GiveBetter x USA

Kids and Car Safety

Vehicle-injury prevention, incident research, education, and safer-design advocacy

Research time: 9 min on GPT-6 Astra Light
  • Research — reviewed programs, finances and impact evidence.

Updated: 2026-09-13

Donate

Summary

What do they do? Kids and Car Safety documents vehicle-related injuries and promotes education, safer vehicle design, and policy changes. Its work includes hot-car deaths, backovers and frontovers, entrapment, submersion, and support for affected families. A hot-car-only working estimate is about $920,000 per 10 USA QALYs, with very wide uncertainty about influence, adoption timing, and funding additionality.

Why we’re interested in this organization:

  • Detecting an unattended child and triggering timely rescue is a concrete, credible route to preventing death.

  • Technical studies show why simple door-based reminders leave gaps that better detection and alerts could address.

  • The recipient operates at a small reported budget and has a sustained, focused presence in a broader safety coalition.

Our main reservations:

  • No field study establishes the organization’s marginal effect on fatalities, rule timing, or device adoption.

  • Existing industry innovation, government work, slow fleet turnover, and inaccessible older vehicles limit the additional benefit.

  • Financial reserves include a material cryptocurrency holding, and the latest located return is unaudited cash-basis reporting rather than a current operating plan.

What do you get for your dollar? $920K per better life: ten additional quality-adjusted life years in USA. Hot-car mortality only; policy effect modeled.

Risk-weighted vehicle-years with stronger child detection; successful detection; functioning alert; timely rescue; deaths prevented. Theoretical scenario coverage, education materials, and historical laws are not treated as measured fatality reductions.

1. What do they do?

Kids and Car Safety collects incident information, distributes safety education, supports affected families, and advocates for vehicle design and policy changes. Its program history includes national Look Before You Lock materials and a Kansas/Missouri Childproof Your Ride initiative, alongside work on entrapment, vehicle submersion, frontovers, and carbon monoxide. This review assesses ordinary support to Kids and Cars Inc, EIN 20-0024715—not purchase of a child detector or a restricted hot-car campaign.

All reported recipient costs are charged. Only prospective hot-car mortality prevention is quantified; pets, other injury mechanisms, nonfatal morbidity, and family wellbeing are not automatically converted into QALYs. Past safety standards are not credited again to new support.

2. Monitoring and information sharing

The organization’s incident database is useful for identifying scenarios and affected families, but documented cases are not a controlled estimate of prevented deaths. Changes in reporting, annual weather, exposure, and technology adoption can change counts without measuring charity impact. NHTSA reports 31 deaths in 2025, while its technical report describes a historical average near 37 annually.

The 2026 regulatory agenda still places the rear-designated seating-position alert rule at prerule stage and lists a July 2026 planned advance notice; a scheduled date is not evidence of completion. The mandate to issue a rule is not an implemented universal direct-detection standard. Some secondary summaries describe universal alerts in 2025; this review does not rely on that characterization.

Better monitoring would connect risk-weighted vehicle penetration, system activation, false alarms, timely rescues, and adoption changes attributable to KACS. Brochures, media impressions, and coalition participation alone cannot establish those links.

3. Qualitative assessment

The mechanism does not require a charity-specific randomized trial to be plausible, but it does require more than a sensor detecting movement. NHTSA tested 12 door-logic systems; none could alert when a child independently entered an unlocked vehicle, and some scenarios produced no alert. Transport Canada’s testing favored radar with strong alerts over simpler systems. Its full report was available only by request, so this review uses the published summary and does not invent measured accuracy.

CHOP researchers examined 354 fatal cases and 10 exemplar scenarios. No single technology covered every scenario; alert recipients still had to intervene. This is retrospective scenario analysis, not a real-world mortality-effect estimate. I therefore use separate subjective detection, timely rescue, and operating-reliability factors rather than calling theoretical coverage lives saved.

I judge a small positive marginal influence plausible given KACS’s sustained focus and documented collaboration. My central opinion credits KACS 20% of a hypothetical 20-percentage-point coalition-induced chance of accelerating stronger protection: an effective4-percentage-point contribution before donor substitution. Zero remains entirely possible.

The CHOP source here is the investigators’ public research summary; journal full text was inaccessible. No unreported field-effect coefficient is attributed to it.

4. What do you get for your dollar?

The hot-car working estimate is approximately $920,000 per 10 USA QALYs. Favorable assumptions give about $38,000; a low-influence scenario exceeds $4.9 billion, with zero and negative outcomes also shown. These are broad subjective scenarios, not confidence bounds or a claim that the expected value is a likely realized outcome.

The model asks whether stronger detection and rescue systems arrive three years earlier than otherwise. It slowly ramps additional protection across the at-risk vehicle population, discounts annual benefits, and allows baseline adoption to catch up. It does not presume every vehicle changes immediately or that KACS is responsible for all future hot-car prevention. The industry’s pre-existing commitment and already-available systems are part of the counterfactual.

About 0.52 discounted expected deaths averted under the full five-year cost-equivalent central model, at 27 QALYs per death, produces about 13.94 USA QALYs. The death-to-QALY value is a judgment based on a long remaining lifespan with annual discounting, not 80 undiscounted healthy years. Other portfolio health is assigned zero net credit, not assumed proven harmless.

The donor-facing cost excludes manufacturers’ incremental hardware and public implementation costs. Because the alternative also adopts later, the incremental social cost is the extra cost of acceleration or stronger features—not necessarily the entire eventual hardware bill. A current social-cost estimate remains unavailable.

Model, assumptions and sensitivity

All reported ordinary recipient costs C=$256,454/year, held constant in real FY2024 dollars for Y=5years. Includes administration/fundraising; direct netted expenses checked. Scenario G=$10,000 is disclosure normalization only. Unpriced volunteer time and hardware/government implementation costs are not claimed included in reported cash expenses; those are separate resource/capacity uncertainties.

National US prevention pathway with g=.99 subjective allowance for exact 50states+DC boundary and exposure coverage. State education remains charged even when not modeled. No pet QALYs or foreign beneficiaries are credited. allPopulationQalys covers this quantified pathway before g, not complete worldwide organizational impact.

F1(t)=min(cap,rate*max(0,t-L1+1)); F0(t)=min(cap,rate*max(0,t-L0+1)); A=sum(t=1..T)[F1(t)-F0(t)]/(1.03^t). H=D*d*r*k*q*A. Q_USA(G)=G/(Y*C)*b*(p*a*H-h)*g. Central: C256454,Y5,D30,d.8,r.7,k.9,q27,b.5,p.2,a.2,rate.05,cap.8,L1=3,L0=6,T20,g.99,h0. Price10*G/Q_USA. Adoption paths are risk-weighted fractions of otherwise-fatal exposures receiving stronger protection, not observed shares of cars sold.

