SUPPLEMENT ARTICLES

The Safety Datalink: A Model for Monitoring Safety

AUTHORS: James Baggs, PhD,a Julianne Gee, MPH,a Edwin Lewis, MPH,b Gabrielle Fowler, MPH,a Patti Benson, MPH,c abstract Tracy Lieu, MD, MPH,d Allison Naleway, PhD,e Nicola P. The Vaccine Safety Datalink (VSD) project is a collaborative project Klein, MD,b Roger Baxter, MD,b Edward Belongia, MD,f Jason Glanz, PhD,g Simon J. Hambidge, MD, PhD,g,h between the Centers for Disease Control and Prevention and 8 man- Steven J. Jacobsen, MD, PhD,i Lisa Jackson, MD, MPH,c aged care organizations (MCOs) in the United States. Established in Jim Nordin, MD, MPH,j and Eric Weintraub, MPHa 1990 to conduct postmarketing evaluations of vaccine safety, the aImmunization Safety Office, Centers for Disease Control and project has created an infrastructure that allows for high-quality re- Prevention, Atlanta, Georgia; bKaiser Permanente Vaccine Study search and surveillance. The 8 participating MCOs comprise a large Center, Oakland, California; cGroup Health Center for Health Studies, Seattle, Washington; dDepartment of Population population of 8.8 million members annually (3% of the US population), Medicine, Harvard Medical School and Harvard Pilgrim Health which enables researchers to conduct studies that assess adverse Care Institute, Boston, Massachusetts; eKaiser Permanente events after immunization. Each MCO prepares computerized data files f Northwest, Portland, Oregon; Marshfield Clinic Research by using a standardized data dictionary containing demographic and Foundation, Marshfield, Wisconsin; gKaiser Permanente Institute for Health Research, Denver, Colorado; hDenver Health medical information on its members, such as age and gender, health Community Health Services, Denver, Colorado; iKaiser plan enrollment, , hospitalizations, outpatient clinic visits, Permanente Southern California, Pasadena, California; and emergency department visits, urgent care visits, and mortality data, as jHealthPartners Research Foundation, Minneapolis, Minnesota well as additional birth information (eg, birth weight) when available. KEY WORDS vaccine safety, immunization, Vaccine Safety Datalink, Other information sources, such as medical chart review, member postmarketing evaluation, surveillance surveys, and pharmacy, laboratory, and radiology data, are often used ABBREVIATIONS in VSD studies to validate outcomes and data. Since 2000, AEFI—adverse event(s) following immunization the VSD has undergone significant changes including an increase in CDC—Centers for Disease Control and Prevention the number of participating MCOs and enrolled population, changes in VSD—Vaccine Safety Datalink MCO—managed care organization data-collection procedures, the creation of near real-time data files, VAERS— Reporting System and the development of near real-time postmarketing surveillance for DDM—distributed-data model newly licensed or changes in vaccine recommendations. Rec- DDF—dynamic data file RCA—rapid cycle analysis ognized as an important resource in vaccine safety, the VSD is working TIV—trivalent inactivated influenza vaccine toward increasing transparency through data-sharing and external OR—odds ratio input. With its recent enhancements, the VSD provides scientific exper- CI—confidence interval tise, continues to develop innovative approaches for vaccine-safety GBS—Guillain-Barré syndrome RR—rate ratio research, and may serve as a model for other patient safety collabor- The findings and conclusions in this article are those of the ative research projects. Pediatrics 2011;127:S45–S53 authors and do not necessarily represent the views of the Centers for Disease Control and Prevention. www.pediatrics.org/cgi/doi/10.1542/peds.2010-1722H doi:10.1542/peds.2010-1722H Accepted for publication Nov 29, 2010 Address correspondence to James Baggs, PhD, Centers for Disease Control and Prevention, 1600 Clifton Rd, Mail Stop D25, Atlanta, GA 30333. E-mail: [email protected] (Continued on last page)

