Words about people

Words for people should describe their circumstances and relationships accurately and respectfully, without reducing them to a diagnosis or a group label.

Inclusive language

Inclusive language avoids words and grammatical structures that may exclude individuals or groups, deliberately or inadvertently. ^{1, 2} It acknowledges diversity, conveys respect, recognises differences and promotes equal opportunities. ³ It also challenges the assumption that a cisgender, heterosexual, able-bodied White man represents everyone. ²

Words convey stereotypes and biases as well as literal meanings, and choices of verbs and grammatical voice can affect who appears responsible for an action. ¹ Othering means viewing, treating or describing people as intrinsically different from or inferior to oneself, through an us-versus-them mentality. ⁴ Language that alienates people can deprive them of representation and exclude their experiences from others' understanding. ⁵ Medical writers can promote inclusive language in documents, visuals and interactions, giving space to people who have been marginalised or stereotyped. ^{6, 7}

Inclusive language can also affect research participation and data quality. ⁸ Generalisations and stigmatising response options may confuse participants, leave them unable to describe themselves accurately, or discourage them from taking part. ⁸ Inclusive wording increases the likelihood of participation by a diverse range of people. ⁸

The following principles guide inclusive wording. ^{5, 9, 6}

• Avoid stereotypes based on group membership, such as “Asians are good at maths”.
• Avoid presenting one group as normal and another as outside the norm, as in “transsexual and normal people”.
• Avoid descriptive euphemisms such as “differently abled”.
• Use updated names for countries, professions, languages and ethnicities, such as high-, low- and middle-income countries rather than first-, second- and third-world countries.
• Avoid passive constructions where they make people objects of others' actions or obscure responsibility, as in “she was beaten”.
• Use gender-neutral pronouns and collective nouns where appropriate, such as “they”, “one”, “people”, “individuals” and “humanity”, and occupational titles such as “Member of Congress”, “mail clerk”, “firefighter” and “police officer”.
• Consider whether adding or removing gender modifiers counters an assumption about who can hold a role, as in “male nurse” or “boss” rather than “female boss”.
• Use person-first wording, such as “Pat has schizophrenia” rather than “Pat is schizophrenic”.

^{5, 9, 6}

The person before the condition

People should not be defined by their diseases, ages or clinical data. ¹⁰ “People with diabetes” is preferable to “diabetics”; the American Diabetes Association recommends this distinction. ^{11, 12, 10, 13, 14} Similar wording avoids using “asthmatic”, “epileptic” or “hypertensive” as nouns. ¹³ A patient's toxicity is a condition, rather than the person's identity: “the patient's acitretin toxicity manifested as cheilitis” avoids describing the patient as “toxic”. ¹⁴

Person-first wording is recommended for patient education, although community preferences also matter. ¹¹ “A person with hearing loss” can be less stigmatising than “a person who is hearing impaired”, while many people use “Deaf” to name their community and culture. ¹¹ Examples proposed for patient education in 2015 paired potentially offensive terms with empathetic alternatives. ¹¹

Potentially offensive term	Empathetic alternative
Amputee	A person with an amputated limb
Bipolar man	A man with bipolar disorder
Cancer patient	A patient with cancer
Autistic child	A child with autism or a child who is on the autism spectrum
Wheelchair-bound woman	A woman who uses a wheelchair

¹¹

Age-specific descriptions such as “infant”, “child” or “adolescent” are more precise than “pediatric patient”. ^{10, 15} “Patients 80 years or older” is more precise and less patronising than “the elderly”. ¹⁶ In a survey of 803 Americans aged 50–75 reported in 1999, “the elderly” was the most objectionable label. ¹⁷ The term “elderly primigravida”, endorsed in 1959 for women aged 35 or older, illustrates how misleading the label can be. ^{16, 17}

