Person:

Kulldorff, Martin

We measured sensitivity and timeliness of a syndromic surveillance system to detect bioterrorism events. A hypothetical anthrax release was modeled by using zip code population data, mall customer surveys, and membership information from HealthPartners Medical Group, which covers 9% of a metropolitan area population in Minnesota. For each infection level, 1,000 releases were simulated. Timing of increases in use of medical care was based on data from the Sverdlovsk, Russia, anthrax release. Cases from the simulated outbreak were added to actual respiratory visits recorded for those dates in HealthPartners Medical Group data. Analysis was done by using the space-time scan statistic. We evaluated the proportion of attacks detected at different attack rates and timeliness to detection. Timeliness and completeness of detection of events varied by rate of infection. First detection of events ranged from days 3 to 6. Similar modeling may be possible with other surveillance systems and should be a part of their evaluation.

Background: Utilizing highly precise spatial resolutions within disease outbreak detection, such as the patients’ address, is most desirable as this provides the actual residential location of the infected individual(s). However, this level of precision is not always readily available or only available for purchase, and when utilized, increases the risk of exposing protected health information. Aggregating data to less precise scales (e.g., ZIP code or county centroids) may mitigate this risk but at the expense of potentially masking smaller isolated high risk areas. Methods: To experimentally examine the effect of spatial data resolution on space-time cluster detection, we extracted administrative medical claims data for 122500 viral lung episodes occurring during 2007–2010 in Tennessee. We generated 10000 spatial datasets with varying cluster location, size and intensity at the address-level. To represent spatial data aggregation (i.e., reduced resolution), we then created 10000 corresponding datasets both at the ZIP code and county level for a total of 30000 datasets. Using the space-time permutation scan statistic and the SaTScan™ cluster software, we evaluated statistical power, sensitivity and positive predictive values of outbreak detection when using exact address locations compared to ZIP code and county level aggregations. Results: The power to detect disease outbreaks did not largely diminish when using spatially aggregated data compared to more precise address information. However, aggregations negatively impacted the ability to more accurately determine the exact spatial location of the outbreak, particularly in smaller clusters (<800 km2). Conclusions: Spatial aggregations do not necessitate a loss of power or sensitivity; rather, the relationship is more complex and involves simultaneously considering relative risk within the cluster and cluster size. The likelihood of spatially over-estimating outbreaks by including geographical areas outside the actual disease cluster increases with aggregated data.

Background: With the objective of identifying spatial and temporal patterns of enzootic raccoon variant rabies, a spatial scan statistic was utilized to search for significant terrestrial rabies clusters by year in New York State in 1997–2003. Cluster analyses were unadjusted for other factors, adjusted for covariates, and adjusted for covariates and large scale geographic variation (LSGV). Adjustments were intended to identify the unusual aggregations of cases given the expected distribution based on the observed locations. Results: Statistically significant clusters were identified particularly in the Albany, Finger Lakes, and South Hudson areas. The clusters were generally persistent in the Albany area, but demonstrated cyclical changes in rabies activity every few years in the other areas. Cluster adjustments allowed the discussion of possible causes for the high risk raccoon rabies areas identified. Conclusion: This study analyzed raccoon variant rabies spatial and temporal patterns in New York that have not been previously described at a focal (census tract) level. Comparisons across the type of spatial analysis performed with various degrees of adjustment allow consideration of the potential influence of geographical factors for raccoon rabies and possible reasons for the highest risk areas (statistically significant clusters).

Background: Clinical microbiology laboratories worldwide constitute an invaluable resource for monitoring emerging threats and the spread of antimicrobial resistance. We studied the growing number of biochemical tests routinely performed on clinical isolates to explore their value as epidemiological markers. Methodology/Principal Findings Microbiology laboratory results from January 2009 through December 2011 from a 793-bed hospital stored in WHONET were examined. Variables included patient location, collection date, organism, and 47 biochemical and 17 antimicrobial susceptibility test results reported by Vitek 2. To identify biochemical tests that were particularly valuable (stable with repeat testing, but good variability across the species) or problematic (inconsistent results with repeat testing), three types of variance analyses were performed on isolates of K. pneumonia: descriptive analysis of discordant biochemical results in same-day isolates, an average within-patient variance index, and generalized linear mixed model variance component analysis. Results: 4,200 isolates of K. pneumoniae were identified from 2,485 patients, 32% of whom had multiple isolates. The first two variance analyses highlighted SUCT, TyrA, GlyA, and GGT as “nuisance” biochemicals for which discordant within-patient test results impacted a high proportion of patient results, while dTAG had relatively good within-patient stability with good heterogeneity across the species. Variance component analyses confirmed the relative stability of dTAG, and identified additional biochemicals such as PHOS with a large between patient to within patient variance ratio. A reduced subset of biochemicals improved the robustness of strain definition for carbapenem-resistant K. pneumoniae. Surveillance analyses suggest that the reduced biochemical profile could improve the timeliness and specificity of outbreak detection algorithms. Conclusions: The statistical approaches explored can improve the robust recognition of microbial subpopulations with routinely available biochemical test results, of value in the timely detection of outbreak clones and evolutionarily important genetic events.

