Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Jan 1.
Published in final edited form as: J Obstet Gynecol Neonatal Nurs. 2025 Sep 5;54(6):606–617.e3. doi: 10.1016/j.jogn.2025.08.003

Patient Disability Status and the Use of Stigmatizing Language in Clinical Notes During Hospital Admission for Birth

Sarah E Harkins 1, Ismael I Hulchafo 2, Jihye Kim Scroggins 3, Caroline Walsh 4, Meghan Didier 5, Maxim Topaz 6, Veronica Barcelona 7
PMCID: PMC12755039  NIHMSID: NIHMS2127704  PMID: 40921416

Abstract

Objective:

To examine the association between patient disability status and use of stigmatizing language in clinical notes from the hospital admission for birth.

Design:

Cross-sectional study of electronic health record data.

Setting:

Two urban hospitals in the northeastern United States.

Participants:

Patients at more than 20 weeks gestation admitted for birth from 2017 to 2019 (N = 19,094).

Methods:

We used a natural language processing algorithm to identify categories of stigmatizing language used in free-text clinical notes (N = 211,841 unique clinical notes). We employed multivariable logistic regression to estimate adjusted odds ratios (aORs) and 95% confidence intervals (CIs) for each stigmatizing language category by disability status, which we determined by ICD-10 (International Classification of Diseases, 10th revision) codes.

Results:

Approximately 3% of patient records (n = 550) included ICD-10 codes for disability. Clinicians were more likely to use stigmatizing language for patients with disabilities compared with patients without disabilities (aOR = 1.75, 95% CI = [1.47, 2.09]). For patients with disabilities compared with patients without disabilities, clinicians were also more likely to use stigmatizing language in the difficult patient category (aOR = 1.96, 95% CI = [1.65, 2.33]) and the unilateral/authoritarian decisions category (aOR = 1.27, 95% CI = [1.06, 1.53]). We found no significant differences for the marginalized language/identities category by patient disability status (aOR = 1.19, 95% CI = [0.87, 1.62]).

Conclusion:

The use of stigmatizing language in birth hospitalization notes differed by patient disability status. Stigmatizing language should be used as a marker of bias and an opportunity for clinicians to reflect on their thoughts, words, and actions. Patient-centered documentation and care practices are needed to improve perinatal health for all.

Keywords: disability discrimination, labor, natural language processing, health equity, obstetric, stigmatizing language

There are stark perinatal health inequities between patients with and without disabilities in the United States (Gleason et al., 2021). Disability is an umbrella term defined in current classification systems as a dynamic interaction between individuals’ body functions or structures and their environment (van der Veen et al., 2023). The prevalence of patients with physical, sensory, or intellectual and developmental disabilities who gave birth in the United States was approximately 3% from medical records data (Dev et al., 2025), although estimates were as high as 20% in survey data (Horner-Johnson et al., 2022). Gleason et al. (2021) used data from the Consortium on Safe Labor (2002–2008; N = 223,385) to evaluate the risk of 23 adverse maternal outcomes among patients with and without physical, sensory, or intellectual disabilities. They reported that patients with any disability had increased risks of nearly all pregnancy complications, including severe maternal morbidity and mortality. Gleason et al. (2023) used the same data to examine risk of adverse neonatal outcomes and found that patients with physical, sensory, or intellectual disabilities had significantly higher risks of preterm birth, low birthweight, and neonatal sepsis compared with patients without disabilities.

Ableism is defined as a system of beliefs, practices, and policies that oppress patients with disabilities (Lundberg & Chen, 2024). This type of identity-based discrimination is a form of clinician bias and may contribute to the observed perinatal health inequities by disability status (Mitra et al., 2016). For example, patients with disabilities reported stigmatizing encounters with clinicians in obstetric care settings, including judgment regarding their decisions to become pregnant and increased surveillance from child protective services (CPS; Harkins et al., 2025). In addition, patients with disabilities described clinicians as being dismissive of their pain concerns during childbirth (Smeltzer et al., 2017) and unwilling to assist with accessibility needs in the hospital birth admission (Saeed et al., 2022).

Ableism and clinician biases about patients with disabilities may also be conveyed through clinicians’ use of stigmatizing language in electronic health records (EHRs). Stigmatizing language is defined as language that communicates implicit or explicit biases, and it can perpetuate social hierarchies (Park et al., 2021). Researchers identified several categories of stigmatizing language in clinical documentation, including words or phrases that reinforce social risk factors without clinical justification, suggest that the patient is “difficult,” imply judgment regarding the patient’s decisions, and emphasize the clinician’s authority and knowledge (Barcelona et al., 2025). It may be harmful for patients to read stigmatizing language and can impede trust in clinicians, particularly because patients now have access to their clinical notes following the passage of the 21st Century Cures Act, which was enacted in 2016 to enhance inter-operability and access to electronic health information (Leonard et al., 2023). Importantly, Goddu et al. (2018) found that the use of stigmatizing language in clinical notes can transmit biases between clinicians, resulting in negative attitudes toward patients and inequitable treatment plans.

Literature Review

In prior studies, researchers found that stigmatizing language was more frequently documented in the clinical notes of patients with marginalized identities in U.S. society. For example, in one large hospital system, at least one clinical note for most patients with substance use disorder contained stigmatizing language (Weiner et al., 2023). In addition, stigmatizing language that labeled the patient as “difficult” or indicated disapproval was more likely to be documented in the clinical notes of Black patients compared with White patients in emergency settings (Sun et al., 2022) and internal medicine care settings (Bilotta et al., 2024).

