North East Medical Services

What do they do? North East Medical Services provides primary and specialty care, pharmacy, dental care and other health services, with culturally and linguistically responsive care. More

Why this approach interests us

• Language and access barriers can prevent people from receiving effective care; responsive services can help bridge that gap.
• Existing liver-health services offer a plausible way to reconnect people with chronic hepatitis B monitoring and indicated treatment.
• Clinical infrastructure provides a concrete starting point for investigating additional patient follow-up.

Our main reservations

• The official donation recipient is a separate Foundation; a general donation has not been linked to additional hepatitis B care.
• The conditional navigation model relies on uncertain evidence transfer and does not establish the impact of ordinary Foundation giving.

What do you get for your dollar? Bay Area health impact per general donation remains unestimated. The separate conditional navigation model is about $1.20 million per 10 QALYs and requires a specific long-term funding commitment; it is not a price for an unrestricted donation. More

The recipient, the organization and the decision being evaluated

The first question is which entity receives the gift. The official NEMS support page directs mailed checks to North East Medical Services Foundation. The linked Give In May campaign identifies EIN 94-3171797, matching the Foundation's original filing. Operating North East Medical Services has EIN 94-1722562. They are related organizations, not interchangeable names for one return.

That distinction resolves an uncertainty in the earlier report. The previous review correctly warned that the Foundation recipient had not been verified. The current original return and official linked campaign provide a stronger identification. They do not, however, establish a restricted hepatitis B gift pathway. Verifying the recipient is necessary for donation readiness but insufficient for proving what the donation changes.

The Foundation's 2024 program description concerns leasing San Francisco facilities to operating NEMS for primary health services. Its Schedule I also reports a cash grant to operating NEMS. These are real routes by which a supporting organization can help clinical care. They are different from hiring a navigator against a defined registry panel. A donor cannot assume the Foundation's entire flexible budget follows the HBV model merely because the operating system runs an HBV program.

The original operating return describes a much larger organization. Managed-care services, primary and specialty medicine, pharmacy and other clinical functions dominate its program activity. NEMS also has related property entities and other affiliates. The goal of this report is not to consolidate that entire structure into an invented single health coefficient. It is to understand the ordinary donation boundary while preserving the narrower accepted clinical calculation as a conditional diagnostic.

The conditional intervention is additional, recurring follow-up for adults with diagnosed chronic hepatitis B who would otherwise miss appropriate monitoring and downstream care. It is not a confirmed new NEMS project. The model assumes a closed initial registry cohort and repeated staffing support. It is not a model of new screening, a single treatment course, all hepatitis services, general primary care or the Foundation's property portfolio.

An ordinary Foundation gift could support a different need. The current campaign names broad clinical and operational purposes and provides examples of prior spending. Those examples improve our understanding of how grassroots gifts may be used, but do not identify a stable allocation fraction. A gift supporting mold mitigation or vaccination should be evaluated through that mechanism, not assigned the modeled health of a long-term HBV navigator.

This report therefore has two layers. The organizational layer covers finances, current services, donation routing and funding conditions. The numerical layer preserves the accepted HBV calibration and adds explicitly unweighted diagnostics. The two layers are related but not collapsed. The headline must state which layer its number belongs to.

The preserved $1.201 million price is not an ordinary-gift expected value. It is also not automatically a Bay-specific price: the original calibration has no explicit beneficiary-geography parameter and was framed around a proposed local cohort. A future verified Bay-restricted cohort could justify a clear geographic boundary. Current broad NEMS operations, including Nevada, do not justify applying that boundary to every unrestricted gift.

There is no probability-weighted mixture in the accepted model. It contains optimistic, central and pessimistic positive scenarios, with additional null and harm diagnostics. Calling their average an expected return would impose weights that were never elicited. Calling the central scenario “the most likely outcome” would similarly add an unsupported probabilistic interpretation. We retain it as a central analyst scenario and leave weighted expectation null.

The giving status is HOLD for a donor seeking a verified low-cost health opportunity. This does not imply that NEMS has low total impact, that all ordinary gifts are ineffective or that property support is unimportant. It means that the actual gift-to-care pathway and the model's clinical transfer are insufficiently established for the stronger recommendation. The report makes those gaps concrete rather than treating uncertainty as an automatic rejection of further investigation.

A useful next decision is whether to pursue a targeted offer. If NEMS can identify an additional registry workload, the staffing and clinical complements it requires, the funding already committed and a credible follow-up period, the conditional model may become more relevant. If its actual priority is another service or capital need, a different model would be appropriate. The donor should not force a long-term navigation story onto an organization whose current request is materially different.

Current services, population and the meaning of activity counts

NEMS describes an integrated health-care system rather than a small single-purpose clinic. Its current organizational page lists medical specialties, pharmacy, dental and vision care, behavioral support and ancillary services. It describes 32 clinics and service-delivery sites across Bay counties and Las Vegas. The Nevada clinic opened in 2024. This expansion is relevant to geography, not a reason to assume a particular out-of-Bay percentage.

Different public sources use different patient totals. The older support-page text says more than 70,000 patients annually; the current campaign says more than 93,000; a 2025 presentation citing the 2024 UDS report says approximately 82,000. These need not be false statements: dates, covered entities, service definitions and updates may differ. They should not be averaged into a pseudo-precise denominator or used interchangeably to calculate marginal cost per patient.

