Asking Compelling Clinical Questions

This is step one, and it is the step everything downstream depends on. A question built badly cannot be searched well, and no amount of careful appraisal recovers from it. The core idea is small and worth holding onto now: know the needle before you go into the haystack. Most of the time people report losing to searching is time spent searching in order to work out what they were looking for. By the end of this half hour you can tell a background question from a foreground one and say where each sends you; build a question in PICOT format and say when to leave an element out; name the five question types and the designs that answer them; separate a clinical question from a research question and a quality improvement question; and say why a good question can still come back empty.

Three videos on the origin of a clinical question, which is almost never curiosity in the abstract. It is your own outcome data not looking the way you expected.

Where a question starts.

EBP 0151 Questions Come From Inquiry

Questions come out of a spirit of inquiry, which is step zero. Without the habit of asking why, there is no question to format.

What usually provokes one.

EBP 0152 Outcome Data Start the Question

Most clinical questions start from outcome data — something in your own numbers that does not look right.

Why the data have to come first.

EBP 0153 Data Set the Expectation

The data establish what you expected. Without that, there is nothing for the finding to be surprising against.

Five videos on what the question is for. Keep the needle-and-haystack image, because it is the one that explains why people lose whole afternoons to searching.

What the question is about.

EBP 0154 What the Question Describes

The question describes a clinical situation and the decision inside it, not the project you might run afterwards.

The single criterion.

EBP 0155 Ask It So It Can Be Answered

A question has to be askable of a database. That is the criterion, and it rules out most of the questions people start with.

The image worth keeping.

EBP 0156 Know the Needle Before the Haystack

Know precisely what you are looking for before you start searching. Searching to find out what you want is how people lose afternoons.

What the question does next.

EBP 0157 The Question Guides the Search

The question determines the terms, the databases and the designs worth looking for. Unit 3 is the whole of that.

What a good question buys.

EBP 0158 Time Saved, Rigor Preserved

A well-built question saves time and preserves rigor at once. It is not a trade between speed and care.

Five videos on the distinction that decides where you look for the answer — a reference work, or a body of research.

The first distinction.

EBP 0159 Two Kinds of Clinical Question

Clinical questions come in two kinds, and knowing which you are asking decides where you go for the answer.

The first kind.

EBP 0160 The Background Question

A background question asks for general knowledge — what a condition is, how a drug works. A textbook answers it.

The second kind, and the one this unit is about.

EBP 0161 The Foreground Question

A foreground question compares specific courses of action for a specific patient. This is the kind PICOT is built for.

Where a foreground question takes you.

EBP 0162 Foreground Questions Lead to a Body of Evidence

Foreground questions lead to a body of research evidence. Background questions lead to a reference work.

Who asks which, and why.

EBP 0163 Novices Ask Background Questions

Novices ask more background questions and experts more foreground ones. That is about accumulated knowledge, not about ability.

Three videos on the format, including the error people make within about a minute of learning it.

The format.

EBP 0164 PICOT

Population, Intervention, Comparison, Outcome, Time. Five elements, and the acronym is the whole apparatus for step one.

Why a format rather than a sentence.

EBP 0165 The Purpose of the Format

The format forces you to name things a database can match. It is a discipline rather than a form.

The error people make first.

EBP 0166 A PICOT Question Is Not a Project

A PICOT question names a comparison, not a plan. Who will run it, what it costs and whether the unit will adopt it come after the answer.

Seven videos, one per letter and two on when the format bends. Watch the C, which is the element people leave vague and the one that causes trouble two units later.

The first element.

EBP 0167 P — Population

Who the question is about, specified tightly enough to matter and loosely enough to find studies.

The second.

EBP 0168 I — Intervention or Issue of Interest

What you are considering doing, or the issue of interest. Not every question involves an intervention.

The third, and the one most often left vague.

EBP 0169 C — Comparison of Interest

What you are comparing against, which is often current practice. A vague comparison is why two studies can appear to disagree when they do not.

The fourth.

EBP 0170 O — Outcome

The outcome, named as something measurable. Better care is not an outcome a database can match.

The fifth.

EBP 0171 T — Time

The time frame over which the outcome is observed.

Where the format bends.

