Disease Burden of PTSD in the US Military Michael C. Freed, Ph.D. Clinical Research Psychologist DoD Deployment Health Clinical Center, Walter Reed Army Medical Center Assistant Professor (Research) Department of Psychiatry, Uniformed Services University of the Health Sciences (USUHS) Scientist Center for the Study of Traumatic Stress, USUHS The views expressed in this presentation are those of the authors and do not necessarily represent the official policy or position of the Deployment Health Clinical Center, Walter Reed Army Medical Center, USUHS, Department of Defense, or the United States Government.

Charles C. Engel, MD, MPH, COL, MC, USA (1,2) Co-Authors Xian Liu, Ph.D. (1,2) Phoebe Kuesters, BA (1) Kristie L. Gore, Ph.D. (1,2) Rachel Goldberg, BA (1) Charles C. Engel, MD, MPH, COL, MC, USA (1,2) (1) Deployment Health Clinical Center, Walter Reed Army Medical Center (2) Department of Psychiatry & Center for the Study of Traumatic Stress (CSTS), Uniformed Services University of the Health Sciences Co-authors; language caveat. Disease Burden—not meant to be pejorative. Persons with PTSD not a burden. It is the “disease,” not the person, which is ‘burdensome.’ Thus, treat/prevent the disease to reduce burden. WHO and World Bank commissioned Harvard University.

Objectives Define the Disability Adjusted Life Year (DALY) as a disease-generic estimate of disease burden. Present the parameters necessary to calculate the disease burden of PTSD. Present coarse disease burden estimates of PTSD in the US military. A lot of front-loaded education about disease burden metrics because they are somewhat foreign to clinicians. A caveat: Language used is that of epidemiologists and economists. If it seems pejorative, it is not meant to be. (e.g., disease burden does not imply that someone with PTSD is a burden).

Why Assess Disease Burden? Quantify a “missed opportunity” or “lost potential” (e.g., years of life lost) resulting from a particular disease. Advocacy for resource allocation—prioritize importance of disease at the population-level. Utilitarian metric, to maximize the health of a society. Measured alone, disease pervasiveness, persistence, and impact do not sufficiently capture population health. Disease burden combines pervasiveness, persistence, and impact into a single unit of measurement. LANGUAGE CAVEAT Disease takes away health—years lived and morbidity. Diseases can be fatal or not, but we need to compare both on the same metric. We want to minimize the burden of any disease on population health through prevention, treatment, and management.

Why Assess Disease Burden? An example: Ischemic Heart Disease* (IHD) IHD ranks as the #1 most “burdensome” disease in the US. Affects both length (mortality) and health-related quality of life (morbidity). Pervasive (incidence) Male Female Total 860,527 (0.66%) 532,499 (0.39%) 1,393,026 (0.53%) Persistent Course (all ages) 2 years Impactful No. of Deaths Yrs of Life Lost Disability Weight (Morbidity)* 536,314 (23%) 2,858,744 (16%) 0.11 IHD #1 out of 400+ diseases and 8 MH disorders *IHD includes and combines Acute MI, Angina 1-year prior to final year, and Heart Failure. (Michaud et al., 2006)

Why Assess Disease Burden? Depression ranks as #4, on the same list as IHD (Michaud et al., 2006). How can a non-fatal disease be compared to IHD? Especially when relatively few deaths are attributable to depression. We do not typically think that years of life lost is on the same scale as health-related quality of life. Disease burden measures population-level morbidity and mortality burdens along a common scale. Compare apples and oranges…

Why PTSD? Military PTSD Burden in the US (Michaud et al., 2006) Ranks 15th (all races, both sexes, just under IHD but above HIV) in terms of morbidity. Ranks 17th in total disease burden (morbidity + mortality burden) in women. Military More pervasive (relatively speaking) More persistent (combat PTSD) More impactful Before we get into what disease burden is.. Argument for why disease burden for PTSD is a no brainer for ptsd

Why PTSD? PERVASIVE Prevalence Incidence (Smith et al., 2008) Vietnam Veterans (NVVRS; Kulka et al., 1990): 9-15% current Desert Storm (as reported in Engel, 2006): 7-15% OEF/OIF (as reported in Hoge et al., 2004): 6-20% VA Primary Care (data collected in 1999; in Magruder et al., 2005): 12% 3 DoD PC Clinics (data collected 2005; Gore et al., in press): 9% Incidence (Smith et al., 2008) New onset 7.6-8.7% in deployers New onset 2.3-3.0% in non deployers New onset within a 2.7 years (mean) timeframe Before we get into what disease burden is.. Argument for why disease burden for PTSD is a no brainer for ptsd

