Ignition Interlocks and Supporting Alcohol Control Technologies

Transcription

1 Ignition Interlocks and Supporting Alcohol Control Technologies Paul R. Marques, PhD. Senior Research Scientist, PIRE Adjunct Professor, UMD-Sch. Public Health

2 Webinar Topics Basic Interlock Operation Interlock installation Rates over Time Recidivism During Interlock Recidivism After Interlock Can Interlocks Solve the DUI Problem? First Offenders - Interlock? USA Interlock Program Types Stories told by the BAC Test Log File Alcohol Biomarkers and Interlock

3 Interlock Basics

4 Basic Features of an Interlock An alcohol sensor that detects BAC.02+ and prevents vehicle startups A running retest feature that requires periodic retesting while underway A tamper detection installation that detects efforts to alter the wiring A data-recording function that logs all activity: BAC results, stops, starts, refusals Anti-circumvention protocols

5 Important Distinction Interlock Devices the hardware Early devices were non-specific (Taguchi semiconductors their calibration would drift and react to many hydrocarbons); today most sensors are alcohol specific fuel-cell devices. Interlock Programs the procedures Laws (court or licensing programs), sanctions, regulations, reporting protocols, information flow, treatment links, consequences; the ways in which the devices are used.

6 Our (PIRE) interlock work (Marques, Voas, Tippetts): 6 Themes, 25 years (1) policy-relevant guidelines and research, (2) studies of interlock ability to control DUI recidivism, (3) the capabilities of the interlock BAC fail records to predict future DUI, (4) impact of a motivational, or other adjunct treatment services to interlock programs, (5) the relationship of interlock BAC data, and alcohol biomarker prediction of recidivism. (6) the impact of State interlock laws on outcomes

7 Installation Rates

8 Interlock Installation Rates - USA

9 Dick Roth chart NM is 60, KS CO 40, AZ 33 Interlocks per 10,000 residents by state Data from 9 of 10 providers; Plot by Richard Roth, PhD;August, New Mexico Arizona Washington Colorado Iowa Maryland North Carolina West Virginia Texas Kansas Nebraska Oregon Virginia Arkansas Louisiana Utah Missouri Alaska Wyoming Oklahoma Florida Michigan Illinois Massachusetts Idaho Montana Pennsylvania Ohio Georgia Nevada California Wisconsin Connecticut New York Delaware Kentucky New Jersey South Dakota New Hampshire Indiana Tennessee Mississippi Minnesota South Carolina Rhode Island Vermont North Dakota Alabama Hawaii DC Maine

10 Number of Interlocks Interlocks Projected to 2020 with Est. DUI Convictions (green) 1,400,000 1,200,000 1,000, , ,000 R² = , ,000 0 Year

11 Interlock Pros and Cons (briefly)

12 Why interlocks are endorsed 140% 120% 10 studies Non-interlock control groups for each study set to 100% 100% 80% 60% 40% 20% 0% 1 OH 2 OR 3 NC 4 AB 5 WV 6 MD 7 AB 8 QC 9 IL 10 NM During Interlock

13 Why we are not satisfied 140% Black line is non-interlock control groups for each study 120% 100% 80% 60% 40% 20% 0% 1 OH 2 OR 3 NC 4 AB 5 WV 6 MD 7 AB 8 QC 9 IL 10 NM During Interlock After Interlock

14 The Details Bar pair # 7 Alberta Interlock

15 Interlocks what we know These programs do reliably reduce recidivism while installed ( 65%) They do not lead to much carry-over once licenses are reinstated The most successful programs get only 50% enrollment (of all DUI convictions) For these (and other) reasons they cannot be a total solution What they do, they do well There is reason to believe they will have a general deterrence effect

16 Alcohol Action and DUI Arrests

17 How select a club are DUIs? What is the chance of arresting an impaired driver? One in (Values rounded) 250 Beitel et al, Self-report Study, Liu? Hingson et al., 1995 Miller, Spicer, Levy, 1999 Voas and Hause, 1987 Borkenstein et al., 1974

18 Let s be conservative on this One in 300 impaired operators are arrested (Voas and Hause) Estimated detection rate of impaired drivers with BAC.10 in a high enforcement zone (Stockton, CA)* undetected detected

19 Idealized BAC curves for Male 2-7 hours expected time to alcohol clearance after 1-4 drinks in an average size male BAC Equivalences 80 mg% = 80 mg/100 ml.08 g/dl =.8 g/l.08 % (near equiv.)

20 Is it fair to impose interlocks on first offenders? It is true that first offenders are somewhat less risky than multiple DUI offenders. As we showed earlier the chance of detecting any impaired driver is very remote maybe 1 in 300 or more How do arrest BAC s differ? How different are first offenders from NON-offenders? What are States doing?

21 BAC distribution: Arrested Drunk Drivers in NM average is.14 Data from Dick Roth

22 DUI % repeated by % 10% 15% 20% First Offenders Pose Much Higher Risk than Non-Offenders % Re-arrested within 3 Years 15% 10% 5% 0% Percent Re-Arrested for DWI Within Three Years vs Priors For Those Convicted of DWI in NM in 2003 New Mexico (% new DUI) based on 2003 status 25% 5x higher 20% None 1st 2nd 3rd 4th 5th+ DUI count in 2003 Conviction Number in 2003 Dick Roth et al., Impact DWI, NM Maryland (alcohol charges) per 1000 drivers 7x higher Prior DUI Bill Rauch et al, Westat, MD

23 Interlock Programs around the US

24 Administrative Programs Some ADVANTAGES Centralized: authority, criteria, monitoring Can be prompt installation after arrest Allows some offenders to drive legally while monitored Applicable to all arrested DWI offenders. Some DISADVANTAGES Large self-selection component Avoided by those who need it most More Administrative Appeal Hearings

25 Judicial Programs Some ADVANTAGES Can be made mandatory (electronic monitoring, jail are motivating alternatives) It eliminates self-selection It has the potential to get more interlocks installed per DUI. (e.g., > 60% of those arrested in NM, > 62% of convictions in Hancock County, IN.) Some DISADVANTAGES Judges often do not apply their laws Overburdened probation departments must monitor, personnel spread too thin Installation not immediately after arrest. Your honor, I don t own a car! And I have no intention to drive!

26 The next six slides are from IIHS Eric Williams presentation in May 2013 at the Association of Ignition Interlock Program Administrators The IIHS (Insurance Institute for Highway Safety) is a highly regarded safety research center in Virginia.

27 Optional alcohol ignition interlock laws April 2013 WA OR CA NV ID MT UT AZ WY CO NM ND SD NE KS OK TX MN IA MO AR LA WI IL MS NH VT ME MI NY MA IN OH PA NJ RI WV CT DE KY VA MD TN NC DC SC AL GA Judicial discretion only repeat offenders AK FL high-bac first offenders high-bac first & repeat offenders all offenders HI Data/slide from Insurance Institute for Highway Safety IIHS

28 Laws mandating alcohol ignition interlock orders April 2013 WA OR CA NV ID MT UT AZ WY CO NM ND SD NE KS OK TX MN IA MO AR LA WI IL MS NH VT ME MI NY MA IN OH PA NJ RI WV CT DE KY VA MD TN NC DC SC AL GA Judicial discretion only repeat offenders AK FL high-bac first offenders high-bac first & repeat offenders all offenders HI Data/slide from Insurance Institute for Highway Safety IIHS

29 Laws mandating alcohol ignition interlock orders April 2013 WA OR CA NV ID MT UT AZ WY CO NM ND SD NE KS OK TX MN IA MO AR LA WI IL MS NH VT ME MI NY MA IN OH PA NJ RI WV CT DE KY VA MD TN NC DC SC AL GA Judicial discretion only repeat offenders AK FL high-bac first offenders high-bac first & repeat offenders all offenders HI Data/slide from Insurance Institute for Highway Safety IIHS

30 Laws mandating alcohol ignition interlock orders April 2013 WA OR CA NV ID MT UT AZ WY CO NM ND SD NE KS OK TX MN IA MO AR LA WI IL MS NH VT ME MI NY MA IN OH PA NJ RI WV CT DE KY VA MD TN NC DC SC AL GA Judicial discretion only repeat offenders AK FL high-bac first offenders high-bac first & repeat offenders all offenders HI Data/slide from Insurance Institute for Highway Safety IIHS

31 Laws mandating alcohol ignition interlock orders April 2013 WA OR CA NV ID MT UT AZ WY CO NM ND SD NE KS OK TX MN IA MO AR LA WI IL MS NH VT ME MI NY MA IN OH PA NJ RI WV CT DE KY VA MD TN NC DC SC AL GA Judicial discretion only repeat offenders AK FL high-bac first offenders high-bac first & repeat offenders all offenders HI Data/slide from Insurance Institute for Highway Safety IIHS

32 Laws mandating alcohol ignition interlock orders April 2013 WA OR CA NV ID MT UT AZ WY CO NM ND SD NE KS OK TX MN IA MO AR LA WI IL MS NH VT ME MI NY MA IN OH PA NJ RI WV CT DE KY VA MD TN NC DC SC AL GA Judicial discretion only repeat offenders AK FL high-bac first offenders high-bac first & repeat offenders all offenders HI Data/slide from Insurance Institute for Highway Safety IIHS

33 Mandate Interlocks? Think for a minute: what does that mean? Mandate meaning that unless the offender installs he has to sit out his suspension period with no driving? Mandate meaning without an interlock he will never drive legally in that state again? Mandate meaning offender must obtain a car (even if he has none) and install before driving again? Mandate meaning he stays in jail unless he installs? Also a law may Mandate but that does not mean it is enforced.

