RESEARCH DOSSIER
Long-Term Care Insurance Fraud Abuse and Provider Billing Controls
Abstract. Long-term care insurance exposes payment systems to risks that arise from the difficulty of observing care, the vulnerability of beneficiaries, the diversity of benefit designs, and fragmented responsibility for verifying delivery. This monograph examines intentional deception, erroneous billing, abusive charging, deficient documentation, and the separate problem of abuse or neglect of care recipients. It establishes the evidentiary requirements for a defensible billing determination, compares private long-term care policies with United States Medicaid-funded personal care, German statutory long-term care insurance, English publicly financed adult social care, and French need assessments, and evaluates available payment-control and investigative evidence. Primary statutes, official audits, model insurance regulation, and peer-reviewed research support the comparative analysis. The central finding is that trustworthy payment integrity depends on the congruence of entitlement, authorization, actual service, correct charge, and fair review, with each step independently open to verification. Aggregate improper-payment estimates, electronic visit records, suspicious classifications, and enforcement allegations measure different events; their analytical separation determines the accuracy of fraud claims. The study develops original comparative interpretations of evidence design, control incentives, adjudication fairness, and empirical validation without introducing a new insurance theory. Its proposed operational tests are expressly analytical applications of established audit, statistical, and legal methods. No claim is made that the study itself measures fraud prevalence or establishes causal reductions from particular controls.
Keywords: Long-term care insurance; insurance fraud; abusive provider billing; Medicaid personal care services; electronic visit verification; claims evidence; improper payments; older adult financial abuse; provider audits; reimbursement insurance; German long-term care insurance; consumer appeals; payment integrity; actuarial exposure; fair enforcement.
Introduction
Long-term care is produced through episodes of assistance, supervision, nursing, rehabilitation, and institutional residence that can persist for years. The person receiving care may depend on others for mobility, cognition, documentation, communication, or financial administration. Under these conditions, invoices and claim files often become the principal observable record of activity that took place outside the payer's direct view. This creates a distinctive problem of verification. A payer must decide whether an entitlement arose, a particular service was delivered, the contractual or statutory requirements were satisfied, and the monetary demand matches that service. The beneficiary must have access to timely payment and safe care, while workers and providers must be able to correct records and challenge erroneous allegations.
Care need itself is heterogeneous. The World Health Organization's World Health Organization analysis of elder abuse distinguishes financial exploitation, neglect, psychological harm, and other forms of mistreatment. Billing manipulation may coincide with these harms but concerns an additional financial representation: what the payer was told happened, by whom, when, for what price, and under which entitlement. The WHO's estimates of abuse across community and institutional settings therefore cannot legitimately be relabeled as a rate of claims fraud.
This investigation concerns the conjunction of provider billing evidence, claim anomaly verification, and consumer rights. The research object is broader than payment coding alone, yet precisely bounded. Private long-term care insurance transfers a defined contractual risk. Medicaid home- and community-based services are publicly funded benefits subject to federal and state program rules. German social long-term care insurance establishes statutory entitlements and provider billing procedures. English local-authority social care funds services through public arrangements, including direct payments, while French long-term care assessment offers a useful empirical comparison for financial incentives. Each system produces different payment obligations and audit powers. Comparisons are valid only after identifying whose funds are transferred, which contractual or legal trigger governs the transfer, and which data document the care.
The study examines primary legal texts as available on November 16, 2026, current oversight material, historical audits, and selected peer-reviewed evidence. Statutory language is treated according to jurisdiction and effective date. Observed audit anomalies are attributed to the examined samples, not projected as global experience. Econometric results are interpreted according to their identification assumptions and population. The manuscript's contribution lies in its systematic reconciliation of claim proof, incentive analysis, and proportional redress, together with specific researchable consequences for policy and operational evaluation.
Historical Development, Conceptual Foundations and Subject Delimitation
From private reimbursement to public care financing
Private long-term care protection developed around the difficulty of financing extended assistance with activities of daily living and cognitive impairment. Its payment obligation is defined by the issued contract, applicable law, benefit triggers, exclusions, maximums, and claims procedures. In the United States, a private policyholder may receive money directly, or a covered service provider may be reimbursed. The NAIC shopper's guide to long-term care insurance distinguishes expense-incurred, indemnity, and disability forms of payment. That distinction determines whether the expense invoice is a condition of the amount payable, an evidentiary indicator of benefit eligibility, or merely contextual information.
For an expense-incurred design, a document showing the kind, date, expense, and provision of the qualifying service can determine the benefit amount. A fixed indemnity payment is ordinarily established by the insured condition and contractually required qualifying criteria; it need not equal a provider's charge. A disability-style payment can be contractually payable despite no service being purchased during the relevant period. A provider billing control that simply requires receipt-for-reimbursement evidence in every design can therefore become a source of wrongful denials. Conversely, an insurer that accepts invoices at face value without assessing eligibility and delivery may encourage excess charges. The relevant proof burden follows the policy's payment rule.
The American Medicaid personal care landscape adds an administrator, a service authorization system, individual attendants, consumer-directed care, and potentially managed care plans. A GAO comparison of personal care safeguards reported in 2016 that four selected states varied in beneficiary monitoring, timesheet signatures, and use of electronic timekeeping. The report established a historical oversight gap and uneven safeguards, rather than a uniform baseline for every state. Subsequent federal EVV requirements expanded the evidence available for specified in-home services.
