It surveys how this term is defined in the published literature — what the sources agree on, where they diverge and what remains contested — and carries its own numbered bibliography. It does not describe what we run. Programme policy is on the core pages.
Our own use of the word is often narrower than general usage; the short definition and the boundary are in the glossary, under Underserved. The review date below matters, because the literature moves.
A definitional review. Citations follow IEEE style; see References.
Abstract
Underserved is a claim about provision: a population receives less of some service than its need warrants. It is routinely used interchangeably with marginalised, vulnerable and disadvantaged, which name different things — exclusion from power, exposure to harm, relative deficit in resources — and license different interventions and different claims. This article separates the four, sets out the unresolved split between access-based and outcome-based definitions, describes how the term is operationalised administratively rather than theoretically, and reviews the literature arguing that the word itself misplaces responsibility. A national illustration is included as neutral context.
I. Definition
Underserved is not a property of people. It is a relation between a population and a service, incomplete until two variables are filled in: underserved by whom, and of what. A community can be underserved in health and adequately served in schooling; the term carries no information until the service is named.
The most workable definition in the peer-reviewed literature comes from health-services research, where the term is used with the greatest precision. A 2024 systematic review adopts it to mean “populations that are underrepresented or disengaged from medical research or services despite a disproportionately high healthcare burden” [2]. Two elements do the work: disproportion — provision is judged against need, not an absolute floor — and despite — the population’s burden establishes an entitlement that provision fails to meet. Neither survives when the term is used as a synonym for “poor”. In United States federal law the parallel construction is statutory: a medically underserved population is “the population of an urban or rural area designated by the Secretary as an area with a shortage of personal health services or a population group designated by the Secretary as having a shortage of such services” [7]. The disjunction returns in §IV — the law recognises that underservice can attach to a place or to a group, and that the two do not coincide.
II. Four terms that are not synonyms
A 2010 qualitative study of voluntary and community-sector practice records “vulnerable”, “transient”, “marginalised”, “refusers”, “hidden”, “forgotten populations”, “underserved”, “special populations” and “disadvantaged populations” all circulating as synonyms for one idea [1]. They are not synonyms.
Underserved — inadequate provision relative to need. A supply-side, service-delivery claim, as above [2], [7]. Remedy: provide the service, or provide it differently.
Marginalised — exclusion from power, participation and social position. The foundational concept analysis defines marginalization as “the process through which persons are peripheralized based on their identities, associations, experiences, and environment”, with seven properties: intermediacy, differentiation, power, secrecy, reflectiveness, voice and liminality [5]. An integrative review extends this to “a process through which persons are pushed to a position in society where both their perceived and objective distance to basic resources; including healthcare and social services, employment, quality education, food and water lead to an increased risk for poor health outcome” [6]. Marginalisation is structural and political; adding service capacity does not address it, because the mechanism is position, not supply.
Vulnerable — exposure and susceptibility to harm. The same review defines vulnerability as “a state of being exposed to and unprotected from health damaging environment”, and treats it as produced by marginalisation rather than as a starting condition: “The cumulative effect of both the creation of margins and living between cultures is the creation of vulnerabilities” [6]. It is the most contested of the four: in research ethics it has been criticised for functioning as a fixed label attached to whole subpopulations rather than as a situational, layered condition, the direction of travel being away from labelling groups and toward identifying which exposures apply to which people in which circumstances (§VI records how thinly this review can source that).
Disadvantaged — relative deficit in resources or outcomes, usually socio-economic. The most cleanly operationalised of the four, because it is defined by position in a distribution rather than by a mechanism. It is comparative by construction, says nothing about who caused the deficit, and — unlike the other three — is measurable without a theory of why.
The consequence is practical. If a population is underserved, the finding indicts a provider; if marginalised, a political settlement; if vulnerable, an exposure; if disadvantaged, it is a measurement and not yet an indictment at all. Writing one word while meaning another transfers the evidence for one claim onto a different claim.
III. The definitional crux: access, or outcome
Is a community underserved because provision is absent, or because results are poor despite provision? The answer changes who counts.
A. Access is not a single quantity
The most-used contemporary framework treats access as “the opportunity to identify, seek, reach, obtain, or use healthcare and to ensure the fulfillment of the needs for these services” [3], as reproduced in a scoping review of its applications [4]. It decomposes access into five supply-side dimensions — “approachability, acceptability, availability/accommodation, affordability, and appropriateness” — mirrored by five demand-side abilities: “to perceive, to seek, to reach, to pay, and to engage, in healthcare” [3], [4]; the review’s assessment is that the framework “is able to take into equal account both the health systems and the patient’s perspective with regard to access” [4]. This dissolves the naive version of the access question. A service can be available and unaffordable; affordable and unacceptable; acceptable and inappropriate to the need presented. “Provision exists” is not a defence against a finding of underservice.
