Timeliness of Nongovernmental versus Governmental Global Outbreak Communications

To compare the timeliness of nongovernmental and governmental communications of infectious disease outbreaks and evaluate trends for each over time, we investigated the time elapsed from the beginning of an outbreak to public reporting of the event. We found that governmental sources improved the timeliness of public reporting of infectious disease outbreaks during the study period.


The Study
The study database consisted of 398 unique human infectious disease outbreak events collected from Disease Outbreak News, published online by the World Health Organization during 1996-2009 (9). For each outbreak, we defi ned the initial source or sources of the fi rst public communication as the individual, organization, or website that fi rst publicly communicated information regarding the disease threat (locally or internationally, orally or in writing). The corresponding date of communication was identifi ed by using outbreak reports disseminated by ProMED-mail (10). All outbreaks were categorized as having been fi rst communicated by >1 nongovernmental or governmental source, or simultaneously by both types of sources. When an outbreak was simultaneously fi rst communicated by nongovernmental and governmental sources (n = 5), the outbreak was repeated in the dataset and each source was given credit. This adjustment increased the number of outbreak events to 403.
To characterize the timeliness of outbreak communications, for each reporting source of an event, we calculated the median time in days, and bootstrapped 95% CI, from outbreak start to public communication (Table 1). Median reporting times were calculated for the entire study period (1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)), before and after public recognition of severe acute respiratory syndrome (SARS) (March 12, 2003), and for each WHO-defi ned geographic region. The effect of the initial reporting source on the timeliness of outbreak communication was quantifi ed by using negative binomial regression after adjusting for geographic region and whether the outbreak occurred before or after SARS. These variables were included in the model on the basis of a priori assumptions that public health infrastructure can vary by geographic and political region and that new pandemic preparedness strategies, including use of informal information to initiate public health responses, were developed in response to the SARS epidemic (11). Interaction terms between each variable were examined but were not included in the fi nal model because none reached statistical signifi cance (p>0.05). Temporal trends were assessed by using univariate negative binomial regression models, stratifi ed by source category. These models included 1 covariate for the year of outbreak start.
Of all initial outbreak reports identifi ed, 137 were excluded from analysis for ≈1 of the following reasons ( Figure 1): 117 (85%) of the excluded reports were missing information on the estimated outbreak start date; 20 (15%) were not found in the ProMED-mail archives; and 1 (1%) outbreak estimated start date occurred after the date of public communication of the outbreak. Of the 266 (66%) outbreaks included in analysis, 163 (61%) were fi rst publicly communicated by governmental sources, and 103 (39%) were fi rst communicated by nongovernmental sources. Chi-square tests showed no signifi cant differences in the proportions of governmental and nongovernmental sources included in the analysis versus those excluded (p = 0.315).
The median time from estimated outbreak start to initial public communication was 10 days shorter for nongovernmental sources (23 days, 95% CI 20-32) than for governmental sources (33 days, 95% CI 30-45), although this difference was not signifi cant according to the Wilcoxon rank-sum test (p = 0.200) ( Table 1). Additionally, multivariate modeling showed no signifi cant difference after covariates were adjusted for (incidence rate ratio [IRR] 0.95, 95% CI 0.77-1.18) ( Table 2). The effect of missing data was assessed in sensitivity analyses for all outbreaks for which we had an estimated outbreak start date (17 of 20). When we used the WHO Disease Outbreak News communication date, our results did not change when crediting either governmental sources (IRR = 0.88, 95% CI 0.71-1.09) or nongovernmental sources (IRR = 1.086, 95% CI 0.882-1.336).
Examination of temporal trends over the study period ( Figure 2) showed that nongovernmental sources generally communicated outbreak signals to the public faster after 1996, although the trend did not reach statistical signifi cance (IRR = 0.96, 95% CI 0.91-1.01). Governmental sources, in contrast, made signifi cant improvements in lessening the time in which they publicly communicated initial outbreak signals (IRR = 0.94, 95% CI 0.91-0.97).

Conclusions
Our data suggest that, from 1996 through 2006, outbreaks reported initially by nongovernmental sources were communicated publicly an average of 10 days earlier than those reported initially by governmental sources.
Though the differences varied, nongovernmental sources tended to report outbreaks faster than governmental sources when we compared outbreaks before and after SARS, or by WHO-defi ned region. The lack of statistically signifi cant differences in initial communication timeliness by source is probably attributable to a lack of statistical power rather than a lack of effect.
Our results also provide support for the International Health Regulations 2005 revisions that allow WHO to use unoffi cial information to request verifi cation from member states. Slightly more than one-third of all unique infectious disease outbreaks in the WHO Disease Outbreak †categories for exclusion are not mutually exclusive; ‡health offi cials, ministries of health, laboratories, hospitals, etc.; §included in sensitivity analysis; ¶includes nongovernmental organizations, individual accounts, ProMED requests for information, and multiple sources. News during this 14-year period were initially reported by informal information sources. Traditional governmental public health reporting mechanisms remain an integral source for outbreak information, accounting for almost two-thirds of all initial reports over this period. Our results also show that these sources made statistically signifi cant improvements in reporting early warnings of outbreak threats more rapidly to the public, which might result in part from a shift toward automated, electronic methods that improve the timeliness of communication (12,13). It is possible that enhancements in nongovernmental outbreak reporting systems also contributed to improvements in governmental outbreak reporting timeliness over the study period, but we were unable to test this assumption with the current data.
This study has potential limitations. We encountered diffi culty in selecting and consistently applying criteria to determine the initial source of public communication from ProMED-mail reports, which could have resulted in misclassifi cation bias. Although other reporting systems that use informal information exist, they either lack a publicly available archive (for example, Global Public Health Intelligence Network) (14) or their database did not cover the entire study period (for example, HealthMap) (15). According to Heymann, et al., 65% of outbreaks recognized by WHO are fi rst identifi ed by informal sources (4), a proportion we did not fi nd. Some outbreak reports were excluded because of missing data. We were able to internally validate the data that remained, but these exclusions limited the study's statistical power. Finally, use of outbreak reports collected from the WHO Disease Outbreak News might limit the generalizability of our fi ndings to all infectious disease outbreaks. Despite these limitations, our data highlight the value of nongovernmental sources as an integral resource for providing timely information about global infectious disease threats, and demonstrate the signifi cant improvements in the timeliness of outbreak reporting made by governmental sources. Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 18, No. 7, July 2012