German social care insurance has a different institutional origin: a statutory financing and entitlement arrangement with designated care funds, negotiated provider compensation, and legally specified billing information. England's direct-payment arrangements place important decisions and accounting responsibilities with an individual or authorized person. The coexistence of these structures makes long-term care billing a useful comparative problem, but category equivalence must not be presumed. A direct-payment misuse allegation in English public social care and a false expense receipt under an American private contract involve different legal obligations even when the same factual act, such as a nonexistent visit, appears in both.
Definitions tied to the evidentiary question
For this monograph, fraud denotes intentional deception to obtain or retain an advantage to which the actor is not entitled under the applicable legal standard. Proof of intent and material representation depends on the governing cause of action. An improper payment denotes a payment that violates a program or contract requirement, is in an incorrect amount, or lacks required supporting information. Abuse in the provider-billing context denotes practices contrary to applicable payment standards or reasonable professional integrity, including systematically excessive or misleading charges. Elder abuse denotes mistreatment of a person in a relationship of trust, including financial exploitation or neglect, and may give rise to safeguarding duties independently of claims adjudication.
These definitions identify several possible outcomes of one missing record. An electronic visit entry may be absent because care was never delivered, because a worker's device failed, because manual correction was used, because an exemption applies, or because the payer's data transfer failed. A claim can then be unsupported, conditionally payable, recoverable, or suspicious without each status having the same meaning. CMS reported that insufficient documentation dominated its 2025 error estimate; this directly contradicts treating the entire estimate as a tally of demonstrated intentional wrongdoing.
The financial objective is similarly differentiated. A payer seeks to prevent unauthorized disbursement and to honor valid claims at the correct time. A claimant seeks continuity of care, understandable evidence requests, and a credible opportunity to challenge decisions. A provider requires consistent coding, advance knowledge of billing rules, and a process for correction. The credibility of a control depends on its performance across all of these interests. Avoidable delay and denial are measurable adverse results of an excessively aggressive fraud program, just as leakage is a measurable result of insufficient scrutiny.
Principal actors and responsibility
The same claim may involve an insured person, legal representative, care recipient, attendant, service agency, supervising clinician, assessment body, payer, public regulator, and independent reviewer. Operational responsibility should be mapped to the authority each actor actually holds. Service authorization does not necessarily certify attendance; attendance does not establish that the right service was provided; and a provider's invoice is a monetary assertion, not independent proof of all constituent events.
A claim investigation therefore needs separate records of authorizations, care plans, staff assignment, recorded attendance, service delivery, invoice production, payment, changes, and appeal. It should identify the author and modification history of each record. A beneficiary's signature can corroborate an event but may not establish the correctness of time, rate, or classification, particularly where dependence or cognitive impairment affects the reliability of the attestation. Such concerns are supported by HHS-OIG's personal care fraud and patient-harm advisory, which joins financial and safety risks in a single operational environment.
Provider Billing Evidence and the Construction of a Verifiable Claim
What a claim must establish
A defensible claim record answers six concrete questions: whether the person and provider were eligible; whether the service was authorized when authorization was required; whether the care episode actually occurred; whether its quantity and classification are supported; whether the charge respects the applicable rate and coverage terms; and whether payment or adjustment already took place. These are familiar audit assertions applied to long-term care, not a newly postulated framework.
The record must preserve four identities: the beneficiary, the person who performed the work, the entity that billed, and the payer that bears the cost. Identity controls are essential because an agency may submit consolidated charges for many attendants and because the authorized representative managing an individual budget may not be the person receiving assistance. Every control should also preserve an event time, a billing time, and, where records change, a correction time. In retrospect, the temporal sequence often explains whether a mismatch indicates late documentation or prospective deception.
German statutory law supplies unusually concrete billing requirements. Under SGB XI section 105, providers participating in care must record the nature, quantity, price, day, and time of billed services and include specified identifying information. The resulting record facilitates comparison of hours, services, staff, and charges. Its legal force is tied to Germany's social care insurance; it is not a universal clause that can be transplanted into every private reimbursement contract.
Temporal proof and visit verification
United States federal Medicaid EVV guidance explains that the 21st Century Cures Act requires electronic verification for qualifying personal care and home health services involving in-home visits, with implementation milestones identified for personal care in 2020 and home health in 2023, subject to the statutory compliance framework. EVV produces evidence of visit attributes and supports authorization and billing comparisons. It does not establish all dimensions of service quality, medical necessity, work performance, or fraudulent intent.
A reliable time control should compare the individual service plan and approved hours with the recorded encounter, the invoice, provider assignment, beneficiary availability, and known institutional admissions. It must allow legitimate corrected entries to remain visible rather than overwriting the historical record. An apparently simultaneous claim for a hospital stay and a home-care visit is a high-value verification target, but a factual conclusion requires checking hospital admission times, discharge times, allowable service exceptions, and whose care was actually billed.
The U.S. Department of Justice's 2026 fraud case summaries contain allegations of home-care claims submitted for periods of hospitalization and for dates of alleged nonperformance. These descriptions identify practical investigative patterns. Charges and complaints do not by themselves establish guilt, and enforcement-selected allegations supply no representative prevalence denominator. Their value in this study is typological: they demonstrate that intersecting independent records can identify a checkable contradiction.
Quantities, rates and service classification
Even when the encounter occurred, the invoice can contain excess units, overlapping shifts, duplicate submissions, unsupported higher-intensity codes, unauthorized tasks, incorrect rates, or charges attributed to another payer. A single high charge does not establish error because contracts, negotiated rates, geographical cost structure, and care intensity differ. A correct audit joins each invoice line to the tariff, contract version, benefit category, worker role, and time-specific authorization that govern it.
