For months, experts in public health and medical ethics, from the National Academy of Medicine to the World Health Organization and scholars at Johns Hopkins and other universities, have been debating which groups should receive the earliest covid-19 vaccines first — and last. Paradoxically, while a considerable amount of energy is being spent on recommending who should receive the vaccine first, almost no attention is being devoted to how to distribute it.

All the current allocation plans recommend giving vaccines to front line health workers in the first wave of an immunization campaign, but plans differ after that. Some recommend vaccinating people over 65 years old first, while others prioritize police and firefighters. Other plans give higher priority to those with risk factors for high covid-19 mortality rates, such as hypertension, cancer and obesity, or those living in crowded or multigenerational housing. The Centers for Disease Control and Prevention intends to wait until the Food and Drug Administration authorizes a vaccine before issuing guidelines on who should receive it, and so far, the CDC has offered no hint of what those guidelines might say.

If that “how” question is not answered until a vaccine is in hand, it will be far too late. Once FDA authorizes a vaccine, there will be enormous pressure for distribution to begin immediately. If the infrastructure is not established to distribute early vaccine doses to the vulnerable and disadvantaged people who stand to benefit most, those supplies will go instead wherever it is politically and logistically expedient — which is what has happened in the past. That would be a disaster. Without a vaccine, there are credible projections that the United States may reach 389,000 deaths by February. A vaccine could help shift that trajectory, but only if early doses go where they can do the most good.

The most urgent questions to make sure the highest-priority groups get the vaccines first are how the initial supplies will be divided among the states, how members of priority groups will be identified and immunized, and how the federal government will oversee and support state and local efforts to distribute the vaccines. The Trump administration has begun planning for some of the daunting hurdles that await vaccine distribution, but its publicly released plans are silent on the resolution of these allocation questions, leaving them to state and localities to resolve.

In the 2009 influenza (H1N1) pandemic, the vaccine was distributed to U.S. states and localities according to their population size, with more populous regions receiving more doses. This approach is simple and politically appealing, but using it again would be a mistake — it would be ethically wrong and also wouldn’t generate the most benefit for the public as a whole.

That’s because the severity of this pandemic, the number of front line health-care workers and the population of people over 65 or with serious comorbidities are not evenly distributed across U.S. states. Nearly one out of five of Floridians is 65 or older, compared with one in nine Alaskans. Similarly, according to the Bureau of Labor Statistics, per capita counts of front line health-care workers (physicians, nurses and respiratory therapists) range from 1 per 100 people in California to 1.58 in Massachusetts. West Virginians have type 2 diabetes at 1.5 times the rate of Coloradans. Half of Alabamians are moderately obese, compared to a third of the residents of Washington, D.C.

With these vast differences, a simple population-based distribution of a covid-19 vaccine to states makes no ethical or practical sense. Doing so would only violate the careful prioritization plans under consideration at the national level.

Once the vaccine is distributed around the country, getting doses into the arms of priority groups will be an extreme logistical challenge, with the possible exception of front line health-care workers, who can be identified and vaccinated at work.

Consider patients with high-risk conditions. How will they be identified, and whose responsibility is it to vaccinate them? Current electronic medical record databases are fragmented and in silos, designed for speedy billing rather than finding high-risk individuals. Relying on groups of specialist physicians or hospitals will miss patients who rely on primary care providers and federally qualified health centers — a population that disproportionately tends to include lower-income individuals and members of minority groups. Will patients with serious comorbidities need to show up at vaccination sites with medications to prove they have diabetes and hypertension? Or will the people tasked with administering scarce early vaccines need to trust patients to honestly self-report having a high-risk medical condition, as happened with the H1N1 vaccine in 2009 — in which case there will be no effective prioritization among high-risk groups? Similar challenges exist with identifying and administering vaccines to other priority groups, from part-time home health aides to the residents of multigenerational homes and homeless shelters.

Ultimately, the difficult work of allocating limited initial supplies of a vaccine will fall on public officials in states, cities, counties and tribal lands. The previous experience with H1N1 does not bode well for fidelity to the prioritization recommendations. A post-mortem federal evaluation found: “State health departments took a variety of approaches in allocating vaccine to local health departments and provider categories … [and] there was considerable local variation regarding eligibility for vaccine, possibly creating confusion among the public.” The result was significant racial and ethnic disparities in the administration of the H1N1 vaccine and, ironically, low vaccination rates among health-care workers.

To ensure the ethical guidelines for allocating scarce covid-19 vaccines are followed, we need to focus more attention now — even before the vaccine is approved — to how we will distribute them. With the primary goal being to reduce premature mortality, the federal government should distribute vaccines to states based on the projected severity of the outbreak over that next month and the best estimates of the number of people in prioritized groups. Anticipating tomorrow’s demands for vaccine doses today will not be simple in a fast-changing pandemic, but those best estimates are certainly more defensible than distributing by raw population size.

State and local officials need federal money, desperately, or none of the plans will work anyway. The CDC estimates states and localities will need at least $6 billion to prepare for the largest vaccination campaign in U.S. history. State-based immunization registries can be leveraged to support equitable and safe vaccine allocation, but their record-keeping and reporting systems are antiquated and need to be upgraded, urgently.

Finally, distributing federal funds and actual vaccines to states and localities should be linked to compliance with detailed federal guidance on the priority groups. States and territories should submit for CDC approval detailed plans on how members of potential priority groups will be identified and administered vaccines, and how the vaccination campaign will be monitored to ensure fair and transparent distribution. Preparing a robust community engagement strategy will be an essential feature of those plans, given that Black and Latino Americans are disproportionately represented among the highest-risk groups and, as a recent poll suggests, they’re also especially wary to get a coronavirus vaccine. These state and local plans should be living documents, modifiable with the support of CDC and adapted as evidence emerges on the best use of these vaccines and population health needs.

The time for preparing for all this is now. In the absence of federal leadership and guidance, U.S. states have taken divergent approaches to many matters in this pandemic, from procuring ventilators to mask mandates to testing to indoor dining in bars and restaurants. The consequences have often been disastrous and often unfair. We can avoid that result when it comes to allocating and distributing vaccines — but only if we work fast.