The Impact of Interlock Installation on Driving Behavior and Drinking Behavior Related to Driving

Abstract

Objective

There is a substantial body of evidence that the recidivism of impaired-driving offenders is reduced while an ignition interlock device (IID) is on their vehicles. This study examines changes in driving behaviors and drinking behaviors used by DWI offenders to manage driving with the IID.

Methods

A total of 166 IID participants who completed two surveys covering the period from arrest to IID installation (T1) and during IID use (T2) were examined. Four domains were covered: demographics, driving environments and transportation needs, reported driving activity, and reported drinking activities. Participants were on average 38 years old, 43% were female, 35% completed college, 34% had an income of more than $50,000, and 83% were employed. For those who provided it, the mean blood alcohol content (BAC) at arrest was .184 g/dL, with only 8 (5%) individuals below .08 g/dL, and 93 (56%) at over .18 g/dL. About 45% were repeat DWI offenders.

Results

Between T1 and T2 there was a slight increase in acknowledging public transportation was available (p=.001), an increase in other individuals driving the interlock-equipped vehicle (p=.002), an increase in the number of vehicles in the household not registered to the DWI offender (p< .001), and an increase in the number of participants who reported that driving was important to their lifestyle (p=.008). Initial (T1) expectations about whether the interlock would be a problem were significantly different from actual experiences reported in T2 (p<.001). With respect to alcohol consumption, 14% reported abstinence at T2 compared to 2 % at T1 (p=.001) and the number of drinks per drinking occasion decreased from a mean of 4.90 at T1 to 3.14 at T2 (p=.001), but the number of drinking occasions increased by a third (p=.003). The number of drinking locations (p=.001), the frequency of stopping after work for a drink (p=.001), and drinking at a bar all decreased (p<.001).

Conclusions

Interlock users make some adjustments in how they drink, the amount they drink, and where they drink. This finding suggests that there may be methods that can be used to extend the benefits of the IID beyond the sanction period.

INTRODUCTION

For more than a decade, the number of alcohol-involved fatal crashes has remained constant at around 10,000 per year, suggesting there is a need for renewed efforts to impede drinking and driving (National Highway Traffic and Safety Administration 2018). One effective program that can profit from renewed effort is the installation of alcohol ignition interlock devices (IIDs) on the vehicles of all drivers arrested for impaired driving. IIDs require the driver to take a breath test each time the vehicle is started, which can prevent engine ignition if the driver has been drinking. Such engine lockouts are commonly set for blood alcohol concentration (BACs) equal to or greater than .025 g/dL in the U.S. The purpose of the device is to prevent high-risk drivers who have been convicted of driving while impaired (DWI) from driving after drinking.

Research studies comparing the recidivism rate of offenders while on the interlock with the recidivism of the same group following removal of the IID and with offenders with suspended licenses have found that while the device is installed, recidivism is substantially reduced. (Elder, et al.,2011, Blais et al. 2013) Research has also shown that while in the interlock program, offenders learn to avoid lockouts and reduce them by half (Marques et al. 1999). Finally, it has been shown that the number of lockouts accumulated while the IID is installed predict the user’s recidivism rate once the interlock has been removed (Marques et al. 2003; Voas 2020). This predictive feature can be used as a screening tool to identify DWI offenders most in need of and likely to profit from a treatment program (Voas et al. 2016; Voas 2020). However, using interlock performance to screen for additional treatment (Voas et al. 2016) has not been widely applied. Instead, states have tended to enact compliance rules which extend the time on the interlock of offenders with high numbers of lockouts (Voas, et al., 2020)

Traditional research studies comparing the recidivism of post-interlock offenders with DWI offenders who receive the traditional license suspension sanction have generally not indicated that those in IID programs have lower recidivism once the interlock is removed. (Coben and Larkin 1999; Elder et al. 2011; Blais et al. 2013). However, recent studies that have focused on the relationship of enacting an interlock law to the occurrence of fatal and alcohol-related crashes have documented 7% to 15% reductions in fatal crashes following the implementation of “all-offender” interlock laws, suggesting the possibility that aside from any effect on recidivism, interlocks (by raising the severity of the sanction for DWI) may act as a general deterrent to all drivers on the road (Kaufman and Wiebe 2016; McGinty et al. 2017; Teoh et al. 2018).