The provider should have a way to submit an explanation before an anomaly becomes an adverse factual finding. For example, two attendants may legitimately work together for a transfer, while a system that assumes one worker per beneficiary would flag the second visit. An overnight service can cross midnight. A telephone check-in may supplement, without being confused with, an in-home episode. A fixed private indemnity benefit may remain payable even if the beneficiary temporarily uses unpaid family care. These are distinct from a fictitious visit or a knowing duplicate request.
The evidentiary unit should also preserve adjustments and reversals. Rebilling after a rejected invoice can appear to be a duplicate if the claim system ignores its correction chain. A valid recovery can be misclassified as a new negative payment. Proper matching uses transaction identifiers, prior-claim links, service-period coverage, and the net amount ultimately borne by the payer.
Invoices as representations and receipts as corroboration
An invoice is evidence that an entity requested a payment; payment evidence shows a transfer; neither is automatically proof of performance. Progress notes and a service plan offer richer content but may be written retrospectively or copied. EVV logs can constrain the sequence but also contain manual exceptions. A beneficiary interview may corroborate the visit while being vulnerable to recall limitations. Convergence among substantially independent sources carries greater evidentiary weight than multiple fields copied from one timekeeping system.
This distinction yields an operational requirement: the investigation file should preserve both the document and its provenance. A billing line that matches a provider-generated timesheet has weaker independence than a billing line that matches an independently registered hospital discharge, beneficiary statement, and policy-specific eligibility determination. Investigators should record data lineage, changes, dates, and source reliability, not merely the number of matching fields.
Institutional Mechanisms and Comparative Legal Arrangements
American private long-term care insurance
Private long-term care claims require the actual policy, issue date, applicable endorsements, and state insurance law. The NAIC Long-Term Care Insurance Model Regulation provides an influential template for benefit-trigger determinations, internal appeals, and prompt payment of clean claims. Model section 32 contemplates payment or a specified written response within 30 business days for an appropriately documented claim, together with further requirements when additional information is requested; it also identifies an exception where the insurer has a reasonable basis supported by specific information that the claim was fraudulently submitted. These are features of the model and must be checked against actual state adoption before asserted as rights in a particular dispute.
The analytical significance is a controlled relationship between prompt payment and justified inquiry. A demand for additional documentation should identify precisely what is needed to decide the claim. An allegation of fraud should rest on particularized contradictory evidence. A claim should not remain indefinitely suspended because an algorithm assigns a high risk score. Separating the contractual benefit trigger from the financial invoice prevents a billing investigation from silently revising the agreed coverage.
United States Medicaid-funded personal care
Medicaid financing often requires a state-specific combination of eligibility, authorized services, agency or consumer-directed model rules, and documentation standards. HHS-OIG's 2016 investigative advisory describes patterns of fraudulent billing and concerns about harm and neglect, recommending consistent worker identification and stronger monitoring. The evidence supports attention to both billing conduct and recipient protection. It does not imply that every worker, beneficiary, or program is fraudulent or that home-based care is inherently less trustworthy.
The 2016 GAO comparison and subsequent federal verification requirements show why control architecture must be dated. A retrospective assessment of conduct in 2016 cannot simply apply a later state's EVV implementation requirement without establishing whether the rule then governed that service. The same principle applies to Medicaid managed care: a state-to-plan capitation transaction and the plan's payment to a care provider are different payment layers, with different data and audit responsibilities.
GAO's 2025 managed-care payment analysis observes that near-zero managed-care improper-payment estimates do not capture some risks in downstream plan-to-provider payments, including services never furnished. Consequently, low values in a narrow reporting system should not be mistaken for comprehensive assurance that care-service billing is accurate.
German social long-term care insurance
Germany's SGB XI provides explicit pathways for examining financial and service integrity. Section 79's billing-audit conditions permit the designated care-fund associations to conduct billing examinations when factual indications of erroneous billing exist, with statutory procedural and documentation conditions. The predicate matters: the examination is connected to specified evidence, preserving a distinction between targeted control and indiscriminate surveillance.
Section 47a's anti-misconduct provisions establish dedicated institutional functions and conditioned information exchanges. They illustrate the need to define both investigatory competence and data-use limits. Information collected for investigating a billing discrepancy acquires risk when repurposed beyond the permitted task. In parallel, section 114's quality inspections address whether care meets standards; billing examinations and care-quality investigations can interact, but their statutory purposes and remedies must be identified separately.
For recipients, section 108's itemized-information rights are especially important. Insured persons can obtain information about charges attributable to them, including a service-and-cost overview extending at least 18 months before a request and comprehensible billing content. The right makes an otherwise opaque three-party relationship more contestable: the payer, provider, and recipient can confront the same recorded services.
England's adult social care direct-payment comparison
English local-authority adult social care is an institutional comparator, not an equivalent private insurance product. Under Care Act 2014 provisions on direct payments, individuals or authorized persons may receive a personal budget under specified conditions, and the governing restrictions require that payments meet assessed care and support needs. Section 33 concerns conditions and potential repayment when these are breached, while section 42 addresses safeguarding enquiries and expressly includes financial abuse.
The care and support statutory guidance recognizes that managing a direct payment may create risks when a household member or representative also has control of the beneficiary's finances. It supports administrative agreements, appropriate monitoring, and safeguarding arrangements. This setting demonstrates how a payment-integrity problem can exist without an insurer adjudicating a policy claim, and why the rights of the person receiving care must remain visible even when another party handles the money.