Clinician bias and discrimination may contribute to high rates of adverse perinatal health outcomes among patients with disabilities.

Racial disparities in the use of stigmatizing language were also noted in an academic internal medicine practice: Black patients had higher odds of their medical records including quotes, judgment words, and language that implied doubt compared with White patients (Beach et al., 2021). These findings highlight how addiction stigma and racism contribute to health inequities. However, there is a dearth of quantitative research that measures how ableism may contribute to perinatal health inequities. We sought to address that knowledge gap in this study by examining the use of stigmatizing language categories as proxy measures of ableism and clinician bias. Understanding documentation patterns of stigmatizing language in the clinical notes of patients with disabilities is a critical first step to identify and mitigate ableism and clinician bias and to promote perinatal health equity. Therefore, the purpose of this study was to examine the association between patient disability status and use of stigmatizing language in clinical notes from the hospital admission for birth. We hypothesized that clinicians would be more likely to use stigmatizing language in the notes of patients with disabilities compared with patients without disabilities.

Methods

Design

We conducted a cross-sectional analysis using EHR data. The institutional review board at the Columbia University Medical Center approved the study procedures (IRB AAAT9870).

Setting

The study was conducted at two urban hospitals in the northeastern United States.

Participants

Participants (N = 19,094) included patients at more than 20 weeks gestation who were admitted for labor and birth from 2017 to 2019.

Researchers have identified several categories of stigmatizing language used in clinical notes that convey clinician bias and can lead to differential treatment.

Data Collection

We analyzed existing EHR data. We included 22 types of clinical notes that had free-text narrative sections such as the obstetric triage note, social work note, and nursing progress note. We excluded structured and templated clinical notes, including medication administration records and pain scales, from this analysis. The final analytic sample included 211,841 unique clinical notes.

Procedures

Patient disability status was the primary exposure under study. We operationally defined disability status as a binary variable (yes/no) according to the presence of ICD-10 (International Classification of Diseases, 10th revision) codes for disability conditions indicated in the EHR. The ICD-10 codes for disability we considered in this analysis were based on a list of ICD-9 codes from prior studies in which researchers measured physical, sensory, and intellectual and developmental disabilities in the EHR (Darney et al., 2017; Gleason et al., 2021). We employed a binary variable to indicate the presence of any disability because patients with physical, sensory, or intellectual and developmental disabilities reported similar experiences of ableism and clinician bias in perinatal health care (Evans et al., 2024; Harkins et al., 2025; Tarasoff et al., 2023). In addition, the number of patients within each disability category in our sample was too small to support inferential statistics. The ICD-10 codes we used to define disability status in our study are available in Supplemental Table S1.

The primary study outcome was the use of stigmatizing language, which we grouped into one or more categories: (a) marginalized language/identities, (b) difficult patient, (c) unilateral/authoritarian decisions, and (d) questioning patient credibility. These language categories are based on prior qualitative analyses of clinical notes from the hospital birth admission (Barcelona, Scharp, et al., 2023; Barcelona et al., 2025). Briefly, the marginalized language/identities category is defined as unjustified documentation of social risk factors that may contribute to marginalization when repeated in free-text narratives. The difficult patient category refers to words or phrases implying that patients or their symptoms are inconvenient to the clinician, whereas the unilateral/authoritarian decisions category includes language that centers on the decisions and knowledge of the clinician instead of promoting patient autonomy and education. The questioning patient credibility category contains language that expresses doubt or dismissal of the patient’s reported symptoms or health histories, and it is often characterized by patient quotes.

We used our well-performing natural language processing model to identify the presence of any stigmatizing language in clinical notes and by each category (Scroggins et al., 2025). Natural language processing is a computational method that is used to analyze and extract meaningful information from human language text within clinical documentation. For this study, we employed an advanced transformer-based language model, ClinicalBERT (Bidirectional Encoder Representations from Transformers; Alsentzer et al., 2019), which demonstrated better performance compared with other models we tested (e.g., BERT, random forest, decision trees) to capture nuanced and context-dependent language patterns in clinical notes. This model was pretrained on large volumes of clinical text and fine-tuned to identify specific categories of stigmatizing language. Unlike simpler rule-based approaches that rely solely on predetermined keywords, our model can understand the contextual meaning of text, which makes it particularly effective for identifying subtle forms of stigmatizing language. The natural language processing model was trained on a gold standard dataset of manually annotated clinical notes and achieved an average F1 score of .78, which indicates strong performance in accurately identifying stigmatizing language categories. We provide further details about the development of the natural language processing model in Supplemental Table S2.

Analysis

We selected covariates based on theory and demographic factors associated with stigmatizing language in prior research (Hulchafo et al., 2025), including health insurance type (Medicaid or private insurance), marital status (single, married, divorced, widowed, or unspecified), age in years (13–19, 20–34, or $35), and race and ethnicity (American Indian or Alaskan Native, Asian or Pacific Islander, Black, Hispanic, multiracial, or White). Race and ethnicity data were missing from 45% of patients in our study sample; therefore, we employed the Bayesian Improved First Name Surname Geocoding algorithm, an imputation method for race and ethnicity using first name, surname, and census block data with high accuracy in our EHR data (Scroggins et al., 2024).

We calculated descriptive statistics for all variables of interest. We then conducted multivariate logistic regressions to calculate unadjusted and adjusted odds ratios and 95% confidence intervals that described the association between patient disability status and stigmatizing language categories (i.e., any category, marginalized language/identities, difficult patient, unilateral/authoritarian decisions, and questioning patient credibility). We set statistical significance to an alpha of .05 and computed all analyses using Python in JupyterLab 3.0.

