For those of you who are engaged in the COVID-19 MS1+3 week, the resources below may be useful.


Internet Book of Critical Care →

  • My top general resource for COVID-19 clinical knowledge
    • Sidenote: IBCC in general is a great resource, especially because it has an accompanying podcast. I highly recommend it for beefing up your critical care knowledgebase.
  • Updated daily (!)


EMRAP Corependium COVID-19 Chapter →

  • Subscription site but I believe this particular content has been made free


PEER REVIEWED LITERATURE (various sub-topics)


COVID-19 Open Research Dataset (CORD-19) →

  • JSON dataset of 30,000+ fulltext COVID-19 research articles
  • Updated daily (or near daily)


Oxford COVID-19 Evidence Service →

  • Curated EBM-based effort to address COVID-related questions
  • Updated daily



Wang, et al. “Detection of SARS-CoV-2 in Different Types of Clinical Specimens.” JAMA. 2020. doi: 10.1001/jama.2020.3786

  • More than a thousand known positive COVID-19 patients were tested using RT-PCR of samples collected via various methods 
    • Nasal swab was only 63% sensitive (but in fairness it was only 8 swabs) 
    • There were ~400 pharyngeal swabs with a sensitivity of just 32%! 
    • This finding (of nasopharyngeal swabs giving false-negatives) is extremely frustrating, even confusing, in light of other findings that certain persons have a very high nasal viral load before they even have symptoms (eg Lancet paper doi: 10.1016/S1473-3099(20)30113-4). Probably all of us have seen news reports of celebrities and athletes who were tested, despite no symptoms, and came back positive.


Tak-Yin Tsang, et al. “Coronavirus-positive Nasopharyngeal Aspirate as Predictor for Severe Acute Respiratory Syndrome Mortality.” 2011. Emerging Infectious Diseases. doi: 10.3201/eid0911.030400

  • Sensitivity of nasopharyngeal swab is 32-50% → “variation in sensitivity makes it difficult for the RT-PCR to be the standard criterion for diagnosis.”


Memish, et al. “Respiratory Tract Samples, Viral Load, and Genome Fraction Yield in Patients With Middle East Respiratory Syndrome.” 2014. Journal of Infectious Diseases. doi: 10.1093/infdis/jiu292

  • Tracheal aspirates yielded significantly higher MERS-CoV loads, compared with nasopharyngeal swab specimens (P = .005) and sputum specimens (P = .0001). 
  • Tracheal aspirates had viral loads similar to those in bronchoalveolar lavage samples (P = .3079). 
  • Bronchoalveolar lavage samples and tracheal aspirates had significantly higher genome fraction than nasopharyngeal swab specimens (P = .0095 and P = .0002, respectively) and sputum samples (P = .0009 and P = .0001, respectively).
  • The genome yield from tracheal aspirates and bronchoalveolar lavage samples were similar (P = .1174).


Corman, et al. “Viral Shedding and Antibody Response in 37 Patients With Middle East Respiratory Syndrome Coronavirus Infection.” 2015. Clinical Infectious Disease. doi: 10.1093/cid/civ951

  • Reportedly similar to Memish study above → I haven’t read this yet, however


Tawfiq & Memish. “Diagnosis of SARS-CoV-2 Infection based on CT scan vs. RT-PCR: Reflecting on Experience from MERS-CoV.” Journal of Hospital Infection. 2020. doi: 10.1016/j.jhin.2020.03.001

  • The currently available RT-PCR kits are variable, offering sensitivities ranging between 45 and 60% [uncited statement; does comport with JAMA study by Wang covered above]
  • “In a study of 336 MERS patients, 89% had a positive result after 1 swab, 96.5% had a positive result after 2 consecutive swabs, and 97.6% had a positive result after 3 swabs.” [ref]
  • “In a study of 51 patients, the positivity rate for a single respiratory swab was 70%, an additional 24% (94% cumulative) after a second test, and an additional 3.9% (98% cumulative) after a third test.” [ref]
    • “CT scan findings compatible with viral pneumonia was seen in 98% of patients.” [ibid]


Xie, et al. “Chest CT for Typical 2019-nCoV Pneumonia: Relationship to Negative RT-PCR Testing.” 2020. Radiology. doi: 10.1148/radiol.2020200343 →

  • Case study of several patients who initially tested RT-PCR negative but has positive CT findings, were treated as presumed positive cases, and days later converted to positive RT-PCR
  • Inclusion criteria of study makes the numbers a likely underestimate of the prevalence of this phenomenon, but nonetheless it demonstrates that a ‘negative’ is not a true negative in a patient for whom there is clinical suspicion. (ie NPV is not great.)


Rothe, et al. “Transmission of 2019-nCoV infection from an asymptomatic contact in Germany.” N Engl J Med. 2020. 

  • I haven’t read this yet, but am archiving it here for later reference if needed


Zou, et al. “SARS-CoV-2 Viral Load in Upper Respiratory Specimens of Infected Patients.” N Engl J Med. 2020. doi: 10.1056/NEJMc2001737

  • Including this in my bibliography as a cautionary point regarding extrapolating from SARS/MERS papers: “Our analysis suggests that the viral nucleic acid shedding pattern of patients infected with SARS-CoV-2 resembles that of patients with influenza and appears different from that seen in patients infected with SARS-CoV.”



