level; diagnosis category and caregiver relationship, both nominal; and caregiver health literacy, which is ordinal.

Preparedness for Caregiving Scale

For preparedness the study uses Archbold's eight-question instrument (Archbold et al., 1990), the same scale named in the Part 1 definitions. Caregivers rate how well prepared they feel for areas such as meeting the person's physical needs, handling emergencies, managing the stress of caregiving and finding information and services, on a scale from 0 (not at all prepared) to 4 (very well prepared). The score is the mean of the eight items, so it ranges from 0 to 4, with higher scores indicating greater preparedness. The scale was chosen because it measures the caregiver's own sense of readiness, which is the quality teach-back is designed to change and the quality a booklet alone does not test. Remr (2026) reported a Cronbach's alpha of .944 and a single-factor structure explaining 71.9% of the variance in a population-based sample of 1,024 Czech adults, and earlier work tied the scale to lower caregiver role strain (Archbold et al., 1990). Because those figures come from different populations, internal consistency will be calculated again in this sample and reported.

Tube-Related Emergency Visit

The primary outcome will be coded 1 if the patient had at least one emergency visit within 30 days of discharge meeting the operational definition from Part 1 and 0 otherwise. Classification depends on judgment, so agreement between the two blinded nurse reviewers will be measured with Cohen's kappa on the first 40 visit records. McHugh (2012) cautions that kappa values in the range traditionally called moderate may be too low for health research, so the reviewers will refine the coding guide and retrain until kappa reaches at least .80 before classifying the remaining records.

Covariates and Fidelity

Caregiver health literacy will be screened with a single question developed by Chew et al. (2004), "How confident are you filling out medical forms by yourself?", answered on a five-point scale. Diagnosis category will be grouped as stroke, head and neck cancer, progressive neurologic disease or other. In the intervention cohort, fidelity will be summarized as the percentage of the six skills for which an accurate return demonstration was documented; caregivers with fewer than five of six skills documented will still be analyzed in their assigned group.

Data Analysis

Data will be analyzed in IBM SPSS Statistics with a two-tailed alpha of .05, matching the power analysis in Part 3. Descriptive statistics will summarize both cohorts: frequencies and percentages for nominal variables, means and standard deviations for normally distributed interval and ratio variables, and medians with interquartile ranges otherwise. The cohorts will be compared at baseline with chi-square tests and independent-samples t tests to describe how far they differ before any outcome is examined.

For the first research question, a chi-square test of independence will set the share of usual-care patients who came back with a tube problem against the share of teach-back patients who did; should any cell in the expected table hold fewer than five cases, Fisher's exact test will replace it. Since the two cohorts were not formed at random, the main analysis will be a logistic regression with group as the predictor and patient age, diagnosis category, caregiver health literacy and distance from the hospital as covariates, with the result expressed as an odds ratio, adjusted for those four covariates, and its 95% confidence interval. For the second question, an independent-samples t test will compare mean preparedness scores after the normality and equal-variance assumptions are checked with the Shapiro-Wilk and Levene tests, and Cohen's d will describe the size of any difference. For the third question, a logistic regression will estimate the odds of a tube-related visit for each one-point increase in preparedness across both cohorts.

Missing data are expected mainly at the day-30 call. Because emergency visits will also be found through the hospital record and the regional health information exchange, the primary outcome can be coded for most dyads even when the call fails. The primary analysis will use all dyads with an outcome; a sensitivity analysis will count every dyad lost to follow-up as having had a visit, the least favorable assumption for the intervention.

Planned Sensitivity and Subgroup Checks

Two further checks will test how sturdy the main result is. First, the logistic regression for the first research question will be repeated with diagnosis category entered as an interaction term, because caregivers of patients after stroke may face different problems from those caring for patients after head and neck cancer, whose tubes are often used alongside some oral intake. Second, the analysis will be repeated after excluding the first two months of the intervention cohort, when nurses were still gaining fluency with the protocol; if the effect grows once those months are removed, that pattern would suggest the protocol takes time to deliver well, which matters for any hospital planning to adopt it. These checks are exploratory, will be labeled as such and will not be used to claim support for the hypotheses if the main analysis does not.

Expected Outcomes

The study is expected to show a lower proportion of tube-related emergency visits and higher preparedness in the teach-back cohort. If both hypotheses are supported and the adjusted odds ratio remains significant, the hospital would have local evidence to make teach-back with return demonstration the standard for all gastrostomy discharges, at little cost beyond nurse training time. If preparedness rises but emergency visits do not fall, the finding would echo the review by Amaratunga et al. (2023), in which better education did not reduce unplanned encounters, and would point toward after-discharge support such as a nurse telephone line as the next intervention to test. A null result on both questions would still be useful, since it would argue against spending training time on teach-back for this problem.

Five limitations are expected. The sequential cohorts cannot rule out changes over time; the study is limited to one hospital and to caregivers who speak English or Spanish; emergency visits outside the regional exchange may be missed when the call fails; preparedness is self-reported; and nurses may teach the usual-care cohort differently because they know a study is planned. Findings will be shared with the three units and the hospital's nursing practice council, presented at a state nursing conference and submitted to a peer-reviewed nutrition or nursing journal so that other rural hospitals can judge whether to adopt the protocol.

References

Amaratunga, H., Bostock, K., Cunich, M., Steffens, D., & Carey, S. (2023). Systematic review of service improvements for home enteral tube feeding in adults. Nutrition in Clinical Practice, 38(2), 329-339. https://doi.org/10.1002/ncp.10900

Archbold, P. G., Stewart, B. J., Greenlick, M. R., & Harvath, T. (1990). Mutuality and preparedness as predictors of caregiver role strain. Research in Nursing & Health, 13(6), 375-384. https://doi.org/10.1002/nur.4770130605

Chew, L. D., Bradley, K. A., & Boyko, E. J. (2004). Brief questions to identify patients with inadequate health literacy. Family Medicine, 36(8), 588-594.

Gray, J. R., & Grove, S. K. (2021). Burns and Grove's the practice of nursing research: Appraisal, synthesis, and generation of evidence (9th ed.). Elsevier.

McHugh, M. L. (2012). Interrater reliability: The kappa statistic. Biochemia Medica, 22(3), 276-282. https://doi.org/10.11613/BM.2012.031

Remr, J. (2026). Psychometric validation of the Caregiver Preparedness Scale in a population-based sample. Nursing Reports, 16(4), Article 115. https://doi.org/10.3390/nursrep16040115