Operational Definitions and Study Variables for CPSP and UHS Synopses

“Pain Will Be Assessed”

Dr. Hammad Ali had rewritten his methods page twice since the last session, and he put it on the table with the confidence of a man who had finally got sampling right. I read down to the heading Operational Definitions. Under it was one line.

“Post-operative pain: pain after surgery will be assessed.”

“Hammad, who will assess it?”

“Sir, the team.”

“When? Two hours after surgery, or the next morning when the patient is sitting up and eating?”

“Whenever we take the round, sir.”

“With what? And how much pain counts as pain?”

He had no answer, and he did not need one, because Dr. Sumaira Talib had already pushed her own synopsis forward. “Sir, mine is worse. I wrote that acute appendicitis will be diagnosed ‘as per standard textbook criteria’. My supervisor wrote ‘Which textbook?’ in the margin.”

“Your supervisor asked the right question,” I said. “Today we deal with the section that most synopses treat as decoration and most examiners treat as the heart of the methods. An operational definition is the instruction manual for your data. If two doctors follow it on the same patient, they should write the same thing on the proforma.”

First, Know Your Variables

“You cannot define what you have not named. A variable is any characteristic that differs from one patient to the next: age, gender, a pain score, whether the appendix was inflamed. Before you write a single definition, sort your variables by the job they do in your study [1].”

Role	What it is	Example from the room
Independent (exposure or intervention)	The variable you think influences the outcome, or the treatment you give	Hammad: low-pressure versus standard-pressure pneumoperitoneum. Bushra: a complete course of antenatal corticosteroids
Dependent (outcome)	The variable you are trying to explain or change	Hammad: shoulder-tip pain score. Bushra: neonatal respiratory distress
Confounding	A third variable linked to both exposure and outcome, which can create or hide an association	Bushra: gestational age at delivery, which affects both who receives steroids and who develops respiratory distress
Descriptive (baseline)	Variables that describe your sample so readers can judge to whom the results apply	Age, gender, BMI, residence

Andrade makes the point plainly: a significant relationship between an independent and a dependent variable does not prove cause and effect, because a confounder may explain part or all of it [1]. “So a confounder you suspect must be measured,” I said, “and anything you measure must be defined.”

Then, Know Their Type

Dr. Zunaira Malik asked the question I was waiting for. “Sir, why does the type matter for the definition?”

“Because the type decides how you record it, how you summarise it and which test you will use. Get the type wrong in the definition and your statistical analysis paragraph will be wrong too [2].”

Type	What it looks like	Reported as	Example
Nominal (categorical)	Categories with no order	Frequency and percentage	Gender; OSMF yes or no; blood group
Ordinal (categorical)	Categories with an order but unequal gaps	Frequency and percentage, or median	ASA class I, II, III; OSMF clinical stage
Discrete (numerical)	Whole-number counts	Mean ± SD or median	Number of doses missed; parity
Continuous (numerical)	Any value on a scale	Mean ± SD, or median (IQR) if skewed	Age in years; systolic BP; appendix diameter in mm

The Five Parts of a Definition

I wrote five words across the board: WHO, WHEN, HOW, CUT-OFF, SOURCE.

“To operationalise a variable is to define it in a way that permits its accurate measurement [1]. In practice, a definition that a CPSP or UHS reviewer will accept answers five questions.”

Who will measure it: the researcher, a named grade of staff, or a person unaware of the group the patient belongs to.
When it will be measured: at enrolment, at a stated hour after surgery, at a stated follow-up visit.
How it will be measured: the instrument, scale, test or record, and the conditions (for example, seated, after five minutes’ rest).
The cut-off or categories, if the variable is to be labelled: what counts as “present”, “adherent” or “positive”.
The source of the criterion: a guideline, a validated scale or the paper you took it from, so the reviewer knows you did not invent it.

“Leave out ‘who’ and you invite measurement bias. Leave out ‘when’ and your patients are measured at different points of recovery. Leave out the cut-off and two data collectors will disagree on the same patient.”

