The use of person-centered language in scientific research articles focusing on opioid use disorder

Abstract

Background:

Opioid Use Disorder (OUD) affects approximately 1% of the population. Despite the prevalence of OUD, it remains a highly stigmatized disorder. Using person-centered language (PCL) - and thereby emphasizing the significance of the person rather than their diagnosis - is a potential strategy to reduce stigma in medical research related to addiction. Thus, we aimed to determine adherence to PCL in OUD-related publications according to the American Medical Association’s guidelines.

Methods:

We performed a systematic search for articles published between January 1, 2016 through December 31, 2020 using the PubMed database. Articles were randomized and screened until we reached 300 articles that met inclusion criteria. Three-hundred articles were screened to meet this goal. Articles were then screened for non-PCL terms, determined a priori.

Results:

The majority (240/300; 80 %) of OUD-related publications were not adherent to the AMA guidelines on PCL. Additionally, the use of emotional language (i.e. suffer, afflicted, etc.) was employed in 48 % (145/300) of articles. Stigmatizing terminology was found in 73 % (218/300) of the OUD related articles in this study. Our study demonstrated a statistically significant correlation between senior author affiliation and adherence to reporting guidelines (i.e., PRISMA, STROBE, etc.).

Conclusion:

A majority of OUD-related publications are not adherent to AMA guidelines on PCL. Language used in these publications is often repeated and replicated in medical education and clinical practice, which directly impacts patient-provider relationships. PCL-adherent language is a tool that both medical researchers and clinicians can use to combat stigma that individuals with OUD may experience.

1. Introduction

Drug overdose is the leading cause of death in the United States for individuals less than 50 years of age (Rummans et al., 2018). Notably, there was a 72 % increase in opioid-related overdose deaths among young adults in 2015 (Hadland et al., 2018). Synthetic opioids, like fentanyl, gained FDA approval in 2005. In 2018, an average of 128 U.S. residents died every day from an opioid overdose, half of which were related to fentanyl (National Institute on Drug Abuse, 2020, 2021a). One-hundred fifty million opioid prescriptions were written in 2019, accounting for nearly half of the United States population to have one prescription (CDC, 2019; National Academies of Sciences, Engineering, and Medicine and Health and Medicine Division, Board on Health Sciences Policy, 2017). In 2017, 44 million people in the U.S. used an illicit drug or prescription medication not prescribed to them; three million of those meet the criteria for opioid use disorder (OUD) (McLellan, 2017; Dydyk et al., 2020).

Despite the prevalence of OUD and opioid overdose, stigmatization— the negative attitudes or beliefs toward specific individuals or groups (Thornicroft et al., 2010) —of individuals experiencing OUD is common. This negative public perception and use of stigmatizing language in the media consequently contributes to low rates of treatment seeking behavior, continuity of treatment, and self-determination in recovery efforts (Crapanzano et al., 2019). One study, discussing barriers to treatment in OUD, uses the term self-stigma, when individuals internalize negative public perceptions and believe themselves to be stereotyped by others; which is sometimes faced by those with substance use disorders (Corrigan and Rao, 2012). Subsequently, this stigma diminishes their likelihood to seek treatment as their self-worth and self-esteem are greatly affected by the perception of others (Corrigan and Nieweglowski, 2018). For example, the public surveyed described individuals with substance use disorder as “dangerous and unpredictable” (Yang et al., 2017). Further, the media’s depiction of OUD contributes to this as well. One study demonstrated that 49 % of news stories used stigmatizing language when talking about opioid use (McGinty et al., 2019). Notably, only 20 % of individuals experiencing OUD sought treatment in 2016, and those same individuals prefer inclusive, person-centered language (PCL) (Ashford et al., 2019). Not only are the ethical benefits of PCL necessary but there is also scientific evidence to support the use of PCL. In multiple randomized studies with nearly 4000 participants combined, it was determined with statistical significance that using PCL had positive connotations while terms such as opioid addict had negative connotations among the participants (Ashford et al., 2018; Goodyear et al., 2018). With the advancement in the understanding and treatment of addiction, using appropriate language when discussing substance use disorders, such as OUD, is a way to demonstrate support and understanding for people with OUD.