The August 2025 community-of-practice presentation, in the NEMS speaker's section, identifies a predominantly Asian patient population, substantial language needs and a large Medicaid share. This is useful context for why culturally and linguistically responsive care may matter. It is not evidence that every patient lacks an alternative provider or that every encounter is caused by philanthropy.

The 2024 original return reports 358,501 medical encounters, 658,573 prescriptions and 899,407 managed-care member-months. Each has a different denominator. Encounters can repeat within a patient; prescriptions can represent refills; member-months count time rather than unique annual people. A person may appear in all three systems. Summing them would create an artificial reach number with no health interpretation.

Other services also have distinct units. Dental encounters, laboratory tests, radiology services, optometry and counseling describe activity delivered, not unique treatment completions. A laboratory test may support the same course of care already counted in a medical visit. Managed-care administration may coordinate services delivered elsewhere. The financial and clinical model must avoid giving a separate full health benefit to every billing or operational unit in that chain.

The payment page explicitly says NEMS is not a free clinic. It describes insurance acceptance, patient cost sharing and sliding-fee arrangements for eligible people. It also describes enrollment help. This is important because an analysis based on “charity provides care that otherwise would not exist” would miss the publicly and privately financed baseline. Donation-supported access may still matter, but its marginal role must be specified.

Hepatitis services are already extensive. The liver-health page describes testing, vaccination, monitoring, fibrosis assessment and treatment support. It distinguishes HBV from HCV and notes that people with HBV may or may not need antiviral treatment depending on their clinical state. The program is not merely an awareness campaign. At the same time, advertised service availability does not establish completed monitoring rates among people who are disengaged.

The page lists fibrosis assessment at multiple Bay clinics, including San Jose as well as San Francisco. That improves geographic specificity compared with an SF-only description. It does not tell us the residence or disease state of the marginal patient. A current referral route can be verified without assuming that every listed appointment is immediately available or that a new gift adds an extra appointment.

NEMS also has a perinatal prevention program, Hep B Moms. Its mechanism is different from adult retention: timely maternal assessment and indicated treatment, infant prophylaxis, vaccination and follow-up. We do not include this branch in the adult numerical estimate. A single referral cannot be assigned an entire lifetime of avoided infection without knowing which element of prophylaxis changes relative to routine care.

The CDA Foundation's ReLink grantee page still lists a twelve-month NEMS project in progress. The description covers diagnosed but untreated people, missing fibrosis assessment, treatment eligibility and loss to follow-up. This is a broader clinical population than the initially inactive cohort in the economic paper. It is evidence of an existing funded effort, not a currently open private donation tranche.

Current operations therefore strengthen the plausibility of a care-access mechanism but also strengthen the counterfactual. NEMS has registries, clinical staff, pharmacies, enrollment assistance and an externally supported re-engagement initiative. A new navigator must improve on that system, not be compared with an empty clinic or no monitoring. The difference could be meaningful for a hard-to-reach subgroup, but the public pages do not quantify it.

We also found evidence of sustained quality-improvement work in latent tuberculosis. The 2025 presentation describes pharmacist monitoring, electronic-record tools and implementation partnerships. Its later slides contain another organization's treatment outcomes; those must not be credited to NEMS. The NEMS section itself is valuable for understanding workflow and barriers, not a license to import unrelated completion figures.

This wider operational picture prevents a common winner-selection error. HBV was selected because a published economic model made it inspectable, not because it has been shown to be NEMS's most valuable marginal activity. Other clinical programs may contribute substantial health. Keeping them unquantified is an honest scope limit, not proof that the HBV calculation represents the entire organization.

Three years of original finances and the supporting-foundation boundary

We inspected original returns for calendar years 2022, 2023 and 2024 for both operating NEMS and NEMS Foundation. These are the latest three available original returns located in this review. The public index separately lists an audit for the period ending December 2025, but the download returned a security refusal through both web and native retrieval. We therefore distinguish “latest returns inspected” from “latest financial period apparently available.”

For operating NEMS, the principal program categories reconcile as follows:

NEMS operator — whole known accounting shares

Calendar years; use operator broader known accounting expense. Do not add Foundation expenses.

NEMS operator — within-program mix

Denominator is PROGRAM expense only; this nested mix must not be added to functional shares.

NEMS Foundation — whole known accounting shares

Separate recipient; calendar years, including identified netted rental costs. Intercompany elimination unresolved.

Shares are calculated as category expense divided by the stated annual denominator. Rounding may make displayed shares differ slightly from 100%.

Sources: original 2022 return, 2023 return and 2024 return, Part III. Other programs include dental, PACE and ancillary services; they are not a residual amount assumed to have no health value.

The supporting functions and known separately netted costs are:

The last column adds known costs reported outside the functional table. It does not add the cost basis of securities sold as if that were new clinical operating expense. It is an accounting expense boundary, not verified cash disbursement, capital expenditure or a complete societal resource total. Program and supporting categories should not be added again to the functional subtotal.

Program expense is approximately 74%, 75.5% and 79.9% of functional expense across these years. That trend is not itself evidence of improved clinical efficiency. The organization changed in scale and mix, especially managed care. Accounting allocations, contracted services, membership and the nature of care can all change. A ratio is descriptive, not a causal measure of what the next dollar produces.

