by revere, cross-posted from Effect Measure
For the first time in medical history we may be seeing an influenza pandemic unfold in real time, but that doesn’t mean we know what we are seeing. There remains some uncertainty about virulence, both in terms of how often it kills and how it kills when it does kill. You’d think both would be easy to determine, but those who have been following along know the problem of how often infection with this virus kills is made almost impossible by not knowing how many people it is infecting. But what about the question of how it kills? There are mainly three scenarios of interest: primary viral pneumonia (the flu virus destroys the deep lung tissue on its own); primary viral pneumonia superimposed by a secondary bacterial infection; death by secondary bacterial invaders with the damage from the flu virus playing little part. In seasonal flu and previous pandemics bacterial infection has played a prominent part in the immediate cause of death. Two small series reporting clinical details of severe cases have been previously reported, 30 hospitalized cases from California and subsequently 10 critical care patients in Michigan. Neither showed evidence of bacterial co-infection. If this virus is acting differently we need to know it.
So what’s the problem? Don’t we have the bodies? Yes and no. First we don’t always know which deaths are flu deaths, for reasons we’ve detailed here many times (here’s one such post). Only a small fraction of deaths are autopsied in the US and death registration and handling are done at the state level. CDC has asked medical examiners and local and state health departments to submit tissue samples from post mortem lung specimens and has just reported on the results from 77 confirmed, but still not representative, fatalities in the May 1 – August 20, 2009 time period. Eight states provided all the cases: California, Hawaii, Illinois, New Jersey, New York, Texas, Utah, and Virginia. What the report showed is different from the results of the two early series and more like what is characteristic of previous pandemics. Bacterial co-infections may be playing a major part in fatalities:
Evidence of concurrent bacterial infection was found in specimens from 22 (29%) of the 77 patients, including 10 caused by Streptococcus pneumoniae (pneumococcus). Duration of illness was available for 17 of the 22 patients; median duration was 6 days (range: 1–25 days). Fourteen of 18 patients for whom information was available sought medical care while ill, and eight (44%) were hospitalized. These findings confirm that bacterial lung infections are occurring among patients with fatal cases of 2009 pandemic influenza A (H1N1) and underscore both the importance of pneumococcal vaccination for persons at increased risk for pneumococcal pneumonia and the need for early recognition of bacterial pneumonia in persons with influenza. (MMWR, CDC)
We learn from this report that only slightly more than half (41) of these lab confirmed fatal cases were diagnosed as swine flu H1N1 before death. Only three quarters had even partial clinical information, so for a quarter of these fatal cases we don’t know much else except they died after being infected with influenza virus. Less than half had preliminary autopsy reports submitted with the clinical specimens. Already you can see that some essential information is missing, even when sections of lung tissue from confirmed cases are available. It is quite possible that some tissue that would show bacterial infection was not included in the sample. It’s hard to be sure this didn’t occur and CDC acknowledges this. The tissue that was submitted was then subjected to a prescribed battery of special tests specially sensitive to detect a handful of bacteria often found in influenza. But CDC also admits that some tissues may have bacterial infections they didn’t test for. The tissue were examined microscopically and then using special assays for S. pneumoniae, Streptococcus pyogenes, Staphylococcus aureus, or Haemophilus influenzae.
As noted, 22 of the 77 cases had evidence of coinfection with one of those four bacteria, most frequently (10 caes) with S. pneumoniae, the organism that the pneumovax vaccine for pneumococcal pneumonia is directed against. The others: six with S. pyogenes, seven with S. aureus, two with Streptococcus mitis, and one with H. influenzae; four cases involved multiple pathogens. Median age of the fatal caes was only 31, the youngest 2 years, the oldest 56. With non pandemic flu, 90% of the fatalities are over age 65. This epidemiology is dramatically different. They were split 50 – 50 male/female.
There was much information unavailable that would have been interesting to know. For example, we only know if medical care was sought for 18 of the 77 patients. It was for 14, meaning that 4 patients died without ever having had any medical care. In my view says something about our health care system, but I have no doubt some of you will blame the victim. We just don’t know and the question is important. Half were sick for 6 days or more, the range being from 1 day to 25 days. So more than half died within a week of illness onset. Of the hospitalized patients, all required ventilators, which isn’t surprising. These were fatally sick people. And of the nine patients for whom information was available, 7 got antibiotics for bacterial infection. So these were people who were treated in intensive care units with modern mechanical ventilation and today’s arsenal of antibiotics and they died anyway. If you think that we don’t have to worry about influenza because we now have modern intensive care units and antibiotics not available in 1918, give this some thought.
These new data, while not representative of all deaths, are giving a different picture than the earlier reports where surprisingly no bacterial co-infections were being found. What’s the explanation? However this might be related to the difficulty in establishing that bacterial infections are involved, which are not detected in many routine tests. CDC reports that less than 10% of patients with clinically diagnosed pneumonia have positive blood cultures. And even with lung tissue, sophisticated methods like PCR and immunohistochemistry has been found to detect infections missed by the usual methods, even after death. When the newer methods are used, as here by CDC, evidence of bacterial infection is being found.
It’s worth emphasizing again that the most common bacterial infection (10/22) was S. pneumoniae, the cause of pneumococcal pneumonia, for which there is a vaccine available (pneumovax). No data was available on whether any of the 22 fatalities had gotten a pneumococcal vaccine, but CDC determined that 16 of the 22 were in categories CDC recommends be vaccinated with pneumococcal vaccine. We have previously discussed this and it is our opinion it is a wise move for anyone over the age of 2. There are different vaccines for children and adults so you have to ask your health care provider. But you should ask. CDC still recommends it only for those between 5 and 64 with certain risk factors, but I fail to see the logic of this (it’s recommended for all children less than 5 and adults older than 64). But even for the recommended group, probably less than 20% have gotten it. Get it, regardless, is our advice.
Is this good news or bad news? First, it has to be confirmed. It’s early news, still. I suppose it could be construed as good news that this isn’t acting like a strange flu virus that kills all by itself with a virulent primary viral pneumonia. But the bad news is that it still kills, and that if you are counting on post-1918 antibiotics and computer controlled mechanical ventilators to save your bacon, you better think again.