Omicron’s scarcest resource was staff, and few hospitals could count them in real time

The week

The news arrived just before the long holiday weekend in the United States. On November 24, 2021, South Africa reported a new variant of SARS-CoV-2, the virus that causes COVID-19, to the World Health Organization. Two days later, on November 26, WHO designated it a variant of concern. That is its label for versions of the virus that may spread more easily, cause more severe illness or weaken the protection of vaccines and treatments. It named the new variant Omicron.

The same day, the United States announced restrictions on travel from eight southern African countries, effective November 29: South Africa, Botswana, Zimbabwe, Namibia, Lesotho, Eswatini, Mozambique and Malawi. Early reports pointed to an unusual number of mutations and rapid spread. Severity and vaccine protection, the two things everyone most wanted to know, were still unknown.

Where hospitals stood that week

Hospitals met the news already stretched. The Delta wave of late summer and autumn had filled intensive care units in many states, and winter was arriving with its usual rise in respiratory illness. Many nurses and other clinicians had already left their jobs or moved to better-paid travel contracts during the pandemic’s first two years.

The workforce was also absorbing a new rule. Earlier that month, on November 4, the Centers for Medicare & Medicaid Services had announced a requirement that staff at most Medicare- and Medicaid-certified providers be vaccinated against COVID-19. Hospitals were collecting vaccination records and processing exemption requests as Omicron emerged. That work added yet another category of staff status to track, held in yet another place.

That requirement was itself in flux during the holiday week. On November 29 and 30, federal judges blocked it in a number of states, and CMS soon paused enforcement while the litigation continued. Hospitals that had built systems to track vaccination status faced the prospect of rules that differed from one state to the next, sometimes inside a single health system.

Federal health officials also widened booster advice. On November 29, the CDC strengthened its recommendation so that all adults were urged to get a booster dose. For hospitals, that meant offering boosters to their own staff, recording them and managing the short absences that side effects sometimes caused, all in the run-up to winter.

Inside hospitals, uncertainty was the hard part. Earlier waves had taught them to plan for patients: beds, ventilators, intensive care. Nobody yet knew whether Omicron would fill those beds the way earlier variants had.

Through December, early reports from South Africa suggested that individual Omicron infections might be less severe than Delta. That would prove to be the case for many patients. It did not ease the pressure on staffing, because a milder illness spreading far faster still sends large numbers of workers home at once.

The weeks that followed

The first US case was identified on December 1, in California, in a traveler who had returned from South Africa. Case counts then rose sharply, and by late December and January hospitals were reporting large numbers of staff unable to work because of infection or exposure. Nurses, doctors, technicians, cooks, cleaners and IT staff were catching the virus or being sent home after contact with someone who had it, often at the same time as patient numbers climbed.

Federal guidance moved to keep pace. On December 23, the CDC updated isolation and quarantine guidance for health care workers, including contingency and crisis options. Isolation is for someone who is infected and quarantine is for someone who has been exposed and might become infected. The contingency and crisis options described when, under staffing shortages, hospitals could bring workers back sooner than normal rules would allow. On December 27, the CDC shortened recommended isolation for the general public.

Two technology shocks landed in the same weeks. On December 9 and 10, the Log4Shell vulnerability became public, sending IT teams everywhere searching for affected systems. On December 11, the workforce software company UKG disclosed a ransomware attack on its Kronos Private Cloud. Employers that used it for timekeeping and scheduling, hospitals and public agencies among them, lost those systems for weeks and fell back on paper and spreadsheets.

Those rule changes and outages mattered, and they raised a practical question every hospital had to answer, daily and sometimes hourly: who, exactly, is out, who is cleared to return, and when? The answer depended less on medicine than on data. In many health systems, the facts needed to answer it sat in four or five different places that did not talk to each other.

Omicron’s scarcest resource turned out to be people, and few hospitals could count them in real time.

Earlier waves had taught health systems to plan around beds, ventilators and intensive care capacity. Omicron behaved differently. It spread fast enough to infect clinicians and support staff in large numbers at the same moment that patient volumes climbed. The question for operations leaders became how many people could work tomorrow, by role and by unit, and how quickly the organization could answer.

