GiveBetter x SF

Changent / Nurse-Family Partnership

National implementation support for Nurse-Family Partnership and Child First

Research time: ~31 min on GPT-6.1 Sol + ~10 min on GPT-5.6 Sol
  • v1: ~10 min on GPT-5.6 Sol (GPT-6 Astra Lite time not recorded)
  • Calibration: ~31 min on GPT-6.1 Sol

Updated: 2026-10-04

Donate

Summary

What do they do? Changent supports two home-visiting programs for families with young children: Nurse-Family Partnership and Child First. Local Nurse-Family Partnership partners provide voluntary nurse visits from pregnancy through a child’s second birthday. Child First teams combine home-based mental-health care with help connecting families to services. More

Why this approach interests us

  • Early childhood and caregiver wellbeing can influence many later outcomes, making effective support potentially valuable.
  • Both programs have randomized evaluations, allowing stronger causal scrutiny than service counts alone.
  • Changent provides training, monitoring and shared infrastructure across an established national delivery network.

Our main reservations

  • Modern trials are mixed or null on important primary outcomes; older findings do not establish contemporary lifetime gains.
  • An unrestricted national donation is not a priced purchase of additional local visits or completed courses.
  • Household resource gains, clinical utility bridges, funding additionality and long-term persistence remain uncertain.

What do you get for your dollar? Our conditional finite-horizon estimate is about $1.2M per DALY nationwide: one modeled equivalent year. Donations support national training, infrastructure and partner services; they do not guarantee a particular family’s treatment. We model small additional course-equivalents, finite health gains and net household support after participation costs. The central scenario is $1.3M; both depend on judgmental additionality and benefit assumptions. The earlier $1.0M headline used a different lifetime-health horizon, while its follow-up-bound estimate was $1.4M. These are not verified donation offers. More

1. What do they do?

What they do

Changent is the national support organization for Nurse-Family Partnership and Child First, not the employer delivering every local visit. Nurse-Family Partnership connects eligible mothers with nurses during pregnancy and continues through the second birthday. Visits address health, development, parenting and family self-sufficiency. Child First pairs a mental-health clinician with a care coordinator to support young children and caregivers experiencing adversity.

The official program descriptions describe free, voluntary services. National-office support includes education, implementation systems, monitoring and network development. A donation to Changent therefore has a different cost boundary from funding a complete course at a particular affiliate.

The 2025 year in review reports 57,087 Nurse-Family Partnership families and 2,738 Child First families. These counts establish substantial reported activity, not additional families attributable to donations. They do not measure unique completed courses, family-years or sustained causal outcomes. Three new and 34 expanded programs are reported, but expansion announcements do not identify which costs an additional unrestricted donation would pay.

2. Monitoring and information sharing

Monitoring and information sharing

Changent shares annual program information, financial reports and descriptions of its delivery systems. This permits basic checks of the operating entity, reported scale and program mix. The legal entity is Nurse-Family Partnership, doing business as Changent, EIN 20-0234163; the support page directs checks to Changent.

A useful donor outcome record would follow families from eligibility and offer through attendance, completion and subsequent health and resources. Visits and contacts are intermediate outputs, not unique beneficiaries or completed courses. We do not convert every contact into another person, or assume all reported families receive the full course.

The annual support illustration uses 2025 reported family totals alongside fiscal expenses for the year ending September 2025. Exact reporting-window alignment is unverified. This is a same-labeled annual approximation, not audited fiscal-year family-years. Older visit and contact totals remain historical diagnostics only and do not determine the current benefit shares.

Reported child or caregiver symptom improvement without a comparison group cannot establish additional impact. Families may improve with time, alternative treatment or changes in circumstances. Randomized evidence is consequently important, but trials conducted in other places or delivery eras still require careful transfer.

Clinical evidence and native outcomes

Nurse-Family Partnership’s older Memphis randomized trial reported fewer preventable-cause deaths through age 20. The event count was sparse and the mortality analysis was not a prespecified primary result. It is a reason to investigate possible benefits, not a measured lifetime survival gain for every current participant.

The modern South Carolina randomized study found no significant improvement in its primary adverse-childhood-outcome composite at two years. The estimated composite difference was approximately +0.4 percentage points, with a confidence interval spanning meaningful improvement and deterioration. This does not prove zero effect through age 20, but it argues against treating older benefits as reliably replicated.

That study also found a modest secondary reduction in emergency-department use. Table 3 reports an adjusted reduction of 2.7 percentage points; the abstract uses a related 2.9-point estimate. We acknowledge this signal rather than describe all utilization as null. An avoided visit is not automatically a large QALY gain or a household cash saving: clinical severity, insurance and out-of-pocket incidence would need to be established before pricing those channels.

