Swine flu: New York City

by revere, cross-posted from Effect Measure

In New York City, an illness termed “mild” for many has killed 7 and put 300 in the hospital. A preliminary analysis of about half of those hospitalized, most (82%) were said to have some underlying medical condition. That’s common with flu, but it’s also a reminder that one of five were otherwise healthy, and unusually for flu, most of them relatively young (mostly under 65). Similarly, the deaths also had underlying medical conditions but were relatively young (median age 43). The two most recent deaths were in the mid 40s.

So not being old is one risk factor. What does “underlying medical condition” mean?

So far, the most common risk factor in New York City has been asthma – an underlying risk factor among 41% of the New Yorkers hospitalized for H1N1 flu. Other important risk factors include being less than 2 years of age (18% of hospitalized patients), having a compromised immune system (13%), having heart disease (12%), or being pregnant. The Health Department recommends that people with asthma, or any of the other conditions listed here, to call a doctor right away if they develop flu-like illness. [snip] Anyone feeling ill can take his or her temperature with a home thermometer. A temperature of 100.4 degrees is a fever. If the fever is accompanied by cough or sore throat, the condition qualifies as flu-like illness. (New York City Department of Health)

While emergency room visits are down from their peak of a week ago, a time when doctor visits for flu were also increased 15 times over what was expected for this time of year, it was clear that flu was still spreading in the city, with visits still higher than normal. Attendance rates in school were still below average although also up from last week. All but 4 of the 50 schools that were closed are now open. In other words, there may be just as much flu out there now as last week, but people are adjusting to it.

So how much flu is “out there”? Not an easy question to answer. One way to get at it would be to do a random sample of New Yorkers and ask them if they’ve had flu-like symptoms. To get an unbiased estimate (i.e., a guess that on average is the same as the true average) would require that any New Yorker have an equal chance of being chosen for the survey. But there is no “master list” of everyone in New York City, so the Health Department used an old but conventional sampling technique called random-digit-dialing for the purpose. In essence this means randomly selecting residential telephone numbers and phoning people to ask them questions. Today, however, this can only be used with land lines (no cell phones) and you have to have a land line phone number to be in the sample, so as a representative sample it’s imperfect. Since there are other causes of “flu-like illness” besides influenza infection, and not all those infected with the influenza virus have classical symptoms (or any symptoms at all), this is another problem. And of course the respondent has to remember they had the symptoms, which if they were mild might have been forgotten or disregarded or remembered now as something different than they were (recall bias). Nonetheless, 1000 people have been questioned this way and the results are being analyzed. With all the problems, it will be interesting to see what they show.

All this goes to show how epidemiology in the real world can be much more difficult, uncertain and complicated than the simple questions it is trying to answer — like how much flu is there in New York City? — would seem.

    Current ye@r *