Every hard decision that winter, from deferring elective surgery to using contingency return-to-work rules, rested on that answer. The problem had three parts: knowing who was available, deploying them safely, and keeping the work beyond the bedside running. All three depended on records that lived in different systems.

A short tour of where those records live comes first. Each of the three parts then gets its own section. A side-by-side scenario shows the difference on a single morning, and the closing sections cover what the blind spot cost, the plumbing that fixes it and the questions to ask before the next surge.

Where a hospital keeps the facts about its people

Ask a hospital executive how many nurses they employ and you will get a quick answer. Ask how many critical care nurses will be able to work the night shift on Thursday on the fourth floor, and the answer may take hours. The difference lies in how the underlying information is stored.

A typical health system holds facts about its workforce in several separate computer systems, each built for a different purpose.

• A workforce management system handles scheduling and timekeeping. It knows who is supposed to work which shift on which unit, and who clocked in.
• Occupational health is the department responsible for staff health and safety: vaccinations, injuries, exposures and, during the pandemic, staff test results and clearance to work. Its records are confidential medical information about employees.
• Human resources holds employment records, including leave of absence.
• A credentialing system tracks licenses and certifications, confirming that a nurse is licensed and that a physician is approved to perform certain procedures.
• A learning management system records training and competency checks, such as whether a nurse has completed critical care training or has been oriented to a particular unit.

Each of these works perfectly well for its own job. The trouble is that a staffing decision during a surge needs facts from all of them at once. Joining data, in the technical sense, means linking records in different systems that refer to the same person, so you can answer one question using all of them. It is the digital version of laying several lists side by side and drawing lines between the entries that match. To do it reliably, every system has to identify each person the same way, typically with a shared employee number.

Why the identifiers do not match

That sounds simple. In practice, contractors, travel nurses, students and residents often had different identifiers in different systems, or none at all. A join that works for permanent staff can fail, unnoticed, for exactly the people a hospital leans on most in a crisis.

The reasons are structural. Agency and travel staff are usually onboarded through a staffing vendor’s system, not the hospital’s HR system. Residents may be administered by a graduate medical education office. Students arrive through school affiliation agreements. Many physicians are members of the medical staff without being employees at all. Each group enters the hospital through a different door, and each door issues its own identifier.

Mergers add another layer. A health system that has acquired hospitals over the years may run two or three scheduling platforms and several HR systems at once. The same nurse who picks up shifts at two campuses can appear as two people.

The vaccination requirement as a dry run

The CMS vaccination rule forced a census many hospitals had never taken. It applied to staff broadly, including many contractors, students, trainees and volunteers who provide care or services. To comply, a hospital needed a list of everyone working on site, with a status for each.

Building that list exposed the same gaps that would matter for Omicron staffing. People who appeared in no HR system had to be found through vendor contracts, school agreements and badge records. Hospitals that completed this census well in November had a head start on the roster they needed in January.

Why the walls exist

Some of the separation is deliberate. Under the Americans with Disabilities Act, employers must keep employee medical information confidential and separate from general personnel files. Occupational health records fall squarely within that rule. A hospital acting as an employer is in a different legal position from the same hospital treating a patient, and HIPAA excludes employment records from its definition of protected health information.

State privacy laws and union agreements can add further limits. The walls between occupational health and scheduling protect staff, and any design that crosses them has to preserve that protection. Hospitals should confirm the details with employment and privacy counsel.

Knowing who was available

Most health systems kept the relevant facts in separate places. Scheduling and timekeeping lived in a workforce management system. Positive tests and exposures went to occupational health, often through a separate portal or a phone line. Leave of absence went through human resources. Return-to-work clearance followed yet another path.

A nurse calling in sick reached a unit manager, who might note it in the schedule, while the same nurse’s positive test sat in an occupational health record nobody in staffing could see. Multiply that by hundreds of staff and dozens of units, and the picture leaders needed simply did not exist anywhere.