Wording that diminishes or blames patients

Care should be described without turning patients into objects or assigning them blame for treatment outcomes. ^{12, 16} A condition is diagnosed; a patient is evaluated or examined. ¹³ Patients take or use drugs, receive a PCA pump, and have their regimens switched; they are not themselves placed on a pump or switched. ^{13, 15} In most instances, “untreated” or “unexposed” can replace “treatment-naive” or “naive”. ¹⁰

“Patient care” avoids the implication that patients themselves are being managed, rather than their blood pressure, disease, pain, care or discharge. ^{16, 10, 13, 18} “Study participants” or “study group” avoids reducing people to “clinical material”. ¹⁶ A case is an instance of disease rather than the person affected; “case” remains appropriate in statistical contexts such as incident cases or case-control studies. ^{16, 13, 19}

A treatment fails, rather than the patient receiving it. ^{16, 12, 20} “Patients whose cardiac resynchronization therapy failed” avoids the blame implied by “patients who failed cardiac resynchronization therapy”. ¹⁶ Similarly, a patient's blood pressure can remain above a target without the patient having “failed to achieve” it. ¹⁰ “Patients who do not respond to conservative therapies” is preferable to “recalcitrant patients”: “recalcitrant” implies resistance to authority or discipline, not lack of response to treatment. ^{21, 20} Nonresponse does not indicate a character or personality flaw. ²⁰

Some editorial recommendations go further than these distinctions. ^{22, 23} One medical editor recommends removing seven terms as disrespectful to patients. ^{22, 23}

1. “Patient management”.
2. “Clinical material”.
3. “Case” used for a patient.
4. “Present” used of a patient.
5. “Failure” attributed to a patient.
6. “Elderly”.
7. “Rule of thumb”.

²³

In this view, “a 37-year-old man with localized melanoma” avoids the power imbalance implied by “a man who presented with localized melanoma”; detail about the examination can be added when needed. ¹⁶ The objection to “rule of thumb” concerns nineteenth-century US court documents referring to wife beating with a switch no larger than a thumb, although the Oxford English Dictionary records the phrase from 1692 as a practical method and does not mention that legal use. ¹⁷ For the editor, the wording of the court documents was enough to reject the phrase. ¹⁷

Other recommendations from a 2008 discussion favoured “patients who did not follow the regimen” over “non-compliant”, “advanced maternal age” over “elderly primigravida”, “report a symptom” over “complain”, and “miscarriage” over “spontaneous abortion”. ¹² For transplantation, “deceased donors” was suggested as probably preferable to “brain-dead patients”, with attention to context. ¹² Gallows humour is inappropriate in journal articles. ¹²

Some editors also prefer “who” to “that” for people and groups of people, because they consider “that” depersonalising, while acknowledging that it need not cause confusion. ^{24, 12, 15}

People receiving care and people providing it

Relationship words should be used only where the relationship they imply exists. ²⁵ “Patient” describes someone receiving care from a health professional or institution; it does not automatically describe every person with a health condition. ^{26, 19} People staying in hospital during treatment are inpatients, while those treated without admission are outpatients. ²⁶

A client engages a professional's services, or receives services from an agency, sometimes without paying. ²⁷ A consumer buys or uses goods or services, and a customer purchases them from a provider. ²⁸ At a veterinary practice, the animal receives care as the patient while the owner engages the veterinarian as the client. ²⁷ A general audience may be more accurately described as a lay audience than as consumers or patients. ²⁸

The distinction between “patient” and “client” can depend on professional usage. ^{27, 29} One editor proposes “patient” for biomedical services and “client” for psychosocial services, but recommends following the profession's standard terminology where the two overlap. ²⁷ “Clients” can be appropriate for people receiving psychological care or taking part in treatment programmes for substance misuse or other disorders. ¹⁹

Commercial terms can also change how care relationships are perceived. ³⁰ In a critique of this vocabulary, “consumer”, “customer” and “provider” imply buying and selling rather than a humanistic relationship, and “provider” obscures professionals' individual expertise and teaching or nurturing roles. ³⁰

