Colorado drugged driving prevalence and impaired driving conviction rates: Effects of impaired driving definitions and a 5-nanogram limit for delta-9 tetrahydrocannabinol

Abstract

Objectives: How prevalent is drugged driving among Colorado drivers convicted of Driving Under the Influence (DUI)? What are the conviction rates of Colorado drivers charged with DUI, including impairment by marijuana’s delta-9 tetrahydrocannabinol (THC)? Is Colorado’s THC permissible inference law effective? To answer these questions, this report analyzes data published primarily in appendices of Colorado drugged driving reports.

Methods: In 2017 Colorado began requiring annual analyses of Driving Under the Influence (DUI) offenses, including causes and judicial consequences of DUI offenses. These analyses are performed by the Division of Criminal Justice’s Office of Research and Statistics (ORS) within the Department of Public Safety. Each analysis requires ORS to link toxicology and court data bases. Data linking enables reporting of charges and convictions by categories including alcohol only, THC only, and polydrug use (two or more drugs simultaneously). Reports have been published annually for 5 years, the latest published in 2023 which covers case filings for 2020.

Results: A rough estimate of one-half of the state’s DUI filings were attributed to drug use and half were attributed to alcohol only. The largest component of drugged driving was polydrug impairment, rather than impairment by a single drug like THC. Conviction rates in 2020 were 91% for alcohol only, 90% for polydrug cases, and 72% for THC only. Blood drug levels and law structure (per se, permissible inference, DUI definition) affected conviction rates significantly by defendant subsets. THC conviction rates in 2020 ranged from 11% to 100%, depending on blood drug levels and the legal charges.

Conclusions: Efforts to educate the public about the dangers of drugged driving should emphasize polydrug impairment, not simply THC impairment. States should analyze data on causes and consequences of DUI arrests to understand what their drugged driving problems are and what they are not. Non-zero drug per se levels and defining DUI as “incapable of safe driving” can severely reduce the effectiveness of DUI laws.

Keywords:

• Drugged driving prevalence
• drugged driving conviction rates
• DUI definitions
• drug per se law

Introduction

Individual states in the USA are responsible for creating and enforcing laws against Driving Under the Influence (DUI). DUI laws initially required courts to prove behavioral impairment to achieve a conviction. For example, a 1930 UK law prohibited driving “under the influence of drink or drugs to such an extent as to be incapable of having proper control of the vehicle” (Jones et al. 2019). The states in the USA adopted various DUI definitions, partly to deal with the difficulty in proving a driver was under the influence. Statutory DUI definitions range from “incapable of safe driving” to numerous forms of “impaired to the slightest degree” as shown in Supplementary material. DUI definitions may even vary between alcohol and other drugs as shown in Supplementary material. DUI definitions became moot beginning in 1938 with the adoption of alcohol legal limits, now .08% in most of the USA, Canada, Mexico and countries that were part of the British Commonwealth. Alcohol legal limits have been supported by studies showing a strong correlation between Blood Alcohol Content (BAC) and crash risk (Zador et al. 2000; Blomberg et al. 2009). Most legal limits make driving with a BAC above legal limits a per se violation of the law so that prosecutors do not need to prove impairment.

DUI definitions may no longer be moot for impaired driving caused by substances other than alcohol. Especially since symptoms of drug impairment are usually more subtle than those of alcohol impairment. The National Institute on Drug Abuse (NIDA) reports that tetrahydrocannabinol (THC) is now the second most common drug after alcohol to be found in the blood of drivers involved in crashes in the US (NIDA 2023). In 1996 California became the first state to enable its citizens to “obtain and use marijuana for medical purposes” (California 1996). In 2014 Colorado became the first state to legalize and commercialize cannabis for recreational purposes when Amendment 64 permitted the “personal use and regulation of marijuana” (Colorado 2012). With more people using THC since its legalization and commercialization (NSDUH 2022, 2014), NIDA’s finding should not be surprising. The importance of NIDA’s finding is tempered by the fact that the mere presence of the drug does not prove impairment. This leads to the first objective of this report, which is to move beyond measures of mere drug presence to understand the role of drugs in driving impairment.