Downloaded from www.aappublications.org/news by guest on SeptemberPEDIATRICS 25, Volume2021 127, Supplement 1, May 2011 S45 High vaccination coverage has signifi- cine safety in the United States. The person within the VSD is assigned a cantly reduced vaccine-preventable VSD is part of the CDC Immunization unique, randomized VSD study identifi- disease morbidity and mortality world- Safety Office. The Immunization Safety cation number that is not linked to wide, especially among children.1,2 Office also includes the Vaccine Ad- their MCO member identification num- Vaccines are generally regarded as verse Event Reporting System (VAERS), ber. VSD study identification numbers safe and effective; however, serious the Clinical Immunization Safety As- can be used to link data on demo- adverse events following immuniza- sessment Network, and the Brighton graphics and medical services. A stan- tion (AEFI) can occur.3,4 Although vac- Collaboration. dardized data dictionary, which en- cine safety is rigorously assessed dur- In this article we provide a review of sures data consistency across sites, is ing prelicensing clinical trials, sample the VSD and focus on the important updated each year by the CDC and the sizes are not adequate to detect rare modifications and enhancements the VSD MCOs. Frequently, medical record adverse events, long-term adverse project has undertaken since 2001. data and, occasionally, participant events are not examined, and popula- survey or interview data are used to tions are not heterogenous.3 In recent THE VSD PROJECT SINCE 2001 validate clinical diagnosis and vaccina- years, public concerns about the Population tion data. safety of vaccines have grown signifi- The VSD population has increased sub- cantly.5 Immunization-safety programs Distributed Data Model stantially since 2001, when 4 new MCOs are an important component of main- From 1991 through 2000, the VSD used joined the project to provide data on taining the public trust in our national a centralized-data model, which re- members younger than 18 years: Kai- immunization program.5 Close monitor- quired each MCO to send its cycle files ser Permanente of Colorado (Denver, ing of vaccine safety also protects the to the CDC annually for merging and CO); Marshfield Clinic Research Foun- public’s health and contributes to safer analyses. When data were needed for a dation (Marshfield, WI); Health Part- vaccines and vaccination practices. specific VSD study, the CDC would send ners Research Foundation (Minneapo- In 1990, the Centers for Disease Con- lis, MN); and Harvard Pilgrim/Harvard a subset of cycle data to the MCO re- trol and Prevention (CDC) National Im- Vanguard (Boston, MA). In 2007, the sponsible for performing the study munization Program created the Vac- VSD population expanded again as all analysis. Because of heightened confi- cine Safety Datalink (VSD) project to participating sites except 2 began pro- dentiality concerns, the centralized- conduct postmarketing evaluations of viding data on members of all ages. data model was replaced by a more vaccine safety.6,7 Initially, the project Currently, data for Ͼ18 million per- secure distributed-data model (DDM) used medical event and demographic sons spanning 16 years are available in 2001 (see Fig 1). information from ϳ6 million children for VSD research. The DDM allows each MCO to assemble younger than 6 years for VSD research and maintain its computerized data from 4 participating managed care or- Data Sources, Collection, and files on a secure server at the site ganizations (MCOs): Group Health Co- Confidentiality rather than transferring data to the operative of Puget Sound (GHC) (Seat- Since inception of the VSD, each MCO CDC,10 and ownership of the data is re- tle, WA); Kaiser Permanente Northwest has prepared annual data files, called tained by the MCOs. Data required for (NWK) (Portland, OR); Kaiser Perma- cycle files, that contain member infor- specific VSD studies are transferred nente of Northern California (KPNC) mation obtained from administrative between the CDC and the VSD sites by (Oakland, CA); and Kaiser Permanente files maintained by the individual using 2 secure methods known as the of Southern California (Los Angeles, MCOs. The cycle files include demo- “indirect” and “direct” methods. With CA).6–9 Later, vaccine-safety studies graphic and medical services informa- the indirect method, CDC and MCO were conducted to include children tion on their members, such as age computers share information through younger than 18 years at all 4 sites and and gender, health plan enrollment, a secure server known as the “hub.” adults 18 years of age and older at vaccinations, hospitalizations, outpa- CDC researchers send computer pro- GHC, KPNC, and NWK. In 2001, 4 addi- tient clinic visits, emergency depart- grams to the hub, which are retrieved tional MCOs joined the VSD, and during ment visits, urgent care visits, mortal- at specified intervals by an MCO com- the last 7 years the VSD has made ity data, and additional birth puter. All computer programs are writ- other changes to enhance its ability to information (eg, birth weight) when ten for a statistical computer program serve as the primary mechanism for available. To ensure confidentiality and (SAS [SAS Institute, Inc, Cary, NC]); data population-based evaluations of vac- comply with federal regulations, each files are also stored by using this plat-

S46 BAGGS et al Downloaded from www.aappublications.org/news by guest on September 25, 2021 SUPPLEMENT ARTICLES

TABLE 1 VSD Strategic Priorities

CDC Evaluate the safety of newly licensed vaccines Evaluate the safety of new vaccine recommendations for existing vaccines Direct Evaluate clinical disorders after SAS programs, logs, output, Assess vaccine safety in special populations at and analytical high risk data sets Hub Develop and evaluate methodologies for vaccine- safety assessment

Indirect bers from several of the participating MCOs and the CDC, develops a compre- hensive study proposal that is pre- sented and reviewed by the members FIGURE 1 of the project. This scientific proposal Representation of the VSD DDM. Sites using the indirect method retrieve SAS programs at regular intervals includes a detailed description of the from the hub. Output is then sent back to the hub where it is retrieved by analysts at the CDC. For sites that hypothesis in question, study design, use the direct method, SAS programs and output are transferred through SAS Connect directly between the sites and the CDC. Both methods use encryption methods to secure data transfers. and analytical plan along with a rigor- ous review of the medical outcomes to be evaluated. All studies meet the nec- form. Each MCO can retrieve and use are updated weekly with new data essary institutional review board and its own SAS programs but cannot ac- from each MCO, although some files Health Insurance Portability and Ac- cess programs or information that be- are updated monthly or quarterly, de- countability Act (HIPAA) requirements. longs to other MCOs. Each MCO sends pending on the capabilities of each Computerized data often supple- its SAS logs, output, and analytical data site’s data systems. Using the DDM, mented with additional medical record subsets back to the hub for retrieval files are accessed by the CDC on an or other data are used to create study- by CDC researchers. With the direct ongoing basis for analysis and/or ex- specific analytical data files that con- method, CDC researchers submit SAS traction of necessary data for each on- tain only an extremely small portion of programs interactively through a se- going study. The DDFs use the same the entire VSD data. Analyses and cure SAS remote session by using SAS standardized data dictionary as the cy- manuscript preparation are typically Connect, an Internet communication cle files and continue to provide the led by a single MCO or the CDC with protocol. Four SAS macros, which are flexibility to conduct various types of input from participating investigators. a collection of SAS program state- studies. Additional tracking and data- ments that can be easily recalled, are quality measures were developed to Study Designs used to facilitate access of the data monitor the new DDFs. The sample size The VSD uses several analytical meth- and retrieval of SAS logs and output. All of the DDF continuously increases as odologies to evaluate vaccine safety. data transfers are conducted securely new data are added to the files, which The calculation of background rates al- by using encrypted methods. date back to 1991. The combination of lows the VSD to conduct multiple types the DDM and DDFs enable the VSD to of observational studies and to calcu- Dynamic Data Files conduct near real-time postlicensure late disease incidence and vaccination Development of the DDM as a secure surveillance, enhance the timeliness of coverage. Early VSD studies typically data-transfer system enabled the VSD certain studies, and increase effi- used observational study designs such to restructure the way data files are ciency in the creation of cycle files. as retrospective cohort studies or collected and used, which led to the case-control studies. Several recent creation of dynamic data files (DDFs) Research and Surveillance VSD studies have used self-control in 2005. DDFs permit the ongoing cap- Process case-series designs, which involve an ture of near real-time event-based The VSD continues to conduct numer- analysis based on a person’s exposure MCO administrative data, including ous studies on a wide range of and control time windows, because ap- data on vaccination, hospitalizations, immunization-safety topics. Table 1 propriate independent control groups emergency department visits, clinical lists the specific strategic priorities may be unavailable or subject to con- visits, MCO enrollment, and certain de- for the VSD. For each study, a team of founding.11,12 Because a large percent- mographic characteristics. Most files VSD investigators, comprised of mem- age of the VSD population is vacci-

