Effect Measure

We still don’t know if we are experiencing a lull in flu or the virus has burned itself out for the season, but it’s as good a time as any to reflect a bit on where we’ve been and where we still need to go. Being otherwise occupied (I’m sure you are sick of hearing about my grant writing obsession but not half as sick as I am about having it!), I’ll start with something relatively straightforward: how CDC did on the epidemiology and surveillance front. Historically this is the agency’s strong suit and so it is expected they would have acquitted themselves well. And pretty much, they did. A lot of good epidemiology got done and the surveillance system more or less worked to provide important information. But this doesn’t mean there isn’t room for improvement.

As for epidemiology, it was an “all hands” effort, but one wonders if the system could have sustained a much more severe event. If there had been a lot of absenteeism and a much bigger demand a lot of work that would have been desperately needed probably would not have gotten done. As it is a lot of non influenza work didn’t get done. One reason for the heavy load on CDC was that the states were not in good shape. State and local public health is where the rubber meets the road in US public health and the road definitely was full of potholes, detours and closures. CDC has a lot of expertise but a thin reserve and there is no way it can be the nation’s health department. So I’ll give them a B, realizing that there wasn’t much they could do to raise their grade. Too much homework and not enough time to study.

The influenza surveillance system was augmented and provided critically important information. But it is a jury rigged system, a mosaic of different sources of information and heavily dependent on state and local health departments (see above). It really needs an overhaul and if Obama’s Electronic Medical Record (EMR) initiative goes forward CDC needs to be at the table. Maybe they are, but somehow I suspect if they are, they aren’t a major player. The EMR is being touted as a major cost saver and rationalizer of medical care, although I am extremely skeptical about those claims. On the other hand it could be an extremely important way to get timely medical information for surveillance purposes, but only if CDC makes sure it is designed to include this need. It’s not just a matter of harvesting information collected for other purposes. The system has to be constructed in such a way that it can be used by CDC. This would include making sure it generates useful information for surveillance and allows it to be gathered in a form that doesn’t compromise patient confidentiality or run afoul of HIPPA or other reasonable privacy concerns. It would be a shame if the information was there but couldn’t be accessed.

As we’ve had too many occasions to remark, however, surveillance in general is always the unsexy poor sister of routine public health. It goes on in the background, can be costly and sometimes feels burdensome and of little value — until you need it. So investing in good health surveillance takes leadership and some fortitude to stick to it in the face of what will certainly seem to be — and will indeed truly be — urgent competition for resources.

Just deciding to surveillance isn’t enough. You have to do it in a smart way but CDC has not always done it intelligently. On occasion they have turned design and implementation of surveillance systems for states over to private contractors who produce rigid, poorly designed, content ignorant and cumbersome systems looking more like research instruments than the kind of simple and flexible systems that CDC’s own surveillance guidelines dictate. That’s CDC’s fault. I’m a researcher, so I’m not against research, but surveillance isn’t research, even if it can be used to do research. Surveillance systems are meant for routine use. A surveillance system needs to be simple and useful and readily implemented by states without costing a fortune and it must be continuously supported, not with start and stop and varying funding. And you can’t demand that they produce visible results immediately. They aren’t that kind of “immediate pay off” affair. That’s just the way it is.

So while the surveillance system sort of worked during this not catastrophic pandemic, I have the feeling that as with the heroic epidemiological effort, we dodged a bullet.

So far. Or until next time.

Comments

  1. #1 Jody Lanard M.D.
    February 5, 2010

    It looks to me like the CDC’s and most states’ surveillance was much better than their messaging about what the surveillance showed over time.

    The January 22 MMWR included data collected from 23-38 states, starting August 30, under the Aggregate Hospitalization and Death Reporting Activity (AHDRA) system, according to interim guidance for reporting influenza-associated hospitalizations and deaths.

    States could report either lab-confirmed hospitalizations and deaths, or hospitalization for pneumonia and influenza syndrome.

    I was mostly interested in the lab-confirmed reports, in order to get a tentative sense of how different age groups were being affected, and to compare this information with HHS and CDC vaccination campaign messages.

    The January 22 MMWR reported age-specific population mortality rate data for the lab-confirmed reports from states as follows:

    Since August 30, cumulative deaths associated with laboratory-confirmed 2009 H1N1 infection per 100,000 population were 0.31 for persons aged 0–4 years, 0.26 for 5–18 years, 0.38 for 19–24 years, 0.60 for 25–49 years, 1.03 for 50–64 years, and 0.65 for ≥65 years. For the period August 30–January 9, the median number of states reporting laboratory-confirmed deaths per week through AHDRA was 34 (range: 23–38).

    Turning those rates into a bar graph shows quite vividly which age group had the highest population mortality.

    Earlier less publicized data showed that by November 14, before vaccine-induced immunity could have played a significant role, the pattern looked much the same.

    The messaging and all the PSA’s all the PSA’s put out by HHS give a very different impression. Peter Sandman and I analyzed this here.

  2. #2 Alex
    February 5, 2010

    About the graph in comment 1. I’m in the 19-24 age group and we’ve been hearing so much about our group having the most deaths, while statistics clearly showed the opposite, that I had stopped taking the CDC, WHO and other organizations seriously. Revere’s posts and the posts of others here always give a better perspective it seems.