B. The same decomposition appears in education
CESCR General Comment No. 13 (E/C.12/1999/10, 1999) states that education at all levels shall exhibit four “interrelated and essential features”, the first being availability: “functioning educational institutions and programmes have to be available in sufficient quantity within the jurisdiction of the State party”, with buildings, sanitation, safe drinking water, trained teachers on domestically competitive salaries and teaching materials [10]. The remaining three — accessibility, acceptability, adaptability — carry the test past whether provision exists at all. Only the availability paragraph was retrieved verbatim for this review (§VI).
The 2020 UNESCO Global Education Monitoring Report locates failure on the system side, listing among barriers to inclusion “differing understandings of the word inclusion, lack of teacher support, absence of data on those excluded from education, inappropriate infrastructure, persistence of parallel systems and special schools, lack of political will and community support, untargeted finance, uncoordinated governance, multiple but inconsistent laws, and policies that are not being followed through” [11]. The emphasised item is §IV’s measurement problem stated as a policy failure.
C. What the choice changes
An access definition counts populations with too few providers, schools or bandwidth. It is administratively tractable and it under-counts: a population covered on paper but poorly served in practice does not appear. An outcome definition counts populations whose results are worse than their need warrants — the “despite a disproportionately high healthcare burden” clause [2]. It catches the paper-covered case and over-reaches, since poor outcomes have causes other than service failure. Neither is self-evidently correct and the literature does not settle it; what is not defensible is switching between them mid-argument.
IV. Operationalisation: how the term is decided in practice
In applied settings underserved is designated administratively rather than defined theoretically. The definition in force is whatever the designation rule computes.
A. Designation schemes
The most formally specified scheme is the United States Medically Underserved Area / Medically Underserved Population designation, made under the statutory definition quoted in §I [7]. In operation, Medically Underserved Areas “may be an entire county or group of contiguous counties, a group of a county or civil divisions, or a group of urban census tracts”, while Medically Underserved Populations “may include specific groups of people who face economic, cultural, or linguistic barriers to health care”; the parallel Health Professional Shortage Area designation covers shortages of “primary care, dental, or mental health providers” [8]. Designation is computed against a composite index for a delineated service area — and the methodology is not settled. A 2007 proposal set out an explicit “replacement framework” for both designations, scoring primary-care utilisation and provider-distribution data to estimate “communities’ effective access to primary care” and reporting that it would reproduce “over 90% of current geographic and low-income population HPSA designations” [9] (abstract read only — §VI). That the replacement was benchmarked on how much of the existing designation it preserved shows how far the operative definition is a property of the formula rather than of the concept.
B. Area or population — and what area-based designation misses
The statute’s disjunction — an area designated as having a shortage, or a population group so designated [7] — is a concession built into the law: if geography were sufficient, the population-group route would be redundant. The hazard is ecological inference. A designation computed over an area assigns the area’s average to everyone in it, and so misclassifies in both directions: underserved people inside well-served areas are excluded, well-served people inside designated areas are included, and the coarser the geographic unit the larger the error. Any headline count of “the underserved” is therefore a count of designated units, not of people. Quantitative estimates of the misclassification rate were not obtained for this review (§VI).
C. A national illustration: Sri Lanka
A 2023 Multidimensional Vulnerability Index derived from a national citizen survey and published by UNDP with the University of Oxford’s Poverty and Human Development Initiative reports that 55.7 per cent of the population — 12.34 million of 22.16 million — are multidimensionally vulnerable, defined as deprivation in “at least three of the 12 weighted indicators”. Seven districts exceed 65 per cent: Ampara, Batticaloa, Kilinochchi, Mullaitivu, Nuwara Eliya, Puttalam and Vavuniya; the largest reported contributions are household debt (33.4 per cent) and adaptive capacity to disaster (48.8 per cent lacking preparedness), and 10.13 million of the vulnerable are rural, concentrated in the Eastern, Northern and North Central provinces [13]. These figures come from a secondary summary; the report itself was not obtained (§VI). Note what the index is and is not: it measures vulnerability in the §II sense — exposure to harm — not underservice. Two Eastern Province districts sit in its worst band, but it does not identify which service is inadequately provided, or by whom.
