Progress 07/01/19 to 08/15/22
Outputs
Target Audience:The growth in drug overdoses in the United States since the late 1970s is well documented. Jalal et al. provide a detailed analysis of the growth in drug overdoses in United States. They find that the opioid death rate (in deaths per 100,000 persons) has grown at an exponential rate over 37 years. This growth rate has masked several nuances. This exponential growth is comprised of several different drug epidemics. In the early part of their analysis they find that drug overdoses are largely attributed to cocaine. This epidemic was largely confined to the Southwestern United States. Their graphical analysis shows a hot-spot in Northern New Mexico and Southern Colorado. This epidemic was followed by significant growth in prescription opioids. The prescription opioid epidemic was prevalent on west coast from 2004 through 2007. From 2008-2011 the overdose death rate due to prescription opioids grew substantially in the midwest (this finding is consistent with results found in this project). From 2012 through 2016, the rate of opioid deaths reached national prominence everywhere except the northern plains. In addition to the dramatic rise in opiod deaths in general, Jala et al. find a bifurcation of the death rate. Historically the heat map for overdose deaths demonstrated significant death rates in the population from 20 to 40 years old; however, the opioid epidemic also showed significant deaths in the age group of 40 to 60 years of age. These results show significant deaths in the younger group in urban areas while the older group are typically white and rural. Another interesting point is that the younger group can be attributed to heroine and synthetic opioids while the older group are mostly due to prescription opioids. Anne Case and Angus Deaton (2015) have extensively researched various components of the opioid epidemic. They examine the increasing death rate in the United States in the 2020s. Their analysis shows an increasing mortality rate for individuals 45 to 54 years of age. Specifically, their analysis shows that the death rate for this group in the United States has increased before 1990 and 2013. In general, the overall death rate for individuals 45 to 54 years old increased by 33.9 percent. While overdoses were not specifically addressed, they found the death rates for poisoning increased by 22.2 percent and chronic liver diseases increased by 5.3 percent. Hence, taken together these results suggested that the decline in overall lifespan observed in the 2020s could be traced to the increased death rates for males from 45 to 54 years of age. Other factors emphasized in later work is the relationship between lifespan and education. Specifically \cite{casedeaton2021} found that the death rate has increased due to differences in education while differences in the death rate has declined with regard to race. From the standpoint of the project, a key factor explaining the growth in opioid death rates since 2010 is the emergence of ``deaths of depair'' The concept of deaths of despair was developed in Case and Deaton (2020). As described by Case and Deaton: Our story of deaths of despair; of pain; of addiction, alcoholism, and suicide; of worse jobs with lower wages; of declining marriage; and of declining religion is mostly a story of non-Hispanic white Americans without a four-year degree..... The less educated are devalued or even disrespected, are encouraged to think of themselves as losers, and may feel that the system is rigged against them (Case and Deaton2020, p.4). They attribute these to a variety of factors including the perception the increased death rate among non-Hispanic white men from 45 to 50 suffered from a feeling of being left behind. They had lost the feeling of inclusion in their places of work due to a shortening of tenure of employment coupled with the feeling of losing political power. Case and Deaton conjecture that these individuals may have felt left behind after the financial collapse of 2008-2010. References A. Case and A. Deaton. Rising morbidity and mortatility in midlife among non-hispancic americans in the 21st century. Proceedings of the National Academy of Sciences, 112(49):15078{15083, 2015. doi: 10.1073/pnas.1518393112. A. Case and A. Deaton. Deatjs of Despair and the Future of Capitalism. 2020. A. Case and M. Deaton. Life expectancy in adulthood is falling for those without a ba degree, but as educational gaps have widened, racial gaps have narrowed. Proceedings of the National Academy of Sciences of the United States, 118(aa):e2024777118, 2021. H. Jalal, J. M. Buchanich, M. S. Robers, L. C. Balmert, K. Zhang, and D. S. Burke. Chainging dyanmics of the drug overdose epidemic in the united states from 1979 through 2016. Science, 361(1218):eaau1184, 2018. Changes/Problems:Same report about the difficulties imposed by missing data. What opportunities for training and professional development has the project provided?
