Figuring out your study design

• Where do I start?

• Literature search - search engine links and strategy

• Case report

• Retrospective

• Prospective 

• Deeper dive - see article

• How do I fulfill my scholarly activity requirement?

Where do I start?

While there is no single approach to coming up with study ideas, a good place to start is with something that tends to bother you on shift. Practice variability and inconsistency, inefficiencies. Ask why does one attending do it this way and another does it totally differently? What does the evidence say about the standard of care? If there is not enough evidence, why not? What are some barriers to studying this? Do we have access to these kinds of patients in the ED and is there a study we can do to help shed light on a better way to manage these patients? Or are there a bunch of small studies without a clear consensus and an opportunity for either a systematic review or a meta-analysis? Or is little known about the topic and a more narrative approach would be useful?

Once you have a question, or a direction, having a framework can help to flesh it out and turn it into a study, or a paper. PICO is a common one (Population, Intervention, Comparison, Outcome), but there are others. Check out this paper for more.

If you are not the creative type, or don't have bandwidth to be creative because trying to survive residency all you can do for the foreseeable future, ask around what other residents are doing, ask your faculty what they're working on, see if there is a project that you like where you can contribute. I cannot underscore how helpful it is if you already have your CITI and IRBNet training complete. This is the biggest barrier to hopping on an already existing study. Please get this done ASAP!

Performing a literature search - helpful links

• Open evidence: An AI powered search engine specifically for clinical trials via NEJM & JAMA. Provides references with links for every search

• CSH/Dignity Library: access to most journal articles you'll need.

• Helen Chang (helen.chang@commonspirit.org): Our amazing librarian. Can get you any article you need even is our library doesn't have it.

Performing a literature search - strategy and tips

Articulating a clinical question can be challenging and the question you will start out with will likely not be the one you choose to eventually pursue. Your initial goal may be to get a sense of how much is known about a specific topic. 

As you explore pay attention to whether or not that knowledge is pertinent to the emergency physician. For example, lots is known about hyponatremia, but most of the management centers on things like serum osmolality and urine electrolytes not readily available to the emergency physician. Also of importance is whether that topic has been studied in patients that you see in Stockton. For example, a study of mostly white male VA patients may not be relevant and that topic may deserve more scrutiny in a more diverse population such as ours.

If you have a general topic in mind, (something like ekg findings in patients with occlusion MI), a review article is a great place to start. As you read, ask yourself how have the authors come to their conclusions, especially if those conclusions are surprising to you. Take a look at their references. Are the studies they're citing robust multi-center trials, or are they single center retrospective trials? Or is it based on in vitro studies or data from clinical trials several decades old that may no longer be applicable to our current practice?

You may have a more specific question such as after a challenging case. Suppose you missed a patient with appendicitis with normal labs and a benign exam, and you wonder if you could have done something differently? Are other clinicians also missing these patients? Your initial search might be on the incidence of patients with acute appendicitis who present to the ED with normal labs and a benign exam. Suppose you find that this is not uncommon but not much more is written on the topic. Your next question might be, what are the characteristics of these patients? Are they patients who are unreliable because of cognitive impairment, very elderly or very young, or something else? What are their clinical outcomes? That could be a study. That could be a narrative review. That could be a publishable paper. (see publication guidance)

Case ReportS

If you come across an unusual case, you may have an opportunity to write it up as a case report. A case report would satisfy your scholarly activity requirement. It may not always be intuitive which cases may qualify, and frequently a literature search is first needed to determine how unique or unusual your case is. Of note, a consent will usually be required for a case report. if you are not sure whether or not the case meets case report criteria, get the consent from the patient or their proxy while on shift! They are in the north doc box on the wall, with large letters S-A-E-M on top. For cases requiring photographs of the patient, a separate consent form may be needed. For more on case reports, check out this article. 

A case report may serve various purposes in its contribution to the overall body of medical knowledge, from novel therapeutic approaches to atypical complications. The Lisfranc joint, for example, is named after Dr Lisfranc who published a case report on a novel technique of foot amputation.

A case series if a group of similar cases with something in common. These typically focus on rare illnesses or rare complications for which a study is not practical due to very small number of patients.

A word of caution on case reports!

For a list of journals that publish case reports, check out this database from OSF

For a paper on how to - check this out

Retrospective

Many questions can be answered with a retrospective design. Retrospective just means that the events you are interested in studying already happened. 

The benefits is that these studies are typically easier to get approval from IRB and generally do not require informed consent and so need a lot less research infrastructure to complete. The downside is that the data you will be working with may be incomplete or wrong, and depending on what you're studying, may require more statistical know-how in order to get meaningful results. 

There are times when retrospective design is optimal - for example many ekg studies use retrospective design since it is helpful to select patients with specific known outcomes, like STEMI or occlusion MI, etc. Similarly, many toxicology or exposure studies have to be retrospective for the same reason.

Overall, this is an excellent choice for residents since many of these are able to be completed during one's tenure in residency. 

Get Dr Bhattacharyya involved early. She is a whiz with statistics and has extensive research experience and can help prevent you from making critical errors early in your design process!

For a how-to on retrospective chart reviews, check out this article

Prospective

Prospective is generally considered the design to get the most definitive answer to a clinical question. With few exceptions, these will require informed consent. There are minimal risk studies that are mostly observational, like our sensory stroke study. These studies are mostly done to collect data on things that would happen with or without the study, and do not influence practitioners' clinical decision making. 

Prospective controlled studies, whether randomized or not, will require informed consent and a greater amount of resources, but are also more likely to yield more precise data than retrospective or purely observational studies.

The more control of a clinical scenario that will be required in order to provide a satisfactory answer, the more ethical considerations are going to come into play. If the question you are looking to answer has had multiple retrospective trials already performed looking into it, and the safety of the interventions has been well-examined and well-established, then this may be the logical next step. However, if this is not the case, and one of the control groups would need to have an intervention whose safety is less than clear, then more work may need to be done before that trial could be ethically justified.

For a deeper dive into various methodologies, click here.