A Comparative Analysis of Hospital Discharge Instruction Methods: Effects on Patient Comprehension and Associations with 30-Day Readmission Rates
Advisor: Farhan Ahmad
Internship Location: Inova
Onsite Mentor: Lauren Landwermeyer
Hospital readmissions and poor patient comprehension of discharge instructions remain persistent healthcare challenges, driven in part by low health literacy and ineffective communication. Studies estimate that roughly one-third of Americans struggle to understand basic medical information, and preventable readmissions cost billions of dollars annually. This project investigates which discharge instruction methods (e.g. teach-back, simplified written materials, multimedia tools, or follow-up calls) most effectively increase patient comprehension and if it leads to reduced 30-day readmission rates. My on-site experience will take place at INOVA Alexandria Hospital, a setting offering direct exposure to real interactions during discharge.
Project Blog
Post 1: Introduction
For my senior project, I am choosing to do an internship/volunteer at INOVA Alexandria Hospital, working in the patient transport department, which includes transporting discharged patients and working in the discharge lounge. Along with this on-site experience, I am aiming to research what methods may improve discharge instruction quality and see if this will perhaps be correlated with a decrease in readmission rates. I aim to do this through a meta analysis of all the different methods and their associated increase in patient comprehension, and then continue to analyze the data to see if that also translates into 30 day readmission rates. What this actually means is I’m attempting to find the most effective discharge instruction method and also if this actually improves patient outcomes by looking at preventable readmissions.
I chose to research this topic because as a volunteer at INOVA, I saw first hand the discharge processe and how messy it tends to be. Patients don’t always seem to be prepared for discharge mentally, or they have misunderstandings with providers when instructions for care at home are explained. This is highly understandable as the transition between hospital care to home contains many variables that add up. Not only must a patient understand all instructions for proper recovery at home, but they must also coordinate additional factors such as transportation home while often in residual pain. My experience at INOVA has exposed me to all the different factors that can add stress and tension to a patient’s discharge process, which led me to my research. Additionally, as someone who has a parent where English is not a first language, I’ve seen firsthand numerous examples of language barriers causing misunderstandings during discharge that could impact quality and efficiency of care at home.
Week 1: Preparation
For the first week of my senior project, I focused on setting up and preparing for my research. I’ve been reviewing a meta analysis video provided by Dr. Hight to familiarize myself with the process more and to further add to my schedule the finer details of completing one. As I was advised in the meta analysis video, I’ve set up my documentation journal where I will note down all decisions I made, why I made it, and the date. This will not only help with transparency about methodology but also help me remember what and why I did things. I then began to delve more deeply into the PRISMA guidelines to solidify my understanding of how I needed to tackle this process. It was extremely intimidating how much details and information required for the flowchart and checklist, but I believe if I stay committed to keeping a documentation journal, it will be very helpful in completing these deliverables.
To begin my actual research, I’m planning on using PubMed (Large database for medical research), Google Scholar (allow for extremely broad searches), CINAHL (nursing journal that is relevant due to role nurses have in discharge), and JAMA (large medical journal). Since I’ve allotted a large chunk of time to finding and screening studies, I may expand further if time allows. I have also been working this week on finalizing search parameters which includes defining concepts. For example, since I hope to compare interventions, I must define/include alternate wordings that relevant studies may use, such as “teach-back method” or “patient education.” I’ve also begun working on inclusion and exclusion criteria which I hope to finish soon that will allow me to screen studies and must be noted in the PRISMA flow chart. Once these have been decided, I will begin the large bulk of my work of reading through and screening studies to include/exclude.
Week 2: Searching for studies
For the second week of my senior project, I began applying the search parameters I had begun developing last week to begin identifying studies for my meta-analysis. I started with PubMed, since it is, as I mentioned last week, one of the largest databases for medical research. Using the concepts I had defined, I employed a search strategy that followed the format of “discharge” (plus synonyms) AND “intervention” (plus synonyms) AND “readmission” (plus synonyms). The use of AND was to prevent casting a too broad net as it requires the resulting studies to contain all three words. Synonyms of each of the three required words/categories were separated by OR since different studies may use each interchangeably or with negligible difference in definition that I will have to further identify during a deeper read later.
While building the search, I noticed that some of the more specific phrases I initially considered were not found in the PubMed phrase index. To avoid unnecessarily limiting my results and potentially losing exposure to relevant studies, I simplified some terms to make the search broader and more inclusive. For example, the phrase “simplified written instructions” was adjusted to the more general term “written instructions.” Making these adjustments helped ensure that relevant studies using slightly different terminology would still appear in the search results.
