RESEARCH DOSSIER
Long-Term Care Activities of Daily Living Trigger Calibration
Abstract. An activities-of-daily-living (ADL) benefit trigger converts an assessment of functional dependence into an insurance eligibility decision. Its calibration determines which limitations count, what degree of assistance is required, how many qualifying activities must be affected, how long the condition is expected to persist, who may certify it, and when payment can begin. This monograph examines those decisions as a connected problem of measurement, contract interpretation, actuarial estimation, and comparative institutional design. The analysis establishes the distinct meanings of clinical impairment, functional limitation, chronic-illness certification, insured entitlement, service use, claim recognition, elimination-period satisfaction, and actual payment. It critically examines the foundational Katz and Lawton–Brody instruments, the World Health Organization's functioning framework, U.S. federal tax law and state regulatory models, Singapore's severe-disability insurance, and the German and Dutch statutory long-term care systems. It uses original source-to-rule comparisons, an explicit assessment dictionary, and reproducible hypothetical threshold examples. Established methods of classification validation, time-to-event analysis, multi-state modeling, and counterfactual contract recalculation are employed without inventing an independent theory. The central finding is that the apparent numerical simplicity of two-of-six or three-of-six ADL thresholds conceals substantial variation in activity taxonomies, assistance standards, cognitive routes, temporal gates, and observation processes. A comparable benefit trigger requires preservation of those dimensions at the item and event level. The manuscript derives implications for benefit predictability, claimant fairness, reserve estimation, product comparability, and scientific reproducibility, with jurisdiction-specific legal conditions and empirical limitations explicitly identified.
Keywords: activities of daily living; ADL benefit triggers; long-term care insurance; functional dependence; substantial assistance; standby assistance; hands-on assistance; severe cognitive impairment; chronic-illness certification; eligibility calibration; elimination period; claim incidence; measurement reliability; insurance contract law; actuarial science; Germany; Singapore; Netherlands; United States.
Introduction
A long-term care insurer promises a contingent benefit whose payment depends on a defined state of need, the performance of specified acts, and the insurer's contractual conditions. The transition from need to payment is mediated by measurement. An older person who needs a hand to enter a bathtub, another who needs somebody standing close by to prevent a fall, and a third who can wash independently but needs continuous supervision for cognitive safety may have materially different eligibility outcomes even when all three receive substantial daily assistance. Likewise, a person with two qualifying limitations may receive benefits under one policy, remain outside a different policy's threshold, and qualify for a public service under an entirely different care-needs evaluation.
The scientific subject of this Research Dossier is the calibration of private long-term care ADL triggers. Calibration means the selection, operationalization, testing, and experience-based evaluation of the observable conditions that convert a contractual promise into an initial or continuing entitlement. The object includes the precise activity list, the definition of deficiency, the count threshold, the certification authority, the expected duration of disability, changes in functioning, procedural review, and the relationship between benefit qualification and an elimination period. It encompasses comparative public programs where they illuminate the meaning and consequences of alternative classifications. Public program eligibility is examined on its own legal basis and is not treated as private insurance wording.
The WHO classification of functioning conceptualizes functioning within bodily, activity, participation, and environmental domains (World Health Organization [WHO], 2001). This international framework supports description of disability and environmental context, while contractual benefit qualification remains a distinct institutional decision. The U.S. qualified-care statute defines the federal tax-law concept of a chronically ill individual through specified ADL and cognitive routes (United States Congress, 2026). The CareShield Life eligibility criteria identify a different three-of-six threshold, with different activity labels and assessment procedures (Central Provident Fund Board [CPF], n.d.). Such differences are constitutive of observed entitlement outcomes.
The inquiry proceeds through six linked questions. What exactly is the assessed object: unaided capacity, usual performance, difficulty, assistance, or supervision? How do an ADL list and a threshold translate that assessment into an entitlement? Which delays arise from expected chronicity, certification, notice, adjudication, and elimination-period accumulation? How can benefits across jurisdictions be compared while respecting their distinct eligibility laws? What kind of empirical dataset can estimate threshold changes without mistaking the resulting administrative selection for underlying population disability? Which decisions are established by documentary evidence and which represent analytical implications requiring future observation?
The method is documentary and critical, supported by explicit formal reasoning and reproducible illustrative calculation. Original legislation and official program rules establish legal and administrative claims. Foundational clinical publications establish the historical measurement genealogy, without implying that their original item scores govern present-day claims. Longitudinal empirical research establishes that observed disability is dynamic. Actuarial experience studies establish the distinction between insurer claim incidence and population need. Where the monograph gives invented profiles or arithmetic, they are labeled hypothetical and carry no empirical rate estimate. There is no original field survey, medical assessment, insurer microdata extract, clinical validation study, or external peer review within this research edition.
Conceptual and Historical Foundations
From clinical function to financial contingency
The measurement of independent daily activity grew from clinical and rehabilitative research. Katz, Ford, Moskowitz, Jackson, and Jaffe's 1963 index of basic daily activities measured bathing, dressing, toileting, transferring, continence, and feeding, providing a standardized vocabulary for basic self-care (Katz et al., 1963). The study described more than 2,000 evaluations of 1,001 individuals and was designed to help characterize illness course, rehabilitation, and function. It did not establish a universal insurance trigger. Later functional instruments addressed more graduated physical performance; Mahoney and Barthel's functional evaluation index belongs to this rehabilitative genealogy (Mahoney & Barthel, 1965). Lawton and Brody's instrumental activities assessment extended the practical description of independence beyond elementary self-care, including tasks essential to autonomous community living (Lawton & Brody, 1969). Its distinction between basic ADLs and instrumental activities of daily living (IADLs) remains methodologically important.
The transfer of a clinical instrument into an insurance contract changes the decision problem. Clinical assessments ask how a person functions, what support is needed, and how ability changes with rehabilitation or environmental adaptation. An insurance contract asks whether specified conditions for a financial promise are satisfied. A clinical score may be comprehensive yet unsuitable for eligibility if the contract defines only particular activities. Conversely, a narrow contract-specific assessment may be valid for adjudication but insufficient to characterize the individual's full support needs. A score of two limitations describes different conditions depending on which two were observed, the response options, the weighting of the tasks, and the assistive environment.
