1. Using Industrial Engineering/Operations Research Modeling Tools in Public Health DIMACS/MBI US - African BioMathematics Initiative: Advanced Study Institute on Economic Epidemiology 21 July 2009 Michael L Washington, PhD Deputy Director, Preparedness Modeling Unit Industrial & Systems Engineer Centers for Disease Control and Prevention August 2009

2. Warning What I say are my ideas and not necessarily the CDC’s or our partners

3. Agenda CDC Preparedness Modeling Unit Industrial and Systems Engineer –Linear Programming –Simulation/Queuing –Logistics –Other research Proposed issues “Don’t need to model” projects

4. CDC’s Mission Is to collaborate to create the expertise, information, and tools that people and communities need to protect their health – through health promotion, prevention of disease, injury and disability, and preparedness for new health threats.

5. Adapted from Dr. Gerberding

6. Safer healthier people Afflicted people with complications Afflicted people without complications Vulnerable people Dying from complications General protection Targeted protection Primary prevention Secondary prevention Tertiary prevention Health Priorities X X XX Current Future CDC: ENHANCING HEALTH PROTECTION! Adapted from Dr. Gerberding

7. Global Impact Adapted from Dr. Gerberding

8. Hotel M Hong Kong Guangdong Province, China A A B J H C D E I K † F G † SARS Transmission Spring 2003 Hospital 2 Hong Kong A H J B 4 other Hong Kong Hospitals Hospital 3 Hong Kong Hospital 1 HK Hospital 4 Hong Kong 2 family members CDE 34 HCWs HCW B Germany Bangkok Singapore United States I L§L§ M§M§ Vietnam Ireland K † 37 HCWs HCW 0 HCWs 28 HCWs 156 close contacts of HCWs and patients Canada G † F 4 family members 10 HCWs 37 close contacts 99 HCWs (includes 17 medical students) 4 HCWs* 3 HCWs HCW 2 family members Unknown number close contacts 2 close contacts SARS Transmission Spring 2003 Adapted from Dr. Gerberding

9. Rodent Shipment Ghana TEXAS 50 Gambian giant rats (GR) 53 rope squirrels (RS) 2 brushtail porcupines (BP) 47 tree squirrels (TS) 100 striped mice (SM) ~510 dormice (DM) NJ RS, BP TS, SM TX-3 RS, SM DM IA † GR, DM TX-4 DM TX-5 DM TX-6 TS, SM DM TX-2 GR TX-8 DM TX-7 DM TX-9 DM Japan DM TX-10 DM MN DM WI DM IL-2 DM IL-1 § GR, DM MI No human cases SC No human cases MO Human cases: 2 confirmed KS Human cases: 1 confirmed IN Human cases: 7 confirmed 9 probable/ suspect IL ‡ Human cases: 8 confirmed 4 probable/ suspect WI Human cases: 17 confirmed 22 probable/ suspect 1 PD traced 200 prairie dogs (PDs) at facility 1 PD traced 11 PDs traced 24 PDs traced 14 PDs traced 42 PDs traced Monkeypox Outbreak 2003 Adapted from Dr. Gerberding

10. Other Global Issues Influenza Climate change HIV/AIDS Tsunami Cancer Polio and all vaccine preventable diseases Many others

11. Agenda CDC Preparedness Modeling Unit (PMU) Industrial/Systems Engineer –Linear Programming –Simulation/Queuing –Logistics –Other research Proposed issues “Don’t need to model” projects

12. PMU’s Vision Promote modeling in public health across all levels Increase the pool of public health modelers Make models usable for the end user

13. The Unit Started in Sep 2008 –Hired Nathanial Hupert, MD, MPH –Funded through preparedness money Hired me back in Apr 2009 (CDC Immunization, 1998-2007; Navy, 2007- 2009) Hired a Public Health Analyst (Aug 2009) Looking to hire a modeler this year Long term, a team of modelers

14. Funded Projects Adapting an SEIR model to evaluate interventions against a pandemic influenza outbreak in the US using ARENA. Estimating economic capacity of large- scale vaccination clinics using Ex-Ante cost function. Identifying geographical areas with high risk of refusing interventions for infectious agent related public health emergencies.

15. Funded Projects Modeling to project city-specific health impacts of increases in the frequency, intensity, and length of heat waves. Predicting occurrence of plague epizootics and understanding how Y. pestis is transmitted during plague epizootics. Risk of yellow fever introduction into Puerto Rico.