Results

The characteristics of the participants (N = 19,094) in our sample are presented in Table 1. We identified at least one ICD-10 code for disability in 550 participants (2.9%). Among those with a disability, most had a physical disability (n = 460, 83.6%), followed by an intellectual or developmental disability (n = 43, 7.8%), a sensory disability (i.e., blind/low vision or deaf/hard of hearing [n = 32, 5.8%]), and multiple disabilities (n = 15, 2.7%). In the total sample, most participants were of Hispanic ethnicity (n = 11,238, 58.9%), were 20 to 34 years of age (n = 13,129, 68.8%), and gave birth vaginally (n = 11,823, 61.9%) to a term infant (n = 16,843, 88.2%). Race and ethnicity data were missing for 121 (0.6%) participants. Approximately half of the participants were single (n = 10,897, 57.1%) and insured by Medicaid (n = 10,689, 56.4%). Slightly less than half of the participants were multiparous (n = 9,530, 49.9%).

Table 1:

Participant Characteristics by Disability Status (N = 19,094)

Total Patients With Disabilities Patients Without Disabilities
Characteristic n % n % n %
Total 19,094 100 550 2.9 18,544 97.1
Exposure
Disability type
 Physical disability 460 2.4 460 83.6 0 0.0
 Intellectual and developmental disability 43 0.2 43 7.8 0 0.0
 Sensory disability 32 0.2 32 5.8 0 0.0
 Multiple disabilities 15 <0.1 15 2.7 0 0.0
Outcomes
Stigmatizing language categories
 Any stigmatizing language 9,472 49.6 341 62.0 9,131 49.2
 Marginalized language/identities 1,738 9.1 49 8.9 1,689 9.1
 Difficult patient 5,525 28.9 240 43.6 5,285 28.5
 Unilateral/authoritarian decisions 4,973 26.0 170 30.9 4,803 25.9
 Questioning patient credibility 16 <0.1 0 0.0 16 <0.1
Demographic covariates and obstetric characteristics
Race and ethnicity
 Hispanic 11,238 58.9 277 50.4 10,961 59.1
 White 4,342 22.7 165 30.0 4,177 22.5
 Black 2,178 11.4 66 12.0 2,112 11.4
 Asian or Pacific Islander 1,189 6.3 38 6.9 1,151 6.2
 American Indian or Alaskan Native 10 0.1 1 0.2 9 0.1
 Multiracial 16 0.1 0 0.0 16 0.1
 Missing 121 0.6 3 0.5 118 0.6
Maternal age, years
 13–19 685 3.6 8 1.5 677 3.7
 20–34 13,129 68.8 368 66.9 12,761 68.8
 ≥35 5,280 27.7 174 31.6 5,106 27.5
Marital status
 Single 10,897 57.1 268 48.7 10,629 57.3
 Married 7,698 40.3 267 48.5 7,431 40.1
 Divorced 89 0.5 2 0.3 87 0.5
 Widowed 11 0.1 0 0.0 11 0.1
 Unspecified 359 1.9 12 2.2 347 1.9
 Missing 40 0.2 1 0.2 39 0.2
Insurance status
 Medicaid 10,689 56.0 274 49.8 10,415 56.2
 Private 8148 42.7 272 49.5 7,876 42.5
 Missing 257 1.4 4 0.7 253 1.4
Mode of birth
 Vaginal 11,823 61.9 296 53.8 11,527 62.2
 Cesarean 7271 38.1 254 46.2 7,017 37.8
Gestational age
 Term (<37 weeks) 16,843 88.2 469 85.3 16,374 88.3
 Preterm (≥37 weeks) 2,251 11.8 81 14.7 2,170 11.7
Stillbirth
 No 18,441 96.6 546 99.3 18,351 99.0
 Yes 653 3.4 4 0.7 193 1.0
Parity
 Nulliparous 8,212 43.0 266 48.4 7,946 42.9
 Multiparous 9,530 49.9 244 44.4 9,286 50.1
 Missing 1,352 7.1 40 7.2 1,312 7.1
Open in a new tab

We present exemplars of stigmatizing language found in the clinical notes of participants with disabilities in Table 2. We selected the marginalized language/identities exemplars for the inclusion of outdated and ableist terminology such as “mental retardation,” “social disarray,” and “psych issues.” In the exemplars for the difficult patient language category, blame is placed on the participant for “refusing” services or education and “complaining” of pain despite being in labor. The unilateral/authoritarian decisions category exemplars reinforce the clinicians’ authority and knowledge by implying judgment on how the participant cared for the infant (e.g., “needed to be counseled”) and mentioning plans to contact CPS. The questioning patient credibility category exemplars include patient quotes that suggest disbelief in the participant’s self-reported health history or disapproval of the participant’s actions.

Table 2:

Definitions and Exemplars of Stigmatizing Language Categories Among Participants With Disabilities

