Pain-Related Suffering: A Definition and Way to Think About It

A recent study proposes a consensus definition and conceptual framework, with validation from machine learning.

What does it mean to “suffer” in the context of pain? Considering the widespread prevalence of pain, and decades of research into its underpinnings, it might be a surprise for many to learn that there is still no universally accepted definition of pain-related suffering, which hampers clinical care and pain research.

But a new study from researchers at University Hospital Heidelberg, Germany, the University of Haifa, Israel, and Dartmouth College, Hanover, US, is proposing a consensus definition and conceptual framework for understanding what has thus far remained elusive and controversial.

The team, including joint first authors Niklas Noe-Steinmüller and Dmitry Scherbakov, used a manual approach – extracting and synthesizing information about pain-related suffering from the scientific literature – to achieve their aim. Then, they used machine learning to validate what they found.

The researchers say that the definition –  “a severely negative, complex, and dynamic experience in response to a perceived threat to an individual’s integrity as a self and identity as a person” – is consistent with current thinking about pain-related suffering. And, their conceptual framework for understanding this suffering involves eight unique dimensions, indicative of the complexity involved in comprehending the phenomenon.

“[T]his article represents a very important step toward improving our conceptualization of this important, multidimensional construct … it also opens new opportunities for research and clinical practice on pain-related suffering,” wrote Francis Keefe, Joseph Winger, and Sarah Kelleher, all from Duke University School of Medicine, Durham, US, in an accompanying commentary. None of the commentary authors were involved in the study.

However, the definition has attracted some criticism. In a strongly worded letter to the editor, Timothy Wideman, McGill University, Montreal, Canada, and co-authors wrote that the proposed definition is “inadequate” and “fatally flawed,” pointing to the lack of involvement of people with pain in the development of the definition, and to the study’s use of “simple, algorithm-like procedures” and AI, “methods [that] effectively prevented any meaningful critical engagement with the literature and anchored the proposed definition of suffering to the broad and indiscriminate past use of the term,” among other criticisms. The study authors responded to those concerns with a vigorous defense in their own letter to the editor.

The research and commentary appeared in the July 2024 issue of PAIN, while the letter and author reply were published in the October 2024 issue of the journal.

Four decades of disagreement
Noe-Steinmüller, who has studied both psychology and philosophy, told Migraine Science Collaborative that the idea for the study came from a collaboration with researchers in Mannheim, a nearby city.

“They asked one straightforward question to patients: ‘Are you suffering?’ or ‘To what extent are you suffering because of your pain?’” And it became clear that the reason they had to use such a straightforward question was because nobody really knows what suffering is,” Noe-Steinmüller said.

“We realized that if we were to continue with [our work], we were going to need a definition.”

In recent decades, pain-related suffering has most commonly been thought of as “the state of severe distress associated with events that threaten the intactness of the person,” a definition put forward by American physician Eric Cassell in 1982.

However, Cassell’s definition received a lot of criticism over the last 40 years, including for failing to consider certain populations like non-verbal infants.

Noe-Steinmüller said there are two key reasons for a lack of consensus about a definition and conceptualization of pain-related suffering.

“One reason is the predominance of medicating pain via opioids. If you really focus on removing pain, why would you even need to work on suffering?” he told MSC.

“The other reason is that the definition proposed by Cassell, which was intuitive and involved concepts that really widened our perspective, wasn’t really a clear enough explanation. But on the whole there has been a feeling that, ‘Oh, someone has already done something’” about defining this, so further examination is not necessary.

Papers, papers everywhere
So the authors took a novel approach to developing a definition and conceptual framework of pain-related suffering.

They began by conducting a manual, transdisciplinary, systematic review of the published literature on pain-related suffering. That yielded 111 peer-reviewed, English-language articles that included a conceptualization of pain-related suffering in the setting of physical pain. There were no specific requirements for the articles to focus on particular pain-related conditions, or to use specific definitions of pain.

The papers included in the study came from a diverse range of disciplines, most often from nursing studies, medical ethics, psychology, and palliative care, but also from the biological sciences, medicine, and the humanities, among others.