During a fast surge, that fragmentation was a binding constraint. Nursing supervisors and incident command, the temporary leadership structure hospitals set up to run an emergency, needed a daily view. Ideally it was shift-level: who was out, who was cleared to return and when, and which units were short. Systems that produced that view from existing data gave leaders time to act. Systems that required phone trees and spreadsheets consumed the attention of the people best placed to solve the problem.

The fix was rarely new software. It was a join, and it needed three things.

1. A unique employee identifier shared across scheduling, occupational health and HR systems, which many organizations lacked for contractors and students.
2. A simple status model: available, isolating, or cleared to return on a given date.
3. A feed from occupational health into the scheduling tool that carried that status without exposing clinical details.

Occupational health records are confidential, so the feed needed to pass only what staffing needed to know. A scheduler does not need to know why a nurse is out, only that she is out and when she can return. Designing the feed that way respects privacy and also makes it far easier to get approved quickly.

Building the join in weeks

Hospitals that built this connection quickly tended to follow a similar sequence. A small team named one owner from nursing operations, one from occupational health and one from IT. The privacy officer and employment counsel agreed in writing on the exact fields to share. The answer was usually just the identifier, the status code and a date.

The first version was often a simple file, generated by occupational health a few times a day and loaded into the scheduling platform. A direct interface came later, if at all. The file approach was crude, and it worked within days rather than months.

Access needed the same care as the content. Only the people who build schedules and run staffing should see the status, and only for the units they manage. A log of who viewed it supports the confidentiality promise made to staff.

Testing mattered more than usual. A wrong status could send an infectious worker to a unit or keep a cleared one at home. Teams that reconciled the feed against a manual list for the first several days caught identifier mismatches before they caused harm.

Health systems that built this join in December could see their staffing gap by unit a day or two ahead. That lead time decided whether elective cases were deferred in an orderly way or cancelled the morning of surgery. For a patient who had arranged time off work, transport and a caregiver, the difference between those two outcomes is enormous. For the hospital, so is the difference in lost revenue, wasted preparation and damaged trust.

Forecasting the gap as well as counting it

A count of who is out today is useful. A forecast of who will be out on Thursday is more useful, because it gives time to act. The ingredients are already in a joined view: scheduled shifts, current isolations with expected return dates, and the recent rate of new staff infections by unit.

Simple projections were enough. A unit where several staff tested positive yesterday is likely to lose more over the next few days. Showing expected returns alongside expected losses lets a supervisor see whether a gap will close on its own or needs agency staff, redeployment or deferral.

The forecast only works if the inputs are timely. A status that arrives two days late turns a forecast into a history lesson.

Using the emergency settings responsibly

The CDC’s December guidance made this data even more important. Contingency and crisis options are, in effect, emergency settings. They allow a hospital facing real shortages to bring staff back to work sooner than it normally would, accepting some added risk in exchange for keeping care running. Using them responsibly means knowing, unit by unit, how short you actually are, so that the emergency setting is switched on where it is needed and nowhere else. A hospital that cannot see its gap clearly is left with two poor choices: apply the crisis rules too broadly, or hesitate while units go short. Either way, the decision is being made on guesswork about the most important variable.

The guidance framed these options as tiers. Conventional practice applied when staffing was adequate. Contingency options shortened time away from work when shortages were expected. Crisis options applied only when shortages were already severe. Moving between tiers was a documented leadership decision, and a record of the staffing data behind it protects the organization when the decision is later questioned by staff, unions, accreditors or regulators.

The data also fed obligations outside the building. Throughout the pandemic, hospitals reported capacity data to the federal government, including whether they faced a critical staffing shortage. A hospital that knew its numbers precisely could report them accurately and quickly. One that did not was guessing in two places at once.

How it differed by type of organization

The shape of the problem varied. A large academic medical center had more systems, more staff types and more students and residents to track. It also had the IT capacity to build a join quickly. A small community or critical access hospital had fewer systems, but often no one free to connect them, and a single absence could close a service line.

Ambulatory networks faced a different version. Their clinics could close for a day when two medical assistants were out, and their staff moved between sites. For them, the key view was site-level coverage for the next morning, not unit-level shifts.

Public hospitals added civil service rules, union agreements and government HR systems that were often older and harder to change. Any status feed had to respect those agreements, which made early involvement of labor relations staff essential.

