Alcohol consumption can have a range of adverse impacts on the consumer, and these have been the subject of considerable study. Recently, researchers have started to focus on alcohol-related harms to people other than the drinker; described as the "collateral damage", "second-hand effects", or "negative externalities" of drinking.1-5Taking a systematic approach to describing and quantifying harm to others from drinking is important for two reasons. The first is to identify problems for specific attention that might otherwise be invisible or neglected. The second is to provide a more complete picture of the burden of drinking in communities to inform decision-making about policy on alcohol control. Advocacy based on harm to "innocent victims" has been a powerful influence in tobacco control.6Despite the obvious relevance to policy, attempts to quantify and cost the many impacts of alcohol on people other than the drinker, or to separate that burden from the overall toll of alcohol in the population have previously been uncommon.7-9 A number of countries, including New Zealand,10 have adapted the methodology of the Global Burden of Disease Comparative Risk Assessment (CRA)11 to demonstrate the scale of alcohol-related health harm, by synthesising data on alcohol's effects into summary measures of burden.The global CRA has shown harm worldwide to be almost equal to that of tobacco using these methods.12 However, because the CRA focuses on health conditions alone, and there is a lack of good data on many alcohol-related outcomes, substantial harm is unaccounted for in these analyses, particularly harm to others.People may be affected by the drinking of their partners, their families, their friends, their work mates, other people they know, or strangers. The collective drinking habits of communities also have an effect on people's lives. The impacts vary widely in nature and severity, from noisy neighbours to child neglect to fatal injuries.Some of those harmed by the drinking of others come to the attention of health and social agencies or the police, and such contacts are recorded in administrative databases. Many other affected individuals leave no discoverable trace of their experience, and the size of the burden will only be uncovered by a systematic approach such as a population-based survey. Thus, service use data and self-report provide complementary views of harms from the drinking of others.As a first step to address the lack of documentation of the range and magnitude of adverse effects of other peoples' drinking this study aimed to: Identify, analyse, and collate data on alcohol-related harm to others from existing administrative and survey databases in New Zealand; and Identify gaps in the data systems needed to estimate the magnitude of harm to others and monitor changes over time. Methods Surveys, research data, and administrative databases were identified through literature searching, examination of websites of relevant agencies, and direct enquiry among those working in research, government agencies and relevant NGOs. Accessible data were analysed, and published or collated data were summarised. No relevant useable data from the time period 2003-8 were excluded. Individual level data Health Behaviour Surveys 2003 and 2004 (HBS03/4)—Data were combined from the 2004 Health Behaviours Survey (HBS) - Alcohol Use and the 2003 Health Behaviours Survey-Drug Use, conducted for the Ministry of Health. The combined sample was made up of 16,480 New Zealand adults, aged 18-65, living in private residential dwellings. Data were collected between September 2003 and August 2004 for the alcohol survey (n=8397) and from April 2003 to November 2003 for the drug survey (n=7083). A full description of the methods used for these surveys is available.13 14 The questionnaires were based on previous National New Zealand Alcohol Surveys 1995 & 2000,15and the same questions were used in the Alcohol Use and Drug Use surveys for all of the items about alcohol consumption and experience of assault reported here. The surveys were weighted to adjust for sampling design, and a scaling factor was used to form new sample weights when combining data from the two surveys. Response rates were 59% for the alcohol survey and 68% for the drug survey. GENACIS-New Zealand survey (Gen07)—This 2007 national survey of 18-70 year olds (n=1924) sampled from the electoral roll and used postal questionnaires. It was carried out by researchers at the University of Otago and the response rate was 49%. The investigators used the core questionnaire from the GENACIS study, an international collaborative study of gender, alcohol and culture16 Further details of methods have been published17Respondents reported their own and their partner's alcohol consumption, and details of the most severe incident of partner aggression by the respondent and towards the respondent