Blood levels of THC in chronic users can be elevated long after acute impairment caused by THC has subsided (Huestis 2007; Sewell et al. 2009). Water-soluble alcohol establishes a uniform concentration across highly perfused tissues and organs so that blood alcohol levels can reliably indicate levels of alcohol in the impaired brain. But fat-soluble THC is rapidly sequestered from the blood by the brain and other fatty tissues, preventing blood THC levels from accurately representing THC levels in the impaired brain. Consequently, unlike with alcohol, there is no correlation between blood THC levels and levels of impairment (AAA 2016; Hartman et al. 2016; Mura et al. 2005; Vandrey et al. 2017).

Nevertheless, several jurisdictions around the world have adopted legal limits on over a dozen different impairing drugs and their metabolites. Rather than basing the drug limits on demonstrated crash risk effects as was done with alcohol, drug limits have usually been based on the minimum levels at which an impairment effect can be documented, the limit of quantitation of forensic toxicology laboratories, or maximum normal therapeutic limits for legal drugs (Gjerde and Strand 2023; Wolff et al. 2013). Due to the lack of good correlation between impairment and the blood content of impairing drugs other than alcohol, there are concerns that establishing a drug per se limit too low would indict non-impaired drivers whereas establishing a drug per se limit too high would exonerate impaired drivers. Colorado is one of the jurisdictions that established a legal limit for THC of 5 nanograms per milliliter of whole blood (Colorado 2013). Both above concerns were debated during testimony prior to establishing the THC legal limit (Waller 2013). The first concern, that of indicting non-impaired drivers, was addressed by formulating the legal limit as a permissible inference, rather than as a per se limit. Whereas a per se law defines guilt, a permissible inference law merely allows a court to infer that the driver was guilty if evidence other than laboratory data supports a finding of guilt. The second concern was judged by legislators to be of little concern, believing that impaired drivers below the legal limit could still be convicted on impairment evidence alone.

Believing that “States cannot estimate the size or characteristics of their drugged driving problem without good data in crashes and arrests,” the Governors Highway Safety Association recommended that States “Track DUID (Driving Under the Influence of Drugs) and DUI separately in crash, arrest, licensing and court data to the extent possible” (Hedlund 2017). This recommendation encouraged Colorado to require its Department of Public Safety to annually publish an analysis of the causes and judicial consequences of DUI convictions (Colorado 2017).

Multiple impairing drug possibilities and inconsistent drug testing make acquiring drugged driving conviction data a non-trivial matter. Colorado is the only state that has attempted to gauge the effectiveness of their drunk and drugged driving laws by publishing data on DUI conviction rates as a function of the toxicology results (Rosenthal and Reed 2023).

A full understanding of the Colorado published data is complicated by the fact that the state has two distinct offenses: Driving Under the Influence (DUI) and Driving While Ability Impaired (DWAI). Colorado’s two impaired driving offenses are summarized on Table 1. DUI is statutorily defined as substantially incapable of safe driving whereas the less serious DWAI offense is statutorily defined as impaired to the slightest degree so that the person is less able to drive safely (Colorado 2013). A DWAI conviction does not necessarily mean that a driver was only moderately impaired however, since DWAI is often used as a plea bargain tool to achieve convictions. DWAI is a lesser offense than DUI and DUI per se. Criminal sanctions for DUI and DWAI are identical except that criminal sanctions for a drivers’ first DWAI offense are substantially reduced. Administrative sanctions are lower for DWAI than for DUI. A DWAI conviction does not cause loss of a driving license. Importantly, both DUI and DWAI are misdemeanors. There is a permissible inference that a driver was DWAI if the BAC was between .05% and .079%, and a presumption that the driver was neither DWAI nor DUI if the BAC was below .05%. Colorado’s legislature applied the 5-ng THC permissible inference law only to DUI, not to DWAI.