The use of the interlock is growing in the U.S. As of 2018, 50% of convicted DWI offenders are being required to install interlocks (Robertson et al. 2018). At the same time, studies of the characteristics of offenders who install interlocks and their behavior while the devices are on their cars gave lagged. This paper describes a pilot study from a larger program (NIAAA, Grant R01 AA022312), ultimately directed at developing support and treatment programs for those with interlocks installed on their vehicle to increase the probability that behavioral changes adopted to cope with the interlock will be carried over to the post-removal period. While research has demonstrated that interlock users reduce their recidivism while the IID is on their vehicle, there has been little work on an assessment of exactly how DWI offenders change their drinking and driving behaviors to produce the observed reduction

There is some evidence suggesting that drivers do not reduce the frequency of driving while the interlock is on the vehicle (Marques et al. 1999). The extent to which offenders with an interlock installed reduce their drinking is unclear. Marques and colleagues (2010) employed alcohol biomarkers in hair to measure change in alcohol consumption while the interlock was installed and found no change. Beck et al., in a retrospective study asked offenders on the interlock to compare current with pre-installation drinking behaviors and found evidence that those reporting change also reported better performance on the interlock. Forsman and Wallhagen (2019) showed a decrease in drinking frequency and consumption per drinking occasion while on the interlock in their study. The current study aims to expand our understanding of the behavior of IID users. This understanding is crucial to the success of the program while IIDs are in place and may indicate how to continue separating drinking from driving after IID removal.

METHODS

Funded by the National Institute on Alcohol Abuse and Alcoholism (NIAAA), the ongoing Managing Heavy Drinking (MHD) project (Grant R01 AA022312) is a longitudinal study of interlock users in an upstate county in New York. The effort includes the collection of data on the participants’ drinking/driving behaviors as they transition from post-arrest conditions to operating with an interlock. The data addressed in the current manuscript are part of that study.

Participants in the current study were convicted of a misdemeanor or felony DWI. Since August 15, 2010, disposition of these cases is supposed to include a conditional discharge that includes the installation and use of an IID in any motor vehicle the person owns or operates. The IID must remain in the vehicle for 12 months, unless otherwise lessened by the court. If an offender has no vehicle registered in their name and does not operate a vehicle, then the interlock requirement is not applied. Checks are done to determine if the person has a vehicle registered in their name and has installed an interlock. If an IID was ordered and not installed the judge can issue a bench warrant for the offender and impose more sanctions.

Potential study participants were informed about the study when they called the agency responsible for monitoring compliance with the court order for IID installation. Flyers were also provided to the IID installers in the area and to the impaired-driving program classes that all DWI offenders are mandated to attend. Potential participants were instructed to text, call, or email the study staff. Individuals were then contacted by study personnel to determine eligibility and book appointments to participate. All eligible participants completed an informed consent process. Survey Monkey was used to collect the data on computers from participants. However, fourteen participants had the survey questions read to them and answered on paper because of difficulty with technology. If transportation to the study site was an issue, a mobile van was used to go to a convenient location for the participant. This study was approved by the Pacific Institute for Research and Evaluation Institutional Review Board.

Study Design and Participants

This study was designed to collect data on the drinking and driving behaviors at the time before interlock installation (T1) and compare those practices after 6–12 months on the interlock (T2). The study was conducted in Erie County, New York, where during the study period, 2016 to 2018, there were 6,585 arrests for alcohol- or drug-related driving offenses with an interlock installation rate of approximately 27%. This provided 1,778 eligible individuals for this study. A total of 186 (10%) of those offenders were recruited for the study and completed the T1 survey, with 166 (89%) completing the T2 survey. Participants were compensated $150 for T1 and $100 for T2.

Measures

A short summary of the survey questions is presented here. (See the Appendix for a full list of questions.) Demographic characteristics including age, sex, education, marital status, income, and employment status were collected. In addition, information about prior convictions for DWI, prior license suspensions or revocations, and instances of being in a crash where someone was injured were also considered.