in the past 2 years. Mean scores for severity, anger and fear associated with these incidents of victimization and perpetration were analysed by gender and involvement of alcohol. Multinomial models estimated associations of drinking patterns with aggression to and from the respondent. Crash Analysis System (CAS)—Descriptive information about traffic crashes and injuries in New Zealand is available from the Crash Analysis System (CAS) of Land Transport New Zealand.18 CAS data are derived from traffic crash reports completed by police attending crashes and although reporting of crashes involving injury within a 24 hour period is mandatory, reporting is known to be incomplete.19 The reports classify injuries as fatal, serious or minor and typically all fatal crashes are reported. CAS data for 2003-2007 were extracted and analysed.20 Numbers of non-fatal crash injuries were adjusted for under-reporting, using conversion factors provided by the Ministry of Transport.21Crashes where a driver or other protagonist had a blood alcohol of more than 0.03g/100ml were considered to be alcohol-involved. Innocent victims in alcohol-related crashes were those who were injured when they were not either a drinking driver or a drunk pedestrian, cyclist or passenger who caused the crash. Motorcyclists were classified as drivers or passengers and combined with car occupants. Average costs of minor, serious and fatal injuries were obtained from the New Zealand Ministry of Transport.21 New Zealand Fire Service Commission (NZFSC)—The involvement of alcohol in fatal unintentional residential fires for the period mid-1997 to mid-June 2003 was reported in a study by Dr Ian Miller in 2005.22 Subsequently we analysed Dr Miller's database for the period mid-1995 to the end of 2006. Only deaths in residential settings and fires of unintentional causation were included (i.e. fires attributed to arson, suicide or homicide were excluded). Data were originally obtained from two sources: The NZFS Fire Incident Risk Management System (FIRMS), which records information collected at or near the time of the incident; and Inquest records, obtained through Coronial Services of the Ministry of Justice. (Inquest records provided a wealth of contextual information, such as behaviour of those involved before and during the fire, intention, and cause of death, that is obtained during the judicial process.) Community Sentiment Surveys—In July 2007, Brett MacLennan (University of Otago PhD candidate) surveyed residents of 7 local government areas of New Zealand to assess community sentiment toward alcohol problems and their regulation. An electoral roll sample (18+ years of age) and postal questionnaires were used. The overall response rate was 58% (n=1306). Further details of methods are available.23 Alcohol harm to others survey (SHORE 08/9)—This nationally representative telephone survey was conducted in 2008/09 by SHORE and Whariki Research Centre, Massey University. The sample comprised 12-80 year olds, using a complex sampling frame similar to the HBS03/04 surveys. The response rate was 64% (n=3068).24 Published data NZ Alcohol and Drug Use Survey 2007 (ADUS07)—This nationally representative survey of New Zealanders aged 16-64 years was carried out by the Ministry of Health in August 2007-April 2008 (n=6784). It measured self-reported alcohol and drug use behaviours among the usually-resident New Zealand population living in private dwellings, using a multi-stage, stratified, probability proportional to size (PPS) sample design. Interviews were conducted in respondents' homes, using a combination of face-to-face computer-assisted personal interview and audio computer-assisted self-interview. Response rate was 60%.25 New Zealand Crime and Safety Survey (NZCASS)—The 2006 NZCASS was conducted by the Ministry of Justice in a nationally representative random sample of 5416 people aged 15 and over living in private households. Face-to-face interviews were conducted in homes in February-June 2006. Participants were asked about being a victim of a type of crime covered by the survey since 1st January 2005. They reported the circumstances and impact of any offences they had experienced. The response rate was 59%.26 New Zealand Police (Police)—In 2009 the New Zealand Police published the National Alcohol Assessment report,27 summarising data for the 2007/8 year on alcohol involvement in crime, based on 15 police databases. These include the National Intelligence Application (NIA), National Homicide Monitoring System (NHMS), New Zealand Alcohol and Drug Abuse Monitoring Programme (NZ-ADAM), Communications and Resource Deployment (CARD), Family Violence database, Alco-link, Tactical Options Reports, Auckland Central Adult Sexual Assault Team (ASAT) Results In 2007, one in six adults aged 16-64 years (18.1%, 16.7-19.4) reported that they had experienced harmful effects on their