Table 1. Colorado impaired driving misdemeanor offenses.

	DUI	DWAI
Description	Driving Under the Influence	Driving While Ability Impaired
Definition	Incapable of safe driving	Impaired to the slightest degree
Alcohol legal limits	≥.08 gm/dL per se	.05 gm/dL permissible inference
THC legal limits	≥ 5 ng/mL permissible inference	N/A

Colorado Revised Statute 42-4-4301.

Methods

In 2017 the Colorado legislature passed HB17-1315. The governor signed it into law as C.R.S. 24-33.5-520. The law mandates ORS to annually publish an analysis of the causes and judicial consequences of substance-affected driving. Reports from calendar years 2018 (Rosenthal and Reed 2020), 2019 (Rosenthal and Reed 2022), and 2020 (Rosenthal and Reed 2023) included sufficient information to enable an understanding of both DUI and DWAI conviction rates. The methods of data collection and reporting conventions are explained in each annual report. Report releases are scheduled to provide a minimum of one year for each citation to reach a judicial disposition before extracting data for analysis.

ORS attempts to link all DUI/DWAI charges with blood toxicology data or evidentiary breath test results. The state’s expressed consent law requires anyone who drives in the state to provide a breath or blood sample to test for intoxicants when a law enforcement officer has probable cause that the driver was impaired. Refusals to comply incur an administrative one-year license suspension which is greater than the 9-month suspension resulting from a DUI conviction. Most drivers have complied with the law, but the Division of Motor Vehicles estimated that 42% of drivers arrested for DUI/DWAI in 2020 refused to comply (Rosenthal and Reed 2023, p. 39). That compares with Colorado’s 28% refusal rate in 2011 and a national average refusal rate of 24% in 2011 (Namuswe et al. 2014). Conviction rates in 2020 were 88.1% for drivers who complied with the expressed consent law and 81.5% for drivers who did not comply (Rosenthal and Reed 2023, Table 28). For the 2023 report of 2020 data, 59% of charged cases were linked to toxicology data and 51% of cases had both a toxicology result match and a DUI/DWAI disposition at the date of data extraction.

Historically, most toxicology testing has been done by means of an evidentiary breath test that is far less expensive than a blood test, provides immediate results, but only tests for alcohol. Blood testing is commonly chosen when a preliminary non-evidentiary breath testing device indicated that the driver had a BAC below .08% or if the arresting officer had evidence that the driver was impaired by drugs other than or in addition to alcohol. These toxicology testing practices provided insufficient data to reliably estimate the prevalence of drugs other than alcohol in DUI/DWAI defendants. For example in 2018, 51% of charged cases were tested for alcohol, but only 7% of charged cases were matched to toxicology tests of both drugs and alcohol (Rosenthal and Reed 2023, p. 40).

Two changes occurred in 2020 to dramatically change drugged driving data availability. First, as a precaution for officers during the early days of the COVID-19 pandemic, several large law enforcement agencies suspended evidentiary breath tests in favor of blood tests. Second, the Colorado Bureau of Investigation (CBI) began comprehensive drug testing on all blood samples beginning July 2019 at no charge to law enforcement agencies. The first full year of such services was 2020. As a result, CBI became the sole blood testing contractor in the state. The Denver Police Department Crime Lab continues to test Denver’s blood samples, but for alcohol only.

Measures

With one exception, this analysis used the conventions established by the ORS reports, including the following:

Convictions for DUI, DUI per se and DWAI were not differentiated in the body of the ORS reports. All were referred to as DUI. This paper analyzes the DUI and DWAI conviction data reported separately in appendices in the 2022 and 2023 ORS reports. Comparable data were not published in the 2020 report but were determined and provided by ORS on request (English 2021).

Findings of Guilty, Deferred, and Deferred/Dismissed were all included within the Guilty finding.

Defendants with a finding of THC only were tested using a multi-drug screen and were found positive for delta-9 tetrahydrocannabinol but negative for all other drugs or drug categories on the drug screen. Defendants who were positive for only THC’s inactive metabolite 11-nor-9 carboxy tetrahydrocannabinol were not included in the THC only category.

