The link between alcohol and interpersonal violence is well established.1 According to a review approximation, in countries in which alcohol is commonly used, over 50% of assailants have been drinking prior to their offence.2 However, the consumption of alcohol as such does not necessarily involve an increased risk of interpersonal violence, but binge drinking seems to do so.3 4

The price of alcohol, through taxation in particular, is one of the central tools of alcohol policy. It is widely recognised that it affects the level of alcohol consumption and hence also of alcohol-related problems.5 Whether alcohol prices also affect the level of interpersonal violence is a question of interest. Evidence shows that lower alcohol prices are associated with higher consumption and higher rates of interpersonal violence, but population-level studies on this relationship are sparse, and natural experiments for examining this issue have been called for.6 One of the few studies in which the focus was directly on the price of alcohol and violent crimes suggested, on the basis of cross-sectional state-level comparisons in the USA, that a 10% increase in beer tax would reduce rapes by 1.3% and robberies by 0.9%, but would have little impact on homicides and assaults.7 On the whole, little is known about the effects of price reductions, as opposed to price increases, on interpersonal violence.

Several studies have focused on the aggregate-level association between alcohol consumption and violence. In Sweden, a relationship between alcohol consumption and homicide and assault rates in 1870–19848 was found and, in Finland, alcohol consumption was one of the four indicators that explained an increase in assault rates in 1950–2000.9 In Norway, an increase in alcohol consumption of one litre per capita predicted an increase of 8% in the violence rate in 1911–200310 and, in five European countries (including Finland) and in Ontario, Canada, homicide rates were positively related to alcohol sales or consumption.11 12

Alcohol use and interpersonal violence are both individual acts, and they are influenced by individual characteristics. However, the environment in which people live also matters. A number of studies have provided evidence that neighbourhood characteristics are related to interpersonal violence,13 14 whereas there is little direct evidence about the association between neighbourhoods and heavy or binge drinking. Indirect evidence in terms of alcohol-related problems, however, shows that this relationship exists.15 Explanations for the rates of interpersonal violence observed in community settings usually focus on one of two neighbourhood features: the characteristics of the people living in the neighbourhoods or the integral characteristics of the places in which they live.16 The former include factors such as area-level poverty, family structure and residential mobility,17 and the latter factors such as the proportion of abandoned housing and the level of retail activity.18 The use of population- rather than place-related characteristics works better in the European welfare setting where cities are better equipped to control their development by means of town planning than in the US context.19

However, the underlying mechanisms linking neighbourhood characteristics and various outcomes have remained uncharted for the most part. Neighbourhood differences are not “naturally” determined, but rather result from social and economic processes influenced by specific policies.20 It has been suggested that spatial embeddedness, internal structural characteristics and social organisational processes are all important in understanding neighbourhood-level variations in rates of violence.11 A review examining social processes related to problem behaviour and health-related outcomes covering over 40 studies emphasised neighbourhood ties, social control, mutual trust, institutional resources, disorder and routine activity patterns more than traditional characteristics such as poverty and residential instability.15

THE FINNISH AND THE HELSINKI METROPOLITAN AREA CONTEXT

Pricing and availability have been two major tools of the authorities in controlling alcohol consumption in the Nordic countries.5 In Finland, the state alcohol off-premise retail company has a monopoly on alcoholic beverages other than those with a low alcohol content, such as beer and cider with an alcohol content of less than 4.7% by volume. Alcoholic beverages have been heavily taxed, and the amount of alcohol that people are allowed to import from abroad has been strictly limited.

