Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Assessing the impact of the Good Samaritan Law in the state of Connecticut: a system dynamics approach

Assessing the impact of the Good Samaritan Law in the state of Connecticut: a system dynamics... Background: Although Good Samaritan laws (GSLs) have been widely adopted throughout the United States, their efficacy in individual states is often unknown. This paper offers an approach for assessing the impact of GSLs and insight for policy-makers and public health officials who wish to know whether they should expect to see outcomes from similar policy interventions. Methods: Utilizing a system dynamics (SD) modeling approach, the research team conducted a policy evaluation to determine the impact of GSLs on opioid use disorder (OUD) in Connecticut and evaluated the GSL based upon the following health outcomes: (1) emergency department (ED) visits for overdose, (2) behavioral changes of bystanders, and (3) overdose deaths. Results: The simulation model suggests that Connecticut’s GSL has not yet affected overdose deaths but has resulted in bystander behavioral changes, such as increased 911 calls for overdose. ED visits have increased as the number of opioid users has increased. Conclusions: The simulation results indicate that the number of opioid-related deaths will continue to increase and that the GSL alone cannot effectively control the crisis. However, the SD approach that was used will allow policymak - ers to evaluate the effectiveness of the GSL over time using a simulation framework. This SD model demonstrates great potential by producing simulations that allow policymakers to assess multiple strategies for combating the opioid crisis and select optimal public health interventions. Keywords: Opioid use disorder, Emergency medicine, Health policy simulation, System dynamics modelling Background and Introduction older misused opioids in 2017 [1], and more than 47, 000 The significant increase in the number of opioid over - people died from opioid overdoses in the same year [2]. dose deaths in the United States over the past few dec- As this death toll continues to rise, policy interventions ades is now widely recognized as a national public health become increasingly important as a means of reducing crisis. Almost 11.4  million Americans aged 12  years or overdose deaths, and policymakers need tools to help guide decision-making. Most importantly, SD modeling is useful for studying resistance to public health inter- ventions [3]. System dynamics (SD) modeling has gained *Correspondence: nasim.sabounchi@sph.cuny.edu Nasim S. Sabounchi and Rebekah Heckmann share first authorship momentum in the health sector due to its potential to Department of Health Policy and Management, Center for Systems address the challenges of decision-making for complex and Community Design, City University of New York (CUNY ) Graduate policy problems [4]. School of Public Health and Health Policy, 55 W. 125th Street, New York, NY 10027, United States of America As part of the Centers for Disease Control and Preven- Full list of author information is available at the end of the article tion’s (CDC) Prescription Drug Overdose Prevention © The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco mmons. org/ publi cdoma in/ zero/1. 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Sabounchi et al. Health Research Policy and Systems (2022) 20:5 Page 2 of 10 for States program, an SD approach was employed to knowledge of the GSL, the likelihood of calling 911 was evaluate the impact of Connecticut’s Good Samaritan three times as high as in events where the bystander did law (GSL) by focusing on the following three health out- not know about the GSL [16]. Thus, it might not be sur - comes: (1) emergency department (ED) visits for drug prising that enacting the GSL in Connecticut has not yet overdose, (2) behavioral changes in bystanders, and (3) resulted in a significant reduction in the number of opi - overdose deaths. oid overdose deaths [17]. Opioid-related overdoses are now the leading cause of In order to better understand the rise in fatalities and preventable death in the United States [5]. The magnitude the impact of the GSL in this complex environment, we of this public health problem is illustrated by the fact that applied an SD approach to account for the numerous fac- the United States, with only 4% of the world’s population, tors that have moderated the impact of the GSL in Con- accounts for 27% of the world’s opioid-involved deaths necticut and to predict the future effectiveness of GSLs. [6, 7]. In an effort to reduce the number of opioid-related deaths, almost all states have enacted some form of a Methods GSL. These GSLs are intended to provide legal protec - According to Homer and Hirsch, “[a] system dynamics tion against liability and arrest for bystanders who give model consists of an interlocking set of differential and assistance during an overdose incident by either calling algebraic equations developed from a broad spectrum of 911 or administering naloxone, in addition to protecting relevant measured and experiential data” [18]. SD mod- first responders and individuals who prescribe naloxone. eling is of particular importance to policymakers because Connecticut’s GSL was originally passed in 2011 and has it helps map out the components of health and preven- been updated and expanded on a yearly basis since 2014 tion systems, explores their interactions, and identifies [8]. policy options that support the most efficient and effec - State-level ecological research has shown a 14–15% tive arrangements of multiple elements within a system lower incidence of opioid overdose deaths in states with [3]. Recently, Homer and Wakeland [19] used an SD GSLs compared to those without these laws [9]. Accord- model to study the United States opioid epidemic and ing to one study, GSLs are necessary in order to encour- reflect upon the unintended consequences of interven - age help-seeking and lifesaving interventions in the event tion effects on opioid use disorder (OUD) and overdose of an overdose; however, GSLs may be challenging to deaths. implement [10]. Moreover, while numerous studies have For the purposes of this analysis, we have developed found that fear of police interactions [11, 12] has been and simulated an SD model using Vensim DSS software, the primary deterrent to people calling 911 during over- version 8.2.1 [20]. The SD modeling approach incorpo - doses, many other factors have also been found to influ - rated measurement of multiple factors and their simulta- ence bystanders. For instance, some people fear that neous variance in order to determine the effectiveness of interactions with law enforcement might jeopardize their the GSL in Connecticut. These factors include the num - housing stability [12] or their employment [11]. People ber of ED visits for opioid drug overdose; the number may also worry about having Child Protective Services of people using illicit drugs and misusing prescription contacted following an overdose with which law enforce- drugs; the number of opioid-involved overdose deaths; ment officers were involved [12]. and the behavioural changes in bystanders, including the In addition to fearing police interactions, a number of number of police officers and members of the public who studies have found that lack of awareness about existing have GSL knowledge. While previous studies have used GSLs is one of the main factors limiting their impact [13]. surveys, questionnaires, and participant interviews to Moreover, even people who know about GSLs are often allow researchers to evaluate the effectiveness of the GSL, still hesitant to call 911 because they are unsure about SD modeling can provide a more structured understand- the specific protections afforded by the law. For example, ing of the effectiveness of the GSL by describing the feed - some states require a review of an individual’s criminal back loops and endogenous sources of system behavior background in order to determine eligibility for immu- that other modes of analysis are not designed to identify. nity [14]. Unfortunately, these details are often unknown in the midst of an overdose, leading to reluctance to call The model emergency medical services. This is a serious barrier The model structure was developed and validated by to the full implementation of harm reduction policies involving several key stakeholders, including staff from because, according to one study, bystander participation the Connecticut Department of Public Health (CT DPH), is necessary during overdose events if help is to be sum- researchers from Yale University, and members of local moned [15]. In fact, the results of another study showed county health departments, during two participatory that, for overdose events where bystanders had proper group model-building (GMB) sessions with the goal of S abounchi et al. Health Research Policy and Systems (2022) 20:5 Page 3 of 10 developing a concept model that would serve as the focus sources of opioids [22, 23]. In the model (Fig.  1), the for the rest of the SD modeling work. Participatory SD assumption was made that, as people who misuse pre- modeling was used to engage stakeholders in systems scription drugs switch to illicit drugs, they would be conceptualization and visual mapping of the dynamics counted as part of the people with illicit drug use disor- that determine community-level opioid-related outcomes der who also misuse prescription drugs group, which is and to identify those dynamics that could be leveraged consistent with the nomenclature and definition for illicit for systems improvement [21]. The concept model devel - drug use as provided by the Substance Abuse and Mental oped within the GMB sessions incorporated overdose Health Services Administration (SAMHSA) [24–26]. deaths and behavioral change in bystanders to study the In Fig.  1, the model components are separated by dot- impact of the Connecticut GSL and served as an impor- ted boundary lines. Located in the upper left portion of tant transitional product that allowed us to incorporate Fig.  1, section A depicts the part of the model that cap- other data sources and perform iterative simulations. tures the change in the number of people being pre- While many factors contribute to both prescription scribed opioids over time (i.e. number of opioid analgesic and illicit drug use, the change in the overall number Rx per 100 Connecticut residents per month). In a man- of opioid drug prescriptions, as well as the rate of this ner similar to the sharp increase in the number of opioid change, certainly impacts the risk of initiation of drug analgesic Rx provided in the mid- to late 1990s that con- misuse. Furthermore, illicit and prescription drug use are tributed to a significant increase in the number of people both affected by the amount of opioid prescribed. This who misuse prescription drugs [27], the change in “num- is evidenced by several studies which have found that, ber of opioid analgesic Rx influences section B, located while some policies lead to decreased OUD by reducing on the right-hand side of the model, which depicts the prescription supplies, other similar policies actually lead number of people who misuse prescription drugs and the to an increased use of narco-trafficked drugs like heroin number of people with illicit drug use disorder who also and fentanyl when individuals with OUD find alternative misuse prescription drugs.  