The national statistical apparatus is organised around a three-way urban / rural / estate sector classification, and the Department of Census and Statistics reports a sector gradient in computer literacy and household computer ownership, alongside a Labour Force Survey 2023 youth unemployment rate of 23.0 per cent [14], [15] — neither read in full (§VI). The estate sector is a separate category precisely because national and rural averages conceal it: §IV.B’s ecological problem embedded in the classification. At the other end of the scale there is no sub-national data at all — the WIPO Global Innovation Index 2025 economy profile ranks Sri Lanka 93rd of 139 economies and 131st on expenditure on education (1.83 per cent of GDP), but reports nothing below national level and carries no technical, vocational or non-formal education indicator [16]. A community can be underserved on every measure that matters locally while remaining invisible in the instrument that governs national policy attention.
V. The critique of the term itself
There is a substantial literature — and for a careful writer, the more useful one — arguing that underserved and its cousins are the wrong words.
A. The deficit is located in the wrong place
The 2010 study of “hard to reach” work found that “the notion ‘hard to reach’ is a contested and ambiguous term” with a “lack of consensus about the meaning of the term”, and that practitioners themselves rejected its direction of blame. One respondent: “it makes it sound like the fault of the non service-user…actually if you just got a ladder and sat next to them that would be fine.” Another: “it is about looking at our own services and looking at why individuals aren’t engaging.” The authors record a service-user quotation that states the objection exactly: “I am not ‘hard to reach’, generally people don’t know how to reach me” [1].
The objection transfers to underserved, which the same paper lists among the interchangeable terms [1]. In the passive voice the agent disappears: a community is underserved by nobody in particular, which reads as a property of the community rather than as a failure with an address. Hence the move in some fields to explicit “underserved by” constructions that name the provider.
B. Guidance now discourages the term outright
American Medical Association guidance groups “underserved” with “disadvantaged” and “under-resourced” as terms “used for decades but which many now find pejorative”, on the ground that they describe “historical disinvestment experienced by some communities” without assigning responsibility for it; the recommended alternatives are “historically and intentionally excluded” and “disinvested” [12].
The substitution is not cosmetic. “Underserved” asserts a shortfall; “historically and intentionally excluded” asserts a shortfall plus an agent plus intent, and therefore requires more evidence. A writer who adopts the stronger term without the stronger evidence has overclaimed; a writer who keeps “underserved” for a case that is actually deliberate exclusion has understated. The choice is evidentiary, not stylistic.
C. The counter-consideration
Against all of the above, “underserved” has one virtue its replacements lack: it is the only term in §II that names a service and therefore names a remedy. “Marginalised” and “historically excluded” are more accurate about causation and less specific about what to do next. Abandoning “underserved” for a structural term gains precision about blame and loses the operational question — which service, provided by whom, is inadequate to this population’s need? The defensible position is to keep the term, keep it relational, and always complete both of its variables.
VI. Limitations
Sources not read in full, and what rests on them. The Levesque framework [3] is cited as reproduced in a scoping review [4] read in full; the original could not be retrieved past a publisher access wall, and no claim here depends on wording found only in the original. The Hall, Stevens and Meleis concept analysis [5] and the Ricketts designation proposal [9] were read as abstracts only and are cited only for claims stated in those abstracts. CESCR General Comment No. 13 [10] was retrieved in part — the availability paragraph verbatim, the other three features by title. The Sri Lanka vulnerability figures [13] come from a secondary summary of the UNDP/OPHI report; the Department of Census and Statistics publications [14], [15] and the WIPO profile [16] were not read for this review and are cited from a bibliographic record only. Each should be obtained before any figure above is republished.
The vulnerability critique is the weakest link. The research-ethics argument against treating vulnerability as a fixed label attached to subpopulations, rather than as layered and situational, is associated with Florencia Luna’s “layers not labels” work; that paper could not be obtained and is deliberately not cited. The disaster-studies critique — that vulnerability frameworks strip agency and reify stereotypes — is omitted for the same reason. Of the four terms in §II, vulnerable is the one this review is least able to substantiate from an authoritative source, and its §II wording should be treated accordingly.
The asset-based counter-tradition is missing. The deficit-framing critique in §V has a longer lineage in asset-based community development (Kretzmann and McKnight, 1993), which argues that “needs maps” teach communities to see themselves as deficient. That literature could not be obtained and is not cited, so §V rests on health-services and professional-guidance sources alone.
No quantitative estimate of area-based misclassification. §IV.B argues the ecological problem structurally, from the statute’s own area/population disjunction [7], [8]. Empirical work quantifying how many deprived individuals fall outside designated deprived areas exists but was not obtained, so no figure is given.