Within other programs, the 2024 return identifies approximately $13.49 million for dental services, $9.29 million for PACE, $5.49 million for laboratory services, $3.06 million for radiology, $2.62 million for health counseling and $2.24 million for optometry, with additional physical therapy expense. Those are service categories, not separate charities. Their outputs may overlap at the patient level and may be financed through different reimbursement arrangements.

Managed-care member-months grew from 702,630 in 2022 to 840,718 in 2023 and 899,407 in 2024. The much larger change in reported managed-care expense is not adequately explained by that count alone. We do not infer a sudden deterioration or improvement in health value from the difference. A proper interpretation would require contract structure, covered services, risk arrangements and changes in accounting scope.

Operating revenue was $242.66 million in 2022, $327.02 million in 2023 and $385.12 million in 2024. In 2024, program-service revenue was $359.79 million, while contributions were $13.05 million. Explicit government grants were $11.75 million. That grant line does not measure all public support because reimbursed clinical and managed-care revenue can also be publicly financed. It would be misleading to characterize NEMS as largely private merely because the grant percentage is small.

The donation Foundation has a different financial scale and function:

Sources: original Foundation 2022, 2023 and 2024 returns. The Foundation's program description is clinic-facility support, and its revenue includes rent from NEMS. Its expense ratio cannot be interpreted like that of a direct-care clinic.

The 2024 Foundation return reports $936,305 of rental income from operating NEMS within program revenue, $69,453 of contributions and $124,935 of investment income. This is not a pure pass-through fundraising vehicle whose every receipt becomes an immediate clinical grant. It owns resources and incurs supporting costs. Conversely, property support may enable important care even if it does not resemble a medical service line.

Foundation Schedule I identifies a $122,100 cash grant to operating NEMS for medical and health services. The total domestic grant expense in Part IX is $153,988. The difference is not resolved here; reporting thresholds and other assistance may be relevant, but we do not assume an explanation. Neither amount is identified as HBV navigation. We therefore do not use the grant percentage as the fraction of an ordinary donation buying the modeled intervention.

The two entities' totals cannot simply be added. Rent paid by one and received by the other, grants, shared services and related transactions would require elimination in a consolidated view. Adding both sets of expenses without checking these flows could count the same resource twice. Using only the Foundation's small expense while crediting all operating NEMS patients would create the opposite error: assigning system-wide benefits to a supporting entity's narrow cost base.

Balance sheets further complicate simple “funding gap” narratives. Operating NEMS reports $470.46 million of assets and $414.70 million of net assets at the end of 2024. The Foundation reports $16.27 million of assets and $11.56 million of net assets. These include property and other noncash or committed resources; they are not current discretionary balances available for any donor-selected purpose.

The Foundation reports no donor-restricted net assets at year end, but unrestricted accounting classification does not mean all property or investment assets are immediately spendable without operational consequences. Operating NEMS reports approximately $1.06 million of donor-restricted net assets. A current liquidity and commitment schedule would be needed to understand financial flexibility. Annual revenue exceeding expense does not prove that a specific access expansion is fully financed, while an appeal for support does not prove an urgent shortfall.

The original returns are valuable because they establish scale, recipient identity and spending categories. They do not establish current marginal clinical costs or an available long-term navigator commitment. The unread 2025 audit remains a concrete evidence gap. A future review should obtain that document and reconcile any changes before presenting these 2024 balances as current.

The clinical mechanism: monitoring is not treatment, and treatment is not cure

Chronic hepatitis B has a dynamic course. Some diagnosed people do not initially meet treatment criteria but still require monitoring because their clinical state can change. A navigation intervention could help them complete appropriate tests and return to care when treatment or surveillance becomes indicated. This is the mechanism modeled, not a claim that every positive HBV test requires immediate medication.

The distinction between clinically inactive infection and an inactive administrative record is essential. The economic source starts people in an inactive clinical health state. A registry record marked inactive may instead represent someone who moved, transferred care, died, cannot be contacted or has advanced disease. Those people do not share the same baseline or expected benefit. A donor-funded registry cleanup needs to classify them before applying a clinical model.

The relevant chain contains several steps: identify an eligible person, establish that needed care is missing, reach the person, complete appropriate monitoring, act on results, sustain indicated treatment or surveillance and change a health outcome. Sending reminders or scheduling visits is insufficient if the person does not attend. Completing a test is insufficient if an indicated treatment is unavailable or not sustained. A model that stops at contacts will tend to overstate clinical benefit.

Current NEMS services already address parts of this chain. The liver-health page describes monitoring, specialist access, fibrosis assessment and help with affordable medication. The ReLink project also addresses diagnosed people missing care. These facts make NEMS a plausible implementer while reducing the appropriateness of a no-care comparison. The new gift should be credited only for the difference beyond these existing systems.

HBV should not be confused with HCV. HCV treatment can produce virologic cure in a finite course for many patients. HBV management often involves long-term suppression and monitoring rather than an analogous short cure. Importing a hepatitis C cure QALY into an HBV navigation calculation would be clinically and economically inappropriate. This report keeps the adult HBV pathway separate.

Nor is a laboratory-monitoring rate the same as medication adherence. A person can complete tests without needing treatment; a person on treatment can miss doses despite attending monitoring; a person can receive appropriate care outside NEMS. A useful outcome registry should distinguish these states. The published economic model makes downstream assumptions that cannot be verified from a two-year count of tests alone.