The case against more plumbing

A reasonable objection is that surges are rare and experienced supervisors manage them with phones and judgment. Building feeds and identifiers costs money and attention that daily operations also need. Many hospitals got through Omicron without a joined view.

They did get through, and at a cost borne by supervisors working long hours, patients whose procedures were cancelled at the door and staff who received contradictory calls. The same joined view also improves ordinary days: holiday coverage, flu season and planned leave. The investment is small relative to those daily uses, which is why it pays even if the next surge never comes.

There is a human side to this as well. Staff who are isolating want to know when they are expected back. Managers want to know whom to call. A clear, shared status, visible to the right people, spares everyone a round of anxious phone calls at a time when nobody has energy to spare.

Deploying them safely

Moving people to where they were needed raised a different data problem. Float pools are staff who are not tied to a single unit and can be sent where needed. They, along with agency nurses, travelers and clinicians redeployed from deferred services, all had to be matched to units by skill, license, certification and orientation.

Those facts were often held in a credentialing system, a learning management system and a manager’s memory, not in the schedule. A supervisor filling an ICU gap at 6 a.m. could not easily tell which float nurse had current critical care competencies and had oriented to that unit.

Health systems that held competency and orientation data in, or linked to, the scheduling system redeployed faster and more safely. Those that did not either moved people slowly or moved them without full confidence in their readiness, which shifts risk onto patients and onto the staff themselves. A nurse placed on an unfamiliar unit, with unfamiliar equipment, at the end of a long run of extra shifts, is set up to struggle through no fault of her own.

What a usable skills record looks like

The practical form is a skills matrix attached to each person in the scheduling system. It lists the units they are oriented to, the patient populations they are competent to care for, key certifications with expiry dates and any restrictions. A supervisor filtering for “critical care, oriented to unit 4 East, available Tuesday night” should get an answer in seconds.

Keeping it current is the hard part. Expiry dates and new orientations need to flow from the learning and credentialing systems automatically, or the matrix decays within months. Nurse educators, who sign off competencies, are the natural owners of its accuracy.

Redeployment also needed structure for the people being moved. Many hospitals paired redeployed clinicians with a unit’s regular staff in team models, rather than giving them a full independent assignment. That approach works better when the schedule shows who is a regular and who is a redeployed partner.

Protecting the people being moved

Safe deployment also concerns the staff themselves. During a surge, the same reliable people are asked again and again to pick up extra shifts. Without visibility of consecutive shifts and overtime hours, a staffing office can unknowingly lean on a small group until they are exhausted.

Many scheduling systems can flag fatigue thresholds set by the hospital, such as a maximum number of consecutive shifts or hours in a period. Making those flags visible to the person filling a gap, and requiring a supervisor’s approval to override them, protects patients and staff together. Union agreements often set such limits already, which gives the data a contractual purpose as well.

Licenses across state lines

Travel nurses add a licensing layer. A nurse must be licensed in the state where the patient is. The Nurse Licensure Compact lets nurses from member states practice in other member states on one multistate license, but not every state belonged. During the pandemic, many states also issued temporary emergency licensing waivers, each with its own terms and end dates.

A staffing office therefore had to verify, for each traveler, which license or waiver allowed practice and when it expired. Holding that in the credentialing record, visible to schedulers, prevents an assignment that should never have been made.

Contract labor as a supply chain

Contract labor was the other half of deployment. Agency and travel staffing became a major cost, bought through a mix of vendors, managed service programs and direct contracts. A managed service program is an arrangement where one company coordinates many staffing agencies on a hospital’s behalf. Many systems lacked an enterprise view of what they were paying for the same role across hospitals, or how long each contract ran.

Treating labor as a supply chain, with approved vendors, rate cards (agreed prices per role and shift), order tracking and spend reporting by unit, gave finance and nursing leaders a common picture. It also exposed the trade-off between paying premium rates now and investing in internal float pools that would cost less over a longer horizon. A hospital buys gloves and syringes with careful tracking of vendors, prices and usage. During that winter, many discovered they were buying people with far less visibility.