However, despite the overlap InitialFraction POD Time to Change Fraction POD - Target Change Fraction POD + Number of Opioid Analgesic Rx per 100CTResidentper Change in Month Fraction POD Fraction Quitting Illicit Fraction Quitting Rx Drug Useper Month Miuseper Month LawEnforcement Officers Fractional Rate LawEnforcement Officers Risk of Initiation with Knowledgeabout GSL - (FR) Switching withoutGSL Knowledge IllicitSubstance Learning through & Naloxone Access Laws Quitting Officer Peers E Quitting Illicit Prescription Misuse Risk of Initiation TotalCTPatrol Drug Use + + Officers (C) ContactRate People with Illicit Initiate Illicit Officers Drug Use People who Drug UseDisorder Misuse whoalsoMisuse + + Fractional Rate Switch to Illicit Prescription Drugs Initiate Miuse Prescription Drugs for OD Drugs Individuals with Fractional Rate for General + + Individuals with + + Population KnowledgeofGSL OD from IllicitDrugs TotalOverdose(OD) GSLKnowledge withoutGSL & Naloxone from Rx Misuse Learning Behavioral & withoutFear Knowledge Access & with Fear TotalODYearly IllicitDrugs throughPeers + Change + Overdose Deaths + ED Visits + Overdose Deaths TotalIllicit Drug - from Misuse + Overdose (N) Total + Narcan Use + Population - - + (C) ContactRate Time Delay in Risk of Overdose ED visitrate + TotalArrests Behavioral Change Death from Misuse D + (ROD) Risk of IllicitDrug 911Calls and Overdose Death - AMRAssistance TotalFear of FR Narcan with ODs calling911 Probability of + + Learning GSLfrom Peers TotalOverdose Monthly Deaths Situational Awareness Perception of Drug Risk Net Change In The Minimum Number Of Perception of Drug DeathToGet Noticed Risk AverageTimeTo Perceive Risk Fig. 1 Simplified illustration of model Sabounchi et al. Health Research Policy and Systems (2022) 20:5 Page 4 of 10 between prescription opioid use and illicit drug use, most GSL from peers). Through these interactions, section E patients with OUD do not necessarily become addicted captures the changes in the number of law enforcement from starting with prescription opioids. Moreover, anal- officers with GSL knowledge, in addition to capturing a ysis of opioid overdose data performed by the CDC’s corresponding change in the number of drug-related Injury Center shows that the second wave of overdose arrests. Conversely, the number of arrests can influence deaths in 2010 involved heroin use, while the third wave, the willingness of bystanders to contact law enforce- which started in 2013, involved synthetic opioids such as ment for help in the event of an overdose. In this way, the illicitly manufactured fentanyl [28]. model demonstrates the means through which people in u Th s, this portion of the model illustrates the number Connecticut are either more or less inclined to call 911 of people who initiate either prescribed or illicit drug use, and take advantage of the protection afforded by the GSL. the number of people who transition from prescription The interaction between fear of police interactions and to illicit drug use, the number of people who quit either knowledge of the GSL is incorporated into the center of type of use by getting into drug treatment programs, and the model in section D, which contains variables for the finally, the number of people who die from overdoses. general population without GSL knowledge, individuals While incorporation of these elements into the SD with knowledge of the GSL and naloxone access and with framework is crucial for its proper functioning, this anal- fear of calling 911, and individuals with GSL knowledge ysis is most concerned with the portions of the model and naloxone access and without fear of calling 911. Just that reflect the effectiveness of the GSL in reducing over - as police officers are made more or less aware of the risks dose deaths. Located at the bottom of Fig.  1, section C of opioid use, harm reduction policies, and naloxone indicates that the net change in the perception of drug risk access through changes in situational awareness, the gen- is influenced by the number of opioid overdose deaths. eral population’s knowledge is also affected. Furthermore, Community members take notice of overdoses and may the rate at which members of the general public and begin thinking about ways to prevent overdose deaths. police officers learn about GSLs is also dependent upon The perceived risk of drug use, in conjunction with the the contact rate in a given region, meaning that more parameter value representing the length of time over interactions throughout the day with people with knowl- which this perception develops (Table 1), impacts the sit- edge of the GSL may result in more people learning about uational  awareness of opioid use. In this way, the model the GSL. The parameter value that addresses contact demonstrates how an individual’s knowledge of the GSL rates can be adjusted to reflect the population density in is either unaffected or improved over time. a region. Section D also shows that, while knowledge of Situational awareness has a clear impact on changes the GSL may increase quickly, it takes time to quell the in behaviour and, therefore, is directly connected to the fear of calling 911 and, thereby, modify bystander behav- left-hand side of the model (sections D and E). In other iour. However, once fear drops, the number of 911 calls words, as the number of people dying from opioid over- and emergency medical services staff arriving at overdose dose changes, the perception of the risk of drug use influ - events will increase. Section D of the model demonstrates ences awareness in a region, determining how much law that, as people are more or less afraid of calling 911 dur- enforcement staff and the public learn about the GSL ing opioid overdoses, the number of 911 calls decreases through peer interactions (i.e.  probability of learning or increases, respectively, thereby impacting the number Table 1 Parameter values Parameter (definition and unit of analysis) Value in the model 95% confidence interval Average time to perceive risk (months) 13.1131 12–17.7579 Risk of initiation of illicit substance (fraction of susceptible users initiating illicit drug use per month) 0.0024 0.0019–0.0028 ED visit rate (average number of times that people who misuse Rx or with illicit drug use disorder 1.37808 1.2096–1.5465 visit ED) Fraction quitting Rx misuse (fraction of users misusing Rx who quit per month) 0.002 0.0015–0.002 Fraction quitting illicit drug use (fraction of users using illicit drugs/misusing Rx who quit per month) 0.0007 0.0003–0.0012 Fractional rate for overdose among nonmedical users of Rx drugs (users per month) 0.0011 0.0011–0.0011 Fractional rate for overdose from illicit drugs and Rx misuse (users per month) 0.0037 0.0037–0.0037 Risk of overdose death from Rx misuse (fraction of Rx misuse OD incidence that leads to death) 0.00001 0.00001–0.0126548 Risk of overdose death from illicit drug use and Rx misuse (fraction of illicit drug OD incidence that 0.2072 0.2021–0.2125 leads to death) S abounchi et al. Health Research Policy and Systems (2022) 20:5 Page 5 of 10 of overdoses during which naloxone is administered, 911 was provided by the following two survey reports: with a differential impact on mortality appreciated. The (1) the CT DPH and Central Connecticut State Univer- detailed model formulations are provided in the online sity’s (CCSU) survey on basic understanding of the GSL supplementary information  (Additional file 1). and the corresponding fear of calling 911 [37] and (2) the High Intensity Drug Trafficking Area’s (HIDTA) Heroin Data and model calibration Response Strategy project report on the GSL’s impact on Since most of the parameters defined in the model were Connecticut policing practices [38]. not available in the relevant literature, model calibra- tion was used to make estimates for the parameter values Modelling and simulation results shown in green in Fig.  1. Constraints on plausible values SD modeling has allowed the research team to capture of the calibrated parameters listed in Table 1 were formu- the complex interrelationships among several key health lated from expert opinion and the literature [29–32]. A outcome measures that drive the opioid epidemic in full list of calibrated parameter values is provided in the Connecticut. These outcomes include ED visits due to online supplementary information (the Additional file  1). overdose, behavioral changes in bystanders, changes in Calibration was performed using Vensim DSS, version perception of the risk of drug use, awareness of harm 8.2.1 [20]. The calibration module in Vensim modeling reduction policies, and overdose deaths. software calculates the optimum values of model param- First, the simulated number of overdose deaths aligns eters using a maximum likelihood estimation approach very closely with data supplied by the OCME. Unfortu- [33] to create simulations which align best with real- nately, these results reveal that the number of overdose world data, replicate historical trends, and create estima- deaths has been increasing over the past 8 years and will tions for the future. continue to grow if no additional action is taken (Fig. 2a). The data used to calibrate this model were collected Additionally, the simulated number of ED visits for over- from numerous sources over different intervals and dose aligns with the data supplied by CHIME discharge describe the status of opioid use in Connecticut from data, which shows that as the number of opioid users 2009 to 2018. Information about overdose deaths was and overdose incidents increases, ED visits also increase collected from the Connecticut Office of the Chief Medi - (Fig. 2b). cal Examiner (OCME). The data concerning the number Secondly, the model demonstrates that, as more opioids of ED visits for overdose came from Connecticut Hospi- are prescribed, the use of illicit and prescription drugs tal Association (CHIME) discharge data. The researchers increases. The simulation indicates that, as more peo - also utilized information from www. CTData. org, includ- ple misuse prescription drugs, an increased number of ing the amount of illicit drug use other than marijuana people switch from prescribed to illegal opioids. For this between 2008 and 2014 (collected by SAMHSA as part of reason, the number of people who misuse prescription the National Survey on Drug Use and Health [NSDUH]) drugs but do not use illicit drugs was shown to decrease [34] and the rate of arrests due to drug law offences from around 2013; and the number of people who use illicit 2010 to 2016 [35]. The most recent SAMHSA reports drugs and also misuse prescription drugs was shown to from 2015 to 2018 provide information about illicit drug increase during the same time period (Fig. 2c). It is inter- use and prescription drug misuse in Connecticut, and esting to note that, around 2013, the rate at which opi- this information was used to validate the simulation oid prescriptions were being written in Connecticut was results [24–26]. Specifically, the illicit drug use described decreasing (Fig.  2d). Consequently, this model supports in the SAMHSA report includes the misuse of prescrip- the hypothesis that an increase in deaths and subsequent tion psychotherapeutics, and the research team used situational awareness could lead to decreased opioid this information to validate the total sum of people who prescriptions; and the corresponding link depicted in the misuse prescription drugs and the number of people model has been shown, thus far, to be an effective means with illicit drug use disorder who also misuse prescription of harm reduction (Fig. 1). drugs in the model. However, there is a mismatch between simulation The rate of opioid prescriptions per 100 Connecticut results and the data points corresponding to illicit drug residents was retrieved from the CDC report on United use and misuse of prescription drugs prior to 2014 States state prescribing rates [36]. Additionally, informa- (Fig.  2e). The simulation results show that the total tion concerning the