That's the frustrating reality of this, and is why I've been wearing an N95 mask and eye protection for my entire shift for the last few months. It is an alternate reality where nothing works and nothing makes any clinical sense. We've had patients present complaining only of worsening depression requesting psychiatric evaluation, only to crash and get intubated due to what turns out to be COVID a few hours later. A patient with a couple weeks of worsening cough/fever/shortness of breath - and family members living in the same household who tested positive for COVID - who had bilateral multifocal pneumonias, all the right serum lab abnormalities and died after testing negative multiple times as a COVID-negative death.
It turns everything you ever learned about hypoxia and ventilator management on its head. People with oxygen saturations of <80% can be chatting on cellphones. I think that the decreased ventilator usage in NY is at least in part due to behavior change by physicians, as the initial forecasts were likely based on an "early intubation" strategy with a trigger based on oxygen saturation and oxygen requirement. That has significantly changed in the last few weeks, and more people are being given high-flow nasal cannula oxygen or CPAP/BIPAP with a more permissive view of hypoxia and a focus on avoiding excessive fluids or high CPAP pressures. HFNC and CPAP were initially avoided due to increased risk of aerosolization, but that risk can be mitigated by staying in full PPE all the time, and some people can avoid intubation/mechanical ventilation using this strategy.
The nasal swabs are universally awful in terms of false negatives - the Chinese have published data that show ~65% sensitivity (positive test if disease is present) for nasopharyngeal swabs compared to CT and clinical diagnosis. That has been my experience. The tests themselves are great if virus is present at a certain threshold level in the sample - that's how they're validated, and they have to have >95% sensitivity (most are higher than this). The problem is that you have to get a threshold level of virus on the swab to come up with a positive result. You can sample perfectly and still come up with nothing, even in someone critically ill with COVID. It's unclear why this is the case.
I had high hopes for widespread antibody testing to clear things up, (and am personally curious about my status) but thus far, the test characteristics here seem underwhelming as well. Hopefully, a reliable, reproducible test will eventually emerge, but as of now, this is looking more like another confusing mess that doesn't move us forward much, if at all.
We've been locked down here in CA since mid-March. Our rate of growth has significantly decreased, but we're still seeing positives frequently. I have to assume that's due to continued asymptomatic spread, as people have become hypervigilant about avoiding anyone who looks like they could even harbor a thought about being sick. Now the South Koreans are talking about reinfections/reactivation in previously recovered patients. Sigh.
TL;DR version. Squirrely virus is VERY squirrely. We are still feeling our way toward optimal management strategies and have only reasonable theories of underlying pathophysiology of both more severe disease and reasons behind testing challenges. Humbling. Frustrating.