Downloaded from www.aappublications.org/news by guest on SeptemberPEDIATRICS 25, Volume2021 127, Supplement 1, May 2011 S47 nated, it is difficult to obtain data on ground rates from the VSD or pub- data sets for analysis by using unvaccinated persons to be used as a lished literature, background rates VSD data through December 31, comparison or control group. Self- from other data sources, concurrent 2000. This program is administered control case-series designs allow re- comparison control groups, or self- through the National Center for searchers to determine if the rate of control methods. Because the compar- Health Statistics of the CDC. More in- AEFI is elevated in the hypothesized ex- isons are conducted weekly, the esti- formation on this program can be posure window compared with the mates are adjusted by sequential found at www.cdc.gov/vaccinesafety/ other time windows. Additional new methods to control for repeated anal- Activities/VSD/Datasharing.html. 13,14 case-only methods are being devel- ysis of the data. When conducting On 2 occasions, the VSD has invited exter- oped and refined to better control for RCA, researchers are on alert for a nal experts from a number of scientific seasonal differences in the uptake of “signal,” which is generated if the ex- disciplines and community members to vaccines relevant to studies that in- pected rate of adverse events is signif- contribute to the design, implementa- volve influenza vaccine and vaccines icantly greater than the control rate tion, and presentation of high-priority given seasonally in advance of school when adjusting for sequential meth- studies. These studies include the study or college attendance. ods and other factors. A VSD data- entitled “Early Thimerosal Exposure and coordinating center was created to Neuropsychological Outcomes at 7 to 10 Postmarketing Surveillance: Rapid handle the increased demand for data Years” and the ongoing thimerosal and management and analysis generated Cycle Analysis autism case-control study.19 A public-use Since 2005, several new vaccines have by these RCA studies. data set from the thimerosal and neuro- been licensed: live, oral, human-bovine Postmarketing Surveillance: VSD developmental outcomes study is avail- reassortant (Ro- Collaboration With the VAERS able at www.cdc.gov/vaccinesafety/ taTeq [Merck & Co, Inc, Whitehouse Concerns/Thimerosal/neuropsychological_ VSD data are also used in conjunction Station, NJ]), the tetravalent meningo- Outcomes.html. with information from VAERS to exam- coccal polysaccharide-protein conju- ine AEFI. VAERS is a passive surveil- gate vaccine (Menactra [Sanofi Pas- EXAMPLE STUDIES lance system that receives adverse- teur, Inc, Swiftwater, PA]), 2 tetanus, event reports from various sources, The VSD conducts rigorous epidemio- diphtheria, and acellular pertussis including vaccine manufacturers, logic studies primarily on a wide range vaccines (Boostrix [GlaxoSmithKline health care providers, immunization of vaccine-safety priorities as well as Biologicals, Rixensart, Belgium] and programs, and vaccine recipients.15 other immunization-related topics, Adacel [Sanofi Pasteur, Toronto, On- Possible associations are examined by including vaccine coverage, disease tario, Canada]), the human papilloma- comparing the number of adverse incidence, research methodology, virus vaccine ( [Merck & Co, events reported to VAERS with back- cost-effectiveness, and medical infor- Inc), and a combination measles- ground rates for these events from matics. Through its innovative ap- -rubella and VSD data.16–18 Summaries of 2 such proaches, the VSD has proven its abil- (ProQuad [Merck & Co, Inc]). Using analyses for the newly licensed ity to adapt and respond to the DDFs, the VSD is conducting near real- RotaTeq and Menactra vaccines are increasingly complex and controver- time postmarketing surveillance for included in this article to illustrate sial topics in immunization research. these 6 newly licensed vaccines. The this collaboration (see “Postmarketing Below are a few examples that illus- studies, referred to as the rapid cycle Monitoring Supplementing VAERS Data trate the capabilities of the VSD. 13,14 projects, are an ac- analysis (RCA) With VSD Background Rates”). tive surveillance system that routinely and Risk of assesses possible associations be- Data-Sharing and Oversight Autoimmune Thyroid Disease20 tween vaccines and predefined poten- Recognized as an important resource A possible link between hepatitis B vac- tial adverse events. in vaccine safety, the VSD is working cine and autoimmune thyroid diseases In the RCA, the observed number of toward increasing transparency. In such as Graves’ disease and Hash- suspected adverse events is compared 2002, the VSD established a data- imoto thyroiditis had been suggested with the expected number of events. sharing program that allows external by a study conducted in Europe and by The expected number of events may be researchers to analyze data sets from reports to the VAERS. Supplementing determined from a variety of sources VSD studies published after August patient interviews and medical record including previously determined back- 2002 or to create novel analytical data with routinely collected automated