France and the classification of care need
Eligibility classification may itself respond to incentives. Roy's French study of strategic need coding linked survey and administrative assessments and found systematic differences associated with who assessed care need. Its interpretation includes incentives for overclassification by some care homes and possible underclassification in community settings, with attention to limitations of the assessment instrument. This finding complicates the convenient assumption that higher coding always reflects intentional fraud. If assessment categories poorly capture complex patient needs, some deviations may arise from the instrument, while others may reflect opportunistic revenue or cost shifting.
The relevant policy question is whether eligibility and intensity judgments can be independently audited, compared with direct functional evidence, and reviewed in light of incentives. This concern applies across countries without making French effect estimates transferable to German, English, or American contracts.
Comparative Case Studies and Measured Financial Consequences
Colorado's 2026 electronic-visit audit
The HHS-OIG Colorado EVV audit examined a sample of 160 personal care claims and reported 91 records with EVV compliance issues, 39 unallowable service errors, and 43 errors set aside for additional determination. The categories are audit findings in the examined sample, not logically disjoint populations, so their counts cannot be added into a single number of erroneous claims. The OIG recommended recovery of an estimated $8,072,870 in federal-share overpayments and determination of the allowability of a further estimated $45,688,080. Colorado disagreed with most findings and recommendations. A set-aside amount is a question requiring resolution, not a confirmed loss.
The case supports a narrow but important proposition: possession of an EVV system does not establish a complete control. The OIG identified gaps involving missing visit-record matches, manual entries, service locations, timesheet-to-worker concordance, approved units, and rates. Recommendations combined system edits, provider monitoring, staff verification, and care-plan controls. The audit therefore treats EVV as one component in a documented reconciliation chain.
Its monetary estimates arise from a particular sampling and projection methodology, a defined claim population, and specified applicable rules. The audit cannot be used to infer the prevalence of fraud among all American long-term care providers, all private LTC insurers, or all Colorado residents. It also does not demonstrate that a proposed system edit would causally save the projected amount in future periods. That hypothesis requires implementation and subsequent evaluation.
Aggregate payment-error statistics and the denominator problem
CMS 2025 Medicaid payment-error reporting gives a rolling national estimate of $37.39 billion, or 6.12%, assembled from the 2023, 2024, and 2025 review cycles. It includes more than long-term care, and the combined numerator includes payment problems other than deliberate fraud. Its 2025 fact sheet attributes 77.17% of the estimated improper-payment total to insufficient documentation. A scientific inference equating $37.39 billion to stolen long-term care benefits would therefore involve errors of scope, classification, and legal characterization.
One denominator consists of reviewed Medicaid transactions and related eligibility determinations; another consists of personal care visit claims; another consists of insured expense-incurred LTC reimbursement; and a fourth consists of adjudicated cases of intentional deception. None can substitute for the others. A cross-jurisdiction fraud rate requires a stable definition of the claim population, unit, period, audit selection, evidence of intent, and treatment of unresolved cases. The present primary sources do not provide a common global denominator.
Medical services furnished to nursing-home residents
HHS-OIG's audit of selected Medicare Part B services identified 91 noncompliant sampled items among 150 reviewed and estimated about $19.5 million in improperly paid selected services during its audit period. This is relevant to overlapping nursing-home financial relationships, but it concerns Medicare Part B payments for medical services furnished to facility residents, not the private long-term care insurance benefit itself. The unit of payment, coding rules, and payer are separate.
The practical risk is a facility or group billing multiple streams for different kinds of care. Reviewers should prevent both duplicate charges and unjustified cross-program conflation. A Medicare psychotherapy service, Medicaid personal assistance hour, private LTC nursing-home reimbursement, and resident room-and-board charge may relate to the same person and date without being interchangeable services. Payment integrity means testing each claim against the appropriate covered service and payer, then identifying prohibited duplication where facts and rules support it.
Incentive differences in French assessments
Roy's 2025 analysis permits a comparative reading of assessment incentives that goes beyond detecting anomalous charges. When the assessment body also receives higher revenue from more severe classifications, diagnostic or functional need categories can become endogenous to payment. When the assessing body bears the cost, there can be pressure in the opposite direction. Roy interprets both directions against the survey data and institutional incentives, while noting the possibility that the classification technology itself fails to express the time required for complex care. The result is important for audit design: adjudicators should examine both excessive awards and suppressed entitlement, not only financially costly overclassification.
External validity remains bounded. The study examines a particular French population, survey design, and care-financing system. It cannot establish a general numerical upcoding fraction for Germany's care grades or American private insurers. Its broader contribution is an empirically supported reason to test assessor incentives instead of assuming that assessments are exogenous and neutral measurements.
Consumer protection as an observable financial outcome
Recovery of overpayments is only one metric. A fair system also measures legitimate-claim payment time, underpayments later reversed, unresolved care disruption, beneficiary costs of appealing, and access to records. German statutory disclosure rights, England's safeguarding framework, and the NAIC model's claims-response requirements illustrate institutional routes for measuring those outcomes.
A scientifically adequate outcome assessment should compare the monetary amounts corrected, the time spent in review, the distribution of errors across beneficiaries and providers, the quality of service continuity, and the privacy cost of evidence collection. A fraud control that prevents a relatively small amount of unauthorized payment while repeatedly delaying approved essential care might fail a welfare evaluation even if a narrow recovery metric appears favorable. Such a conclusion requires measured costs and benefits; the proposition is an analytical criterion, not an empirical estimate from the present sources.