Why PTSD? PERSISTENT US Disease burden (Michaud et al., 2006): 4 years (males) 5 years (females) NCS (Kessler et al., 1995): 3 years in treatment 5.33 no treatment NVVRS & HVVP (from Schnurr et al., 2003): 18.54 years Before we get into what disease burden is.. Argument for why disease burden for PTSD is a no brainer for ptsd

Why PTSD? IMPACTFUL Decrements in quality of life Symptom severity correlated with poorer functioning (Magruder et al., 2005) Symptom improvement was synchronous (vs lagged) with improvement in QoL (Schnurr et al., 2006) Other physical and mental health problems (Grieger et al., 2006; Gillock et al., 2005; Schnurr & Green, 2004) Increased use of medical services (Gillock et al., 2005) Increased employment absenteeism and higher medical costs (Berndt et al., 2000; Walker et al., 2003) Before we get into what disease burden is.. Argument for why disease burden for PTSD is a no brainer for ptsd

Why PTSD? TREATABLE PREVENTABLE? Psychotherapy (Bradley et al., 2005; Bisson et al., 2007) Pharmacotherapy (Davidson, 2006; Ipser et al., 2006; Davis et al., 2006) PREVENTABLE? Before we get into what disease burden is.. Argument for why disease burden for PTSD is a no brainer for ptsd

What is Disease Burden? Summary measure of population health that estimates the “gap” or difference between actual population health and some specified norm or goal (WHO, 2001). Morbidity + Mortality Hypothetical cohort of 100,000 people. Blue line is survival curve of population if no one had disease We differentiate disease+ from disease-. Just a description based on incidence and persistence in population. We could think of this division as red hair vs. all other hair. Says nothing yet. In this case, though we will assume the disease is fatal. And we can see how many years of life are lost (YLL) in the population. 2 “spaces” are just reciprocal.

What is Disease Burden? Valuing health… morbidity How many years in a disease state (i.e., disability) is equivalent to 1 year of perfect health for a given person with the disease? Disease morbidity is important too. Mortality is just one negative outcome. Morbidity is another. Discuss weighing system in a few minutes. Here, note that a “disability free” year is worth 0. A dead year is worth 1. We ask, how many years in a disease state (disability) is equivalent to 1 year of perfect health. In the example, approx. 0.33. Conversely, we ask, what is the decrement (disability weight) should we attach to being diseased. Here it is ~0.66.

What is Disease Burden? = Mortality + Morbidity Burden on the same scale. Disease Burden To put morbidity and mortality together, we just add. Dashed lines are the added burden from morbidity. Thus, we can examine the disability adjusted life years (morbidity + mortality of the population) in relation to the entire population. Or, we could value other diseases too. See, the DALY depends on the pervasiveness, persistence, and impact of the disease.

Morbidity (Disability Weight) Disease Burden DALYs Morbidity (Disability Weight) Impactful Course of Illness Persistent Discounting (Optional) Mortality Age Weighting [numerator] Incidence Pervasive Most components make intuitive sense, but let’s elaborate on the disability weight…

+ What is Disease Burden? Disability Adjusted Life Year (DALY) = Mortality Years of Life Lost (YLL) (Fatal diseases only) Morbidity Years Lost Due to Disability (YLD) (Fatal and Non-Fatal Diseases) + Pervasiveness Incidence x Persistence Course Impact Disability Weight Age Adjustment Discount Rate Simple arithmetic: multiplication and addition Optional & Controversial • DMED • Smith et al. (2008) Combat versus Non-combat estimates • GBD Study (Murray & Lopez, 1996) • Replication Study (Michaud et al., 2006)

Example of a Person Tradeoff How Health is Valued Example of a Person Tradeoff (adapted from Freed et al., in press; Sanderson & Andrews, 2001) Choose Either A or B Intervention B Extend the life of n individuals with the disabling condition of interest for 1-year. All n individuals will die after 1-year. If don’t choose Intervention B, then all will die today. Intervention A Extend the life of 1000 healthy people for 1 year. All 1000 people will die after 1-year. If don’t choose Intervention A, then all will die today. How we value health 1 year in perfect health is worth how many years disabled? Conversely, how many ‘disabled’ years are worth 1 year in perfect health. Value is relative to other diseases anchored by perfect health and death. Value diseases and disability—NOT people. Remember GBD commissioned by WHO and World Bank… economics. Other methods too. Disability weight = 1-1000/n, where the respondent is indifferent between A and B. It is a ratio of 1000 : n, where the minimum n = 1000 (best health) and the maximum n = 1 million (indicating worst health). The disability weight therefore falls between 0.001 and 1.

