The problem with "objective" and "subjective" measures - objectively speaking

I know there is a lot on your mind right now, such as kids starting online school, how to focus when the world is very noisy, and of course one of the biggest issue of them all: what’s the deal with clinical studies using the somewhat arbitrary terms ‘subjective’ and ‘objective’ when discussing outcome measurements? I totally agree with you, they don’t explain a whole lot about the measurements and create an incomplete perspective of the outcomes. What a perfectly convenient time for everyone to try and solve this problem. 

I would argue that the terms should be replaced with descriptions such as 'patient-derived' and 'clinician-derived' as it is a little easier to understand what is happening. In this blog post, we’ll review what the terms objective and subjective mean, why they are defined this way, and why I believe they should be replaced moving forward. 

What are objective and subjective outcomes?

Objective measurements are impartial, usually quantifiable outcomes recorded with some kind of diagnostic instrument. Examples in medicine include blood work to determine cholesterol levels, sphygmomanometer for blood pressure, and wearable devices that measure step count.

Subjective measurements, on the other hand, usually rely on human judgment of some kind. Often, researchers think of patient-reported outcomes as a classic example of subjective measurements, such as “How am I feeling today?”, but doctors' assessments of their patient’s overall well being can just as easily fall into this category. There is often greater variability in how ‘subjective’ outcomes are recorded, ranging from patient journals and informational interviews.

But what do we call a term that’s objectively subjective? Sure, my Fitbit tells me how many steps I took (an objective measure), but what exactly does the step count tell me? How “healthy” I am? How physically active I am? Technically, it only tells me how many steps I took and nothing else, and everything else is a small logical jump. I took a lot of steps today therefore I am healthy. 

What about a term that is subjectively objective? If I ask you, “How many steps did you take today?”, this would technically be a “subjective” outcome even if you tried your best to count accurately. However, you could still make the same logical conclusions as before. I took a lot of steps today and I feel healthy. 

For a deeper understanding of the ‘subjective’ and ‘objective’ labeling, it is helpful to have a fundamental understanding of key terms. Lucky for you, we get to revisit our favorite school subject - statistics - and learn more about the concepts of reliability, validity, and bias. (No need to thank me! You’re welcome!) These terms really drive whether a researcher labels a measurement as 'objective' or 'subjective'.

Reliability means that a measurement device (could be a diagnostic tool or personal response) will repeatedly capture the same information each time when no change occurs. 

Validity refers to the quality of the measurement as a proxy I’m trying to measure. Does the outcome measurement really assess the underlying scientific question?

Bias is how accurate (or inaccurate) the measurement is to the “truth”. (We’ll save the philosophical rabbit hole of “What is truth?” for another blog post!) (But actually, stay tuned.)

We’ll use an example as it will be much easier to understand these terms, and then you can tell all your friends how statistically savvy you are! :)

Let’s say I want to measure my weight using a digital weight scale - an objective measurement for all intents and purposes. I weigh myself once, and the scale says ‘200’ lbs (Quarantine hasn’t helped much). Then I weigh myself two more times, and if they both read ‘200’ then the scale would be reliable. Presumably my weight hasn’t changed, and the scale reflects this as such. 

Next, if I wanted to know my weight, the digital scale is a valid measurement, as a weight scale is a good proxy to my actual weight. While the scale is valid for my weight, it is common in many facets of life to use a scale for many more outcomes to which it is not as valid. 

Too often, a weight scale is used for many more decisions than just understanding my body mass. A digital scale is much less valid for assessing one’s overall physical health, the progress people make during a workout routine, or an individual’s love and self-worth - even though each of these may be a more relevant outcome of interest.

This notion is important to consider later on, as I believe sometimes researchers can get so caught up in, “Is this a valid measurement?” And while the answer may be “Yes”, the measurement could still be an insufficient proxy to the ultimate outcome of interest. 

Finally, if the scale said ‘200’ lbs, but in reality I weighed 195 lbs, the scale would be biased, and the measurement doesn’t reflect the truth. If I weighed 200 lbs and the scale said ‘200’ lbs, it would be unbiased. 

This example is a bit of an oversimplification, as these ideas have rich scientific literature behind them. Especially in health data, it may very well be impossible to truly know ‘the truth’ or achieve a perfectly valid assessment, but these goals are important to understand as we explore outcome measurements. 

Why replace the terms 'objective' and 'subjective' moving forward?

In healthcare, due to the complexity of many different situations and health outcomes, I believe ‘objective’ and ‘subjective’ create the wrong perspective when describing outcome measurements for several reasons. 

First, for better or worse, there sometimes is a consensus in the research community that objective measurements are better than subjective measurements, primarily related to the reliability, validity, and bias terms described earlier. While sometimes objective measurements are more reliable, valid, and unbiased, this is not always the case (1,2). 

Returning to our step-count example, a Fitbit could be used after knee surgery to see how well a patient is recovering. While the step-count is technically an objective measurement of movement and an incredibly helpful tool to gauge one aspect of knee health, it needs to be supported by patient inputs such as their pain perspectives, general mobility, and need for assistance. Otherwise the step-count paints an incomplete picture. 

Second, objectivity and subjectivity are not binary categories, but rather exist on a continuum between highly objective and highly subjective (3), leading to a high degree of variability in describing the qualities of outcomes. There is also no definite consensus on what constitutes an 'objective' and 'subjective' outcome in clinical research (4). The terms themselves are a bit ambiguous and in clinical trials have to be defined more specifically, such as how the outcome is derived and who is responsible for collecting the data (patient v. clinician v. wearable device).  

Third, objective measurements don’t always exist in every situation. More complex chronic conditions, for example, autoimmune, behavioral, or respiratory complications, use quality of life measurements such as pain management, fatigue levels, or feelings of anxiety to assess overall well-being. Even though objective measurements don’t exist, there still can be ways to capture pain levels in a reliable, valid, and unbiased way (5, 6).

Finally, patient-reported outcomes often fall into the subjective category which, as mentioned earlier, can get a bad reputation as inherently “worse” than an objective measure. This is at cross-purposes with the ultimate goal of research, drug development, and clinical decision making, as they are specifically designed to improve how patients feel. It only makes sense we would want a patient-derived outcome, because that’s the primary goal in the first place. 

Using the terms 'patient-derived' and 'clinician-derived' 

I suggest using the terms 'patient-derived' for outcomes collected from the patient and 'clinician-derived' when obtaining them from a clinical assessment. Adding the capture method into the description of the measurement removes our inherited thoughts about the objective and subjective debate, so we can instead focus on the important qualities of the measurements: whether they are reliable, valid, and unbiased. 

Perhaps we can even expand these definitions with more descriptive adjectives like “human-captured” vs. “device-captured”. This informs everyone involved in the study and data collection about where the data is coming from and how it will be collected. 

Using this type of terminology here at Folia helps us understand and communicate, with data clarity, the insights in the wonderful data that our patients and caregivers capture. 

Consider all the variables that go into overall pulmonary health. We can capture an array of patient-derived outcomes such as their coughing status, mucus production, and difficulty or ease of breath. We also have clinically derived information like FEV1 or prescribed treatment plans, which indicate similar but different avenues of respiratory well-being. Finally, we can use an at-home device like pulse oximeter to monitor on days away from the clinic. 

The information collected by an individual in the course of living with a condition is so much more than just 'objective' or 'subjective'. It is an amalgamation of rich observations, stemming from the patient perspective, clinical results, and diagnostic tools being used by community members to gain an improved understanding of that most important of possessions - their health. 

Have questions? Comments? Want to learn more? Reach out to us here!