Person-centered language and pediatric ADHD research: a cross-sectional examination of stigmatizing language within medical literature

Journal selection and publication randomization and reduction

Two authors (MH and SG) conducted a systematic search utilizing a cross-sectional study design via PubMed (which encompasses MEDLINE) on March 8, 2021, utilizing an adapted, previously published search string [24]: ((“attention deficit disorder with hyperactivity”[MeSH Terms] OR (“attention” [All Fields] AND “deficit” [All Fields] AND “disorder” [All Fields] AND “hyperactivity” [All Fields]) OR “attention deficit disorder with hyperactivity” [All Fields] OR “ADHD” [All Fields]) OR “ADDH” [All Fields]) OR “ADHD” [All Fields]) OR “attention deficit” [All Fields] OR “brain dysfunction” [All Fields]) AND (“humans” [MeSH Terms] AND “English” [Language] AND (“infant” [MeSH Terms] OR “child” [MeSH Terms] OR “adolescent” [MeSH Terms]))) AND (2014:2021[PDAT]). We chose to search PubMed because it is one of the most widely utilized, publicly accessible platforms for biomedical literature [25].

Studies published between January 1, 2014 and March 7, 2021 were filtered, following the publication of the DSM-5 [26] in May 2013—subsequent to the inclusion of PCL guidelines in the AMAMS [19]. Journals were selected for inclusion to retrieve the most frequent sample research focused on ADHD if they have at least 20 search returns from the search strategy. This process has been utilized in multiple studies as a way to glean the journals that most often publish content on the topic of interest that researchers and healthcare professionals in that field are likely to encounter [23, 27, 28].

Google Sheets were utilized for screening and extraction purposes for the randomized search returns. Articles were included in our study if they were original research articles, including research letters, brief reports, case reports, and editorials focused on ADHD pertaining to humans. Studies focused on animal modeling or genetics (which did not address human subjects or populations) and studies not in English were excluded. Only studies published in English were included because PCL guidelines within the AMAMS may not be translatable to other languages. Separate Google Sheets were utilized by two authors (CC and KR) to conduct screening and data procedures in a masked manner until data from 300 articles had been extracted. The investigators were unmasked to meet and verify that there were no discrepancies between the two data sets. They then went through each article that did not match, and they came to an agreement on how each article should be classified. All discrepancies were resolved by revisiting the manuscript text and making changes so that 100% agreement was reached. Data collection was completed on July 1, 2021.

Data extraction

For our primary research question, we assessed adherence to or deviance from guidelines presented in the AMAMS [21]: The AMAMS promotes putting the person first by utilizing person-first language (i.e., avoidance of labels), emotive language (i.e., suffers from, problems with, afflicted by), and euphemistic descriptions of individuals. We systematically searched for the following terms, partial terms, and phrases in conformity with the AMAMS along with the following words identified in the previously published literature that were emotive or had negative connotations [29, 30]: “gifted,” “odd,” “eccentric,” “abnormal,” “retardation,” “sick,” “unstable,” “challenging,” “talkative,” “disturbed,” “difficult*” (“difficult child” or “child is difficult”; not has difficulties with x or y), “crazy,” “disabled,” “mental” (“mental patient”; “mentally ill”; “mental illness”), “spastic,” “suffer*” (“sufferer”; “suffers from”), “problem” (“problems with”; “problem child”) (where “problems with” refers to an individual having “problems with” a child who has ADHD, not a child having “problems with” concentration/attentiveness/etc.), and “afflicted” (with/by). We did not include non-PCL that was included within direct quotes from participants within studies.

The article-level elements extracted were: (1) type of article: original (full) article (including brief report or research letters), case reports, or editorial; (2) type of research: systematic review, clinical trial, cross-sectional, etc; (3) type of intervention (pharmacologic, nonpharmacologic, or no treatment); (4) employment type of first and last author: private, public, or government; and (5) funding statement/source. We defined “government” employment as one associated with a governmental organization, “private” as associated with for-profit organizations with no federal funding, and “public” as employment being within a nonprofit organization or university.

Data analysis

We reported the following: (1) total number of journals and articles from the PubMed searches; (2) number of journals included in our investigation; (3) number of studies that were screened; and (4) number of studies that were included based on the eligibility criteria from each selected journal. Articles including one or more of the labels or terminology listed above were coded as not adherent to the PCL guidelines. Utilizing publications from January 1, 2014 to March 7, 2021, we calculated the proportion of articles without deviance from the AMAMS [21] guidelines relevant to PCL compared to the total number of articles in the included study sample. Further, to evaluate the most common forms of deviance from AMAMS guidelines within these articles, we calculated frequencies, percentages, and 95% confidence intervals. We utilized Fisher’s exact tests to identify potential associations between adherence to PCL and the article and journal characteristics listed above. The Type 1 error rate for analyses were set at 0.05, which were performed utilizing STATA 16.1.