Surveillance for liver cancer is another distinct mechanism. Its value depends on risk selection, test performance, follow-up, stage shift and effective treatment, not merely an ultrasound being ordered. The current model's lifetime health calibration implicitly aggregates disease pathways. It does not independently reconstruct surveillance benefits. That is one reason to call the adaptation a heuristic rather than a new disease-state simulation.

Perinatal prevention is potentially important but separate. Maternal antiviral treatment, infant immunoprophylaxis and timely vaccination each have their own baseline and eligibility. The previously reviewed trial of maternal tenofovir found few infant infections when both groups received prompt infant prophylaxis. A navigation program should therefore model the care actually changed, not assume every maternal referral prevents infection. No infant benefit is added to the retained adult model.

The broader health system provides other plausible pathways, including blood-pressure management, vaccination, dental treatment, vision care and latent TB prevention. Their clinical efficacy does not automatically identify a donation effect. A publicly reimbursed prescription may be effective while an ordinary Foundation gift has little influence on whether it is dispensed. Conversely, a small practical improvement could help care reach people who otherwise miss it. The missing variable is the gift's effect on completed care, not whether medicine can work.

A source-informed model may still use judgments when local measurements are incomplete. The problem is not that every parameter must be directly observed. The problem is making the judgment correspond to the correct unit and counterfactual. A bounded additional-monitoring prior is more defensible than labeling all existing patients additional. A finite duration with recurring costs is more defensible than buying lifetime follow-up with one salary payment.

This review retains those safeguards and exposes their limitations. It does not claim to resolve the causal effect of NEMS navigation. It identifies a specific route through which additional support could matter and the evidence needed to distinguish that route from already financed care, administrative activity or a different Foundation purpose.

The published economic calibration and its unresolved source problems

The numerical anchor comes from Starinieri and colleagues' 2026 economic model, Cost-Effectiveness of Interventions to Improve Retention in Care for Chronic Hepatitis B in the United States. The V2 review freshly inspected its primary full-text XML through Europe PMC after the PMC page returned a browser-check barrier. This is a published decision model, not a randomized NEMS outcome evaluation.

The source compares current practice with electronic reminders, patient navigation and a combined intervention in a hypothetical cohort of 100,000 diagnosed adults. The cohort begins in clinically inactive HBV and is followed over a lifetime. Its results depend on assumed improvements in monitoring, subsequent treatment and disease progression. Those assumptions are important causal links, not background details that can be ignored when importing the result.

The retained navigation QALY anchor is calculated from Table 3: 1,871,632 minus 1,854,238 discounted QALYs, divided by 100,000 people, giving 0.17394 per initial person. This is not the QALY gain from one completed blood test or one newly treated patient. It is a lifetime difference under the source's sustained intervention scenario. Multiplying it by every annual patient without addressing duration would repeatedly credit the same long-run benefit.

The source's navigation monitoring rate is 65.35%, compared with 37% under current practice, a 28.35-percentage-point difference. The local model scales health by a smaller assumed incremental monitoring difference relative to this reference. That linear relationship is unvalidated. A ten-point gain among lower-risk or already well-managed people may not have the same health effect as ten points among people who have become treatment eligible.

Several source inconsistencies remain. The abstract gives a different navigation ICER from the results section and Table 3. The combined strategy's displayed total QALY difference does not match its difference row. The stated relationship between the baseline and reminder monitoring rate is also inconsistent with the description of a 29% relative increase. The review reports these problems rather than selecting whichever number makes the donor calculation most attractive.

The numerical adaptation uses the navigation total difference consistently, not the combined-strategy difference row. That prevents one immediate arithmetic ambiguity, but does not validate the source's causal monitoring effects or annual disease trajectories. A substantial transfer discount can express uncertainty; it cannot prove that an erroneous source estimate has been corrected. A stronger future analysis would reproduce the disease model or obtain clarifications and local data.

The source attributes intervention monitoring rates to NEMS-related community-health-center evidence. The cited local conference abstract describes care characteristics among 3,457 NEMS patients with a primary-care visit in 2022–2023. The earlier accepted source audit found testing and prescribing descriptions, not the causal pre/post navigation comparison claimed by the economic paper. This review retains that limitation explicitly rather than presenting the local citation as causal validation.

Selection into that descriptive cohort matters. People with no primary-care visit are excluded, yet they may be the people a re-engagement program most needs to reach. A two-year testing count among attenders cannot establish annual monitoring adherence among all registry patients. The distinction is especially important when the intervention is intended to reconnect people lost to follow-up rather than improve tests among people already attending.

The source also assumes sustained high annual adherence to treatment and care. An organization-level donor model must account for how long the additional support is financed and whether the downstream treatment persists. The accepted adaptation does this imperfectly by charging recurring support costs and allocating the already discounted lifetime health over a finite window. It is a transparent heuristic, not an empirical annual trajectory.

The salary inputs likewise require careful interpretation. The source provides a navigator salary anchor, allocated supervision and analyst support. These are not a NEMS quote, and salary alone is not a fully loaded program budget. The local cost multiplier is intended to cover benefits, overhead and local conditions. Whether it is sufficient depends on the actual delivery design, staffing, technology and clinical support.

The published economic ICER includes a health-care-system perspective and downstream costs. The donor adaptation instead prices philanthropic navigation capacity and excludes much of the subsequent clinical resource use. The two prices answer different questions. It would be incorrect to say the published system ICER is a current donor price, or to describe the navigation-only calculation as a complete societal cost-effectiveness estimate.