16. Agenda CDC Preparedness Modeling Unit Industrial/Systems Engineer (ISE) –Linear Programming –Simulation/Queuing –Logistics –Other research Proposed issues “Don’t need to model” projects

17. What is an IE Industrial engineering is concerned with the design, improvement and installation of integrated systems of people, materials, information, equipment and energy. It draws upon specialized knowledge and skill in the mathematical, physical, and social sciences together with the principles and methods of engineering analysis and design, to specify, predict, and evaluate the results to be obtained from such systems. www.eng.ufl.edu

18. What is an SE Systems Engineering = Operations Research (OR) INFORMS: OR is the discipline of applying advanced analytical methods to help make better decisions. By using techniques such as mathematical modeling to analyze complex situations, OR gives people the power to make more effective decisions and build more productive systems. www.eng.ufl.edu

19. What is an ISE UF defines Industrial and Systems Engineering as a discipline concerned with the design and application of analytical methods to the analysis, design, and optimization of systems. The tools used include statistics, simulation, engineering methods, mathematical modeling, and optimization. True to our roots, we take these tools one-step further and implement our solutions with the use of algorithms, databases and user interfaces. After all, solutions are only useful when implemented. www.eng.ufl.edu

20. Tools (mainly IE) Manufacturing and efficiency techniques (but, has been in healthcare since 1908) Time studies Ergonomics Occupational health/safety Quality (i.e., LQA, control charts) –Monitor errors created by staff/department –Determine if a bio-terrorist attack has occurred –Immunization coverage Statistics

21. Tools (more SE) Simulation Optimizations Geographical information systems Economic analyses Decision analyses Linear and other programming

22. Agenda CDC Preparedness Modeling Unit Industrial/Systems Engineer –Linear Programming –Simulation/Queuing –Logistics –Other research Proposed issues “Don’t need to model” projects

23. Linear Programming Linear Integer and binary Quadratic Conic Mixed

24. Linear Programming Probably the number one operations research tool used Great for scheduling, determine product mix, and other things How is it being used in relationship to vaccinations? Scheduling

25. Vaccine Age Birth 1 mo 2 mos 4 mos Hepatitis B 1 Diphtheria, Tetanus, Pertussis 2 Haemophilus influenzae Type b 3 Inactivated Polio 4 Measles, Mumps, Rubella 5 Varicella 6 Pneumococcal 7 Hepatitis A 8 Influenza 9 Hep B #1 Hep B #2 4-6 yrs 6 mos 12 mos 15 mos 18 mos 24 mos 11-12 yrs 13-18 yrs Hep B #3 DTaP TdDTaP Hib IPV MMR #1 MMR #2 Varicella PCV Hepatitis A series Influenza (yearly) This schedule indicates the recommended ages for routine administration of currently licensed childhood vaccines, as of December 1, 2001, for children through age 18 years. Any dose not given at the recommended age should be given at any subsequent visit when indicated and feasible. Indicates age groups that warrant special effort to administer those vaccines not previously given. Additional vaccines may be licensed and recommended during the year. Licensed combination vaccines may be used whenever any components of the combination are indicated and the vaccine’s other components are not contraindicated. Providers should consult the manufacturers' package inserts for detailed recommendations. Vaccines below this line are for selected populations range of recommended ages catch-up vaccination Hep B series MMR #2 Varicella PPV PCV only if mother HBsAg ( - ) preadolescent assessment Recommended Childhood Immunization Schedule United States, 2002

26. http://www.vaccineselection.com/ Jacobson – Math, Math, IE Sewell - ?, ?, OR Schedule vaccination visits to minimize cost, but meet minimal vaccination requirements (combo, time between, not same time, no extra-vac different brand req.) Price vaccines with this tool and Monte Carlo

27. http://www.cdc.gov/vaccines/ recs/scheduler/catchup.htm Engineer – ?, ?, IE Child Vaccine Catch-up Scheduler Depending upon age, vaccine req. changes Smalley – ?, ?, IE Adults and Canadian children (varies by Providence)

