County-level aggregate costs arising from Indiana’s opioid crisis

This article examines the impact of Indiana’s ongoing opioid misuse crisis at the county level. The purpose of this analysis is twofold: First, we provide local leaders detailed information on the full extent of damages local communities have experienced from the opioid crisis thus far to offer perspective of the scope of the problem at the local level. Second, we present an algorithm leaders can use to estimate the rate at which economic damages continue to accumulate per community.

We defined “community” as the county for three primary reasons:

1. Information was available allowing us to assess damages at the county level.
2. County-level data can be used as a point of reference to reflect local labor markets within Indiana.
3. Decision-making at the county level will be crucial in generating steps to end the opioid epidemic.

We began by considering the statewide analysis by Brewer and Freeman, also published in this issue.¹ In that study, the total cumulative economic damages in Indiana arising from the epidemic from 2003 through 2017 were estimated to total $43.3 billion. Three distinct categories of damages were included in that sum:

1. Total aggregate direct costs associated with non-lethal opioid misuse.
2. Losses to gross state product stemming from labor market tightness and associated losses to workforce productivity (GSP losses).
3. The accumulating costs associated with deaths from opioid overdoses.

Neither cost estimates for pain-and-suffering nor the investment costs necessary to repair accumulating damages were considered. These and perhaps other categories of damages from the opioid crisis—such as losses to productivity in family members of those deceased or injured from the crisis, medical costs associated with hepatitis and other non-HIV blood-borne illnesses, as well as costs associated with increased use of mental health counseling and antidepressants—have yet to be directly connected to the crisis in ways such that corresponding costs can be measured with a reasonable degree of accuracy.

Figure 1 shows the estimated aggregate damages at the county level between 2003 and 2017. (More results, including per capita calculations, are found following the methodology).

Figure 1: Cumulative economic damages arising from the opioid epidemic, 2003 to 2017

Source: Authors’ calculations

Methodology for allocating damages to counties

In order to allocate statewide economic damages to the 92 Indiana counties each year, we considered three categories of damages separately, and the allocation method for each is outlined in the following sections.

1. Direct costs: Funerals, first response, acute hospitalization, long-term treatment, neonatal abstinence syndrome, foster care, arrests and court costs, property loss, incarceration, HIV and overdose death wage losses.
2. Indirect costs: Lost GSP in tight labor markets.
3. Present value of all lost future productivity of past opioid-related casualties.

Direct costs (Category 1)

Because detailed county-level information was absent for the various subcategories of direct costs, we apportioned the state-level direct costs across counties based on a combination of the data items that were available on a county-level basis. For several years of data, we obtained information on county-level, non-fatal, opioid-related emergency room (ER) visits (available for years 2009-2016) and county-level reported opioid deaths (available for years 2011-2016).

We calculated county-level opioid fatalities each year by multiplying the estimated statewide opioid fatalities by the proportion within the state of reported opioid fatalities per county, as reported by the Indiana State Department of Health (ISDH). As discussed in the statewide analysis, ISDH numbers from the early years of the crisis underreported overdose deaths attributable to opioids.

We calculated cumulative deaths in each county by summing the annual deaths over the crisis period (2003-2017). For years 2003-2010, when county-level opioid deaths were not reported, we estimated the opioid-related deaths in each county by multiplying state-level opioid deaths by the average proportion of state-level opioid deaths each county had in years 2011-2013. (We did not make actuarial adjustments for this component because such modifications would not have produced material changes to the outcome of the county-level allocation of damages.)

We allocated costs at the county level by the following algorithm:

where Direct costs_c denotes direct costs for county c and Allocation_c is the sum of state-level direct costs. Allocation_c is the allocation rule, given as follows:

Population allocation reflects the proportion of state population county c comprises. Severity allocation factor (SAF) denotes the ratio of the weighted sum of emergency room visits and cumulative overdose deaths (2003 to a given year of damages) in county c in a given year to the state-level weighted sum of ER visits and cumulative overdose deaths in the same year. Thus, for each county in each year, 20 percent of the allocation of state-level economic damages was based on the relative population in each county and the remaining 80 percent of the allocation was based on county-level reporting of the two direct cost categories available (i.e., ER visits and cumulative deaths). We included a population allocation in the weight because labor markets and communities transcend county lines; thus, counties experience spillover effects from damages in neighboring counties and across the state as a whole. Further, this adjustment smooths underreporting of statistics in counties where such underreporting may have been present.