Individuals were asked about whether public transportation was available and if they used it, whether they rode a bicycle, and whether they had friends or family who offered transportation. In terms of vehicles, questions asked about the number of vehicles that were registered to the participant, the number of vehicles that were part of the household, and if they drove a vehicle that was registered to someone else. Other questions asked whether they drove for family or friends, drove their children to school, or drove a spouse or significant other to work, if they typically drove to work, how many days they drove in a typical week, and the frequency they drove after 10 p.m., excluding work-related driving.

Importance of driving was assessed by asking how important driving was to their lifestyle, entertainment, or to drive to sporting events. At T1, study participants were asked how they expected the interlock to impact their life (driving-related behaviors, drinking-related behaviors, personal issues, social issues, and whether the cost of the IID was a problem). These same questions were asked for the period they had the interlock on the vehicle (T2), with reference to whether the interlock actually had an impact on their life.

The study also considered drinking events and how these may have been influenced by the interlock: the average number of days in a month participants drank 5 or more drinks (4 or more for women) on an occasion; number of days they drank last week; and for the previous month, number of days they drank at lunch or after work; number of locations visited when drinking; number of drinks per drinking occasion; and whether someone else drove the interlock-equipped vehicle when they went out to a drinking location.

Data Analyses

Paired-samples t-tests were used for the continuous measures for comparisons between expectations and/or behavior at T1, and actual experiences and behaviors at T2. McNemar tests for the dichotomous measures and McNemar-Bowker tests for the ordinal measures were applied. When there were fewer than 5 respondents per cell for ordinal measures, the low frequency cell was collapsed with the higher adjacent cell for both time periods. For some of the Likert scaled questions, variables were recoded to assess lower (1,2,3) versus higher values (4,5).

RESULTS

Demographics

A short summary of results for demographic and personal characteristics is presented here. For a more detailed description, see the Appendix (Table A1). The mean number of days between surveys for T1 and T2 was 304. Average age was 37.9 years, 43% were female, 86% white, 52% single/never married, 83% employed, 35% held college degrees, and 34% had incomes of $50,000 or more. Additionally, the majority lived in the suburbs (66%), and approximately 4% were veterans. With respect to driving history, the mean years of driving was 20.7, with 45% having a prior arrest for an impaired-driving offense, 43% had prior license suspensions, and 18% had a history of being involved in a prior crash where injuries occurred. Finally, for those who provided a BAC, the mean BAC at the arrest that resulted in the court-ordered interlock being installed was .184 g/dL, with 56% having a BAC greater than .18 and only 8 (5.1%) having a BAC below .08, the legal limit. In addition, 18.2% reported using illicit or illegal drugs the night they were arrested, and 23% reported being involved in some type of crash.

Transportation

All significant effects and marginal trends are shown in Tables 1 and 2, with results for all variables shown in Tables A2, A3, A4, and A5 in the Appendix. Table 1 shows that only awareness of the availability of public transportation and the number of vehicles owned by others showed significant increases at T2

Table 1.

Significant Effects for vehicle and Interlock Measures T1-T2 Comparison (n=166).

		T1	T2	p value
Public Transportation Available	% Yes	59.9	72.2	0.001
# Cars in House Owned by Others	Mean (SD)	1.10 1.30)	1.58 (0.92)	< .001
	% None	34.6	11.7	< .001
Typically Drive to Work	% Yes	93.3	87.2	0.006
Importance of Driving to Lifestyle	% Quite a Bit/Extremely	81.2	90.3	0.008
Interlock a Problem	% A Problem	43	62.2	<.001
Annoyed Because of Interlock	% Very Annoyed	36.4	45.7	0.081
Late for Work	% Never	49.3	58.8	0.06
Other Interlock-Related Expenses	% Some/Considerable	26.7	36.3	0.066
Others Drive Car while interlock is on it.	% Yes	31.8	45	0.002

T1 and T2 indicate responses to interviews taken before or at the time the interlock was installed (T1) and while the interlock was in place (T2). SD denotes Standard Deviation

Table 2.