Contract terms deserved the same scrutiny. Crisis rates, minimum guaranteed hours, cancellation notice and the right to extend all shape the real cost. A contract with guaranteed hours costs money even when census drops. A short cancellation window gives flexibility but may lose the nurse to a competitor. Finance and nursing should agree on those terms centrally rather than unit by unit.

The counter-argument deserves a hearing. Contract labor kept many units open that winter, and some leaders saw rate tracking as a distraction during an emergency. The case for visibility is about better buying, not a ban on agency staff. A health system that can see its rates across hospitals can stop paying different prices for the same role in the same week.

Keeping the work beyond the bedside running

The third part was everything outside direct care that also depended on people. Laboratory testing demand rose sharply, both for patients and for staff seeking clearance to return. Environmental services, pharmacy, food service and patient transport ran short at the same time as nursing. Patient access, call centers and portal message pools were flooded with requests for tests, results and appointment changes.

IT and security teams were affected too. Fewer staff meant slower patching, thinner monitoring and longer help desk queues, during the same weeks that Log4Shell demanded urgent work. Log4Shell was a serious flaw in a widely used piece of software that records activity logs, buried inside a great many other products. Organizations everywhere had to find and fix affected systems quickly. Hospitals had to do that with depleted teams.

When the workforce system itself goes down

The UKG Kronos outage made the point sharper. For affected employers, the very system that held schedules and time records was unavailable through the holidays and into January. Supervisors rebuilt schedules from exported files and paper. Payroll teams estimated hours and corrected them later.

The outage showed that the workforce management platform is a critical system. It deserves downtime procedures as careful as those for the electronic health record. That means regular exports of the next few weeks’ schedules, a manual timekeeping process people have rehearsed, and contract terms with the vendor that cover recovery time and data return.

It also showed a dependency many hospitals had not mapped. A cloud workforce platform used by thousands of employers is a concentration risk. A single attack on the vendor took out scheduling for many organizations at once, during a staffing crisis.

Runbooks and cross-training

Some health systems had documented critical runbooks, cross-trained at least two people on each essential platform, and set clear rules for deferring non-urgent projects. They kept operating more smoothly than those relying on a few experts. A runbook is a written, step-by-step guide to running or restoring a system. When the one person who knows how to restart the lab interface is home sick, the runbook is what lets someone else do it.

Deferral rules need to be written in advance. A short list of project categories that stop during a declared surge frees staff for operations without a debate each time. Change freezes on non-essential systems reduce the chance that a thin team breaks something it cannot quickly fix.

Mapping critical roles outside nursing

Every hospital has small teams whose absence stops larger ones. Sterile processing technicians clean and package surgical instruments, and without them operating rooms cannot run. Perfusionists run heart-lung machines. Interface analysts keep lab and pharmacy systems talking to the health record. A handful of absences in any of these groups can halt services far larger than the group itself.

A simple map of these roles, with the minimum safe staffing for each, belongs in the same workforce view as nursing. It lets incident command see that a cardiac surgery list is at risk because of a perfusion gap, not a nursing one.

Digital channels as relief valves

Digital channels helped absorb demand where they were ready. Online scheduling for testing, automated result release and portal messages answered by templates freed phone lines for patients who needed a person. Where those channels were not ready, the phone queue became the bottleneck, and staff pulled to answer phones were staff not available elsewhere. The workforce view had to include every function whose shortage could stop care, well beyond nursing.

Staff testing created a demand of its own. Employees needed a fast way to report results, book tests and learn when they could return. A simple employee portal or hotline that wrote directly into occupational health records reduced both the phone load and the risk of transcription errors.

Two hospitals on the same Monday

Consider two hypothetical community hospitals of similar size on a Monday morning in early January.

At the first, the nursing supervisor arrives at 5:30 a.m. to a stack of voicemails and texts. She opens the schedule, then a separate spreadsheet that occupational health emails each evening, then her own notes from the weekend. Some names appear in one place and not another. Two travel nurses are not in the occupational health list at all because they have no employee number. By 8 a.m. she has a rough picture, but the operating rooms have already started calling in patients for elective procedures. Three are cancelled at the door.