28. Integer Programming Hurricane response (Stilp – Math, Math, IE) Assigning Applied Public Health Team –Consist of multiple skilled people (epi, hygienist, engineers, healthcare worker, etc…) and size differ –Sub-teams requires certain skills, else incur a penalty –Have to travel, so consider travel conditions (roads closed, distance, mode of travel) Objective is to minimize time to do their job Working to make this web based and stand alone Greedy algorithm –Decisions need to be made fast –Computers may not be powerful

29. Agenda CDC Preparedness Modeling Unit Industrial/Systems Engineer –Linear Programming –Simulation/Queuing –Logistics –Other research Proposed issues “Don’t need to model” projects

30. Simulation (Queuing Theory) & Optimization Emergency departments Disease spread and progression Redesign the all women’s hospital in Afghanistan (Goldsman – Math, Math, IE) Mass vaccination/prophylaxis clinics (many)

31. http://www.isr.umd.edu/Labs/CIM/ projects/clinic/software.html Queuing Theory Herrman –Math, ?, IE

32. http://www.simfluenza.org Muckstadt – Math, Math, OR Hupert – ?, MPH, MD Queuing Theory

33. http://emergency.cdc.gov/agent/sm allpox/vaccination/maxi-vac/ Simulation Washington – IE, IE, IE Meltzer - ?, ?, Econ

35. Agenda CDC Preparedness Modeling Unit Industrial/Systems Engineer –Linear Programming –Simulation/Queuing –Logistics –Other research Proposed issues “Don’t need to model” projects

36. Project Management, Transportation, Logistics We try to transport and store vaccine and vaccinate as many kids as possible based upon the resources you have and the country. Sometimes, you have to think out the box to improve the process.

37. Mobile Communities Courtesy of Dr Washington Is it worth it to track and vaccinate?

38. Courtesy of Dr. Washington Managing resources Hupert (2007) Mass Medication Modeling in Response to Public Health Emergencies: Outcomes of a Drive-thru Exercise

39. Supply and Demand/Resource Allocation Why do hospitals divert? Why is there a line in the emergency room? Kidney allocation Vaccine stockpile (Jacobson and Sewell) – $1.5 b needed for 6 month stockpile

40. Supply and Demand/Resource Allocation Dispersal of HIV funds in developing countries (Lasry - ?, ?, IE) SEIR model Two level optimization model –Global –Region (high risk/low risk) –Two types of strategies Equity – population based Optimal Results – Opt/Opt then Equity/Opt

41. Difficult for People Travel Courtesy of Dr Washington Logistics

42. Health Center Locations Where should I locate a community health center (Scherrer – IE, IE, IE. Griffin - ?, ?, IE) Optimize individual decisions (I want it here!) Look at it from a network Network models exists Up to 20% improvement in care

43. Influenza Pandemic Swann – IE, IE, IE Keskinocak – IE, IE, OR Where it’s likely to starts? How it spreads? Stations to vaccinate Distribution network – food, vaccine, etc.

44. Agenda CDC Preparedness Modeling Unit Industrial/Systems Engineer –Linear Programming –Simulation/Queuing –Logistics –Other research Proposed issues “Don’t need to model” projects

45. Timing of Testing and Treatment of Hepatitis C Markov model of the disease (Swann – IE, IE, IE. Griffin - ?, ?, IE) Asymptomatic, but infectious Cost-effectiveness of test/treatment –Cost of testing –Cost of treatment –Cost of disease Yes, for some populations

46. Closing Bathhouses Unlikely to Decrease Number of Sex Acts Bernoulli process model of HIV transmission (Swann – IE, IE, IE. Griffin - ?, ?, IE) Remove bathhouses and sex clubs Decrease transmission by 1% However, 1% increase use of condoms would decrease transmission by more than 1%

47. Economic Evaluation Cost/Benefit analyses of vaccines (Zhou - ?, ?, OR) Benefit of vaccinating sub-populations –HepB and prisoners –HPV and women –HepC and restaurant workers –Lyme Disease and everyone Prioritize vaccines (the most valuable)?