The year 2004 was a milestone in Finnish alcohol policy. On 1 January, it became legal to import practically unlimited amounts of alcoholic beverages for one’s own consumption due to the deregulation of import quotas within the European Union. On 1 March, taxes on alcohol were reduced by an average of 33%. The off-premise retail price of spirits went down by an average of 36%, wines by 3% and other alcoholic beverages by between 13 and 28%. The reason for the tax cuts was that Estonia was to join the European Union on 1 May, after which the deregulation of import quotas would also apply to alcohol imports from Estonia. This was expected to lead to a considerable increase in imports because of the proximity of the two countries and the significantly lower price of alcohol in Estonia. Tax cuts were aimed at restraining these imports. A unique natural experiment—such as that carried out in Finland in 2004, involving a radical reduction in the full price of alcohol (including both retail price and indirect costs)—has thus opened up the possibility of assessing the causality between the price of alcohol and interpersonal violence.

The increase in total alcohol consumption (the sum of recorded and estimated unrecorded consumption) in Finland was estimated to be 9.6% in 2004 and a further 1.9% in 2005.21 There are no data available on the total consumption in the Helsinki Metropolitan Area, but recorded sales increased by 5.4% (7.3% at the national level) in 2005 compared with 2003 (National Product Control Agency for Welfare and Health, unpublished table). If the ratio between recorded sales and unrecorded consumption was similar in the Helsinki Metropolitan Area to that in the whole country, this would translate to an increase of 8.7% in total consumption. Owing to its proximity to the Estonian capital (which means that imports may have increased more than average) and its roles as a centre of nightlife and tourism (which means that part of the changes in on-premise sales may be attributable to changes in the consumption of visitors to Helsinki), sales statistics do not necessarily reflect consumption by the metropolitan inhabitants, or its changes, reliably. Even if there is uncertainty in the Helsinki Metropolitan Area about the extent of the increase in alcohol consumption—the central intervening variable between alcohol price and violence—it is of great interest to examine what effects the price change had on interpersonal violence and also whether the effects varied by type of area.

Helsinki is the capital of Finland and had a population of 559 000 on 1 January 2005. Espoo and Vantaa are neighbouring cities of Helsinki that had 227 000 and 185 000 inhabitants respectively. Helsinki, Espoo, Vantaa and Kauniainen constitute the Helsinki Metropolitan Area with a population of 972 000, which was 18.6% of the total population of Finland in 2005. Finland is characterised by an ethnically homogeneous population and a low rate of economic inequality.22 An expressed objective of local housing and city planning policies in Helsinki is to prevent and reduce socioeconomic segregation among residential areas.23 As a result of the national context and local policies, the Helsinki Metropolitan Area does not feature areas that could be described as “truly disadvantaged” or “hypersegregated”.24 25 A previous study has, however, reported differences in mortality between areas in the Helsinki Metropolitan Area.26

The main purpose of this study was to assess to what extent the changes in the full price of alcohol affected the change in rates of interpersonal violence and two indicators of less severe disorderly conduct (custody due to alcohol intoxication and disturbing behaviour), and how these changes varied at the small-area level in the Helsinki Metropolitan Area. We had three research questions, the first of which was more general in nature. We examined whether:

1. measures of area-level social disadvantage were associated with interpersonal violence rates

2. the reduction in the price of alcohol was associated with interpersonal violence

3. the effects of the price reduction on interpersonal violence were associated with measures of area-level social disadvantage.

METHODS

Study sites and period

The sample for this study comprised 86 small areas (tracts) from the Helsinki Metropolitan Area, of which 33 belonged to Helsinki, 27 to Espoo and 26 to Vantaa and a small municipality within Espoo (Kauniainen). The tracts had populations ranging from 486 to 36 522, with a mean of 10 981, and were based on the administrative area division of the municipalities, which is used for policy purposes. We obtained data on the sociodemographic characteristics from administrative databases.27 28

The study period extended from the beginning of 2002 until the end of 2005. The total period was divided into two subperiods: (1) before the change in the pricing of alcohol (2002–2003) and (2) after the change (2004–2005). We thus had two symmetric periods to investigate in our analysis and could avoid potential bias due to the seasonal variation in interpersonal violence. A before–after comparison is more appropriate here than an interrupted time series analysis because time series analysis needs a clearly longer study period than is available in this study. Time series analysis requires at least 100 observations in order to detect a moderate effect of an intervention;29 but we only have access to 48 measurement points in our data.