Next, the team used qualitative research methods to identify common themes regarding how the articles defined pain-related suffering, as well as how they differentiated various types or dimensions of pain-related suffering.

After reviewing and considering the key elements of the 60 articles that explicitly defined pain-related suffering, the research team decided on their definition, again: “A severely negative, complex, and dynamic experience in response to a perceived threat to an individual’s integrity as a self and identity as a person.”

The investigators’ conceptual framework for pain-related suffering included eight dimensions, comprising social, physical, personal, spiritual, existential, cultural, cognitive, and affective dimensions, along with two to four descriptors that further characterized each dimension. For example, the existential dimension of pain-related suffering was further described as being related to dissatisfaction with life, existential concerns, and the fear of death.

Machine learning for validation
Noe-Steinmüller and colleagues then used two machine learning approaches to cross-validate their definition and conceptual framework.

“The idea was that the credibility and the impact of the conceptual work could be improved if there was a systematic and less subjective approach to reviewing and synthesizing conceptual data,” he told MSC.

Collaborators Dmitry Scherbakov, Alexandra Zhuravlyova, and Pavel Goldstein, all from the University of Haifa in Israel, had extensive experience with machine learning through the PainStory project. PainStory is an online platform that collects and analyzes the “pain stories” of people with chronic pain using natural language processing. The goal is to improve how researchers and clinicians understand and address chronic pain.

For the current study, large language models (called GPT-3.5 and GPT-3), when provided with the same source material used in the manual analysis, validated the definition of pain-related suffering. The outputs of the large language models were more than 90% similar to the researchers’ manual definition.

Another machine learning approach that can identify topics in large amounts of text produced a conceptual framework for pain-related suffering that was 82% similar to the manually generated framework, proposing the same eight dimensions and all but four of the descriptors the team had previously identified.

Noe-Steinmüller was pleasantly surprised with the match between the manual and machine learning results for the conceptual framework, and welcomes criticism of their approach, such as expressed in the letter to the editor by Wideman and colleagues.

“This was nice, because we could clarify [our rationale for using machine learning] in our responses to such letters. It was exactly what we hoped for – that it would drive further discussion about using machine learning in this fashion,” Noe-Steinmüller said.

Noe-Steinmüller was quick to point out that he and his co-authors are in no way suggesting that artificial intelligence replace other kinds of analysis. Rather, he and his colleagues are using modern technology to supplement and strengthen their overall work.

Where to from here?
Based on the definition and conceptual framework, development of a measure of pain-related suffering for use in clinical and research settings is “[a] logical and important next step in this research,” according to the commentary authors.

“Such a tool might prove invaluable in capturing the individual variations in pain-related suffering that occur across patients who differ in background (sex, racial/ethnic background, or socioeconomic status) and pain context (e.g., acute versus chronic pain or pain at end of life versus intermittent headache in a young adult,” they wrote. “A brief measure also could help clinicians identify high levels of pain-related suffering in a timely fashion and tailor treatments accordingly. A well-validated, brief measure could also stimulate additional research in this area.”

Noe-Steinmüller pointed to the importance of systematically examining patient perspectives and incorporating those findings into the integrated definition and conceptual framework, using AI.

“Our idea would be to do something similar to Pavel Goldstein’s work with the PainStory platform – to collect patient narratives about suffering and use natural language processing to feed that into whatever expert group would come together in the next step,” instead of jumping straight to developing a measure, he said.

Lincoln Tracy, PhD, is a researcher and freelance writer based in Melbourne, Australia. Follow him on X (formerly known as Twitter) @lincolntracy.

Image credit: 123RF Stock Photo.

References
Defining suffering in pain: a systematic review on pain-related suffering using natural language processing.
Noe-Steinmüller et al.
PAIN. 2024 July. 165(7):1434-49.

Pain-related suffering: new insights into what it means and new opportunities for research and clinical practice.
Keefe et al.
PAIN. 2024 July; 165(7):1429-30.