Significant Effects for Drinking Measures T1-T2 Comparison (n=166).

		T1	T2	p value
Past 30 Days Stopped for Drink After Work	Mean (SD)	2.37 (4.80)	1.21 (3.58)	<.001
# Days Had a Drink Last Week	Mean (SD)	1.28 (1.57)	1.70 (1.83)	0.003
# Drinks on Days They Drank	Mean (SD)	4.90 (3.53)	3.14 (2.99)	<.001
# Locations When Drinking	Mean (SD)	2.58 (1.00)	2.10 (0.82)	<.001
Stopped Drinking	% Yes	1.9	14.4	<.001
Drink in a Bar	% Yes	37.5	23.8	<.001
Prefer Beer	% Yes	54	60.7	0.099

T1 and T2 indicate responses to interviews taken before or at the time the interlock was installed (T1) and while the interlock was in place (T2). SD denotes Standard Deviation

Driving Behaviors

Importance of driving for lifestyle showed a significant increase. However, for driving to work, there was a significant decrease (see Table 1)

Interlock Performance

The only significant effects found for the interlock-related questions were for whether the interlock was a problem, where there was a significant increase in affirmative responses and for others driving the vehicle while the interlock was installed, with an increase at T2. Marginal effects were found for the IID being annoying, being late for work, or more expenses related to the IID than expected.

Drinking Behaviors

Given that the quantity and frequency of drinking may have already been suppressed by the recent conviction for DWI, it is not surprising that there were no significant changes on several drinking related measures. However, there was a significant increase for days drinking in the week prior to the survey, as well as significant decreases in the number of drinks consumed on each drinking occasion; the number of days participants stopped for a drink after work; the number of locations visited when they drank; and the percentage of individuals who typically drank at a bar and a marginal increase for preferring beer (see Table 2).

DISCUSSION

The objective of this study was to identify the behavioral changes in alcohol consumption or driving, or their combination, that produce the recidivism reductions while on the interlock that have been documented in the literature. Aside from recidivism, few studies have assessed the changes in drinking or in driving behaviors while offenders are on the IID. The current results are similar to those of Forsman and Wallhagen (2019), who found reductions in quantity and frequency of drinkingwhile on the interlock. In this study we found significant reductions in the number of days the offenders drank, the number of drinks they had on a day when they drank, and that 14 percent stopped drinking.

Considering specific drinking/driving behaviors, our results extend those of the Beck et al. (2015) one-time, retrospective study by prospectively documenting changes from a survey conducted prior to or at installation to a survey 3 to 11 months later. Beck and colleagues (2015) found that “adapters,” who reported changing their usual drinking location from bars and restaurants to their homes, versus “non-adapters,” who did not change their drinking location, reported having fewer drinks per occasion, drinking solo or only with a spouse/significant other, and changing their drinking plans and habits after installation. In the current prospective study, using the full range of poor to good performers, we found that, in contrast to Beck et al., there were no significant changes between T1 and T2 on a number of drinking/driving frequency measures (e.g., the percentage of participants who reported driving within 2 hours of drinking, driving after drinking too much, riding with a driver who had too much to drink, someone else driving the car when they go out drinking or drinking at home). However, we did find significant behavioral changes in more subtle behaviors related to drinking— decreases in the number of drinks consumed on each drinking occasion, and a reduction in the number of days participants stopped for a drink after work, the number of locations visited when they drank, and the percentage of individuals who typically drank at a bar—and a marginal increase for preferring beer over spirits.

While the great majority of the participants reported continuing to drink while on the interlock, 14% indicated they stopped drinking in the T2 survey. This carries with it some ramifications for treatment programs focused on abstinence as the goal. Using a harm reduction approach to treatment may reduce the problems by not demanding abstinence as the only goal; it can help individuals be more accepting of treatment. This approach may also decrease the quantity of alcohol consumed by the individual, which should lead to reduced strain on the body and impairment, and potentially lead to better choices when making decisions about separating drinking from driving. In 2016, Voas and colleagues advocated for additional alcohol treatment based on failed attempts to start the interlock-equipped vehicle due to an IID-recorded positive breath test for alcohol (Voas et al. 2016). The program that was implemented in Florida showed a significant reduction in subsequent arrests over the 12- to 48-month follow-up period once the interlock was removed from the vehicle. However, some DWI researchers suggest that treatment needs to be multimodal and should address a broad range of problems (Woodall et al. 2007; Mullen et al. 2015). Wells-Parker and Williams (2002) suggest the need to add brief interventions to increase the effectiveness of interventions for DWI offenders. Future work should assess the specific impact of the types of treatment for these individuals. The findings from this study may help provide information for how this approach may be implemented.