The WHO's ICF conceptual framework defines functioning and disability in relation to the person's health and environment (WHO, 2001). This implies an important distinction between capacity in a standardized environment and performance in the usual environment. Care provided by a spouse may change observable performance without restoring underlying physical capacity; equipment may enable independent completion of a task without eliminating all risks. Assessment instructions must disclose which construct is being measured. A person's diagnosis remains relevant evidence, yet diagnosis alone generally does not supply a complete item-by-item contract classification. The WHO also explains the interaction of intrinsic capacity and environment in its functional ability account (WHO, 2020).
The object measured by a benefit trigger
A functional deficit is a limitation in a task under a stated assessment rule. Dependence means that performance requires assistance as defined by that rule. Chronic illness is a legal or contractual status established through temporal and substantive criteria. Entitlement is the legal consequence of satisfying the applicable coverage and procedural requirements. Claim incurrence is an event defined for insurance administration and actuarial recording. Benefit payment is a cash flow arising after eligibility, any contractual deferral, and adjudication. Each term represents a different variable.
For clarity, denote the underlying functional condition of person i at time t by F_i(t), and the observed ADL item vector by A_i(t). Let C_i(t) represent the observed cognitive and supervision condition. The contract's eligibility rule is E_i(t) = g(A_i(t), C_i(t), D_i(t), Q_i(t); theta), where D indicates qualifying duration or its prospective certification, Q indicates admissible certification and process, and theta contains the exact contract wording. Service setting, provider conditions, coverage in force, exclusions, and benefit structure may enter downstream payment rules. The equation is a bookkeeping expression for existing contractual reasoning, not a proposed new theoretical doctrine. An identical clinical condition can yield different E under different theta or different evaluation protocols.
A count threshold is therefore a partial description. A faithful eligibility definition is at minimum a tuple: activity set; item definitions; qualifying assistance; aggregation and alternative cognitive route; temporal condition; authorized assessor; applicable contract or statute; and the effective policy date. Additional payment conditions must be recorded alongside rather than hidden inside the clinical trigger. This distinction is particularly important when comparing insurance sold decades apart with contemporary publicly standardized eligibility.
The historical tax and regulatory boundary in the United States
The United States' 1996 Health Insurance Portability and Accountability Act framework was reflected in Internal Revenue Code §7702B. The current statutory chronic-illness conditions specify a licensed health care practitioner's certification of inability to perform at least two ADLs for a period of at least 90 days due to loss of functional capacity, or substantial supervision needed because of severe cognitive impairment; the statute also specifies a possible comparable-disability route subject to regulations. At least five of the six identified ADLs must be taken into account by a qualified contract when applying the functional route. The law identifies eating, toileting, transferring, bathing, dressing, and continence and requires certification within the preceding 12 months. The 90-day term is part of the substantive expectation for qualifying functional incapacity, rather than itself a universal 90-day claim waiting period.
The Treasury Department's account of Notice 97-31 safe-harbor definitions shows the practical significance of distinguishing hands-on help from standby help and substantial supervision from mere forgetfulness (U.S. Department of the Treasury, 1997). The notice supported implementation of the 1996 reforms through operational definitions. California's insurance regulator further identifies hands-on and standby assistance as material in tax-qualified benefit triggers (California Department of Insurance, n.d.). Item thresholds, qualifying assistance, and cognitive-route treatment must therefore be analyzed as jointly operative.
An additional source of complexity arises from the National Association of Insurance Commissioners' Long-Term Care Insurance Model Regulation (NAIC, 2017). Section 29 states a model consumer-protection standard under which eligibility is not to be more restrictive than a deficiency in no more than three listed ADLs or the presence of cognitive impairment. Section 30 separately addresses qualified contracts, including the federal chronic-illness route. A model regulation is a template for jurisdictions; its existence is not proof of a single nationwide enacted rule. California's published insurance guidance illustrates the need to read state consumer protections and policy terms alongside the federal baseline. Any actual claim or historical portfolio study must resolve the state, policy issue cohort, applicable law, and contract wording before stating the effective standard.
Need, eligibility, and economic risk
A trigger operates between latent need and the insurer's payable event. For the policyholder, it influences the probability and timing of financial assistance. For the insurer, it influences claim incidence, duration, payment delay, reserves, and contractual interpretation risk. For regulators, it affects clarity, comparability, and opportunities for contesting a denial. For society, it affects substitution between private coverage, public services, and unpaid family support. These effects arise through existing financial and institutional relationships; their direction depends on data, product structure, and the population at risk.
An initially attractive claim that a less demanding ADL threshold always increases total insurance payout ignores offsetting changes to price, coverage take-up, lapses, family support, and benefit duration. Holding the portfolio and contract otherwise fixed, an expanded qualifying set weakly increases the number of people eligible at a given assessment point; it does not by itself establish a net welfare gain or a commercially sustainable tariff. Finkelstein and McGarry's long-term care selection analysis demonstrates that private information about risk and taste for coverage can have different effects in observed insurance samples (Finkelstein & McGarry, 2006). It is consequently unsound to carry a population prevalence rate directly into an insured-loss assumption without an exposure and selection adjustment.
ADL Definitions and Assessment
The six-item list is a measurement dictionary, not a self-executing threshold
The widely cited six basic ADLs are bathing, dressing, eating or feeding, transferring, toileting, and continence. Their apparent commonality can obscure clinically and legally material differences. The NAIC policy-definition provisions describe bathing as including getting into or out of the tub or shower, continence as including associated hygiene and care of devices, dressing as including braces or fasteners, eating as getting food into the body, toileting as access and hygiene, and transferring as movement into or out of a bed, chair, or wheelchair (NAIC, 2017). The item is the entire contract-defined task, not an evaluator's unannounced simplification.
The following crosswalk expresses a documentary comparison, not equivalence between clinical measures and all insurance policies.
| Function | Frequent U.S. tax-qualified/NAIC label | Singapore CareShield Life label | Principal comparability question |
|---|---|---|---|
| Personal washing | Bathing | Washing | Does the assessment include entering or exiting a bathing facility? |
| Clothing management | Dressing | Dressing | Are braces, prosthetic devices, and fasteners included? |
| Taking food | Eating | Feeding | Is food preparation separated from getting food into the body? |
| Toilet-related activity | Toileting | Toileting | Are access, transfer, cleansing, and continence management treated together? |
| Changing bodily position | Transferring | Transferring | Is movement specifically between bed, chair, and wheelchair assessed? |
| Bowel and bladder control/hygiene | Continence | No separate ADL item in the core six | Is continence embedded in toileting or counted separately? |
| Movement within the home | No separate ADL item in the core six | Walking or moving around | Can inability to walk qualify independently of transfers? |
Singapore's six qualifying activities include walking or moving around instead of a separately enumerated continence item (CPF, n.d.). Both lists contain six entries, yet they represent different dimensions. A person needing help with continence management and bathing, while moving independently indoors, may have two qualifying U.S. items and only one Singapore item, before considering the countries' differing assistance standards. A person unable to walk independently indoors but maintaining continence may exhibit the reverse directional difference at the item level. An empirical crosswalk must preserve both raw observations, not equate their summary counts by arithmetic.