No single authoritative definition exists. No multilateral body located in this review publishes a formal, general definition of underserved communities. Health-services research is the only field found to define it against a stated criterion [2], and that definition is scoped to healthcare. Use of the term outside health is therefore an analogy rather than an application — which is the strongest practical reason to name the service every time the word is used.
References
[1] S. M. Flanagan and B. Hancock, “‘Reaching the hard to reach’ — lessons learned from the VCS (voluntary and community sector). A qualitative study,” BMC Health Services Research, vol. 10, art. 92, 2010, doi: 10.1186/1472-6963-10-92.
[2] Z. Kayani, A. Willis, S. O. Salisu-Olatunji, S. Jeffers, K. Khunti, and A. Routen, “Reporting and representation of underserved groups in intervention studies for patients with multiple long-term conditions: a systematic review,” Journal of the Royal Society of Medicine, vol. 117, no. 9, pp. 302–317, 2024, doi: 10.1177/01410768241233109.
[3] J.-F. Levesque, M. F. Harris, and G. Russell, “Patient-centred access to health care: conceptualising access at the interface of health systems and populations,” International Journal for Equity in Health, vol. 12, art. 18, 2013, doi: 10.1186/1475-9276-12-18. ⚠️ Not read in full — cited as reproduced in [4].
[4] A. Cu, S. Meister, B. Lefebvre, and V. Ridde, “Assessing healthcare access using the Levesque’s conceptual framework — a scoping review,” International Journal for Equity in Health, vol. 20, art. 116, 2021, doi: 10.1186/s12939-021-01416-3.
[5] J. M. Hall, P. E. Stevens, and A. I. Meleis, “Marginalization: a guiding concept for valuing diversity in nursing knowledge development,” ANS. Advances in Nursing Science, vol. 16, no. 4, pp. 23–41, 1994, PMID: 8092811, doi: 10.1097/00012272-199406000-00005. ⚠️ Abstract read only.
[6] F. Osei Baah, A. M. Teitelman, and B. Riegel, “Marginalization: conceptualizing patient vulnerabilities in the framework of social determinants of health — an integrative review,” Nursing Inquiry, vol. 26, no. 1, art. e12268, 2018, doi: 10.1111/nin.12268.
[7] Health Centers, 42 C.F.R. § 51c.102 (definitions; “medically underserved population”).
[8] Indiana Primary Health Care Association, “HPSA MUA Designations Indiana — Health Shortage Areas.” [Online]. Available: https://www.indianapca.org/about-chcs/hpsa-mua/
[9] T. C. Ricketts, L. J. Goldsmith, G. M. Holmes, R. Randolph, R. Lee, D. H. Taylor, and J. Ostermann, “Designating places and populations as medically underserved: a proposal for a new approach,” Journal of Health Care for the Poor and Underserved, vol. 18, no. 3, pp. 567–589, 2007, PMID: 17675714, doi: 10.1353/hpu.2007.0065. ⚠️ Abstract read only.
[10] UN Committee on Economic, Social and Cultural Rights, General Comment No. 13: The Right to Education (Art. 13 of the Covenant), E/C.12/1999/10, 1999. ⚠️ Retrieved in part — paragraph 6(a) verbatim only.
[11] UNESCO, Global Education Monitoring Report 2020 — Inclusion and Education: All Means All. Paris, France: UNESCO, 2020.
[12] A. Robeznieks, “Try these 7 equity-focused language options to engage patients,” American Medical Association, Nov. 1, 2021.
[13] UNDP and Oxford Poverty and Human Development Initiative, Understanding Multidimensional Vulnerabilities: Impact on People of Sri Lanka — Multidimensional Vulnerability Index derived from the National Citizen Survey 2022–2023, 2023. ⚠️ Cited from a secondary summary (Sri Lanka Brief, 2 Sept. 2023); the report itself was not obtained.
[14] Sri Lanka Department of Census and Statistics, Labour Force Survey Annual Report 2023. ⚠️ Not read in full — cited from a bibliographic record.
[15] Sri Lanka Department of Census and Statistics, Computer Literacy Statistics, Annual Bulletin 2024. ⚠️ Not read in full — cited from a bibliographic record. ⚠️ Edition note: an earlier version of this review cited the 2023 bulletin, which the 2024 edition supersedes. The sector gradient described above is qualitative and holds in both; the two editions report different figures, so no number from either is reproduced here.
[16] World Intellectual Property Organization, Global Innovation Index 2025 — Sri Lanka economy profile, 2025. ⚠️ Not read in full — cited from a bibliographic record.