Overall, results show few changes in driving behaviors related to basic transportation needs (e.g., no differences between T1 and T2 for driving family or friends; driving the spouse to work; driving on the job, driving after 10 p.m.; number of days driving in a typical week or driving to sporting events), however there were several indications of change in driving behaviors to accommodate the interlock. Participants reported more occasions of others driving the interlock-equipped vehicle than they had anticipated, suggesting that they may have recruited friends to drive from drinking locations. They also reported more access to public transportation, which may actually indicate an increased use of public transportation options that had always been available to them. This may reflect the need to seek out alternative methods of transportation to meet the new requirements associated with the interlock such as a conditional license which limits where interlock users are permitted to drive such as work, treatment, education classes.

Another finding was an increase in the number of vehicles registered to others in the household. Because DWI offenders are required to install interlocks on all vehicles in their name, the increase in vehicles registered to others might indicate a reluctance by family members to drive an interlock-equipped vehicle and/or a method to avoid using the interlock-equipped vehicle by the offender. Other available vehicles in the household could offer them a vehicle to drive without an interlock. However, evidence within this study shows that the percent of offenders who admitted to driving other vehicles did not change across time, suggesting that the increase in household vehicles may not have been a trigger for driving other vehicles (Table A2).

Individuals also were more likely to indicate that driving was important to their lifestyle in the second survey. The added burden of the interlock may have underscored how reliant they are on driving. On the other hand, individuals also showed a decrease in driving to work, although the decrease was not statistically significant. The experience of the interlock may have reduced their desire to drive to work, finding alternatives for this transportation need.

Participants’ expectations for how the interlock would impact their lives differed significantly from how they reported it did influence their lives after 6–12 months on the IID. Compared to T1 expectations, at T2 individuals were more likely to indicate a moderate or a great problem while the interlock was installed, with some feeling annoyed, and some who experienced greater expenses related to the interlock than what they had expected at installation. Among the problems reported was an increase in being late to work. The specific experiences were not revealed in the study but could have been associated with difficulty in starting the engine or public transportation delays. The differences between the expected and actual impact of the interlock might provide insights into future behavior. The interlock being experienced as more of a problem and more annoying than expected could add to its perceived sanction severity and support the deterrent effect documented in the Kaufman and Wiebe study (2016). On the other hand, while the negative interlock experience could induce some offenders to abandon the safety behaviors adopted while on the interlock and revert to pre-interlock behaviors after IID removal, it might induce others to avoid future DWI offenses. A better understanding of these possible paths is needed.

There are some limitations in this study. The sample was relatively small (n=166) and was self-selected (not random), which indicates a need to use caution when considering the results. The current study had a higher rate of female participation (43%) than might be expected; the overall current rate of female DWI offenders in Erie County for the same time period of this study was 27%. The higher rate of female participants is an issue that needs further research. It is a question of whether females install interlocks at a higher rate than that of the general DWI population or that females may be more receptive to recruitment efforts for research studies. In terms of repeat offender status, the overall rates tend to be in the 33 to 35% range (STOP DWI, Erie County), while for the current sample the rate was around 45%. Given that the interlock is a court-ordered sanction for misdemeanor or felony DWI offenses, it makes some sense that the percent of repeat offenders in the study might be higher than the overall population of DWI offenders. Furthermore, while this study is unique in its identification of behavior changes associated with IIDs, the data collected do not allow for comprehensive explanations about why the changes took place. Although we could identify associations between behavioral changes and IID expectations, assessing the direction of these associations (i.e., causality) was not always possible. This is important for future work that the MHD study will address in subsequent publications.