IADL assessment—shopping, meal preparation, finances, communication, and other complex tasks—has a distinct scientific role. The Lawton–Brody original instrument addresses activities with important implications for safe community living (Lawton & Brody, 1969). IADL limitations often signal dependence before a basic self-care threshold is reached, but their inclusion in a covered event must derive from the contract or program rule. A comparison that silently adds IADLs to an ADL count changes the threshold rather than measuring it more completely.
Assistance levels and observation protocols
Assessment may distinguish independent activity, completion with ordinary aids, verbal prompting, supervision from a distance, standby presence within reach, partial physical assistance, and total physical assistance. That hierarchy is descriptive; a specific policy may combine or differentiate categories. An evaluator must record what the claimant actually needs, why assistance is required, how frequently assistance is needed, and which protocol is in force. For a person at risk of falling during bathing, the difference between standby and hands-on intervention is particularly consequential. A recording scheme that captures only whether physical contact occurred can miss a documented need for another person to be immediately available.
A reliable assessment also specifies the reference period. Capacity demonstrated once under controlled conditions can differ from usual performance across a week. Fluctuating symptoms, medication effects, fatigue, pain, environmental obstacles, and cognitive changes can produce variability. The appropriate decision requires the legally specified construct: inability, need for substantial help, or some other condition. Repeated observations may be valuable, but their results cannot be substituted for the contract's required certification without a clear legal mapping.
The scientific distinction between validity and reliability is fundamental. Reliability concerns consistency when the assessed state is held sufficiently stable; validity concerns whether the instrument measures the intended construct. The published interRAI reliability evaluation used a Korean setting and double-assessed a subsample, reporting substantial to nearly perfect agreement across many items (Kim et al., 2015, first published online in 2014). Those results support the possibility of disciplined repeatable assessment under specified training conditions. They do not establish that all insurers' proprietary ADL forms are interchangeable or that interrater agreement automatically validates a financial entitlement rule.
Measurement design should explicitly separate five sources of variation: the patient's condition; the assessor's understanding of the item; the observation setting; changes in available assistance or equipment; and the applicable decision rule. Only the first represents the person's underlying functional evolution. The others can induce a different recorded score without an equivalent health change. Recording an assessment version, assessor qualification, item-level finding, information source, and date allows those effects to be investigated.
Cognitive impairment as an independent route
Severe cognitive impairment can create a need for substantial supervision when a person retains physical ability to bathe, eat, dress, or transfer. For a U.S. qualified contract, the federal supervision criterion supplies an alternative qualification route (United States Congress, 2026). A claimant's number of physical ADL impairments consequently cannot be the only variable in a valid claims assessment. The nature of the risk—wandering, inability to recognize hazards, unsafe medication handling, or severe judgment impairment—must be described against the policy's definition, with suitable professional certification. The presence of a diagnosis alone is not identical to the requisite degree of supervision.
Singapore's administrative assessment guidance on cognitive conditions explains that accredited assessors are trained to consider cognitive impairment and fluctuating functioning within the six-ADL evaluation (Ministry of Health Singapore, n.d.). That is a materially different architecture from a U.S. policy providing an independent severe-cognitive-impairment trigger. The analyst must not infer that equal clinical severity produces equal automatic access in both systems, or that Singapore's model has exactly the same separate cognitive benefit clause as U.S. tax-qualified private insurance.
Evidence standard and procedural integrity
The defensible unit of evidence is an item-specific observation linked to the decision rule. The underlying record should state the activity; the person's ability with ordinary aids; required assistance; reason and frequency; date and observation window; assessor's professional role; relevant clinical or safety documentation; and the governing policy clause. It should also record conflicting testimony or different behavior in different settings without quietly averaging away the disagreement. Clinical notes, family observations, home-care records, and formal evaluations may illuminate different aspects of need. Their credibility should be assessed according to provenance, relevance, temporal proximity, and methodological quality.
The NAIC model also addresses assessments by licensed or certified professionals and procedures for contesting adverse decisions. The model independent-review provisions place procedural scrutiny alongside functional classification (NAIC, 2017). Actual independent-review rights depend on applicable enacted law and policy. A decision with a seemingly accurate total score can still be scientifically opaque if no item-level basis or definitional explanation is recorded. Transparent claims records are therefore important evidence for both consumer protection and actuarial validation.
Trigger Thresholds and Waiting Rules
Eligibility as a sequence of separately defined decisions
The phrase two-of-six ADL trigger abbreviates a series of legal and evidentiary conditions. The governing form identifies the recognized ADLs, the degree of assistance that constitutes inability, the physical or cognitive route to entitlement, who may assess the condition, the duration or prognosis of impairment, the qualifying service requirements, any elimination period, available benefit limits, and the date from which payments are calculated. A properly calibrated trigger records each condition as a distinct field. The arithmetic count alone cannot decide a claim.
The U.S. Internal Revenue Code provision is especially instructive (United States Congress, 2026). Section 7702B(c)(2) defines one route for a chronically ill individual through inability to perform at least two ADLs without substantial assistance for a period of at least 90 days, as certified by a licensed health care practitioner; a separate route involves substantial supervision to prevent threats to health and safety arising from severe cognitive impairment. Certification must fall within the preceding twelve months. Qualified services are provided pursuant to a prescribed plan of care. The statute requires that determination under the physical-ADL route take account of at least five enumerated activities. It does not establish a nationwide universal promise to pay everyone with two ADL limitations: contractual coverage, policy terms, applicable state law, insured status, and benefit commencement remain material.
A scientifically precise reading treats the ninety-day provision as a duration or prognosis condition attached to the chronic-illness classification. It is distinct from a policy's elimination period, which controls when otherwise covered benefits become payable under the contract. The two periods might overlap in calendar time, but they answer different questions. Confusing them systematically shifts the estimated start of payment and exaggerates the apparent protection supplied by a policy.