Interpersonal violence and disorderly conduct

Our main outcome measure was interpersonal violence, but we also included two less severe indicators of disorderly conduct in our analyses. Data on interpersonal violence in the administrative areas were obtained from the Helsinki Police Department. The size of these areas was smaller on average in the Helsinki Metropolitan Area than in the municipal administrative areas but, by merging them, we obtained corresponding data for the purpose of analysis and also achieved correspondence in terms of area units. The data on interpersonal violence and disorderly conduct consisted of the number of various acts by tract and month in the years 2002–2005.

The data on interpersonal violence and disorderly conduct were obtained from three sources in the Helsinki Police Department. First, we considered crimes recorded by the police that were specified as offences against the Penal Code. The following outcome measures from this source were included in the analysis: assault and battery, including the subgroups assault in private homes and in public places, robbery and extortion, disturbance of domestic peace and rape. The second source of data comprised emergency call-outs related to domestic violence, disturbing behaviour and vandalism, and emergency responses in total. Approximately 80% of these cases were reported to the police emergency centre by the public. Third, we included a particular outcome measure: taken into custody due to alcohol intoxication was recorded as a police task but not as a crime.

Area characteristics

The choice of seven sociostructural area characteristics was based on prior knowledge obtained from previous studies on area differences in association between alcohol and crimes,30 as well as from studies on the effect of neighbourhood characteristics on mortality in the Helsinki Metropolitan Area.26 Six of the area characteristics were measures of social disadvantage: for those aged 15 years and over, we measured the proportion of people with basic education, mean personal income (€1000), manual class membership and unemployment (the last two being characteristics of the labour force), the proportion of single-parent families and the proportion of homes that were not owner-occupied. We also used a measure of residential instability (outmigration). All the area characteristics were measured in 2002–2003, except for manual class membership in 2000. All the area characteristics are relatively stable over time, and the ranking of neighbourhoods on these variables has changed very little. The area characteristic “proportion of 15–29 year olds” was left out because of its very minor effect.

Analyses

The frequencies and rates per risk population per year and per 1000 persons were calculated for all the different categories of interpersonal violence and disorderly conduct. Multiple linear regression analyses were performed using Stata, version 8, in order to evaluate the association between the area characteristics and assault, assault in private homes, domestic violence and custody due to alcohol intoxication. The choice of these outcome measures was based on their importance or their observed associations in the preliminary analyses. In order to take account of the correlatedness of the data over time, we used the “cluster” option, which affects the estimated standard errors and variance–covariance matrix of the estimators, but not the estimated coefficients.31 Weighting was used in linear regression analysis in order to take account of differences in the size of small-area populations.

We used two different models in assessing the association between area-level social disadvantage and interpersonal violence and disorderly conduct. First, we conducted the analyses separately for each of the area characteristics, chosen on the basis of earlier studies, which established their relevance in theoretical terms.26 27 Second, we used a backward selection regression procedure, with an incremental removal significance criterion of p>0.05, to determine the best model with suitable independent variables.

We excluded the centre of Helsinki from the analyses of assault and custody due to intoxication because of its role as the centre of amusements and nightlife in the Helsinki Metropolitan Area. If we had not done so, it would have biased the estimates of assault and custody rates in areas of concentrated disadvantage, as the centre of Helsinki represents a “good” tract with its low unemployment rate and high mean income, among other things.

We calculated rates for three indicators of interpersonal violence and for custody due to alcohol intoxication and changes in them over time according to the high, intermediate and low categories in the area characteristics. The high and low categories each covered 20% of the population aged 15 years and over and the intermediate category 60%. The change over time was tested by means of linear regression focusing on the interaction between time and area characteristics. Time was used here as a categorical variable and area characteristics as a continuous variable with values of 1 for the high-status category, 2 for the intermediate and 3 for the lowest status category.