After finalizing the search terms, the initial search on PubMed produced 320 results. To narrow these results to higher-quality and more relevant studies, I applied article type filters. Specifically, I limited the results to randomized controlled trials (RCTs), clinical trials, and systematic reviews, as these study types tend to be more robust and reliable and are often used in meta-analyses. I also set the publication date to the past fifteen years (2011–2026) to ensure the research reflects more current healthcare practices. Applying these filters reduced the results to 70 studies.
From there, I conducted a preliminary title screening to determine whether each study was broadly relevant to my research question. During this stage, I removed studies that clearly fell outside the scope of my project, such as pediatric-focused research, since I am aiming to focus on adult patients. This screening eliminated 17 studies, leaving a smaller and more manageable group for deeper abstract review in the coming weeks.
Week 3: Abstract Review
For the third week of my senior project, I began a deeper level of screening by beginning to read the abstracts of the studies I gathered from PubMed. As I read through the abstracts, I started creating tags to help organize the studies based on key characteristics. These tags included the type of intervention used (such as teach-back or written instructions), the patient population, and the specific hospital department or treatment context. Creating this tagging system has been helpful in identifying patterns across studies and will likely make data extraction and comparison easier later on as I can group studies accordingly. At the same time, I removed studies that I felt were not in the scope of my project.
In addition to screening PubMed results, I expanded my search to other databases. I used Google Scholar with the same search query I made earlier. Since Google Scholar offers fewer filtering options, I limited my review to the top 50 results sorted by relevance, which still provided a broad range of potentially useful studies. Based on this experience, I decided not to continue searching CINAHL due to subscription barriers, especially since Google Scholar already casts a broad net. Similarly, I am working to finish reviewing studies from JAMA and incorporating any relevant ones into my master document. However, I believe more will be added as I sift through the studies more deeply which may point to more relevant studies that I can use.
I will admit that I was not very productive this week, and I do feel slightly behind my original schedule. However, I have a clear plan moving forward. My goal is to complete a more thorough abstract review of any added documents to further narrow down my sources as well. If I have time, I hope to additionally do another search on PubMed with more specific search queries to address any issues that could arise from using a longer one, as Mr. Ahmad suggested. Looking ahead to next week, I plan to begin full-text readings of selected studies in order to start preparing my Introduction, Literature Review, and Methods sections for submission by the March 27th due date.
Week 4: Refining Screening Criteria
During Week 4, after completing an initial abstract screening, I wanted to refine the inclusion criteria to better align with my central research question. Through my deeper reading with the literature, it became clear that there is broad consensus regarding the positive relationship between discharge education interventions and patient comprehension. I had originally wanted to address that as one of the products of my research, but now believe it would become redundant to do so. As a result, I focused my research only to determine which specific intervention types were most strongly associated with reductions in hospital readmission. I believe that during my initial abstract screening, I had included too many that would have not given me the necessary information.
Accordingly, studies were excluded if they did not report outcomes related to readmission. Earlier screening stages had been overly liberal, allowing inclusion of studies that described discharge interventions but did not measure readmission in either the abstract or reported results, or alternatively, mentioned the use of multiple interventions at the same time, which could make finding the effect of solely one intervention difficult. The criteria were therefore made more stringent to ensure that all studies would be relevant to my focus.
To further improve the relevance and completeness of the studies, I also did backward citation searching using prior systematic reviews and meta-analyses I had eliminated in a title review (due to my decision not to include these study types) with research aims closely aligned to this project. I looked at reference lists from these reviews to identify additional primary studies that may not have been noted in the initial database searches, with them undergoing the same screening processes.
Week 5: Article Review
For the fifth week of my senior project, my progress was more limited than usual since I was out of town for college visits. Although I was not able to dedicate as much time to the active research, I did take steps to keep my project organized and moving forward. At this point, my pool of studies is more or less finalized, and I have sorted all originally included articles into a single document. Alongside each study, I noted my rationale for why it was included or excluded. Keeping this level of documentation is important for transparency and ensures that my methodology could be replicated by someone else following the same process.