Category Definition Exemplars
Marginalized language/identities Unjustified documentation of social risk factors in narrative notes that may contribute to marginalization “Autism–high functioning but required early intervention/special needs-psych issues”
“Mild mental retardation with ‘personality disorder’ and depressive disorder—not currently medicated, not followed by Psych”
“Social disarray/cognitive impairment/drug use/teenage pregnancy with minimal care”
Difficult patient Language that implies the patient is problematic, noncompliant, or the patient’s symptoms or requests are inconvenient to clinicians “Patient refused sign language interpreter for entire night shift”
“Observed pt [patient] straight cath [catheter] technique and discovered that she is not using sterile techniques AT ALL (does not wash hands, blindly sticks catheter all over introitus until it finds urethra)—refused to go through teaching”
“Complains of uterine contractions and repeatedly was spoken to her about the reasons, also offered and gave her medication”
Unilateral/authoritarian decisions Language that centers the decisions and opinions of the clinician and minimizes patient autonomy “Pt picked up newborn without supporting the newborn’s neck and had to be counseled. Also, pt seemed unaware that she should attend to newborn when the newborn was crying and in obvious distress during the interview.”
“Patient and spouse are aware that ACS [adult and child protective services] will be contacted once patient delivers”
“Instructed that it’s unsafe practice [to hold infant while sleeping], but patient got irritated. R.N. [registered nurse] told patient that there was incidence of baby fall, but in spite of the teaching she continues to sleep with infant. She became irritated every time with the safety reminder [and] is also annoyed every time a procedure is done l ike VS [vital signs] is taken.”
Questioning patient credibility Expressing doubt or dismissal of patients’ reported symptoms or health histories “She was told that ‘her pelvis is too small for a vaginal delivery’ and taken for cesarean delivery”
“Pt self d/c’ed [discharged (medication)] due to finding out she was pregnant and did not want ‘effects of the meds to effect baby’ given she has 2 special needs children while on medication”
“Patient does share that [she] has been at Mommy and Me programs; however, gave different reasons as to why she left each facility”
Open in a new tab

We also present the frequencies of stigmatizing language categories in Table 1. We identified the use of at least one stigmatizing language category for nearly half of the participants (n = 9,472, 49.6%). In addition, for approximately one third of the sample, we identified language that represented the difficult patient (n = 5,525, 28.9%) and unilateral/authoritarian (n = 4,973, 26.0%) categories. When we examined frequencies by disability status, we found stigmatizing language in the clinical notes of nearly two thirds of participants with disabilities (n = 341, 62.0%) compared with slightly less than half of the participants without disabilities (n = 9,131, 49.2%). The difficult patient category was the most common type of stigmatizing language used for participants with disabilities (n = 240, 43.6%) and without disabilities (n = 5,285, 28.5%). The frequency of the marginalized language/identities category was similar between participants with disabilities (n = 49, 8.9%) and without disabilities (n = 1,689, 9.1%). We identified language that indicated questioning of patient credibility in less than 1% of clinical notes for participants with disabilities (n = 0) and without disabilities (n = 16).

We report results from the unadjusted and multivariable analyses of participant disability status and stigmatizing language in Table 3. In unadjusted analyses, the clinical notes of participants with disabilities were significantly more likely to include any stigmatizing language compared with the clinical notes of participants without disabilities (odds ratio [OR] = 1.68, 95% confidence interval [CI] = [1.41, 2.00]). The clinical notes of participants with disabilities were also significantly more likely to include difficult patient language (OR = 1.94, 95% CI = [1.64, 2.31]) and unilateral/authoritarian decisions language (OR = 1.28, 95% CI = [1.07, 1.54]) compared with the clinical notes of participants without disabilities. We found no significant differences by disability status for the marginalized language/identities category (OR = 0.98, 95% CI = [0.73, 1.31]).

Table 3:

Unadjusted and Adjusted Logistic Regression Models Examining Stigmatizing Language Categories by Participant Disability Status

95% CI 95% CI
Stigmatizing Language Category OR LL UL p aORa LL UL p
Stigmatizing language, any
 Patients with disabilities 1.68 1.41 2.00 <.001 1.75 1.47 2.09 <.001
 Patients without disabilities Reference Reference
Marginalized language/identities
 Patients with disabilities 0.98 0.72 1.31 .87 1.19 0.87 1.62 .27
 Patients without disabilities Reference Reference
Difficult patient
 Patients with disabilities 1.94 1.64 2.31 <.001 1.96 1.65 2.33 <.001
 Patients without disabilities Reference Reference
Unilateral/authoritarian decisions
 Patients with disabilities 1.28 1.07 1.54 .01 1.27 1.06 1.53 .01
 Patients without disabilities Reference Reference
Questioning patient credibilityb
 Patients with disabilities – –
 Patients without disabilities Reference Reference
Open in a new tab

Note. OR = odds ratio; CI = confidence interval; aOR = adjusted odds ratio; LL = lower limit; UL = upper limit.

a

Adjusted for patient insurance type, marital status, age, race, and ethnicity.

b

There was an insufficient number of notes with language for the questioning patient credibility category to provide estimates.

Our findings remained consistent after adjustment for covariates. The clinical notes of participants with disabilities were significantly more likely to include any stigmatizing language compared with the clinical notes of participants without disabilities (adjusted odds ratio [aOR] = 1.75, 95% CI = [1.47, 2.09]). The clinical notes of participants with disabilities were also significantly more likely to include difficult patient language (aOR = 1.96, 95% CI = [1.65, 2.33]) and language from the unilateral/authoritarian decisions category (aOR = 1.27, 95% CI = [1.06, 1.53]). The odds that clinical notes included language from the marginalized language/identities category increased for participants with disabilities after adjustment for covariates; however, this was not statistically significant (aOR = 1.19, 95% CI = [0.87, 1.62]).

Discussion

The purpose of our study was to examine the association between disability status and the use of stigmatizing language in clinical notes from the hospitalization for birth. We observed that the clinical notes of participants with disabilities were more likely to include stigmatizing language than the notes of participants without disabilities. Specifically, words or phrases used to label the participants as “difficult” and to reinforce the unilateral/authoritarian decisions of the clinician were more likely to be included in the clinical notes of participants with disabilities compared with participants without disabilities.