Clinical plausibility remains despite these limitations. Monitoring can identify changes that matter, and navigation can reduce practical barriers. But the correct evidential conclusion is a mechanism worth investigating, not a validated NEMS donor coefficient. The model should be read as a conditional calculation that makes the consequences of its assumptions inspectable.

An independent review should focus on the source-to-model chain rather than only reproduce the final division. Which population does the anchor represent? What additional monitoring is truly beyond current care? Which treatments follow? How long are they sustained? Which costs are financed by the donor and by others? An exact numerical result can be computationally correct while the answers to these questions remain uncertain.

Monitoring quality, patient denominators and decision-relevant evidence

The strongest monitoring framework would follow a deduplicated clinical cascade rather than count every touchpoint as a separate benefit. For HBV, relevant stages include diagnosed people eligible for follow-up, those successfully contacted, those receiving indicated tests, those found treatment-eligible, those starting appropriate treatment and those maintaining it. Imaging and cancer surveillance should be tracked separately where indicated. A completed test is not the same endpoint as preventing liver disease.

A useful dataset would distinguish existing patients already receiving adequate care from those genuinely re-engaged. It would record the time since last appropriate care and the most plausible alternative-care pathway. If an intervention advances treatment by six months, the counterfactual is six months of delay, not necessarily no treatment for life. If a patient would never have returned without help, that is a different and potentially larger claim requiring stronger evidence.

Repeated visits, prescriptions and member months cannot be summed into unique patients. The original returns report different service units for medical, pharmacy, managed care, laboratory and other programs. One person may appear in several categories and across years. The current campaign's annual patient figure and the 2024 UDS presentation also use different periods or boundaries. None can be converted directly into a donor-created cohort without a documented definition.

The August 2025 community-practice presentation provides a useful glimpse of NEMS quality-improvement infrastructure. Its NEMS segment describes multilingual care, a substantial Medicaid population, pharmacy medication-management follow-up and collaboration on latent tuberculosis work. It also identifies practical barriers such as adverse effects, outside-pharmacy refills, contact arrangements, language and travel. These details make the delivery problem more concrete than an abstract assumption that navigation automatically works.

The presentation contains multiple organizations. Later pages describe East Los Angeles activities, not NEMS. Their referral and completion counts must not be credited to NEMS. Within the NEMS segment, a historical cascade graphic has ambiguous treatment-completion labeling; we do not treat it as a verified completed-course count. This is a small but important source-fidelity point: a promising number in a shared presentation is not necessarily an organization's outcome.

The same presentation describes public and academic collaboration. That helps establish that care improvement is occurring, but it also reinforces the need to include existing quality-improvement work in the baseline. A donor cannot claim the entire effect of a system already supported by clinical staff, public partners and research infrastructure. The incremental contribution may be valuable, but it should be measured against that system.

For a future evaluation, outcomes could include timely completion of indicated monitoring, treatment initiation among eligible people, retention at prespecified intervals and reasons for noncompletion. Clinical appropriateness matters: more medication is not necessarily better when treatment is not indicated. Patient-reported burden and adverse effects should accompany activity measures. Language access and ease of navigation can be valuable process outcomes, but should not be converted into a fixed QALY award without a defensible bridge.

A comparison group or credible phased implementation would substantially improve interpretation. Before-and-after changes can reflect changes in patient mix, insurance, staffing, electronic records or guidelines. If a controlled design is impractical, a clear baseline, prespecified outcomes, denominator reconciliation and documented alternative explanations are still better than a success-only story. Missing follow-up should not automatically be coded as success or as permanent harm.

For donor monitoring, financial and clinical reporting need to connect. A report should identify the actual incremental staff time and resources purchased, the delivery period and the patients reached by that increment. The organization-wide patient total is useful context but not evidence that a restricted tranche achieved its intended effect. Conversely, a narrow grant report should not imply that it captures all organizational value.

The immediate evidence priorities are therefore modest and concrete: reconcile the Foundation's current funding plan; specify any additional HBV service offer; provide a deduplicated local care cascade; explain alternative-care timing; and identify the recurrent financing needed for durable follow-up. These are higher-value next steps than adding decimal places to the current model or assigning a probability distribution without new information.

Review methods, source freshness and unresolved work

This V2 review preserves the accepted model's three scenario outputs exactly. It adds validation, explicit ordinary-gift unknowns, timing and allocation diagnostics, six original financial-return reconciliations and current operational evidence. It does not retune coefficients, assign scenario probabilities or manufacture a whole-portfolio residual. The original executable model and original input file remain adjacent to the wrapper for replication.

The latest three original returns located cover 2022 through 2024 for each entity. A later audit is indexed but could not be downloaded; it is recorded as an unresolved source, not as reviewed evidence. Financial figures are accounting expense and revenue, not a verified cash-flow budget. Where costs are separately netted against revenue, identified rental and gaming costs are added transparently. Investment transaction proceeds and cost basis are not treated as ordinary clinical operating expense.

Current official pages and the campaign were reviewed during this pass. The economic paper's full XML was freshly inspected. The local AASLD abstract critique is retained from the accepted source audit rather than represented as a fresh reread of that large conference PDF. The shared tuberculosis presentation was read in memory and its organizational sections distinguished. No outreach was conducted.