S48 BAGGS et al Downloaded from www.aappublications.org/news by guest on September 25, 2021 SUPPLEMENT ARTICLES data, the VSD was able to investigate this attended events were found. Further parent interviews. Information on po- relationship through a multisite case- analysis including chart review and a tential confounding factors was ob- control study.20 Cases were initially iden- subanalysis of 28 820 children with no tained from the interviews with moth- tified through VSD cycle data and vali- underlying medical conditions that ers and from medical records. dated through medical record review, would put them at increased risk of com- Standardized tests that assessed 42 and then telephone interviews were con- plications of influenza vaccination re- neuropsychological outcomes, includ- ducted to verify hepatitis B vaccination vealed that children vaccinated with TIV ing speech and language measures, status. The study analyzed 355 cases of were not at increased risk of gastritis/ verbal memory, fine motor coordina- Graves’ disease, 418 cases of Hashimoto duodenitis compared with the entire tion, tics, and behavioral regulation, thyroiditis, and 1102 frequency-matched study population.21 This study supported were administered to the children. controls and revealed that having ever the Advisory Committee on Immunization Only 5% of the statistical tests (19 of received hepatitis B vaccine did not in- Practices vaccination recommendation 378) showed significant associations; crease the risk of either Graves’ disease by providing reassurance to support the 12 tests revealed a positive associa- or Hashimoto thyroiditis.20 This study’s safety of universally immunizing all chil- tion, and 7 tests revealed a negative results reveal the ability to collect com- dren aged 6 to 23 months with influenza association. The vast majority of the prehensive vaccine information and to vaccination. tests revealed no association, and the accurately identify and confirm cases 5% that showed associations could be through alternative data-collection Early Thimerosal Exposure and explained by chance alone. The results methods. Neuropsychological Outcomes at 7 added to accumulated evidence that 19 to 10 Years thimerosal does not cause neuropsy- Safety of Trivalent Inactivated In 1999, the Public Health Service and chological deficits in children.19 This Influenza Vaccine in Children Aged the American Academy of Pediatrics study’s results further show the ability 21 6 to 23 Months called on vaccine manufacturers to of the VSD to supplement administra- As a result of the decision by the CDC’s remove thimerosal, a mercury- tive data with data from other sources Advisory Committee on Immunization containing preservative, from vac- to conduct rigorous studies and test Practices to recommend routine vacci- cines. The decision to remove thimero- vaccine-safety hypotheses. nation of all children aged 6 to 23 sal was a precautionary measure,22 months with trivalent inactivated influ- and subsequent studies have found Postmarketing Monitoring enza vaccine (TIV) in 2004, the VSD no significant association between Supplementing VAERS Data With proved to be well suited to assess the thimerosal and neuropsychological VSD Background Rates safety of TIV in this population. In one of deficits.19,23–31 In an effort to better im- The VSD has collaborated with VAERS the largest population-based TIV stud- prove on previous studies, the VSD was in conducting postmarketing surveil- ies to date, the VSD conducted a retro- able to rigorously assess the relation- lance of potential AEFI for newly spective cohort study of 45 356 chil- ship between thimerosal exposure licensed vaccines. The potential asso- dren who received a total of 69 359 and neuropsychological functioning ciation between RotaTeq and intussus- influenza vaccinations between Janu- through a retrospective cohort study ception was assessed by comparing ary 1, 1991, and May 31, 2003.21 Self- with extensive assessments and inter- the number of VAERS intussusception control case-series methods were views among the study population and reports to the number of intussuscep- used for this analysis. Cycle files were their mothers. The study population in- tion cases expected to occur by chance analyzed to identify medically attended cluded 1047 children between the ages alone.16,32 To determine the expected events seen in clinic, emergency de- of 7 and 10 years who were enrolled in number of cases that would occur by partment, or hospital settings after 4 of the VSD MCOs and had received chance alone, the VSD first determined vaccination with TIV. Preliminary anal- vaccinations as infants when thimero- the background rates of natural intus- yses revealed that gastritis/duodenitis sal was used as a preservative in many susception (International Classifica- was more likely to occur in the 14 days childhood vaccines. For the analysis, tion of Diseases, Ninth Revision code: after TIV (matched odds ratios [ORs]: the children were grouped according 560.0) by using VSD data files from 5.50 [95% confidence interval (CI): to their level of mercury exposure 2000 to 2004, when no rotavirus vac- 1.22–24.81] for control period 1 [0–3 (low, medium, and high), which was cine was in use. The analysis was days] and 4.33 [95% CI: 1.23–15.21] for determined from VSD MCO immuniza- stratified into 3 age groups, because control period 2 [1–14 days]). No other tion records, medical records, per- background rates of natural intussus- significant associations with medically sonal immunization records, and ception and the number of doses ad-

