New York City’s first COVID-19 wave cannot be understood by counting emergency-room visits, reading billing codes literally, or treating hospital beds like empty seats at a baseball game. The clinical reality was more complicatedand far more devastating.
A Spreadsheet Can Be Accurate and Still Tell the Wrong Story
In November 2022, the Brownstone Institute published an analysis questioning whether New York City hospitals were truly overwhelmed during the spring 2020 COVID-19 wave. Its author pointed to declining emergency-department traffic, relatively low percentages of visits coded as COVID-19, and differences between COVID-like illness admissions and recorded COVID hospitalizations.
At first glance, those observations sound provocative. Total emergency-room visits did fall. Not every patient with respiratory symptoms was admitted. Early coding categories did not line up neatly. The problem is not necessarily that the underlying numbers were invented. The problem is that they were interpreted without sufficient attention to testing limitations, coding changes, patient acuity, hospital workflow, expanded capacity, and the extraordinary death toll occurring at the same time.
Numbers do not interpret themselves. Sadly, spreadsheets have yet to complete an internal medicine residency.
A more complete examination supports the conclusion offered by NYC hospitalists and public-health researchers: New York experienced an enormous wave of severe COVID-19 that placed hospitals, emergency medical services, intensive care units, and frontline workers under extreme pressure.
Flaw One: Using Total ER Visits as a Measure of Hospital Strain
Lower volume does not mean lower workload
The Brownstone analysis emphasized that overall visits to New York City emergency departments dropped sharply in spring 2020. That part is unsurprising. National data showed a 42% reduction in emergency-department visits during an early four-week pandemic period compared with the same weeks in 2019. The decline was particularly pronounced in the Northeast and in the region containing New York and New Jersey.
People stayed home. Schools and businesses closed. Traffic accidents, workplace injuries, sports injuries, and other routine emergencies declined. Some patients used telemedicine, while others delayed care because they feared exposure to the virus.
Yet the categories increasing during this period included COVID-19, pneumonia, respiratory failure, cardiac arrest, and lower respiratory disease. The waiting room could therefore contain fewer people while the treatment areas were packed with patients needing oxygen, continuous monitoring, invasive lines, emergency dialysis, vasopressors, intubation, or prolonged hospitalization.
One critically ill patient may consume more personnel, equipment, medications, oxygen, and bed time than several patients with sprains, headaches, or minor infections. Comparing those visits as though each required an equal amount of care is like comparing a bicycle repair with rebuilding a jet engine because both arrived through the same garage door.
Acuity matters more than the raw head count
Hospital strain is driven by several interacting factors:
- How sick patients are when they arrive
- How many require intensive care or mechanical ventilation
- How long each patient occupies a bed
- Whether trained nurses, respiratory therapists, physicians, and technicians are available
- Whether oxygen, medications, ventilators, dialysis equipment, and protective gear can meet demand
- How quickly patients can be transferred, discharged, or replaced by incoming cases
A hospital can remain below a theoretical bed ceiling and still operate under crisis conditions. Beds do not care for patients. People doand the right people cannot be manufactured overnight.
Flaw Two: Treating Early COVID Billing Codes as Complete Clinical Records
The Brownstone article argued that only a small percentage of spring 2020 emergency-department visitors received a COVID-19 diagnosis. That conclusion relied heavily on encounters coded with U07.1, the specific ICD-10-CM code for COVID-19.
However, U07.1 did not become effective in the United States until April 1, 2020. Using that code as the primary filter for a period beginning in March automatically loses part of the outbreak. Before April, clinicians and coders were instructed to document diagnoses such as viral pneumonia, acute bronchitis, lower respiratory infection, acute respiratory distress syndrome, shortness of breath, fever, or exposure to a communicable disease.
Testing was also scarce during the critical early weeks. Initial eligibility rules focused heavily on travel history or recognized contact with a confirmed patient. Community transmission was already accelerating, but many symptomatic New Yorkers could not obtain timely laboratory confirmation. Some results took days to return.
A patient discharged before receiving a result might have been coded with pneumonia, bronchitis, or an unspecified viral respiratory infection. Searching only for a later COVID-specific code is therefore not a reliable way to count all suspected or actual COVID cases.
Billing data answer billing questions
Administrative codes are useful for surveillance and reimbursement, but they are not substitutes for reviewing clinical notes, imaging, laboratory findings, oxygen requirements, treatment records, and discharge summaries. Coding practices also changed rapidly as testing improved and official guidance evolved.
During a crisis, physicians focused on keeping patients alive. Producing a perfectly polished collection of billing codes was not always the first item on the morning checklistsomewhere behind stabilizing blood pressure and locating another oxygen tank.
Flaw Three: Assuming the COVID-Like Illness Definition Captured Every Serious Case
Another Brownstone argument compared COVID hospitalizations with admissions categorized as COVID-like illness, or CLI. The NYC surveillance definition generally required fever plus cough or shortness of breath.
That definition was specific, but it was not sensitive enough to identify every hospitalized COVID patient. In a large New York-area hospital cohort, only about 31% of patients had a fever at triage. Many arrived with profound oxygen deficiency, weakness, confusion, gastrointestinal symptoms, or other manifestations without meeting a fever-based surveillance definition.
Consequently, a patient could have laboratory-confirmed COVID-19 pneumonia and require hospitalization while failing to qualify as a CLI admission. The difference between those datasets does not automatically prove incidental infection, widespread miscoding, or hospital-acquired transmission. It may simply reveal that the categories measured different things.
This is a classic denominator problem dressed in a lab coat. When definitions are incomplete or change over time, ratios can look dramatic while providing little support for the conclusion attached to them.
Flaw Four: Ignoring Capacity That Was Created During the Emergency
Brownstone also argued that citywide acute-care and ICU occupancy remained below the expanded system’s stated maximum. But expanded capacity was not evidence that hospitals avoided a crisis. It was evidence that they responded to one.
At NewYork-Presbyterian/Weill Cornell Medical Center, researchers reported 109 adult ICU beds before the pandemic. Operating rooms and post-anesthesia care units were converted into expansion ICUs, with 60 additional beds operating at the peak. Anesthesia machines were used as ventilators, and staffing shortages were among the reasons expansion was difficult.
At NewYork-Presbyterian/Columbia, ICU capacity was doubled. An operating suite became an 80-bed intensive care unit, while catheterization laboratories and medical-surgical units were adapted for critical care. Approximately 1,200 Columbia clinical staff members were redeployed, including specialists working outside their normal departments.
These changes were not routine spring cleaning. They required construction, new gas connections, additional monitoring systems, reassigned personnel, canceled procedures, emergency training, and entirely new care teams.
Citywide averages also concealed differences between hospitals. Facilities with large emergency departments in heavily affected neighborhoods experienced far greater pressure than specialty hospitals with limited emergency intake. Combining all beds into one citywide denominator can make capacity appear comfortable even while individual emergency departments and ICUs are struggling.
Moreover, expanding the denominator during a surge creates a misleading comparison. If a hospital doubles ICU capacity and then occupies most of those new beds, claiming it was never overwhelmed because a few emergency spaces remained is statistical sleight of hand.
The Excess-Mortality Evidence Is Difficult to Explain Away
Between March 11 and May 2, 2020, New York City recorded 32,107 deaths. Public-health investigators estimated that 24,172 were above the expected seasonal baseline. That figure included 13,831 laboratory-confirmed COVID-19 deaths and 5,048 probable COVID-19 deaths, plus 5,293 additional excess deaths not classified in either category.
Some of the remaining deaths may have resulted indirectly from delayed medical care, disrupted services, or fear of visiting a hospital. Others may have been unrecognized COVID-19 deaths during a period of limited testing. Exact attribution is difficult, but the scale and timing are incompatible with the suggestion that the event was primarily panic, coding confusion, or ordinary seasonal illness.
New York City’s official 2020 vital-statistics report later placed the COVID-19 mortality rate at 241.3 deaths per 100,000 residentshigher than the city’s reported 1918 influenza mortality rate. Overall life expectancy fell by 4.6 years between 2019 and 2020, with even larger declines among Black and Hispanic New Yorkers.
Cardiac arrests provide another warning signal
Out-of-hospital cardiac arrests requiring resuscitation occurred at roughly three times the 2019 rate during the initial wave. On April 6, 2020, the city recorded 305 such arrests, almost ten times the number on the corresponding date in 2019.
Researchers cautioned that not every arrest could be directly attributed to SARS-CoV-2. Delayed care and disruption of chronic disease management probably contributed. That distinction matters. Nevertheless, both direct infections and indirect healthcare disruption are components of a pandemic’s total impact. Neither supports minimizing what happened.
“With COVID” Versus “From COVID” Is Not Solved by Suspicion
Questions about incidental infections are reasonable, particularly in later periods when widespread transmission and routine screening identified asymptomatic cases. But applying that later-pandemic framework to spring 2020 ignores the intense alignment among respiratory illness, test positivity, hospital expansion, critical-care demand, and excess mortality.
A national CDC review of more than 378,000 death certificates listing COVID-19 found that 94.5% included at least one additional diagnosis. Among certificates with another diagnosis, 97.3% contained a plausible condition in the chain of events leading to death, a significant contributing condition, or both. Common entries included pneumonia, acute respiratory failure, hypertension, and diabetes.
Comorbidities do not mean COVID-19 was irrelevant. A person with diabetes who develops fatal viral pneumonia did not die because a positive test wandered innocently onto the chart. Preexisting conditions increase vulnerability; they do not erase the triggering disease.