Background: Drug adverse event (AE) signal detection using the Gamma Poisson Shrinker (GPS) is commonly applied in spontaneous reporting. AE signal detection using large observational health plan databases can expand medication safety surveillance. Methods: Using data from nine health plans, we conducted a pilot study to evaluate the implementation and findings of the GPS approach for two antifungal drugs, terbinafine and itraconazole, and two diabetes drugs, pioglitazone and rosiglitazone. We evaluated 1676 diagnosis codes grouped into 183 different clinical concepts and four levels of granularity. Several signaling thresholds were assessed. GPS results were compared to findings from a companion study using the identical analytic dataset but an alternative statistical method—the tree-based scan statistic (TreeScan). Results: We identified 71 statistical signals across two signaling thresholds and two methods, including closely-related signals of overlapping diagnosis definitions. Initial review found that most signals represented known adverse drug reactions or confounding. About 31% of signals met the highest signaling threshold. Conclusions: The GPS method was successfully applied to observational health plan data in a distributed data environment as a drug safety data mining method. There was substantial concordance between the GPS and TreeScan approaches. Key method implementation decisions relate to defining exposures and outcomes and informed choice of signaling thresholds.

Background: To implement effective control measures, timely outbreak detection is essential. Shigella is the most common cause of bacterial diarrhea in Argentina. Highly resistant clones of Shigella have emerged, and outbreaks have been recognized in closed settings and in whole communities. We hereby report our experience with an evolving, integrated, laboratory-based, near real-time surveillance system operating in six contiguous provinces of Argentina during April 2009 to March 2012. Methodology To detect localized shigellosis outbreaks timely, we used the prospective space-time permutation scan statistic algorithm of SaTScan, embedded in WHONET software. Twenty three laboratories sent updated Shigella data on a weekly basis to the National Reference Laboratory. Cluster detection analysis was performed at several taxonomic levels: for all Shigella spp., for serotypes within species and for antimicrobial resistance phenotypes within species. Shigella isolates associated with statistically significant signals (clusters in time/space with recurrence interval ≥365 days) were subtyped by pulsed field gel electrophoresis (PFGE) using PulseNet protocols. Principal Findings In three years of active surveillance, our system detected 32 statistically significant events, 26 of them identified before hospital staff was aware of any unexpected increase in the number of Shigella isolates. Twenty-six signals were investigated by PFGE, which confirmed a close relationship among the isolates for 22 events (84.6%). Seven events were investigated epidemiologically, which revealed links among the patients. Seventeen events were found at the resistance profile level. The system detected events of public health importance: infrequent resistance profiles, long-lasting and/or re-emergent clusters and events important for their duration or size, which were reported to local public health authorities. Conclusions/Significance: The WHONET-SaTScan system may serve as a model for surveillance and can be applied to other pathogens, implemented by other networks, and scaled up to national and international levels for early detection and control of outbreaks.

Background: Guillain-Barré Syndrome (GBS) can be triggered by gastrointestinal or respiratory infections, including influenza. During the 2009 influenza A (H1N1) pandemic in the United States, monovalent inactivated influenza vaccine (MIV) availability coincided with high rates of wildtype influenza infections. Several prior studies suggested an elevated GBS risk following MIV, but adjustment for antecedent infection was limited. Methods: We identified patients enrolled in health plans participating in the Vaccine Safety Datalink and diagnosed with GBS from July 2009 through June 2011. Medical records of GBS cases with 2009–10 MIV, 2010–11 trivalent inactivated influenza vaccine (TIV), and/or a medically-attended respiratory or gastrointestinal infection in the 1 through 141 days prior to GBS diagnosis were reviewed and classified according to Brighton Collaboration criteria for diagnostic certainty. Using a case-centered design, logistic regression models adjusted for patient-level time-varying sources of confounding, including seasonal vaccinations and infections in GBS cases and population-level controls. Results: Eighteen confirmed GBS cases received vaccination in the 6 weeks preceding onset, among 1.27 million 2009–10 MIV recipients and 2.80 million 2010–11 TIV recipients. Forty-four confirmed GBS cases had infection in the 6 weeks preceding onset, among 3.77 million patients diagnosed with medically-attended infection. The observed-versus-expected odds that 2009–10 MIV/2010–11 TIV was received in the 6 weeks preceding GBS onset was odds ratio = 1.54, 95% confidence interval (CI), 0.59–3.99; risk difference = 0.93 per million doses, 95% CI, −0.71–5.16. The association between GBS and medically-attended infection was: odds ratio = 7.73, 95% CI, 3.60–16.61; risk difference = 11.62 per million infected patients, 95% CI, 4.49–26.94. These findings were consistent in sensitivity analyses using alternative infection definitions and risk intervals for prior vaccination shorter than 6 weeks. Conclusions: After adjusting for antecedent infections, we found no evidence for an elevated GBS risk following 2009–10 MIV/2010–11 TIV influenza vaccines. However, the association between GBS and antecedent infection was strongly elevated.