The remaining work is substantive rather than cosmetic: obtain the inaccessible audit if available through an authorized route, resolve the actual marginal Foundation allocation, establish local causal care improvements and replace the timing proxy with a stronger clinical bridge. This report can be published transparently as a conditional-program review, but should not appear in an ordinary-gift Bay ranking with an implied complete expected return.

Qualitative value and limits of the health-only lens

NEMS's language capacity and community familiarity may reduce barriers that are poorly represented by standard utilization statistics. A patient who can communicate symptoms, understand follow-up instructions and trust the care team may experience better care even when no new procedure is counted. These features plausibly support clinical effectiveness. We do not assign them a separate utility increment on top of modeled treatment benefits, because doing so could count the same mechanism twice.

The breadth of the organization is both a strength and a modeling challenge. Primary care, pharmacy, dental care, behavioral health, managed care and elder services may interact. A donor might support infrastructure that benefits several pathways. A narrow HBV model cannot capture that portfolio, and a large patient denominator cannot substitute for it. The appropriate response is to state the partial scope, not to assume all other activities have no health effect.

There may also be benefits outside QALYs: financial protection, reduced administrative burden, caregiver time, dignity, community stability and improved access to public entitlements. These are not included in the numerical result. Their omission limits comparisons with organizations whose principal effects are social or economic. It does not justify adding an arbitrary residual health value to a clinical model.

The absence of a complete ordinary-gift estimate is not a negative judgment on NEMS's clinical quality. It is a statement about what this evidence can support for a particular donor question. A provider can deliver important services while the marginal effect of a small unrestricted gift remains difficult to identify. Conversely, a small organization with a clear marginal offer may be easier to model without necessarily having better care.

Our qualitative stance is therefore interest with a hold on recommendation. The infrastructure and disease mechanism warrant serious investigation. The recipient boundary, public baseline, causal calibration and sustained funding requirements prevent the present number from being treated as a donor-ready return. A future specific offer could change that assessment without requiring a change in the organization's mission or a favorable reinterpretation of the current evidence.

Preserved model, finite funding and validation

The V2 model preserves the original implementation and input file alongside a wrapper. All three original scenario outputs are compared exactly in tests. The central result remains $1,201,090.5888876051 per 10 QALYs. No salary, panel size, monitoring response, transfer factor or duration was changed to make the result more attractive.

The central scenario uses a $65,000 annual navigator salary, $22,900 of allocated annual supervision, $3,593.44 of analyst support in the first year and $2,524.88 in later years. These are multiplied by a local fully loaded cost factor of 1.5 and divided by an initial panel of 2,500 people. Costs recur over twenty funded years and are discounted at 3%.

The initial panel is closed. People who die, move or otherwise leave are not replaced with new entrants whose benefits are added for free. The model also does not reduce the fixed staffing bill as the original cohort attrits. This convention is cautious about cost but simplified operationally. A real program might redeploy capacity; modeling that would require an explicit replacement-cohort structure rather than silently reusing the original cohort denominator.

The health anchor is 0.17394 already discounted lifetime QALYs per initial person. The central local monitoring increment is 0.10, divided by the source reference increment of 0.2835. Causal transfer is 0.5 and funding additionality is 0.5. These are separate judgments about local downstream health applicability and whether the gift changes provision rather than replacing other funds.

The timing fraction uses year-weighted discounted terms over a thirty-five-year reference window. Only the first twenty years are credited in the central scenario. Normalizing the credited weights by all reference weights avoids discounting the already discounted lifetime anchor a second time. It does not establish that the selected time profile matches the source model.

The central calculation is therefore:

Net QALYs per initial person = 0.17394 × (0.10 / 0.2835) × 0.5 × 0.5 × timing fraction − additional harms.

Present-value donor cost per initial person is divided by that net health and multiplied by ten. The approximate central cost is $808 per initial registry person, while the health increment is about 0.00673 QALY. These are calculated quantities under the scenario, not measured NEMS costs and outcomes.

The finite funding rule is important. A one-year grant does not buy the twenty-year central result. Shorter funded periods reduce both credited health and recurring costs according to the specified timing scheme. Longer support can improve the ratio if later health gains are important, but it requires a correspondingly longer financial commitment. No current twenty- or thirty-five-year NEMS offer was verified.

The timing restriction is not a claim that treatment stops helping immediately after funding ends. It is a conservative accounting convention that credits no health beyond the funded window in this adaptation. Earlier care could create a later survival benefit, while missed continuity could also weaken earlier gains. A disease-state model would handle these possibilities more faithfully. The wrapper does not pretend to add that clinical sophistication.

The original positive scenarios combine different assumptions. The optimistic case has a larger panel, lower cost multiplier, full source-like monitoring improvement, high funding additionality and a long support horizon. The pessimistic positive case combines a smaller panel, higher cost, weaker monitoring gain, lower transfer and shorter funded duration. They are joint judgments, not confidence bounds or independently sampled parameters.

Null cases set monitoring improvement, causal transfer or funding additionality to zero. A harm case subtracts 0.01 QALY per initial person, exceeding the central modeled gain and producing negative net health. The result does not display a finite positive-health price when net health is zero or negative. This avoids turning a harmful scenario into a superficially attractive ratio through sign manipulation.

V2 adds finite-output checks to the preserved engine. The original input validator rejects many invalid values but can still produce nonfinite derived quantities under extreme numerical inputs. The wrapper rejects such outputs without changing ordinary results. Tests cover malformed scenario containers, NaN, infinity, null numerical fields, tiny panel size, extreme costs, invalid fractional years and overflow-prone timing parameters.