Our findings support those of previous researchers who studied disability and stigmatizing language in clinical practice. Despite the passage of Rosa’s Law in 2010 (Friedman, 2016), which required changing all references of “mental retardation” in federal health, education, and labor policy to “intellectual disability,” clinicians have continued to use that outdated phrase and other ableist terms such as “special needs,” “psychotic,” and “suffering” to describe patients with disabilities (Agaronnik et al., 2019; Lagu et al., 2022). In addition, the use of stigmatizing language has been shown to influence clinician attitudes and care practices toward patients with disabilities and chronic conditions. For example, in a study of 413 medical students and residents, Goddu et al. (2018) found that physicians-in-training exposed to stigmatizing language in a clinical note were more likely to report negative attitudes and less aggressive pain management compared with those exposed to neutral language. Similarly, Glassberg et al. (2013) reported that among 655 emergency physicians, those who used the stigmatizing term “sickler” frequently (43.3%) or always (8.7%) were more likely to have negative attitudes toward the care of patients with sickle cell disease compared with those who never used this language (13.1%).

Stigmatizing language that labeled them as “difficult” and minimized their autonomy occurred more frequently in the clinical notes of patients with disabilities.

We expand on prior research findings by providing novel insights into how language may negatively influence perceptions of care among patients with disabilities during labor and birth. For example, we found that the records of participants with disabilities were more likely to include language that labeled them as “difficult” and indicated “complains of pain.” This result aligns with qualitative findings in which patients with disabilities reported that their pregnancy and labor pain concerns were repeatedly dismissed by clinicians, which resulted in traumatic and painful childbirth experiences (Long-Bellil et al., 2017; Smeltzer et al., 2017).

In addition, we found that the records of participants with disabilities were more likely to include unilateral/authoritarian language. This category includes threats of CPS involvement (Barcelona, Scharp, et al., 2023; Barcelona et al., 2025), and we identified harmful exemplars of this language in our study. Our finding supports recent reports that patients with disabilities are disproportionately represented in CPS investigations, out-of-home care, and termination of parental rights (LaLiberte et al., 2024). Rates differ depending on the type of disability, with patients with emotional or behavioral disorders, learning disabilities, or intellectual and developmental disabilities overrepresented in CPS investigations at a rate of two to six times higher than patients without disabilities (LaLiberte et al., 2024). These investigations have been described as harmful by patients with disabilities, and some have withheld disclosing their disability status to clinicians out of fear of being referred to CPS (Harkins et al., 2025). Documenting notices or threats to CPS may perpetuate perinatal health disparities given that patients with disabilities have reported avoiding essential services, such as postpartum mental health care and social services, to prevent a potential CPS investigation (Harkins et al., 2025).

It is also concerning that the clinical notes of nearly half of the participants included stigmatizing language from the birth hospitalization regardless of their disability status. This finding highlights how birth equity is necessary to improve care and outcomes for all patients, particularly those with marginalized identities. For example, in a recent study we found that the birth notes of Black patients were more likely to include stigmatizing language than those of White patients (Hulchafo et al., 2025). In addition, it has been well-documented that racially and ethnically minoritized patients, patients with higher body weights, patients with substance use disorders, and single mothers encounter bias and discrimination from clinicians in perinatal care settings (Barcelona, Horton, et al., 2023). Therefore, it is critically important to address systemic bias toward patients with any marginalized identity to advance perinatal health equity for all.

Implications

We provide several recommendations to improve clinician documentation practices for patients with disabilities. Importantly, disability-inclusive language guidelines are nuanced, and asking all patients which terms they prefer to use is essential (Lepard et al., 2024). Person-first language is commonly recommended in clinical documentation (Vanka et al., 2025), and this type of language is used in the Convention on the Rights of Persons with Disabilities (United Nations, 2006). However, identity-first language is preferred by some patients with specific disabilities, including those who are deaf/hard of hearing and those who are autistic (Andrews et al., 2022; Taboas et al., 2023). In addition, clinicians should document the term “disability” and avoid euphemisms (e.g., “special needs,” “differently abled”) because the alternatives were developed and used by those without disabilities and are offensive and patronizing (Andrews et al., 2019, 2022). Similarly, clinicians should avoid making subjective assessments of disabilities (e. g., “probable learning disability”) and must first consult with patients about a new diagnosis of a disability before documenting the condition in the clinical note. This approach ensures that patients are fully informed and involved in their own health care knowledge and decisions. Furthermore, this aligns with recommendations by clinicians to support and maximize autonomy for patients with disabilities (Barbera et al., 2024).

In addition, recent guidelines for patient-centered documentation should be followed to improve clinical documentation practices for all patients, including those with disabilities (Vanka et al., 2025). For example, through a series of focus groups that included physicians experienced with writing open notes, patients accustomed to reviewing their notes, medical student educators, and resident physicians, Vanka et al. (2025) identified 10 discrete themes that can help clinicians to use and spread patient-centered documentation. Focus group attendees recommended that clinicians document objectively and avoid judgmental terms such as “argumentative” and descriptors such as “pleasant” and to consider whether comments on personality and appearance are helpful in providing high-quality clinical care (Vanka et al., 2025). Clinicians also should avoid phrases that imply skepticism of the patient’s perspective, including “patient claims” and “patient denies,” and words that convey bias or judgment such as “refused” (Ferná ndez et al., 2021) and “noncompliant”” (Vanka et al., 2025). These commonly documented terms place blame on the patient and do not take into consideration other factors that affect patient care such as financial stability and adequate insurance (Fernández et al., 2021; Vanka et al., 2025). In sum, there is an urgent need to incorporate patient-centered documentation practices into education for health care professionals to support trust in clinicians and improve perinatal health outcomes for patients with disabilities (Havercamp et al., 2021).