RESULTS

The effect of the price reduction on interpersonal violence and disorderly conduct

Table 1 gives the descriptive statistics of the area characteristics used in the analysis.

The change in rates for interpersonal violence and disorderly conduct from 2002–2003 to 2004–2005 varied according to the type of crime or recorded task (table 2). Custody due to alcohol intoxication, assault and battery together with its subgroups (assault in private homes and public places and disturbing behaviour) increased by 3–5%, whereas domestic violence decreased by 7% and robbery, violating domestic peace and rape by over 10%. Emergency call-outs in total remained at the level that prevailed before the price reduction. However, all the estimates of change, except in the case of violating domestic peace, were statistically insignificant.

Area characteristics and interpersonal violence

Four dependent variables were chosen for further analyses on the basis of their importance or their observed associations in the preliminary analyses. The effect of area characteristics on three types of interpersonal violence and custody due to intoxication before the price reduction was examined by fitting two different regression models (table 3). When all the measures of social disadvantage were added into the model one by one (model 1), all except outmigration and manual class had a significant relationship to all four outcome variables. However, when all measures of social disadvantage were put in simultaneously, the effects overlapped (model 2). When we controlled for other factors that remained significant in the model, we found that the assault rate was higher in tracts with a higher proportion of people with basic education and outmigration, while other factors lost their significance. The controlled model for assault in private homes revealed a higher crime rate in tracts with a higher proportion of single-parent families. As far as domestic violence was concerned, the final model indicated a higher rate in tracts with a high proportion of manual class membership and unemployment. The controlled model for custody due to intoxication revealed a higher rate in tracts with people of a lower educational level and a higher outmigration rate.

The price reduction, area characteristics and interpersonal violence

Table 4 confirms the findings shown in table 3 that rates of all types of interpersonal violence were higher in low-status than in high-status tracts before the changes in the pricing of alcohol. Different indicators of interpersonal violence changed differently, depending on the area characteristics, from 2002–2003 to 2004–2005. We observed very little change in incidences of assault in high-status tracts. Interaction analysis showed that variations in change according to area measures were proportionally significant only in the manual class: the assault rate increased more in tracts with a high proportion of inhabitants of that status. The difference in the direction of change was similar for those with a low educational level, although not statistically significant.

Changes in incidences of assault in private homes did not differ by area characteristics. As far as domestic violence was concerned, there was some indication that the change was more favourable in tracts with higher unemployment and lower income. A similar pattern also emerged for other area characteristics (except outmigration), ie the rate of domestic violence decreased most in lower status tracts, but the change was not statistically significant.

The change in custody due to alcohol intoxication was the most adverse in tracts with high outmigration, and was close to zero in those with low outmigration. The manual class and home ownership variables showed the same pattern, but the association between the change and the categories of area characteristics was not statistically significant. That many of the seemingly sizeable differences remained non-significant was at least partly due to the distribution of the categories: intermediate categories covered 60% of the population.

DISCUSSION

This study showed that a drastic reduction in the cost of alcohol did not increase the rates of interpersonal violence and disorderly conduct in the Helsinki Metropolitan Area. Several studies that have examined the effects of price increases6 have uniformly suggested that changes in the price of alcohol have consequences in terms of alcohol-related harm, such as interpersonal violence. A reduction in the price of foreign-produced spirits in Switzerland has been reported to have been followed by increased self-reported alcohol-related problems (the effects on violence were not examined).32 Previous studies6 have raised expectations that the association between price and interpersonal violence is two-way, ie if an increase in the price of alcohol is observed to have been followed by a lower level of interpersonal violence, a decrease would result in higher levels. According to the findings of this study, this does not necessarily seem to be true.