Child First’s original trial reported improvements in some child and caregiver outcomes. More recent multisite randomized evidence, conducted in the pandemic context, found largely null primary family outcomes and suggestive secondary findings. The 2023 working paper and later peer-reviewed publication concern the same underlying study, with differing reported sample counts; they are not two independent replications.

Our current clinical illustration retains uncertain signed harm, null, cautious, central and favorable worlds. Small finite utility bridges for Child First are analyst judgments, not trial-measured QALYs. The Nurse-Family Partnership mortality bridge uses a finite nine-healthy-year allowance in the follow-up-bound calculation. We do not add an actuarial lifetime automatically, or interpret absence of long follow-up as proof that later effects vanish.

The older lifetime calculation remains inspectable. It credits the observed survival window plus a judgmental share of the later actuarial tail. Choosing the finite calculation for the current comparison is a conservative horizon judgment, not new empirical evidence settling persistence. The price change must be read with that change in display boundary.

3. Qualitative assessment

Qualitative assessment

The programs address important early-life problems through sustained relationships rather than a one-off interaction. Established delivery networks and randomized evaluation are strengths. We are interested in whether those systems can reliably improve outcomes for families whose alternatives are limited.

The main reservation is the gap between evidence that a program can help and evidence that a particular national donation adds effective delivery. Recent trials also make the benefit estimate less straightforward than the older positive findings might suggest. Financial scale or an attractive favorable scenario cannot close that causal gap.

Changent’s reporting permits verification of current operations and broad expense boundaries. It does not establish a fully priced marginal expansion opportunity, local capacity or the complete net household-resource effect. The current conditional comparison should guide further diligence, not substitute for it.

4. What do you get for your dollar?

What do you get for your dollar?

A donation could support nurse education, affiliate implementation, monitoring and shared infrastructure across both programs. The immediate unit we model is an additional support-funded course-equivalent, not a guaranteed death averted. Whether national support actually expands effective local delivery is a major uncertainty.

The full donation stays in the donor-cost numerator once. The model scales annual organization activity by additional support cash divided by total annual expense. Funding additionality is an explicit prior: a donation may replace other funding, support reserves or maintain existing systems rather than expand family services.

For Nurse-Family Partnership, reported annual families are divided by an assumed 2.5-year course length. This steady-state conversion is not observed completions. Realization assumptions then reduce the funded course-equivalents. Child First uses reported families with separate realization and uniqueness assumptions; it does not use visits as people.

FY2025 total expense is $31,183,784. Direct program expense is $17,009,763 for Nurse-Family Partnership and $4,920,100 for Child First. Shared expense is allocated proportionally across those two program lines for the conditional support illustration. That allocation is not a marginal procurement quote or proof that all support spending creates the same expansion response.

The current central illustration yields 0.07 clinical QALYs and 0.01 income-equivalent years per illustrative $100,000, totaling 0.08 combined years. Dividing the donation by that benefit gives $1.3M per DALY. The subjective weighted illustration yields 0.07 clinical QALYs plus 0.010988 resource-equivalent years, totaling 0.081615 and giving $1.2M per DALY.

These small modeled amounts reflect national-support additionality and uncertain program effects, not the value of every family already served. The weights are decision priors, not probabilities estimated from current donation outcomes. Null and adverse scenarios remain in the calculation; the 5% favorable stress world contributes about 89.55% of the weighted total. That concentration makes the weighted result especially sensitive to uncertain positive assumptions.

External affiliate delivery costs are shown separately. The weighted gross-resource illustration is approximately $1.8M per DALY, versus $2.3M in the central world. Historical delivery-cost ranges are sensitivities, not contemporary priced offers. They include delivery resources before unsuccessful participation, subtract modeled internal support expense once and keep the donation once. Neither donor cash nor this partial gross calculation is a complete societal cost accounting.

Health and net household resources

We assess health and net income or consumption separately before combining them. A healthy-year-equivalent is a normative welfare comparison, not an extra measured clinical QALY. The mapping follows the repository’s Coefficient Giving-inspired comparison, with a $50,000 reference-income parameter and $100,000 healthy-year value, logarithmic resources and finite discounted duration.

Baseline household annual resources of $20,000 are an explicit prior, not an observed average for the modeled families. Household money is counted once per family, not multiplied by children or visits. The method values net resources received, not gross donor spending or insurer claims.

The 2025 randomized benefits-access study found approximately 0.6 additional months of SNAP participation over two postpartum years. We illustrate $90 per year using an assumed $300 monthly value, spread over two years. The assumed dollar value and transfer to current program families are not trial-measured cash effects. We do not separately add WIC, referral counts or the same welfare already captured by health.