As I finish full-text readings of these studies, I believe that I may still exclude a few that, upon deeper review, do not fully align with my research question. During this stage, I have also encountered an unexpected challenge. While attempting to access several full articles, I found that a few were blocked behind subscription paywalls, which I expected. I checked for access through the public library system, but unfortunately, these articles were not available there either, which surprised me. This has introduced an additional task for me, which is to search for other methods of accessing these studies, such as even contacting the author themselves. If I am ultimately unable to obtain certain articles, this may become a limitation of my project that I will need to acknowledge in my final paper.
This weekend, I hope to finish constructing my data extraction table. I will design it to include all relevant variables from the studies (intervention type for example) and incorporate the necessary formulas that will allow me to record and then compare data across articles. Setting up this table will prepare me to begin the complex data extraction process more easily.
Week 6: Full Text Read and Data Extraction cont.
For this week, I continued to focus on doing full text reads and coding studies, or extracting the relevant data. During my full text reads, I continued screening out studies that I deemed not in the scope of my question or containing the necessary data I needed for my analysis. For example, one of the studies mentioned that the intervention group had seen lower rates of readmission, and since the standard definition of readmission rates in most studies seems to be measured at 30 days, I included it past the abstract screening despite noting that outcomes were assessed at 3 and 6 months after the trial began. I had hoped that the study might continue or have measured readmission rates at the 30 day mark, but after doing a full text read, I had to exclude it since it did not include the information I needed. Another significant reason is that quite a few studies include or group more than one intervention in their experiment. Instead of just simply a visual discharge method as the intervention, they may also include follow up calls, which I classify as a different intervention that I would like to measure the effects of. Since there is no way to differentiate the impacts of the two, I cannot include the study.
The data extraction is also much more difficult and tedious than I suspected, even after warnings from Dr. Hight and Mr. Ahmad. I feel as if it is best practice to also record explanatory variables as well for potential later uses. This means including characteristics of the study group such as location, age, and disease type. This could allow me later on to investigate if there is a significant relationship between patient type and readmissions beyond discharge methods. For example, perhaps older patients prefer or are more comfortable with direct human interaction and so the use of a teach back method could potentially be much more effective than the use of electronic/video formats, and vice versa for younger patients. While I have not fleshed out exactly how this may work, I want to be prepared, and so I am trying my best to note all this down, but it requires significantly more reading and thought for each study.
Week 7: Data Extraction and a New Discharge Method
This week, I continued my full reading of the studies and worked on data extraction. At the same time, I continued adding to my final paper, especially expanding my literature review and explaining the importance of discharge education in more detail as I continue to come across new, important information in my full read.
For example, I added information about a discharge method I was not as familiar with: pharmacist counseling. I learned that older adults transitioning from hospitals to community care are at especially high risk of adverse drug events (ADEs), which can account for up to 70% of post-discharge complications (Pevnick et al.) These patients often have very complex medication regimens and fewer resources and/or support to manage them safely, which makes pharmacist involvement during discharge another important area to consider.
However, this led to a challenge in my study selection. Some of the pharmacist counseling studies do not evaluate pharmacist education as a “single” intervention. Instead, they sometimes utilize a broader care model where the pharmacist also uses methods like teach-back or follow-up phone calls. This makes it difficult to categorize the study under just one discharge method, since it overlaps with other intervention types I want to analyze. I am now working through how to decide whether to include these studies and how to classify them in a way that remains consistent with my research question and methodology. My thinking is that I will aim to include most given that the focus of the intervention is still on drug management and the prevention of ADEs.
Week 8: Data Extraction, Risk of Bias, and Final Product
This week, I continued working through full-text readings of my selected studies while actively extracting data into my table. I additionally am attempting to figure out the risk of bias tool Cochrane. This requires a lot of reading and understanding of study design that further increases the time and attention each study requires.
At the same time, I made additions to my final paper. Since I have come further along in my project, I could add more onto my challenges and limitations. For example, I reflected on how conducting a meta-analysis alone may have introduced bias, since standard practice involves multiple reviewers independently screening studies and resolving disagreements. Without a second, independent perspective, there is a greater possibility that some studies were excluded or included based on my own interpretation alone.
I also added discussion of time constraints and how they limited the number of studies I was able to fully review. Because full-text reading and data extraction is time-intensive, I cannot realistically analyze every potentially relevant study in depth, which may have resulted in missing data. Additionally, I noted that my developing experience with meta-analysis and healthcare research may have negatively impacted my ability to apply inclusion criteria consistently, especially when studies assessed complex intervention models such as the pharmacist care models I discussed last week that were difficult to categorize.