Hammad’s Pain, Measured Properly

“Hammad, in the trial session you were going to score the pain yourself, and you knew which patient had which pressure. What did we say?”

“That my hopes would leak into the scores, sir. The outcome assessor should not know the group.”

“Then write that into the definition. And choose an instrument. The 0 to 10 numerical rating scale is simple, quick and widely validated for adult pain, and it can be asked verbally at the bedside, which matters on a post-operative ward [3].”

He started writing, then stopped. “Sir, should I convert it into mild, moderate and severe? My senior’s thesis did that.”

“Only if you have a reason. Your primary outcome is a score, and a score carries more information than a label. Cutting a continuous variable into categories throws away precision and statistical power [4]. Andrade accepts it only in particular situations, for example when the variable cannot be measured accurately or when there is a clear clinical or public health need for categories [4]. Keep the score as your primary outcome. If you want a category as well, say ‘moderate to severe pain (NRS 4 or above)’ as a secondary outcome, and give the source of that cut-off.”

His definition, by the end of the hour:

Shoulder-tip pain will be assessed by the ward nurse on duty, who will not know the group allocation, at 6, 12 and 24 hours after the end of surgery. The patient will be asked to rate pain felt at the tip of either shoulder on the 0 to 10 numerical rating scale, where 0 is no pain and 10 is the worst pain imaginable [3]. The score will be recorded on the proforma as a number. Rescue analgesia given in the first 24 hours (drug, dose and time) will also be recorded from the treatment chart.

Sumaira’s Two Tests

“Sumaira, a diagnostic accuracy study has two tests, so it needs two definitions, and each must say what ‘positive’ means.”

“Sir, for the ultrasound I can use a non-compressible appendix with an outer diameter of more than 6 mm?”

“That is the criterion most studies use. Now add who and when. If the radiologist knows the surgeon has already decided to operate, what happens to his report?”

“He reads it as positive,” she said. “Review bias, from the diagnostic accuracy session.”

“So the ultrasound is done before the decision is recorded, by a stated grade of radiologist, and the histopathologist reports without seeing the ultrasound result. Then cite the paper your criteria come from, not a textbook.”

Her revised pair:

Positive ultrasound for acute appendicitis will be recorded when a consultant radiologist, performing graded-compression ultrasound before the operative decision is entered in the file, finds a non-compressible, blind-ending tubular structure with an outer diameter greater than 6 mm in the right iliac fossa.
Positive histopathology (reference standard) will be recorded when the consultant histopathologist, unaware of the ultrasound result, reports neutrophilic infiltration extending into the muscularis propria of the resected appendix.

Bushra and the Variable Nobody Defines

Dr. Bushra Fatima had defined respiratory distress carefully: tachypnoea above 60 breaths per minute, grunting, chest retractions or nasal flaring, with a need for supplemental oxygen within the first 24 hours of life, assessed by the paediatric resident on duty. I read it twice and found nothing to change.

“Now your exposure. What is a complete course?”

“Two doses of betamethasone 12 mg intramuscularly, 24 hours apart, with delivery at least 24 hours after the second dose. Taken from the treatment chart.” She smiled. “I did that one too.”

“And gestational age? It is your most important confounder.”

The smile went. “I wrote it in the proforma, sir. I did not define it.”

“How will you know it? From the last menstrual period that a frightened mother half remembers in the labour room, or from a first-trimester dating scan?”

“Some have a scan, most don’t.”

“Then say so. Gestational age will be taken from a first-trimester ultrasound where available, and otherwise from the last menstrual period, and record which method was used for each mother. A reviewer can accept an imperfect measure. He will not accept an undefined one.”

Junaid’s Adherence

Dr. Junaid Rashid read his aloud before I could ask. “Adherence will be assessed by the researcher at the OPD visit. A patient taking at least 80 percent of prescribed antihypertensive doses in the last 30 days will be labelled adherent.”

“Who, when, cut-off. Good. How?”

“I ask the patient.”