friendships or social life, home life or financial position in the past year due to someone else's alcohol use (ADUSO7). This was higher than the proportion experiencing any harmful effects from their own drinking (12.2%), and differentially affected women (22.8% vs 17% of men) and younger people, with 35% of women between 18 and 24 years of age reporting harm. There was no overall association with socioeconomic deprivation, but those living in the lowest socioeconomic quintile reported significantly more harm to home life and financial position than the highest socioeconomic quintile as measured by NZDep06. Violence and crime: police data New Zealand Police reported that in the 2007/2008 year at least 31% of all recorded offenders were affected by alcohol (118,829 of 377,911 offences). This finding was consistent between three different data sources. (NIA, Alco-Link, NZ-ADAM). However, alcohol status was unknown in more than a quarter of cases. The proportion of offences perpetrated under the influence of alcohol amongst those with known status was estimated to be 46%, and this was considered a more realistic estimate overall.27 In the case of violent offending, the offender had consumed alcohol before committing the offencein at least 33% of cases (n=20,447). Police reported that in 49.5% of 489 homicides occurring between Jan 1999 and June 2008 either a suspect or a victim consumed alcohol prior to the incident (NHMS). The proportion of offenders affected by alcohol was greater (44%) than the proportion of victims who were affected by alcohol (35%). Almost one half (49.3%; n=241) of all homicides were family violence related homicides and 37% of these (n=89) involved either a suspect or a victim drinking alcohol prior to the offence (NHMS). Of alcohol-involved homicides, 56% occurred in residential areas, 31% in public places, 7% in licensed premises and 6% unknown location. All but one of the homicides on licensed premises were alcohol-related (NHMS). There were 19,388 recorded family violence assault victims in the 2007/2008 year, of whom 82% were women. The proportion of offenders affected by alcohol was recorded as 34%, and the proportion of victims who were affected by alcohol was estimated to be between 14% and 16% (Family Violence Database). Data from the Auckland Central police district revealed that 28% of victims of sexual offences were judged to have consumed alcohol prior to the incident (ASAT). At the same time, the national data on all recorded sexual offences (n=3652) showed that the proportion of sexualoffenders affected by alcohol at the time of offending was 15%. The low proportion of offenders recorded to have been affected by alcohol was considered an artefact of the delay in apprehending the offenders (ASAT). Table 1. Police data: proportion (%) of offences where the offender, the victim, or either the offender or the victim were affected by alcohol Type of offence Either (%) Offender (%) Victim (%) Violent offending Family violence Homicide Family violence related homicide Sexual offence 49.5 37 33 34 44 15* 14-16 35 28 *known to be very incomplete In incidents where police officers used tactical options ("use of force") it was estimated that 59% of offenderswere affected by alcohol (Tactical Options Report), and the proportion was 35% in cases when police officers employed Tasers (electro-shock weapon). Across all categories of offences, where alcohol was involved 42% of alleged offenders reported having had their last drink at a private residence, 18% at licensed premises, 18% in a public place, and for 20% location was unknown (Alco-link). Violence and crime: self-reported assault Health Behaviour Surveys—Using self-reported data from HBS 03/4 we estimated the 12-month prevalence of physical assault to be 6.8% (6.2-7.4) for men and 3.0% (2.6-3.3) for women, between the ages of 18 and 65 years. More than half (54%) of assaults involved a perpetrator who had been drinking. The 12-month prevalence of sexual assault was 0.4% (0.2-0.5) for men and 1.0% (0.8-1.2) for women, with 57% perpetrator drinking. For both types of assault, alcohol use was more strongly associated with a perpetrator from outside the family. These findings suggest that more than 62,000 physical assaults and 10,000 sexual assaults occur in New Zealand every year where the offender has been drinking. About half of all physical assaults reported in the surveys, whether or not they involved alcohol, were by a stranger. The distributions of "person responsible" differed by involvement of alcohol. In particular, where alcohol is not involved assaults were more likely to involve a member of the respondent's family; 35% compared with 23% in the alcohol-involved group. Alcohol-involved assaults were more likely to occur in a pub, bar or club, or on the street than assaults not involving alcohol, which more commonly occurred at the respondent's