Downloaded from www.aappublications.org/news by guest on SeptemberPEDIATRICS 25, Volume2021 127, Supplement 1, May 2011 S49 ministered varied substantially ac- cur by chance alone. Only 2 of the in- POWER CONSIDERATIONS cording to age. Using these data, the tussusception cases were confirmed With the addition of 4 new MCOS in expected number of background cases by medical chart review. This analysis 2001, the size of the VSD cohort in- was calculated by multiplying VSD suggested that there was no evidence creased and, with it, the statistical background rates for each age group that RotaTeq vaccine is associated power available to detect rare vaccine- according to the estimated number of with an increased risk for intussus- associated adverse events. Two exam- vaccine doses administered to that ception or other prespecified ples, which illustrate the capacity of 32–34 age group. The findings suggested that events. Between March 2005 and the VSD to detect rare events, follow. there is no association between Rota- September 2008, Ͼ570 000 Menactra The intussusception background rate Teq vaccination and intussusception, doses were delivered in participating of infants 6 to 35 weeks old, the age because the number of cases of intus- MCOs, and no cases of GBS after med- during which RotaTeq vaccine is given, susception reported to the VAERS (32) ical record review were observed is 32.4 per 100 000 person-years, as es- was not elevated above the expected among vaccine recipients aged 11 to timated from VSD data. The number of number of cases (52).16,32 Similar anal- 19 years within 6 weeks of vaccination infants aged 6 to 35 weeks in the VSD yses were conducted to compare the (0.9 cases would be expected during cohort is ϳ95 000. At 90% vaccination potential association between Menac- that period). During the same period, 5 coverage and 80% power, it would re- tra and Guillain-Barré syndrome unconfirmed cases of GBS were identi- quire 4.3 years to detect an RR equiva- (GBS). For the study time period, the fied among an unvaccinated compari- lent to 2.0, 1.4 years to detect an RR of background incidence rate of GBS was son group of Ͼ900 000 persons aged 3.0, and 0.7 years to detect an RR of 4.0. determined to be 0.11 per 100 000 11 to 19 years (Dr Lieu, update on Men- person-months in the VSD. By dividing actra RCA through September 2008, However, the VSD has limited capabil- the VAERS reporting rate of GBS (0.20 personal communication, October 20, ity to detect very rare AEFI in minimal per 100 000 person-months) by the 2008 and refs 14, 17, and 18). The re- time periods for minimal risk ratios. background incidence rate of GBS sults of neither study suggest an as- For example, the background rate of (0.11 per 100 000 person-months), the sociation of serious AEFI with these GBS among 11- to 19-year-olds is 1.3 to VSD determined the reporting rate ra- vaccines. In February 2008, VSD in- 1.4 per 100 000 person-years (unpub- tio (RR) to be 1.77 (95% CI: vestigators presented the Advisory lished data). The average monthly co- 0.96–3.07).17,18 Although the data sug- Committee on Immunization Practices hort of 11- to 19-year-olds in the VSD gest a possible small increased risk of with preliminary results from an anal- cohort is between 870 000 and GBS in persons aged 11 to 19 who re- ysis performed after a possible signal 1 000 000 (average monthly 11- to 19- ceived Menactra vaccination, the find- of seizure was observed in the VSD year-old cohort). At 70% vaccination ings should be viewed with caution.17,18 ProQuad RCA study.35 On the basis of coverage and 80% power, it would re- Ͼ43 000 administered doses of the quire ϳ13 years to detect an RR of 2.0, Postmarketing Monitoring Using combination measles-mumps-rubella 4 years to detect an RR of 3, and 2 years VSD RCA and varicella vaccine (MMRV), the at- to detect an RR of 4. In addition to providing background tributable risk for seizures on days 7 CONCLUSIONS rates to supplement the VAERS, the to 10 after MMRV was calculated as 1 VSD is also conducting RCA studies to per 2000 doses compared with the The VSD has conducted important monitor the safety of several vaccines measles-mumps-rubella and varicella vaccine-safety research since its in- including RotaTeq, Menactra, Gardasil, vaccines administered separately but ception in 1990, and since 2001 it has Adacel and Boostrix, ProQuad, and sea- at the same visit.35 The VSD continues made changes that have enhanced its sonal influenza vaccination. Between to investigate this association and con- ability to answer urgent questions May 2006 and May 2008, Ͼ205 000 tinues to monitor the potential risk of about vaccine safety and other doses of RotaTeq were administered GBS after Menactra. The VSD is now in immunization-related issues and to in- orally to infants at ages 2, 4, and 6 the process of developing RCA studies form US . months in VSD-monitored MCOs. Only 5 for Kinrix (GlaxoSmithKline Biologicals, The ability of the VSD to adapt to a cases of intussusception within 30 Rixensart, Belgium), Pentacel (Sanofi changing environment is exemplified days of vaccination were reported Pasteur Ltd, Toronto, Canada), Rotarix by the development of the DDM, the among RotaTeq recipients; in contrast, (GlaxoSmithKline Biologicals, Rixen- creation of DDFs, and the implementa- on the basis of historical background sart, Belgium), and pandemic H1N1 in- tion of RCA projects. The DDM, a sys- rates, 6.75 cases were expected to oc- fluenza vaccination. tem for securely and quickly trans-

S50 BAGGS et al Downloaded from www.aappublications.org/news by guest on September 25, 2021 SUPPLEMENT ARTICLES ferring data within the VSD, was devel- critical for the strategic priorities of Health Insurance Plans, funded by oped in response to heightened confi- the VSD (listed in Table 1), further stud- the CDC. dentiality concerns. The DDM enabled ies that compare population charac- We thank all the organizations and the development of the DDFs. Together, teristics and disease incidence rates their staff for participating in the VSD the DDM and DDFs enabled the imple- to additional non-MCO sources char- project, including the following key mentation of RCA, which is allowing re- acterizing the national population members: Rich Platt, MD, MS, Kather- searchers to monitor events in near could be beneficial for VSD studies, ine Yih, PhD, Virginia Hinrichsen, MPH, real-time after the introduction of new especially those that describe esti- Rich Fox, MPH, Renny Li, MS, RuiHua vaccines and new vaccine recommen- mates of vaccine coverage and dis- Yin, MS, Sharon Greene, PhD, and Irene dations. The VSD continues to refine ease incidence. Shui, MPH (Harvard Pilgrim/Harvard previously used methodologies and The VSD is one of several worldwide Vanguard), Leslie Kuckler and Amy Bu- develop new methodologies, such as research programs that are able to tani (Health Partners Research Foun- the maximized sequential probability conduct vaccine-safety studies by us- dation), Jim Donahue, PhD, DVM, Jer- ratio test (maxSPRT), which is applica- ing large databases. Other established emy McCauley, MBA, and Stephanie ble to the VSD’s RCA.13,14,36 With the re- programs include the US Defense Med- cent expansion of the VSD, the project Irving, MS (Marshfield Clinic Re- ical Surveillance System,57–59 the UK is well positioned to conduct RCA stud- search Foundation), Christina General Practice Research Database,60 ies and traditional VSD analyses on Clarke, Matthew Daley, MD, Dave Mc- postlicensure of child- new vaccines being introduced for Clure, PhD, Jo Ann Shoup, MS, and hood vaccinations in Denmark,61 and teenagers and young adults. Although Stan Xu, PhD (Kaiser Permanente the Vietnam Vaccine Data Link.62 the majority of VSD studies focus on of Colorado), Y. Craig Cheetham, The VSD provides scientific expertise, hypotheses related to vaccine safety, PharmD, Wansu Chen, MS, Marlene continues to develop innovative ap- the VSD has also demonstrated its Lugg, DrPH, Mike Marcy, MD, and Lina proaches for vaccine safety, is the pri- ability to conduct studies on vaccine Sy, MPH (Kaiser Permanente of mary mechanism for population- coverage,37–41 disease incidence,42–49 Southern California), Paula Ray, MPH based evaluations of vaccine safety in methodology,14,36,50 cost-effectiveness,51–53 (Kaiser Permanente of Northern Cal- the United States, and may serve as a and medical informatics.10,54 ifornia), Rachel Gold, PhD, MPH, John model for other patient-safety collab- Mullooly, PhD, Lois Drew, BA, and As described previously, the comput- orative research projects. erized vaccination records have been Karen Riedlinger, MPH (Kaiser Per- shown to be of high quality,55 whereas Information about the VSD project manente Northwest), Jennifer Nel- the accuracy of computerized records can be obtained at www.cdc.gov/ son, PhD, Darren Malais, and Onchee to determine medical outcomes var- vaccinesafety/Activities/VSD.html. Yu, MS (Group Health Cooperative), ies.56 The VSD recognizes this limita- and Karen Broder, MD, MPH, Sophia tion and uses additional data sources, ACKNOWLEDGMENTS Greer, MPH, Paul Gargiullo, PhD, John especially medical chart review, to val- This study was supported, in part, by Iskander, MD, MPH, and William idate administrative data. Although not the VSD contract with America’s Thompson, PhD (CDC). REFERENCES