More research is needed to determine whether stigmatizing language is associated with adverse pregnancy outcomes. Although use of stigmatizing language undermines the patient–provider relationship for many reasons, demonstrating its link to pregnancy outcomes would provide critical evidence of how bias contributes to perinatal health inequities. In addition, clinician training on patient-centered documentation and the potential use of artificial intelligence to remove stigmatizing language in clinical notes are important steps toward more equitable documentation. However, language encompasses more than written documentation given that it also reflects verbal communication and behaviors that effect pregnancy care, which we previously discussed (Barcelona, Horton, et al., 2023). Therefore, documenting with more neutral language is unlikely to eliminate all stigmatizing care practices, and perinatal inequities will persist without more comprehensive efforts.

Instead, we recommend that stigmatizing language should be viewed as a marker of bias and an opportunity for clinician training and self-reflection to improve the quality of perinatal care. For example, artificial intelligence tools may flag stigmatizing language in real time to encourage clinicians to recognize and address their underlying biases. In addition, patient-centered care is an essential practice to mitigate stigmatizing behaviors toward patients during pregnancy. This type of care prioritizes clear communication, shared decision-making, and patient autonomy, and it was associated with positive physical and mental health in the postpartum period (Attanasio et al., 2022). Ultimately, clinicians must recognize the influence of their biases and take deliberate action to deliver compassionate patient-centered care that promotes perinatal health equity.

Limitations

The generalizability of our findings to patients with disabilities who receive labor and birth care outside of the urban northeastern United States is limited. In addition, we defined disability based on ICD-10 codes present in the EHR at the time of labor and birth. It is possible that we missed some participants with disabilities without documented ICD-10 codes. In addition, the relatively low prevalence of disability in the study precluded us from examining differences in stigmatizing language documentation by the type of disability. Finally, it is possible that our natural language processing algorithm did not detect all instances of stigmatizing language, particularly for the questioning patient credibility category, which is more nuanced, and no specific words or phrases exemplify this category. Future researchers should expand on this work by using self-report measures of disability, determining differences in the use of stigmatizing language by disability type, and examining associations between stigmatizing language and adverse perinatal health outcomes among patients with disabilities.

Conclusion

We identified that stigmatizing language was more likely to be documented in the clinical notes of participants with disabilities compared with participants without disabilities during the hospital admission for birth. The clinical notes of participants with disabilities were more likely to include language that labeled them as “difficult” and language that supported the unilateral and authoritarian decisions of the clinician. These findings highlight the need to promote patient-centered documentation and care practices to mitigate clinician bias and improve perinatal health outcomes for all. More research is needed to explore differences in the use of stigmatizing language by disability type and to examine associations between ableism and adverse perinatal health outcomes.

Supplementary Material

Supplementary files
Author's perspective
Download video file (153.9MB, mov)
1

Note: To access the supplementary materials that accompany this article, visit the online version of the Journal of Obstetric, Gynecologic, & Neonatal Nursing at http://jognn.org and at https://doi.org/10.1016/j.jogn.2025.08.003.

FUNDING

Supported by Columbia University Data Science Institute Seed Funds and a Gordon and Betty Moore Foundation grant (GBMF9048). Sarah E. Harkins is supported by the National Institute of Nursing Research training grant Reducing Health Disparities Through Informatics (T32NR007969).

Footnotes

CONFLICT OF INTEREST

The authors report no conflicts of interest or relevant financial relationships.

Contributor Information

Sarah E. Harkins, Obstetrics and Gynecology, Columbia University Irving Medical Center, New York, NY..

Ismael I. Hulchafo, School of Nursing, Columbia University, New York, NY..

Jihye Kim Scroggins, School of Nursing, University of North Carolina at Chapel Hill, Chapel Hill, NC..

Caroline Walsh, Quincy, Massachusetts..

Meghan Didier, Sisters by Heart, Valley Center, CA..

Maxim Topaz, School of Nursing, Columbia University, New York, NY..

Veronica Barcelona, School of Nursing, Columbia University, New York, NY..