Considering the unidirectional communication from author to reader in research manuscripts, it is especially important for authors to be cognizant of the language they choose. Especially considering the audience of many OUD related publications are the very people that help patient’s fight addiction - physicians, psychiatrists, therapists, and addiction researchers. The use of stigmatizing language in this setting reinforces the acceptability of stigmatizing language in professional practice (National Institute on Drug Abuse, 2021b). To reduce stigmatization, the American Medical Association (AMA), adopted the use of PCL within medical literature. According to the AMA’s Manual of Style (AMAMS), individuals with disabilities and diseases should not be labeled by their disease, and stigmatizing and emotional language should be avoided (Frey and Young, 2020). Further, authors of medical literature should use inclusive language in their writing in accordance with PCL. For example, ‘individuals with opioid use disorder’ (Frey and Young, 2020), rather than labeling the person by a trait of addiction. This employment of PCL promotes a culture of inclusion and reduces stigmatization of individuals who are often minimized with labels.

The primary objective of this study is to determine the adherence to or deviance from PCL in scientific research regarding opioid use disorder following the publication of the U.S. Surgeon General’s report, Facing Addiction in America: The Surgeon General’s Report on Alcohol, Drugs, and Health (Substance Abuse and Mental Health Services Administration (US) and Office of the Surgeon General (US), 2017), which specifically calls for reducing stigma related to substance-use disorders through teaching and promoting the use of “non-shaming language that underscores the medical nature of addiction.” The secondary objective is to identify factors that may make an article more or less likely to include non-PCL adherent language. This is critical as there has not been a study of PCL adherence specific to OUD related research to date and this literature is directly disseminated to healthcare professionals for patient care purposes.

2. Methods

2.1. Principle aims and study registration

We aimed to quantify the use of PCL in scientific research articles focusing on OUD treatment and determine associations between PCL-adherence and article characteristics in order to determine pertinent factors that make them more likely to use stigmatizing language. Our methodology was submitted to an Institutional Review Board and determined to not meet the criteria of human subjects research. This study was carried out and written in compliance with Strengthening and Reporting of Observational Studies in Epidemiology (STROBE) guidelines. (von Elm et al., 2014) The complete study protocol has been uploaded a priori to the Open Science Framework (OSF.io/twx8f) and has been made publicly available.

2.2. Selection of journals and inclusion criteria

A systematic search was conducted using PubMed (which encompasses MEDLINE) on January 23, 2021 using a cross-sectional design. Articles published between January 1, 2016 and December 31, 2020 were identified for possible inclusion in the study. We searched for specific keywords and terms using the following search strategy: ((opioid* or analgesics[mesh] or buprenorphine or codeine or fentanyl or hydrocodone or hydromorphone or methadone or morphine or oxycodone or oxymorphone or tapentadol) AND (dependence or addiction or misuse) or “opioid-related disorders”[MeSH Terms] OR “opioid use disorder”[Text Word])) AND (opioid use disorder). To ensure that the articles chosen for inclusion originated from journals with an expansive history of publication in opioid research. Journals without 20 or more publications regarding opioid research in the four-year period were excluded from the analysis. The search returns were then randomized and the first 300 articles that met our inclusion criteria were included for the study. The following types of articles were screened to be included in the study: original research articles, case reports, brief reports, and research letters. Editorials and letters to the editor were excluded from the study. In addition, articles were not limited to the United States but had to be written in English and had to be related to human research with qualitative or quantitative results. Any article that did not pertain to opioid research or OUD was deemed unrelated and was therefore excluded. After the completion of article screening, data extraction began.

2.3. Extraction of data

Data was extracted systematically by two authors (BT and MP). Each article was thoroughly screened for adherence to or deviation from the guidelines set forth by AMAMS - the use of labels and emotional language (ANONIMO, 2009). Specifically, we searched articles for pre-determined labels that, according to AMAMS guidelines, would make the paper non-adherent to PCL. These labels included: ‘addict’, ‘user’, ‘junkie’, ‘clean’, ‘abuse’, ‘habit’, ‘druggie’, ‘consumer’, ‘alcoholic’. Each article was then screened for terms that employed emotional language. Search terms consistent with emotional language were as follows: ‘suffer’, ‘afflicted’, and ‘problem’. The aforementioned terms were chosen based on guidance from AMAMS guidelines (ANONIMO, 2009) and the National Institute of Drug Abuse (National Institute on Drug Abuse, 2021b). To identify characteristics that may have increased an article’s likelihood of using non-PCL language, the following metrics were noted from each article: article type, type of research, type of intervention, class of opioids, source of funding, the employer of the first and last author, and adherence to STROBE, PRISMA, CONSORT, or EQUATOR reporting guidelines. Every step of the article screening and data extraction process was performed in a double blind and masked fashion. After completion, data extractors (BT and MP) were un-masked and the data was discussed until complete agreement was reached between extractors. Any conflicts were resolved by discussion with MH, who served as a mediator.