Defendants with a finding of alcohol only were tested by law enforcement agencies using an Intoxilyzer™ 9000 evidentiary breath testing device, by the Denver Crime Lab using conventional blood assays, or by CBI as part of its 14-panel drug screen. The first two testing sources did not test for drugs other than alcohol. Only CBI tested for drugs other than alcohol.

Analysis

Frequency of convictions by toxicology results and charges are presented for the three years for which we have data. Conviction rates by category for 2018-2020 were compared without statistical hypothesis testing. The Department of Public Safety did not determine conviction rates by DUI and DWAI categories until 2018.

Results

Drug prevalence in DUI filings

Table 2 shows how the toxicology results varied as the testing practices changed beginning July 2019. One laboratory performing drug tests only used varying drug screens as requested by its law enforcement agency customers, one of which did not include THC in the screen.

Table 2. Changes in reported drug prevalence as a function of testing policies.

		2018	2019	2020
Test protocol	% drug and alcohol tested	7	20	37
	% alcohol tested only	51	40	22
	% drug tested only	5	2	0
Positive test results	% alcohol only	75.3	73.7	59.2
	% THC only	6.4	5.7	6.7
	% Other single drug	2.7	1.9	3.2
	% Polydrug	14.5	17.0	29.3

(Rosenthal and Reed 2023 Figure 9, Table 21).

Table 3 shows 2020 DUI/DWAI case filings by drug category using two different data sets, “All tests” and “Full drug tests.” The “All tests” column contains data from all three sources of DUI toxicology testing: the Colorado Bureau of Investigation (CBI), The Denver Crime Lab (DEN), and the Colorado Department of Health and the Environment (CDPHE). CBI tests all blood samples for alcohol and a full drug panel. DEN tests blood samples for alcohol only. CDPHE gathers data from all the state’s evidentiary breath testing devices and reports consolidated data from all evidentiary breath testing in the state. The “Full drug tests” column contains data only from CBI.

Table 3. DUI case filings in 2020 by drug category.

	All tests		Full drug tests
Drug category	Number	%	Number	%
None detected	205	1.6%	183	2.4%
Alcohol only	7,360	59.2%	2,647	34.4%
THC only	831	6.7%	831	10.8%
Single other drug	398	3.2%	398	5.2%
Polydrug	3,646	29.3%	3,646	47.3%
Alcohol + THC	1,471	11.8%	1,471	19.1%
Alcohol + other	842	6.8%	842	10.9%
THC + other	556	4.5%	556	7.2%
Alcohol + THC + other	499	4.0%	499	6.5%
Polydrug not alcohol or THC	278	2.2%	278	3.6%
Total	12,440		7,705

“All tests” include CBI, CDPHE, and Den data (Rosenthal and Reed 2023, Table 21).

Of the 12,440 cases, 7,705 were tested for alcohol and other drugs, and 4,735 were tested for alcohol only. “Full drug tests” include CBI data only, were tested for both drugs and alcohol, and were provided by the Department of Public Safety under an Open Records Act request.

DUI conviction rates

The overall conviction rate for DUI/DWAI defendants in 2020 was 88%. But Table 4 shows a large difference in conviction rates as a function of identified intoxicant. Specifically, the lowest conviction rates are seen for THC only filings, which ranged from 72% to 78%. In comparison, the conviction rate for polydrug and alcohol only filings was 90% or higher for all three years. Alcohol only data include cases not tested for drugs other than alcohol. Any other single drug (besides THC or alcohol) had a conviction rate from 77% to 81%.

Table 4. Convictions and conviction rates by impairing substances identified.

