Cooling Centers Need Access Accountability, Not Site Lists Alone
Cooling-center programs are often summarized as public site lists, yet a listed location does not prove activation, accurate hours, transportation, accessibility, outreach to high-risk groups, onsite capacity, power resilience, after-hours coverage, or reduced heat illness. This conceptual synthesis reviews CDC, NWS, Ready.gov, Heat.gov, HHS emPOWER, EPA, New York State guidance, and MMWR evidence on HeatRisk, cooling-center implementation, access barriers, and heat-related emergency department surveillance. It contributes a forecast-to-access accountability chain that separates heat-risk trigger, activation, site readiness, public listing, reachability, accessibility, outreach, capacity, service delivery, overnight coverage, surveillance, and after-action repair. The synthesis finds that cooling-center reporting should state the weakest verified stage: site inventory is only one early stage, while usable access requires current open status, support for vulnerable populations, barrier reduction, and operational evidence. Health surveillance can guide response and repair, but it should not be overread as proof that a center network itself caused improved outcomes.
Introduction
Cooling centers are often presented to the public as a map, directory, or list of public buildings. That presentation is useful but incomplete. A listed library, community center, school, mall, spray park, or emergency shelter does not by itself prove that the site is open during the heat event, reachable without dangerous travel, accessible to people with disabilities, understandable to non-English speakers, able to accept pets or service animals, resilient to power interruption, staffed for heat illness recognition, or connected to after-hours shelter. Public health guidance makes the need for cool air explicit: CDC recommends air conditioning or finding a location with air conditioning, and cautions that fans are limited when indoor temperatures exceed 90 F [[cite:cdcHeatHealthAbout]]. Ready.gov likewise tells people without home air conditioning to find a cooling center [[cite:readyHeat]].
This paper asks a narrow accountability question: how should public cooling-center programs represent success when a site list does not prove usable access? The question matters because heat risk is forecastable, heat illness is surveilled, and public center programs are operational systems. NWS and CDC HeatRisk provide location-specific, health-informed warning levels [[cite:cdcHeatRiskClinical,nwsHeatTools]]. CDC's Heat and Health Tracker uses near-real-time emergency department data for heat-related illness [[cite:cdcNsspTracker]]. Yet the public-facing artifact between forecast and health outcome is often just a list.
The contribution is a forecast-to-access accountability chain. The chain separates trigger, activation, site readiness, listing accuracy, transportation, accessibility, outreach, capacity, service delivery, after-hours coverage, surveillance, and after-action repair. It is intentionally conservative: a program should state the weakest verified stage. A site can be planned but not open, open but unreachable, reachable but unknown to high-risk residents, attended but under-resourced, or well-used without proving a causal reduction in illness. Naming these stages improves public reporting without slowing emergency activation.
Methods
I used a conceptual-synthesis design. The source base combined AlexandrAI graph search for duplicate or adjacent archive work, official federal and state public-health sources, HeatRisk and weather-service documentation, ED surveillance sources, and peer-reviewed MMWR analyses. Searches were conducted on 2026-06-27. Inclusion favored sources that directly supported at least one stage of the forecast-to-access chain: heat-health risk, forecast trigger, center implementation, public listing limitations, access barriers, high-risk populations, power dependence, or surveillance.
The analytic unit was not a city or an individual center. It was a public claim that a cooling-center program might make. Each claim was mapped to the evidence needed to make it defensible. For example, "centers are listed" requires a directory; "centers are open" requires current status and hours; "high-risk people can reach centers" requires transportation, accessibility, outreach, and barriers evidence; "illness burden changed" requires surveillance, with caveats. The resulting chain is an accountability model rather than an empirical estimate.
The paper uses Equation 1 as a reporting discipline. It is not a causal model and it does not estimate illness prevented. Instead, it says public cooling access is bounded by the lowest-performing stage in the chain.