Claim Anomaly Verification and the Economics of Billing Controls
From anomaly signal to tested explanation
A screening system selects discrepancies worth investigation. Examples include inconsistent identity numbers, claim periods overlapping hospitalization, recorded hours exceeding approved units, implausible overlap between workers, repeated rounded visits, abrupt shifts in service intensity, and unusually high corrections. Each signal admits competing explanations. Staffing shortages, complex care needs, regional travel, coding changes, device limitations, historical contract amendments, and uneven documentation resources can create patterns resembling deliberate manipulation.
The correct workflow is to identify the signal, reconstruct the governing rule, request the most discriminating evidence, preserve the response and corrections, and determine the appropriate disposition. A signal's usefulness is measured by the additional evidence it helps investigators find. A rule should state its expected failure modes and exceptions before deployment. Repeated manual-entry flags, for example, may indicate a provider's device problem rather than inflated service provision; the distinction can be tested using repair logs, backup records, and independent visit corroboration.
Shekhar, Leder-Luis, and Akoglu's explainable Medicare fraud-detection study investigates unsupervised provider-level hospital claims screening. It provides evidence that anomaly methods can prioritize audit attention without a large set of confirmed fraudulent labels. The unit is a hospital billing pattern within Medicare, however. Its performance cannot be presumed for home attendants, residential long-term care, or private LTC policies with fixed indemnity triggers. The evidentiary justification for a review must be supplied in the target context.
False positives, prevalence and investigative capacity
Consider an illustrative audit population in which 1% of claims truly contain a material wrongful payment, a screening rule identifies 80% of those claims, and correctly clears 95% of legitimate claims. In 10,000 claims, the rule flags 80 of 100 problematic claims and 495 of 9,900 legitimate claims. Its positive predictive value is 80 / 575, approximately 13.9%. These inputs are hypothetical, not measured long-term care performance. They illustrate why a seemingly strong sensitivity and specificity can produce numerous false allegations when the target event is uncommon.
The consequence is operational rather than rhetorical. Investigative labor is scarce; staff must prioritize evidence-rich cases. A model trained on previously investigated providers can learn the selection criteria used by investigators rather than actual misconduct. A broad fraud label combining missing forms, incorrect rates, disputed services, and criminally proven deceit corrupts training targets. Precision, recall, calibration, differential error rates, and review burden should be reported separately.
If control cost is C, the expected preventable unauthorized payment is L, the probability of effective detection and prevention is q, and expected harm from wrongful interruption of valid care is H, a simplified comparison is qL - C - H. This is a standard expected-cost accounting identity applied to the problem; it does not establish numerical values or a new formal doctrine. Its importance is to force each economic claim to identify what evidence is needed before a control is called cost-effective. Additional implementation burdens, privacy costs, provider exit, and downstream hospitalization should be incorporated when the decision context makes them material.
Verification depth and proportional review
Routine mechanical checks can be made before payment: duplicate identifiers, authorization windows, tariffs, provider enrollment, incompatible time overlaps, and arithmetic consistency. Medium-risk cases may need supporting documentation or independent beneficiary confirmation. High-risk and well-supported allegations may justify formal audit, recovery, referral, or safeguarding investigation under the governing law. Each level should use the lowest intrusive evidentiary method sufficient to answer the specific claim question.
An audit trail needs to record what was originally submitted, what was requested, who responded, the basis for the decision, and how the disposition changed after appeal. A human reviewer should be able to reconstruct the calculation and cited rule. A score without explanatory features or associated contractual criteria does not establish either a false claim or an entitlement to withhold a benefit.
Care continuity changes the stakes of provisional measures. When a disputed portion of an invoice can be separated from a valid portion, the reviewer should examine whether the governing contract or statute supports prompt payment of the undisputed amount. Where services must continue for safety, administrative actions should be assessed for their effect on care recipients. Such operational recommendations require tailoring to the legally authorized process and cannot override statutory rules.
Provider behavior and the boundaries of deterrence
Payment systems shape the behavior of their participants. Fixed reimbursements can encourage less documentation-intensive care; time-based systems may encourage inflated hours; severity-adjusted payments can encourage more intensive classification; under-resourced budgets can encourage downcoding and cost shifting. Roy's empirical work demonstrates that the direction of classification incentives matters. None of these incentives prove that a particular provider acts dishonestly. The relevant research task is to test whether observed practice changes when payment rules change, while accounting for patient severity and provider composition.
A successful control should be resistant to simple substitution. If an agency responds to a new timekeeping requirement by inflating service categories rather than hours, a program monitoring only timestamp accuracy will miss the change. If excessive verification burdens induce reputable home-care providers to leave a market, an apparently lower fraud exposure may coincide with reduced care supply and additional beneficiary harm. These possible effects warrant longitudinal measurement before policy conclusions.
Consumer Rights, Safeguarding and Proportional Enforcement
Claimant participation and intelligible records
A person receiving long-term care has a direct interest in what is claimed on their behalf even where the payer pays the agency. The German right to care billing information illustrates how a service recipient can inspect descriptions, costs, and submitted bill content in comprehensible form. That right supports error detection by someone with firsthand knowledge while remaining consistent with protections for personal data. Transparency may also expose discrepancies that would otherwise remain concealed in provider-to-payer exchanges.
Private LTC claimants need accurate information about the benefit trigger, policy version, invoices, coverage limit, documentation requirement, and precise reason for adverse determinations. A reviewer should consider whether cognitive impairment or disability affects the person's ability to answer a routine demand for documents. A legal representative may assist, but a risk of exploitation by the representative must also be evaluated. These concerns support clear authorizations, auditable communications, and appropriate safeguarding referral rather than treating silence as evidence of fraud.