	2018		2019		2020
Cause	Nbr	%	Nbr	%	Nbr	%
Overall	13,964	91%	14,440	92%	9,475	88%
Alcohol only	11,775	92%	11,792	93%	6,653	92%
Polydrug	2,239	91%	2,722	91%	3,029	90%
THC only	978	76%	907	78%	700	72%
Other single drug	410	80%	300	81%	302	77%

Convictions are judicial outcomes of either guilty, deferred, or deferred dismissed for either DUI or DWAI. Alcohol only conviction rates are based on all laboratory assays, including those that did not test for other drugs. (Rosenthal and Reed 2023 Table 35).

Effect of THC and alcohol legal limits

Table 5 shows conviction rates as a function of the concentration of intoxicants identified above and below the legal limits of 5-ng/mL for THC and .08% for alcohol. THC conviction rates for drivers who tested below the 5-ng limit (36%) were consistently less than half the conviction rates of other categories.

Table 5. Conviction rates for alcohol and THC by legal limits.

Drug Drug Level		2018	2019	2020	Average
Alcohol only	BAC .08+	95%	96%	95%	95%
Alcohol only	BAC .05-.079	76%	79%	75%	77%
THC only	5-ng+	89%	89%	83%	87%
THC only	< 5-ng	36%	39%	33%	36%

Conviction rates are based on convictions of either DUI or DWAI. (Rosenthal and Reed 2023, Appendices I & J).

Convictions for both substances were lower when the driver tested below the legal limits of .08% for alcohol (77%) or 5-ng for THC (36%).

Effect of offense definition

The conviction rates by drug and offense are shown on Table 6. The 3-year average drunk driving conviction rates were lower for DUI (91%) than for DWAI (97%).

Table 6. Conviction rates for alcohol and THC by offense.

Drug Offense		2018	2019	2020	Average
Alcohol only	DUI	91%	91%	90%	91%
Alcohol only	DWAI	97%	98%	97%	97%
THC only	DUI	46%	51%	42%	46%
THC only	DWAI	98%	98%	100%	99%
Polydrug w/ THC	DUI	86%	79%	87%	84%
Polydrug w/ THC	DWAI	100%	100%	100%	100%

(Rosenthal and Reed 2023, Appendices I & J).

Combined effect of legal limits and offense definition

The combined effects of offense and legal limits are summarized in Table 7. The 3-year average DUI conviction rate for drivers impaired by THC above the permissible inference limit was 65% compared with a 92% conviction rate for drivers above alcohol’s legal limit.

Table 7. Convictions (nbr) and conviction rate (CR) by drug, concentration and offense.

		2018		2019		2020		3 yr Avg CR
	Charge	Nbr	CR	Nbr	CR	Nbr	CR
THC only	DUI THC 5-ng+	172	69%	155	70%	117	57%	65%
	DUI THC< 5-ng	13	8%	8	8%	11	11%	9%
	DWAI 5-ng+	444	99%	370	99%	312	100%	99%
	DWAI < 5-ng	70	93%	52	93%	35	97%	94%
Alcohol only	DUI BAC .08+	4,924	92%	5,032	93%	3,125	91%	92%
	DUI BAC .05-.079	34	29%	24	26%	22	29%	28%
	DWAI BAC 08+	3,565	100%	4,094	100%	2,382	100%	100%
	DWAI BAC .05-079	625	84%	670	85%	353	83%	84%

(Rosenthal and Reed 2023, Appendices I & J).

The 3-year average THC conviction rate for drivers testing below 5-ng/mL was only 9% for DUI but 99% for DWAI regardless of the blood THC level.

Discussion

Many states publish annual DUI reports but Colorado’s is unique in its linking of forensic laboratory data to court cases. Other states must adopt this or a similar practice to confirm the conclusions from this Colorado-centric study. This is important because as noted in the Introduction, laws vary widely among the states. Social norms also vary widely, from states like Idaho that enforce Federal drug laws to Colorado that has commercialized sales of marijuana products, actively promotes drug use through its Cannabis Business Office (Colorado 2023) and provides grants to new drug businesses (Mehrtens 2023).

