Monday, May 15, 2006

Clusters and Indonesia

Clusters of bird flu cases are always worrisome. There is no strict definition of a "cluster." Essentially it is a group of cases closer together in space or in time than one would expect to see "by chance." Usually "cluster" is further taken to mean a space-time cluster, i.e., cases grouped together that occur in a specific span of time, the span usually scaled to take account of the serial time of the disease, the time from one case to another when disease transmission is efficient.

But clustering in space and time doesn't necessarily mean the disease is being transmitted person to person. Epidemiologists most often use case clusters to indicate some common source for the disease, such as chemicals in the air or water or a food poisoning outbreak. When considering cancer clusters, the statistical problem of detecting a genuine cluster is not at all straightforward. Imagine for a moment throwing darts at a dart board. Some of the darts will be closer together than others but be a "cluster" only by chance. In addition, disease cases can only occur where there are people. This seemingly obvious and innocent observation makes interpreting disease clusters devilishly difficult. Cancer cases plotted on a map are always clustered, for example, because where people live is clustered. Thus we have to disentangle the clustering of the underlying population from any additional clustering.

Even the more straightforward problem of infectious disease clusters requires interpretation. The relatively large cluster of bird flu cases in Indonesia at the moment is worrisome but not necessarily indicative of a change in the virus. The cluster comprises eight members of one family, six of whom have died. The index case had onset on April 27, a 40 year old woman. The cause of her death has not been confirmed. Five of the other seven cases had dates of onset on May 3 and 4. Date of onset of the other two were not available to us at this time. A nurse with flu like symptoms has also been reported but the diagnosis was not confirmed and indications are she is negative for H5N1. (Some Indonesian MetroTV news links here, here and here for those able to read the language.)

Given these minimal facts, there remain several possibilities. The mother (index case) could have passed on the disease to her seven relatives in an instance of human to human transmission. This is the big worry. Or they could all have been infected by a common source. Sick poultry is mentioned in several news reports (Wall Street Journal, subscription required, alas). The gap between the index case and the other cases might be a difference in timing of two separate exposures or an abnormally short incubation period for the mother (who might have been more heavily exposed) and the rest of the family, or both. Or the dates of onset might be incorrect.

If this is bird to human (or possibly pig to human; there are pigs in this region of Indonesia, which has more Christians than other parts), a cluster this size (the largest recorded to date) could signal more efficient bird to human transmission or an unusual susceptibility of these individuals said to be related by blood. A change in the virus to more efficient bird to human transmission might be signaled by the appearance of more clusters. More efficient human to human transmission might be signaled by the appearance of larger clusters with secondary cases, i.e., cases that seem to come from a connection with a cluster or by disease appearing among health care workers. We'll have to wait and see and hope the Indonesian authorities are paying sufficient attention to this case and have obtained help from international experts.

There is a curious lack of attention in the English language newspapers in Indonesia (e.g., The Jakarta Post). Whether this is attention to the volcano crisis, bird flu fatigue or a combination, I don't know. This is a situation to keep our eyes on.

Update, 5/16/06, 6 am EDST: Dr. Andrew Jeremijenko, an epidemiologist currently in Indonesia, has provided additional information and a translation of some current Indonesian language news articles in the comments. He notes that at least two of the cases in this cluster were hospitalized and then went home. Indonesia continues to have difficulty coping with their bird flu situation and the optimism of Nabarro over the actions of Vietnam and Thailand doesn't seem to apply there.