rate of administration and dose of number of people who use illicit drugs and/or misuse naloxone used was supplied by the American Medical prescription drugs (i.e., the sum of people who mis- Response (AMR) transportation company, which serves use prescription drugs and the number of people with a large portion of the state of Connecticut. Lastly, the illicit drug use disorder who also misuse prescription information on knowledge of the GSL and fear of calling drugs in the model) has increased overall, but data show Sabounchi et al. Health Research Policy and Systems (2022) 20:5 Page 6 of 10 Overdose Deaths (Total Number per Month) Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 MonthlyOve rdoseDeath s- Simulate d MonthlyOve rdoseDeath s- OCME Data Part a Overdose EDVisits (Total Number per Year) Illicit Drug UseandNon-Medical Use of Prescription Drugs 8000 140000 6000 100000 80 000 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 Number of People whoMisusePrescript ionDrugs an dnot UseIllicit Drugs- Simulated Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 Number of People with IllicitDrugUse Disorder whoalsoMisusePre scriptionDrugs-Simulated Number of Overdose ED Visits perYear- Simulated Number of Overdose ED Visits perYear- CHIMEData Part c Part b Number of Peoplewith Illicit DrugUse andthe misuse of prescription psychotherapeutics or theuse of cocaine(includingcrack), heroin,hallucinogens, inhalants, or methamphetamine -CTDATA .org & SAMHSA Data (National Opioid Analgesic Prescriptions (Number per100 Connecticut Residentper Survey on Drug Use and Health) Month) 5 115000 2 95000 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 Number of People with IllicitDrugUse Ot herthanM arijuana an dmisuseofprescriptions Number of People with IllicitDrugUse an dthe misuse of prescriptions Number of Opioid AnalgesicRxper Month- Simulated Number of Opioid AnalgesicRxper Month- CPMR S& DCPData Part e Part d Fig. 2 Simulation results for overdose deaths and drug use a decrease between 2009 and 2013. One explanation collection methods before and after 2014. However, the is that, although the data sources for illicit drug use are SAMHSA reports were the most reliable data source all based on SAMHSA reports, the SAMHSA NSDUH available for our modeling purposes and, hence, were uti- reports introduced an independent multistage area prob- lized in this study. ability sample as the first level of stratification in 2014 Also, the model predicts that GSL knowledge will within each state. Thus, the data points between 2015 continue to grow from the 2017 estimates that indicate and 2018 are based on an updated data collection pro- that 60% of the general public and 74% of police officers cess; and there are likely inconsistencies between data in Connecticut knew about the GSL [37, 38] (Fig.  3b, c). S abounchi et al. Health Research Policy and Systems (2022) 20:5 Page 7 of 10 This is a prediction based on the model structure and same direction. As more people perceive the high risk estimates, not related to a specific educational plan. As of drug use, they become more situationally aware. As a shown in Fig. 1, the model structure shows a positive link consequence, the probability of learning about the GSL from perception of drug risk to situational awareness for from peers increases, which contributes to an increased drug risk. A positive link describes a causal relationship number of law enforcement officers and members of the in an SD model when the cause and effect change in the general public who learn about the law. We calibrated Narcan Use Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 Number of EmergencyIncidents with Narc an Adminsitration -Simulated Number of EmergencyIncidents with Narc an Adminsitration -AMR Data Part a Officers CorrectKnowledge of GSLabout providingimmunityfromcriminal BasicUnderstanding of GSLamong Public charges 1 0.9 0.9 0.8 0.8 0.7 0.7 0.6 0.6 0.5 0.5 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 % of GeneralPopulationwithGSL Knowle dge- Simulated % of GeneralPopulationwithGSL Knowle dge- CT DPH& CCSU Data % of LawEnforceme nt Office rs with CorrectGSL Kn owledge- Simulated % of LawEnforceme nt Office rs with CorrectGSL Kn owledge- HIDTAData Part b Part c Rate of Drug Arrests(per10,000Connecticut Residentsper Year) Rate of 911Calls forOverdoses (FractionofIncidents that areCalled) 4.5 0.8 0.7 3.5 0.6 2.5 0.5 0.4 1.5 0.3 0.2 0.5 0.1 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 Rate of Drug Arrests- Simulated Rate of Drug Arrests- CTDATA.org Data Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 Fraction of Overdose incidentsthatresultin911 Calls Part d Part e Fig. 3 Simulation results for behavioral changes Sabounchi et al. Health Research Policy and Systems (2022) 20:5 Page 8 of 10 the model to identify the strength of these causal links situationally relevant and may be used to evaluate what-if by estimating the parameter values highlighted in green simulation scenarios. Policy  makers may use this model in Fig. 1, including the average time that it takes for GSL to test new interventions that might be used to address knowledge to change and the threshold number of over- the opioid crisis. Additionally, once more robust data on dose fatalities at which changes in awareness towards the the behavioral impact of the GSL become available, those risk of drug use and the benefit of GSLs will occur. data can be used to produce an even more reliable model Policy  makers may infer from the model that, because in the future. the number of overdose deaths is increasing, situational While existing data show that the GSL has not yet awareness will change, leading to an increase in officer reduced the number of overdose deaths, the model’s and public awareness of the GSL. Subsequently, the num- simulations indicate that the high number of deaths will ber of people with GSL knowledge and without fear of likely foster an increased awareness of the GSL, leading calling 911 is anticipated to increase. Also, the rate of to decreased fear of calling 911 and increased naloxone drug arrests per 10,000 for drug law offences has been administration. However, the model also  suggests that decreasing, possibly due to the growth of GSL awareness, the overall trend of increased deaths may continue to but is predicted to remain stable in the future (Fig. 3d). grow despite this increased awareness. This prediction is Finally, the results of the simulation that represent the supported by the model, as well as by many other studies, use of naloxone correspond to the data received from indicating that fear of police interactions is the primary AMR, which show an overall increase in the number of reason that bystanders do not call 911. overdose events during which naloxone is administered Fortunately, the model demonstrates that interven- (Fig. 3a). The model explains that this increase is partially tions like the GSL, which protects bystanders against due to an increase in overdose incidents since 2009 and liability for providing assistance during overdoses, may partially due to a similar increase in the number of 911 represent a partial solution to this problem. However, calls since the start of the simulation (Fig.  3e). However, additional interventions are needed to improve the effec - even accounting for an increase in emergency medical tiveness of the GSL. For example, although the model’s assistance and administration of naloxone, the number of results indicate that the rate of drug-related arrest has overdose deaths has continued to increase (Fig.  2a); and slowed in recent years, future interventions, such as risk of overdose remains stable (Table 1), contrary to our increased training for police officers, may still be neces - initial hypothesis that had assumed negative causal links sary. One study claims that GSLs and other harm reduc- from Narcan use to risk of overdose death from misuse tion policies are necessary but insufficient, primarily due (ROD) and risk of illicit drug overdose death in Fig. 1. The to problems with implementation and awareness [10]. increase seen over time in the dose of naloxone adminis- While efforts to alter the course of the opioid epidemic tered for an overdose from an average dose of 1.18 mg in will require ongoing research concerning the numerous 2010 to 1.94  mg in 2018 is based on data received from interventions that could be applied to this problem, this AMR and may be related to the high potency of the new analysis illustrates how SD modeling may be beneficial illicit drugs that have become available over the past few in aiding policy  makers who are tasked with decision- years. making in the setting of complex challenges. Because more time must pass in order to observe the long-term Discussion effectiveness of the GSL, an SD modeling approach can SD modeling is an analytical tool that helps policy  mak- be used to make predictions about its long-term impact ers approach difficult decisions in the presence of the by employing a simulation framework. Additionally, uncertainties that complex problems create [3]. The SD since the model structure and feedback loops are relevant approach is ideal for evaluating the delayed impact of the to the opioid epidemic in general, the model parameters GSL on behavioral changes because it allows research- can be calibrated towards historical data trends for other ers to investigate the long-term effects of policy inter - geographical regions or states and, thus, be customized ventions using a simulation framework. Because of the for different locations and settings. In this way, pol - ability to simulate with SD, policy makers can infer that, icy makers can utilize this model to test future trends and although the evidence currently demonstrates mixed determine the best solutions for various public health effects of the GSL, the overall amount of drug use and problems. number of overdose deaths will increase if no additional policies are implemented. Limitations Moreover, because the model’s simulation results align This SD model included feedback processes and dynam - with real-world data and can be used to replicate his- ics important to understanding the impact of GSLs and torical trends, it is reasonable to infer that the model is providing insight for policy  makers and public health S abounchi et al. Health Research Policy and Systems (2022) 20:5 Page 9 of 10 Acknowledgements officials. While this model provides a useful foundation We would like to thank Dr. Brian J. Biroscak, PhD, MS, MA, at the Weitzman for answering targeted research questions about GSLs Institute Community Health Center, Inc., and Dr. Jason Buckert, MD, for their and related policies, our analysis had several limitations. advice and contributions to this study. First, identifying relevant existing data sources for some Authors’ contributions model indicators that span the time horizon of the study Nasim S. Sabounchi, Rebekah Heckmann, Gail D’Onofrio, and Robert Heimer has been challenging. Also, some data sources, such as originated the study, contributed to the concept and design, and supervised all aspects of its implementation. Nasim S. Sabounchi and Rebekah Heck- the SAMHSA reports [24–26], have updated their data mann developed the system dynamics simulation model and processed and collection process, which has likely led to inconsisten- analyzed the simulation results. Jennifer Walker contributed to the drafting cies in data collection methods. While the current model of the manuscript, and all authors contributed to the critical revision of the manuscript for important intellectual content. All authors read and approved structure and feedback loops appear to replicate histori- the final manuscript. cal trends well, future iterations of the