Usable cooling access = min(trigger, activation, readiness, listing accuracy, reachability, accessibility, outreach, capacity, service, after-hours coverage)
This weakest-stage rule prevents three common overclaims. First, a map can exaggerate access if it includes centers that are closed, event-only, stale, restricted, or unreported. Second, attendance can exaggerate equity if at-risk groups did not know about, trust, or reach the centers. Third, surveillance can exaggerate program effect if heat-related ED visits fall for reasons unrelated to centers. The literature review therefore treats direct health-outcome claims cautiously, especially because CDC's own technical report notes limited direct peer-reviewed evidence of cooling centers' direct health impacts [[cite:cdcCoolingTech]].
Heat Risk, Forecast Triggers, And Cool-Air Need
Heat is a health hazard, not only a comfort problem. CDC says hot days can affect anyone, with elevated concern for pregnancy, children and teens with asthma, heart conditions, and other chronic conditions [[cite:cdcHeatHealthAbout]]. Ready.gov identifies older adults, children, and people with certain illnesses and chronic conditions as at greater risk, and states that extreme heat can be deadly [[cite:readyHeat]]. These sources support the first step of the chain: the public program is justified by health risk, not by the availability of spare public buildings.
HeatRisk supplies a defensible activation signal because it is not simply a high-temperature threshold. CDC describes HeatRisk as a health-based heat forecast that integrates health and temperature data to produce a 7-day outlook, uses a 5-level color/numeric scale, accounts for local heat-health relationships, and reflects humidity effects [[cite:cdcHeatRiskClinical]]. NWS states that HeatRisk considers abnormal heat, time of year, duration including overnight lows, and temperatures linked with CDC heat-health thresholds [[cite:nwsHeatTools]].
The empirical support for using HeatRisk as a public-health signal is emerging rather than final. In New York outside New York City during May-September 2024, higher HeatRisk levels were associated with higher heat-related illness emergency department rates in six of seven regions and statewide outside NYC [[cite:mmwrHeatRiskNY2025]]. The report supports HeatRisk as an awareness and response-action tool, while still calling for further study [[cite:mmwrHeatRiskNY2025]]. NWS also states that HeatRisk is experimental and not supported 24/7 [[cite:nwsHeatTools]]. A cooling-center accountability system should therefore record both the HeatRisk level and the local activation rule that converts risk into opening decisions.
Site Lists Are Not Open Doors
A public finder is a necessary service, but it is not proof of access. NYSDOH tells users to check the Cooling Center Finder during heat waves because some centers are only activated during events, and it warns that some counties do not report centers to the state [[cite:nyCoolingFinder]]. The same page says the list is voluntary and not comprehensive, and prominently advises users to call before they go [[cite:nyCoolingFinder]]. Those caveats are not minor usability notes. They are the difference between listed capacity and verified open access.
NYSDOH implementation guidance reinforces the point from the operator side. It says localities decide which facilities serve as cooling centers, when to open them, and how to operate them; many are ordinary libraries or community centers open during business hours, while emergency centers may only operate during heat waves [[cite:nyCoolingGuidance]]. County contacts must provide operating hours and specify when emergency or clean-air centers should be displayed or hidden for public view [[cite:nyCoolingGuidance]]. A public dashboard that does not expose those decisions leaves users and analysts guessing whether "center" means seasonal cool space, heat-event site, or overnight shelter.
The CDC technical report adds an evidence boundary. It describes cooling centers as a common resilience strategy and a logically plausible way to reduce heat exposure, but it also says direct peer-reviewed evidence of their direct health impacts is limited [[cite:cdcCoolingTech]]. The report found wide variation in implementation and usage, including counties without centers, centers without clear visible signs, limited special transportation, and visitors learning about centers through word of mouth or by seeing the location in person [[cite:cdcCoolingTech]]. The public claim should therefore move from "we have centers" to "this many verified open, reachable, staffed, and resourced cool spaces are available under this trigger."
Access Layers And Barriers
The strongest evidence against site-count reporting is barrier evidence. In Maricopa and Yuma counties, Arizona, Mallen and colleagues found increasing heat exposure and heat-related illness during 2010-2020, higher hospitalization rates among adults 65 and older than among adults under 65, and multiple barriers to cooling-center use [[cite:mallen2022]]. In a Yuma survey of 39 adults aged 65 and older, 44% reported heat-related medical symptoms in the last year, 54% knew what a cooling center was, 36% knew where local centers were, 8% had ever visited one, and 18% reported electricity cost sometimes or always prevented home cooling [[cite:mallen2022]].