2.4. Analysis of data

Statistical analysis was performed on April 2, 2021, using STATA 16.1 (StataCorp, LLC., College Station, TX). Data analysis began by determining the percentage of articles that were adherent to PCL compared to the total number of included articles. We then calculated the frequency (represented as a percentage) of adherence to PCL for each of the non-PCL adherent labels, listed above under ‘Extraction of Data’. Additionally, we determined the frequency in which emotional language was used in each of the articles. Chi-square analysis and Fisher exact tests were employed to identify any relationship that existed between the presence of non-PCL adherent terms and particular article characteristics, like article type, type of intervention, article funding, and article adherence to AMA or ICMJE guidelines.

3. Results

Our original systematic search resulted in 6371 articles from 1273 journals. After we excluded articles that originated from journals with fewer than 20 publications regarding OUD, there were 2876 articles from 54 journals. The articles were then randomized and data was extracted systematically from each article until we reached 300 that met inclusion criteria. Of the 468 articles screened, articles were excluded because they were unrelated to opioids (14), were not human research (16), were editorials or commentaries (71), articles were not related to OUD (59), and articles were excluded for other reasons (8; Fig. 1).

Fig. 1.

Flow Diagram of Methods.

3.1. Study characteristics

Of the 300 included articles, 209 (209/300; 70 %) were original research articles. The majority of the studies were observational (179/300; 60 %). We searched each article for adherence to methodological reporting guidelines (i.e. STROBE, PRISMA, CONSORT, and EQUATOR); 94 % (282/300) of articles did not reference adherence to the aforementioned guidelines. Grants were the most common funding source for the included studies (NIH, AHRQ, etc.) (129/300, 43 %).

3.2. PCL-adherence

Based on the results from our systematic search, only 20 % (60/300) of the articles were fully adherent to AMA guidelines on PCL. Among the most common terms that were non-adherent to PCL were: the label “user”, emotional terminology (i.e. suffer, afflicted, etc.), and language suggesting helplessness like “abuse” and “problem.” The non-PCL term that was most commonly employed in OUD related research articles was “abuse” (139/300; 46 %). Labels (i.e. ‘addict’, ‘user’, ‘junkie’, etc.) were used in 57 (161/300; 60 %) of the OUD related articles in this study. The word ‘user’ was the most commonly used label with its use in 138 of the 300 (46 %) included articles. The use of emotional language (i.e. suffer, afflicted, problem, etc.) was employed in 145 of the articles (145/300; 48 %). A full list of non-PCL terms and their usage are in Fig. 2.

Fig. 2.

Occurrence rate (%) of non-person-centered language within scientific research articles focused on Opioid Use.

3.3. Associations of PCL and study characteristics

There were statistically significant associations between the employer of the last author and the use of PCL adherent language (X² = 10.2; P = 0.006), and the mention of reporting guidelines (Fisher’s, P = .01). We found no significant associations between the classifications of opioids being studied, nor type of research or article. Complete findings from our statistical analysis can be found in Table 1.

Table 1.

Article characteristics and associations of adherence to PCL language.