The NAIC's Long-Term Care Insurance Model Regulation specifies in Section 29 that an LTC policy's benefits should not require more than three ADL deficiencies, or cognitive impairment, under its model standard (NAIC, 2017). This should be read as an upper restrictiveness boundary in model guidance, not a rule that every policy applies exactly three ADLs. A two-ADL policy and a three-ADL policy can both be evaluated against the model, subject to the precise form and adopting jurisdiction. The federal qualified-contract tax definition and the NAIC insurance-benefit model must be examined together while retaining their different regulatory purposes. The NAIC's text is a model that jurisdictions may adopt or modify; it is not itself a uniformly enacted federal insurance law.
Dependence severity, threshold choice, and boundary cases
A threshold is the consequence of a particular decision rule applied to a defined observation. Denote item-level dependence for individual i, activity j, and assessment time t by , taking value one when the contract-defined assistance criterion is satisfied and zero otherwise. With covered items, the recorded count is
For a simple count rule, physical eligibility at time t may be written , where k is the contractual threshold. This is an analytical representation of a count-based term, not a substitute for policy language. The real eligibility indicator also depends on the relevant assistance definition, qualifying duration, assessment and certification rules, and alternative cognitive route. A broader expression is
where represents applicable certification and timing requirements, the separate cognitive qualification when provided, and other qualifying conditions. Multiplication denotes conjunction of binary conditions; logical OR protects the independence of the two eligibility routes. This expression is a descriptive reconstruction of established contractual logic, not a new actuarial or legal doctrine.
For one population assessed at one instant using the same six-item dictionary and severity criterion, the set with at least three ADL dependencies is contained in the set with at least two. The difference is the set with exactly two recorded dependencies. This mechanical monotonicity does not establish that moving from two to three decreases ultimate lifetime claim cost by the same proportion. Delayed onset, episodes that recover, death, different claims durations, different product prices, and behavioral effects intervene. It also does not permit numerical comparison between the United States and Singapore, because the item dictionaries and assistance thresholds differ.
Several boundary cases merit explicit decision records. A person dependent in bathing and dressing for several weeks may satisfy an observed two-item score without meeting an applicable ninety-day chronic-illness condition. A person who physically completes five tasks but wanders into dangerous environments may meet a policy's separate cognitive trigger. A person using ordinary assistive technology may perform an activity independently for the purposes of one form yet require assistance under another form's assessment convention. Two assessments separated by a short interval can produce different scores if a condition fluctuates. None of these cases can be resolved reliably by reducing the file to a single number.
Waiting time, elimination periods, and the payable event
An elimination period is a contract-specified interval or accumulation condition that must be satisfied before benefits commence. The federal Administration for Community Living explains the practical importance of elimination-period alternatives, including distinctions between calendar-day and service-day counting (Administration for Community Living [ACL], n.d.). A requirement of ninety qualifying service days is economically different from ninety elapsed calendar days, particularly for home care delivered intermittently. The policy may also contain provisions addressing consecutive days, one or multiple periods of care, waiver of certain days, or a lifetime versus per-episode elimination period. The contract and applicable law govern each detail.
To isolate the effect, consider an expressly hypothetical covered period beginning on day 1 and lasting 120 calendar days. A ninety-calendar-day elimination condition can leave up to thirty post-elimination covered days, depending on the form's start-of-day convention. Suppose instead that eligible services occur on only four days of each week, with the policy requiring ninety service days: by the end of 120 calendar days the claimant may have accrued only about sixty-nine service days, so no benefit has yet become payable. This comparison assumes a uniform four-of-seven service schedule and ignores other contract rules. It demonstrates how two policies with the same printed number can generate markedly different timing. It is not an estimate of the prevalence of home-care claims.
A claim database should keep separate dates for functional onset, qualifying duration, certification, first covered service, claim notification, claim acceptance, elimination-period satisfaction, and first payment. The SOA research on LTC incidence illustrates why claim and exposure definitions and experience period must be established before interpreting incidence trends (Society of Actuaries [SOA], 2018). Payment start can lag onset because of contractual waiting rules even if no adjudication delay occurs. Administrative delay is an additional, separately measured phenomenon.
Recovery, recurrence, and supervision
ADL impairment is not always permanent. The longitudinal study of incident ADL/IADL disability followed a U.S. population sample and found substantial subsequent recovery and death (Ankuda et al., 2020). The reported figures concern incident ADL or IADL disability in that study's population and should not be represented as insured, tax-qualified, two-ADL claim experience. Their significance here is structural: the path through dependence, recovery, recurrent dependence, and death can cross a contract threshold repeatedly. A short temporary episode may never qualify after duration and elimination rules; a prolonged episode may become payable; and a later recurrence may be handled as a new or continuing claim according to the policy.
A prospective transition analysis distinguishes the time spent in a documented dependent state, time in a certified eligible state, time accumulating elimination days, time receiving payment, and time after termination or recovery. It also recognizes death as a competing exit event. The benefit trigger is therefore a time-dependent and condition-dependent decision rather than a once-for-all clinical label. This conclusion follows from established event-history reasoning and the source definitions, without introducing a new theory of care dependence.
Cross-Market Entitlement Outcomes
The United States: statutory qualification and private insurance performance
In the United States, private LTC insurance operates through policies and state-level insurance regulation, while federal tax law establishes the qualifying-contract framework. At a high level, the federal chronic-illness definition describes two-ADL and cognitive routes, but policy benefits depend on coverage, plan-of-care and service requirements, contractual waiting rules, benefit caps, and applicable rights (United States Congress, 2026). The NAIC consumer information separately describes LTC insurance design and regulatory considerations (NAIC, n.d.). A comparative study should neither infer automatic approval from a tax statute nor treat model insurance regulation as an enacted nationwide rule.
Claims analysis must distinguish all applicants from accepted insureds, all accepted claims from paid claims, and new claims from ongoing benefit recipients. A more generous functional trigger may increase the eligible population at an assessment date while affecting total spending through duration, utilization and benefit exhaustion. The SOA long-term care experience database provides insurer-experience material and methodological context, but aggregate published experience cannot by itself identify the causal cost of changing a single ADL threshold. Such identification would require policy-wording strata, harmonized assistance definitions, age and duration adjustment, and adequate claims-exposure linkage.
A population-based analogue also demands care. The OECD's long-term care needs assessment describes need across countries and severity categories (OECD, 2024). Those categories represent survey and administrative constructs, not frequencies of payable events under a U.S. commercial policy. Survey prevalence is informative for overall care burden and distribution but requires additional mappings to predict insured incidence.