The SAF is based on the following formula:

where,

Thus, the allocation of economic damages into specific counties included emergency room visits and cumulative deaths, weighting each based on the relative cost each (deaths and ER visits) comprised within the total state-level direct costs. ER visits proxied for all annual non-death costs. GSP losses associated with death, which have been relatively severe, have been categorized separately.

The rationale for this procedure was to develop an algorithm that accurately incorporated the cost of each death relative to the cost of each non-lethal emergency room visit. We took this rationale and infused it within an allocation function, which divided the total state costs into the counties vis-à-vis each county’s distinctive annual mix of opioid overdose deaths and opioid-related emergency room visits.

Over the course of the epidemic, we see the death adjustment ratio increasing over time. In 2009, this annual number was estimated to be 1.20. The number in 2010 was also estimated to be 1.20, while it was measured at 0.95 in 2011. From here, the number generally increased annually to 1.16 (2012), 1.49 (2013), 1.89 (2014), 1.71 (2015), and then to 5.15 in 2016. The SAF in 2017 incorporated the death adjustment ratio from 2016 of 5.15.

The increases in this measurement over time reflect two elements: better categorical reporting of opioid-related emergency room visits and more emphasis on quick emergency response to opioid misuse incidents, which more often results in non-lethal emergency room visits these days versus in earlier years of the crisis.

For years 2003-2008, in which we did not have county-level ER visits, we used the 2009-2011 three-year average to allocate costs in the early years of the epidemic.

Indirect costs (Category 2)

The indirect cost category included lost GSP to the state resulting from opioid misuse and the associated loss of workers during periods of labor market tightness. For this category, we allocated state-level damages across the 92 counties by the following algorithm:

Allocation_c is defined in the direct cost section. Labor effect_c is given as follows:

where UE_s is the state-level unemployment rate and UE_c is the unemployment rate for county c, in any given year, measured as the average unemployment rate across all 12 months. Thus, a county with an unemployment rate lower than the statewide unemployment rate received a larger allocation of GSP loss. GSP labor losses are statewide annual losses due to opioid misusers’ inability to contribute at full production levels.

Notes to categories 1 and 2

Crossfooting direct and indirect costs: The above methodology for estimating the labor effect resulted in a 1-2 percent imprecision, in that the resultant sum of the county damages estimates did not add exactly to the state damages estimate. Therefore, to correct this 1-2 percent imprecision, we added to each county a pro rata amount based on each county’s proportionate damages, such that the sum of damages in all 92 counties equaled the state total damages in each corresponding year.

Present value of economic damages: Initially, we allocated economic damages each year within each category (deaths, direct non-lethal categories and GSP) in nominal dollars for a given year. We then applied present value factors based upon GSP growth using mid-year convention. Selection of these factors are explained in the statewide analysis.

Present value of future lost productivity from past opioid casualties (Category 3)

We estimated that an additional $18.9 billion in statewide economic damages—calculated as a present value as of December 31, 2017—resulted from accumulating deaths throughout the epidemic period. These losses stem from elimination of economic output that would have otherwise occurred via future expected lifelong productivity, based upon the average age of decedents each year and the expected productivity of adults in Indiana. This damages estimate, which derives particularly from future losses, was not added into the annual allocations. Consequently, we present this number as a separate category, having allocated the statewide totals to counties based upon each county’s accumulated deaths as of 2017.

Exploring the results

Download the appendix spreadsheet to view the estimated county-level annual economic damages for all three categories over the entire crisis period for all 92 Indiana counties—both nominally and on a per capita basis. (All damages amounts are valued as of December 31, 2017.)

For quick reference, Table 1 lists the 20 counties having the greatest nominal aggregate economic damages.

Table 2, meanwhile, lists the 20 counties suffering the greatest aggregate economic damages per capita.

Figure 2 visualizes these per capita data statewide.

Figure 2: Per capita cumulative economic damages arising from the opioid epidemic, 2003 to 2017

Source: Authors’ calculations

Reflections on the county-level estimates

Damage estimates originated from reports on opioid-related deaths per county and opioid-related emergency room visits per county. Such reports have improved over time, with 2016 reports providing perhaps the most accurate accounts of opioid-related activity throughout the state since the epidemic began. We have made efforts to compile cost estimates and allocate them using scientific methods, including inference and deduction. In some counties, such as Spencer and Martin, the numbers appear quite low—which could be because opioid misuse has not harmed those counties as much as elsewhere in the state or the numbers may be underreported.