Article characteristics	Articles with Non-PCL (240) No. (%)	Articles adhering to PCL (60) No. (%)	Total No. (%)	Statistical Test
Type of Article				Chi-square
Brief Report	17, (5.7)	7, (2.3)	24, (8.0)	2.1, P = .53
Case Report	10, (3.3)	2, (0.7)	12, (4.0)
Original Research	209, (69.7)	49, (16.3)	258, (86)
Research Letter	4, (1.3)	2, (0.7)	6, (2.0)
Type of Research				Chi-square
Clinical Trial	20, (6.7)	7, (2.3)	27, (9.0)	2.4, P = .66
Literature review	5, (1.7)	0, (0.0)	5, (1.7)
None	28, (9.3)	9, (3.0)	37, (12.3)
Observational/Cross.	178, (59.3)	42, (14.0)	220, (73.3)
Systematic Review/m..	9, (3.0)	2, (0.7)	11, (3.7)
Type of Intervention				Chi-square
Drug/Pharmacologic	22, (7.3)	6, (2.0)	28, (9.3)	0.1, P = .97
No Treatment of Any Kind	192, (64.0)	48, (16.0)	240, (80)
Non-Pharmacologic	26, (8.7)	6, (2.0)	32, (10.7)
Classification of Opioids				Chi-square
Illicit Opioids	24, (8.0)	6, (2.0)	30, (10.0)	7.29, P = .12
MAT	112, (37.3)	38, (12.7)	150, (50.0)
opioids - combination	20, (6.7)	5, (1.7)	25, (8.3)
opioids unspecified	21, (7.0)	4, (1.3)	25, (8.3)
Prescription opioids	63, (21.0)	7, (2.3)	70, (23.3)
Adherence to Reporting Guidelines				Fisher’s Exact
None	230, (76.7)	52, (17.3)	282, (94.0)	0.01
Other	0, (0.0)	2, (0.7)	2, (0.7)
PRISMA	4, (1.3)	2, (0.7)	6, (2.0)
STROBE	6, (2.0)	4, (1.3)	10, (3.3)
Article funding				Chi-square
No funding	35, (11.7)	4, (1.3)	39, (13.0)	9.2, P = 0.10
No funding statement	60, (20.0)	11, (3.7)	71, (23.7)
Grant	99, (33.0)	30, (10.0)	129, (43.0)
Mixed	4, (1.3)	3, (1.0)	7, (2.3)
Industry	13, (4.3)	1, (0.3)	14, (4.7)
University	29, (9.7)	11, (3.7)	40, (13.3)
First author employment				Chi-square
Government	41, (13.7)	7, (2.3)	48, (16.0)	10.3, P = 0.06
Private	82, (27.3)	34, (11.3)	116, (38.7)
Public	117, (39.0)	19, (6.3)	136, (45.3)
Last author employment				Chi-square
Government	38, (13.0)	5, (1.7)	43, (14.7)	12.6, P = 0.02
Private	80, (27.3)	35, (12.0)	115, (39.3)
Public	116, (39.6)	19, (6.5)	135, (46.1)
Publishing journal recommends author adherence to AMA/ICMJE				Chi-square
No	158, (52.7)	36, (12.0)	194, (64.7)	0.7, P = 0.40
Yes	82, (27.3)	24, (8.0)	106, (35.3)
H-index				T-test
Mean (SD)	128.5 (124.6)	130.7 (97.2)	128.9 (119.5)	−.13, P = .90

4. Discussion

Our investigation demonstrated that 80 % of peer-reviewed research articles regarding OUD were not adherent to PCL guidelines. Additionally, our study demonstrated the use of emotional language in nearly half of OUD related articles (145/300; 48 %). Emotional language such as ‘abuse’ implies that the individual is deliberately and willfully engaging in the act of misusing opioids; thus, giving the perception that this is a moral failure rather than an actual disease process affecting the individual. The use of emotional language to describe OUD evokes a feeling of vulnerability and powerlessness to the reader, furthering the stigma that exists around individuals with OUD, and with substance-use disorder in general (McLellan, 2017). Interestingly, if an article used one non-PCL adherent term, it was more likely to use multiple non-PCL terms. Of the 240 non-adherent articles, 65.8 % (158/240) used more than one stigmatizing term.

Among the most common terms that were non-adherent to PCL was the label “user”. The term ‘user’ is akin to ‘drinker’ for people with alcohol use disorder and suggests an individual’s identity lies in their disease, depriving them of their identity as an individual. Further, the expression “opioid user” stigmatizes the individual it refers to and gives the perception that the individual is “recklessly engaging in willful misconduct,” rather than battling a disease (Kelly and Westerhoff, 2010). Generally, labels (i.e. ‘addict’, ‘user’, etc) were used in 60 % of included articles related to OUD in this investigation. A study by Hartwell et al. in 2020, similarly found that found that nearly 80 % of scientific articles pertaining to alcohol use disorder (AUD) were not adherent to PCL (Hartwell et al., 2020). In that study, the label “drinker” was used in over 50 % of the non-PCL adherent articles (Hartwell et al., 2020). Similar findings in this separate investigation into PCL in studies related to AUD, demonstrates the pervasive nature of non-PCL language in medical research. Labels heighten the stigmatization of individuals with OUD that already exists. In a study regarding stigmatization of individuals with mental illness, labels were linked to decreased acceptance and tolerance (Crapanzano et al., 2019).

The use of stigmatizing language, like labels and emotional language, is especially harmful when used by medical professionals, as it may hinder a person with OUD from seeking treatment. Further, medical professionals have a duty to treat their patients fairly and without bias. Stigmatizing phrases, including labels and emotional language, risks impeding on the care medical professionals are able to provide due to the bias they create. When PCL is not used, barriers are created and the quality of patient care declines as a result (Crocker and Smith, 2019). A study focusing on attitude toward a cohort of people found that participants were much more accepting of “people with mental illness” compared to groups referred to as “the mentally ill” (Granello and Gibbs, 2016). Even more consequential, a study that specifically included physicians, found that stigmatizing language in medical records led to more unfavorable attitude toward patients. The study even demonstrated differences in prescribing patterns based on the physician’s perception of the patient. (Goddu et al., 2018) Given that OUD is a chronic medical illness and not merely a moral failure, the use of PCL is vital in helping patients, physicians, and the general public help those affected by OUD. Hereby, shifting the focus to the healing of the individual rather than their illness (Crocker and Smith, 2019; Olsen and Sharfstein, 2014).