model may test the alignment of current or additional feedback loops. Funding This study was funded under contract to the Connecticut Department of Pub- Additionally, the model was designed based upon expert lic Health (DPH) on behalf of the CDC Prescription Drug Overdose Prevention input and GMB sessions conducted with key stakehold- for States (CDC Grant Number: 1U17CE002720-02). The content presented ers. A next step could adopt a more inclusive approach, does not necessarily reflect DPH or CDC policy. and other relevant stakeholders such as patients, provid- Availability of data and materials ers, law enforcement officers, and first responders could All data generated or analyzed during this study are included in this published be invited to contribute to the modeling process and cor- article and its additional information file. responding  validation while exploring specific questions related to the impact of opioid policies on reducing over- Declarations dose risk and fatality. Ethics approval and consent to participate Not applicable. Consent for publication Conclusions Not applicable. SD modeling has been proven to be a useful approach for assessing the effectiveness of public health policy Competing interests The authors declare that they have no competing interests. interventions through its utilization of a simulation framework. While other analytical methods may require Author details research involving study participants and clinical tri- Department of Health Policy and Management, Center for Systems and Com- munity Design, City University of New York (CUNY ) Graduate School of Public als, SD modeling allows for the prediction of the future Health and Health Policy, 55 W. 125th Street, New York, NY 10027, United effectiveness of interventions through its ability to rep - States of America. Department of Emergency Medicine, Yale University licate historical trends. While investigating the impact School of Medicine, New Haven, CT, United States of America. Bingham- ton University-State University of New York, Binghamton, NY, United States of the GSL on overdose deaths, ED visits and bystander of America. Department of Epidemiology of Microbial Diseases, Yale Univer- behavior in Connecticut is the main purpose of this anal- sity School of Public Health, New Haven, CT, United States of America. ysis, this model has also demonstrated great potential by Received: 21 May 2021 Accepted: 9 December 2021 producing simulations that reveal multiple strategies to aid policy  makers in determining the best public health interventions for combating the opioid crisis. References 1. Bose J, Hedden SL, Lipari RN, Park-Lee E. Key substance use and mental Abbreviations health indicators in the United States: results from the 2017 National GSL: Good Samaritan law; SD: System dynamics; SAMHSA: Substance Abuse Survey on Drug Use and Health. Rockville: Center for Behavioral Health and Mental Health Services Administration; ED: Emergency department; OUD: Statistics and Quality, Substance Abuse and Mental Health Services Opioid use disorder; CDC: Centers for Disease Control and Prevention; AMR: Administration; 2018. Contract No.: No. SMA 18-5068. American Medical Response; GMB: Group model-building. 2. Scholl L, Seth P, Kariisa M, Wilson N, Baldwin G. Drug and opioid-involved overdose deaths—United States, 2013–2017. Morb Mortal Wkly Rep. Supplementary Information 2019;67(5152):1419. 3. Sterman JD. Learning from evidence in a complex world. Am J Public The online version contains supplementary material available at https:// doi. Health. 2006;96(3):505–14. org/ 10. 1186/ s12961- 021- 00807-w. 4. Atkinson J-A, Wells R, Page A, Dominello A, Haines M, Wilson A. Applica- tions of system dynamics modelling to support health policy. Public Health Res Pract. 2015;25(3):e2531531. Additional file 1. Section 1: Model Formulation. Section 2: Estimation 5. Watson DP, Ray B, Robison L, Huynh P, Sightes E, Brucker K, et al. Lay and Calibration of Model Parameters. responder naloxone access and Good Samaritan law compliance: postcard survey results from 20 Indiana counties. Harm Reduct J. 2018;15(1):18. Sabounchi et al. Health Research Policy and Systems (2022) 20:5 Page 10 of 10 6. Gomes T, Tadrous M, Mamdani MM, Paterson JM, Juurlink DN. The burden and Medicine Division; Board on Health Sciences Policy; Committee on of opioid-related mortality in the United States. JAMA Netw Open. Pain Management and Regulatory Strategies to Address Prescription 2018;1(2):e180217-e. Opioid Abuse; 2017. 7. WHO. Information sheet on opioid overdose. Geneva: World Health 28. CDC. Opioid data analysis and resources. Atlanta: Centers for Disease Organization; 2018. Control and Prevention, National Center for Injury Prevention and Con- 8. DMHAS. Opioid overdose prevention/naloxone (Narcan) initiative. trol; 2021. Hartford: Connecticut Department of Mental Health & Addiction Services; 29. Heimer R, Barbour R, Palacios WR, Nichols LG, Grau LE. Associations 2019. between injection risk and community disadvantage among suburban 9. McClellan C, Lambdin BH, Ali MM, Mutter R, Davis CS, Wheeler E, et al. injection drug users in southwestern Connecticut, USA. AIDS Behav. Opioid-overdose laws association with opioid use and overdose mortal- 2014;18(3):452–63. ity. Addict Behav. 2018;86:90–5. 30. Krawczyk N, Eisenberg M, Schneider KE, Richards TM, Lyons BC, Jackson K, 10. Banta-Green CJ, Beletsky L, Schoeppe JA, Coffin PO, Kuszler PC. Police et al. Predictors of overdose death among high-risk emergency depart- officers’ and paramedics’ experiences with overdose and their knowledge ment patients with substance-related encounters: a data linkage cohort and opinions of Washington State’s drug overdose–naloxone–Good study. Ann Emerg Med. 2020;75(1):1–12. Samaritan law. J Urban Health. 2013;90(6):1102–11. 31. Larochelle MR, Bernson D, Land T, Stopka TJ, Wang N, Xuan Z, et al. Medi- 11. Koester S, Mueller SR, Raville L, Langegger S, Binswanger IA. Why are cation for opioid use disorder after nonfatal opioid overdose and associa- some people who have received overdose education and naloxone tion with mortality: a cohort study. Ann Intern Med. 2018;169(3):137–45. reticent to call Emergency Medical Services in the event of overdose? Int 32. Weiner SG, Baker O, Bernson D, Schuur JD. One-year mortality of patients J Drug Policy. 2017;48:115–24. after emergency department treatment for nonfatal opioid overdose. 12. Latimore AD, Bergstein RS. “Caught with a body” yet protected by law? Ann Emerg Med. 2020;75(1):13–7. Calling 911 for opioid overdose in the context of the Good Samaritan 33. Dogan G. Bootstrapping for confidence interval estimation and hypoth- Law. Int J Drug Policy. 2017;50:82–9. esis testing for parameters of system dynamics models. Syst Dyn Rev. 13. SAMHSA. Preventing the consequences of opioid overdose: understand- 2007;23(4):415–36. ing 911 Good Samaritan Laws. Substance Abuse and Mental Health 34. CTData. Illicit drug use reports the prevalence of the consumption of Services Administration’s Center for the Application of Prevention Tech- illicit substances by age range. Connecticut Data Collaborative; 2020. nologies task order; 2017.http:// data. ctdata. org/ datas et/ illic it- drug- use. 14. Hawk KF, Vaca FE, D’Onofrio G. Focus: addiction: reducing fatal opioid 35. CTData. Drug arrests reports the number and rate (per 10,000) of arrests overdose: prevention, treatment and harm reduction strategies. Yale J due to drug law offenses, per age range. Connecticut Data Collaborative; Biol Med. 2015;88(3):235. 2020. http:// data. ctdata. org/ datas et/ drug- arres ts. 15. Karamouzian M, Kuo M, Crabtree A, Buxton JA. Correlates of seeking 36. CDC. U.S. state prescribing rates: Centers for Disease Control and Preven- emergency medical help in the event of an overdose in British Columbia, tion, National Center for Injury Prevention and Control; 2017. https:// Canada: findings from the Take Home Naloxone program. Int J Drug www. cdc. gov/ drugo verdo se/ maps/ rxsta te2009. html. Policy. 2019;71:157. 37. Hawk KF, Doernberg M, D’Onofrio G, Heimer R, Diaz-Matos LF, Jenkins M, 16. Jakubowski A, Kunins HV, Huxley-Reicher Z, Siegler A. Knowledge of the et al. Knowledge of Connecticut’s Good Samaritan Law among connecti- 911 Good Samaritan law and 911-calling behavior of overdose witnesses. cut residents who have witnessed an overdose. 2019. Subst Abuse. 2018;39(2):233–8. 38. HIDTA. Heroin Response Strategy—911 Good Samaritan Law Corner- 17. Backus L. Fatal drug overdoses rising in CT: Town-by-town view of 2019 stone Project—Preliminary State Report Connecticut. High Intensity Drug deaths: ctpost; 2020. https:// www. ctpost. com/ local/ artic le/ Fatal- drug- Trafficking Areas (HIDTA) Program—WHite House Office of National Drug overd oses- rising- in- CT- Town- by- town- 15066 922. php. Control Policy; 2019. 18. Homer JB, Hirsch GB. System dynamics modeling for public health: back- ground and opportunities. Am J Public Health. 2006;96(3):452–8. Publisher’s Note 19. Homer J, Wakeland W. A dynamic model of the opioid drug epidemic Springer Nature remains neutral with regard to jurisdictional claims in pub- with implications for policy. Am J Drug Alcohol Abuse. 2020;47:5–15. lished maps and institutional affiliations. 20. Vensim. Ventana Systems, Inc.; 2021. 21. Hovmand PS, Andersen DF, Rouwette E, Richardson GP, Rux K, Calhoun A. Group model-building ‘Scripts’ as a collaborative planning tool. Syst Res Behav Sci. 2012;29(2):179–93. 22. Pitt AL, Humphreys K, Brandeau ML. Modeling health benefits and harms of public policy responses to the US opioid epidemic. Am J Public Health. 2018;108(10):1394–400. 23. Chen Q, Larochelle MR, Weaver DT, Lietz AP, Mueller PP, Mercaldo S, et al. Prevention of prescription opioid misuse and projected overdose deaths in the United States. JAMA Netw Open. 2019;2(2):e187621-e. 24. SAMHSA. 2015–2016 national survey on drug use and health: model- based prevalence estimates (50 states and the District of Columbia). Rockville: Substance Abuse and Mental Health Services Administration (SAMHSA), Center for Behavioral Health Statistics and Quality, National Re Read ady y to to submit y submit your our re researc search h ? Choose BMC and benefit fr ? Choose BMC and benefit from om: : Survey on Drug Use and Health; 2016. 25. SAMHSA. 2017–2018 national survey on drug use and health: model- fast, convenient online submission based prevalence estimates (50 states and the district of Columbia). thorough peer review by experienced researchers in your field Rockville: Substance Abuse and Mental Health Services (SAMHSA), Center for Behavioral Health Statistics and Quality, National Survey on Drug Use rapid publication on acceptance and Health; 2018. support for research data, including large and complex data types 26. SAMHSA. 2016–2017 national survey on drug use and health: model- • gold Open Access which fosters wider collaboration and increased citations based prevalence estimates (50 states and the district of Columbia). Rockville: Substance Abuse and Mental Health Services Administration maximum visibility for your research: over 100M website views per year (SAMHSA), Center for Behavioral Health Statistics and Quality, National Survey on Drug Use and Health; 2017. At BMC, research is always in progress. 27. NASEM. Pain management and the opioid epidemic: balancing societal Learn more biomedcentral.com/submissions and individual benefits and risks of prescription opioid use. Washington (DC): National Academies of Sciences, Engineering, and Medicine; Health http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Health Research Policy and Systems Springer Journals

Assessing the impact of the Good Samaritan Law in the state of Connecticut: a system dynamics approach

Loading next page...