  3. #3 Andrew Jeremijenko
    February 6, 2010

    Revere,
    I found this article quite interesting. A young doctor in an ICU ran a test on a pregnant women, got an unusual result and then ran a few more tests on other people with severe disease. Might help us understand why some people especially pregnant women may get more severe disease in H1N1.
    Andrew Jeremijenko

    Clin Infect Dis. 2010 Feb 1. [Epub ahead of print]

    Association between Severe Pandemic 2009 Influenza A (H1N1) Virus Infection and Immunoglobulin G(2) Subclass Deficiency.
    Gordon CL, Johnson PD, Permezel M, Holmes NE, Gutteridge G, McDonald CF, Eisen DP, Stewardson AJ, Edington J, Charles PG, Crinis N, Black MJ, Torresi J, Grayson ML.

    Infectious Diseases, 2Intensive Care, 3Respiratory, and 4Pathology Departments, Austin Health, 5Department of Obstetrics and Gynaecology, Mercy Hospital for Women, 6Victorian Infectious Diseases Service, Royal Melbourne Hospital, University of Melbourne, Melbourne, 7Intensive Care Unit, Bendigo Health, 8Pathology Department, Alfred Health, 9Department of Epidemiology and Preventive Medicine, Monash University, and Department of Medicine, University of Melbourne, Melbourne, Australia.

    Background. Severe pandemic 2009 influenza A virus (H1N1) infection is associated with risk factors that include pregnancy, obesity, and immunosuppression. After identification of immunoglobulin G(2) (IgG(2)) deficiency in 1 severe case, we assessed IgG subclass levels in a cohort of patients with H1N1 infection. Methods. Patient features, including levels of serum IgG and IgG subclasses, were assessed in patients with acute severe H1N1 infection (defined as infection requiring respiratory support in an intensive care unit), patients with moderate H1N1 infection (defined as inpatients not hospitalized in an intensive care unit), and a random sample of healthy pregnant women. Results. Among the 39 patients with H1N1 infection (19 with severe infection, 7 of whom were pregnant; 20 with moderate infection, 2 of whom were pregnant), hypoabuminemia , anemia, and low levels of total IgG , IgG(1) , and IgG(2) (15 of 19 vs 5 of 20; ; mean value +/- standard deviation [SD], g/L vs g/L; were all statistically significantly associated with severe H1N1 infection, but only hypoalbuminemia and low mean IgG(2) levels remained significant after multivariate analysis. Follow-up of 15 (79%) surviving IgG(2)-deficient patients at a mean (+/-SD) of days (R, 38-126) after the initial acute specimen was obtained found that hypoalbuminemia had resolved in most cases, but 11 (73%) of 15 patients remained IgG(2) deficient. Among 17 healthy pregnant control subjects, mildly low IgG(1) and/or IgG(2) levels were noted in 10, but pregnant patients with H1N1 infection had significantly lower levels of IgG(2) Conclusions. Severe H1N1 infection is associated with IgG(2) deficiency, which appears to persist in a majority of patients. Pregnancy-related reductions in IgG(2) level may explain the increased severity of H1N1 infection in some but not all pregnant patients. The role of IgG(2) deficiency in the pathogenesis of H1N1 infection requires further investigation, because it may have therapeutic implications.

  4. #4 revere
    February 6, 2010

    Andrew: I saw this and thought it was interesting, too, but this is cross sectional data so it is impossible to know if the immunoglobulin differences are a cause or a result of the more severe disease (or both are caused by something else). So the bottom line is one I agree with: it is something that should be looked at more closely to see what is going on. My intention was to keep an eye out for further data that might shed some light on this. Glad to hear from you. Hope all is well.

  5. #5 Mom
    February 6, 2010

    As has been pointed out by the Reveres and others, the cases counted as influenza-related deaths likely represent only the tip of the iceberg of the actual deaths caused by influenza. It is true for this year’s H1N1 outbreak, and thus has it always been. My grandfather died of kidney failure a month after getting infected during the 1968 pandemic. He had recovered from his flu illness, but at 86, the rest of his systems gradually shut down until his kidneys quit and he died. His death certificate listed “natural causes” and “kidney failure”. He was not counted as one of the 33800 US fatalities of the pandemic.

    Bottom line is that we don’t have a clue how many people die in a “normal” year, and we don’t have a clue how many people have died this year, and if anybody tells you otherwise, they don’t have a clue, either.

  6. #6 Alex
    February 6, 2010

    @Mom: Well yeah but we do know to a certain extent. I mean you don’t see people dropping like flies all around you from H1N1. As far as I know, since the Black Death, society didn’t need to shut down until a pandemic was over. The H1N1 pales in comparison to the 1918 flu, which itself pales in comparison to the Black Death. BD killed 50-60% of the pop. in Europe only, depending which sources you use. It took about 2 centuries for Europe to recover. Even with all the drama about H1N1, we still have an overpopulation problem. So we do know, within a certain range, how many die and from what, in both pandemic and normal years.

  7. #7 BostonERdoc
    February 6, 2010

    I give the CDC a B+.
    As far as incorporating data in hospital, clinic EMRs for surveillance– I dont think it would work well. We use it (i.e. EMR) at my shop and many times docs dont get to writing there notes, encounters, etc until several days after the patient encounter because of time constraints and pressure to see more patients (the almighty RVU/hr). Writing notes on computer takes forever (12 minutes per patient in my situation with many typos) because of all of the government documentation requirements that have little relevance to the practice of medicine.

    CDC should–and I believe they are begining to –be heavily into social media avenues for surveillance especially in emerging infection hot spots.