C
256454 USD/year all recipient reported expenses (observed). FY2024 Form990 total functional expenses; direct cost lines require no addback. [kacs-99024]
Y
5 years of recipient expenditure (judgment). Finite sustained advocacy effort producing one adoption acceleration; not repeatedly credited every year.
historical-deaths
37 reported USA deaths/year (observed). NHTSA report historical mean; surveillance quantity rather than ascertainment-complete burden. [nhtsa-door-test]
deaths-2025
31 reported USA deaths (observed). Latest completed-year count reported by NHTSA. [nhtsa-heatstroke]
D
30 future baseline deaths/year (judgment). Near recent31 and below historical37; allows existing prevention progress.15–40 tested; weather and ascertainment uncertain. [nhtsa-heatstroke] [nhtsa-door-test]
d
0.8 probability of detection in relevant otherwise-fatal exposure (judgment). Mechanistic working value, not measured field sensitivity; research shows substantial scenario heterogeneity. [chop-tech25] [tc-test25]
r
0.7 timely effective rescue conditional on detection and operating system (judgment). Allows unavailable/ignoring caregiver, failed escalation, and deliberate leaving; a warning is not a rescue. [chop-tech25]
k
0.9 fraction reliably operating and not disabled (judgment). Allows sensor failure, nuisance-alert disablement and real-world degradation; technical accuracy alone is insufficient. [nhtsa-door-test] [tc-test25]
q
27 discounted QALYs per averted child death at event date (judgment). Approximately a long 75-year future with utility around 0.9 and 3% annual discount, rounded; not a patient-specific life table. 20–30 QALYs tested. Nonfatal morbidity and family health are omitted.
rate
0.05 additional risk-weighted protection fraction/year after adoption starts (judgment). Slow replacement and unequal access to newer vehicles; not an observed market-penetration statistic.
cap
0.8 maximum risk-weighted population share covered (judgment). Leaves20% beyond this diffusion pathway because of older vehicles, noncovered settings or access constraints; distinct from detection efficacy.
L1
3 future year of first stronger protection under accelerated path (judgment). No asserted legal compliance date. Forward-looking model despite current prerule status. [nhtsa-agenda26]
L0
6 future year of first comparable protection absent campaign acceleration (judgment). Counterfactual adoption still happens through industry and government. A3-year difference is a subjective opportunity, not a measured delay. [industry-reminders] [nhtsa-agenda26]
T
20 years of adoption-difference accounting (judgment). Finite horizon. At central year20 adoption differs by5percentagepoints; later residual benefit omitted.
p
0.2 incremental coalition probability of the modeled acceleration (judgment). Subjective policy/implementation influence, not unconditional adoption probability;0–.5 scenarios. [kacs-policy] [nhtsa-agenda26]
a
0.2 KACS share of coalition-induced acceleration probability (judgment). Focused longstanding group but other advocates, engineers, regulators and manufacturers dominate much of delivery; tested.05–.3. [kacs-policy] [kacs-partners25]
b
0.5 net additional effort per dollar over five years (judgment). Allows reserves, grant and donor substitution, and unpriced volunteer capacity. No verified marginal operating plan. [kacs-99024] [kacs-99024d] [kacs-99024o]
h
0 other-portfolio net QALYs lost per five-year full-cost effort (judgment). Central zero net credit for other work; adverse cases retain losses even if no acceleration occurs.
g
0.99 fraction of credited health in50states+DC (judgment). Near-national pathway; small boundary allowance, not a measured spending share.
central-adoption-path
[{"t":1,"with":0,"without":0,"discountedDifference":0},{"t":2,"with":0,"without":0,"discountedDifference":0},{"t":3,"with":0.05,"without":0,"discountedDifference":0.04575708296765798},{"t":4,"with":0.1,"without":0,"discountedDifference":0.08884870479156888},{"t":5,"with":0.15000000000000002,"without":0,"discountedDifference":0.1293913176576246},{"t":6,"with":0.2,"without":0.05,"discountedDifference":0.12562263850254815},{"t":7,"with":0.25,"without":0.1,"discountedDifference":0.12196372670150303},{"t":8,"with":0.30000000000000004,"without":0.15000000000000002,"discountedDifference":0.11841138514709035},{"t":9,"with":0.35000000000000003,"without":0.2,"discountedDifference":0.11496250985154402},{"t":10,"with":0.4,"without":0.25,"discountedDifference":0.11161408723450876},{"t":11,"with":0.45,"without":0.30000000000000004,"discountedDifference":0.10836319148981428},{"t":12,"with":0.5,"without":0.35000000000000003,"discountedDifference":0.1052069820289459},{"t":13,"with":0.55,"without":0.4,"discountedDifference":0.10214270099897664},{"t":14,"with":0.6000000000000001,"without":0.45,"discountedDifference":0.09916767087279288},{"t":15,"with":0.65,"without":0.5,"discountedDifference":0.09627929210950763},{"t":16,"with":0.7000000000000001,"without":0.55,"discountedDifference":0.0934750408830171},{"t":17,"with":0.75,"without":0.6000000000000001,"discountedDifference":0.09075246687671557},{"t":18,"with":0.8,"without":0.65,"discountedDifference":0.08810919114244235},{"t":19,"with":0.8,"without":0.7000000000000001,"discountedDifference":0.05702860268119245},{"t":20,"with":0.8,"without":0.75,"discountedDifference":0.02768378770931675}] year, protected-risk fraction with/without, discounted difference (judgment). Explicit inspectable annual path; A=1.7247803796467673 discounted protected-risk years.
unpriced-resource-inputs
null USD of volunteer time and accelerated implementation (unknown). Reported cash costs do not value all donated labor or manufacturers’ incremental engineering/hardware; required for social-cost rather than donor-cost estimate. [kacs-99024] [tc-test25]

Subjective three-year acceleration expected value: Cost: $10K; USA QALYs: 0.10872641468101513; all-population QALYs: 0.10982466129395467. Hot-car mortality only, all recipient reported costs. Inputs={"C":256454,"Y":5,"D":30,"d":0.8,"r":0.7,"k":0.9,"q":27,"b":0.5,"p":0.2,"a":0.2,"rate":0.05,"cap":0.8,"L1":3,"L0":6,"T":20,"g":0.99,"h":0}. Discounted incremental protected-risk years A=1.7247803796467673; full accelerated-adoption health H=704.1243421869962 QALYs. These are analyst scenarios, not measured charity effects.