In a survey study, over 70 % of individuals with heroin use disorder in a substance use detoxification program described themselves as ‘addicts’ (Pivovarova and Stein, 2019). The public has put this stigmatizing label on individuals that struggle with opioid use for so long that these individuals now see themselves as such - a label. It is vital that, moving forward, medical research adheres to PCL in OUD research. Healthcare providers can be pillars of hope and healing for individuals with OUD and opioid misuse rather than propagating their stigmatization through the use of non-PCL language. However, it is evident, through this investigation, that PCL is routinely omitted from medical research articles. When this information is disseminated to healthcare professionals, patients are affected as these providers now have implicit bias and negative perceptions toward those with a particular disease when PCL is not used (Hyams et al., 2019).

Only 35 % of articles in our study required ICMJE guidelines, which requires the use of inclusive PCL in order to reduce stigmatization This may represent a potential mechanism for the substantial number of OUD related articles that were not adherent to PCL (ICMJE, 2021a). Several respected organizations, like AMAMS, ICMJE, and SAMHSA support the use of PCL (ICMJE, 2021b). Unfortunately, despite the endorsement from these organizations, our study demonstrates the deficits that still exist in the incorporation of PCL into medical research. AMAMS guidelines give clear insight on using inclusive language (Frey and Young, 2020). In Table 2., examples from the National Institute on Drug Abuse and the American Psychological Association are outlined to offer inclusive, PCL-adherent, alternatives to stigmatizing phrases too often used in our vernacular (National Institute on Drug Abuse, 2021b; Anon, 2021). It is well documented that physicians are not immune to implicit bias as several studies have proven (Cooper et al., 2012). The use of PCL, in adherence to the aforementioned guidelines, is one way that we can reduce the stigmatization that individuals with OUD face in healthcare (Jost et al., 2009).

Table 2.

Recommendations of person-centered language in opioid use research to reduce stigma.

Instead of this...	Use this...
Addict	Person with substance use disorder, Patient
User	Person with substance use disorder, Patient
People suffering from opioid abuse	People who have Opioid Use Disorder
Problem with opioid misuse	Challenges that people with Opioid Use Disorder face
Opioid substitution replacement therapy	Opioid agonist therapy, Medication treatment for Opioid Use Disorder
Drug Abuse	Drug misuse, Drugs used other than prescribed
Drug Habit	Substance use disorder
Drugie or Junkie	Person with opioid use disorder, Person with substance use disorder
Clean	Abstinent, Being in remission
Clean drug screen	Negative drug screen
Dirty drug screen	Positive drug screen
Former addict	Person in recovery, Person who previously used drugs

Stigmatizing language is embedded in the current culture of treatment and recovery. For example, at the beginning of most Alcohol Anonymous meetings, a person often introduces themselves as an “alcoholic” or “addict.” In the clinical setting, urine drug screens are often referred to as “clean” and “dirty” by both patients and medical staff. Public perception is heavily influenced by stigma and will not change to a more science-informed perspective if those leading the fields of addiction science do not adhere to person-empowering language. Treatment is futile if people do not access it due to public perception.

4.1. Strengths and limitations

Limitations include inherent subjectivity in the search terms for our data. We confronted the subjectivity by strictly adhering to our protocol in a systematic fashion. Strengths of our study are that the protocol is open to the public, ensuring the study is reproducible. We also had over 50 different journals included in our study ensuring that there was adequate randomization of articles and that there was adequate representation of journals in the field of research related to opioid use. Another strength of our study was the duplicated and double-blind data extraction, which helped ensure the accuracy of our data.

5. Conclusion

The purpose of this study was to demonstrate the presence of stigmatizing language within medical research articles related to OUD and the harm that said language can have on individuals with OUD. Our aim is that authors eliminate stigmatizing language from their research articles and that editors of journals refuse to publish articles with stigmatizing language in order to provide the most patient-centered care to individuals with opioid use disorder. We hope to emphasize the importance of addressing people with OUD in a non-stigmatizing manner and in the way that the individual chooses, in adherence with PCL guidelines. Doing so will strengthen the patient-physician relationship and assist in breaking down barriers in addiction medicine. This will encourage patients with OUD to seek the help they deserve.