Ebola: Protecting us from epidemic data

Day 51: Protecting the public from epidemiological data

It is rather astonishing here how much of the data I and others collect cannot be shared. It is not that the data is about things people don’t know. It is not that the data is about secret findings… it is about keeping the press and academics from saying stupid things and attributing it to WHO.

We have done a series of sensitivity analyses of the major CDC created surveillance system, the VHF, that everyone knows is incomplete. It is consistently 10 or 20% lower than the weekly district reports probably because somehow 10 or 20% of cases fail to be properly documented and transferred on through the steps of the surveillance process (e.g. the case report does not get filled out, or gets lost, or gets missed at the data entry office on the district level….or because half the freakin’ country have one of about six first and last names so they see 17 people named Mohammed Kumara from that district and think that their Mohammed Kumara has already been entered…). But, say out loud with data what everyone knows, for example that most of the cases are not being detected, and wow do the cheap seats in Atlanta and New York start rattling.

The most stunning censorship is the reluctance of anyone to show the epidemic curve. This is because the data takes time to get into the system (both the CDC VHF system and the MoHS “call to the Districts every day” system). It takes days for suspect cases to be tested and confirmed or sent home. It takes days for the data to get entered into the database. There are all kinds of delays… that when presented as an Epi curve, constantly give the impression that the outbreak has peaked and is coming down over the last two weeks.


Above is an example from New York Times of an apparent drop off in cases, but not a true one. It is due to counting errors.

(The original image was missing) This is mostly based on the MoHS data so the dip is not as dramatic as the VHF database. Most every reporter and fancy-pants modeler that has never done contact tracing or data entry form hospital forms will be tempted to report that the outbreak appears to be going down. This is actually what going up looks like. It would be fine if poorly informed people misinterpreted and then moved on to the crossword puzzle, but more often, the press officer and we in the office need to squander time explaining about data flow and delays. And then there are aberrant events.

If one looks at the week of September 21 above, it looks like there was a spike in suspected cases and then a dip. In fact, there was an outbreak of coding errors combined with the largest lab near Port Loko (the district with the highest incidence) shutting down for the week. Thus, those suspected cases from the week of September 21 mostly died without ever being laboratory confirmed and thus they will stay suspect forever. Since back in September, it took 4 or 5 days to get a lab test back (it is much shorter now), and then the data record had to be updated, typically cases that appeared as suspect in the week of Sept. 21 would mostly change to confirmed in the week of Sept. 28, but that never happened that week resulting in an artifact of apparently elevated numbers of suspect cases one week and fewer cases the next. There are issues like this or bigger in every data source I see every day… so people just do not share data and findings… not with the press, not with your peers who might share it with the press, and heaven knows not with Geneva!

Ah, the things they don’t teach you in intro Epi!



Why the Texas case will not cause an ebola outbreak

I’ve scanned about 500 blog posts tagged #ebola since August 2014 and I can say with confidence that there are a lot of paranoid uninformed Americans out there. These fears are founded on misinformation and suspicion of the system designed to protect the public, not on facts. Here are the facts about yesterday’s case:

The system worked as it should

  • The Ebola carrier left Monrovia, Liberia (Sep 19) without symptoms and arrived in Dallas, Texas (Sep 20) without symptoms. His temperature was taken in both airports and did not indicate a fever.
  • Fever is one of the early symptoms of ebola. If a patient has no fever, he cannot spread ebola.
  • The carrier visited a hospital on Sep 24 with symptoms. This hospital was prepared for ebola, had briefed its staff, and had an isolation ward ready. Because the symptoms were run-of-the-mill flu stuff (no hemmoragic fever), he was sent home.
  • His family brought him back on Sunday, Sep 28 and he was admitted. His blood tested positive for the ebola virus. PCR is a very precise test. Only a 99% exact DNA match will test positive.
  • All people who contacted him during the days he was contagious are being monitored for symptoms. It only included his direct household relatives.

Ebola can only be spread through direct contact. Ebola does not spread through the air. So the only people who could be exposed are known and being monitored.

Why USA is different from Liberia

  • Access to healthcare is much better in America. There is one doctor for every 400 Americans, compared to one doctor for every 20,000 Liberians.
  • Every hospital bed in Liberia has been filled for weeks. Sick Liberians are not likely to visit a hospital anymore, and so the virus spreads faster there. For how the number of beds affects models to count the number of cases and predict disease spread, look at this academic paper. “By September 30, 2014, without additional interventions and using the described likelihood of going to an ETU, approximately 670 daily beds in use (1,700 corrected for underreporting) will be needed in Liberia and Sierra Leone (Figure 2).” There are an estimated 2.5 ebola sufferers for every one that got a hospital bed in September, 2014.
  • Obamacare has reversed the trend of making hospitals only available to the rich and the employed. At the moment, every seriously sick American (citizen or not) will go to a hospital for treatment, making quarantine possible. If large segments of society expect to be turned away from hospitals, denied treatment, they will stop coming, and ebola WILL spread in America.

Here is the CSPAN briefing about the Ebola case in Texas from September 30th: