Info

0 5 10 15 20 25 30 Lag («1 lisysj figure 24.2 Effects of ozone levels of greater than 20 parts per billion in air between 10 AM and 6 PM on future absenteeism.

0 5 10 15 20 25 30 Lag («1 lisysj figure 24.2 Effects of ozone levels of greater than 20 parts per billion in air between 10 AM and 6 PM on future absenteeism.

passive cigarette smoking with more frequent absences due to respiratory illnesses. These researchers' studies conclusively show that absenteeism data, when combined with information on the types of symptoms being experienced, provides a highly precise view into the health of a school-age population.

Other work has looked at the association between school absenteeism and influenza outbreaks. Overall, monitoring systems using school absenteeism appear to have considerable power to detect influenza outbreaks. Lenaway and coauthors described a surveillance system for Boulder Valley School District in which schools, on a weekly basis, reported rates of absences due to illnesses (Lenaway and Ambler, 1995). These rates were compared to the rates of influenza A and B observed in sentinel clinical practices. Over a five-year period, reports of absenteeism increases paralleled (two of five years) or pre-ceeded by one week (three of five years) rates of influenza-like illnesses observed in the sentinel medical care system.

Bescuilides et al. (2005) attempted to replicate this experience in New York City using absenteeism reports from city schools. This was a more difficult environment for monitoring than suburban Boulder, Colorado, with high rates of absenteeism from non-illness-related causes and poor follow-up. Their study noted differences in the quality of elementary school and high school data, with many more unexplained absences in high schools. Absenteeism was slightly higher on Mondays and Fridays and nearly twice as high on days with parent teacher conferences, state exams, or school half-days. Analyses that controlled for temporal variability in absenteeism found statistically increased absenteeism in three of four influenza A outbreaks studied. Satscan statistics applied to elementary school absenteeism data frequently detected clusters of schools with increased absenteeism—up to three clusters per day with a median of five schools in each cluster. Similar findings occurred in high schools. These results suggest that spatial analyses may be over sensitive and at the minimum produce more "hits" in New York City than the Department of Health could investigate.

In Japan, Takahashi et al. (2001) developed an outbreak detection system for elementary schools. This system differs from the prior two in that it did not use attendance estimates. Rather, teachers and nurses use the system to report daily counts of children with influenza-like illness in classes. Counts of these illnesses were well correlated with cases identified by an influenza surveillance system (r=0.85) with good sensitivity and specificity for outbreaks of both influenza A and B. This finding provides further evidence that researchers may need to combine absenteeism data with information on symptoms that cause absenteeism to achieve adequate performance.

Was this article helpful?

0 0

Post a comment