The complete saved result file is compared exactly. This is stronger than checking only a rounded central headline: scenario output, annual cost schedules and timing fractions must all agree. Separate financial tests reconcile program components and the known expense addbacks for all six original returns. These tests demonstrate computational consistency, not the truth of the clinical priors.

The wrapper also includes a scope-transfer grid. It applies allocation fractions of 1, 0.5, 0.1 and 0 to the hypothetical navigation component, and Bay shares of 1, 0.9 and 0.5. The full $100,000 gift remains in the numerator. These values are illustrative algebra, not an inferred Foundation allocation or a preferred Bay share. The grid shows why a conditional program price cannot be relabeled an ordinary-gift price without additional assumptions.

Ordinary Foundation expected QALYs, ordinary Foundation Bay share, current marginal offer and weighted expectation remain null. Null is intentional. It should not be replaced with the conditional central result by an API, index or renderer. The source and model files are retained so an independent reviewer can reproduce both the accepted baseline and the new diagnostics without editing the publication checkout.

Results, timing and what the threshold does not mean

The preserved central ratio is $1,201,090.5888876051 per 10 modeled QALYs. This is a conditional program estimate. It does not have an observed Bay multiplier, a probability distribution or a verified donor funding offer. Its precision is computational rather than epistemic: reporting cents does not mean the evidence supports distinctions at that scale. The rounded $1.20 million figure is appropriate in prose, with the exact value retained for replication.

The model crosses neither the $100,000 nor the $1 million threshold in its central scenario. More importantly, those thresholds cannot be used to rank an unrestricted Foundation gift directly against a whole-organization estimate elsewhere. The intervention here is a specifically assumed stream of recurring navigation resources for a fixed cohort. A broad gift can reach property support, other clinical services, prevention or infrastructure. Those uses have neither been assigned the navigation effect nor declared valueless.

The funded-duration diagnostics preserve other central inputs. One funded year gives $11,564,063.58 per 10 QALYs; five gives $3,895,956.62; ten gives $2,177,186.79; thirty-five gives $760,439.80. These are not independent empirical estimates of care programs with those lengths. They expose the consequences of the model's assumed distribution of benefit over time and its recurring cost schedule. A thirty-five-year commitment cannot be described as an available opportunity simply because its modeled ratio looks better.

Holding the central twenty-year funding window fixed, uniform temporal benefit weights give $760,620.09 per 10 QALYs, while a more delayed shape gives $1,990,867.33. The difference is a warning about the unresolved bridge from a published lifetime model to a shorter donor commitment. We do not average the shapes or select the one that crosses a threshold. Nor do these tests imply that stopping navigation instantly removes every clinical gain. They are approximations to a missing explicit disease-state simulation, not a validated withdrawal model.

The signed diagnostics set funding additionality, monitoring increment or causal transfer to zero separately. Each removes modeled benefit without turning a positive cost into a favorable price. The harm scenario permits negative net QALYs. A zero-year scenario has no recurring program purchase. These cases matter because navigation could fail to improve useful care or consume time and resources without additional clinical benefit. Negative outcomes should remain visible rather than being omitted from an average of successful scenarios.

The allocation/geography grid retains a $100,000 gift in every row and scales only quantified HBV benefits. It illustrates the algebra of devoting all, half, one tenth or none of a gift to the hypothesized pathway, and of Bay shares of one, 0.9 or 0.5. Neither dimension is an estimate of Foundation behavior. No local data justify choosing a preferred cell. An allocation of one is especially demanding: the official campaign describes multiple purposes and does not promise exclusive HBV navigation.

This grid is not a completed whole-gift model. It excludes the health of other funded activities, and it cannot establish what those activities are. A small quantified component may coexist with meaningful unquantified benefits; a favorable component may also be unavailable to a donor. The correct conclusion is a conditional clinical hypothesis with unresolved ordinary-gift impact, not a comprehensive estimate with a silently zero residual.

A better next model would specify an actual marginal care cohort, current untreated need, treatment eligibility, alternative-care timing and a funded delivery period. It would then propagate those patient states through a clinical natural-history model or a defensible finite outcome bridge. Until then, the existing result is useful for identifying the strength of assumptions required, not for confidently choosing NEMS over another organization.

Public financing, resource costs and overlapping attribution

NEMS is a health-care provider that bills for care, not a clinic whose entire clinical output is purchased by charitable donations. Its payment page explicitly describes insurance and sliding-fee arrangements. The operating return's program-service revenue is much larger than its contributions, and its government-grant line captures only one form of public support. Medi-Cal and Medicare-related reimbursement may enter program revenue. A calculation that treats only the grant line as the publicly funded baseline would substantially misdescribe the financing system.

Public payment does not make philanthropy irrelevant. A reimbursable appointment can remain inaccessible if a patient cannot navigate scheduling, language, eligibility or follow-up. Conversely, the availability of reimbursement can mean that a donor replaces an activity the provider would otherwise finance. The question is whether a specific additional dollar changes completed useful care relative to this baseline. Neither the existence of a public payer nor the existence of a barrier answers it automatically.

