TY - JOUR AU - Law, Royal K. AU - Burkom, Howard AU - Schier, Josh PY - 2017/05/02 Y2 - 2024/03/28 TI - Improving Detection of Call Clusters through Surveillance of Poison Center Data JF - Online Journal of Public Health Informatics JA - OJPHI VL - 9 IS - 1 SE - Novel algorithms, statistical or mathematical methods DO - 10.5210/ojphi.v9i1.7596 UR - https://ojphi.org/ojs/index.php/ojphi/article/view/7596 SP - AB - <div style="left: 90px; top: 335.758px; font-size: 14.1667px; font-family: sans-serif; transform: scaleX(1.08129);" data-canvas-width="63.778333333333336">Objective</div><div style="left: 105px; top: 350.709px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.958947);" data-canvas-width="379.91883333333357">Our objective was to compare the effectiveness of applying the</div><div style="left: 90px; top: 367.376px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.03218);" data-canvas-width="394.51616666666615">historical limits method (HLM) to poison center (PC) call volumes</div><div style="left: 90px; top: 384.043px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.00076);" data-canvas-width="253.80999999999997">with vs without stratifying by exposure type.</div><div style="left: 90px; top: 415.758px; font-size: 14.1667px; font-family: sans-serif; transform: scaleX(1.11768);" data-canvas-width="82.63416666666666">Introduction</div><div style="left: 105px; top: 430.709px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.03611);" data-canvas-width="379.63125">The Centers for Disease Control and Prevention (CDC) uses the</div><div style="left: 90px; top: 447.376px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.986563);" data-canvas-width="395.71041666666656">National Poison Data System (NPDS) to conduct surveillance of</div><div style="left: 90px; top: 464.043px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.02844);" data-canvas-width="394.6110833333332">calls to United States PCs. PCs provide triage and treatment advice</div><div style="left: 90px; top: 480.709px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.00783);" data-canvas-width="393.8588333333332">for hazardous exposures through a free national hotline. Information</div><div style="left: 90px; top: 497.376px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.987436);" data-canvas-width="395.4185833333332">on demographics, health effects, implicated substance(s), medical</div><div style="left: 90px; top: 514.043px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.00077);" data-canvas-width="321.03083333333325">outcome of the patient, and other variables are collected.</div><div style="left: 105px; top: 530.709px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.98924);" data-canvas-width="378.10408333333334">CDC uses automated algorithms to identify anomalies in both pure</div><div style="left: 90px; top: 547.376px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.01943);" data-canvas-width="394.128">call volume and specific clinical effect volume, and to identify calls</div><div style="left: 90px; top: 564.043px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.956822);" data-canvas-width="394.4340000000001">reporting exposure to high priority agents. Pure and clinical effect</div><div style="left: 90px; top: 580.709px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.999869);" data-canvas-width="393.46500000000003">volume anomalies are identified when an hourly call count exceeds a</div><div style="left: 90px; top: 597.376px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.997649);" data-canvas-width="261.66683333333333">threshold based on historical data using HLM.</div><div style="left: 351.66px; top: 597.751px; font-size: 8.5px; font-family: serif;">1</div><div style="left: 355.91px; top: 597.376px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.997752);" data-canvas-width="127.51274999999998">Clinical toxicologists</div><div style="left: 90px; top: 614.043px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.02282);" data-canvas-width="394.48641666666634">and epidemiologists at the American Association of Poison Control</div><div style="left: 90px; top: 630.709px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.988825);" data-canvas-width="395.8265833333332">Centers and CDC apply standardized criteria to determine if the</div><div style="left: 90px; top: 647.376px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.02838);" data-canvas-width="394.22999999999996">anomaly identifies a potential incident of public health significance</div><div style="left: 90px; top: 664.043px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.982339);" data-canvas-width="395.60416666666663">(IPHS) and to notify the respective health departments and local</div><div style="left: 90px; top: 680.709px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.02822);" data-canvas-width="394.5289166666666">PCs as needed. Discussions with NPDS users and analysis of IPHS</div><div style="left: 90px; top: 697.376px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.966118);" data-canvas-width="395.19050000000016">showed that alerting based on pure call volume yielded excessive</div><div style="left: 90px; top: 714.043px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.987015);" data-canvas-width="393.09525">false positives. A study using a 5-year NPDS call dataset assessed the</div><div style="left: 90px; top: 730.709px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.02833);" data-canvas-width="394.6054166666665">positive predictive value (PPV) of the call volume-based approach.</div><div style="left: 90px; top: 747.376px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.997499);" data-canvas-width="385.6166666666669">This study showed that less than 4% of anomalies were IPHS.</div><div style="left: 475.75px; top: 747.751px; font-size: 8.5px; font-family: serif;">2</div><div style="left: 90px; top: 764.043px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.01497);" data-canvas-width="393.9608333333329">A low PPV can cause unnecessary waste of staff time and resources</div><div style="left: 90px; top: 780.709px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.00095);" data-canvas-width="198.30500000000004">analyzing false positive anomalies.</div><div style="left: 105px; top: 797.376px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.958198);" data-canvas-width="379.87775000000016">As an alternative to pure call volume-based detection where all</div><div style="left: 90px; top: 814.043px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.99032);" data-canvas-width="393.19016666666664">calls to each PC are aggregated for anomaly detection, we considered</div><div style="left: 90px; top: 830.709px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.02941);" data-canvas-width="394.55583333333345">separating calls by toxicologically-relevant exposure categories for</div><div style="left: 90px; top: 847.376px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.995279);" data-canvas-width="393.0853333333334">more targeted anomaly detection. We hypothesized that this stratified</div><div style="left: 90px; top: 864.043px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.00079);" data-canvas-width="302.93999999999994">approach would reduce the number of false positives.