The NAIC model's appeal and claims language links timely decisions with notice and review procedures, subject to individual states' adoption. The model's specific fraud-related exception is a reminder that suspicion requires a reason grounded in specific information. A fair adverse determination should state the disputed material fact, the governing rule or policy term, the evidence considered, and the route to reconsideration.
Elder abuse and provider misbilling
Financial manipulation can be accompanied by neglect, coercion to sign a timesheet, diversion of funds, or intentional deprivation of services. WHO's elder abuse evidence summary provides the public-health categories, while the HHS-OIG personal care advisory documents the coexistence of fraudulent billing and patient harm within investigated cases. A billing investigation should therefore have a pathway for urgent protection when available information indicates present risk to an individual, irrespective of how quickly a financial overpayment can be adjudicated.
English Care Act safeguarding provisions make financial exploitation part of the statutory safeguarding concept. The comparison illustrates why institutions responsible for repayment may need to coordinate with institutions responsible for personal safety. The purpose and legal authority for information sharing must be recorded: a legitimate safeguarding concern does not create an unlimited license to combine all private information across agencies.
Privacy, surveillance and the limits of digitized verification
The promise of EVV is greater temporal observability. Its costs may include intrusive geolocation, administrative friction, inability to accurately record exceptions, and altered working conditions. Mateescu's interview-based study of Medicaid homecare surveillance reports that visit verification can shape workers' and recipients' movements and increase bureaucratic burdens. This qualitative study does not quantify EVV's overall fraud-prevention effects, but it is directly relevant to the design and welfare cost of verification.
Proportional controls can use minimal needed records, limited retention, meaningful correction routes, restrictions on secondary use, and human review of exceptions. The degree of intrusion should correspond to the claim-specific risk and statutory authority. A system should be tested for its effect on workers with limited connectivity, persons in rural areas, individuals who require irregular care patterns, and households with cognitive or communication barriers. That inquiry concerns substantive quality and equality of access, not only data-protection paperwork.
Investigations, recovery and accountability
Different suspected events call for different remedies. An arithmetic mistake may warrant correction; an unsupported expense may warrant supplementary records or repayment under the applicable rule; a repeated knowing false statement may warrant formal investigation; a credible risk of neglect may warrant immediate safeguarding action. Civil recovery, contract sanctions, administrative penalties, criminal prosecution, and professional discipline follow separate legal tests. A reviewer should not use the severity of a possible sanction to infer the strength of the evidence.
Where a person is accused, procedural accuracy includes notice, source disclosure as legally required, opportunity to submit contradictory records, independent escalation, and preservation of the original and revised decisions. Prosecutors' allegations remain allegations unless adjudicated. The payer's obligation to investigate suspected misuse should proceed alongside its obligation to distinguish error from intent. This separation improves both evidentiary reliability and legitimacy of enforcement.
Source Validation, Causal Inference and Research Limitations
Documentary hierarchy and version control
The research gives primary legal texts precedence for present legal obligations, current agency guidance for official implementation descriptions, completed audits for their specified populations, and peer-reviewed empirical papers for the questions their methods actually identify. Each category has limits. A model insurance regulation has persuasive design relevance without becoming binding law. A current consolidated statute explains current duties without proving that identical language applied to a claim years earlier. An audit establishes findings conditional on selection, records examined, methodology, and the auditee's response.
The comparative legal analysis uses German SGB XI sections 47a, 79, 105, 108, and 114 as viewed in November 2026; Care Act 2014 Part 1 and current English statutory guidance; the NAIC 641 model; and CMS's published implementation guidance. Individual disputes require reconstruction of the precise prior statutory version, state adoption, issued policy, benefit schedule, service date, and rate agreement. The present monograph does not substitute an undated generic statement for any of those dated instruments.
Every substantive external source in this edition is listed in References and separately identified in the research evidence ledger. The public text uses human-readable source links instead of private research IDs. Directly verified URLs indicate an accessible institutional record, source landing page, published abstract, or other identified material. An inaccessible full text is not silently treated as if its methods and appendices had been independently inspected.
Audit samples and nonrepresentative enforcement data
Claims targeted by investigators usually have a higher prior level of suspicion than claims selected from the full payment population. Results of targeted review therefore should not be interpreted as random-sample incidence. Similarly, audit projection depends on whether a statistical sample was representative of its stated sampling frame and how errors were valued. The Colorado 2026 figures include separate categories of recorded issues, allowable determinations, and unresolved questions; they cannot be mechanically reaggregated as counts of intentional misconduct.
Prosecution summaries are even more selected: they reflect cases that crossed charging thresholds and received enforcement attention. Their role is to define potential failure modes and illustrate the evidentiary contrasts that investigators use. They do not support a claim that the most frequently prosecuted arrangement is the most common misconduct in the underlying population. This matters because policy designed around conspicuous cases may neglect more widespread administrative errors or beneficiary underpayment.
Identification of causal effects
To say that a particular verification intervention reduced false payments requires a counterfactual: what would have occurred in comparable care arrangements without that intervention? Simple before-and-after recovery totals are confounded by policy changes, claim volume, payer mix, provider exit, enforcement intensity, staffing levels, and changes in classification. Randomized or credible quasi-experimental rollout designs, matched comparisons with appropriate assumptions, or carefully identified discontinuities may help. Even these designs should report effects on valid-claim delay, care availability, privacy costs, and beneficiary outcomes.
A credible evaluation would define the unit as claim, service episode, beneficiary, provider, or payer before estimation. It would specify whether the outcome is a documentary error, a verified non-delivered service, an excess amount, a proved false representation, or a criminal judgment. It would predefine exclusions and outcome adjudication protocols and preserve separate measures for sensitivity, specificity, positive predictive value, false denial, corrected payment, and investigation cost.