Stronger and faster marginal influence: Cost: $10K; USA QALYs: 2.6200655378688564; all-population QALYs: 2.6465308463321784. Five-year acceleration, higher reach and responsiveness; optimistic scenario not a confidence bound. Inputs={"C":256454,"Y":5,"D":40,"d":0.9,"r":0.9,"k":0.95,"q":30,"b":0.8,"p":0.5,"a":0.3,"rate":0.07,"cap":0.8,"L1":2,"L0":7,"T":20,"g":0.99,"h":0}. Discounted incremental protected-risk years A=3.0625650750183766; full accelerated-adoption health H=2827.972590271969 QALYs. These are analyst scenarios, not measured charity effects.

Little influence with weak coverage: Cost: $10K; USA QALYs: 0.00002035665303514452; all-population QALYs: 0.000020562275793075274. Only one-year acceleration, lower rescue and donor additionality. Inputs={"C":256454,"Y":5,"D":15,"d":0.5,"r":0.3,"k":0.7,"q":20,"b":0.1,"p":0.05,"a":0.05,"rate":0.03,"cap":0.5,"L1":5,"L0":6,"T":20,"g":0.99,"h":0}. Discounted incremental protected-risk years A=0.33481129372935403; full accelerated-adoption health H=10.546555752474651 QALYs. These are analyst scenarios, not measured charity effects.

Only one-year acceleration: Cost: $10K; USA QALYs: 0.037318532898213776; all-population QALYs: 0.037695487775973514. Keeps all central inputs except counterfactual arrives sooner. Inputs={"C":256454,"Y":5,"D":30,"d":0.8,"r":0.7,"k":0.9,"q":27,"b":0.5,"p":0.2,"a":0.2,"rate":0.05,"cap":0.8,"L1":3,"L0":4,"T":20,"g":0.99,"h":0}. Discounted incremental protected-risk years A=0.5920021691957862; full accelerated-adoption health H=241.67896555248774 QALYs. These are analyst scenarios, not measured charity effects.

Ten percent additional effort: Cost: $10K; USA QALYs: 0.021745282936203023; all-population QALYs: 0.021964932258790933. Most support replaces reserves or other funding. Inputs={"C":256454,"Y":5,"D":30,"d":0.8,"r":0.7,"k":0.9,"q":27,"b":0.1,"p":0.2,"a":0.2,"rate":0.05,"cap":0.8,"L1":3,"L0":6,"T":20,"g":0.99,"h":0}. Discounted incremental protected-risk years A=1.7247803796467673; full accelerated-adoption health H=704.1243421869962 QALYs. These are analyst scenarios, not measured charity effects.

Same adoption with or without effort: Cost: $10K; USA QALYs: 0; all-population QALYs: 0. No technology benefit even if policy advocacy is active. Inputs={"C":256454,"Y":5,"D":30,"d":0.8,"r":0.7,"k":0.9,"q":27,"b":0.5,"p":0.2,"a":0.2,"rate":0.05,"cap":0.8,"L1":3,"L0":3,"T":20,"g":0.99,"h":0}. Discounted incremental protected-risk years A=0; full accelerated-adoption health H=0 QALYs. These are analyst scenarios, not measured charity effects.

No KACS marginal policy contribution: Cost: $10K; USA QALYs: 0; all-population QALYs: 0. Other actors deliver identical acceleration. Inputs={"C":256454,"Y":5,"D":30,"d":0.8,"r":0.7,"k":0.9,"q":27,"b":0.5,"p":0.2,"a":0,"rate":0.05,"cap":0.8,"L1":3,"L0":6,"T":20,"g":0.99,"h":0}. Discounted incremental protected-risk years A=1.7247803796467673; full accelerated-adoption health H=704.1243421869962 QALYs. These are analyst scenarios, not measured charity effects.