34 Interlock Log Files the Data Record

35 The Interlock Data Record What is it? BAC levels: pass/fail, time/date stamp, failure to comply with requirements, engine on/off etc What can it tell us? Who warrants more watching? Who is using the interlock car and when? How many BAC tests are in there? Average 7/day x 365 = 2550/year/driver How can we use it to improve safety? Adjust interlock requirements or monitoring to reflect performance on interlock program.

36 First and Repeat Offenders give 6-8 BAC tests daily* Quebec BAC Tests per day * About half are start up tests Frequency Std. Dev = 4.63 Mean = 8.8 N =

37 How many BAC tests are taken over time? (each bar is 2 months) Alberta data 2009 So about 5-6 tests per day (half are starts)

38 BAC lockouts.04 (failure rate decline) - 2 month intervals Alberta data 2009

39 Interlocks: Fewer Fails, Less Recidivism The preceding slide demonstrated that the rate of failed BAC tests declines sharply over the duration of the interlock program. This effect has been found in every one of the 10 or so studies that examined it. Combine this with evidence of reduced recidivism during interlock. Recipe for enthusiasm.

40 Stories in the Data File day of the week Interlock BAC Tests.02 g/dl Tuesday low New Mexico 11 million tests

41 More on Tuesday 5.5 million tests Alberta 0.70% 0.60% 0.50% 0.40% 0.30% 0.20% 0.10% 0.00% Hifail >.08% Fail % Warn % SUN Mon Tues Wed Thurs Fri SAT 18.8 million tests Québec 1.00% 0.90% 0.80% 0.70% 0.60% 0.50% 0.40% 0.30% 0.20% 0.10% 0.00% Hi fail >.08% Mid Fail % Low Fail % DIMANCHE Lundi Mardi Mercredi Jeudi Vendredi SAMEDHI

42 USA 2007: Fatal Crashes with Driver BAC.08 by day of week; N=9884 Tuesday low

43 More stories in the Data File hr of day Alberta (fail >.04%) Interlock Fails Interlock Fails Workday (Monday-Friday) 7-8 am high 300 Weekend (Saturday-Sunday) high gm/dl 100 gm/dl Count HOUR of Day 8 10 noon Count HOUR of Day 8 10 noon X axis = 24 Hours of the Day

44 New Mexico, BAC.02% by hr of day - based on 11million tests Monday - Friday Saturday - Sunday 7-8 am high high

45 Morning high BAC most common: AB, QC, NM, TX Based on 40 million BAC tests am high Hour of Day Quebec All Tests Quebec Tests>=.02% Alberta All Tests Alberta Tests >=.02 Texas All Tests TX tests>.02%

46 Predicting Driver Risk: Combining Data File and Future Recidivism

47 24 Mo. AFTER INTERLOCK Predicting new DUI by the rate of failed BAC tests during interlock N=720 Offenders in Each Bar Data from Marques, Voas, Tippetts (2003) Addiction, Quebec Prediction Study Support from NIAAA DURING INTERLOCK BAC Rate.02%

48 7200 DUIs, Same data showing a third (time) axis months after interlock 18.8 million BAC tests Quebec

49 What the data say BAC tests average 6-8 per day or > 2500/yr. Failed BAC tests, as a percent of total tests taken, decline over time in the program. People with the highest fail rates during interlock have the highest recidivism after the interlock. Failed BAC test occur most often in the workday morning hours 7-9AM, weekend hours This morning failed test effect is believed due to prior night high consumption levels (and a potential treatment need indicator). About 20% of a cohort of interlock DUI drivers is likely to be high risk based on fail rates.

50 Alcohol Biomarkers with Interlock and Recidivism

51 Defined: Interlock BAC Fail Groups by Total Fail Rate (BAC.04 g/dl) Alberta Three BAC risk groups instead of ten 26.8% 52.8% 20.5%

52 Alcohol Biomarkers Markers of alcohol exposure that extend the duration of evidence long after Blood Alcohol Concentration (BAC) has fallen to zero. Different types and found in different sources: blood, serum, urine, hair, fingernails. Used widely in European nations to document drinking levels.