 
/lp/springer-journals/assessing-the-impact-of-the-good-samaritan-law-in-the-state-of-NWbrGe05wr

References (52)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2022
eISSN
1478-4505
DOI
10.1186/s12961-021-00807-w
Publisher site
See Article on Publisher Site

Abstract

Background: Although Good Samaritan laws (GSLs) have been widely adopted throughout the United States, their efficacy in individual states is often unknown. This paper offers an approach for assessing the impact of GSLs and insight for policy-makers and public health officials who wish to know whether they should expect to see outcomes from similar policy interventions. Methods: Utilizing a system dynamics (SD) modeling approach, the research team conducted a policy evaluation to determine the impact of GSLs on opioid use disorder (OUD) in Connecticut and evaluated the GSL based upon the following health outcomes: (1) emergency department (ED) visits for overdose, (2) behavioral changes of bystanders, and (3) overdose deaths. Results: The simulation model suggests that Connecticut’s GSL has not yet affected overdose deaths but has resulted in bystander behavioral changes, such as increased 911 calls for overdose. ED visits have increased as the number of opioid users has increased. Conclusions: The simulation results indicate that the number of opioid-related deaths will continue to increase and that the GSL alone cannot effectively control the crisis. However, the SD approach that was used will allow policymak - ers to evaluate the effectiveness of the GSL over time using a simulation framework. This SD model demonstrates great potential by producing simulations that allow policymakers to assess multiple strategies for combating the opioid crisis and select optimal public health interventions. Keywords: Opioid use disorder, Emergency medicine, Health policy simulation, System dynamics modelling Background and Introduction older misused opioids in 2017 [1], and more than 47, 000 The significant increase in the number of opioid over - people died from opioid overdoses in the same year [2]. dose deaths in the United States over the past few dec- As this death toll continues to rise, policy interventions ades is now widely recognized as a national public health become increasingly important as a means of reducing crisis. Almost 11.4  million Americans aged 12  years or overdose deaths, and policymakers need tools to help guide decision-making. Most importantly, SD modeling is useful for studying resistance to public health inter- ventions [3]. System dynamics (SD) modeling has gained *Correspondence: nasim.sabounchi@sph.cuny.edu Nasim S. Sabounchi and Rebekah Heckmann share first authorship momentum in the health sector due to its potential to Department of Health Policy and Management, Center for Systems address the challenges of decision-making for complex and Community Design, City University of New York (CUNY ) Graduate policy problems [4]. School of Public Health and Health Policy, 55 W. 125th Street, New York, NY 10027, United States of America As part of the Centers for Disease Control and Preven- Full list of author information is available at the end of the article tion’s (CDC) Prescription Drug Overdose Prevention © The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco mmons. org/ publi cdoma in/ zero/1. 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Sabounchi et al. Health Research Policy and Systems (2022) 20:5 Page 2 of 10 for States program, an SD approach was employed to knowledge of the GSL, the likelihood of calling 911 was evaluate the impact of Connecticut’s Good Samaritan three times as high as in events where the bystander did law (GSL) by focusing on the following three health out- not know about the GSL [16]. Thus, it might not be sur - comes: (1) emergency department (ED) visits for drug prising that enacting the GSL in Connecticut has not yet overdose, (2) behavioral changes in bystanders, and (3) resulted in a significant reduction in the number of opi - overdose deaths. oid overdose deaths [17]. Opioid-related overdoses are now the leading cause of In order to better understand the rise in fatalities and preventable death in the United States [5]. The magnitude the impact of the GSL in this complex environment, we of this public health problem is illustrated by the fact that applied an SD approach to account for the numerous fac- the United States, with only 4% of the world’s population, tors that have moderated the impact of the GSL in Con- accounts for 27% of the world’s opioid-involved deaths necticut and to predict the future effectiveness of GSLs. [6, 7]. In an effort to reduce the number of opioid-related deaths, almost all states have enacted some form of a Methods GSL. These GSLs are intended to provide legal protec - According to Homer and Hirsch, “[a] system dynamics tion against liability and arrest for bystanders who give model consists of an interlocking set of differential and assistance during an overdose incident by either calling algebraic equations developed from a broad spectrum of 911 or administering naloxone, in addition to protecting relevant measured and experiential data” [18]. SD mod- first responders and individuals who prescribe naloxone. eling is of particular importance to policymakers because Connecticut’s GSL was originally passed in 2011 and has it helps map out the components of health and preven- been updated and expanded on a yearly basis since 2014 tion systems, explores their interactions, and identifies [8]. policy options that support the most efficient and effec - State-level ecological research has shown a 14–15% tive arrangements of multiple elements within a system lower incidence of opioid overdose deaths in states with [3]. Recently, Homer and Wakeland [19] used an SD GSLs compared to those without these laws [9]. Accord- model to study the United States opioid epidemic and ing to one study, GSLs are necessary in order to encour- reflect upon the unintended consequences of interven - age help-seeking and lifesaving interventions in the event tion effects on opioid use disorder (OUD) and overdose of an overdose; however, GSLs may be challenging to deaths. implement [10]. Moreover, while numerous studies have For the purposes of this analysis, we have developed found that fear of police interactions [11, 12] has been and simulated an SD model using Vensim DSS software, the primary deterrent to people calling 911 during over- version 8.2.1 [20]. The SD modeling approach incorpo - doses, many other factors have also been found to influ - rated measurement of multiple factors and their simulta- ence bystanders. For instance, some people fear that neous variance in order to determine the effectiveness of interactions with law enforcement might jeopardize their the GSL in Connecticut. These factors include the num - housing stability [12] or their employment [11]. People ber of ED visits for opioid drug overdose; the number may also worry about having Child Protective Services of people using illicit drugs and misusing prescription contacted following an overdose with which law enforce- drugs; the number of opioid-involved overdose deaths; ment officers were involved [12]. and the behavioural changes in bystanders, including the In addition to fearing police interactions, a number of number of police officers and members of the public who studies have found that lack of awareness about existing have GSL knowledge. While previous studies have used GSLs is one of the main factors limiting their impact [13]. surveys, questionnaires, and participant interviews to Moreover, even people who know about GSLs are often allow researchers to evaluate the effectiveness of the GSL, still hesitant to call 911 because they are unsure about SD modeling can provide a more structured understand- the specific protections afforded by the law. For example, ing of the effectiveness of the GSL by describing the feed - some states require a review of an individual’s criminal back loops and endogenous sources of system behavior background in order to determine eligibility for immu- that other modes of analysis are not designed to identify. nity [14]. Unfortunately, these details are often unknown in the midst of an overdose, leading to reluctance to call The model emergency medical services. This is a serious barrier The model structure was developed and validated by to the full implementation of harm reduction policies involving several key stakeholders, including staff from because, according to one study, bystander participation the Connecticut Department of Public Health (CT DPH), is necessary during overdose events if help is to be sum- researchers from Yale University, and members of local moned [15]. In fact, the results of another study showed county health departments, during two participatory that, for overdose events where bystanders had proper group model-building (GMB) sessions with the goal of S abounchi et al. Health Research Policy and Systems (2022) 20:5 Page 3 of 10 developing a concept model that would serve as the focus sources of opioids [22, 23]. In the model (Fig.  1), the for the rest of the SD modeling work. Participatory SD assumption was made that, as people who misuse pre- modeling was used to engage stakeholders in systems scription drugs switch to illicit drugs, they would be conceptualization and visual mapping of the dynamics counted as part of the people with illicit drug use disor- that determine community-level opioid-related outcomes der who also misuse prescription drugs group, which is and to identify those dynamics that could be leveraged consistent with the nomenclature and definition for illicit for systems improvement [21]. The concept model devel - drug use as provided by the Substance Abuse and Mental oped within the GMB sessions incorporated overdose Health Services Administration (SAMHSA) [24–26]. deaths and behavioral change in bystanders to study the In Fig.  1, the model components are separated by dot- impact of the Connecticut GSL and served as an impor- ted boundary lines. Located in the upper left portion of tant transitional product that allowed us to incorporate Fig.  1, section A depicts the part of the model that cap- other data sources and perform iterative simulations. tures the change in the number of people being pre- While many factors contribute to both prescription scribed opioids over time (i.e. number of opioid analgesic and illicit drug use, the change in the overall number Rx per 100 Connecticut residents per month). In a man- of opioid drug prescriptions, as well as the rate of this ner similar to the sharp increase in the number of opioid change, certainly impacts the risk of initiation of drug analgesic Rx provided in the mid- to late 1990s that con- misuse. Furthermore, illicit and prescription drug use are tributed to a significant increase in the number of people both affected by the amount of opioid prescribed. This who misuse prescription drugs [27], the change in “num- is evidenced by several studies which have found that, ber of opioid analgesic Rx influences section B, located while some policies lead to decreased OUD by reducing on the right-hand side of the model, which depicts the prescription supplies, other similar policies actually lead number of people who misuse prescription drugs and the to an increased use of narco-trafficked drugs like heroin number of people with illicit drug use disorder who also and fentanyl when individuals with OUD find alternative misuse prescription drugs.  