1. Roush SW, Murphy TV. Historical compari- Safety Datalink Team. Risk of anaphylaxis 8. Chen RT, Glasser JW, Rhodes PH, et al. sons of morbidity and mortality for vaccine- after vaccination of children and adoles- Vaccine Safety Datalink project: a new preventable diseases in the United States. cents. Pediatrics. 2003;112(4):815–820 tool for improving vaccine safety moni- toring in the United States. The Vaccine JAMA. 2007;298(18):2155–2163 5. Cooper LZ, Larson HJ, Katz SL. Protecting Safety Datalink Team. Pediatrics. 1997; 2. Centers for Disease Control and Prevention. public trust in immunization. Pediatrics. 99(6):765–773 Impact of vaccines universally recom- 2008;122(1):149–153 9. Wassilak SG, Glasser JW, Chen RT, Hadler mended for children: United States, 6. Chen RT, DeStefano F, Davis RL, et al. The 1990–1998. MMWR Morb Mortal Wkly Rep. SC. Utility of large-linked databases in vac- Vaccine Safety Datalink: immunization re- cine safety, particularly in distinguishing in- 1999;48(12):243–248 search in health maintenance organiza- dependent and synergistic effects. The Vac- 3. Chen RT, Mootrey G, DeStefano F. Safety of tions in the USA. Bull World Health Organ. cine Safety Datalink Investigators. AnnNY routine childhood vaccinations: an epidemi- 2000;78(2):186–194 Acad Sci. 1995;754:377–382 ological review. Paediatr Drugs. 2000;2(4): 7. DeStefano F. The Vaccine Safety Datalink 10. Baggs J, Lewis E, Amundson JH, et al. Dis- 273–290 project. Pharmacoepidemiol Drug Saf. tributed data models for the Vaccine Safety 4. Bohlke K, Davis RL, Marcy SM, et al; Vaccine 2001;10(5):403–406 Datalink (VSD). Presented at: 10th Annual

