I've been reflecting lately on how we teach search strategies - particularly the ubiquitous PICO framework that has become the foundation of evidence-based medicine education.
PICO (Population, Intervention, Comparison, Outcome) is practically sacred text in health sciences education. It's how we introduce first-year medical and health sciences students to the concept of systematic searching. It provides a clean, structured approach to formulating clinical questions and translating them into search strategies. But lately, I've been questioning whether this is always the best introduction to the nuanced art of literature searching.
The P and I components of PICO make perfect sense as search concepts. Defining who you're studying (e.g. patients with type 2 diabetes, pregnant women in their third trimester) and what intervention you're investigating (a specific medication, a surgical technique, a behavioral intervention) creates a solid foundation for any clinical search.
The C element is useful when directly applicable, though sometimes separate searches followed by synthesis might yield better results than forcing both interventions into a single query.
But it's the O: Outcome that increasingly troubles me as a search concept. By including specific outcomes in our search strategies, aren't we inherently biasing our results toward studies that emphasised those particular endpoints? What about unexpected findings, secondary outcomes, or adverse effects that might be clinically significant but weren't the primary focus?
Moreover, the hoped-for outcome is often implied by the intervention and population themselves. When we search for antihypertensive medications in patients with elevated blood pressure, the desired outcome (blood pressure reduction) is inherent to the clinical question. Adding explicit outcome parameters might unnecessarily restrict our results.
In those cases where the outcome is already implied, the search is often filled with overly specific yet non-descriptive keywords such as reduc* (reduce/reducing), improv* (improve, improving) etc. What if the author didn't express themselves with the exact arbitrary words we chose to include in such a search?
Here's an example search for a population of adolescents with depression, and the intervention of cognitive behavior therapy:
(adolescents[MeSH] OR teenager*[tiab] OR youth[tiab] OR "young people"[tiab])
AND
(depressi*[tiab] OR "depressive disorder"[MeSH])
AND
("cognitive behavioral therapy"[MeSH] OR CBT[tiab] OR "cognitive therapy"[tiab] OR "cognitive behavi?r therapy"[tiab])
AND
(reduc*[tiab] OR decreas*[tiab] OR improv*[tiab] OR efficacy[tiab] OR effectiveness[tiab] OR outcome*[tiab] OR remission[tiab] OR response[tiab])
Each line makes sense as a self-contained concept except for the last one. The point of this line is what I find hardest to explain to students, and I think it's because I don't quite believe in it myself.
I understand why PICO is taught early. It provides structure, it's easy to remember, and it aligns with the broader evidence-based medicine framework. But I wonder if we're doing students a disservice by embedding potential search biases right from the start of their training.
Advanced searchers know when to selectively apply or omit various PICO elements, particularly outcomes. We recognize that sometimes the best approach is to search broadly on Population and Intervention, then apply outcome considerations during the screening process rather than in the initial search.
Yet we're teaching novices to include all elements as if they're equally valuable in search construction. By the time they learn the nuances and exceptions, many have already internalized practices that might limit their literature review effectiveness.
What if we evolved our teaching approach? Perhaps we could:
This approach would better reflect the practice of experienced searchers while avoiding inadvertently teaching students to create potentially biased searches from the outset.
Don't get me wrong - PICO has tremendous value as a conceptual framework. It helps clinicians articulate clear clinical questions and consider all important elements of those questions. But there's a difference between what makes a good clinical question and what makes an optimal search strategy.
As librarians and educators, perhaps we need to be more explicit about this distinction. We can honor PICO's role in question formulation while teaching more nuanced approaches to search construction - approaches that leave room for discovery, serendipity, and comprehensive evidence retrieval.