Some clinicians find “provider”, “covered lives”, “mid-level provider” and “customer” impersonal or disrespectful. ²⁹ One language columnist prefers “health care professional” to “health care provider”, which can suggest a health maintenance organisation, and stresses that nurses and physical therapists are professionals too. ³¹ A clinical writing guide advises against calling an anesthesiologist an anesthetist, a family physician a family practitioner or an orthopedic surgeon an orthopod. ²⁹

“Caregiver” avoids assuming that a child's primary carer is the mother or even a parent or legal guardian. ³² A caregiver can be a family member or paid helper who regularly looks after a child or someone who needs care. ³²

People taking part in research

“Participants”, “subjects”, “patients”, “clients” and “volunteers” are not interchangeable in clinical reporting. ³³ “Patients” is inappropriate for healthy volunteers in an early clinical study because they are healthy. ³³ Survey respondents and people in community research may likewise not be patients. ¹⁹ Specific descriptions such as children, adults, respondents or patients with breast cancer can provide greater precision. ¹⁹

By 2025, “participants” had increasingly replaced “subjects” over the preceding three decades. ^{34, 35} “Subject” can describe a person whose responses are studied, but also someone under authority or acted upon; those connotations can suggest a power imbalance. ³⁶ One person who spent two years in a clinical trial preferred “participant” because taking part required effort, while “subject” suggested that she was simply an object of study. ³⁷ Conversely, some writers consider “subject” a more honest acknowledgement of vulnerability and the need for research protections, or more specific than “participant”, which can also include investigators, coordinators and committee members. ³⁶

There is no single best term for everyone taking part in research, although “participants” is often the clearest and most respectful choice. ³⁸ The wording should acknowledge people's contributions, autonomy, humanity and individuality. ³⁸ Someone can be both an investigator's patient and a research subject when a doctor–patient relationship also exists. ³⁹

“Volunteer” describes someone who comes forward or takes part of their own free will. ^{33, 39} It may fit nontherapeutic research or a comparison group, but not people who could not consent or willingly participate, including people who have died or whose families consented on their behalf. ³⁶ The NIH defines a healthy volunteer as someone without known significant health problems who takes part in research to test an intervention and whose information can provide a comparison. ³⁶ The AMA Manual of Style's “control participant” need not be healthy: the person lacks at least some characteristics under study or does not receive the intervention, but provides a comparison. ³⁶

“Sample” may fit analyses of large deidentified datasets, established statistical terms or descriptions of study designs. ¹⁹ “Consumer” can occasionally describe people consuming internet information, but should not replace “patient”. ^{19, 36}

Sex, gender and sexual orientation

Sex refers to biological attributes associated with chromosomes, gene expression, hormone function and reproductive or sexual anatomy. ⁴⁰ Gender refers to socially constructed roles, behaviours and identities. ^{40, 4} Gender identity is a person's innermost sense of being male, female, neither, a blend of genders or no gender. ⁴ Sexual orientation concerns emotional, romantic or sexual attraction. ⁴ For sexual behaviour, descriptions such as “men who have sex with men” and “women who have sex with women” distinguish behaviour from identity. ^{41, 42}

The 2016 SAGER guidelines recommend careful use of “sex” and “gender” to avoid confusing the terms, and use “sex” for animals. ⁴³ JAMA Network interim guidance reported in 2023 similarly used “sex” for biological factors and “gender” for identity or psychosocial and cultural factors. ⁴⁴ For pregnancy research in which gender identity was not assessed, it recommended “pregnant participants”, “pregnant individuals” or “pregnant patients”, as appropriate. ⁴⁴

Pregnancy, menstruation or testicular cancer does not establish how a person identifies their gender. ⁴⁵ Not all women have ovaries, and not everyone with ovaries is a woman. ⁴⁵ Participant information about contraception should therefore describe the people, anatomy or physiological process actually relevant, rather than assume gender from anatomy. ⁴⁵