Singapore: severe-disability assessment under CareShield Life
Singapore's CareShield Life is a mandatory or automatically enrolled national long-term care insurance arrangement for eligible cohorts, with rules differing from the contractual architecture of U.S. private products. The official CPF claim eligibility explanation states that an insured person must be assessed by a Ministry of Health-accredited assessor as unable to perform at least three of six specified ADLs (Central Provident Fund Board [CPF], n.d.). Those activities are washing, dressing, feeding, toileting, walking or moving around, and transferring. The CareShield Life benefit description describes continuing monthly cash payouts while severe-disability requirements remain satisfied (CPF, n.d.).
Singapore's Agency for Integrated Care guidance specifies full assistance with three activities and the accredited assessment procedure (Agency for Integrated Care, 2026). This feature prevents a simplistic three-versus-two comparison: both the set of assessed tasks and the degree of help differ from the common shorthand used for U.S. qualified LTC coverage. Moreover, Singapore's toileting definition incorporates management of bowel and bladder function; continence therefore has a potential role within that activity even though it is not a standalone item. Saying that Singapore ignores continence would misstate the official description.
The Ministry of Health's 2025 CareShield Life review discusses the implications of broadening eligibility and maintaining the three-ADL threshold (Ministry of Health Singapore, 2025). Its projections illustrate policy trade-offs, but they should be presented as ex ante official estimates dependent on assumptions rather than observed effects of an implemented threshold change. Assessment access, claims administration, reassessment, and beneficiaries' preferred care arrangements are part of the operational outcome, distinct from the threshold itself.
Germany: multidomain statutory dependency grades
Germany's social long-term care insurance provides a valuable methodological contrast because its dependency determination is based on impairment of independence and abilities across multiple domains. Section 14 of Social Code XI establishes a need for assistance due to health-related impairments expected to last at least six months (Federal Republic of Germany, 2026). Section 15 establishes the assessment instrument and care-grade determination. The statutory weighting schedule assigns 10% to mobility, a combined maximum of 15% to cognitive/communication or behavioral/psychological domains, 40% to self-care, 20% to illness- or therapy-related demands, and 15% to everyday life and social relationships (Federal Republic of Germany, 2026).
Those weights do not make German grade eligibility a six-item ADL count. A person may have relatively preserved basic physical self-care while experiencing material limitations in cognition or treatment-related demands. Conversely, a U.S. claimant could satisfy a contract-defined two-ADL rule without crossing a particular German care-grade threshold. Accurate comparisons therefore require either parallel administration of both assessment instruments or validated crosswalks with explicitly reported disagreement. One cannot convert German weighted points into a number of U.S. ADLs through a constant multiplier.
The six-month criterion in German statutory care need is a separate duration condition from the ninety-day U.S. tax-qualified chronic-illness criterion. The respective definitions are associated with distinct social insurance and private insurance institutional functions. No international clinical equivalence follows from comparing their respective time units.
The Netherlands: access grounded in continuous supervision and care needs
The Dutch Long-term Care Act (Wlz) provides entitlement to intensive long-term care for people requiring permanent supervision or twenty-four-hour care in proximity, subject to assessment by the responsible body. The Dutch government's Wlz program explanation and care-needs assessment guidance describe an institutional gateway different from a predetermined count of basic ADLs (Government of the Netherlands, n.d.). The substantive need for continuous supervision or nearby care, the assessment decision, and the applicable service system define this entitlement.
A person may require near-continuous supervision because of cognitive or behavioral risks without being recorded as dependent in three physical self-care activities. Conversely, a person with three ADL dependencies may receive assistance at planned intervals without meeting the Dutch need for continuous supervision or twenty-four-hour proximity. Neither country necessarily classifies the two persons identically. The insurance scientist should compare needs and entitlements as parallel measures of related but different constructs.
Comparative decision table and limits of transferability
| System | Principal assessment gate | Functional dictionary and severity | Duration and timing | Nature of outcome |
|---|---|---|---|---|
| U.S. tax-qualified LTC framework | At least two ADLs with substantial assistance, or severe cognitive supervision, plus certification | Federal six-item list; at least five considered for physical route; policy/state details govern payments | At least 90 days for ADL chronic-illness criterion; separate policy elimination terms | Tax-qualified status and eligibility framework; private payments depend on form |
| Singapore CareShield Life | At least three of six qualifying ADLs and accredited assessment | Full assistance under official severe-disability assessment; movement included | Payments continue while the qualifying severe disability persists | Statutory insurance cash benefit |
| Germany SGB XI | Statutory care-grade determination | Multidomain weighted assessment of independence and abilities | Expected need at least six months | Social LTC insurance grade and associated benefits |
| Netherlands Wlz | Permanent supervision or twenty-four-hour nearby care with authorized needs assessment | Continuous care need rather than fixed ADL count | Ongoing nature of intensive care need | Statutory long-term care entitlement |
The table compares rule structure at the cited sources' current versions. It does not claim uniform eligibility among all insurers, states, cohorts, or national programs. Its analytical use is to show how the choice of clinical construct, assistance criterion, observer, time rule, and benefit design determine which cases become payable. The OECD's comparative assessment methodology documents why cross-country care-need classifications require care when definitions and survey measures vary (OECD, 2024).
Evidence Validation and Sensitivity
Defining the estimand and its denominator
The main empirical question is generally not how many older adults need assistance. It is how changes in a stated contractual assessment rule affect the incidence, persistence, timing, and cost of payable insured episodes in a defined portfolio. The estimand should be stated before data collection. For example, a portfolio analyst might seek the one-year change in first claim incidence among insured lives in force, under a counterfactual replacement of a three-ADL trigger with a two-ADL trigger, holding policy cohort, age, gender, assistance definition, cognition, service eligibility, and elimination terms fixed. That is a counterfactual question requiring item-level longitudinal and insurance contract data.
Let indicate that insured individual i is alive, insured, and at risk of first benefit initiation at time t, after accounting for the applicable exposure definition. An empirical annualized incidence estimate of first payable entries is
This expression requires numerator and denominator to refer to identical insured populations, policy rules, calendar periods, and event definitions. A claims-paid file without exposure data does not supply a valid annualized incidence denominator. An in-force census on December 31 is not automatically equivalent to insured person-years at risk. An active-life exposure definition can exclude people already claiming, while a total-life exposure basis may include them; rates calculated using the two cannot be compared without reconciliation.