The Coronavirus Curve – Numberphile →

  • Probably the best SIR tutorial / explanation I have seen so far
  • Includes links to open source modeling app and the code/model used for video


Simulating an epidemic →

  • Very helpful simulation video from mathematician playing with SIR modeling
  • Although he is not an expert in epi modeling, he walks the viewer through his own discovery process, which is helpful, since he is more able to get the modeling working than most viewers would be (and he is compressing many days of his learning into 20 minutes of video)
  • Includes some interesting revelations, such as: which of the following is more important…
    • One fifth the rate of citizens going to a central congregation area (ie shopping) vs
    • Half the rate of transmissibility (ie handwashing)
      • They end up having equivalent results!
  • “While the disease still exists, as soon as people let up and they go back to their normal lives, if nothing is in place to contain the cases, few though they might be, you’ll just get a second wave.”
    • This comports with my understanding and the statements from experts elsewhere in this document (see: Imperial College paper and MSRI talk) 


COVID-19: The Exponential Power of Now – With Prof. Nicholas Jewell [MSRI] → (Mathematician turned epidemiologist, UC Berkeley & London School of Tropical Hygiene) →

  • Very informative and realistic talk from an expert
  • At some point near the second half he says something equivalent to:
    • Once the first wave breaks and things start getting better, you’re going to see every epidemiologist worth their salt begging people to understand that the second wave will be right around the corner if we don’t plan for it and take mitigating steps.


Imperial College COVID-19 Response Team → “Impact of non-pharmaceutical interventions (NPIs) to reduce COVID- 19 mortality and healthcare demand” March 16, 2020 →

  • Report from Imperial College team (dozens of authors) that estimates 2.2 million US deaths if nothing is done
  • They conclude a huge number of lives can be spared if extreme distancing measures are put in place, but these must be continued indefinitely until a vaccine is available
    • Their models suggest that if you titrate the distancing based on available ICU beds (ie turn off distancing, wait until ICUs fill, then turn back on) you only need to distance 2/3 of the time
    • It seems possible to me that with EXCEPTIONALLY good and freely available testing it may be possible to return huge swaths of the country back to normal(ish) life (with immediate testing and contact tracing for any suspect cases that pop up), rather than the crude “2/3 on, 1/3 off for everybody” model they suggest but that is my own intuition
  • They suggest in no uncertain terms that there will be a second wave as soon as measures are relaxed. See green and orange lines below, which come after distancing is turned off (non-blue background)

Estimating actual COVID 19 cases (novel corona virus infections) in an area based on deaths →

  • Great talk from Sal Khan on the lag between real cases and case testing
  • Uses actual Wuhan data to demonstrate the lag and how this leads to epidemic spread
  • Makes the obvious but important point that if case fatality rate = 1%, then true number of cases is at least 100x the number of fatalities.
    • Consider Benton County (where I live) which has “56 cases” currently and “5 deaths”
    • True number of Benton cases should be at least 500
    • (Is probably much higher because of difference between CFR and IFR, see MSRI talk above)


Gabriel Goh’s SEIR interactive calculator (screenshot below) →



Oxford Mathematician explains SIR disease model for COVID-19 (Coronavirus) [PART 1] →

  • Okay overview of SIR, but his lack of health background gets somewhat in the way, as he continually returns to the transmission coefficient to push handwashing and other NPI, when these are just a part of the equation of how transmission can be reduced


Oxford Mathematician explains SIR Travelling Wave disease model for COVID-19 (Coronavirus) [PART 2] → 

  • See above, although to his credit I believe his field of expertise is in fluid modeling, so this may be more up his alley.



UCSF School of Medicine Grand Rounds — Recorded on March 19, 2020 → “UCSF Experts on the Epidemiology, Science, & Clinical Manifestations of COVID-19, and UCSF Response” →

  • 1hr 40mins grand rounds with a dozen topic experts
  • Very decent overview, nothing groundbreaking (assuming you have read the IBCC chapter)
  • UCSF using just surgical mask precautions for COVID-19 patients
  • One expert paraphrases literature as estimating ~75% sensitivity (!) for RT-PCR of COVID-19
    • This is confirmed by my primary literature search (it may be even worse)


UCSF School of Medicine Grand Rounds — Recorded on March 20, 2020 → “Overview of Covid-19 from the UCSF Infectious Diseases Division” →

  • 20 minute overview that seems to be directed toward medical students (?)


Stanford Medicine Grand Rounds — Recorded on March 25, 2020 → 3/25/2020 Coronavirus (COVID-19) Grand Rounds – Stanford Department of Medicine →

  • Informatic estimation of population prevalence at ~18:30
  • Lag between intervention and peak demand is 12-14 days, leading to a 5-30x increase in demand from time of intervention to peak (and this assumes perfect intervention)
  • Interesting proprietary tool shown at ~24mins time mark
  • “What is the sensitivity of the Stanford test?”
    • “Currently this is unclear”
    • Analytic sensitivity is very good
  • PPE Oven study mentioned [this one] → “don’t try this at home [yet?]”