Race and ethnicity

Race and ethnicity terminology varies across countries, studies and clinical genetics practice. ⁴⁶ Race is a social construct historically associated with perceived physical characteristics such as skin colour and used as a tool of oppression. ⁴ Ethnicity can include nationality, tribal affiliation, religion, language and traditions. ⁴ The AMA Manual of Style's August 2021 revision treats racial and ethnic categories as social constructs rather than biological categories, with limited utility in research and practice but possible value in examining racism, disparities and inequities. ^{47, 48} It calls for accurate, clear and precise terminology, consistent and justified designations, and continued review as language and social norms change. ^{48, 49, 50}

Recommendations differ over collective terms. ^{1, 51, 52} Inclusive-language advice published in 2021 suggested replacing “non-white” with “people of colour”, “Black” or “Brown”, and a 2023 conference account offered “communities of color” as an alternative to “minority communities”. ^{1, 51} The AMA guidance summarised in 2023 instead cautions that terms such as “people of color” and “brown” may erase differences among groups, and recommends avoiding them unless used in data collection and then defining them where possible. ⁵²

The AMA guidance summarised in 2023 gives the following recommendations for terminology. ^{52, 53}

1. Capitalise racial and ethnic categories, including Black and White, except where inappropriate or inflammatory, as in “white supremacy”. List categories alphabetically, with “other” and “unknown” last, and define the categories within “other”. Be as specific as possible; where small numbers could identify participants, numbers and percentages may be withheld if this is documented. Query an undefined “other” group.
2. Use racial and ethnic terms as adjectives, such as “Asian patient” or “patients who are Black”, rather than as nouns.
3. Do not hyphenate geographic combinations such as “African American”, whether used as nouns or compound modifiers.
4. Avoid abbreviating categories unless space requires it, and explain abbreviations in the text, table footnotes or figure legends.
5. Avoid collective terms that carry negative connotations or obscure differences, including “mixed race”, “minorities” as a noun, “people of color”, “brown”, “Black, Indigenous, and people of color” (BIPOC), “Black, Asian, and minority ethnic” (BAME), and “Black and minority ethnic” (BME), unless used in data collection and defined where possible.
   • Prefer “multiracial” or “multiethnic” to “mixed race” when their component categories are defined or the terms were predefined for participants' self-selection.
   • Prefer specific categories to collective terms. Where collective terms are needed, use “racial and ethnic minority groups” or “individuals”; define “underserved” or “underrepresented populations” at first mention; give “minoritized” a noun to modify; and explain why “historically marginalized” applies.
   • Prefer “Hispanic”, “Latino”, “Latina”, “Latinx” or “Latine” to “Spanish speaking”. Follow author preference rather than reflexively exchanging these terms, and retain the terms prespecified in research.
   • Regional descent is acceptable when used in data collection, but a specific country or region is preferable when known and relevant, especially for Asian, Middle Eastern or North African ancestry. Do not add “American” automatically, although it fits people who identify as Arab American, Asian American or similar.
   • For Indigenous people of North America, “American Indian” or “Alaska Native” is generally preferred to “Native American”; “Indigenous” is also acceptable. Use appropriate designations elsewhere, such as Native Hawaiian and Pacific Islander, Indigenous peoples of Canada or Aboriginal people, and name the nation or people where appropriate.
   • Avoid collectively describing racial and ethnic minority groups as “non-White”. Name the groups compared; explain the rationale and included categories if a White/non-White comparison is justified.
   • Explain the rationale for a justified comparison limited to two racial or ethnic groups.