For payment costs, a prospective value of covered claims can be represented through the established actuarial discounted cash-flow identity
where collects transition, contract, utilization, and valuation assumptions, is the discount factor, and is covered benefit amount. The benefit-trigger threshold can alter the payable indicator, but claim cost also depends on survival, episode duration, care utilization, eligible services, policy maximums, lapses, and reserves. This is ordinary expected-present-value analysis, not an original pricing theory.
A transparent synthetic threshold calibration
Consider a purely illustrative cohort of 10,000 insured people observed at a fixed assessment date. Each has been assessed against the same six-item ADL dictionary, with the same assistance standard. The following fictional distribution contains no insurer data and is deliberately chosen for transparent arithmetic.
| Number of qualifying ADL dependencies | Hypothetical individuals | Share of cohort |
|---|---|---|
| Zero | 7,600 | 76.0% |
| One | 1,000 | 10.0% |
| Two | 650 | 6.5% |
| Three | 350 | 3.5% |
| Four | 220 | 2.2% |
| Five | 120 | 1.2% |
| Six | 60 | 0.6% |
| Total | 10,000 | 100.0% |
Under a simple physical threshold of at least two, 1,400 people meet the instantaneous count criterion; under at least three, 750 people do. The difference is 650 people, or 6.5% of the cohort, entirely corresponding to the exactly-two group. Relative to the three-ADL count, the two-ADL count is 86.7% higher (650 divided by 750). None of these people has yet been determined to satisfy certification, required persistence, insured status beyond the cohort definition, or an elimination period. Thus 14.0% and 7.5% are hypothetical cross-sectional count proportions, not private LTC claim incidence, approval rates, or premium changes.
Now introduce a different illustration: assume 80% of the exactly-two group satisfies the applicable functional-duration condition, 90% of those obtain valid certification, and 75% of that certified subgroup completes a qualifying elimination condition during the follow-up window. If each percentage were conditional on all preceding stages, the number attributable to the exactly-two group reaching that defined sequence would be . If the conditional parameters were instead 60%, 85%, and 50%, the corresponding result would be , an expected count of approximately 166. The numerical difference is driven by invented conditional assumptions. It illustrates the sensitivity of payable eligibility to temporal and administrative stages; it provides no evidence of real-world rates or causation. In a complete model, death, cognitive triggers, changed assistance criteria, repeated claims, care-setting eligibility, and state transitions must be incorporated explicitly rather than hidden inside these percentages.
The exercise also indicates what a threshold sensitivity analysis can answer mechanically and what requires empirical data. Given a fixed item-level count distribution and fixed severity definition, the instantaneous count difference is identified. The resulting difference in payable episode incidence requires longitudinal observations and contractual event reconstruction. The difference in lifetime present value additionally requires duration, benefit, utilization, mortality, selection, expense, and discount assumptions.
Measurement error around the cutoff
Eligibility based on a discrete cutoff is especially sensitive to observations near that cutoff. Suppose two trained assessors independently classify dependence and disagree on exactly one ADL for a person whose initial count is two. Under a two-ADL contract, disagreement can move the classification from eligible to ineligible if the second assessor records one. Under a three-ADL contract, the same disagreement can move a person scored two to eligible if the second assessor records three. Which threshold has greater portfolio-level misclassification depends on the density of cases near its cutoff and the error structure of each task. A reliable audit therefore tabulates item-specific disagreement and the number of decisions changed by the disputed item, rather than reporting only agreement on total scores.
Diagnostic or evaluator-related error should be separated from legitimate fluctuations in need. Repeat assessments made weeks apart may disagree because the claimant's functioning changed; same-day assessments may disagree because assessor instructions were inconsistent. A validation sample should record each mechanism and use adjudication independent of the initial evaluator where practicable. The interRAI reliability study supplies one empirical precedent for structured assessment research, but its results cannot be transplanted without validation to private insurer benefit determinations or different item dictionaries (Kim et al., 2015).
Sensitivity to assistance definitions can be evaluated by classifying the same documented episodes under several prespecified rules: hands-on assistance only; hands-on plus standby; and substantial supervision under the explicit cognitive clause. Comparing these simulated results is meaningful only when the source records include the relevant detail. A dataset that records only 'help: yes/no' cannot reconstruct more stringent gradations. An analyst should not fill missing physical-assistance observations with an unverified default and then present the resulting difference as identified.
Selection, confounding, and causal limitations
Insured portfolios are selected populations. People choose whether to purchase LTC insurance, products vary by eligibility at issue and underwriting, and lapses alter the remaining risk pool. The published study of private information in LTC insurance by Finkelstein and McGarry (2006) shows the relevance of multidimensional private information to LTC coverage and selection. Accordingly, a comparison of claims rates between two-product groups cannot assume the thresholds caused the observed difference: age, health, contract vintage, insured duration, price, product design, benefit levels, selection, and reporting procedures may differ.
A defensible causal design, where data permit, would require a well-defined policy-rule change with adequate controls or a credible threshold-specific quasi-experiment. It must inspect pre-change balance, concurrent contract and administration changes, differential selection, repeated-assessment behavior, and observation windows. If the policy rule change also modifies premium, benefit amount, or underwriting, its isolated effect on claims is especially difficult to identify. Even a careful matched observational analysis may remain descriptive when rule assignment is endogenous.
Censoring compounds the problem. A person can lose coverage, die, exhaust benefits, move to another service setting, or leave the administrative observation window. Those events have different meanings for incidence and duration. Survival bias appears when only long-lived claimants can accumulate enough care days to qualify. Death is a competing event when the estimand is first payable claim; a lapsed policyholder can be censored or modeled as an explicit state according to the estimand. Claim termination may be followed by recovery or subsequent entry. A robust analysis reports event definitions, censoring treatments, assumptions, and sensitivity to competing events, not a single unqualified 'ADL incidence rate.'
Documentary and numerical audit protocol
For each source, the evidence record should retain its institutional author, title, version, jurisdiction, source URL, access or verification date, relevant paragraph or table, and exact conclusion supported. For each policy, preserve version, issue period, amendments, state endorsements, and the rule governing the observed claimant. For individual assessments, preserve instrument version, item coding, data provenance, evaluation date, assessor role, and adjudication outcome. For insurer data, preserve definitions of exposure, claim incurrence, payment date, termination, and duplicate episodes. These are the components of a reproducible calibration audit.