Measurement should permit independent verification of the adjudicated ground truth and avoid leakage from the model's screening variables into the outcome classification. Particularly in long-term care, classification can be influenced by assessor incentives, and case mix can be related to both reimbursement and ability to document care. These features create genuine identification problems, not simply software-engineering issues.
Limits of the comparative evidence
The sample combines legal systems with different coverage structures, time periods, and payment units. It is selected for institutional contrast and source accessibility, not statistically representative of global long-term care insurance arrangements. German statutory provisions and English social care direct payments illuminate governance choices but are not legally interchangeable with U.S. private contracts. French causal interpretations concern assessment incentives in a specific system, and U.S. anomaly-detection results concern Medicare hospital billers rather than long-term care service claims.
The study does not have proprietary insurer claim files, linked person-level records across all jurisdictions, administrative adjudication samples large enough to determine fraud prevalence, or independent access to protected investigation records. It therefore makes no original numerical claim about international fraud prevalence, no calibrated actuarial savings forecast, and no finding of guilt against any particular person or agency. Its methodological strength lies in tracing which claims can and cannot be defended by the available sources and constructing operationally testable research questions.
Integrated Synthesis and Open Scientific Questions
What the comparative record establishes
The sources establish a recurring relationship among three problems. First, the benefit itself must be correctly defined: private expense-incurred reimbursement, fixed indemnity, public service authorization, and statutory care payments do not require identical proof. Second, the represented service must be verified with appropriate independent records and a defensible record of corrections. Third, the person affected by payment control must be able to understand and contest the relevant account, while immediate safety concerns are addressed under their own legal authority.
German section 105 demonstrates detailed provider recording duties, and section 108 demonstrates beneficiary access to charge information. The NAIC model illustrates response and appeal provisions for private contracts, subject to adoption. Medicaid EVV makes visit activity more observable but requires reconciliations beyond a simple electronic timestamp. England's direct payments show how representatives and personal budgets create financial misuse risks even outside ordinary commercial insurance. Roy's French study demonstrates that who classifies need can affect the measured eligibility category. These are converging findings about information, incentives, and accountability, not evidence of a single uniform global legal regime.
The Colorado OIG audit supplies a concrete example: the relevant correction recommendations targeted missing connections among visits, attendants, service plans, hours, and rates, while the state disputed many findings. The 2025 CMS data show why improper payments should remain distinct from fraud. The literature on EVV surveillance and provider anomaly detection adds a further operational discipline: technical detection cannot be assessed solely by the quantity of flags.
Consequences for insurer and public-payer practice
The immediate implication is that evidence requests should reflect the payment design and the precise fact at issue. Under expense-based coverage, invoices and receipts can be central; under fixed indemnity, they may have a different relevance. Provider contracts and administrative systems should preserve a complete correction trail and explicit authorization-to-billing links. Review thresholds should be chosen with regard to the base rate of true material misconduct, scarce investigative capacity, and the cost of erroneous interruption.
Organizations should publish or internally track differentiated performance: total claims checked, confirmed corrections, recovered and refunded amounts, overturned denials, provider response time, appeals, beneficiary harm indicators, and data-protection incidents. Such disaggregation can expose a program that reports substantial gross recoveries yet carries excessive valid-claim delays or records disputes repeatedly reversed. Comparative analysis should retain the responsible payer, service type, and year for every measure.
Open scientific questions
A first empirical question is whether the incremental introduction of independent beneficiary billing visibility reduces verified excess charges after controlling for changes in reporting and care use. The German information-rights arrangement provides an institutional example, while causal identification would require independently measured complaints and outcomes.
A second question concerns the comparative effect of strict real-time attendance verification versus a more permissive system with audited manual exceptions. Outcomes should include verified false payments, service supply, care continuity, worker retention, and the incidence of disputed legitimate visits. Interview evidence suggests that labor and privacy costs matter, but those costs need systematic quantitative measurement.
A third question is whether independent care-needs assessment changes both overclassification and underclassification in financially different institutions. Roy's French evidence motivates extensions into other financing systems with suitable exogenous assignment or policy variation. Comparability requires common functional measures, attention to cognitive impairment, and observation of appeals.
A fourth question is how much of measured payment error arises from record quality, actual nonperformance, or deliberate deception after independent adjudication. This is essential for appropriate enforcement thresholds and for unbiased modeling. The required data are unusually difficult to obtain because investigatory files, clinical and functional records, and private policy terms are dispersed and often confidential.
A fifth question concerns the distribution of enforcement costs. Restrictions and audit intensity may affect small providers, rural households, and persons with complex care needs differently. A policy can improve average billing accuracy while widening access disparities. A complete welfare analysis must recognize both dimensions.
Conclusion
Long-term care provider billing is an evidence problem embedded in a relationship of care. A valid payment decision requires the correct entitlement, a verifiable event where service evidence is contractually or legally relevant, an authorized quantity and charge, and a review process that can correct mistaken conclusions. These requirements operate across private insurance and public long-term care systems through different legal mechanisms.
The primary sources support concrete controls: recording service type, quantity, time, identity, and price; reconciling visits with authorizations, rates, and other records; protecting the audit trail; allowing beneficiaries to inspect claims made on their behalf; investigating specific inconsistencies; and preserving proportionate appeal, privacy, and safeguarding procedures. They also support clear limits on what may be inferred. An improper payment is not by definition fraud. A detected anomaly is a question for verification. A criminal charge is a claim to be adjudicated. A model insurance regulation is not a substitute for enacted state law. An audit of a selected population cannot establish a worldwide prevalence rate.