The first question addressed in this analysis was the prevalence of drugged driving in individuals charged with DUI/DWAI. Table 2 shows that as the percent of drug testing rose from 7% to 37% of the total and testing of alcohol only dropped from 51% to 22% of the total, the reported percentage of DUI/DWAI attributed to alcohol only dropped from 75.3% to 59.2% and the percentage of DUI/DWAI attributed to polydrug impairment rose from 14.5% to 29.3%. Not surprisingly, the reported incidence of polydrug impairment was low when few samples were tested for all drugs.

The “All tests” data set in Table 3 suggests 59.2% of the cases were due to alcohol only and 1.6% were undecided, leaving 39.2% of the cases as drugged driving (including polydrugs), whereas the “Full drug tests” data set suggests 34.4% were due to alcohol only, and 63.2% were due to drugs. Neither is correct. Neither data set purports to be a representative sample of drivers arrested for DUI/DWAI or even a representative sample of drivers arrested for DUI/DWAI who received a toxicology test. The “All tests” data set undercounts drugged driving because 4,713 of the cases were tested only for alcohol, so that number may include polydrug cases of alcohol plus other drugs. The “Full drug tests” data set overcounts drugged driving because of the possibility that subjects were selected for full drug testing only because the arresting officer had evidence that drugs may have been involved, which has been the historical practice. The most we can say from these data is that the prevalence of drugged driving in individuals charged with DUI/DWAI was somewhere between about 39% and 63%, or very roughly about half. Any number in this range is higher than inferences from earlier studies shown on Table 2 or an early 30% estimate based on a study of vehicular homicide and vehicular assault cases (Wood and Salomonsen-Sautel 2016).

The final question addressed was the DUI/DWAI conviction rates and the effect of Colorado’s 5-ng THC permissible inference law. During Colorado’s legislative debate over the 5-ng permissible inference level, it was made clear that the intent of passing a THC permissible inference law rather than a THC per se law was to aid convictions of THC impaired drivers but also to ensure that drivers would not be convicted of DUI based solely on toxicology laboratory results.

Table 7 reveals that the 3-year average DUI conviction rate for drivers impaired by THC who were above 5-ng was 65% compared with a 92% conviction rate for drivers above alcohol’s legal limit. As intended by legislators, it appears that the permissible inference structure ensured that convictions of drivers were not based simply on THC toxicology results, or the conviction rates would be closer in value.

Prosecutors frequently offer DWAI charges as a plea bargain for drivers charged with DUI. Defendants often accept a guilty plea to DWAI in exchange for dropping the DUI charges. The success of these efforts to conserve judicial resources can be seen on Table 7 where 43% of drivers with a BAC over the alcohol .08% per se limit were found guilty of DWAI, rather than the more serious crimes of DUI or DUI per se. Those drivers would likely have been found guilty of DUI per se had it not been for plea bargaining. This means that, in spite of the legal definitions of DUI and DWAI, it is not necessarily true that a DWAI conviction means that a driver was only slightly impaired.

The 3-year average THC DUI conviction rate for drivers below the permissible inference level was 9%, whereas the THC DWAI conviction rate for drivers below the permissible inference level was 94%. This suggests that the 5-ng law is an effective bar to DUI convictions, since the 5-ng limit does not apply to DWAI charges. It also suggests that drivers below 5-ng were “impaired to the slightest degree” but not “incapable of safe driving,” but note that plea bargaining makes DUI and DWAI definitions less definitive than they might otherwise be. Separating the effect of the 5-ng law from the different impairment definitions on THC conviction rates cannot be done using Colorado data alone. We could gain a better understanding of that issue if we had drugged driving conviction data from other states. Data in the Supplementary material show that 5 states statutorily define DUI as “incapable of safe driving,” an additional 4 states use that definition only for drugged driving, whereas 22 states use a less rigorous statutory definition such as “less safe to drive.” None publish drugged driving conviction data to enable quantification of the effect of differing DUI definitions.

Colorado had no measures of DUI/DWAI conviction rates prior to passing its 5-ng law. Therefore, there is no way to ascertain if the law made THC convictions easier, as was hoped when the law was passed. The 46% overall THC conviction rate for DUI irrespective of blood THC level suggests it may not have been effective in achieving that objective unless THC convictions were lower than 46% prior to the law’s enactment.