However, despite the overlap InitialFraction POD Time to Change Fraction POD - Target Change Fraction POD + Number of Opioid Analgesic Rx per 100CTResidentper Change in Month Fraction POD Fraction Quitting Illicit Fraction Quitting Rx Drug Useper Month Miuseper Month LawEnforcement Officers Fractional Rate LawEnforcement Officers Risk of Initiation with Knowledgeabout GSL - (FR) Switching withoutGSL Knowledge IllicitSubstance Learning through & Naloxone Access Laws Quitting Officer Peers E Quitting Illicit Prescription Misuse Risk of Initiation TotalCTPatrol Drug Use + + Officers (C) ContactRate People with Illicit Initiate Illicit Officers Drug Use People who Drug UseDisorder Misuse whoalsoMisuse + + Fractional Rate Switch to Illicit Prescription Drugs Initiate Miuse Prescription Drugs for OD Drugs Individuals with Fractional Rate for General + + Individuals with + + Population KnowledgeofGSL OD from IllicitDrugs TotalOverdose(OD) GSLKnowledge withoutGSL & Naloxone from Rx Misuse Learning Behavioral & withoutFear Knowledge Access & with Fear TotalODYearly IllicitDrugs throughPeers + Change + Overdose Deaths + ED Visits + Overdose Deaths TotalIllicit Drug - from Misuse + Overdose (N) Total + Narcan Use + Population - - + (C) ContactRate Time Delay in Risk of Overdose ED visitrate + TotalArrests Behavioral Change Death from Misuse D + (ROD) Risk of IllicitDrug 911Calls and Overdose Death - AMRAssistance TotalFear of FR Narcan with ODs calling911 Probability of + + Learning GSLfrom Peers TotalOverdose Monthly Deaths Situational Awareness Perception of Drug Risk Net Change In The Minimum Number Of Perception of Drug DeathToGet Noticed Risk AverageTimeTo Perceive Risk Fig. 1 Simplified illustration of model Sabounchi et al. Health Research Policy and Systems (2022) 20:5 Page 4 of 10 between prescription opioid use and illicit drug use, most GSL from peers). Through these interactions, section E patients with OUD do not necessarily become addicted captures the changes in the number of law enforcement from starting with prescription opioids. Moreover, anal- officers with GSL knowledge, in addition to capturing a ysis of opioid overdose data performed by the CDC’s corresponding change in the number of drug-related Injury Center shows that the second wave of overdose arrests. Conversely, the number of arrests can influence deaths in 2010 involved heroin use, while the third wave, the willingness of bystanders to contact law enforce- which started in 2013, involved synthetic opioids such as ment for help in the event of an overdose. In this way, the illicitly manufactured fentanyl [28]. model demonstrates the means through which people in u Th s, this portion of the model illustrates the number Connecticut are either more or less inclined to call 911 of people who initiate either prescribed or illicit drug use, and take advantage of the protection afforded by the GSL. the number of people who transition from prescription The interaction between fear of police interactions and to illicit drug use, the number of people who quit either knowledge of the GSL is incorporated into the center of type of use by getting into drug treatment programs, and the model in section D, which contains variables for the finally, the number of people who die from overdoses. general population without GSL knowledge, individuals While incorporation of these elements into the SD with knowledge of the GSL and naloxone access and with framework is crucial for its proper functioning, this anal- fear of calling 911, and individuals with GSL knowledge ysis is most concerned with the portions of the model and naloxone access and without fear of calling 911. Just that reflect the effectiveness of the GSL in reducing over - as police officers are made more or less aware of the risks dose deaths. Located at the bottom of Fig.  1, section C of opioid use, harm reduction policies, and naloxone indicates that the net change in the perception of drug risk access through changes in situational awareness, the gen- is influenced by the number of opioid overdose deaths. eral population’s knowledge is also affected. Furthermore, Community members take notice of overdoses and may the rate at which members of the general public and begin thinking about ways to prevent overdose deaths. police officers learn about GSLs is also dependent upon The perceived risk of drug use, in conjunction with the the contact rate in a given region, meaning that more parameter value representing the length of time over interactions throughout the day with people with knowl- which this perception develops (Table 1), impacts the sit- edge of the GSL may result in more people learning about uational  awareness of opioid use. In this way, the model the GSL. The parameter value that addresses contact demonstrates how an individual’s knowledge of the GSL rates can be adjusted to reflect the population density in is either unaffected or improved over time. a region. Section D also shows that, while knowledge of Situational awareness has a clear impact on changes the GSL may increase quickly, it takes time to quell the in behaviour and, therefore, is directly connected to the fear of calling 911 and, thereby, modify bystander behav- left-hand side of the model (sections D and E). In other iour. However, once fear drops, the number of 911 calls words, as the number of people dying from opioid over- and emergency medical services staff arriving at overdose dose changes, the perception of the risk of drug use influ - events will increase. Section D of the model demonstrates ences awareness in a region, determining how much law that, as people are more or less afraid of calling 911 dur- enforcement staff and the public learn about the GSL ing opioid overdoses, the number of 911 calls decreases through peer interactions (i.e.  probability of learning or increases, respectively, thereby impacting the number Table 1 Parameter values Parameter (definition and unit of analysis) Value in the model 95% confidence interval Average time to perceive risk (months) 13.1131 12–17.7579 Risk of initiation of illicit substance (fraction of susceptible users initiating illicit drug use per month) 0.0024 0.0019–0.0028 ED visit rate (average number of times that people who misuse Rx or with illicit drug use disorder 1.37808 1.2096–1.5465 visit ED) Fraction quitting Rx misuse (fraction of users misusing Rx who quit per month) 0.002 0.0015–0.002 Fraction quitting illicit drug use (fraction of users using illicit drugs/misusing Rx who quit per month) 0.0007 0.0003–0.0012 Fractional rate for overdose among nonmedical users of Rx drugs (users per month) 0.0011 0.0011–0.0011 Fractional rate for overdose from illicit drugs and Rx misuse (users per month) 0.0037 0.0037–0.0037 Risk of overdose death from Rx misuse (fraction of Rx misuse OD incidence that leads to death) 0.00001 0.00001–0.0126548 Risk of overdose death from illicit drug use and Rx misuse (fraction of illicit drug OD incidence that 0.2072 0.2021–0.2125 leads to death) S abounchi et al. Health Research Policy and Systems (2022) 20:5 Page 5 of 10 of overdoses during which naloxone is administered, 911 was provided by the following two survey reports: with a differential impact on mortality appreciated. The (1) the CT DPH and Central Connecticut State Univer- detailed model formulations are provided in the online sity’s (CCSU) survey on basic understanding of the GSL supplementary information  (Additional file 1). and the corresponding fear of calling 911 [37] and (2) the High Intensity Drug Trafficking Area’s (HIDTA) Heroin Data and model calibration Response Strategy project report on the GSL’s impact on Since most of the parameters defined in the model were Connecticut policing practices [38]. not available in the relevant literature, model calibra- tion was used to make estimates for the parameter values Modelling and simulation results shown in green in Fig.  1. Constraints on plausible values SD modeling has allowed the research team to capture of the calibrated parameters listed in Table 1 were formu- the complex interrelationships among several key health lated from expert opinion and the literature [29–32]. A outcome measures that drive the opioid epidemic in full list of calibrated parameter values is provided in the Connecticut. These outcomes include ED visits due to online supplementary information (the Additional file  1). overdose, behavioral changes in bystanders, changes in Calibration was performed using Vensim DSS, version perception of the risk of drug use, awareness of harm 8.2.1 [20]. The calibration module in Vensim modeling reduction policies, and overdose deaths. software calculates the optimum values of model param- First, the simulated number of overdose deaths aligns eters using a maximum likelihood estimation approach very closely with data supplied by the OCME. Unfortu- [33] to create simulations which align best with real- nately, these results reveal that the number of overdose world data, replicate historical trends, and create estima- deaths has been increasing over the past 8 years and will tions for the future. continue to grow if no additional action is taken (Fig. 2a). The data used to calibrate this model were collected Additionally, the simulated number of ED visits for over- from numerous sources over different intervals and dose aligns with the data supplied by CHIME discharge describe the status of opioid use in Connecticut from data, which shows that as the number of opioid users 2009 to 2018. Information about overdose deaths was and overdose incidents increases, ED visits also increase collected from the Connecticut Office of the Chief Medi - (Fig. 2b). cal Examiner (OCME). The data concerning the number Secondly, the model demonstrates that, as more opioids of ED visits for overdose came from Connecticut Hospi- are prescribed, the use of illicit and prescription drugs tal Association (CHIME) discharge data. The researchers increases. The simulation indicates that, as more peo - also utilized information from www. CTData. org, includ- ple misuse prescription drugs, an increased number of ing the amount of illicit drug use other than marijuana people switch from prescribed to illegal opioids. For this between 2008 and 2014 (collected by SAMHSA as part of reason, the number of people who misuse prescription the National Survey on Drug Use and Health [NSDUH]) drugs but do not use illicit drugs was shown to decrease [34] and the rate of arrests due to drug law offences from around 2013; and the number of people who use illicit 2010 to 2016 [35]. The most recent SAMHSA reports drugs and also misuse prescription drugs was shown to from 2015 to 2018 provide information about illicit drug increase during the same time period (Fig. 2c). It is inter- use and prescription drug misuse in Connecticut, and esting to note that, around 2013, the rate at which opi- this information was used to validate the simulation oid prescriptions were being written in Connecticut was results [24–26]. Specifically, the illicit drug use described decreasing (Fig.  2d). Consequently, this model supports in the SAMHSA report includes the misuse of prescrip- the hypothesis that an increase in deaths and subsequent tion psychotherapeutics, and the research team used situational awareness could lead to decreased opioid this information to validate the total sum of people who prescriptions; and the corresponding link depicted in the misuse prescription drugs and the number of people model has been shown, thus far, to be an effective means with illicit drug use disorder who also misuse prescription of harm reduction (Fig. 1). drugs in the model. However, there is a mismatch between simulation The rate of opioid prescriptions per 100 Connecticut results and the data points corresponding to illicit drug residents was retrieved from the CDC report on United use and misuse of prescription drugs prior to 2014 States state prescribing rates [36]. Additionally, informa- (Fig.  2e). The simulation results show that the total tion concerning the rate of