I’m an advocate of using computer models to help us think about what might or could happen during various pandemic flu scenarios, but it is a technique with drawbacks. For one, it can suggest that some things might be possible that are either very difficult to do or aren’t feasible. This happened in 2005 when some models were published in Science and Nature that suggested a pandemic could be nipped in the bud before it started. Most people thought that what was required was unrealistic but it put WHO in a bind. They had to marshal their resources to show they were willing to try or go down valiantly. These models can also be misunderstood or some results taken out of a very nuanced context for a good headline. That’s what has happened for a really interesting modeling exercise that was just published in the Proceedings of the National Academy of Sciences (PNAS).
Consider this Reuters story:
Pandemic flu plan would put Chicago on lockdown
By Julie Steenhuysen
CHICAGO (Reuters) – Containing an influenza pandemic in a large U.S. city like Chicago would require widespread school closings, quarantines of infected households and bans on public gatherings, U.S. researchers said on Monday.
But, if done quickly and well, such steps could reduce infections by as much as 80 percent, said researcher Stephen Eubank of Virginia Tech in Blacksburg, Virginia, based on a computer simulation of just such an event.
“If you implement it early and people comply, you can save a lot of people. You can make it look a lot more like a seasonal flu than the 1918 pandemic,” said Eubank, referring to a global flu epidemic that started in 1918 and killed between 40 million and 100 million people. (Reuters)
The idea of Chicago in a “lockdown” is preposterous and not even remotely suggested in the story. So what was the story really about? It’s really quite an interesting paper. Three groups used independently developed but very sophisticated computer models to gauge the impact of two very different sorts of pandemic flu interventions: medical interventions, like the use of antivirals (e.g., Tamiflu) either prophylactically for exposed persons or as treatment for those diagnosed with flu; and non-pharmceutical measures, like school closures, prohibiting public events and isolation of cases or quarantining their well family members. Using the same general range of assumptions, the three groups ran simulations and compared their results. The idea was to see how sensitive the outcomes were to the different kinds and implementations of the models for the same event. They tested several thresholds for when to wheel the measures into action, one as low as .01% of the population down with flu, ratcheting it up as high as 10%. If your city had 1,000,000 the low (.01%) end would start the interventions as soon as 100 cases were diagnosed in an outbreak, the high end (10%) wouldn’t kick in until there were 100,000 cases. A 1% threshold would be 10,000 cases in a city of 1 million. In 1918 it is reported that Chicago had more cases than that in a single week in October.
The models are quite involved, with a large number of moving parts. Here’s a sample:
School Closure. All schools, including primary, middle, and high schools, are closed at a particular threshold community cumulative illness attack rate. Once the schools are closed, children are expected to stay at home with a certain compliance rather than to increase community contacts. Compliance is modeled by the reduction in community contacts achieved–assumed to be 30%, 60%, or 90%. In the UW/LANL model, day care centers and small play groups of preschool children are also closed, and the same compliance rates apply. The other two models do not explicitly model day care centers and small play groups.
Liberal Leave Policy. All symptomatic individuals retire to the home from the workplace one day after becoming ill.
Workplace Social Distancing. At a particular threshold community cumulative illness attack rate, workplace contacts are reduced by a certain percent. In the baseline combination scenarios, the workplace contacts are reduced by 50%. Workplaces are not closed. Social distancing in the workplace might eventually be accomplished by staggering the arrivals of workers at work, encouraging people to work at home, or other measures. (Halloran et al., PNAS)
There’s lots more to this paper and if you are interested you will find it fascinating and well within your reach. There are some technicalities but the gist is quite accessible, although it takes concentration to sort it all out. Without any intervention at all, and depending to some extent on the basic reproductive number, R0 (the average number of new cases each infective produces in a susceptible population), the proportion of the population eventually affected ranges from 40% to 60% (the latter with what is probably an unrealistically high R0 = 3.0). One intervention scenario, described as baseline, begins to intervene at a threshold of 1% of the population symptomatic, with 60% of the actual cases being counted. Everyone is treated with antivirals and all household members are prophylaxed. Schools but not workplaces are closed. Workplace absenteeism or transmission reduction is assumed to be 50%. Liberal leave policies are universal, however. If you are sick you stay home. Assuming compliance of a home quarantine for households with a case of flu and children staying home after schools are closed of only 30% (i.e., most people wouldn’t comply) but a 60% compliance with isolation of those who are sick (meaning that 40% of the sick would be out and about to some extent), there is an 80% – 90% reduction at the lower and more likely R0 values (1.9 to 2.1). Even at fairly high R0 levels there is greater than a 50% knockdown of cases.
There didn’t seem to be marked qualitative variation with these three different models. The application of targeted use of antivirals (treat cases and their household contacts), close schools and discourage children from going out into the community and encourage liberal sick leave policies go a surprisingly long way. This is not a mandatory quarantine and isolation order. There is no enforced “lockdown.” On the contrary, it assumes that only 30% of the households with a sick person will not go out and about and only 60% of the sick will stay home. When a person gets sick they will leave work after one day of illness — but only half the time. Other results in the paper show that lowering the threshold to .01% helps but not much. 1% seems like a reasonable threshold.
These are not mild interventions, to be sure, but if the R0 were in the expected region of 2.0, this much would reduce cases by more than 80% — according to these models. But of course they aren’t the real thing. They are scenarios inside a computer. So there is the obligatory disclaimer:
We caution against overinterpretation of the modeling results, even where the three models suggest similar effectiveness of interventions. Because of the uncertainties in the models, the results need to be viewed more as helping to structure thinking about pandemic planning, rather than being predictive of the precise effectiveness of different policies.
Nice to say this, although it is not at all clear exactly what it means. How are these results supposed to help us structure our thinking about pandemic planning? One thing it says to me is that it is conceivable much could be accomplished to slow a pandemic without mandatory and coercive policies. Martial law wouldn’t be needed and in fact might be counter productive as people flee the authorities. Social pressure is much more likely to bring about the desired social distancing than is the National Guard (they are all in Iraq, anyway).
One thing for sure. People will need support if they are going to reduce their contact with each other in the community. There will need to be mechanisms to supply and care for households with sick people in them, adequate social sick leave policies, and some kind of accommodations made for working parents when schools are closed. In other words, we’ll need a robust, working and effective public health and social service infrastructure. Without it, all the authorities will be able to do is preside over chaos. With it, we can conceivably get through some very rough times together.