The use of an automated surveillance algorithm could more accurately determine the beginning of influenza season, according to recent data.
Researchers collected weekly counts of laboratory-confirmed influenza A from the Children’s Hospital of Colorado and Johns Hopkins Hospital’s children’s hospital between 2001 and 2013. Using this data, they developed the Above Local Elevated Respiratory Illness Threshold (ALERT) algorithm, a method to predict and declare the beginning of an influenza outbreak. The system automatically observes collected influenza data weekly, declaring the beginning of influenza season once cases exceed a predetermined cutoff.
“All the extra precautions cost time and money, so you don’t want to declare flu season too early,” Nicholas G. Reich, PhD, biostatistician at the University of Massachusetts Amherst, said in a press release. “For hospitals, there is a strong incentive to define a really clear period as flu season. It does not start the moment you see the first case in the fall. If you begin the full response too early, you set yourself up for a long slog. You want to be as effective and efficient as possible in your preparations and response.”
Nicholas G. Reich
Researchers applied a 90%-of-cases threshold to the 2011-2012 and 2012-2013 influenza seasons for both hospitals. This resulted in a threshold of four cases per week for Children’s Hospital of Colorado, whereas Johns Hopkins Hospital was set to six cases weekly.
During these two seasons, 77% and 89% of cases reported at Children’s Hospital of Colorado fell within the period designated by the ALERT system, respectively. At Johns Hopkins Hospital, these values were 71% and 91%.
Researchers said the strength of the algorithm was its focus on an individual institution’s case rate, as opposed to more widespread surveillance systems such as Google Flu Trends. Additionally, the system’s variable parameters make it applicable to multiple potential scenarios.
“Ultimately, the power of the ALERT algorithm lies in its simplicity, flexibility and generalizability,” the researchers wrote. “However, some technological extensions to the algorithm as it stands now could enhance its utility and empower public health and medical practitioners.”
Disclosure: The researchers report no relevant financial disclosures.