At the second, the supervisor opens a single staffing view. Overnight, occupational health statuses flowed into the schedule as “isolating” or “cleared from” a given date, with no clinical detail. Travel nurses appear because the hospital assigned them identifiers when they started. The view shows that two units will be short on Tuesday and one critical care nurse returns Wednesday. On Sunday afternoon, leaders had already decided to defer a handful of elective cases on Tuesday and called those patients to rebook. A float nurse whose record shows recent ICU orientation is assigned to cover the gap.

Later that morning, the second hospital’s incident command meets. The chief nursing officer shows unit-level shortages for the next three days. The group decides to use contingency return-to-work options on two units only, and records the data behind the decision. The supply chain lead reports agency rates across the system’s hospitals, and the group agrees one rate for the week.

Neither hospital had more staff than the other. One simply knew sooner what it had.

What the blind spot cost

Without a live count of available staff, hospitals bought premium contract labor without knowing its rates or duration. The blind spot touched almost everything a board or executive team cares about.

• Patients and safety. Matching the right skills to the right unit is a patient safety issue, and orderly deferral of procedures is a patient experience issue.
• Money. Premium contract labor bought without visibility of rates and duration is expensive. Cancelled procedures lose revenue and waste preparation.
• Staff. People asked to work in unfamiliar settings without support, or to chase information by phone at dawn, burn out. That cost shows up later in turnover.
• Privacy and compliance. Employee health information is sensitive. A well-designed status feed shares only what is needed. A hastily circulated spreadsheet of who tested positive may share far too much.
• Resilience. The same weakness that hurt during Omicron will hurt during the next surge, whether it is a different infection, a severe weather event or a cyberattack that takes systems offline.

Costs that arrive later

Some costs appeared after the wave. Pay errors from manual timekeeping had to be found and corrected, and wage-and-hour rules on overtime still applied. Staff who felt their status had been mishandled, or their health information shared too widely, raised grievances. Vaccination and exemption records collected in a hurry had to be reconciled when rules changed.

Each of those costs is smaller when the underlying records are joined, minimal and well governed. They are larger when the organization assembled its picture from spreadsheets passed around by email.

Plumbing that pays off in a crisis

Most of the fix is plumbing, and the hospitals that had it in place before the surge saw the benefit.

• Every person who works shifts, including contractors, travelers and students, has one identifier used across scheduling, occupational health and HR.
• A small, agreed set of availability statuses flows automatically from occupational health to scheduling, without clinical detail.
• Competency and unit orientation are visible to whoever assigns float and agency staff.
• Contract labor is managed like any other major purchase, with approved vendors, rate cards and spend reporting by unit and role.
• Every essential IT, laboratory and support platform has at least two trained people and a written runbook.
• The workforce picture covers lab, pharmacy, environmental services, transport, call centers and IT as well as nursing.
• The workforce management platform has tested downtime procedures and recent schedule exports held outside the vendor’s system.

Sequencing the work

The work can be phased. In the first month, a hospital can agree the status codes, issue identifiers to every contractor and student already on site, and stand up a simple file feed from occupational health. In the first quarter, it can connect competency records to scheduling and bring contract labor into one rate card.

Over the first year, it can replace file feeds with interfaces, add forecasting and rehearse downtime procedures for the workforce platform. Each phase delivers value on ordinary days, so the program does not depend on the next emergency to justify it.

Owners and measures

Plumbing decays without owners. The chief nursing officer usually owns the staffing view and its use. Occupational health owns the accuracy of status. HR owns the identifier for employees, and supply chain or a contingent workforce office owns it for contractors. IT owns the feeds and their monitoring.

A few measures show whether the plumbing works. One is the share of shift workers with a matching identifier in every system. Another is the time between a staff member reporting a positive test and the status appearing in the schedule. A third is the share of contract hours bought at the standard rate. Reviewing them quarterly keeps the system ready between surges.

The failures are mirror images.

• Buying a new platform when the real problem is that existing systems do not share an identifier.
• Leaving contractors out of the data, so the people most often redeployed are the hardest to see.
• Over-sharing health information in the rush to get a picture.
• Counting only nurses, and discovering that a shortage of lab technicians or transport staff stops care just as effectively.
• Relying on heroes, the one supervisor or engineer who knows how everything fits together.
• Assuming the scheduling vendor will always be up, with no plan for running staffing on paper.