Ongoing opioid misuse damages estimation tool

To help local leaders understand the scope of the opioid problem in their particular communities, we offer a downloadable tool that can be used to calculate damages going forward.

The tool is composed of two options. The first option is a comprehensive damages tool that includes the present value of future known labor contribution losses, because GSP contribution through the duration of expected lifespan is known to be lost. The second option is an annual damages tool that does not include the known losses from future work periods of prior deaths. This second option includes only annual costs occurring within the damages year in question, without consideration for losses that will occur in future periods.

This tool requires no personal information about misusers. Rather, pertaining to individuals, the tool only requires tallies of emergency room visits related to opioid misuse and opioid overdose deaths.

The full array of required information to use the opioid misuse damages estimation tool includes the following items, which must be obtained by the user for the period in question. (Items followed by an asterisk are provided by county in the tool itself).

1. County unemployment rate in the period wherein a damages estimate is sought (UE_c).
2. State unemployment rate in the period wherein a damages estimate is sought (UE_s).
3. Cumulative decedents in the county under investigation through the end of the year preceding the year of analysis (Deaths_pre-y).*
4. Additional opioid overdose casualties in the period wherein a damages estimate is sought (Deaths_y).
5. Total cumulative opioid overdose decedents through the end of the year in which the damages estimate is desired (Cumulative decedents_y).*
6. Emergency room visits related to opioid misuse in the period wherein a damages estimate is sought (ER visits_y).
7. The per capita allocation per county of GSP losses attributable to a given county for year 2017 (Population GSP factor_c).*
8. An adjustment for inflation to reflect expected costs in a given county in a given future year (1 + inflation)^t.²

The tool incorporates current-year GSP losses arising from work absence due to death. The average age of decedents through 2017 was 42 years, resulting in 23 years of lost wages and resultant GSP. For newly deceased (in the current year), the average age is expected to be 39 years at time of death, resulting in 26 years of lost wages and resultant GSP. Our tool reflects this assumption by allocating 46.94 percent of the deaths damages constant ($94,143) to the pool of decedents from prior years, and it allocates the remaining 53.06 percent of the deaths damages constant ($94,143) to the newly reported overdose deaths in the year subject to analysis. (More details about the underlying calculations are available inside the tool.)

Download the ongoing opioid misuse damages estimation tool »

Conclusion

While limitations to the fullest comprehensiveness of this study certainly exist, we believe the estimates presented herein are reasonable reflections of damages. Thus, we are confident the findings contained within the present study add value by allowing local leaders to gauge the level of opioid misuse-related damages within each of their communities, so they can plan for investments and budget local capital commensurate with the size of the local problem.

Certainly, leaders in counties wherein the damages have been relatively large may wish to seek additional information about solutions for slowing the rates of misuse, as well as solutions for the victims of the epidemic, who find themselves at the mercy of addiction to some of the most deadly and most costly drugs humankind has seen.

Illuminating the level of cumulative damages is merely the first step on the road to recovery. We believe the effective solution set to this vast and costly epidemic will include thoughts and ideas transcending boundaries and bridging communities. Through efforts made by decision makers, medical leaders and first responders—alongside associates from regional think tanks, research universities and nonprofit organizations—our communities can develop and implement effective solutions, raising our state in its productive output capacity, as well as leading others in directions yielding improved growth and progress.

Notes

1. Brewer, R. M., & Freeman, K. M. (2018). Cumulative economic damages from 15 years of opioid misuse throughout Indiana. Indiana Business Review, 93(1). Retrieved from www.ibrc.indiana.edu/ibr/2018/spring/article1.html
2. The inflation adjustment would require compounding, with 2018 estimates requiring only “1 + inflation” without the need to apply exponential calculations. However, for years 2019 and beyond, because the estimation tool uses 2017 GSP damages numbers, the inflation adjustment would require exponential calculation to reflect a damages estimate in a year wherein an estimate is desired. For instance, in 2020, “t” (the exponent) would be 2020-2017, or 3. It is true actual inflation rates vary from year to year, and indeed, GSP growth rates—the actual rates that would be relevant in this application—are also bound to vary each year. The problem with GSP rates is that they are not reliably available in contemporary sources, while inflation numbers are typically available more often without significant lag. Technically, several options exist for taking care of the inflation adjustment, but using inflation rates to adjust the damages estimates within the tool will provide a reasonably accurate reflection of costs in dollars as of the year in question.