REFERENCES

  1. Agaronnik N, Campbell EG, Ressalam J, & Iezzoni LI (2019). Exploring issues relating to disability cultural competence among practicing physicians. Disability and Health Journal, 12 (3), 403–410. 10.1016/j.dhjo.2019.01.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Alsentzer E, Murphy JR, Boag W, Weng WH, Jin D, Maumann T, & McDermott M (2019). Publicly available clinical BERT embeddings [preprint]. arXiv:1904.03323. [Google Scholar]
  3. Andrews EE, Forber-Pratt AJ, Mona LR, Lund EM, Pilarski CR, & Balter R (2019). #SaytheWord: A disability culture commentary on the erasure of “disability.” Rehabilitation Psychology, 64(2), 111–118. 10.1037/rep0000258 [DOI] [PubMed] [Google Scholar]
  4. Andrews EE, Powell RM, & Ayers K (2022). The evolution of disability language: Choosing terms to describe disability. Disability and Health Journal, 15(3), Article 101328. 10.1016/j.dhjo.2022.101328 [DOI] [PubMed] [Google Scholar]
  5. Attanasio LB, Ranchoff BL, Paterno MT, & Kjerulff KH (2022). Person-centered maternity care and health outcomes at 1 and 6 months postpartum. Journal of Women’s Health, 31(10), 1411–1421. 10.1089/jwh.2021.0643 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Barbera JP, Cichon B, Ankam N, & Schwartz BI (2024). Equitable care for patients with disabilities: Considerations for the gynecologic health care professional. Obstetrics & Gynecology, 143 (4), 475–483. 10.1097/aog.0000000000005493 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Barcelona V, Horton RL, Rivlin K, Harkins S, Green C, Robinson K, … Topaz M (2023). The power of language in hospital care for pregnant and birthing people: A vision for change. Obstetrics & Gynecology, 142(4), 795–803. 10.1097/AOG.0000000000005333 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Barcelona V, Scharp D, Idnay BR, Moen H, Goffman D, Cato K, & Topaz M (2023). A qualitative analysis of stigmatizing language in birth admission clinical notes. Nursing Inquiry, 30 (3), Article e12557. 10.1111/nin.12557 [DOI] [PubMed] [Google Scholar]
  9. Barcelona V, Scroggins JK, Scharp D, Harkins SE, Goffman D, Aubey J, & Topaz M (2025). Secondary qualitative analysis of stigmatizing and nonstigmatizing language used in hospital birth settings. Journal of Obstetric, Gynecologic, and Neonatal Nursing, 54(1), 112–122.e4. 10.1016/j.jogn.2024.10.003 [DOI] [PubMed] [Google Scholar]
  10. Beach MC, Saha S, Park J, Taylor J, Drew P, Plank E, … Chee B (2021). Testimonial injustice: Linguistic bias in the medical records of Black patients and women. Journal of General Internal Medicine, 36(6), 1708–1714. 10.1007/s11606-021-06682-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Bilotta I, Tonidandel S, Liaw WR, King E, Carvajal DN, Taylor A, … Hansen M (2024). Examining linguistic differences in electronic health records for diverse patients with diabetes: Natural language processing analysis. JMIR Medical Informatics, 12, Article e50428. 10.2196/50428 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Darney BG, Biel FM, Quigley BP, Caughey AB, & Horner-Johnson W (2017). Primary cesarean delivery patterns among women with physical, sensory, or intellectual disabilities. Women’s Health Issues, 27(3), 336–344. 10.1016/j.whi.2016.12.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Dev A, Horner-Johnson W, Schaefer A, Ganduglia-Cazaban C, Stukel TA, Goodman DC, & Leyenaar JK (2025). Identifying pregnant women with disabilities and maternal and newborn outcomes. JAMA Network Open, 8(3), Article e252159. 10.1001/jamanetworkopen.2025.2159 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Evans M, Tarasoff LA, Lunsky Y, Welsh K, Proulx L, Havercamp SM, … Brown HK (2024). Disability justice and collective access to labour and delivery care: A qualitative study. BMC Pregnancy and Childbirth, 24(1), Article 832. 10.1186/s12884-024-07036-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Fernández L, Fossa A, Dong Z, Delbanco T, Elmore J, Fitzgerald P, … DesRoches C (2021). Words matter: What do patients find judgmental or offensive in outpatient notes? Journal of General Internal Medicine, 36(9), 2571–2578. 10.1007/s11606-020-06432-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Friedman C (2016). Outdated language: Use of “mental retardation” in Medicaid HCBS waivers post-Rosa’s Law. Intellectual and Developmental Disabilities, 54(5), 342–353. 10.1352/1934-9556-54.5.342 [DOI] [PubMed] [Google Scholar]
  17. Glassberg J, Tanabe P, Richardson L, & Debaun M (2013). Among emergency physicians, use of the term “sickler” is associated with negative attitudes toward people with sickle cell disease. American Journal of Hematology, 88(6), 532–533. 10.1002/ajh.23441 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Gleason JL, Grewal J, Chen Z, Cernich AN, & Grantz KL (2021). Risk of adverse maternal outcomes in pregnant women with disabilities. JAMA Network Open, 4(12), Article e2138414. 10.1001/jamanetworkopen.2021.38414 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Gleason JL, Grewal J, Chen Z, Cernich AN, & Grantz KL (2023). Risk of adverse neonatal outcomes among pregnant women with disabilities. International Journal of Epidemiology, 52(1), 203–213. 10.1093/ije/dyac183 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Goddu AP, O’Conor KJ, Lanzkron S, Saheed MO, Saha S, Peek ME, … Beach MC (2018). Do words matter? Stigmatizing language and the transmission of bias in the medical record. Journal of General Internal Medicine, 33(5), 685–691. 10.1007/s11606-017-4289-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Harkins SE, Hahn AL, Didier M, Walsh C, Barcelona V, Melton K, & George M (2025). A qualitative meta-synthesis of the perinatal healthcare experiences of people with disability. Disability and Health Journal, 18(3), Article 101828. 10.1016/j.dhjo.2025.101828 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Havercamp SM, Barnhart WR, Robinson AC, & Whalen Smith CN (2021). What should we teach about disability? National consensus on disability competencies for health care education. Disability and Health Journal, 14(2), Article 100989. 10.1016/j.dhjo.2020.100989 [DOI] [PubMed] [Google Scholar]