Method Pervasiveness: Incidence Impact: Disability Weights Provider diagnosis from Defense Medical Epidemiological Database (DMED) Ambulatory data by branch and age in 2007 First occurrence diagnosis ICD 309.81 Self-report questionnaire from the Millennium Cohort Study (MCS), sensitive criteria (Smith et al., 2008) By branch PCL-C with DSM-IV criteria (at least moderate distress) Impact: Disability Weights GBD (estimation): 0.11 (Murray & Lopez, 1996) Replication (actual PTO): 0.66 (Sanderson & Andrews, 2001) Our model Pervasiveness: provider diagnosis vs. self-report questionnaire. We go with conservative provider diagnosis. Impact: GBD estimate vs. Replication is actual PTO. We like actual.

Method Persistence Mortality: Years in PTSD vs. Not GBD: 4 years (males), 5 years (females) NCS: 3 years in treatment & 5.33 no treatment (Kessler et al., 1995) NVVRS & HVVP: 18.54 years (from Schnurr et al., 2003) PTSD screener study: 38% combat PTSD (Gore et al., in press) Our estimate: 9.67 years Mortality: Years in PTSD vs. Not Assessed with a life table using rates from a US white male population in CY2004. Our model Persistence: We estimate 9.67 years because military has higher combat ptsd. Mortality: white male life table, used in VA paper. We know military population is selected, so US white male might not be the best. And, people leave the military.

Method Model parameters for sensitivity analyses Base Case Worst Case Incidence from DMED Disability weight: 0.66 Sanderson & Andrews (2001) Worst Case Incidence from MCS, sensitive criteria Disability weight: 0.66 Best Case Disability weight from GBD: 0.11 Our model

Results All person years lived in 2007 for each branch = red + green + blue. Red + Green is years in PTSD Blue + Green is all healthy years Red is YLD, or the disease burden.

Results Wide variation in YLD due to wide variation in incidence and disability weight estimates. Freed et al. (not used in these sensitivity analyses) estimated the health utility (~ 1-disability weight) of PTSD in veterans and it appears closer to the replication method.

*data from Michaud et al., 2006 Results US estimates from Michaud et al 2006. Everything is relative here, and proportional to a military the size of the US population. Shouldn’t’ PTSD receive more attention? Even at the best case, ti still nearly exceeds IHD. *data from Michaud et al., 2006

Discussion & Conclusion Prioritize diseases based on pervasiveness, persistence, and impact Differences in service branches due to differences in pervasiveness (in our model) If model parameters change, then estimates change PTSD in military (scaled) vs. IHD in general population Disability weight: 0.11 (estimation in GBD) vs. 0.66 (actual PTO in replication study) Outcome useful for policy makers and health economists Also can and should be considered by providers as a way to better advocate for funding

Progress not Perfection Limitations Service utilization from DMED data has its drawbacks Patchwork model parameters US white male life table may not be representative of a military population Disability weights may not reflect military preferences We do not have the overall disability measurements within the military. Thus, percentages of YLDs may be inflated. Crude sensitivity analysis did not account for discounting or age weighting Progress not Perfection

Future Directions A comprehensive dataset to include mortality and morbidity in servicemembers and veterans. Service member preferences for most relevant diseases within DoD and VA. Comparison of resources devoted to the prevention, treatment, and researching of the diseases being studied. Longitudinal trend in disease burden as a function of resources devoted to prevention, treatment, and research.

Questions, Information, & Assistance DoD Deployment Health Clinical Center Walter Reed Army Medical Center Building 2, Room 3E01 6900 Georgia Ave, NW Washington, DC 20307-5001 (202) 782-6563 E-mail: pdhealth@amedd.army.mil Website: www.PDHealth.mil Provider Helpline 1-866-559-1627 Patient Helpline 1-800-796-9699 Correspondence regarding this presentation: Michael C. Freed, Ph.D., EMT-B Clinical Research Psychologist Deployment Health Clinical Center (202) 356-1012 x40318 Michael.Freed@amedd.army.mil

Incidence Rates Used Branch DMED MCS 1-Year US Burden of Disease (Smith et al., 2008) US Burden of Disease (Michaud et al., 2006) Population Size All 0.80% 1.31% 1,360,798 USA 1.35% 1.82% 510,390 USN 0.36% 1.07% 334,077 USAF 0.27% 0.67% 333,664 USMC 1.04% 1.37% 182,667 US Population (Males) 0.09% 129,810,215 US Population (Females) 0.23% 135,473,568