In the central Nurse-Family Partnership resource world, a 50% positive overlap adjustment reduces the $90 support allowance to $45; $30 of participation burdens then leaves $15 net per year. Maternal earnings and out-of-pocket medical gains are not established central dollar effects. A zero entry excludes that channel from the restricted illustration; it does not prove there is no benefit.

For Child First, the central illustrative $300 annual saved take-home-pay allowance is a speculative, heavily attenuated bridge from suggestive employment-retention evidence. It is not a measured earnings effect. A 50% positive overlap reduction followed by $60 of care, fees and time burdens leaves $90 net for one year. No later-life child earnings or permanent wage premium is assumed.

Only positive resource rows receive the overlap adjustment before the signed sum. Negative earnings, lost transfers, fees, travel and care costs remain full-sized, including when another positive row makes the overall household balance positive. The signed logarithmic transformation is applied once; the helper receives no second overlap reduction.

Nurse-Family Partnership resources last two years with a half-year delay; Child First resources last one year without additional delay. Both use 3% discounting. Service realization is applied once, without inflating trial intention-to-treat effects into larger per-completer claims. Adverse and no-positive-credit cases preserve participation burdens.

Transfers may have payer costs and displace alternatives. The illustration does not add taxpayer savings, GDP, hospital charges, future child wages or baseline survivor earnings. Wider fiscal and displacement effects remain unresolved. Clinical survival and household resources are not monetized twice.

5. Funding and previous grants

Funding and previous grants

The ordinary donation route supports national work across both programs. It cannot be assumed to fund a specific Bay Area affiliate. Prior Nurse-Family Partnership local shares remain unverified diagnostics, and Child First’s local health-and-income allocation is unidentified. We therefore leave total Bay and San Francisco welfare unknown rather than fabricate a local price from national routing.

FY2025 revenue was below expenses by $2,489,261. This accounting deficit is signed and retained. It does not by itself show an unfunded high-impact expansion. Changent also reports substantial net assets; liquidity, restrictions, commitments and funding plans determine what those assets can support.

The annual report’s roughly $740,000 net-asset decline is different from revenue minus expenses. Those quantities should not be interchanged. A useful giving plan would specify additional national support, partner delivery, other funding, full costs, families reached and outcome follow-up.

Spending breakdown and financial appendix

The latest three reviewed original Form 990 returns cover fiscal years ending September 30, not calendar years inferred from IRS form labels.

Fiscal yearRevenueExpenseRevenue minus expenseNet assets
2025$28,694,523$31,183,784−$2,489,261$110,471,423
2024$42,539,779$30,712,013$11,827,766$111,212,082
2023$40,565,834$40,030,887$534,947$92,666,428

Sources: original FY2025, FY2024 and FY2023 returns, Part I. Three-year average annual expense is approximately $33,975,561. FY2025 direct program amounts sum to $21,929,863; the remaining $9,253,921 of total expense is outside those direct program lines. FY2025 functional expense is $21,929,863 program services, $6,464,521 management and $2,789,400 fundraising, reconciling to the full expense. These are legal-entity accounts, not affiliate delivery budgets or marginal donation allocations.

Revision record and model appendix

The original Nurse-Family Partnership-only, follow-up portfolio and published lifetime portfolio outputs, their scenario worlds, mortality sensitivities and source hashes are preserved. The old follow-up donor estimate was $1.4M per DALY. The different lifetime prior gave $1.0M. Current finite health plus signed resources gives $1.2M in the weighted illustration and $1.3M centrally.

The change updates expense and reported service denominators, includes finite signed household-resource scenarios and selects the follow-up clinical boundary for the main comparison. It does not identify the ordinary donation’s expected value or prove the lifetime tail is absent. Current-input lifetime health remains a separately labeled sensitivity, not the new combined headline.

The current adapter validates a finite signed accounting surplus before calling the legacy calculator. It temporarily supplies the legacy helper’s compatible historical diagnostic value, then restores the actual current deficit in all returned current inputs. The original engine and historical outputs are unchanged. Invalid accounting fields are rejected; negative deficits are not changed to zero.

The independent reviewer read the original fiscal returns and benefits-access evidence. Annual-report PDF access returned a challenge in that session; its rounded net-asset figure is reconciled with the original return rather than claimed as a newly inspected PDF page. Modern and Memphis manuscripts were checked through indexed original text and abstracts where direct access was challenged; prespecification and transferability remain uncertain.

Actual source research, modeling, independent audit and report-writing intervals are recorded separately. Tests, builds, publication, unrelated diagnosis and overnight inactivity are not research minutes.

