The eastern metro area has 4,000 residents and one community health coordinator position. Day seventeen of the heat emergency, and she's making rounds through a neighborhood where people have stopped calling ambulances.
They need help. The air quality index hasn't dropped below 250 in two weeks. People with COPD are struggling. People with heart failure are decompensating. They're not calling for it.
She arrives at the Martinez house at 0800. Mrs. Martinez, 71, COPD, oxygen saturation at 89%. The coordinator adjusts the oxygen flow, checks medication adherence, teaches the daughter how to recognize warning signs that would require hospital transport. They discuss the cooling schedule—Mrs. Martinez will spend afternoons at her sister's house three blocks away, the only home in the network with functioning air conditioning.
"Should we call an ambulance?" the daughter asks.
The coordinator checks vitals again. Oxygen saturation holding at 91% with adjusted flow. Respiratory rate elevated but stable. "Not yet. We can manage this here if you're comfortable with it."
The daughter nods. She's comfortable with it. More comfortable than she would be watching her mother get processed through the regional hospital network's triage protocol.
During the 2029 Phoenix crisis, the daughter's best friend watched her mother get reassigned from P1 to P3 status as her condition deteriorated. The triage officer explained that intensive resources would be better used on patients with higher survival probability. The decision was ethically sound according to protocol criteria. The decision was also the moment an entire network of neighbors and friends learned that the healthcare system would abandon you when things got bad enough.
The protocol treats this as an education problem, a communication failure that can be solved with better messaging about how triage criteria work. People understand the protocol, though. They understand that systematic allocation based on SOFA scores optimizes for aggregate survival by treating patients as interchangeable units with measurable characteristics. They understand that when you become a unit with unfavorable characteristics, the system will allocate its scarce resources elsewhere.
So they're not engaging with the system. Mrs. Martinez is stable at home, not taking up a hospital bed the regional network desperately needs. The coordinator's intensive relationship-based care is inefficient by protocol standards—her time could be "better used" processing more patients through systematic assessment. But the forty-three patients on her rounds are following heat safety protocols, monitoring their own symptoms, calling before situations become emergencies. They're doing this because they trust that their healthcare provider sees them as people rather than SOFA scores.
This is the feedback loop the protocol can't measure: systematic triage optimizes for immediate survival during discrete surge events. But treating patients as interchangeable units erodes the trust that determines whether communities engage with healthcare systems during crises. And whether communities engage determines capacity across repeated surge events.
The coordinator can point to evidence that community health workers serve as crucial bridges during disasters. Can demonstrate that relationship-based care maintains the social cohesion communities need for long-term climate adaptation. She can't prove that maintaining these relationships will matter more over decades of recurring crises than maximizing survival during any single surge.
The hospital administrators have data showing that systematic triage reduces mortality compared to overwhelmed individual decision-making. They're right about that. During surge events, when demand exceeds capacity by 60%, systematic allocation produces better aggregate outcomes than chaos. "Better aggregate outcomes during surge events" assumes that surge events are discrete episodes rather than the new baseline, though. It assumes that optimizing for each individual crisis produces optimal outcomes across decades of compounding crises.
The coordinator finishes her rounds at 1600. Forty-three patients stable at home. Mrs. Martinez's oxygen saturation holding at 92%. The network of neighbors coordinating cooling access, medication reminders, overnight monitoring for elderly residents living alone. Some of these patients would probably survive just as well if processed through the regional protocol. Some might not survive the next surge, when relationship-based approaches can't scale to meet demand.
They trust their healthcare system enough to engage with it, though. They follow the protocols recommended to them. They call when things get worse, before they become emergencies. And when this surge ends and the next one begins—and the projections show more coming, longer duration, higher intensity—they'll still maintain the relationships that let communities organize collective survival.
The heat map still glows red. The coordinator pulls up tomorrow's schedule. Forty-three patients who need twice-daily visits. Forty-three patients who aren't calling ambulances because they have something the protocol can't provide: someone who will still see them as worth caring for when the measurements say otherwise.