administration and dose of number of people who use illicit drugs and/or misuse naloxone used was supplied by the American Medical prescription drugs (i.e., the sum of people who mis- Response (AMR) transportation company, which serves use prescription drugs and the number of people with a large portion of the state of Connecticut. Lastly, the illicit drug use disorder who also misuse prescription information on knowledge of the GSL and fear of calling drugs in the model) has increased overall, but data show Sabounchi et al. Health Research Policy and Systems (2022) 20:5 Page 6 of 10 Overdose Deaths (Total Number per Month) Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 MonthlyOve rdoseDeath s- Simulate d MonthlyOve rdoseDeath s- OCME Data Part a Overdose EDVisits (Total Number per Year) Illicit Drug UseandNon-Medical Use of Prescription Drugs 8000 140000 6000 100000 80 000 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 Number of People whoMisusePrescript ionDrugs an dnot UseIllicit Drugs- Simulated Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 Number of People with IllicitDrugUse Disorder whoalsoMisusePre scriptionDrugs-Simulated Number of Overdose ED Visits perYear- Simulated Number of Overdose ED Visits perYear- CHIMEData Part c Part b Number of Peoplewith Illicit DrugUse andthe misuse of prescription psychotherapeutics or theuse of cocaine(includingcrack), heroin,hallucinogens, inhalants, or methamphetamine -CTDATA .org & SAMHSA Data (National Opioid Analgesic Prescriptions (Number per100 Connecticut Residentper Survey on Drug Use and Health) Month) 5 115000 2 95000 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 Number of People with IllicitDrugUse Ot herthanM arijuana an dmisuseofprescriptions Number of People with IllicitDrugUse an dthe misuse of prescriptions Number of Opioid AnalgesicRxper Month- Simulated Number of Opioid AnalgesicRxper Month- CPMR S& DCPData Part e Part d Fig. 2 Simulation results for overdose deaths and drug use a decrease between 2009 and 2013. One explanation collection methods before and after 2014. However, the is that, although the data sources for illicit drug use are SAMHSA reports were the most reliable data source all based on SAMHSA reports, the SAMHSA NSDUH available for our modeling purposes and, hence, were uti- reports introduced an independent multistage area prob- lized in this study. ability sample as the first level of stratification in 2014 Also, the model predicts that GSL knowledge will within each state. Thus, the data points between 2015 continue to grow from the 2017 estimates that indicate and 2018 are based on an updated data collection pro- that 60% of the general public and 74% of police officers cess; and there are likely inconsistencies between data in Connecticut knew about the GSL [37, 38] (Fig.  3b, c). S abounchi et al. Health Research Policy and Systems (2022) 20:5 Page 7 of 10 This is a prediction based on the model structure and same direction. As more people perceive the high risk estimates, not related to a specific educational plan. As of drug use, they become more situationally aware. As a shown in Fig. 1, the model structure shows a positive link consequence, the probability of learning about the GSL from perception of drug risk to situational awareness for from peers increases, which contributes to an increased drug risk. A positive link describes a causal relationship number of law enforcement officers and members of the in an SD model when the cause and effect change in the general public who learn about the law. We calibrated Narcan Use Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 Number of EmergencyIncidents with Narc an Adminsitration -Simulated Number of EmergencyIncidents with Narc an Adminsitration -AMR Data Part a Officers CorrectKnowledge of GSLabout providingimmunityfromcriminal BasicUnderstanding of GSLamong Public charges 1 0.9 0.9 0.8 0.8 0.7 0.7 0.6 0.6 0.5 0.5 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 % of GeneralPopulationwithGSL Knowle dge- Simulated % of GeneralPopulationwithGSL Knowle dge- CT DPH& CCSU Data % of LawEnforceme nt Office rs with CorrectGSL Kn owledge- Simulated % of LawEnforceme nt Office rs with CorrectGSL Kn owledge- HIDTAData Part b Part c Rate of Drug Arrests(per10,000Connecticut Residentsper Year) Rate of 911Calls forOverdoses (FractionofIncidents that areCalled) 4.5 0.8 0.7 3.5 0.6 2.5 0.5 0.4 1.5 0.3 0.2 0.5 0.1 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 Rate of Drug Arrests- Simulated Rate of Drug Arrests- CTDATA.org Data Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16 Jan-17 Jan-18 Jan-19 Jan-20 Jan-21 Fraction of Overdose incidentsthatresultin911 Calls Part d Part e Fig. 3 Simulation results for behavioral changes Sabounchi et al. Health Research Policy and Systems (2022) 20:5 Page 8 of 10 the model to identify the strength of these causal links situationally relevant and may be used to evaluate what-if by estimating the parameter values highlighted in green simulation scenarios. Policy  makers may use this model in Fig. 1, including the average time that it takes for GSL to test new interventions that might be used to address knowledge to change and the threshold number of over- the opioid crisis. Additionally, once more robust data on dose fatalities at which changes in awareness towards the the behavioral impact of the GSL become available, those risk of drug use and the benefit of GSLs will occur. data can be used to produce an even more reliable model Policy  makers may infer from the model that, because in the future. the number of overdose deaths is increasing, situational While existing data show that the GSL has not yet awareness will change, leading to an increase in officer reduced the number of overdose deaths, the model’s and public awareness of the GSL. Subsequently, the num- simulations indicate that the high number of deaths will ber of people with GSL knowledge and without fear of likely foster an increased awareness of the GSL, leading calling 911 is anticipated to increase. Also, the rate of to decreased fear of calling 911 and increased naloxone drug arrests per 10,000 for drug law offences has been administration. However, the model also  suggests that decreasing, possibly due to the growth of GSL awareness, the overall trend of increased deaths may continue to but is predicted to remain stable in the future (Fig. 3d). grow despite this increased awareness. This prediction is Finally, the results of the simulation that represent the supported by the model, as well as by many other studies, use of naloxone correspond to the data received from indicating that fear of police interactions is the primary AMR, which show an overall increase in the number of reason that bystanders do not call 911. overdose events during which naloxone is administered Fortunately, the model demonstrates that interven- (Fig. 3a). The model explains that this increase is partially tions like the GSL, which protects bystanders against due to an increase in overdose incidents since 2009 and liability for providing assistance during overdoses, may partially due to a similar increase in the number of 911 represent a partial solution to this problem. However, calls since the start of the simulation (Fig.  3e). However, additional interventions are needed to improve the effec - even accounting for an increase in emergency medical tiveness of the GSL. For example, although the model’s assistance and administration of naloxone, the number of results indicate that the rate of drug-related arrest has overdose deaths has continued to increase (Fig.  2a); and slowed in recent years, future interventions, such as risk of overdose remains stable (Table 1), contrary to our increased training for police officers, may still be neces - initial hypothesis that had assumed negative causal links sary. One study claims that GSLs and other harm reduc- from Narcan use to risk of overdose death from misuse tion policies are necessary but insufficient, primarily due (ROD) and risk of illicit drug overdose death in Fig. 1. The to problems with implementation and awareness [10]. increase seen over time in the dose of naloxone adminis- While efforts to alter the course of the opioid epidemic tered for an overdose from an average dose of 1.18 mg in will require ongoing research concerning the numerous 2010 to 1.94  mg in 2018 is based on data received from interventions that could be applied to this problem, this AMR and may be related to the high potency of the new analysis illustrates how SD modeling may be beneficial illicit drugs that have become available over the past few in aiding policy  makers who are tasked with decision- years. making in the setting of complex challenges. Because more time must pass in order to observe the long-term Discussion effectiveness of the GSL, an SD modeling approach can SD modeling is an analytical tool that helps policy  mak- be used to make predictions about its long-term impact ers approach difficult decisions in the presence of the by employing a simulation framework. Additionally, uncertainties that complex problems create [3]. The SD since the model structure and feedback loops are relevant approach is ideal for evaluating the delayed impact of the to the opioid epidemic in general, the model parameters GSL on behavioral changes because it allows research- can be calibrated towards historical data trends for other ers to investigate the long-term effects of policy inter - geographical regions or states and, thus, be customized ventions using a simulation framework. Because of the for different locations and settings. In this way, pol - ability to simulate with SD, policy makers can infer that, icy makers can utilize this model to test future trends and although the evidence currently demonstrates mixed determine the best solutions for various public health effects of the GSL, the overall amount of drug use and problems. number of overdose deaths will increase if no additional policies are implemented. Limitations Moreover, because the model’s simulation results align This SD model included feedback processes and dynam - with real-world data and can be used to replicate his- ics important to understanding the impact of GSLs and torical trends, it is reasonable to infer that the model is providing insight for policy  makers and public health S abounchi et al. Health Research Policy and Systems (2022) 20:5 Page 9 of 10 Acknowledgements officials. While this model provides a useful foundation We would like to thank Dr. Brian J. Biroscak, PhD, MS, MA, at the Weitzman for answering targeted research questions about GSLs Institute Community Health Center, Inc., and Dr. Jason Buckert, MD, for their and related policies, our analysis had several limitations. advice and contributions to this study. First, identifying relevant existing data sources for some Authors’ contributions model indicators that span the time horizon of the study Nasim S. Sabounchi, Rebekah Heckmann, Gail D’Onofrio, and Robert Heimer has been challenging. Also, some data sources, such as originated the study, contributed to the concept and design, and supervised all aspects of its implementation. Nasim S. Sabounchi and Rebekah Heck- the SAMHSA reports [24–26], have updated their data mann developed the system dynamics simulation model and processed and collection process, which has likely led to inconsisten- analyzed the simulation results. Jennifer Walker contributed to the drafting cies in data collection methods. While the current model of the manuscript, and all authors contributed to the critical revision of the manuscript for important intellectual content. All authors read and approved structure and feedback loops appear to replicate histori- the final manuscript. cal trends well, future