No influence plus1QALY portfolio harm: Cost: $10K; USA QALYs: -0.003860341425752766; all-population QALYs: -0.003899334773487643. Illustrative harm from risk compensation, distraction, privacy-related health effects or other portfolio actions; not an observed estimate. Inputs={"C":256454,"Y":5,"D":30,"d":0.8,"r":0.7,"k":0.9,"q":27,"b":0.5,"p":0.2,"a":0,"rate":0.05,"cap":0.8,"L1":3,"L0":6,"T":20,"g":0.99,"h":1}. Discounted incremental protected-risk years A=1.7247803796467673; full accelerated-adoption health H=704.1243421869962 QALYs. These are analyst scenarios, not measured charity effects.

Only ten years of incremental coverage: Cost: $10K; USA QALYs: 0.05399640678078787; all-population QALYs: 0.054541825031098855. Tests long-horizon dependence. Inputs={"C":256454,"Y":5,"D":30,"d":0.8,"r":0.7,"k":0.9,"q":27,"b":0.5,"p":0.2,"a":0.2,"rate":0.05,"cap":0.8,"L1":3,"L0":6,"T":10,"g":0.99,"h":0}. Discounted incremental protected-risk years A=0.8565714528540457; full accelerated-adoption health H=349.6867299131356 QALYs. These are analyst scenarios, not measured charity effects.

Counterfactual: Existing government campaigns, simple reminders, manufacturer innovation, other charities and eventual stronger detection continue. Central compares the same adoption ramp beginning in future year3 rather than6; it credits temporary earlier coverage, not permanent prevention of all baseline deaths. D30 incorporates a modest decline from historical37 deaths, close to recent31, rather than projecting an immutable historic rate.

Attribution: p=.2 is a subjective coalition-induced chance of the specified three-year acceleration; a=.2 is KACS’s share of that causal increment. Effective contribution .04 is not the chance any regulation happens. b=.5 then discounts effort that reserves, government/private grants, or other donors would fund anyway. All portfolio costs are charged, so no second hot-car budget share is applied.

Working expected value, not an experimentally estimated charity return or a median realized policy outcome. Adoption and causal attribution dominate; broad scenarios include no acceleration, no influence, funding substitution, and adverse net health elsewhere. h is not discounted by policy success, preserving downside even when p=0.

Sensitivity

  • Central A=1.72478038 and H=704.124342 QALYs before policy/funding credit. Effective p*a=.04 and b=.5 give13.941662 USAQALYs per five-year expense-equivalent cost $1,282,270.
  • One-year rather than three-year acceleration increases price to approximately $2.68M per10USAQALYs. b=.1 increases the central price fivefold. Matching counterfactual adoption removes modeled benefit.
  • To meet $1M/10, five-year effort requires12.8227 netUSAQALYs; at 27 Q/death this is.474915 discounted expected deaths. The $100k/10 threshold requires128.227 QALYs, or4.74915 deaths.
  • At central clinical/adoption inputs and b=.5,g=.99,h0, effective p*a must be at least.03679 for $1M/10 and.36790 for $100k/10. Central.04 narrowly clears the former and is far below the latter.
  • Central net other-portfolio harm above p*a*H=28.164974 QALYs per five-year effort erases estimated benefit before b/g. Zero-other-health is not a guaranteed lower bound.
  • A simple reminder is not direct detection. Treating industry’s2025 reminder commitment as universal effective protection, or treating statutory rulemaking deadlines as actual deployment, invalidates this model.
  • Hardware/public costs cannot be called zero. For an incremental social-cost calculation, compare cost and benefit of earlier/stronger adoption against later adoption, including other uses of sensors; do not debit every future detector while crediting only a brief acceleration.

Unresolved inputs

  • Causal timing or specification changes due to KACS rather than other coalition members, regulators or manufacturers.
  • Current marginal expenditure/capacity plan, donor and public-grant substitution, restrictions and availability of cryptocurrency holdings.
  • Risk-weighted penetration of direct detection, reliability/disablement, caregiver response, and observed fatality/nonfatal health effects.
  • Current discounted marginal hardware/public-resource costs and net non-hot-car portfolio health effects.

5. Funding and previous grants

The FY 2024 return reports $256,454 expenses: $179,537 program, $27,906 management, and $49,011 fundraising. This is cash-basis reporting; the return says statements were not independently audited or reviewed. FY 2023 requires adding $4,517 direct fundraising-event cost to $258,364 functional expenses, for $262,881 gross. FY 2022 expenses were $265,887. The three-year mean is $261,741.