A useful numerical audit checks five identities: the six item-count categories sum to the defined population; cohorts are mutually exclusive when required by the estimand; the number completing sequential eligibility stages cannot exceed the preceding conditional stage; paid episodes are reconciled to claim acceptance and waiting rules; and source denominators match numerators in population and time. Published insurer studies may use specialized exposure definitions or adjustments, so their methodological notes should be applied before deriving any quantity. The SOA 2000–2016 aggregated database is a source for understanding commercial experience material rather than an authorization to invent policy-level causal results (SOA, 2020).
Scientific Synthesis and Research Agenda
The operative scientific result
The decisive scientific result of this source-based analysis is that a long-term care insurance ADL trigger is a compound measurement-and-contract rule. Its observable result depends on the ADL dictionary, the assistance standard, the evaluator and evidence protocol, duration and certification conditions, cognitive alternatives, elimination mechanics, eligible services, and continuing benefit limits. A stated number such as two or three describes only the count component. This is an analytical synthesis of established clinical, statutory, contractual, and actuarial material; it does not purport to establish a new theory, doctrine, or universal standard.
The comparison demonstrates that the same ordinary words—bathing, dressing, eating, toileting, transferring—can be embedded in different legal tasks. Singapore adds indoor movement to its six while its toileting definition includes bowel and bladder management; the U.S. federal list contains a separate continence item. Germany uses weighted, multidomain dependency assessment, while the Netherlands emphasizes a defined need for continuous supervision or care nearby. Directly comparing headline thresholds across these systems therefore confounds the construct being measured with the rule used to award financial entitlement. That is a documentary finding supported by the original rules, not a numerical claim about international generosity.
Implications for policy drafting and claims operations
Good policy drafting makes the underlying measurement reproducible. The wording should define each item unambiguously, specify whether ordinary aids are taken into account, describe substantial assistance and supervision, distinguish the cognitive route, specify an assessor and certification process, identify the persistence condition, explain the elimination period's counting convention, and state when benefit entitlement begins and ends. Operational forms should mirror those definitions. Claims professionals should retain reasons for item-by-item decisions and provide a comprehensible path for correcting evidence or contesting an adverse finding according to the applicable jurisdiction.
Threshold accessibility must also be evaluated together with service availability. An insured may qualify clinically while residing in a location with limited covered-care capacity, or may rely on informal assistance that is treated differently by reimbursement and cash-benefit designs. The OECD's care affordability analysis highlights the significance of costs and financing arrangements for older persons (OECD, 2024). The value of a claim trigger cannot be inferred solely from the rate of medical or functional qualification; practical benefit realization depends on how the policy pays and which services qualify.
Implications for actuarial and supervisory comparisons
An actuarial assumption labeled 'two-ADL incidence' is incomplete without its assessment and contractual metadata. Experience studies should classify policy cohorts by trigger, item dictionary, severity definition, cognitive clause, qualifying duration, elimination mechanics, benefit type, and effective policy version. Actuaries should model each stage between functional need and benefit payment when available data support it. Where only aggregate data exist, they should state the identification limits rather than using a point adjustment that appears empirically calibrated without evidence. Scenario analysis may quantify assumed uncertainty, provided the underlying assumptions remain visible.
A supervisor comparing insurers should similarly distinguish claim denials due to absence of functional impairment, inadequate professional certification, uncompleted elimination periods, noncovered services, exhausted limits, or policy discontinuance. Combining those reasons as 'failure to meet ADL trigger' conceals distinct consumer and insurance risks. Oversight of assessor training and appeals can be quantitatively relevant when borderline classifications are frequent, although its actual effect must be measured in the relevant portfolio rather than presumed.
A testable research agenda
Several studies would materially improve the evidence base. One would administer matched U.S. and Singapore ADL dictionaries to the same consented functional-assessment sample and measure item-level and threshold-level agreement, including how Singaporean continence-related toileting maps to U.S. continence and toileting. Another would link repeated clinical scores, certification, insurance policy versions, elimination-day accounting, and benefits to estimate first payable-entry probabilities with competing death and lapse. A third would randomize assessor training protocols, where ethically and legally appropriate, to measure decision consistency around two- and three-item thresholds. A fourth would examine appeal outcomes and error correction using blinded independent review and stratification by claimant needs. A fifth would study the effect of service-day versus calendar-day elimination rules on benefit onset and the proportion of eligible home-care claimants receiving payments.
Each study must define a target population, event date, raw assessment items, policy terms, observation horizon, and treatment of missing data before drawing conclusions. A linked comparison between public care-need classifications and commercial insurance claims would be especially informative, but its results would reflect the particular national and insurer settings. Replication across settings is required before asserting broader transferability. Original primary-source analysis can guide these research designs now; causal, actuarial, or clinical effect sizes await the necessary data.
Conclusion
Activities of daily living provide indispensable observable evidence for long-term care protection, but the financial consequence of an ADL limitation is established by the entire insurance or statutory eligibility rule. The distinction between a person's ability to perform an activity, the assessor's recorded classification, certified chronic impairment, contract-specific qualification, completed elimination period, and actual payment is the foundation of accurate trigger calibration.
The primary sources establish three especially consequential distinctions. First, the U.S. federal qualified-LTC framework includes a two-ADL chronic-illness condition or an alternative severe-cognitive-impairment condition, while state implementation and policy wording shape payment; the ninety-day chronic-illness element is analytically separate from the elimination period. Second, Singapore's CareShield Life applies three of six specifically defined activities under accredited severe-disability assessment, using a different activity dictionary and assistance convention. Third, German and Dutch statutory care systems use different constructs of continuing need. These differences preclude automatic numerical conversion of threshold counts across markets.
The original hypothetical analysis identifies the exact-two-ADL population as the immediate difference between fixed two- and three-item count rules, while showing why that cross-sectional result does not identify claim incidence or lifetime cost. The defensible actuarial object is a payable episode under explicit exposure, certification, timing, service, and policy rules, modeled with appropriate recovery and competing risks. Transparent source linkage, individual-item assessment, contract-version control, and distinct administrative event dates make the analysis auditable. They also support more intelligible consumer decisions and sounder supervisory comparisons.
This monograph's contribution is its integrated original analytical reading of established functional measurement, legal entitlement, insurance operations, and actuarial inference. No new clinical scale, legal doctrine, actuarial theory, proprietary dataset, measured causal effect, or externally peer-reviewed claim is introduced.