Journal reduction and screening

Our systemic search returned a total of 9,353 articles from 1,371 journals. After journals with less than 20 items in the query were removed, 88 journals were retained from January 1, 2014 to March 7, 2021. From the remaining journals, 5,308 combined articles were randomized, and the first 500 were selected for screening. Of these articles, 311 met the inclusion criteria and 189 did not. Articles met the inclusion criteria if they were peer-reviewed and focused on children with ADHD. Articles were excluded if they did not meet the inclusion criteria. Reasons for exclusion included 78 articles about adult populations, 70 articles unrelated to ADHD, and 41 articles not involving human research. This can be found in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Flow Diagram in Figure 1.

Figure 1:

PRISMA flow diagram for systematic search, article selection, and outcomes. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Study characteristics

Of the 311 articles within our sample, 291 were original research (93.6%), and 20 articles were case reports or editorials (6.4%). The majority of the articles by type of research were cross-sectional (237/311; 76.2%) and were funded through grants (168/311; 54.0%). Out of the remaining articles, 12 were clinical trials (3.9%), 19 were literature reviews (6.1%), 12 were not research articles (3.9%), and 31 were systematic reviews (10.0%). In terms of funding, 43 received no funding (13.8%), 5 received government funds (1.6%), 16 receive industry funding (5.1%), 13 received university funding (4.2%), and 66 had no funding statement (21.2%). Of the 311 articles, 56 reported pharmacologic interventions (18.0%), 237 reported no treatment (76.2%), and 18 reported nonpharmacologic interventions (5.8%). First-author employment was most commonly found to be public (304/311; 97.8%) with the remaining being government (5/311; 1.6%) or private (2/311; 0.6%). Last-author employment was most commonly private (304/311; 97.8%) with the remaining being government (4/311; 1.3%). These values can be found in Table 1.

Adherence to PCL guidelines

Among the sample of articles on children diagnosed with ADHD, we determined that 131 of the 311 (42.1%) articles were adherent to PCL guidelines. Emotive language most frequently appeared among the articles including ‘problems with,’ which was found in 47.6% (148/311) of articles, and ‘suffers from,’ which appeared in 5.8% (18/311). Other labeling and stigmatizing terminology occurred less frequently, with the label ‘mentally ill or mental patient’ being found in 3.9% (12/311) and ‘difficult child’ or ‘child is difficult’ found in 1.6% (5/311) of the articles. ‘Gifted,’ ‘sick,’ and ‘talkative’ were each found in 0.3% (1/311) of articles. ‘Challenging’ and ‘disabled’ were each found in 1.3% (4/311). The term ‘afflicted with/by’ was found in 0.6% (2/311) of articles. These values can be found in Figure 2.

Figure 2:

The percent of articles (n=311) with stigmatizing language in published research focused on children with ADHD. The following terms were also included in the systematic search of articles but were not found: Odd, Eccentric, Abnormal, Retarded/Retardation, Unstable, Disturbed, Crazy, and Spastic.

Associations of PCL and article characteristics

To determine the relationship between the article characteristics and the use of PCL language, we performed Fisher’s exact tests. From these analyses, we found no statistically significant associations between PCL adherence and the first author’s employment, type of article, type of research, type of intervention, article funding, or the employment of the last author (Table 1).

Limitations and future directions

A limitation of our study involves the subjectivity of human interpretation of emotive language as defined by the AMA. In an attempt to mitigate these risks, investigators in charge of data extraction were trained in PCL, and the systematic searching of publications utilizing predefined terms was adopted. While our list of predefined terms was based on previous literature, it may not encompass all stigmatizing terminology utilized in ADHD research. For example, the term ‘disruptive’ may be seen as having a negative connotation and was found in 42 papers; however, it is utilized in the DSM-5 and is part of the name of a diagnosis, therefore we excluded it from the analysis. Additionally, the use of solely PubMed articles, as well as the reduction of articles utilized to determine PCL use, may have limited the breadth of the ADHD research we investigated. However, this selection process was utilized to select the articles with the most current and well-researched information on children with ADHD. Strengths of our study include the use of publicly available protocols for reproducibility and the randomization of article selection. Future research should cover the importance of PCL utilization in adherence with AMAMS in journals, clinical practice, and medical education to stress the positive correlation with person-centered care and how it benefits the patient, specifically research on how PCL use in journals, clinical practice, and medical education can impact the mental and physical health of each patient. Future research may also explore how PCL is utilized with a variety of medical conditions and research, particularly with highly stigmatized medical conditions, by investigating word choice and avoidance of negative associations. Additionally, development of PCL guidelines for journals as well as authors would be beneficial for implementing PCL use in medical literature.