EBP 0172 Not Every Question Needs Every Element

Some questions have no meaningful comparison or no time element. Forcing all five produces a worse question.

When the T earns its place.

EBP 0173 When Time Is Part of the Question

Time belongs in the question when the outcome depends on the period — mortality at thirty days is a different question from mortality at five years.

Eight videos. The type of question determines which designs can answer it, which is why this is settled before any searching happens.

Five types, and why the type matters.

EBP 0174 Five Types of Clinical Question

Clinical questions come in five types, and the type determines which study designs can answer it.

The first type.

EBP 0175 Intervention Questions

Does this treatment work. Answered best by trials and syntheses of trials.

The second.

EBP 0176 Prognosis or Prediction Questions

What is likely to happen over time. Answered by cohort studies following people forward.

The third.

EBP 0177 Diagnosis Questions

Whether a test identifies the condition accurately. A different design question again.

The fourth.

EBP 0178 Etiology Questions

What causes or contributes to a condition. Often answered by case-control work, looking backward.

The fifth, and the one people forget.

EBP 0179 Meaning Questions

What an experience is like for the people living it. Answered by qualitative designs, and unit 7 is entirely this.

There are templates.

EBP 0180 Templates for Each Type

Each question type has a template you can fill. Using one is not a shortcut, it is how you avoid omitting an element.

How long a question should be.

EBP 0181 Too Few Words, Too Many Words

Too few words and the search returns everything; too many and it returns nothing. The yield tells you which way to move.

Five videos on a wider distinction. This one returns in unit 10 and again in unit 22, and getting it wrong sends you to the wrong method entirely.

A wider distinction, and a load-bearing one.

EBP 0182 Three Kinds of Question in Healthcare

Clinical, research and quality improvement questions. Three kinds, and mixing them up sends you to the wrong method entirely.

What makes a question clinical.

EBP 0183 A Clinical Question Has No Direction

A clinical question is genuinely open. It does not presume which arm wins, and that openness is what lets the evidence answer it.

What makes it research.

EBP 0184 A Research Question Has Direction

A research question carries a hypothesis and therefore a direction. That is proper for research and wrong for a clinical question.

What makes it quality improvement.

EBP 0185 A Quality Improvement Question Is Local

A quality improvement question asks about this setting. It does not seek to generalise, which is a feature.

Why the type has to be settled first.

EBP 0186 Getting the Type Right Early

Getting the type wrong sends you to the wrong designs, the wrong databases and the wrong appraisal criteria. It is cheap to fix now and expensive later.

Three videos on what people believe about PICOT that is not true. The third is the honest one.

The first of three errors.

EBP 0187 Misconception: PICOT Is Fill-in-the-Blank

PICOT is not a form. It is a discipline for making a question answerable, and filling boxes without that discipline produces an unanswerable question in the right shape.

The second.

EBP 0188 Misconception: Every Question Needs All Five Letters

Not every question has a comparison or a time frame. The format serves the question, not the other way round.

The third, and the honest one.

EBP 0189 Misconception: A Good Question Guarantees an Answer

A well-built question can return nothing, because the evidence may not exist. That is a finding rather than a failure of the question.

One procedure, which takes a question from an outcome that looked wrong to a sentence a database can match.

The unit assembled into one worked question.

EBP 0190 Building a PICOT Question

One question built from an outcome that looked wrong, through type and elements, to a sentence a database can match.

That is step one. Against the outcomes. A background question asks for general knowledge and a reference work answers it. A foreground question compares courses of action for a specific patient, and it leads to a body of research. Which kind you are asking decides where you go. You can build in PICOT: population, intervention, comparison, outcome, time. And you know when to leave one out, because not every question has a comparison or a time frame, and forcing all five produces a worse question. Five types, and the type sets the design. Intervention questions go to trials. Prognosis goes to cohorts. Diagnosis asks about accuracy. Etiology often goes to case-control. And meaning questions go to qualitative work, which unit 7 takes up in full. Three kinds of question in healthcare: clinical, which is genuinely open; research, which carries a hypothesis and therefore a direction; and quality improvement, which is local by design. And a well-built question can return nothing at all, because the evidence may not exist. That is a finding about the literature rather than a fault in your question. Unit 3 is step two: taking this question into the databases and finding what is actually there.