Nothing Reported
How have the results been disseminated to communities of interest?
Nothing Reported
What do you plan to do during the next reporting period to accomplish the goals?
Nothing Reported
Impacts
What was accomplished under these goals?
One of the most challenging parts of this project was the construction of the data. This study used two primary data sets. The first dataset includes data a farm level input demand, output supply, input prices and output prices. This firm level data was constructed from the 2012 and 2017 Censuses of Agriculture. The most significant problem was missing variables at the county level. Some variables where systematically missing. For example, county level yields and prices for field crops were included in the census data but yields for fruits and vegetables were not. As a results, we used the state level data if available and the national data if necessary. In addition, some of the data were missing due to confidentiality restrictions. For example, the national data indicated that there was 7,339 acres of artichokes, but the county level data only accounted for 7,139 acres. Several counties were listed as missing. To fill in the missing information we devised a method of allocating shares of missing state vegetable crop acreage to the share of missing vegetable acres at the county level. Similar difficulties occurred for the livestock sector. In addition to the difficulties in the agricultural sector, opioid death rate data are "suppressed" for counties with less than ten opioid deaths. Hence, our procedures had to be amended for selection bias. One of the most significant outputs for this project is in the final stages as the project draws to a close. Specifically, a paper titled "Are Opioids a Rural or Urban Problem?" This paper uses a public health approach to estimate whether the rural opioid epidemic is a function of the opioid epidemic in metropolitan centers or whether it either drives the opioid epidemic in urban centers or is independent. To do this the study Area Health Resource File (AHRF) classification of counties. Within this specification we define rural counties as NonCore - Nonmetroplitan counties. The opioid death rate typically is lower than the death rate for Metropolitan and Fringe counties with the exception of 2010 and 2011 where rural counties had the highest death rate (defined in number of opioid deaths per 100,000 persons). After that time period, the rural opioid death rate falls to the smallest number from 2015 through the end of the sample. Other evidence indicates that the deaths in rural areas are typically older individuals and are due to prescription opioids (as opposed to synthetic opioids such as Fentanyl). Given this background, the study constructs the breakout defined as 10 or more opioid deaths per 100,000 (breakout is a general public health definition). Using this definition, the analysis follows the growth of the opioid epidemic over time. In 2000 opioid breakouts are confined to urban areas. Staring in 2005, the opioid breakouts are starting to leave the metropolitan areas and spreads into fringe counties. By 2010 opioid break outs are starting to spread into rural counties particularly in the Ohio River Valley area. By 2015, the entire Midwest is experiencing an opioid breakout. The actual breakout pattern is consistent with previous results in this project. Specifically, the Ohio River Valley roughly corresponds to the Eastern Upland, Heartland, and Northern Crescent regions. In addition to the reported projects, we are currently working on extending the data to address specific concerns. First, as often noted in this project data suppression is a concern. Specifically, the CDC suppressed data when fewer than ten opioid deaths occur in any county. As a result, there is a selection problem. One possibility of addressing this suppression problem is to use outside data to fill-in the suppressed data and then adjust the specification to test whether the filled in data are different than the reported data. We care currently working on a Bayesian Hierarchical model to fill in the suppressed death rate data. A second concern is the several areas in the 2012 census were affected by a severe drought. We are in the process of creating a weather variable using the Palmer Drought Severity Index to correct for this drought event at the county level.