It remains unclear why the rates of interpersonal violence did not increase in the Helsinki Metropolitan Area after a rapid change in pricing and availability and an increase in consumption. We have identified two possible factors that may explain this. First, in the Finnish context, the association between alcohol consumption and violence is not as strong as was expected on the basis of earlier studies. This factor is supported by a recent study based on time series analysis conducted at the national level, which did not demonstrate an increase in violent crimes in 2004.33 Second, there is some evidence that the increase in alcohol consumption that followed the price reduction was especially strong among heavy drinkers. For example, increase in custody due to intoxication resulted from an increased number of arrests among “regulars”, while the number of arrests among the more occasional arrestees did not increase.34 In addition, after the price reduction, alcohol-related mortality increased particularly strongly in the groups that had already experienced the highest levels of alcohol-related mortality, ie males living alone, without a job and those on disability pensions.35 The increase was also mainly due to alcoholic liver cirrhosis.36 It may thus be that the possible increase in the case of interpersonal violence due to increased consumption is confined to a relatively limited group of heavy drinkers. Further studies should try to identify these population subgroups.

This study provides support for previous work16 37 demonstrating a significant relationship between measures of area-level social disadvantage and interpersonal violence. It extends these findings by specifying and quantifying these relationships. The association was found to be different with regard to different types of interpersonal violence and for different area characteristics: a low educational level and a high outmigration level were the most salient. The role of education in creating inequality and contributing to other sociostructural characteristics is well established at the individual level.38 39 Moreover, a low educational level among individuals is reported to have an association with other types of harm, such as alcohol-related mortality.40 The findings of this study correspond with previous findings on the association between residential instability and interpersonal violence.15 It is also noteworthy that rates of assault in private homes were strongly related to a high proportion of single-parent families, and rates of domestic violence to a high proportion of unemployment and manual class membership when other characteristics were held constant.

Most previous studies on the association between interpersonal violence and disadvantage have been conducted in the USA and in metropolitan contexts in which disadvantage could be assumed to be concentrated to a much larger extent than in a Nordic welfare setting, where the state intervenes more in market processes in order to reduce poverty41. Finland is one of the most egalitarian societies measured by income inequality.42 Accordingly, the current study extends the relationship between measures of area-level disadvantage and interpersonal violence by providing support for the observation that this association exists even in a Nordic welfare state.

Overall, the differences in impact of the reduction in alcohol prices on interpersonal violence and disorderly conduct between high-, intermediate- and low-status areas were small. An adverse development in assault occurred in low-status tracts according to manual class membership and educational level, and in custody due to intoxication according to outmigration and manual class. In terms of domestic violence, however, the results were encouraging in two ways: the rate decreased on the whole, and most in the low-status tracts in which it was highest, although the association between time and area characteristics was statistically significant only in the case of unemployment and income. This favourable trend could be interpreted as a result of increased public intolerance of domestic violence, which has catalysed actions within society, in terms of legislation for instance, in order to reduce it.

Two issues need to be considered in the interpretation of our findings. First, the data on interpersonal violence that we used included only crimes about which the police had information, ie there was a report of an offence. Many criminal events fail to enter the records, however: they may not to be known to the police at all, or the police may not record them as crimes for a variety of reasons.43 However, the coverage of police records is high as regards severe crimes, such as the acts of interpersonal violence included in our analyses. Second, there are factors that affect the compilation of statistics on the number of crimes, such as changes in police control and in legislation. Our study period was too short to witness any remarkable changes in the readiness to report crimes, however, and there were no significant changes in legislation or police control concerning the types of crime used in our study during the study period. The frequent variation in police control was one reason for excluding drunken driving in our analyses, for instance.

Conclusions

Although previous studies indicate that the price of alcohol is a major tool affecting alcohol-related problems, this was not found in the case of interpersonal violence when alcohol prices decreased and consumption increased. Furthermore, the differences in the change in interpersonal violence between high- and low-status tracts were small. Consequently, it would appear that a drastic reduction in the price of alcohol does not necessarily lead to detrimental consequences in interpersonal violence or to an adverse development in areas of social disadvantage. However, additional studies on other types of harm, such as mortality and morbidity, are needed in order to reach a fuller understanding of the impact of the availability of cheaper alcohol.