References
Administration for Community Living. (n.d.). Receiving long-term care insurance benefits. U.S. Department of Health and Human Services. https://acl.gov/ltc/costs-and-who-pays/what-is-long-term-care-insurance/receiving-long-term-care-insurance-benefits
Agency for Integrated Care. (2026). How to qualify CareShield Life claims? Government of Singapore, Ask.gov.sg. https://ask.gov.sg/aic/questions/cmt2f6flw008k4g01dmombzp9
Ankuda, C. K., Levine, D. A., Langa, K. M., Ornstein, K. A., & Kelley, A. S. (2020). Caregiving, recovery, and death after incident ADL/IADL disability among older adults in the United States. Journal of Applied Gerontology, 39(4), 393–397. https://pubmed.ncbi.nlm.nih.gov/30741065/
California Department of Insurance. (n.d.). Long term care insurance. https://www.insurance.ca.gov/01-consumers/105-type/95-guides/05-health/01-ltc/ltc-insurance.cfm
Central Provident Fund Board. (n.d.). CareShield Life. Government of Singapore. https://www.cpf.gov.sg/member/healthcare-financing/careshield-life
Central Provident Fund Board. (n.d.). When am I eligible to make a CareShield Life/ElderShield claim? Government of Singapore. https://www.cpf.gov.sg/service/article/when-am-i-eligible-to-make-a-careshield-life-eldershield-claim
Federal Republic of Germany. (2026). Sozialgesetzbuch (SGB) XI, Anlage 2: Bewertungssystematik. Official federal law publication. https://www.gesetze-im-internet.de/sgb_11/anlage_2.html
Federal Republic of Germany. (2026). Sozialgesetzbuch (SGB) XI, § 14: Begriff der Pflegebedürftigkeit. Official federal law publication. https://www.gesetze-im-internet.de/sgb_11/__14.html
Federal Republic of Germany. (2026). Sozialgesetzbuch (SGB) XI, § 15: Ermittlung des Grades der Pflegebedürftigkeit, Begutachtungsinstrument. Official federal law publication. https://www.gesetze-im-internet.de/sgb_11/__15.html
Finkelstein, A., & McGarry, K. (2006). Multiple dimensions of private information: Evidence from the long-term care insurance market. American Economic Review, 96(4), 938–958. https://doi.org/10.1257/aer.96.4.938
Government of the Netherlands. (n.d.). How can I apply for a Wlz care-needs assessment? https://www.government.nl/documents/question-and-answer/how-can-i-apply-for-a-wlz-care-needs-assessment
Government of the Netherlands. (n.d.). Long-term Care Act (Wlz). https://www.government.nl/themes/family-health-and-care/nursing-homes-and-residential-care/long-term-care-act-wlz
Katz, S., Ford, A. B., Moskowitz, R. W., Jackson, B. A., & Jaffe, M. W. (1963). Studies of illness in the aged: The index of ADL—A standardized measure of biological and psychosocial function. JAMA, 185, 914–919. https://pubmed.ncbi.nlm.nih.gov/14044222/
Kim, H., Jung, Y.-I., Sung, M., Lee, J.-Y., Yoon, J.-Y., & Yoon, J.-L. (2015). Reliability of the interRAI Long Term Care Facilities (LTCF) and interRAI Home Care (HC). Geriatrics & Gerontology International, 15(2), 220–228. First published online August 27, 2014. https://pubmed.ncbi.nlm.nih.gov/25163513/
Lawton, M. P., & Brody, E. M. (1969). Assessment of older people: Self-maintaining and instrumental activities of daily living. The Gerontologist, 9(3), 179–186. https://doi.org/10.1093/geront/9.3_Part_1.179
Mahoney, F. I., & Barthel, D. W. (1965). Functional evaluation: The Barthel Index. Maryland State Medical Journal, 14, 61–65. https://pubmed.ncbi.nlm.nih.gov/14258950/
Ministry of Health Singapore. (2025). CareShield Life 2025 Review: Frequently asked questions. https://www.moh.gov.sg/managing-expenses/schemes-and-subsidies/careshield-life/careshield-life-2025-review-faqs/
Ministry of Health Singapore. (n.d.). Assessment guidance on cognitive conditions and severe disability. Ask.gov.sg. https://ask.gov.sg/moh/questions/clpjhuuov009shipnyk1tgv7u
National Association of Insurance Commissioners. (2017). Long-Term Care Insurance Model Regulation (MO-641). https://content.naic.org/sites/default/files/committee_related_documents/ltci_model_reg_641.pdf
National Association of Insurance Commissioners. (n.d.). Long-term care insurance. https://content.naic.org/insurance-topics/long-term-care-insurance
Organisation for Economic Co-operation and Development. (2024). Growing long-term care needs across OECD countries. In Is care affordable for older people? OECD Publishing. https://www.oecd.org/en/publications/is-care-affordable-for-older-people_450ea778-en/full-report/component-5.html
Organisation for Economic Co-operation and Development. (2024). Is care affordable for older people? OECD Publishing. https://www.oecd.org/en/publications/is-care-affordable-for-older-people_450ea778-en.html
Organisation for Economic Co-operation and Development. (2024). Methodological annex. In Is care affordable for older people? OECD Publishing. https://www.oecd.org/en/publications/is-care-affordable-for-older-people_450ea778-en/full-report/component-9.html
Society of Actuaries. (2018). Researching long-term care insurance incidence. https://www.soa.org/resources/experience-studies/2018/researching-ltc-insurance-incidence/
Society of Actuaries. (2020). 2000–2016 long-term care aggregated databases. https://www.soa.org/resources/experience-studies/2020/2000-2016-ltc-aggregate-database/
U.S. Department of the Treasury. (1997). Treasury Benefits Tax Counsel J. Mark Iwry, House Ways and Means Subcommittee on Health: Testimony concerning Notice 97-31 and long-term care insurance provisions. https://home.treasury.gov/news/press-releases/sp092597
United States Congress. (2026). 26 U.S.C. § 7702B: Treatment of qualified long-term care insurance. Office of the Law Revision Counsel. https://uscode.house.gov/us/usc/119-118/t26/s7702B/
World Health Organization. (2001). International Classification of Functioning, Disability and Health. https://www.who.int/classifications/international-classification-of-functioning-disability-and-health
World Health Organization. (2020). Healthy ageing and functional ability. https://www.who.int/news-room/questions-and-answers/item/healthy-ageing-and-functional-ability