Publications
- Type:
Journal Articles
Status:
Published
Year Published:
2019
Citation:
M. C. Wang, C. B. Moss, J. F. Oehmke, A. Schmitz, G. D'Onofrio, and L. A. Post. Prescription rate and its effect on the opioid overdoses death rate: Implications of pharmaceutical financial incentives. Internal Medicine Review, 5(3):1{13, 2019. doi: 10.18103/imr.v5i3.805.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2021
Citation:
T. Cao, C. B. Moss, A. Schhmitz, J. F. Oehmke, and L. A. Post. Opioids and despair: Limited resource agriculture and opioid deaths. Paper Presented at the Agricultural and Applied Economics Association Meetings, Austin Texas, 2021.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2022
Citation:
T. Cao, C. B. Moss, A. Schmitz, J. F. Oehmke, and L. Post. Did the opioid epidemic change farm production decisions: Application of the differential multiproduct system. Southern Agricultural Economics Association Meetings, New Orleans, 2022a.
- Type:
Conference Papers and Presentations
Status:
Submitted
Year Published:
2022
Citation:
T. Cao, C. B. Moss, A. Schmitz, J. F. Oehmke, and L. A. Post. Did the opioid epidemic change agricultural technology: Application of the distance functions. Paper Submitted to the Agricultural and Applied Economics Meetings, Anaheim 2022b.
- Type:
Journal Articles
Status:
Published
Year Published:
2022
Citation:
Post, L.A., A. Lundberg, C.B. Moss, C.A. Brandt, I.Quan, L. Han, and M. Mason. 2022. Geographic Trends in Opioid Overdoses in the US From 1999 to 2020. JAMA Network Open. 2022;5(7):e2223631. DOI[10.1001/jamanetworkopen.2022.23631.
|
Progress 07/01/20 to 06/30/21
Outputs
Target Audience:Agricultural decision makers along with state-level policy makers. Changes/Problems:The missing data has been a significant problem. Many counties are "suppressed" because of confidentiality concerns (both by the USDA and by the CDC). What opportunities for training and professional development has the project provided?The result that increases in the opioid death rate decreases the agricultural demand for labor raises the question: Is the reduction in labor due to the reducdtion in the supply of labor in these rural communities, or has the opioid epidemic reduced the quality of agricultural labor. How have the results been disseminated to communities of interest?
Nothing Reported
What do you plan to do during the next reporting period to accomplish the goals?Before completing the project under our extension we are planning three additional projects: (1) We will use a distance function to test whether or not the onset of the opioid epidemic has effected the production frontier using a distance function formulation. (2) We will use a public health approach to examine the urban/rural dynamics of the opioid epidemic. And (3) we will construct two additional spatial datasets (a) a spatial dataset to produce estimates of the rate change of the opioid death rate for agricultural counties east of the Rocky Mountains, and (b) we will create a county level dataset of the change in the Palmer Drought Severity Index to try to adjust for the significant drougth experience through much of the midwest in 2012.
Impacts
What was accomplished under these goals?
Given the data work described above, the research project generated two research products between August 2020 and July of 2021. The first project examines the effect of the share of limited resource agriculture on the number of opioid deaths in each county considering the censoring of the opioid data. Specifically, we used a double hurdle formulation [Cragg, 1971[ which uses probit model to estimate whether the dependent variable is greater than zero and then a linear equation to estimate the effect of a specified set of parameters. In general, the "Craggit" model is a two-equation simultaneous maximum likelihood specification. The results presented in Cao et al. 2021 demonstrate that selection (i.e., the case where the number of opioid deaths are not zero) is positively related to the county's population, but negatively related to the county being classified as a farm county. Hence, rural counties have a higher probability of being "suppressed" in the CDC cause of death data set. However, when counties are reported, the opioid death rate is higher for counties with more limited resource farmers. These results are consistent with Case and Deaton's [2020] concept of deaths of despair. Apart from these results, the results of this study indicate that the opioid deaths are related to the farms resource typology. One of our initial hypotheses was that the opioid epidemic would be higher where stoop labor (such as the hand harvest of fruits or vegetables) was prevalent. However, the results indicate that the rate of opioid deaths was lower in the Fruitful Rm (i.e., California, Florida, Oregon, and Washington). However, the opioid death rates where highest in the Eastern Uplands, Heartland, and Northern Crescent. These areas largely describe the middle Mississippi River Valley coupled with the Ohio River Valley. Research completed after July 2021 indicates that this regionalization can be explained by the dispersion of opioids from metropolitan counties into fringe metropolitan areas and then into rural settings. The result is also consistent with popular literature such as Hillbilly