All right. So now, let me let me switch gears a little bit and talk about how do we report out economic impact estimates. So we know we have medical costs, non-medical costs, and productivity losses, but there are many different ways in the literature that you'll find that people are reporting cost of illness. The first one is prevalence-based costs of violence. These are cross, this is using cross-sectional data. It includes all costs within a specific time period. And, typically, we're talking about one year's worth of time. And, it's very helpful for linking resources required for treatment within a given time period. So, for example, if we wanted to answer the question, how much does the medicated program pay every year for violent episodes, we would use the prevalence-based reporting. If we wanted to track how much do we annually pay for Medicaid, for violent episodes between men and women over time, we would also want to use this prevalence-based reporting. Alternatively, we can use an incidence-based reporting, or in this case, we call this lifetime costs of violence. So to, to do this type of reporting requires longitudinal data. It includes an assessment of the lifetime cost for a cohort of new victims of violence. So in other words you would asses, okay, in 2010 we had 5,000 more victims of violence. Now we're going to follow those 5,000 people to determine what are their lifetime costs associated with the violence that they experienced in that initial year. So the, the, the times that you would use this in, incidence-based reporting is when you want to talk about what is the amount of money that we can save if we prevent the violent episode from occurring in the first place. So we really want to think about what is the lifetime stream of costs that we're preventing. Very different research question that you're answering with incidence-based reporting compared to prevalence-based reporting. Now, the Economic Impact Methods that one also sees in the literature are the following. The first is summing all of the medical costs for a victim of violence. And this is usually done with cross-sectional data. The second one is summing only those diagnosis-specific costs for the victims. So in other words, we only care about those costs that are associated with, with the violence episode. And that's usually done with longitudinal data. And then the third way is, is using attributable fraction. And so I'll go into each one of these with examples. So, the first example, some all medical costs using cross-sectional data. So this example comes from Amy Bonomi out of Ohio State. This is a paper she published in 2008 where she was looking at the costs associated with child abuse. So her study objectives were to estimate the healthcare use and cost associated with child abuse. She defined child abuse as one typically does, physical, sexual, or both. And she was looking at the cost of child abuse into late middle age. So she was looking at an adult cohort. So, this was a retrospective cohort analysis. She included over 3,000 women in her sample. This is a randomly sampled group from Group Health Cooperative participated in a telephone survey to assess abuse history and health. And they also had access for these 3,000 plus women, access to their health care utilization over that period of time in that one year. So here's how they, how they retrospectively assessed whether one was, had been a victim of child abuse. Before you were 18, were you punched, kicked, choked, or did you receive more serious punishment? And then the, a question about sexual violence as well. And just let me note that these are typical ways that one retrospectively assesses child maltreatment in an adult cohort. It would be preferable to ask, many more questions to get at the child abuse experience but these are the only two that were included in her survey. So the child abuse exposures were physical, sexual, physical and sexual, and then the reference group which is those women reporting no physical or sexual abuse. The the health plan use data, so this is the healthcare utilization. So, this is getting out the medical cost were calculated between '92 and 2002, so they had ten years worth of data. And then, just like the medical cost that I mentioned before, they looked at all of those medical cost for which they had claims in their claims database. So primary care, specialty care, mental health, etc. So the annual health care cost they allocated for each unit of service delivered. They adjusted all of their costs to 2004 US dollars which is what I'm presenting here. They did unadjusted annual health care use and cost. They did some they did some adjustments based on relative risks and incidence rate ratios. And they adjusted for age, education, and calendar year. So, let's take a look at their results. So, you can see in this first column is the No abuse group. There's approximately 2,000 women in that group. The second column is physical abuse only, third column is sexual abuse only, and the last column is experiencing both physical and sexual abuse. The age across all of these cohorts are the same. You can see that there is a somewhat of a difference in the percent of the population that's white, no difference in highs, in education level. She assessed depression I'm not sure what scale she used but she assessed depression and you can see depression is higher in the abuse groups, which one would expect. She also assessed BMI and found that BMI was statistically higher in the abuse groups. And that is also something that we have found in the literature that there is a correlation between childhood sexual abuse, physical abuse, and then future BMI, which is body mass index. And now here are the costs. So if you, so just from the previous side, slide, if you know that depression is higher in abuse groups and you know that BMI is higher in the abuse groups, then you should expect that costs would be higher as well, since those two health outcomes drive costs. So again, we have the, this, the violence abuse are in the last three columns. You can see that costs, annual costs are higher for all groups. And for the, for the group experiencing both physical and sexual abuse, they experience the higher, the highest annual medical costs. So remember, this is an example of assessing all medical costs, not just diagnosis-specific medical costs. So it could be that for those persons who are experiencing maltreatment as a kid, they have some other things that are correlated with that abuse over time like depression. Like obesity that would not get an, a diagnosis code as violence as an adult, it gets a diagnosis code as something else. Hence, why it's important for you to look at total medical costs when you're doing a cross-sectional analysis such as this. So the pros of doing this approach are good for relative