^{52, 53}

“Caucasian” refers to a region of Eurasia and should not be used as a synonym for White. ⁴⁹ “African descendant” has been proposed to convey origin, culture and traditions, but its suitability has been questioned where those traditions are not practised. ⁴⁹ CDC advice reported in 2022 favoured “disproportionately affected groups” over “high risk group”, “people from racial or ethnic groups” over “racial or ethnic groups”, and “people from racial and ethnic minority groups” over “minority”. ⁵⁴

Other inclusive alternatives

Inclusive terminology should distinguish behaviour, identity and the characteristic actually being described. ^{42, 51} “People at risk of developing diabetes” names the relevant risk more precisely than “at-risk patients”. ⁵¹ Examples include “person living with HIV” rather than “HIV-positive person”, “drug misuse” rather than “drug abuse”, and “negative” or “positive” drug tests rather than “clean” or “dirty” results. ⁴² Further suggested alternatives include “mental health condition” for “mental illness”, “intellectual or developmental disability” for “mental retardation”, and “wheelchair user” for “wheelchair-bound”. ⁴²

For sexual orientation and gender, alternatives include “gay”, “lesbian”, “bisexual”, “pansexual”, “asexual” or “queer” for identity; “transgender man” or “transgender woman” rather than “female-to-male” or “male-to-female”; “transgender” as an adjective rather than “transgendered”; and “sexual orientation” rather than “sexual preference” or “lifestyle”, which suggest a choice. ^{42, 51} Other examples are “people who identify as LGBTQ+” and “gender-affirming”, “gender-affirmation” or “gender-confirmation surgery”. ⁵¹ Neutral forms of address and relationship terms include “everyone”, “folks”, “spouse”, “partner” and “parent”. ⁴²

“Older adults” or “older people”, “people experiencing homelessness” and “people experiencing poverty” avoid collective labels such as “the elderly”, “the homeless” and “the poor”. ^{42, 55} “People with low income” is another suggested alternative to “needy” or “disadvantaged”. ⁵⁵ Preferences about “fat” and “obese” remain contested; some activists reclaim “fat”, so asking people how they wish to be described is recommended. ⁵⁵ “Person of short stature” or “little person” are alternatives to “dwarf” or “midget”. ⁵⁵

Other suggested alternatives include “Romani” rather than “gypsy”, and a specific country or nationality rather than a broad label where appropriate. ⁴² “Assigned female at birth” or “assigned male at birth” distinguishes assignment at birth from gender identity. ⁴² “Transgender people and cisgender people” avoids describing one group as normal, and “all genders” avoids assuming there are only two. ⁴²

Health-related alternatives include “someone with a drug problem or addiction”, “someone who misuses heroin”, “a person with Down syndrome”, “a person with dementia”, and “a person who cannot speak” or who uses synthetic speech. ⁴² “Died from an AIDS-related illness” or “complications from AIDS” is suggested in place of “died of AIDS”, and “has” or “lives with” a condition in place of “suffers from” it. ⁴²

Checking wording with the community

Preferred language should be checked with the people it describes rather than assumed. ⁵⁶ For example, the parent of a child with a rare condition may dislike “subject” and prefer “patient” or “study participant”. ⁵⁶ Patient-community discussions on social media can reveal sensitivities about words and imagery. ⁵⁶ A criticism of a magazine cover as “ableist and tone deaf” was itself criticised by a representative of the deaf community, illustrating differing sensitivities. ^{56, 57}

A wording check can ask whether a term is relevant, accurate, inclusive, respectful and thoughtful; if it fails any criterion, its inclusion and possible revision should be reconsidered. ⁵⁸ Practical steps include the following. ⁵⁸

• Respect self-identification and allow selection of more than one race or ethnicity where possible.
• Create an inclusive-language guide with alternatives to generalised terms such as “minorities” and “non-White”.
• Recognise stigmatising language such as “mixed race” and “at-risk patients”.
• Expand race, ethnicity and gender-identity response options in surveys.
• Replace “Other” with “A race or ethnicity not listed” or “A gender identity not listed”.
• Alphabetise survey options to avoid implying a hierarchy.

⁵⁸

Language and its acceptability change, and some formerly offensive terms, such as “queer”, have been reclaimed. ^{59, 55} Keeping terminology under review is therefore part of inclusive writing. ^{50, 55}