Elegy by T.D. Vance. The last empirical estimation started during the report period was the estimation of the effect of changes in the opioid death rate on the demand for agricultural inputs at the county level and the supply of agricultural outputs. We estimated this effect by reformulating the multiproduct differential demand system that Moss and Suh [2020] used to estimate the changes in banking legislation on the supply of bank outputs and the demand for inputs. The multiproduct differential demand model is theoretically based on changes in the first order condition. The exact specification is that the share of an input with respect to variable cost times the logarithmic change in the level of input demanded is a function of the change in quantity of outputs produced and the change in input prices. The exact specification can be thought of as an extension Rotterdam demand model. To investigate the effect of the opioid crisis, we appended the logarithmic change in the opioid deaths to the standard multiproduct differential demand speciation. The results indicate that the change in the opioid death rate has a significant effect on the demand of agricultural inputs. As one may expect, increases in the opioid death rate is associated with a reduction in the quantity of labor used by agriculture. The general findings are reported in Cao et al. 2022 which will be presented at the Southern Agricultural Economics Association meetings in New Orleans.
Publications
- Type:
Conference Papers and Presentations
Status:
Accepted
Year Published:
2021
Citation:
T. Cao, C. B. Moss, A. Schmitz, J. F. Oehmke, and L. A. Post. Opioids and despair: Limited resource
agriculture and opioid deaths. Paper Presented at the Agricultural and Applied Economics Annual Meetings in Austin Texas, 2021.
|
Progress 07/01/19 to 06/30/20
Outputs
Target Audience:The research contained in this project analyzes the effect of the opioid epidemic on commercial agriculture and rural communities. The first objective of this research is to examine the relationship between the type of agriculture and the prevalence of opioid abuse. To do this the research develops three aggregate agricultural outputs: commodity agriculture (i.e., corn, soybeans, and wheat), fruit and vegetable agriculture, and livestock agriculture. The dominant hypothesis is that the type of agriculture implies different characteristics of the labor demand. Opioid abuse affects and is affected by these differences. At one level fruit and vegetable agriculture tends to be labor intensive. Required field operations such as harvest implies a large amount of "back labor" which may lead to injuries that introduce workers opioids to treat pain. Alternatively, commodity agriculture in certain areas may increase the level of limited resource farmers. Increased limited resource farming may increase the incidence of despair which increases the incidence of opioid addiction. In addition to the interaction between production agriculture and opioid abuse, the level of opioid abuse appears to follow a traditional geographical pattern which is higher in Appalachia and the Ohio River valley. This pattern is related to limited resource farms and tobacco production. However, the pattern is also related to demographic characteristics and other economic characteristics such as reliance on coal mining. The target audience is then determined by these research parameters. First, while the results of this research may be interesting to farmers, it is unlikely that the information generated by this research will provide alternative actions. It is more likely that the information generated by this research will lead local policy makers such as county commissioners and state-level policymakers such as legislators to make different policy decisions. At the national level, we anticipate the information generated by this research will provide information to inform agricultural policies regarding limited resource farmers and agricultural labor. Specifically, agricultural policy has traditionally been commodity agriculture policy with price supports to program crops. Increased incidence of opioid abuse among limited resource farming may point to a program such as a "base income" support program among limited resource farmers. Agricultural labor policy may be modified to focus on policies that further limit field injuries associated with higher valued fruit and vegetable agriculture. Changes/Problems:As stated in our accomplishment section, the most significant challenge has been missing price data for inputs and outputs at the county level. While this challenge has not changed our research approach, it has added an additional level of work and complication. What opportunities for training and professional development has the project provided?Work on this project has led to four growth opportunities: (1) Increased proficiency of use of APIs, (2) increased proficiency in Bayesian econometrics - particularly in the use of Bayesian Kriging both as a data tool and as a specification for empirical analysis, (3) use of High Performance Computing Technology (HiPerGator at the University of Florida), and (4) Renewed expertise in index number theory with spatial (e.g., cross-sectional) consistency. How have the results been disseminated to communities of interest?