“Ask what, exactly? ‘Do you take your medicines?’ gets a yes from almost everyone. Write the question, or the counting method, into the definition: number of daily doses prescribed, number missed in the last 30 days by the patient’s own report, checked against the remaining tablets if the patient has brought the strip. Then put that same question on the proforma, word for word.”

“And blood pressure, sir? I have it as a secondary variable.”

“Then define it like a measurement, not a number. Which apparatus, calibrated when? Seated, after five minutes’ rest, which arm, how many readings, and whether you use the mean of two. If three people take BP in your OPD in three different ways, you are measuring the doctors, not the patients.”

The Proforma Rule

Zunaira had been comparing her proforma with her definitions while we talked. “Sir, I have ‘socioeconomic status’ and ‘oral hygiene: good, fair, poor’ on the proforma. Neither is defined anywhere.”

“Then you have found the rule examiners check most often. Every item on the proforma must be one of two things: something read directly and unambiguously, like age in completed years or gender, or something defined in your operational definitions. ‘Good oral hygiene’ is an opinion until you give it an index, an examiner and a cut-off. If you cannot define it, ask yourself whether you need it at all. Every extra variable is extra work at the bedside and an extra question from the reviewer.”

“And my exposure, areca nut?”

“Asked by you, at recruitment, with the same question to cases and controls: type of product, how many times a day and for how many years. Define a ‘current chewer’ with a frequency and a minimum duration, give your source, and ask it exactly the same way in both groups. In a case-control study, a question asked more carefully of the cases is recall bias you created yourself.”

Five Mistakes Reviewers Spot First

Mistake	Example	Fix
Textbook definition instead of a measurement	“Anaemia is a decrease in red cell mass”	State the test, the threshold and its source
No who or when	“Pain will be assessed”	Name the assessor and the time points
Circular definition	“Wound infection: infection of the wound”	List the signs or criteria that make it present
Undefined proforma items	“Oral hygiene: good, fair, poor”	Define it with an index, or remove it
Continuous outcome cut without reason	Pain recorded only as mild, moderate or severe	Record the number; add categories only as a stated secondary outcome

Hammad looked at his page. One line had become a paragraph. “Sir, it is longer.”

“It is longer because now someone else could do your study,” I said. “That is what a method is for.”

Key Takeaways

Name every variable and its role first: independent (exposure or intervention), dependent (outcome), confounding and descriptive.
Know each variable’s type (nominal, ordinal, discrete or continuous), because it decides how you record, summarise and test it.
A usable operational definition answers five questions: who measures, when, how, the cut-off or categories, and the source of the criterion.
Build bias protection into the definition: an assessor unaware of the group in a trial, a radiologist unaware of the operative decision, the same question asked of cases and controls.
Keep a continuous outcome as a number unless there is a stated reason to categorise it; categories lose precision and power.
Every proforma item must either be read directly or be defined in the operational definitions. If you cannot define it, consider removing it.

References:

Andrade C. A student’s guide to the classification and operationalization of variables in the conceptualization and design of a clinical study: Part 1. Indian J Psychol Med. 2021;43(2):177-179.
Kaliyadan F, Kulkarni V. Types of variables, descriptive statistics, and sample size. Indian Dermatol Online J. 2019;10(1):82-86.
Hawker GA, Mian S, Kendzerska T, French M. Measures of adult pain: Visual Analog Scale for Pain (VAS Pain), Numeric Rating Scale for Pain (NRS Pain), McGill Pain Questionnaire (MPQ), Short-Form McGill Pain Questionnaire (SF-MPQ), Chronic Pain Grade Scale (CPGS), Short Form-36 Bodily Pain Scale (SF-36 BPS), and Measure of Intermittent and Constant Osteoarthritis Pain (ICOAP). Arthritis Care Res (Hoboken). 2011;63(Suppl 11):S240-S252.
Andrade C. A student’s guide to the classification and operationalization of variables in the conceptualization and design of a clinical study: Part 2. Indian J Psychol Med. 2021;43(3):265-268.

Rate this article

How useful was it for you?