6. Sources

  1. 2024 Year in Review. Changent. Published: 2024 reporting year; retrieved: 2026-09-10. Inherited reference; not all sources freshly reread.
  2. FY2024 public-inspection Form 990. Changent. Published: Tax year ending 2024-09-30; retrieved: 2026-09-10. Inherited reference; not all sources freshly reread.
  3. program description. Original source. Published: See source; retrieved: 2026-10-04. Source facts and transferred priors are distinct.
  4. program description. Original source. Published: See source; retrieved: 2026-10-04. Source facts and transferred priors are distinct.
  5. checks/financial reports. Original source. Published: See source; retrieved: 2026-10-04. National support organization, not automatic unrestricted local affiliate delivery grant.
  6. 2025 program service totals/expanded eligibility. Original source. Published: See source; retrieved: 2026-10-04. Annual reported families are not measured family-years or completed courses; outcomes are uncontrolled.
  7. Effect of Prenatal and Infancy Nurse Home Visitation on Mortality of Mothers and Children. JAMA Pediatrics. Published: 2014; retrieved: 2026-09-10. Inherited reference; not all sources freshly reread.
  8. Swanson etal JAMA Pediatrics2025 treatment effects, DOI10.1001/jamapediatrics.2025.1600. Original source. Published: See source; retrieved: 2026-10-04. Follow-up2years cannot settle rare mortality throughage20 or laterlife.
  9. A randomized controlled trial of Child FIRST. Child Development. Published: 2011; retrieved: 2026-09-10. Inherited reference; not all sources freshly reread.
  10. Abstract, Xia etal DOI10.1037/fam0001393; online2025-11-10. Original source. Published: See source; retrieved: 2026-10-04. Later peer-reviewed sample differs from2023working paper. Suggestive job-loss/substance findings, null psychological/childbehavior/welfare and other economic indicators; not independent replication.
  11. Child First. ACF Prevention Services Clearinghouse. Published: Current evidence record; retrieved: 2026-09-10. Inherited reference; not all sources freshly reread.
  12. Child First. Blueprints for Healthy Youth Development. Published: Current profile; retrieved: 2026-09-10. Inherited reference; not all sources freshly reread.
  13. Costs of Nurse-Family Partnership Scale-Up. Prevention Science. Published: 2015; retrieved: 2026-09-10. Inherited reference; not all sources freshly reread.
  14. NFP Multidomain Impact in California. Nurse-Family Partnership. Published: 2019; retrieved: 2026-09-10. Inherited reference; not all sources freshly reread.
  15. California 2024 State Profile. Nurse-Family Partnership. Published: Late-2023 data in 2024 profile; retrieved: 2026-09-10. Inherited reference; not all sources freshly reread.
  16. Changent filing record, EIN 20-0234163. Internal Revenue Service, mirrored by ProPublica Nonprofit Explorer. Published: FY2024 filing received 2025-07-18; retrieved: 2026-09-10. Inherited reference; not all sources freshly reread.
  17. 2022 period life table. U.S. Social Security Administration. Published: 2025 supplement; 2022 mortality; retrieved: 2026-09-10. Inherited reference; not all sources freshly reread.
  18. Original Form 990: 2025-09-30. IRS filing via ProPublica. Published: 2026-08-04; retrieved: 2026-10-04. Original fiscal return; not a marginal funding offer.
  19. Original Form 990: 2024-09-30. IRS filing via ProPublica. Published: 2024-09-30; retrieved: 2026-10-04. Original fiscal return; not a marginal funding offer.
  20. Original Form 990: 2023-09-30. IRS filing via ProPublica. Published: 2023-09-30; retrieved: 2026-10-04. Original fiscal return; not a marginal funding offer.
  21. p7 Financials. Original source. Published: See source; retrieved: 2026-10-04. Recognizes philanthropy when committed, not received/spent; net-asset change differs from revenue minus expense due other changes.
  22. Abstract; Data and methods; DOI10.1093/haschl/qxaf038, published2025-04-02. Original source. Published: See source; retrieved: 2026-10-04. Source facts and transferred priors are distinct.
  23. Table4 p16; Table5 p17-18; discussionp22-24. Original source. Published: See source; retrieved: 2026-10-04. Pandemic context; jobloss p<.10; no general earnings-dollar effect or proof of duration beyond12months. Most clinical and other economic outcomes null.

Annual expenses: years and sources

Average annual expenses (three consecutive fiscal years): $33,975,561. Organization size is separate from the modeled cost-effectiveness of a donation.

Changent / Nurse-Family Partnership

Latest three original Form 990 returns for fiscal years ending September 30; national support organization, not affiliate delivery budgets.

Average annual expenses are separate from conditional cost-effectiveness and marginal funding capacity.