Nothing Reported
What do you plan to do during the next reporting period to accomplish the goals?I believe that the data work will require the rest of the summer of 2020. The code to impute output prices typically requires about 2 days on either the HiPerGator or one of my significant research desktops. In August 2020, the goal is to compute the input and output indices. The effect of opioid abuse on agriculture in the United States can then be estimated in September and October. The specification that we will use was recently published in Moss, Charles B. and Dong Hee Suh. 2019. Effect of Compliance Cost on the Supply of Bank Credit to Agriculture: A Differential Approach. American Journal of Agricultural Economics 102(3): 713-726. [DOI: 10.1002/ajae.12074] The new opioid death rate data from Northwestern University will be used in this effort. In addition researchers at Northwestern University use the death rate data in combination with various demographic data to analyze the effect rural demographic factors.
Impacts
What was accomplished under these goals?
Most of the work in the first year has focused on data generation. To accomplish objective 1, the study will generate a system of three output level, output price, input level, and input price indices for each of the 3,160 counties in the United States that produce some level of agricultural output. To accomplish this, researchers at the University of Florida have used an Application Programming Interface (API) to download county level data. In the fall of 2019 and spring of 2020, we developed a list of significant crops to characterize commodity agriculture (i.e., corn, soybeans, wheat, and cotton), fruit and vegetable agriculture (i.e., strawberries, blueberries, citrus), and major livestock outputs. The quantity produced and total value of each product were downloaded from the USDA's data site. In the late spring, work shifted to specifying the inputs. In addition to labor, we downloaded the amount of money spent on fuel, and agricultural chemicals (including fertilizers). In addition, work in the spring 2019 shifted to deriving the price for each input and output. For many crops, prices are simply defined as the revenue divided by quantity marketed. However, several counties have missing values for either revenue or production due to confidentiality conditions (e.g., when fewer than ten farmers in a county produce a particular crop the revenue value is withheld). Overt he last three months researchers at the University of Florida have been developing a Bayesian Kriging approach to filling in the missing prices using a form of a gravity model (e.g., based on distances between counties, the center of the state, and marketing hubs). This approach has been computationally intense. In addition to the difficulties in commodity agriculture, the researchers have adopted the same Bayesian Kriging approach to impute county level input prices for fuel and agricultural chemicals. Similar challenges have arisen in the collection of opioid overdose data. The original research relied on the Center for Disease Control, Cause of Death data. Like the USDA data, the Cause of Death data has missing data due to a form of confidentiality criteria. To describe the confidentiality question, the overdose death rate data are reported in deaths per 50,000. In smaller counties releasing the data could (hypothetically) allow for the naming of overdose death victims. The problem with this "missingness" is that agricultural counties in several areas of the countries tend to be smaller counties. Hence, agricultural counties are more likely to be missing. Researchers at Northwestern University have spearheaded the effort to obtain better data on opioid death rates by directly obtaining the data from state-level authorities. Research linking the two groups - efforts at the University of Florida generated the percentage of limited resource farmers in each county coupled with a detailed set of county classifications form the USDA. Coupling this data with the death rate data from Northwestern, we are currently estimating the relationship between limited resource agriculture and opioid death rates using the Bayesian Kriging formulation.
Publications
- Type:
Journal Articles
Status:
Published
Year Published:
2019
Citation:
Wang, Michael C., Charles B. Moss, James F. Oehmke, Andrew Schmitz, Gail DOnofrio, and Lori A. Post. 2019. Prescription Rate and Its Effect on the Opioid Overdoses Death Rate: Implications of Pharmaceutical Financial Incentives. Internal Medicine Review 5(3): 1-13. [DOI: 10.18103/imr.v5i3.805]
|