On the other hand, the 99% THC DWAI conviction rate (more than double the THC DUI conviction rate) for drivers impaired by THC only is strong evidence that a 5-ng legal limit was unnecessary as long as a DWAI conviction is deemed a satisfactory resolution to a DUI arrest. Prosecutors do not rely on the 5-ng law. 73% of all THC only convictions were for DWAI, not for DUI.

It is unclear why a prosecutor would charge a defendant with DUI and not both DUI and DWAI, accepting a guilty plea for DWAI if there is insufficient evidence that the driver was incapable of safe driving. One possible reason is that prosecutors have had a strong incentive to avoid a DWAI charge in cases of vehicular homicide and vehicular assault. Whereas DUI vehicular homicide and assault are Class 3 and 4 felonies respectively, DWAI vehicular homicide and assault were not even unlawful in Colorado until March 2022 when they were made Class 4 and 5 felonies respectively, with legal sanctions one-half what they are for DUI vehicular homicide and assault.

Limitations

Conviction rates are based on court filings, not on arrests. Arrests dismissed by prosecutors due to insufficient evidence are not included.

Toxicology data availability was limited by a high rate of refusal to comply with the state’s expressed consent law.

The first objective of this study, determining the prevalence of drugged driving in Colorado drivers convicted of DUI/DWAI, was limited to determining a very wide range estimate of prevalence due to limited data. Adoption of policies to test all DUI/DWAI suspects with a full panel of drugs including alcohol would be needed to tighten that estimate and to be able to perform statistical hypothesis testing of the estimate. Expense makes adoption of such policies unlikely.

Colorado did not begin collecting and analyzing DUI/DWAI data until four years after legalizing and two years after commercializing the sales of THC products. There is therefore no way to determine the impact of the state’s 5-ng THC permissible inference law before and after legalization and commercialization.

Comparisons of THC conviction rates for DUI and DWAI are muddled by two distinct differences between DUI and DWAI charges: DUI has a THC permissible inference level while DWAI does not, and the two charges have very different definitions. These two differences limit the power of inferences drawn by comparing data from different population subsets.

Conclusions

States should analyze and publish their drugged driving prevalence and conviction rates by impairing substance. Only by analyzing local data can a state understand what its drugged driving problem is and what it isn’t. Only by doing so can a state understand if its laws and policies are effective.

Available data do not permit us to determine the prevalence of drug impaired driving in drivers charged with DUI/DWI in Colorado. Nevertheless, by linking toxicology and judicial databases we can estimate the prevalence to be in the 39% to 63% range. The primary component of drugged driving was polydrug driving, with its primary component being driving under the combined influence of alcohol and THC.

Colorado’s 3-year average drugged conviction rates for driving under the influence of marijuana’s THC varied from 9% to 99% depending on the blood THC level, and the statute being charged. Charging a driver with DUI (incapable of safe driving) when the blood THC level was below the 5-ng/mL permissible inference level had a three-year average 9% conviction rate, compared with a 99% conviction rate of DWAI (less safe to drive) regardless of the blood THC level.

There are two principal differences between DUI and DWAI: different definitions of the offenses (incapable of safe driving and less safe to drive), and the application of the 5 ng/mL THC permissible inference law to DUI only. We cannot quantify the effects of these two differences on DUI/DWAI conviction rates. We know that if we use both an “incapable of safe driving” definition and a 5 ng/mL THC legal limit, we have substantially lower conviction rates than if we avoid both practices. However, we cannot yet specify the relative importance of these two differences.

Acknowledgements

The author gratefully acknowledges Statistical Analyst Allison Rosenthal, and Research Director Jack Reed from Colorado’s Office of Research and Statistics for their creation of the Driving Under the Influence of Drugs and Alcohol reports upon which this paper is based.

Disclosure statement

The author has no competing interests to declare.