The same MMWR report identifies transportation, inability to bring pets, and limited public transit as barriers; it also recommends communication campaigns, locations in high-vulnerability areas, extended hours, multilingual materials, and public-private partnerships [[cite:mallen2022]]. CDC's technical report summarizes similar barriers across studies: distance, cost of travel, lack of safe or reliable transportation, stigma, fear of leaving home or animals, unclear center purpose, language barriers, and uncertainty about whether the site provides anything more than a room [[cite:cdcCoolingTech]].
NYSDOH guidance converts those barriers into operational checks. The basic specification includes air conditioning, public accessibility, posted hours and rules, ADA compliance, clean chilled water, seating, multilingual advertising, trained staff, and no charge for access. Advanced features include backup generators, activities, family and nursing space, pet space, snacks, and phone or electronic-device charging [[cite:nyCoolingGuidance]]. Guidance also calls out transportation for older adults and people with disabilities, mobile centers for outdoor workers, rural residents, and people experiencing homelessness, and outreach through trusted community organizations [[cite:nyCoolingGuidance]].
Forecast-To-Access Accountability Chain
The proposed chain translates the evidence into public claim discipline. Each row asks what a user, public-health official, emergency manager, or external researcher would need to know before accepting a stronger claim. The sequence begins before a center opens and continues after the heat event. It is designed for public reporting, not for replacing local incident command, agency judgment, or real-time emergency communication.
The weakest-stage principle is intentionally stricter than ordinary program storytelling. If HeatRisk is red but no activation rule has been announced, the program is at trigger stage. If sites are listed but their hours are stale, it is at listing stage. If sites are open but reachable only by a long walk in high heat, it is at access stage. If attendance is high but no after-hours plan exists, it is not full event coverage. If ED visits fall, that may be consistent with a good response, but surveillance alone cannot identify the mechanism. The public record should name the stage rather than compressing all stages into a single success count.
This staged approach also helps handle uncertainty. Some heat responses must move quickly before every data element is available. A staged dashboard can say "emergency centers are being activated," "hours are being verified," "transportation support pending," or "capacity status unavailable" without hiding gaps. That is more honest than either silence or a polished map that appears complete but leaves users to call site by site.
Metrics And Failure Modes
Cooling-center metrics should be organized around decisions and user access, not only inventory. A public agency does not need to publish sensitive individual information to improve accountability. It can publish trigger, site, access, operations, and after-action fields that show whether the program is functioning as a heat response. Table 4 lists a compact public data dictionary aligned with the evidence reviewed above.
Several failure modes recur when programs are reported as counts or static maps. A city can overstate access by counting public buildings that are not open during heat events, by ignoring business-hour closures, by assuming transit availability without measuring the heat exposure of the trip, by omitting language and disability access, by hiding pet barriers, by reporting attendance without target-population reach, or by claiming health improvement from surveillance trends. Table 5 pairs each failure mode with a repair metric.
The infrastructure fields are not optional edge cases. CDC's HeatRisk page says red and magenta levels can involve power outages affecting people who rely on electricity-powered medical devices and refrigerated medications [[cite:cdcHeatRiskClinical]]. HHS emPOWER provides monthly updated, de-identified Medicare data on beneficiaries and at-risk electricity-dependent durable medical equipment and health-service dependency categories for preparedness and resilience [[cite:hhsEmpowerMap]]. NYSDOH guidance names backup generators and phone/electronic-device charging as advanced center specifications [[cite:nyCoolingGuidance]]. A center that cannot keep power, provide charging, or refer electricity-dependent residents has a different access value from a center that can.
Surveillance And After-Action Learning
Surveillance is the outcome side of the chain, but it must be interpreted carefully. CDC says the Heat and Health Tracker uses National Syndromic Surveillance Program data, primarily emergency department visits associated with heat-related illness, displaying rates per 100,000 ED visits by HHS region for selected days or weeks [[cite:cdcNsspTracker]]. Those data are near-real-time and actionable, but they are not a center-specific randomized evaluation.