48. Agenda CDC Preparedness Modeling Unit Industrial/Systems Engineer –Linear Programming –Simulation/Queuing –Logistics –Other research Proposed issues “Don’t need to model” projects

49. Global Syphilis Prevent stillbirths Cheap treatment syphilis, to US Infrastructure (back of HIV/AIDS) Cost (take from HIV/AIDS funds) Model –Establish baseline –Provide suggestions Ms. Malik

50. Polio When should we switch from the live vaccine to the inactivated (or should we)? Will the vaccine virus replace the wild viruses? Dr. Washington

51. Influenza Modeling multiple influenza viruses One virus may dominate another Vaccination strategy –Focus on pandemic, seasonal, or both –In the US, not an issue Also model scarce resources (i.e., ventilators, beds, etc…) Could be an issue in the US Dr. Goldsman

52. Perinatally-Acquired HIV Infections 70-80% loss to follow-up in prenatal care services in South India At least 2% HIV positive Missed opportunity to prevent perinatal HIV transmission Model Economic impact of perinatally-acquired HIV infections if they continue prenatal care Dr. Sneller

53. TB Care TB and HIV is common Model The economic impact and feasibility of conducting TB tests and providing treatment at HIV clinics, especially for children < 15 The last two are very political Dr. Sneller

54. Agenda CDC Preparedness Modeling Unit Industrial/Systems Engineer –Linear Programming –Simulation/Queuing –Logistics –Other research Proposed issues “Don’t need to model” projects

55. Push packages anywhere in the U.S. within 12 hours Vendor Managed Inventory (VMI) – follow-on stores of medical supplies to augment push packages that arrive within 24-48 hours of activation Strategic National Stockpile (SNS)

56. Cold Chain Issues (Inventory and Logistics) Shipping requirements (insulated, temp, dry ice, short delivery time?) Condition on Arrival (frozen?) Storage requirement (light, freeze, re- freeze?) Shelf life (1, 1.5, 2, 3 years?) Reconstitution (yes/no, shake) Shelf life after reconstitution or opening (immediate, after some time/expiration date, vial size) Supplies and disposal

57. Are the SNS Optimally located? Number of warehouses Location Infrastructure Supplies/materials/medicine I have no idea

58. Mass Prophylaxis/Vaccination Created a mass prophylaxis model –Created in Feb 2001 –Told by boss “Not important” –Anthrax attach Aug 2001 Created a mass influenza vaccination model –Created in 2005 –“States know how to do this” (Apr 2007) –How we do this and cost (now) –Paper published this month

59. Public Vaccine Funding Previous method of funding –What did we get last year –Add a certain percentage to this year Always had to borrow from future funds Pushed for different method since 1998 Finally in 2005, a new method –Increased CDC’s vaccine budget from $1.05 b to $1.63 b, which is 20% of the CDC’s budget (Washington and Zhou) –Why so resistant to change funding equation?

60. H1N1 Vaccine Distribution Decision made in May 2009 (we wanted to model) –One distributor/multiple manufacturers No repack vaccine at sites No need to worry about skilled labor –Ship infinite clinics, any size shipment Last week –Distributer “Cost too much” May not have the capacity (300 m/yr vs 600 m/4 mth) –Demand unknown (possible wasted and reship vaccine) A model could have answered these questions May have something for anti-virals

61. “The report builds on a growing realization with in the health care community of the critical role information/communications technologies, systems engineering (OR) tools, and related organizational innovation must play to address the interrelated quality and productivity crises facing the health care system.” National Academy of Engineering and Institute of Medicine of the National Academies (2005). Building a better delivery systems: A new engineering/health care partnership.

62. Thank You Michael L Washington, PhD Deputy Director, Preparedness Modeling Unit mwashington@cdc.gov Courtesy of Dr Washington

63. Google Maps for TSP http://www.tsp.gatech.edu/maps/index.htm l http://www.tsp.gatech.edu/maps/index.htm l Optimization Game for TSP http://www.tsp.gatech.edu/games/index.ht ml http://www.tsp.gatech.edu/games/index.ht ml Arena http://www.arenasimulation.com/http://www.arenasimulation.com/ Promodel http://www.promodel.comhttp://www.promodel.com Lindo http://www.lindo.comhttp://www.lindo.com CPLEX http://www.ilog.comhttp://www.ilog.com

64. Interest is Health Systems SHS (http://www.shsweb.org/)http://www.shsweb.org/ ASQC HCD (www.healthcare.org) INFORMS (www.trinity.edu/aholder/HealthApp) ME group (www.himss.org)www.himss.org ME listserve (finance.groups.yahoo.com/group/hme/)finance.groups.yahoo.com/group/hme/ HS forum (www.eweek.org)www.eweek.org