FY 2024 revenue was $597,894, including $323,028 noncash support and $21,003 government grants. [Schedule M](https://projects.propublica.org/nonprofits/full_text/202513039349301566/IRS990Schedule M) identifies a $300,000 cryptocurrency donation. [Schedule D](https://projects.propublica.org/nonprofits/full_text/202513039349301566/IRS990Schedule D) records $303,125 in Coinbase/Bitcoin at year end; this is not its present market value. Cash and savings totaled $167,395, versus $428,767 net assets overall. Donor restrictions are not separately classified under this reporting presentation; none are assumed freely spendable merely from net assets.

The ordinary giving page illustrates $50 for materials reaching 500 families. That is not full recipient cost per family or a proven health return. Staff/consultant and volunteer capacity, future reserve use, and replacement donors determine what additional support changes. The central 50% additionality is an opinion over five years, not an observed staffing commitment.

Annual expenses

Organization-level spending, including programs, administration and fundraising. The research list averages three comparable, consecutive full fiscal years when available.

  • FY 2022: $266K; Kids and Cars Inc, EIN20-0024715, 12-month period, Recipient cash-basis Form990 functional expenses plus direct costs netted against revenue. Source
  • FY 2023: $263K; Kids and Cars Inc, EIN20-0024715, 12-month period, Recipient cash-basis Form990 functional expenses plus direct costs netted against revenue. Source
  • FY 2024: $256K; Kids and Cars Inc, EIN20-0024715, 12-month period, Recipient cash-basis Form990 functional expenses plus direct costs netted against revenue. Source

6. Sources

  1. Kids and Cars Inc FY2024 Form990. Kids and Cars Inc / IRS via ProPublica. Published: not stated; retrieved: 2026-09-13.
  2. Kids and Cars Inc FY2023 Form990. Kids and Cars Inc / IRS via ProPublica. Published: not stated; retrieved: 2026-09-13.
  3. Kids and Cars Inc FY2022 Form990. Kids and Cars Inc / IRS via ProPublica. Published: not stated; retrieved: 2026-09-13.
  4. FY2024 ScheduleD: other investments. Kids and Cars Inc / IRS via ProPublica. Published: not stated; retrieved: 2026-09-13.
  5. FY2024 ScheduleM: noncash contributions. Kids and Cars Inc / IRS via ProPublica. Published: not stated; retrieved: 2026-09-13.
  6. FY2024 ScheduleO: consultants and governance. Kids and Cars Inc / IRS via ProPublica. Published: not stated; retrieved: 2026-09-13.
  7. Current activities and mission. Kids and Car Safety. Published: not stated; retrieved: 2026-09-13.
  8. Accomplishments and history. Kids and Car Safety. Published: not stated; retrieved: 2026-09-13.
  9. Hot cars: federal legislation and technology. Kids and Car Safety. Published: not stated; retrieved: 2026-09-13.
  10. Hot-car incident documentation. Kids and Car Safety. Published: not stated; retrieved: 2026-09-13.
  11. Ordinary donation page and illustrative education costs. Kids and Car Safety. Published: not stated; retrieved: 2026-09-13.
  12. 2026 regulatory agenda: Rear Designated Seating Position Alert, RIN2127-AM49. NHTSA / OIRA. Published: not stated; retrieved: 2026-09-13.
  13. Prevent hot-car deaths: current burden and prevention. NHTSA. Published: not stated; retrieved: 2026-09-13.
  14. Testing rear-door-logic based unattended child reminder systems, DOT HS813669. Prasad and Snyder / NHTSA. Published: not stated; retrieved: 2026-09-13.
  15. New research evaluates technological solutions to prevent pediatric vehicular heatstroke. Jalaj Maheshwari / Children's Hospital of Philadelphia CIRP. Published: 2025-07-23; retrieved: 2026-09-13.
  16. Technical report: testing occupant monitoring systems. Transport Canada / National Research Council Canada. Published: 2025-09-04; retrieved: 2026-09-13.
  17. 2024 industry heatstroke report announcement. Alliance for Automotive Innovation. Published: 2024-11-04; retrieved: 2026-09-13.
  18. National Heatstroke Prevention Day partner event. Safe Kids Worldwide. Published: 2025-05-01; retrieved: 2026-09-13.