  23. Horner-Johnson W, Dissanayake M, Marshall N, & Snowden JM (2022). Perinatal health risks and outcomes among US women with self-reported disability, 2011–19. Health Affairs, 41(10), 1477–1485. 10.1377/hlthaff.2022.00497 [DOI] [PubMed] [Google Scholar]
  24. Hulchafo II, Scroggins JK, Harkins SE, Moen H, Tadiello M, Cato K, … Barcelona V (2025). Stigmatizing and positive language in birth clinical notes associated with race and ethnicity. JAMA Network Open, 8(5), Article e259599. 10.1001/jamanetworkopen.2025.9599 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Lagu T, Haywood C, Reimold K, DeJong C, Walker Sterling R, & Iezzoni LI (2022). “I am not the doctor for you”: Physicians’ attitudes about caring for people with disabilities. Health Affairs, 41(10), 1387–1395. 10.1377/hlthaff.2022.00475 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. LaLiberte T, Piescher K, Mickelson N, & Hwa Lee M (2024). The overrepresentation of parents with disabilities in child protection. Children and Youth Services Review, 158, Article 107446. 10.1016/j.childyouth.2024.107446 [DOI] [Google Scholar]
  27. Leonard SM, Zackula R, & Wilcher J (2023). Attitudes and experiences of clinicians after mandated implementation of open notes by the 21st Century Cures Act: Survey study. Journal of Medical Internet Research, 25, Article e42021. 10.2196/42021 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Lepard T, Dugan M, Cleveland E, & Menke C (2024). Clinical genetic counselors’ use of people- and identity-first language in regard to patients’ identification with disability. Journal of Genetic Counseling, 34(2), Article e1972. 10.1002/jgc4.1972 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Long-Bellil L, Mitra M, Iezzoni LI, Smeltzer SC, & Smith LD (2017). Experiences and unmet needs of women with physical disabilities for pain relief during labor and delivery. Disability and Health Journal, 10(3), 440–444. 10.1016/j.dhjo.2017.02.007 [DOI] [PubMed] [Google Scholar]
  30. Lundberg DJ, & Chen JA (2024). Structural ableism in public health and healthcare: A definition and conceptual framework. The Lancet Regional Health—Americas, 30, Article 100650. 10.1016/j.lana.2023.100650 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Mitra M, Long-Bellil LM, Iezzoni LI, Smeltzer SC, & Smith LD (2016). Pregnancy among women with physical disabilities: Unmet needs and recommendations on navigating pregnancy. Disability and Health Journal, 9(3), 457–463. 10.1016/j.dhjo.2015.12.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Park J, Saha S, Chee B, Taylor J, & Beach MC (2021). Physician use of stigmatizing language in patient medical records. JAMA Network Open, 4(7), Article e2117052. 10.1001/jamanetworkopen.2021.17052 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Saeed G, Brown HK, Lunsky Y, Welsh K, Proulx L, Havercamp S, & Tarasoff LA (2022). Barriers to and facilitators of effective communication in perinatal care: A qualitative study of the experiences of birthing people with sensory, intellectual, and/or developmental disabilities. BMC Pregnancy and Childbirth, 22(1), Article 364. 10.1186/s12884-022-04691-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Scroggins JK, Hulchafo II, Harkins S, Scharp D, Moen H, Davoudi A, … Barcelona V (2025). Identifying stigmatizing and positive/preferred language in obstetric clinical notes using natural language processing. Journal of the American Medical Informatics Association, 32(2), 308–317. 10.1093/jamia/ocae290 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Scroggins JK, Hulchafo II, Topaz M, Cato K, & Barcelona V (2024). Addressing bias in preterm birth research: The role of advanced imputation techniques for missing race and ethnicity in perinatal health data. Annals of Epidemiology, 94, 120–126. 10.1016/j.annepidem.2024.05.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Smeltzer SC, Wint AJ, Ecker JL, & Iezzoni LI (2017). Labor, delivery, and anesthesia experiences of women with physical disability. Birth, 44(4), 315–324. 10.1111/birt.12296 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Sun M, Oliwa T, Peek ME, & Tung EL (2022). Negative patient descriptors: Documenting racial bias in the electronic health record. Health Affairs, 41(2), 203–211. 10.1377/hlthaff.2021.01423 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Taboas A, Doepke K, & Zimmerman C (2023). Preferences for identity-first versus person-first language in a US sample of autism stakeholders. Autism, 27(2), 565–570. 10.1177/13623613221130845 [DOI] [PubMed] [Google Scholar]
  39. Tarasoff LA, Lunsky Y, Welsh K, Proulx L, Havercamp SM, Parish SL, & Brown HK (2023). Unmet needs, limited access: A qualitative study of postpartum health care experiences of people with disabilities. Journal of Advanced Nursing, 79(9), 3324–3336. 10.1111/jan.15642 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. United Nations. (2006). Convention on the Rights of Persons With Disabilities. https://www.un.org/disabilities/documents/convention/convoptprot-e.pdf [DOI] [PubMed]
  41. van der Veen S, Evans N, Huisman M, Welch Saleeby P, & Wid-dershoven G (2023). Toward a paradigm shift in healthcare: Using the International Classification of Functioning, Disability and Health (ICF) and the capability approach (CA) jointly in theory and practice. Disability and Rehabilitation, 45(14), 2382–2389. 10.1080/09638288.2022.2089737 [DOI] [PubMed] [Google Scholar]
  42. Vanka A, Johnston KT, Delbanco T, DesRoches CM, Garcia A, Salmi L, & Blease C (2025). Guidelines for patient-centered documentation in the era of open notes: Qualitative study. JMIR Medical Education, 11, Article e59301. 10.2196/59301 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Weiner SG, Lo YC, Carroll AD, Zhou L, Ngo A, Hathaway DB, … Wakeman SE (2023). The incidence and disparities in use of stigmatizing language in clinical notes for patients with substance use disorder. Journal of Addiction Medicine, 17(4), 424–430. 10.1097/adm.0000000000001145 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary files
NIHMS2127704-supplement-Supplementary_files.docx (28KB, docx)
Author's perspective
Download video file (153.9MB, mov)
1
NIHMS2127704-supplement-1.pdf (264KB, pdf)

RESOURCES