Seeing the gap before it arrives

• If a third of a given unit called in sick tonight, how quickly would we know, and from which system?
• Do our scheduling, occupational health and HR systems identify people the same way, including contractors and students?
• Can a supervisor see, at 6 a.m., which available staff are trained and oriented for a specific unit?
• For each essential system, who are the two people who can run it, and where is the runbook?
• When we share staff health status for scheduling, what exactly do we share, and who can see it?

Most of these can be answered with a short exercise: pick one unit, one night and one scenario, and trace how the information would reach the supervisor. Where the trace breaks is where the work starts.

Hospitals have long treated beds and supplies as things to be counted in real time. Omicron made the case that people deserve the same treatment, handled with care for their privacy and their wellbeing.

With hindsight

The Omicron wave peaked in the United States in January 2022 and fell quickly, but its effect on hospital staffing lasted much longer. The surge itself lasted weeks. The strain it put on the people who worked through it did not end when case counts dropped.

Policy kept moving

The rules around the workforce kept changing in the months that followed. On January 13, 2022, the Supreme Court allowed the CMS staff vaccination requirement to take effect, while blocking a separate OSHA rule for large employers. Hospitals then had to finish collecting records and resolving exemptions. CMS withdrew the requirement in 2023, and the federal public health emergency ended on May 11, 2023.

The federal government also sent help. In mid-January 2022, the administration announced additional military medical teams to support hospitals in several states. Those teams were welcome and small relative to the national need, which kept the focus on what hospitals could do with their own staff.

The workforce did not snap back

Burnout, turnover and the cost of contract labor remained pressures for many systems long after the wave. Staff who had worked extra shifts, covered unfamiliar units and watched colleagues fall ill carried that experience with them, and some chose to leave. Each departure made the next shortage a little easier to trigger and the next contract a little more necessary.

A 2023 study by the National Council of State Boards of Nursing estimated that about 100,000 registered nurses had left the workforce during the pandemic. Demand and pay rates for travel nurses later fell from their peaks, but many systems carried a higher share of contract and new-graduate staff than before.

States also legislated. New York had already required hospitals to form clinical staffing committees in 2021. Oregon enacted hospital nurse-to-patient ratios in 2023. Rules like these increase the need for accurate, unit-level staffing data, because compliance has to be shown shift by shift.

What hospitals built

Many hospitals responded with workforce management platforms, internal staffing pools and unit-level demand forecasting. Forecasting here means predicting how many staff each unit will need on a given day, based on patterns in admissions and patient needs. Some turned to virtual nursing models to stretch scarce clinical staff. In a virtual nursing model, an experienced nurse works remotely, often by video, handling tasks such as admissions paperwork, patient education and monitoring, so that nurses at the bedside can focus on hands-on care.

The Kronos outage also left a mark. Workforce platforms moved closer to the list of systems that need tested downtime procedures and vendor resilience reviews. Hospitals learned that a staffing crisis and a scheduling outage can arrive in the same week.

By this point in 2026, staff availability data is treated as operational infrastructure at well-run systems, alongside bed management and supply chain. That is a meaningful shift in category. Workforce data used to be thought of mainly as an HR and payroll matter. It is now, at the better-run organizations, something leaders expect to see on the same screens as beds and supplies.

The risk with crisis-built plumbing is that it lapses when incident command stands down. A feed built as a temporary workaround, owned by no one, is easy to switch off and hard to rebuild when the next respiratory season arrives. The hospitals that kept their gains made the feeds part of normal operations, with named owners.

What held up

The finding from that winter holds: in a surge, the limiting resource is often people, and the speed of response depends on how quickly an organization can see where they are. Less durable was the hope that the problem would be solved by buying a platform. Hospitals that bought new scheduling software without fixing identifiers for contractors and students found the same blind spots in the new system.

Outside healthcare, the same is true of any operation that runs on skilled staff in specific places. The time to connect those records is before the surge, not during it.