comparisons and understanding of the impact on the health care system in general. But the con is that is doesn't necessarily allow you to isolate what is really driving the cost. So is it the case that BMI is so highly correlated with abuse categories, and BMI then is correlated with high health care costs, or is it the mental health? From this type of analysis, we don't know. It points to what might be driving the cost but we don't know for sure. So that brings us to the next type of analysis one might do which is summing only those costs that are diagnosis specific. So in other words there's an E code, which is an external cause code in the claims data that is specific for violence or there's other, some other code being used that says this is for the violent episode. So this is, so how one does this is the total of related medical and productivity costs for all the victims. And you can provide a percentage of total costs that are violence related. So, in other words, going back to the Bonomi example, if I know that total costs for a sexual abuse group is $2800, I, if I knew what the diagnosis-specific codes were, I could go in and say, 50% or 60% of these total costs are related to the, to the previous child maltreatment. Okay. So let's look at an example of summing diagnosis-specific cost only. This is a study that was published in from the Centers for Disease Control and Prevention in 2002. It's looking at the cost of intimate partner violence in the United States. The data sources were for incidences they used the National Violence Against Women Survey and for costs, which are presented here in 1995 dollars. We used the Medical Expenditure Panel Survey, which is a very typical data set that one uses when you're interested in medical claims. They have lots of information, not just on the claims, but it ties it to diagnosis codes, it ties it to pair. So I would be able to pull out what percentage is Medicaid, Medicare, third party. We used the National Medical Expenditure Survey, which is another good, a good database for claims, Medicare claims data, and then the US statis, statistical abstract. The cost parameters are, should be very familiar to you by now because these are direct medical costs and productivity losses, as we mentioned before. We looked at ED visits, hospitalizations, MD visits, dental, physical therapy, etc. We also included mental health care. And then productivity losses, we looked at work losses, and then we made some adjustments for household productivity loss. So here's the incidence data that we're, that we got from the National Violence Against Women Survey. For every 1,000 women age 18 and over, there are 3.2 rape victimizations annually, 44 physical assaults, and 5 stalkings. And just know that some women reported more than one, and said the and so that's important to note here, except that did not happen on the stalkings part. All right. So here's some of our results. So if you look at just the direct costs for victimization, and this is using the societal perspective. So I haven't talked to you about perspective yet, so let me take a moment to do that. When we're assessing costs and if we want to take a perspective of society, then we would be including all costs of the violence, regardless to whom they accrue. So, for example, if you think about you, you having health insurance, when you go to the doctor, if you, if you're lucky to have health insurance, if you go to the doctor and you have a $500 doctor visit, you may only pay $100 out of pocket. The insurance company may pick up a couple of hundred dollars. And if you have a third party payer, your employer, your employer may pick up part of that cost as well. So this is, including all costs regardless to whom they in, accrue. As you can see here, the difference between medical care and mental health, this is, this is per person cost. For stalking, which does not typically result in a physical as, physical abuse, we only have mental health cost. But they're fairly high. The rape costs are the highest of all the different violence categories. But we also wanted to present our costs from a business perspective as well. At this time, in the early in the early 2000s when this paper was released, CDC was working very closely with the Washington Business Group on Health to get them engaged in doing violence prevention within the workplace to particularly around intimate partner violence. And so, one of the things we wanted to do was to, to go to the Washington Business Group on Health and say, hey, the cost of intimate partner violence are pretty high and you pick up a large burden of those cost. So these are the costs for the business perspective. And the way that we did this was we looked at those parts of the medical claim that were that were paid by the third party, which is your your employer-based health insurance. So here you can see the costs are fairly high. This is looking at productivity losses from a societal perspective. You can see that for stalking, those productivity losses are considerably higher than the other two violence categories. Which makes sense again, because so much of stalking occurs at the work, at the workplace. So, there is a productivity hit there. And then, from the business perspective, the productivity losses again, are substantial particularly for that last category of violence. And then, we wanted to give them total economic burden. So what do these figures look like nationally? So this is total cost of intimate partner violence, $5 billion. And you can see that of the $5 billion, $4 billion is for physical assaults, $361 million for stalking, and $330 million for rape. How does this translate to the business perspective? Shows that businesses end up paying for over 50% of the total cost of intimate partner violence. So this is a really compelling argument for the CDC to have with Washington Business Group on Health to get them really excited and interested in doing interventions around preventing intimate partner violence. So what costs are missing? Even though these estimates are fair, you know very high, $5 billion, there's still a lot of costs missing. In this analysis we did not include criminal justice costs, which we know for violence is a very substantial cost. They've been estimated to be at $18 billion per year. And we did not include productivity losses associated with fatalities. We did include productivity losses for the morbidity experienced by intimate partner violence but not fatalities. And if we had included that, we would've added an, another one, almost $1 billion to our total. And then from the business perspective, although we did, we're able to calculate that 52% of the total costs of intimate partner violences were incurred by the businesses. We still were missing cost to replace workers, and then the decrease on the job productivity.