Downloaded from www.aappublications.org/news by guest on SeptemberPEDIATRICS 25, Volume2021 127, Supplement 1, May 2011 S51 HMO Research Network Conference. May 5, Recommendations regarding the use of Practices meeting; June 25, 2008; Atlanta, 2004; Dearborn, MI vaccines that contain thimerosal as a pre- GA 11. Farrington CP, Nash J, Miller E. Case series servative. MMWR Morb Mortal Wkly Rep. 34. Irving SA, Belongia EA, Shui IM, et al. Rapid analysis of adverse reactions to vaccines: a 1999;48(43):996–998 cycle analysis of pentavalent rotavirus vac- comparative evaluation. Am J Epidemiol. 23. Andrews N, Miller E, Grant A, Stowe J, Os- cine safety in the Vaccine Safety Datalink 1996;143(11):1165–1173 borne V, Taylor B. Thimerosal exposure in Population. Presented at: the joint 46th 12. Farrington CP. Control without separate infants and developmental disorders: a ret- Annual Infectious Diseases Society of controls: evaluation of vaccine safety using rospective cohort study in the United king- America Conference and 48th Annual In- case-only methods. Vaccine. 2004; dom does not support a causal association. terscience Conference on Antimicrobial 22(15–16):2064–2070 Pediatrics. 2004;114(3):584–591 Agents and Chemotherapy. October 25, 2008; Washington, DC 13. Davis RL, Kolczak M, Lewis E, et al. Active 24. DeStefano F. : evidence surveillance of vaccine safety: a system to does not support a causal association. Clin 35. Klein N. Vaccine Safety Datalink project: detect early signs of adverse events. Epide- Pharmacol Ther. 2007;82(6):756–759 evaluation of MMRV and febrile seizures. miology. 2005;16(3):336–341 Presented at: Advisory Committee on Immu- 25. Heron J, Golding J; ALSPAC Study Team. nization Practices February 2008 meeting; 14. Lieu TA, Kulldorff M, Davis RL, et al; Vaccine Thimerosal exposure in infants and devel- February 27, 2008; Atlanta, GA Safety Datalink Rapid Cycle Analysis Team. opmental disorders: a prospective cohort 36. Glanz JM, McClure DL, Xu S, et al. Four differ- Real-time vaccine safety surveillance for study in the United kingdom does not sup- ent study designs to evaluate vaccine safety the early detection of adverse events. Med port a causal association. Pediatrics. 2004; were equally validated with contrasting lim- Care. 2007;45(10 suppl 2):S89–S95 114(3):577–583 itations. J Clin Epidemiol. 2006;59(8): 15. Varricchio F, Iskander J, DeStefano F, et al. 26. Madsen KM, Lauritsen MB, Pedersen CB, et 808–818 Understanding vaccine safety information al. Thimerosal and the occurrence of 37. Centers for Disease Control and Prevention. from the Vaccine Adverse Event Reporting autism: negative ecological evidence from Rapid assessment of influenza vaccination System. Pediatr Infect Dis J. 2004;23(4): Danish population-based data. Pediatrics. coverage among HMO members: northern 287–294 2003;112(3 pt 1):604–606 California influenza seasons, 2001–02 16. Centers for Disease Control and Prevention. 27. Meadows M. IOM report: no link between through 2004–05. MMWR Morb Mortal Wkly Postmarketing monitoring of intussuscep- vaccines and autism. FDA Consum. 2004; Rep. 2005;54(27):676–678 tion after RotaTeq vaccination: United 38(5):18–19 38. Jackson LA, Neuzil KM, Baggs J, et al. Com- States, February 1, 2006–February 15, 2007. 28. Parker SK, Schwartz B, Todd J, Pickering LK. pliance with the recommendations for 2 MMWR Morb Mortal Wkly Rep. 2007;56(10): Thimerosal-containing vaccines and autis- doses of trivalent inactivated influenza vac- 218–222 tic spectrum disorder: a critical review of cine in children less than 9 years of age 17. Centers for Disease Control and Prevention. published original data [published correc- receiving influenza vaccine for the first Update: Guillain-Barré syndrome among re- tion appears in Pediatrics. 2005;115(1): time: a Vaccine Safety Datalink study. Pedi- cipients of Menactra meningococcal conju- 200]. Pediatrics. 2004;114(3):793–804 atrics. 2006;118(5):2032–2037 gate vaccine—United States, June 39. Centers for Disease Control and Prevention; 2005–September 2006 [published correc- 29. Stehr-Green P, Tull P, Stellfeld M, Mortenson Advisory Committee on Immunization Prac- tion appears in MMWR Morb Mortal Wkly PB, Simpson D. Autism and thimerosal- tices. Rotavirus vaccination coverage and Rep. 2006;55(43):1177]. MMWR Morb Mortal containing vaccines: lack of consistent evi- adherence to the Advisory Committee on Im- Wkly Rep. 2006;55(41):1120–1124 dence for an association. Am J Prev Med. 2003;25(2):101–106 munization Practices (ACIP)-recommended 18. Centers for Disease Control and Prevention. : United States, Febru- 30. Taylor B. Vaccines and the changing epide- Update: Guillain-Barré syndrome among re- ary 2006–May 2007. MMWR Morb Mortal miology of autism. Child Care Health Dev. cipients of Menactra meningococcal conju- Wkly Rep. 2008;57(15):398–401 2006;32(5):511–519 gate vaccine—United States, October 40. Lin ND, Kleinman K, Chan KA, et al; Vaccine 2005–February 2006. MMWR Morb Mortal 31. Verstraeten T, Davis RL, DeStefano F, et al; Safety Datalink Team. Variation in hepatitis Wkly Rep. 2006;55(13):364–366 Vaccine Safety Datalink Team. Safety of B immunization coverage rates associated 19. Thompson WW, Price C, Goodson B, et al; thimerosal-containing vaccines: a two- with provider practices after the temporary Vaccine Safety Datalink Team. Early thimer- phased study of computerized health main- suspension of the birth dose. BMC Pediatr. osal exposure and neuropsychological out- tenance organization databases [published 2006;6:31 comes at 7 to 10 years. N Engl J Med. 2007; correction appears in Pediatrics. 2004; 41. Lin ND, Kleinman K, Chan KA, et al; Vaccine 357(13):1281–1292 113(1):184]. Pediatrics. 2003;112(5): Safety Datalink Group. Impact of the intro- 20. Yu O, Bohlke K, Hanson CA, et al. Hepatitis B 1039–1048 duction of pneumococcal vaccine and risk of autoimmune thyroid 32. Haber P, Patel M, Izurieta HS, et al. Postlicen- on immunization coverage among infants. disease: a Vaccine Safety Datalink study. sure monitoring of intussusception after BMC Pediatr. 2005;5:43 Pharmacoepidemiol Drug Saf. 2007;16(7): RotaTeq vaccination in the United States, 42. Belay ED, Holman RC, Clarke MJ, et al. The 736–745 February 1, 2006, to September 25, 2007. Pe- incidence of Kawasaki syndrome in West 21. Hambidge SJ, Glanz JM, France EK, et al; Vac- diatrics. 2008;121(6):1206–1212 Coast health maintenance organizations. cine Safety Datalink Team. Safety of triva- 33. Belongia E, Irving S, Shui I, et al. Rapid cycle Pediatr Infect Dis J. 2000;19(9):828–832 lent inactivated influenza vaccine in chil- analysis of pentavalent rotavirus (RotaTeq) 43. Parashar UD, Holman RC, Bresee JS, et al. dren 6 to 23 months old. JAMA. 2006; vaccine safety in the Vaccine Safety Datalink Epidemiology of diarrheal disease among 296(16):1990–1997 population: preliminary results. Presented children enrolled in four West Coast health 22. Centers for Disease Control and Prevention. at: Advisory Committee on Immunization maintenance organizations. Vaccine Safety

S52 BAGGS et al Downloaded from www.aappublications.org/news by guest on September 25, 2021 SUPPLEMENT ARTICLES