The retained funding-additionality factor of 0.5 is a judgment, not a number derived from the public-revenue share. It is intended to address whether private funding creates additional delivery rather than substitutes for other resources. The separate causal-transfer factor addresses whether the modeled monitoring improvement is real and transferable. These factors describe different questions, but applying both does not establish that they have been independently measured. They should be revisited against an actual funding plan rather than mechanically inherited into every new program.

Downstream medication, laboratory work, imaging and specialist services have resource costs even when another payer bears them. The preserved calculation prices navigation staff and associated support, not the entire treatment pathway. It therefore cannot be called a full societal cost-effectiveness analysis. The published economic study uses a health-care perspective, whereas the adapted model uses a narrower donor-program perspective. Those boundaries are not interchangeable.

A future resource analysis should first identify which downstream services are actually additional. Counting all routine care costs as newly caused by navigation would overstate resources if patients would have received them anyway. Counting none would understate them when re-engagement genuinely creates additional treatment. The same counterfactual must govern both benefits and costs. Existing reimbursed care belongs in the comparison arm; additional care belongs in the intervention's incremental resource ledger.

The Foundation's relationship with the operator introduces another accounting issue. Rent paid by the operator is revenue to the Foundation; a grant paid by the Foundation is support received by the operator. Adding both entities' expenditures without elimination can count an internal transfer twice. We provide the separate original returns because they describe different legal recipients, not because their totals can be combined into an impact denominator without consolidation.

The Foundation's unrestricted net assets are not a verified stock of liquid money available for new navigation. Property and financing obligations matter. Likewise, the operator's balance sheet does not establish a cash-created capacity offer. Strong reserves can reduce the likelihood that a small gift is decisive, but cannot prove that every useful program is fully financed. A shortage should not be inferred from a deficit, and a surplus should not be treated as proof that no marginal activity exists.

Finally, navigation can overlap with existing NEMS staff, insurers, community partners and the separately funded re-engagement project. An additionality assessment should describe which work these actors already perform. It should not award the same treatment gain to each funder independently. Where a gift supports a shared causal chain, the donor-facing claim should explain its marginal role rather than claiming sole ownership of the entire health outcome.

Where an ordinary gift goes and the next funding questions

The official support page directs checks to North East Medical Services Foundation. The linked Give in May campaign identifies that Foundation with EIN 94-3171797, matching the original returns. This verifies a public donation route and legal recipient; it does not verify an HBV restriction, a current navigation vacancy or a price for an additional patient. Donors should not be led from the conditional model to that route with an implication that the modeled purchase is automatically available.

The campaign provides useful specificity about actual uses of philanthropy. It describes medical, dental and behavioral services, language access, outreach, technology and other initiatives. Staff quotations dated 2026 say that 2025 donations supported flu vaccination and mold mitigation at an older Chinatown clinic site. These are plausible care-support activities, not an observed count of additional healthy patients. The page does not state how much went to each activity, how much other funding was available, or how many additional people benefited.

The displayed campaign fundraising total exceeds its displayed goal. That observation should not be translated into either “fully funded organization” or “urgent remaining gap.” A campaign target is not a reconciled marginal budget. It may cover a limited period or purpose, and a website may remain visible after an appeal ends. We retain the information as evidence of a real giving channel and concrete historical purposes, not as a funding-capacity estimate.

The Foundation's 2024 Schedule I reports a $122,100 cash grant to NEMS for medical and health purposes. Its functional statement reports a larger domestic-grants line, leaving a difference not reconciled by the inspected Schedule I detail. We do not allocate that difference to HBV or invent an explanation. The grant description is broad, and its amount is neither a current navigation budget nor a marginal offer. The property-support program in the Foundation return is another reason not to assume a one-to-one flow into incremental clinical staff.

The CDA Foundation's re-engagement page separately lists a twelve-month NEMS HBV project in progress. Its described targets include untreated diagnosed people, missing fibrosis assessment, indicated treatment and lost follow-up. This is encouraging evidence that the relevant care gap is recognized and an implementer exists. It is also financing in the no-gift world. The page does not establish a particular unfilled tranche for a new donor, and a round-wide grant total must not be attributed to NEMS.

An actionable request would begin with the current marginal purpose: if an additional unrestricted gift arrives now, what activity changes, over what period, and under whose budget authority? If HBV navigation is proposed, the next question is whether a restricted gift can legally and operationally purchase that activity through the Foundation, including its share of administration, space and supervision. Neither question requires assuming that the organization should create a special program solely to fit this model.

The clinical capacity question is equally important. A staffing plan should identify the number of additional eligible people it can support, not just contacts or names in a database. It should distinguish finding a patient, obtaining monitoring, determining treatment eligibility, initiating indicated treatment and sustaining it. A plan that produces more screening in people already effectively managed may have less incremental value than one that closes an untreated high-risk care gap. That difference is not visible in a generic annual patient total.

The donor should also ask what would happen without the proposed gift. Would an insurer, existing grant, operating reserve, another donor or a planned staff hire cover the work? Would patients obtain equivalent care elsewhere, and after what delay? The useful answer is a bounded counterfactual with uncertainty, not an assurance that every donation helps. We have not contacted the organization, and none of these questions is presented as answered.

A final question concerns durability. The model's central cost buys recurring support for twenty years. A one-time appeal cannot be assumed to guarantee that stream. A shorter grant could fund an initial cohort with later operating support, but that continuation needs to be described and costed rather than silently assumed. Alternatively, a donor could fund a shorter intervention with a different health model. The current numerical result should not be repackaged as the return on a single annual gift.