Background: In disease surveillance, the prospective space-time permutation scan statistic is commonly used for the early detection of disease outbreaks. The scanning window that defines potential clusters of diseases is cylindrical in shape, which does not allow incorporating into the cluster shape potential factors that can contribute to the spread of the disease, such as information about roads, landscape, among others. Furthermore, the cylinder scanning window assumes that the spatial extent of the cluster does not change in time. Alternatively, a dynamic space-time cluster may indicate the potential spread of the disease through time. For instance, the cluster may decrease over time indicating that the spread of the disease is vanishing. Methods: This paper proposes two irregularly shaped space-time permutation scan statistics. The cluster geometry is dynamically created using a graph structure. The graph can be created to include nearest-neighbor structures, geographical adjacency information or any relevant prior information regarding the contagious behavior of the event under surveillance. Results: The new methods are illustrated using influenza cases in three New England states, and compared with the cylindrical version. A simulation study is provided to investigate some properties of the proposed arbitrary cluster detection techniques. Conclusion: We have successfully developed two new space-time permutation scan statistics methods with irregular shapes and improved computational performance. The results demonstrate the potential of these methods to quickly detect disease outbreaks with irregular geometries. Future work aims at performing intensive simulation studies to evaluate the proposed methods using different scenarios, number of cases, and graph structures.

Abstract Purpose To develop the infrastructure to conduct timely active surveillance for safety of influenza vaccines and other medical countermeasures in the Sentinel System (formerly the Mini‐Sentinel Pilot), a Food and Drug Administration‐sponsored national surveillance system that typically relies on data that are mature, settled, and updated quarterly. Methods: Three Data Partners provided their earliest available (“fresh”) cumulative claims data on influenza vaccination and health outcomes 3–4 times on a staggered basis during the 2013–2014 influenza season, collectively producing 10 data updates. We monitored anaphylaxis in the entire population using a cohort design and seizures in children ≤4 years of age using both a self‐controlled risk interval design (primary) and a cohort design (secondary). After each data update, we conducted sequential analysis for inactivated (IIV) and live (LAIV) influenza vaccines using the Maximized Sequential Probability Ratio Test, adjusting for data‐lag. Results: Most of the 10 sequential analyses were conducted within 6 weeks of the last care‐date in the cumulative dataset. A total of 6 682 336 doses of IIV and 782 125 doses of LAIV were captured. The primary analyses did not identify any statistical signals following IIV or LAIV. In secondary analysis, the risk of seizures was higher following concomitant IIV and PCV13 than historically after IIV in 6‐ to 23‐month‐olds (relative risk = 2.7), which requires further investigation. Conclusions: The Sentinel System can implement a sequential analysis system that uses fresh data for medical product safety surveillance. Active surveillance using sequential analysis of fresh data holds promise for detecting clinically significant health risks early. Limitations of employing fresh data for surveillance include cost and the need for careful scrutiny of signals. © 2015 The Authors. Pharmacoepidemiology and Drug Safety Published by John Wiley & Sons Ltd.

Background: Many different test statistics have been proposed to test for spatial clustering. Some of these statistics have been widely used in various applications. In this paper, we use an existing collection of 1,220,000 simulated benchmark data, generated under 51 different clustering models, to compare the statistical power of several disease clustering tests. These tests are Besag-Newell's R, Cuzick-Edwards' k-Nearest Neighbors (k-NN), the spatial scan statistic, Tango's Maximized Excess Events Test (MEET), Swartz' entropy test, Whittemore's test, Moran's I and a modification of Moran's I. Results: Except for Moran's I and Whittemore's test, all other tests have good power for detecting some kind of clustering. The spatial scan statistic is good at detecting localized clusters. Tango's MEET is good at detecting global clustering. With appropriate choice of parameter, Besag-Newell's R and Cuzick-Edwards' k-NN also perform well. Conclusion: The power varies greatly for different test statistics and alternative clustering models. Consideration of the power is important before we decide which test statistic to use.