Datalink Team. Pediatr Infect Dis J. 1998; 50. McClure DL, Glanz JM, Xu S, Hambidge SJ, used to monitor childhood vaccine safety. 17(7):605–611 Mullooly JP, Baggs J. Comparison of epide- Methods Inf Med. 2004;43(2):163–170 44. Averhoff F, Zucker J, Vellozzi C, et al. Ade- miologic methods for active surveillance of 57. Davila JC, Payne DC, Zhang Y, et al. Quality quacy of surveillance to detect endemic ru- vaccine safety. Vaccine. 2008;26(26): assessment of nonanthrax vaccination data bella transmission in the United States. Clin 3341–3345 in the Defense Medical Surveillance System Infect Dis. 2006;43(suppl 3):S151–S157 51. Lieu TA, Black SB, Ray GT, Martin KE, Shine- (DMSS), 1998–2004. Vaccine. 2008;26(12): 45. Mullooly JP, Bridges CB, Thompson WW, et field HR, Weniger BG. The hidden costs of 1577–1584 al; Vaccine Safety Datalink Adult Working infant vaccination. Vaccine. 2000;19(1): 58. Payne DC, Rose CE Jr, Aranas A, et al. Assess- Group. Influenza- and RSV-associated hospi- 33–41 ment of anthrax vaccination data in the De- talizations among adults. Vaccine. 2007; 52. Prosser LA, Bridges CB, Uyeki TM, et al. Val- fense Medical Surveillance System, 25(5):846–855 ues for preventing influenza-related mor- 1998–2004. Pharmacoepidemiol Drug Saf. 46. Bohlke K, Davis RL, DeStefano F, Marcy SM, bidity and vaccine adverse events in chil- 2007;16(6):605–611 Braun MM, Thompson RS. Epidemiology of dren. Health Qual Life Outcomes. 2005;3:18 59. Rubertone MV, Brundage JF. The Defense anaphylaxis among children and adoles- 53. Prosser LA, Bridges CB, Uyeki TM, et al. Medical Surveillance System and the De- cents enrolled in a health maintenance or- Health benefits, risks, and cost- partment of Defense serum repository: ganization. J Allergy Clin Immunol. 2004; effectiveness of influenza vaccination of glimpses of the future of public health sur- 113(3):536–542 children. Emerg Infect Dis. 2006;12(10): veillance. Am J Public Health. 2002;92(12): 47. Mullooly JP, Riedlinger K, Chun C, Weinmann 1548–1558 1900–1904 S, Houston H. Incidence of herpes zoster, 1997–2002. Epidemiol Infect. 2005;133(2): 54. Hinrichsen VL, Kruskal B, O’Brien MA, Lieu 60. Wong IC, Murray ML. The potential of UK clin- 245–253 TA, Platt R. Using electronic medical records ical databases in enhancing paediatric to enhance detection and reporting of vac- medication research. Br J Clin Pharmacol. 48. Izurieta HS, Thompson WW, Kramarz P, et al. cine adverse events. J Am Med Inform As- 2005;59(6):750–755 Influenza and the rates of hospitalization soc. 2007;14(6):731–735 for respiratory disease among infants and 61. Hviid A. Postlicensure epidemiology of child- young children. N Engl J Med. 2000;342(4): 55. Mullooly J, Drew L, DeStefano F, et al. Quality hood vaccination: the Danish experience. 232–239 of HMO vaccination databases used to mon- Expert Rev Vaccines. 2006;5(5):641–649 49. O’Brien MA, Uyeki TM, Shay DK, et al. Inci- itor childhood vaccine safety. Vaccine 62. Ali M, Canh GD, Clemens JD, et al; Vaccine dence of outpatient visits and hospitaliza- Safety Datalink Team. Am J Epidemiol. 1999; Safety Datalink Group. The use of a comput- tions related to influenza in infants and 149(2):186–194 erized database to monitor vaccine safety young children. Pediatrics. 2004;113(3 pt 1): 56. Mullooly J, Drew L, DeStefano F, et al. Quality in Viet Nam. Bull World Health Organ. 2005; 585–593 assessments of HMO diagnosis databases 83(8):604–610

(Continued from first page) PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275). Copyright © 2011 by the American Academy of Pediatrics FINANCIAL DISCLOSURE: Dr Baxter has received research grants from Sanofi Pasteur, MedImmune, Novartis, GlaxoSmithKline, Pfizer, and Merck; Dr Jackson has received research funding related to vaccines from the CDC, National Institutes of Health, Wyeth (Pfizer), Novartis, Sanofi Pasteur, and GlaxoSmithKline and has served as an advisory board member for Wyeth (Pfizer), Novartis, and GlaxoSmithKline; Dr Jacobsen has received grant funding from and served as an unpaid consultant to Merck Research Laboratories; Dr Klein has received research support from GlaxoSmithKline, Merck & Co, Sanofi Pasteur, Wyeth (Pfizer), Novartis, and MedImmune; and Mr Lewis in the past 3 years has worked on grants funded by Merck, Wyeth (Pfizer), Novartis, Sanofi Pasteur, GlaxoSmithKline, and MedImmune (now AstraZeneca). The other authors have indicated they have no financial relationships relevant to this article to disclose.

Downloaded from www.aappublications.org/news by guest on SeptemberPEDIATRICS 25, Volume2021 127, Supplement 1, May 2011 S53 The Vaccine Safety Datalink: A Model for Monitoring Immunization Safety James Baggs, Julianne Gee, Edwin Lewis, Gabrielle Fowler, Patti Benson, Tracy Lieu, Allison Naleway, Nicola P. Klein, Roger Baxter, Edward Belongia, Jason Glanz, Simon J. Hambidge, Steven J. Jacobsen, Lisa Jackson, Jim Nordin and Eric Weintraub Pediatrics 2011;127;S45 DOI: 10.1542/peds.2010-1722H originally published online April 18, 2011;

Updated Information & including high resolution figures, can be found at: Services http://pediatrics.aappublications.org/content/127/Supplement_1/S45 References This article cites 58 articles, 11 of which you can access for free at: http://pediatrics.aappublications.org/content/127/Supplement_1/S45# BIBL Subspecialty Collections This article, along with others on similar topics, appears in the following collection(s): Infectious Disease http://www.aappublications.org/cgi/collection/infectious_diseases_su b Permissions & Licensing Information about reproducing this article in parts (figures, tables) or in its entirety can be found online at: http://www.aappublications.org/site/misc/Permissions.xhtml Reprints Information about ordering reprints can be found online: http://www.aappublications.org/site/misc/reprints.xhtml

Downloaded from www.aappublications.org/news by guest on September 25, 2021 The Vaccine Safety Datalink: A Model for Monitoring Immunization Safety James Baggs, Julianne Gee, Edwin Lewis, Gabrielle Fowler, Patti Benson, Tracy Lieu, Allison Naleway, Nicola P. Klein, Roger Baxter, Edward Belongia, Jason Glanz, Simon J. Hambidge, Steven J. Jacobsen, Lisa Jackson, Jim Nordin and Eric Weintraub Pediatrics 2011;127;S45 DOI: 10.1542/peds.2010-1722H originally published online April 18, 2011;

The online version of this article, along with updated information and services, is located on the World Wide Web at: http://pediatrics.aappublications.org/content/127/Supplement_1/S45

Pediatrics is the official journal of the American Academy of Pediatrics. A monthly publication, it has been published continuously since 1948. Pediatrics is owned, published, and trademarked by the American Academy of Pediatrics, 345 Park Avenue, Itasca, Illinois, 60143. Copyright © 2011 by the American Academy of Pediatrics. All rights reserved. Print ISSN: 1073-0397.

Downloaded from www.aappublications.org/news by guest on September 25, 2021