iterations of the model may test the alignment of current or additional feedback loops. Funding This study was funded under contract to the Connecticut Department of Pub- Additionally, the model was designed based upon expert lic Health (DPH) on behalf of the CDC Prescription Drug Overdose Prevention input and GMB sessions conducted with key stakehold- for States (CDC Grant Number: 1U17CE002720-02). The content presented ers. A next step could adopt a more inclusive approach, does not necessarily reflect DPH or CDC policy. and other relevant stakeholders such as patients, provid- Availability of data and materials ers, law enforcement officers, and first responders could All data generated or analyzed during this study are included in this published be invited to contribute to the modeling process and cor- article and its additional information file. responding  validation while exploring specific questions related to the impact of opioid policies on reducing over- Declarations dose risk and fatality. Ethics approval and consent to participate Not applicable. Consent for publication Conclusions Not applicable. SD modeling has been proven to be a useful approach for assessing the effectiveness of public health policy Competing interests The authors declare that they have no competing interests. interventions through its utilization of a simulation framework. While other analytical methods may require Author details research involving study participants and clinical tri- Department of Health Policy and Management, Center for Systems and Com- munity Design, City University of New York (CUNY ) Graduate School of Public als, SD modeling allows for the prediction of the future Health and Health Policy, 55 W. 125th Street, New York, NY 10027, United effectiveness of interventions through its ability to rep - States of America. Department of Emergency Medicine, Yale University licate historical trends. While investigating the impact School of Medicine, New Haven, CT, United States of America. Bingham- ton University-State University of New York, Binghamton, NY, United States of the GSL on overdose deaths, ED visits and bystander of America. Department of Epidemiology of Microbial Diseases, Yale Univer- behavior in Connecticut is the main purpose of this anal- sity School of Public Health, New Haven, CT, United States of America. ysis, this model has also demonstrated great potential by Received: 21 May 2021 Accepted: 9 December 2021 producing simulations that reveal multiple strategies to aid policy  makers in determining the best public health interventions for combating the opioid crisis. References 1. Bose J, Hedden SL, Lipari RN, Park-Lee E. Key substance use and mental Abbreviations health indicators in the United States: results from the 2017 National GSL: Good Samaritan law; SD: System dynamics; SAMHSA: Substance Abuse Survey on Drug Use and Health. Rockville: Center for Behavioral Health and Mental Health Services Administration; ED: Emergency department; OUD: Statistics and Quality, Substance Abuse and Mental Health Services Opioid use disorder; CDC: Centers for Disease Control and Prevention; AMR: Administration; 2018. Contract No.: No. SMA 18-5068. American Medical Response; GMB: Group model-building. 2. Scholl L, Seth P, Kariisa M, Wilson N, Baldwin G. Drug and opioid-involved overdose deaths—United States, 2013–2017. Morb Mortal Wkly Rep. Supplementary Information 2019;67(5152):1419. 3. Sterman JD. Learning from evidence in a complex world. Am J Public The online version contains supplementary material available at https:// doi. Health. 2006;96(3):505–14. org/ 10. 1186/ s12961- 021- 00807-w. 4. Atkinson J-A, Wells R, Page A, Dominello A, Haines M, Wilson A. Applica- tions of system dynamics modelling to support health policy. Public Health Res Pract. 2015;25(3):e2531531. Additional file 1. Section 1: Model Formulation. Section 2: Estimation 5. Watson DP, Ray B, Robison L, Huynh P, Sightes E, Brucker K, et al. Lay and Calibration of Model Parameters. responder naloxone access and Good Samaritan law compliance: postcard survey results from 20 Indiana counties. Harm Reduct J. 2018;15(1):18. Sabounchi et al. Health Research Policy and Systems (2022) 20:5 Page 10 of 10 6. Gomes T, Tadrous M, Mamdani MM, Paterson JM, Juurlink DN. The burden and Medicine Division; Board on Health Sciences Policy; Committee on of opioid-related mortality in the United States. JAMA Netw Open. Pain Management and Regulatory Strategies to Address Prescription 2018;1(2):e180217-e. Opioid Abuse; 2017. 7. WHO. Information sheet on opioid overdose. Geneva: World Health 28. CDC. Opioid data analysis and resources. Atlanta: Centers for Disease Organization; 2018. Control and Prevention, National Center for Injury Prevention and Con- 8. DMHAS. Opioid overdose prevention/naloxone (Narcan) initiative. trol; 2021. Hartford: Connecticut Department of Mental Health & Addiction Services; 29. Heimer R, Barbour R, Palacios WR, Nichols LG, Grau LE. Associations 2019. between injection risk and community disadvantage among suburban 9. McClellan C, Lambdin BH, Ali MM, Mutter R, Davis CS, Wheeler E, et al. injection drug users in southwestern Connecticut, USA. AIDS Behav. Opioid-overdose laws association with opioid use and overdose mortal- 2014;18(3):452–63. ity. Addict Behav. 2018;86:90–5. 30. Krawczyk N, Eisenberg M, Schneider KE, Richards TM, Lyons BC, Jackson K, 10. Banta-Green CJ, Beletsky L, Schoeppe JA, Coffin PO, Kuszler PC. Police et al. Predictors of overdose death among high-risk emergency depart- officers’ and paramedics’ experiences with overdose and their knowledge ment patients with substance-related encounters: a data linkage cohort and opinions of Washington State’s drug overdose–naloxone–Good study. Ann Emerg Med. 2020;75(1):1–12. Samaritan law. J Urban Health. 2013;90(6):1102–11. 31. Larochelle MR, Bernson D, Land T, Stopka TJ, Wang N, Xuan Z, et al. Medi- 11. Koester S, Mueller SR, Raville L, Langegger S, Binswanger IA. Why are cation for opioid use disorder after nonfatal opioid overdose and associa- some people who have received overdose education and naloxone tion with mortality: a cohort study. Ann Intern Med. 2018;169(3):137–45. reticent to call Emergency Medical Services in the event of overdose? Int 32. Weiner SG, Baker O, Bernson D, Schuur JD. One-year mortality of patients J Drug Policy. 2017;48:115–24. after emergency department treatment for nonfatal opioid overdose. 12. Latimore AD, Bergstein RS. “Caught with a body” yet protected by law? Ann Emerg Med. 2020;75(1):13–7. Calling 911 for opioid overdose in the context of the Good Samaritan 33. Dogan G. Bootstrapping for confidence interval estimation and hypoth- Law. Int J Drug Policy. 2017;50:82–9. esis testing for parameters of system dynamics models. Syst Dyn Rev. 13. SAMHSA. Preventing the consequences of opioid overdose: understand- 2007;23(4):415–36. ing 911 Good Samaritan Laws. Substance Abuse and Mental Health 34. CTData. Illicit drug use reports the prevalence of the consumption of Services Administration’s Center for the Application of Prevention Tech- illicit substances by age range. Connecticut Data Collaborative; 2020. nologies task order; 2017.http:// data. ctdata. org/ datas et/ illic it- drug- use. 14. Hawk KF, Vaca FE, D’Onofrio G. Focus: addiction: reducing fatal opioid 35. CTData. Drug arrests reports the number and rate (per 10,000) of arrests overdose: prevention, treatment and harm reduction strategies. Yale J due to drug law offenses, per age range. Connecticut Data Collaborative; Biol Med. 2015;88(3):235. 2020. http:// data. ctdata. org/ datas et/ drug- arres ts. 15. Karamouzian M, Kuo M, Crabtree A, Buxton JA. Correlates of seeking 36. CDC. U.S. state prescribing rates: Centers for Disease Control and Preven- emergency medical help in the event of an overdose in British Columbia, tion, National Center for Injury Prevention and Control; 2017. https:// Canada: findings from the Take Home Naloxone program. Int J Drug www. cdc. gov/ drugo verdo se/ maps/ rxsta te2009. html. Policy. 2019;71:157. 37. Hawk KF, Doernberg M, D’Onofrio G, Heimer R, Diaz-Matos LF, Jenkins M, 16. Jakubowski A, Kunins HV, Huxley-Reicher Z, Siegler A. Knowledge of the et al. Knowledge of Connecticut’s Good Samaritan Law among connecti- 911 Good Samaritan law and 911-calling behavior of overdose witnesses. cut residents who have witnessed an overdose. 2019. Subst Abuse. 2018;39(2):233–8. 38. HIDTA. Heroin Response Strategy—911 Good Samaritan Law Corner- 17. Backus L. Fatal drug overdoses rising in CT: Town-by-town view of 2019 stone Project—Preliminary State Report Connecticut. High Intensity Drug deaths: ctpost; 2020. https:// www. ctpost. com/ local/ artic le/ Fatal- drug- Trafficking Areas (HIDTA) Program—WHite House Office of National Drug overd oses- rising- in- CT- Town- by- town- 15066 922. php. Control Policy; 2019. 18. Homer JB, Hirsch GB. System dynamics modeling for public health: back- ground and opportunities. Am J Public Health. 2006;96(3):452–8. Publisher’s Note 19. Homer J, Wakeland W. A dynamic model of the opioid drug epidemic Springer Nature remains neutral with regard to jurisdictional claims in pub- with implications for policy. Am J Drug Alcohol Abuse. 2020;47:5–15. lished maps and institutional affiliations. 20. Vensim. Ventana Systems, Inc.; 2021. 21. Hovmand PS, Andersen DF, Rouwette E, Richardson GP, Rux K, Calhoun A. Group model-building ‘Scripts’ as a collaborative planning tool. Syst Res Behav Sci. 2012;29(2):179–93. 22. Pitt AL, Humphreys K, Brandeau ML. Modeling health benefits and harms of public policy responses to the US opioid epidemic. Am J Public Health. 2018;108(10):1394–400. 23. Chen Q, Larochelle MR, Weaver DT, Lietz AP, Mueller PP, Mercaldo S, et al. Prevention of prescription opioid misuse and projected overdose deaths in the United States. JAMA Netw Open. 2019;2(2):e187621-e. 24. SAMHSA. 2015–2016 national survey on drug use and health: model- based prevalence estimates (50 states and the District of Columbia). Rockville: Substance Abuse and Mental Health Services Administration (SAMHSA), Center for Behavioral Health Statistics and Quality, National Re Read ady y to to submit y submit your our re researc search h ? Choose BMC and benefit fr ? Choose BMC and benefit from om: : Survey on Drug Use and Health; 2016. 25. SAMHSA. 2017–2018 national survey on drug use and health: model- fast, convenient online submission based prevalence estimates (50 states and the district of Columbia). thorough peer review by experienced researchers in your field Rockville: Substance Abuse and Mental Health Services (SAMHSA), Center for Behavioral Health Statistics and Quality, National Survey on Drug Use rapid publication on acceptance and Health; 2018. support for research data, including large and complex data types 26. SAMHSA. 2016–2017 national survey on drug use and health: model- • gold Open Access which fosters wider collaboration and increased citations based prevalence estimates (50 states and the district of Columbia). Rockville: Substance Abuse and Mental Health Services Administration maximum visibility for your research: over 100M website views per year (SAMHSA), Center for Behavioral Health Statistics and Quality, National Survey on Drug Use and Health; 2017. At BMC, research is always in progress. 27. NASEM. Pain management and the opioid epidemic: balancing societal Learn more biomedcentral.com/submissions and individual benefits and risks of prescription opioid use. Washington (DC): National Academies of Sciences, Engineering, and Medicine; Health

Journal

Health Research Policy and SystemsSpringer Journals

Published: Jan 6, 2022

Keywords: Opioid use disorder; Emergency medicine; Health policy simulation; System dynamics modelling

There are no references for this article.