home. Medical attention was sought for 15% of physical assaults involving drinking by the assailant and 10% of those not involving drinking (p=0.17). Police involvement was reported for similar proportions of assaults with and without drinking by the perpetrator (26% vs 28%; p=0.70). Overall, in one year, about 17,000 assaults by someone who has been drinking involved police, and 10,500 required medical attention.28 NZCASS—In the NZ Crime and Safety Survey conducted by the Ministry of Justice29 41% of victims of interpersonal violence reported that the offender was under the influence of alcohol. This was most common for offences by strangers (49%) followed by sexual offences against women (44%), partner offences (37%), and offences by people well known (31%). On average 20% of victims of interpersonal crimes reported themselves to have been drinking alcohol prior to the violence. The offender only was reported to have been drinking in 31% of offences by strangers, 27% of sexual offences against women, 19% of offences by people well known and 17% of partner offences. The proportion of offences when only the victim was drinking was very small and similar across offence categories. Table 2. The involvement of alcohol in interpersonal violence (NZCASS) Variables Offences by strangers (n=426) Offences by partners (n=276) Offences by people well known (n=296) Sexual offences against women (n=137) All offences(n=1135) Person drinking Offender only Victim only Both offender and victim Neither offender nor victim Don't know % 31 <1 18 44 7 % 17 2 19 58 5 % 19 2 13 53 14 % 27 5 17 41 9 % 23 2 17 50 9 Adapted from Table 13 in Family Violence Statistics Report 200929 For offences involving force or threat of force, alcohol was reported to be more involved in crime in public places (47 %) than in private places (31%). Offences occurring in places of entertainment were most likely to involve an offender who had been drinking (79%), and a victim who reported drinking themselves (45%). Table 3. Proportion of victims of aggression reporting the involvement of alcohol at the time of offence (NZCASS)] Person drinking Public places % Private places % Places of entertainment % Offender Victim 47 26 31 9 79 45 Violence and crime: partner aggression Data from a 2007 general population survey (Gen07) showed approximately 15% of men and 12% of women reported an aggressive act by a partner, and 11% of men and 16% of women reported be

Alcohol consumption can have a range of adverse impacts on the consumer, and these have been the subject of considerable study. Recently, researchers have started to focus on alcohol-related harms to people other than the drinker; described as the "collateral damage", "second-hand effects", or "negative externalities" of drinking.1-5Taking a systematic approach to describing and quantifying harm to others from drinking is important for two reasons. The first is to identify problems for specific attention that might otherwise be invisible or neglected. The second is to provide a more complete picture of the burden of drinking in communities to inform decision-making about policy on alcohol control. Advocacy based on harm to "innocent victims" has been a powerful influence in tobacco control.6Despite the obvious relevance to policy, attempts to quantify and cost the many impacts of alcohol on people other than the drinker, or to separate that burden from the overall toll of alcohol in the population have previously been uncommon.7-9 A number of countries, including New Zealand,10 have adapted the methodology of the Global Burden of Disease Comparative Risk Assessment (CRA)11 to demonstrate the scale of alcohol-related health harm, by synthesising data on alcohol's effects into summary measures of burden.The global CRA has shown harm worldwide to be almost equal to that of tobacco using these methods.12 However, because the CRA focuses on health conditions alone, and there is a lack of good data on many alcohol-related outcomes, substantial harm is unaccounted for in these analyses, particularly harm to others.People may be affected by the drinking of their partners, their families, their friends, their work mates, other people they know, or strangers. The collective drinking habits of communities also have an effect on people's lives. The impacts vary widely in nature and severity, from noisy neighbours to child neglect to fatal injuries.Some of those harmed by the drinking of others come to the attention of health and social agencies or the police, and such contacts are recorded in administrative databases. Many other affected individuals leave no discoverable trace of their experience, and the size of the burden will only be uncovered by a systematic approach such as a population-based survey. Thus, service use data and self-report provide complementary views of harms from the drinking of others.As a first step to