53 Alcohol markers to be discussed 2 general categories of markers indirect (chronic exposure) Due to organ damage or biological adjustment after repeated alcohol exposure examples: ALT, AST, MCV (poorer), GGT, %CDT (better) direct (recent exposure if in blood or urine, long exposure if in hair) directly reflect consumption via minor metabolic pathways, or are only formed in presence of alcohol examples: ethanol, PEth, EtG, EtS, FAEE (all good)

54 Mean, Median: 5 biomarkers, 6 categories of consumers. Marker levels with solid lines reference left axis; marker levels with dashed lines reference right axis

55 Driver Fitness Criteria Abroad Biomarkers as Decision Aids Sweden GGT, CDT, MCV, ALT, AST levels above 97.5% Percentile of normal Netherlands GGT, CDT Belgium CDT, GGT, AST, ALT, MCV France GGT, CDT, MCV Germany newly added EtG urine, EtG hair Comparative detection: urine 2.1%, hair 12.7% Norway, Finland marker panels Likely many others in Europe Australia, Canada studying evidence USA still MIA

56 Interlock data plus biomarker data 26.8% 52.8% 20.5%

57 Ten biomarkers by 3 interlock risk groups in Z (standard) scores *** *** *** *** *** *** ns * *** *** *** P<.0005

58 Biomarker evidence Not much change in drinking over the course of an interlock program But: This is not yet replicated by others This was documented in one jurisdiction Not enough statistical power to tease out different types of drivers (other than priors) If confirmed by others it means interlocks are only a temporary preventative

59 Blue baseline Green 8 months Over 8 months of interlock - no change in %CDT or GGT %CDT Mean +/- 2SE baseline to end GGT Mean +/- 2SE baseline to end

60 Blue baseline Green 8 months No change over 8 months on interlock in PEth or γ%cdt PEth Mean +/- 2SE baseline to end γ-cdt Mean +/- 2SE baseline to end

61 Sadly This means that while the interlock reduces recidivism and the rate of failed BAC tests, it does not appear to reduce drinking per se. Participants seem to drink more strategically but not less alcohol. This likely explains why recidivism rates return to control levels once interlocks are removed and licenses reinstated.

62 A (AUC) Recidivism Prediction 0.75 biomarkers Dui and self-report 0.7 Hair EtG A' ROC

63 A (AUC) BAC Fail Rate Prediction biomarkers Dui and self-report 2 methods PEth γ%cdt

64 Hair ethylglucuronide (hetg) Highest sensitivity (Høiseth et al., 2009) Most accurate detection of heavy drinking (Pirro et al., 2011) Best choice for assessing fitness in drinking drivers (Liniger et al., 2010) Detected 12.7% positives relative to 2.1% positives (in urine EtG) among German DUI drivers at relicensing (Dufaux et al., 2012) 30 pg/mg hetg is cutoff for alcohol abuse (Soc. Hair Testing 2010) Compare with current DUI sample: median 22 pg/mg, mean 51pg/mg. One third are over 50 pg/mg, two thirds over 30 pg/mg

65 Phosphatidylethanol (PEth) Formed only in presence of ethanol, so only ethanol can alter levels (Varga et al., 1998) Half life about 4 days, detectable in blood for 4 weeks after sobriety (Hansson et al., 1997) Highest specificity and highest sensitivity alcohol marker (Wurst et al., 2010) Significantly correlated with 9 other alcohol biomarkers, 16 psychometric scales, and rates of failed interlock BAC tests (Marques et al., 2010) Threshold ethanol intake to raise PEth in blood estimated at 50 g/day for 3 weeks (Viel, 2012)

66 Conclude from Biomarker data Hair EtG is strongest contender for predicting recidivism, dependence, and patterns of heavy drinking over extended time base. PEth seems the solid best predictor of failed BAC tests on interlock (and therefore drinkdriving proclivity). The interlock record and biomarker data should be used to help triage certain drivers (maybe 20-30%) toward alcohol treatment services.

67 Interlocks with Biomarkers Interlock is still a very good intervention that reduces recidivism while installed but it will not change drinking for the average person required to use one Supplemental evaluation with monitoring is warranted using objective biological methods to assess drinking levels It will be necessary to include competent expert treatment services to attain real behavior change There is no better time to do that than when someone is under interlock restriction

68 Some closing thoughts Annual interlock growth at about 10% continues apace much legislative support from MADD. Interlock is effective intervention for many Focusing only on arrested, convicted DUI offenders is unlikely to make the roadways safe from impaired drivers. Also, alcohol is just one drug, new designer drugs are available daily each often need guidelines. Medicines taken legally (opiates, benzos) are another large source of driver error. Impaired driver solutions may need to await autonomous cars to get rid of the driver