What a Better Data Analysis Would Do
A credible assessment of New York City’s first COVID-19 wave should triangulate multiple forms of evidence rather than elevating one administrative dataset above everything else. At minimum, it should examine:
- All-cause and cause-specific excess mortality
- Testing eligibility, availability, turnaround time, and positivity
- Changes in ICD and surveillance definitions
- Patient acuity, oxygen use, ventilation, dialysis, and length of stay
- Hospital-level rather than only citywide capacity
- Temporary ICU beds added during the emergency
- Staffing, medication, equipment, and supply constraints
- EMS call volume and out-of-hospital deaths
- Neighborhood, racial, occupational, and socioeconomic disparities
It should also acknowledge uncertainty. Some deaths were indirect. Some early treatments changed as evidence improved. Some mitigation policies produced serious economic, educational, and psychological costs. Public-health decisions deserve rigorous review.
But criticism of policy is not strengthened by minimizing the disease burden. One can debate school closures, mandates, communication failures, nursing-home policies, or the timing of restrictions without pretending that New York’s hospitals were merely hosting a citywide convention of nervous coughers.
Experience-Based Perspective: What the Hospital Data Cannot Fully Show
The following is a composite reflection informed by published accounts from New York clinicians and hospital operational reports. It is not presented as the author’s personal eyewitness testimony.
A normal hospital morning begins with structure. Teams review overnight events, check laboratory results, plan discharges, and discuss which patients may need a higher level of care. During New York City’s first COVID-19 wave, that structure remained on paper, but almost everything surrounding it changed.
Medical floors became COVID units. Operating rooms became intensive care spaces. Physicians from unrelated specialties refreshed critical-care skills because the usual teams could not absorb the volume. Nurses cared for unfamiliar patient populations while wearing protective equipment for hours. Respiratory therapists moved from ventilator to ventilator, managing machines that had become some of the most valuable objects in the building.
Patients often arrived with frighteningly low oxygen levels. Some were awake and speaking despite severe hypoxemia; others deteriorated rapidly. A patient who looked stable during morning rounds might require emergency intubation that afternoon. Once a patient entered intensive care, the bed could remain occupied for two or three weeks. Admission was not a quick overnight visit followed by a cheerful breakfast tray and a taxi home.
Proning an intubated patient required a coordinated team to turn a sedated person onto the abdomen without dislodging the breathing tube, intravenous lines, or monitoring equipment. Kidney failure could require continuous dialysis. Blood-pressure support required medications that had to be adjusted minute by minute. Every additional patient increased not only the census but the number of hands, devices, and decisions required around the clock.
Families usually could not enter the hospital. Physicians and nurses became the physical connection between patients and loved ones who were waiting beside telephones. Clinical updates carried unusual emotional weight because a relative might never again see the patient awake. Staff held phones near beds, arranged video calls, and delivered news that would ordinarily have been shared in person.
Meanwhile, the emergency department could appear statistically “quieter” because minor visits had vanished. That quiet was deceptive. Behind closed doors were patients needing escalating oxygen, emergency procedures, and rapid placement in an ICU that might not have existed three weeks earlier.
Frontline clinicians also worried about becoming sick, infecting their families, or leaving colleagues short-staffed. Some isolated from children or partners. Others slept in temporary housing. The emotional strain did not replace the data; it explained what the data represented.
This is why experience and quantitative analysis should not be treated as enemies. Personal testimony alone cannot establish population-level causation. Aggregate statistics alone cannot reveal how clinical systems function under pressure. The strongest account comes from combining death records, hospital operations, patient outcomes, and the observations of trained professionals who understood what was happening at the bedside.
When those forms of evidence converge, dismissing the experience as panic or faulty memory becomes increasingly implausible.
Conclusion: Debate the Response, but Do Not Rewrite the Emergency
The Brownstone analysis identified real features of the early pandemic data: emergency-department visits fell, surveillance definitions were imperfect, and administrative datasets did not always align. Those observations could have supported a useful investigation into how crisis data are collected and interpreted.
Instead, the analysis drew sweeping conclusions from incomplete indicators. It treated fewer total ER visits as evidence against overload, relied on a diagnostic code unavailable during part of the study period, assumed a fever-dependent definition captured nearly every serious case, and evaluated occupancy without adequately accounting for emergency capacity expansion or hospital-level variation.
The broader evidence tells a consistent story. New York City experienced an abrupt rise in severe respiratory disease, intensive-care demand, out-of-hospital cardiac arrests, and excess deaths. Hospitals converted operating rooms into ICUs, redeployed thousands of workers, and operated under crisis-care conditions. Communities already facing healthcare and economic inequality suffered the greatest losses.
Historical review is essential. Public-health agencies and medical institutions made mistakes, and future pandemic planning must learn from them. But learning requires skepticism disciplined by contextnot skepticism that discards every inconvenient fact until the spreadsheet finally agrees with the preferred conclusion.
Editorial note: This article evaluates historical public-health claims using established mortality data, hospital reports, clinical research, and documented frontline experiences. It is intended for informational purposes and does not provide personal medical advice.