address the lack of documentation of the range and magnitude of adverse effects of other peoples' drinking this study aimed to: Identify, analyse, and collate data on alcohol-related harm to others from existing administrative and survey databases in New Zealand; and Identify gaps in the data systems needed to estimate the magnitude of harm to others and monitor changes over time. Methods Surveys, research data, and administrative databases were identified through literature searching, examination of websites of relevant agencies, and direct enquiry among those working in research, government agencies and relevant NGOs. Accessible data were analysed, and published or collated data were summarised. No relevant useable data from the time period 2003-8 were excluded. Individual level data Health Behaviour Surveys 2003 and 2004 (HBS03/4)—Data were combined from the 2004 Health Behaviours Survey (HBS) - Alcohol Use and the 2003 Health Behaviours Survey-Drug Use, conducted for the Ministry of Health. The combined sample was made up of 16,480 New Zealand adults, aged 18-65, living in private residential dwellings. Data were collected between September 2003 and August 2004 for the alcohol survey (n=8397) and from April 2003 to November 2003 for the drug survey (n=7083). A full description of the methods used for these surveys is available.13 14 The questionnaires were based on previous National New Zealand Alcohol Surveys 1995 & 2000,15and the same questions were used in the Alcohol Use and Drug Use surveys for all of the items about alcohol consumption and experience of assault reported here. The surveys were weighted to adjust for sampling design, and a scaling factor was used to form new sample weights when combining data from the two surveys. Response rates were 59% for the alcohol survey and 68% for the drug survey. GENACIS-New Zealand survey (Gen07)—This 2007 national survey of 18-70 year olds (n=1924) sampled from the electoral roll and used postal questionnaires. It was carried out by researchers at the University of Otago and the response rate was 49%. The investigators used the core questionnaire from the GENACIS study, an international collaborative study of gender, alcohol and culture16 Further details of methods have been published17Respondents reported their own and their partner's alcohol consumption, and details of the most severe incident of partner aggression by the respondent and towards the respondent in the past 2 years. Mean scores for severity, anger and fear associated with these incidents of victimization and perpetration were analysed by gender and involvement of alcohol. Multinomial models estimated associations of drinking patterns with aggression to and from the respondent. Crash Analysis System (CAS)—Descriptive information about traffic crashes and injuries in New Zealand is available from the Crash Analysis System (CAS) of Land Transport New Zealand.18 CAS data are derived from traffic crash reports completed by police attending crashes and although reporting of crashes involving injury within a 24 hour period is mandatory, reporting is known to be incomplete.19 The reports classify injuries as fatal, serious or minor and typically all fatal crashes are reported. CAS data for 2003-2007 were extracted and analysed.20 Numbers of non-fatal crash injuries were adjusted for under-reporting, using conversion factors provided by the Ministry of Transport.21Crashes where a driver or other protagonist had a blood alcohol of more than 0.03g/100ml were considered to be alcohol-involved. Innocent victims in alcohol-related crashes were those who were injured when they were not either a drinking driver or a drunk pedestrian, cyclist or passenger who caused the crash. Motorcyclists were classified as drivers or passengers and combined with car occupants. Average costs of minor, serious and fatal injuries were obtained from the New Zealand Ministry of Transport.21 New Zealand Fire Service Commission (NZFSC)—The involvement of alcohol in fatal unintentional residential fires for the period mid-1997 to mid-June 2003 was reported in a study by Dr Ian Miller in 2005.22 Subsequently we analysed Dr Miller's database for the period mid-1995 to the end of 2006. Only deaths in residential settings and fires of unintentional causation were included (i.e. fires attributed to arson, suicide or homicide were excluded). Data were originally obtained from two sources: The NZFS Fire Incident Risk Management System (FIRMS), which records information collected at or near the time of the incident; and Inquest records, obtained through Coronial Services of the Ministry of Justice. (Inquest records provided a wealth of contextual information, such as behaviour of those involved before and during the fire, intention, and cause of death, that is obtained during the judicial process.) Community Sentiment Surveys—In July 2007, Brett MacLennan (University of Otago PhD candidate) surveyed residents of 7 local government areas of New Zealand to assess community sentiment toward alcohol problems and their regulation. An electoral roll sample (18+ years of age) and postal questionnaires were used. The overall response rate was 58% (n=1306). Further details of methods are available.23 Alcohol harm to others survey (SHORE 08/9)—This nationally representative telephone survey was conducted in 2008/09 by SHORE and Whariki Research Centre, Massey University. The sample comprised 12-80 year olds, using a complex sampling frame similar to the HBS03/04 surveys. The response rate was 64% (n=3068).24 Published data NZ Alcohol and Drug Use Survey 2007 (ADUS07)—This nationally representative survey of New Zealanders aged 16-64 years was carried out by the Ministry of Health in August 2007-April 2008 (n=6784). It measured self-reported alcohol and drug use behaviours among the usually-resident New Zealand population living in private dwellings, using a multi-stage, stratified, probability proportional to size (PPS) sample design. Interviews were conducted in respondents' homes, using a combination of face-to-face computer-assisted personal interview and audio computer-assisted self-interview. Response rate was 60%.25 New Zealand Crime and Safety Survey (NZCASS)—The 2006 NZCASS was conducted by the Ministry of Justice in a nationally representative random sample of 5416 people aged 15 and over living in private households. Face-to-face interviews were conducted in homes in February-June 2006. Participants were asked about being a victim of a type of crime covered by the survey since 1st January 2005. They reported the circumstances and impact of any offences they had experienced. The response rate was 59%.26 New Zealand Police (Police)—In 2009 the New Zealand Police published the National Alcohol Assessment report,27 summarising data for the 2007/8 year on alcohol involvement in crime, based on 15 police databases. These include the National Intelligence Application (NIA), National Homicide Monitoring System (NHMS), New Zealand Alcohol and Drug Abuse Monitoring Programme (NZ-ADAM), Communications and Resource Deployment (CARD), Family Violence database, Alco-link, Tactical Options Reports, Auckland Central Adult Sexual Assault Team (ASAT) Results In 2007, one in six adults aged 16-64 years (18.1%, 16.7-19.4) reported that they had experienced harmful effects on their friendships or social life, home life or financial position in the past year due to someone else's alcohol use (ADUSO7). This was higher than the proportion experiencing any harmful effects from their own drinking (12.2%), and differentially affected women (22.8% vs 17% of men) and younger people, with 35% of women between 18 and 24 years of age reporting harm. There was no overall association with socioeconomic deprivation, but those living in the lowest socioeconomic quintile reported significantly more harm to home life and financial position than the highest socioeconomic quintile as measured by NZDep06. Violence and crime: police data New Zealand Police reported that in the 2007/2008 year at least 31% of all recorded offenders were affected by alcohol (118,829 of 377,911 offences). This finding was consistent between three different data sources. (NIA, Alco-Link, NZ-ADAM). However, alcohol status was unknown in more than a quarter of cases. The proportion of offences perpetrated under the influence of alcohol amongst those with known status was estimated to be 46%, and this was considered a more realistic estimate overall.27 In the case of violent offending, the offender had consumed alcohol before committing the offencein at least 33% of cases (n=20,447). Police reported that in 49.5% of 489 homicides occurring between Jan 1999 and June 2008 either a suspect or a victim consumed alcohol prior to the incident (NHMS). The proportion of offenders affected by alcohol was greater (44%) than the proportion of victims who were affected by alcohol (35%). Almost one half (49.3%; n=241) of all homicides were family violence related homicides and 37% of these (n=89) involved either a suspect or a victim drinking alcohol prior to the offence (NHMS). Of alcohol-involved homicides, 56% occurred in residential areas, 31% in public places, 7% in licensed premises and 6% unknown location. All but one of the homicides on licensed premises were alcohol-related (NHMS). There were 19,388 recorded family violence assault victims in the 2007/2008 year, of whom 82% were women. The proportion of offenders affected by alcohol was recorded as 34%, and the proportion of victims who were affected by alcohol was estimated to be between 14% and 16% (Family Violence Database). Data from the Auckland Central police district revealed that 28% of victims of sexual offences were judged to have consumed alcohol prior to the incident (ASAT). At