The original analytical contribution of this monograph is its integrated reading of evidence provenance, provider incentives, payer decision rules, and claimant protection across otherwise dissimilar financing systems. The governing conclusion is that payment integrity and care integrity must be evaluated together. A rigorous insurer or public payer investigates disputed charges while maintaining accurate, timely fulfillment of legitimate benefits and safeguarding the person whose care is at stake. Future research should test those outcomes with transparent denominators, independent adjudication, jurisdiction-specific rules, and credible causal designs.
References
Centers for Medicare & Medicaid Services. (2025). Fiscal year 2025 improper payments fact sheet. https://www.cms.gov/newsroom/fact-sheets/fiscal-year-2025-improper-payments-fact-sheet
Centers for Medicare & Medicaid Services. (2025). PERM error rate findings and reports. https://www.cms.gov/data-research/monitoring-programs/improper-payment-measurement-programs/payment-error-rate-measurement-perm/perm-error-rate-findings-and-reports
Centers for Medicare & Medicaid Services. (n.d.). Electronic visit verification. Medicaid.gov. https://www.medicaid.gov/medicaid/home-community-based-services/home-community-based-services-guidance-additional-resources/electronic-visit-verification
Department of Health and Social Care. (n.d.). Care and support statutory guidance. GOV.UK. https://www.gov.uk/government/publications/care-act-2014-supporting-implementation/care-and-support-statutory-guidance
Federal Republic of Germany. (2026). Sozialgesetzbuch XI, section 105: Abrechnung pflegerischer Leistungen (consolidated text). https://www.gesetze-im-internet.de/sgb_11/__105.html
Federal Republic of Germany. (2026). Sozialgesetzbuch XI, section 108: Auskünfte an Versicherte (consolidated text). https://www.gesetze-im-internet.de/sgb_11/__108.html
Federal Republic of Germany. (2026). Sozialgesetzbuch XI, section 114: Qualitätsprüfungen (consolidated text). https://www.gesetze-im-internet.de/sgb_11/__114.html
Federal Republic of Germany. (2026). Sozialgesetzbuch XI, section 47a: Stellen zur Bekämpfung von Fehlverhalten im Gesundheitswesen (consolidated text). https://www.gesetze-im-internet.de/sgb_11/__47a.html
Federal Republic of Germany. (2026). Sozialgesetzbuch XI, section 79: Wirtschaftlichkeits- und Abrechnungsprüfungen (consolidated text). https://www.gesetze-im-internet.de/sgb_11/__79.html
Government of the United Kingdom. (2014). Care Act 2014, c. 23, Part 1 (as amended). https://www.legislation.gov.uk/ukpga/2014/23/part/1
Mateescu, A. (2024). Working against the clock: Digital surveillance in US Medicaid homecare services. Journal of Sociology, 60(3), 560–576. https://doi.org/10.1177/14407833231226097
National Association of Insurance Commissioners. (2022). A shopper's guide to long-term care insurance. https://content.naic.org/sites/default/files/publication-ltc-lp-shoppers-guide-long-term.pdf
National Association of Insurance Commissioners. (2024). Long-term care insurance model regulation (Model 641; model text). https://content.naic.org/sites/default/files/model-law-641.pdf
Office of Inspector General, U.S. Department of Health and Human Services. (2016). Investigative advisory on Medicaid fraud and patient harm involving personal care services. https://oig.hhs.gov/reports/all/2016/investigative-advisory-on-medicaid-fraud-and-patient-harm-involving-personal-care-services/
Office of Inspector General, U.S. Department of Health and Human Services. (2026). Colorado could improve its electronic visit verification system and claimed federal Medicaid reimbursement for millions of dollars in personal care services that did not comply with federal and state requirements (A-07-24-03260). https://oig.hhs.gov/reports/all/2026/colorado-could-improve-its-electronic-visit-verification-system-and-claimed-federal-medicaid-reimbursement-for-millions-of-dollars-in-personal-care-services-that-did-not-comply-with-federal-and-state-requirements/
Office of Inspector General, U.S. Department of Health and Human Services. (2026). Medicare Part B services provided to patients residing in nursing homes: Selected billing oversight findings (A-06-21-04002). https://oig.hhs.gov/reports/work-plan/browse-work-plan-projects/medicare-part-b-services-provided-to-patients-residing-in-nursing-homes/
Roy, D. (2025). Strategic coding in the assessment of long-term care needs: Evidence from France. Health Economics, 34(6), 1035–1063. https://doi.org/10.1002/hec.4950
Shekhar, S., Leder-Luis, J., & Akoglu, L. (2026). Can machine learning target health care fraud? Evidence from Medicare hospitalizations. Journal of Policy Analysis and Management, 45(1), e70078. https://doi.org/10.1002/pam.70078
U.S. Department of Justice. (2026). 2026 national health care fraud takedown: Case summaries. https://www.justice.gov/fraud/2026-national-hcf-case-summaries
U.S. Government Accountability Office. (2016). Medicaid personal care services: CMS could do more to harmonize requirements across programs (GAO-17-28). https://www.gao.gov/products/gao-17-28
U.S. Government Accountability Office. (2025). Medicaid managed care: Improper payment estimate (GAO-25-107770). https://www.gao.gov/products/gao-25-107770
World Health Organization. (2024, June 15). Abuse of older people. https://www.who.int/news-room/fact-sheets/detail/abuse-of-older-people