</div><div style="left: 90px; top: 895.758px; font-size: 14.1667px; font-family: sans-serif; transform: scaleX(1.07287);" data-canvas-width="58.23916666666666">Methods</div><div style="left: 105px; top: 910.709px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.02619);" data-canvas-width="379.3295">We derived our exposure categories based on the criteria that the</div><div style="left: 90px; top: 927.376px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.964065);" data-canvas-width="395.0899166666667">categories must: 1) relate to hazardous exposures of public health</div><div style="left: 90px; top: 944.043px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.04094);" data-canvas-width="397.2049999999998">importance, 2) reflect categories based on clinical effects and</div><div style="left: 90px; top: 960.709px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.954846);" data-canvas-width="394.65358333333313">treatment modalities, 3) avoid high priority exposures that may be</div><div style="left: 90px; top: 977.376px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.02906);" data-canvas-width="394.43966666666677">triggered by single calls, 4) be compatible with exposure substance</div><div style="left: 90px; top: 994.043px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.998795);" data-canvas-width="393.42533333333336">identification codes currently used by PCs and NPDS, and 5) include</div><div style="left: 90px; top: 1010.71px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.01858);" data-canvas-width="394.1223333333335">enough calls for meaningful tracking. We queried all calls reporting</div><div style="left: 90px; top: 1027.38px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.02212);" data-canvas-width="396.73466666666593">exposures to the proposed categories between January 1, 2009</div><div style="left: 90px; top: 1044.04px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.940039);" data-canvas-width="394.6124999999998">and July 31, 2015 for ten PCs. We applied the HLM method after</div><div style="left: 90px; top: 1060.71px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.03029);" data-canvas-width="394.58558333333303">stratifying by exposure category and tabulated the number of alerts</div><div style="left: 90px; top: 1077.38px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.00837);" data-canvas-width="393.71999999999986">triggered for each category during the study period. We then applied</div><div style="left: 90px; top: 1094.04px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.02844);" data-canvas-width="394.3390833333333">the HLM method for the ten PCs on all combined exposure calls to</div><div style="left: 90px; top: 1110.71px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.970536);" data-canvas-width="394.65216666666663">represent the traditional non-stratified approach. We compared the</div><div style="left: 90px; top: 1127.38px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.999854);" data-canvas-width="393.52591666666666">combined alert burden generated by stratifying by exposure category</div><div style="left: 90px; top: 1144.04px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.00405);" data-canvas-width="393.63641666666666">with the alert burden for the non-stratified approach for varying time</div><div style="left: 90px; top: 1160.71px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.00384);" data-canvas-width="379.2841666666665">windows (1-, 2-, 4-, 8- and 24-hours). We conducted analysis in R.</div><div style="left: 510px; top: 335.758px; font-size: 14.1667px; font-family: sans-serif; transform: scaleX(1.08488);" data-canvas-width="51.17">Results</div><div style="left: 525px; top: 350.709px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.00739);" data-canvas-width="378.69483333333335">We derived a total of 20 exposure categories, including chemicals</div><div style="left: 510px; top: 367.376px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.989699);" data-canvas-width="393.0754166666666">(n=4), drugs of abuse (n=6), pesticides (n=3), gas/fume/vapors (n=2),</div><div style="left: 510px; top: 384.043px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.00378);" data-canvas-width="393.6505833333333">contaminated food/water (n=1), and others (n=4). Call counts during</div><div style="left: 510px; top: 400.709px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.982914);" data-canvas-width="392.96916666666675">2015 for these categories ranged from approximately 5,000 to 90,000.</div><div style="left: 510px; top: 417.376px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.03625);" data-canvas-width="394.24133333333333">Table 1 shows the total number of alerts triggered for each method</div><div style="left: 510px; top: 434.043px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.975089);" data-canvas-width="395.39875">by time windows. There was a marked reduction of alert burden</div><div style="left: 510px; top: 450.709px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.00301);" data-canvas-width="393.56416666666667">when first stratifying by exposure category for time windows shorter</div><div style="left: 510px; top: 467.376px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.03547);" data-canvas-width="394.5841666666665">than eight hours compared to the alert burden for the non-stratified</div><div style="left: 510px; top: 484.043px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.00143);" data-canvas-width="55.4625">approach.</div><div style="left: 510px; top: 515.758px; font-size: 14.1667px; font-family: sans-serif; transform: scaleX(1.10336);" data-canvas-width="85.01416666666667">Conclusions</div><div style="left: 525px; top: 530.709px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.02846);" data-canvas-width="381.8327499999999">Stratification of call volume by exposure category and time</div><div style="left: 510px; top: 547.376px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.05386);" data-canvas-width="390.40216666666663">window suggests potential improvement over traditional non-</div><div style="left: 510px; top: 564.043px; font-size: 14.1667px; font-family: serif; transform: scaleX(0.982646);" data-canvas-width="395.3350000000002">stratified approach by having a lower alert burden. Further work</div><div style="left: 510px; top: 580.709px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.04432);" data-canvas-width="394.57991666666595">should focus on refining the exposure categories, refining the time</div><div style="left: 510px; top: 597.376px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.01618);" data-canvas-width="394.29375000000016">window for surveillance, and assessing other detection performance</div><div style="left: 510px; top: 614.043px; font-size: 14.1667px; font-family: serif; transform: scaleX(1.00693);" data-canvas-width="155.4366666666667">metrics, such as sensitivity.</div><div style="left: 510px; top: 642.194px; font-size: 12.5px; font-family: serif; transform: scaleX(1.00701);" data-canvas-width="386.7249999999996">Table 1: Alert burden comparison for the non-stratified vs stratified approach</div> ER -