the same time, the national data on all recorded sexual offences (n=3652) showed that the proportion of sexualoffenders affected by alcohol at the time of offending was 15%. The low proportion of offenders recorded to have been affected by alcohol was considered an artefact of the delay in apprehending the offenders (ASAT). Table 1. Police data: proportion (%) of offences where the offender, the victim, or either the offender or the victim were affected by alcohol Type of offence Either (%) Offender (%) Victim (%) Violent offending Family violence Homicide Family violence related homicide Sexual offence 49.5 37 33 34 44 15* 14-16 35 28 *known to be very incomplete In incidents where police officers used tactical options ("use of force") it was estimated that 59% of offenderswere affected by alcohol (Tactical Options Report), and the proportion was 35% in cases when police officers employed Tasers (electro-shock weapon). Across all categories of offences, where alcohol was involved 42% of alleged offenders reported having had their last drink at a private residence, 18% at licensed premises, 18% in a public place, and for 20% location was unknown (Alco-link). Violence and crime: self-reported assault Health Behaviour Surveys—Using self-reported data from HBS 03/4 we estimated the 12-month prevalence of physical assault to be 6.8% (6.2-7.4) for men and 3.0% (2.6-3.3) for women, between the ages of 18 and 65 years. More than half (54%) of assaults involved a perpetrator who had been drinking. The 12-month prevalence of sexual assault was 0.4% (0.2-0.5) for men and 1.0% (0.8-1.2) for women, with 57% perpetrator drinking. For both types of assault, alcohol use was more strongly associated with a perpetrator from outside the family. These findings suggest that more than 62,000 physical assaults and 10,000 sexual assaults occur in New Zealand every year where the offender has been drinking. About half of all physical assaults reported in the surveys, whether or not they involved alcohol, were by a stranger. The distributions of "person responsible" differed by involvement of alcohol. In particular, where alcohol is not involved assaults were more likely to involve a member of the respondent's family; 35% compared with 23% in the alcohol-involved group. Alcohol-involved assaults were more likely to occur in a pub, bar or club, or on the street than assaults not involving alcohol, which more commonly occurred at the respondent's home. Medical attention was sought for 15% of physical assaults involving drinking by the assailant and 10% of those not involving drinking (p=0.17). Police involvement was reported for similar proportions of assaults with and without drinking by the perpetrator (26% vs 28%; p=0.70). Overall, in one year, about 17,000 assaults by someone who has been drinking involved police, and 10,500 required medical attention.28 NZCASS—In the NZ Crime and Safety Survey conducted by the Ministry of Justice29 41% of victims of interpersonal violence reported that the offender was under the influence of alcohol. This was most common for offences by strangers (49%) followed by sexual offences against women (44%), partner offences (37%), and offences by people well known (31%). On average 20% of victims of interpersonal crimes reported themselves to have been drinking alcohol prior to the violence. The offender only was reported to have been drinking in 31% of offences by strangers, 27% of sexual offences against women, 19% of offences by people well known and 17% of partner offences. The proportion of offences when only the victim was drinking was very small and similar across offence categories. Table 2. The involvement of alcohol in interpersonal violence (NZCASS) Variables Offences by strangers (n=426) Offences by partners (n=276) Offences by people well known (n=296) Sexual offences against women (n=137) All offences(n=1135) Person drinking Offender only Victim only Both offender and victim Neither offender nor victim Don't know % 31 <1 18 44 7 % 17 2 19 58 5 % 19 2 13 53 14 % 27 5 17 41 9 % 23 2 17 50 9 Adapted from Table 13 in Family Violence Statistics Report 200929 For offences involving force or threat of force, alcohol was reported to be more involved in crime in public places (47 %) than in private places (31%). Offences occurring in places of entertainment were most likely to involve an offender who had been drinking (79%), and a victim who reported drinking themselves (45%). Table 3. Proportion of victims of aggression reporting the involvement of alcohol at the time of offence (NZCASS)] Person drinking Public places % Private places % Places of entertainment % Offender Victim 47 26 31 9 79 45 Violence and crime: partner aggression Data from a 2007 general population survey (Gen07) showed approximately 15% of men and 12% of women reported an aggressive act by a partner, and 11% of men and 16% of women reported be