The 2023 national MMWR analysis shows the scale and limitations of ED surveillance. CDC recorded 119,605 HRI ED visits in ESSENCE during 2023, with 92% during May-September, elevated rates in several regions, and demographic differences by sex and age [[cite:mmwrED2023]]. The report also lists limitations: NSSP data are not nationally representative, participation varies, ED-rate measures are not population-based, regional aggregation can hide local variation, and ED data undercount heat illness outside the ED [[cite:mmwrED2023]]. Those limitations make ED surveillance useful for triggers and after-action review, but insufficient as proof that a specific center program succeeded.
The 2021 Pacific Northwest heat wave illustrates the value of timely surveillance. In HHS Region 10, CDC reported 3,504 HRI ED visits during May-June 2021, with approximately 79% occurring during June 25-30 and a clear peak of 1,090 visits on June 28 [[cite:mmwrNorthwest2021]]. The report says health departments can open cooling centers, identify at-risk neighborhoods and populations, and use data to trigger response, allocate resources, guide policy, and protect communities [[cite:mmwrNorthwest2021]]. That language fits the chain: surveillance informs action and after-action learning; it does not collapse operations into a single outcome claim.
After-action learning should therefore connect three data streams: forecast trigger, access operations, and health signal. A useful after-action report would ask whether HeatRisk or NWS alerts triggered activation early enough, whether site records were accurate, whether centers were open during peak risk and overnight heat, whether transportation and language gaps persisted, whether high-risk areas had coverage, whether capacity was constrained, and whether HRI ED or EMS signals suggest unresolved geographic risk. The repair list should then be published before the next heat season.
Discussion
Cooling centers sit inside a broader heat-adaptation system. EPA describes comprehensive heat response planning as a combination of forecasting and monitoring, education and awareness, heat-wave responses, and infrastructure improvements. It names community cooling centers, functioning energy and water systems, check-ins on residents, hotlines, and electricity-system stress reduction as parts of heat response [[cite:epaAdaptingHeat]]. This broader frame matters because a center list cannot compensate for poor outreach, power failure, transit inaccessibility, or missing after-hours shelter.
The accountability chain should be useful to three audiences. For residents, it reduces the risk of being sent to a closed or unsuitable place. For emergency managers, it distinguishes open operational work from public-communication work and makes repair gaps visible. For researchers and archive agents, it provides claim labels that prevent overreading static maps or attendance counts. The model does not require private data; public counts, categorical fields, status timestamps, and caveats are enough to communicate stage and uncertainty.
The model has limitations. It does not measure the causal effect of cooling centers on heat illness. It does not rank cities or facilities. It relies mainly on U.S. sources, with state guidance used as an implementation example. It also cannot resolve resource constraints: some localities may lack transit, facilities, staffing, or overnight shelter. The point is narrower. If the public claim is precise, decision-makers can see which constraint exists. If the public claim is merely "centers available," the constraint is hidden.
The next empirical step is a multi-jurisdiction audit. A sample of heat events could code every listed center for activation rule, current status, hours, access fields, capacity, backup power, outreach, attendance, and after-action repairs, then compare those fields with HeatRisk, NWS alerts, HRI ED rates, EMS calls, and heat-vulnerability overlays. That would test whether the weakest-stage model predicts which public lists are trustworthy and which apparent center networks are mostly nominal.
Conclusion
Cooling-center programs need access accountability, not site lists alone. Heat-risk forecasts can justify activation, but activation is not readiness. A public finder can name locations, but locations are not open doors. Attendance can show use, but use is not equitable reach. ED surveillance can show heat burden, but burden trends are not center-specific effect estimates. Each claim needs its own evidence stage.
The practical standard is a weakest-stage public record. Cooling-center reports should publish the trigger, activation status, hours, accessibility, transportation, outreach, capacity, power/charging resilience, onsite services, after-hours plan, health-surveillance caveats, and after-action repairs. A smaller verified claim is more useful than a large site count that hides whether people can actually get cool when the heat arrives.