12th May 2017, Volume 130 Number 1455

Taisia Huckle, Khoon Ching Wong, Karl Parker, Sally Casswell

The New Zealand context

New Zealand is a high-income country with over 4.2 million residents and a per capita alcohol consumption, for those 15 years of age and above, of 9.5 litres of absolute alcohol in 2008.1 Analysis has suggested that in New Zealand, as elsewhere, alcohol is one of the most important risk factors for avoidable injury and mortality in early and middle adulthood, and contributes substantially across the life course.2 The contribution of alcohol to the burden on health services in New Zealand, such as the emergency department, is high.3 In 2008, an estimate of the monetary costs of alcohol in New Zealand showed that alcohol cost New Zealand between 3.6 and 4.5 billion dollars.4

Consumption differs by age and socioeconomic status and contributes to the inequalities found in New Zealand society.5,6 The impacts of heavy alcohol use go beyond the drinker to those in the drinkers’ environment; the health and wellbeing of New Zealanders is lower in those most exposed to heavy drinkers compared to those not exposed.7

Alcohol’s harm to others

The importance of measuring alcohol’s harm to others has received increasing emphasis in the research literature; not only because it’s an important way in which the health and wellbeing of individuals may be reduced, but because estimates of alcohol’s harm to others is likely to be important for informing Global Burden of Disease and Injury estimates. Currently, alcohol-attributable fractions measure harm caused to the drinker while estimates of harm to others are excluded, meaning these statistics are underestimates of the total impact of alcohol. A first step is to begin to quantify alcohol’s harm to others; effects otherwise described as the “collateral damage”, “second-hand effects” or “negative externalities” of drinking.8

A number of specific studies have now been conducted assessing alcohol’s harm to others. Studies from New Zealand and Australia reported that those with a heavy drinker in their life experienced reduced health and wellbeing.7,9 In the European Union (EU), very conservative estimates of harm to others (based mainly on drink-driving, homicides and fetal alcohol syndrome) find that between 3% and 4% of the overall alcohol-attributable deaths in the EU are caused by harm to others.10 In six European countries—Denmark, Finland, Iceland, Norway, Sweden and Scotland—the proportion of the survey respondents experiencing physical harm by a drunken person ranged from 2.6% in Denmark through to 5.7% in Finland.11

Service use by those with heavy drinkers in their lives

While some impacts of alcohol’s harm to others are known, there are other areas that have been less well investigated, including the burden placed on services because of another’s drinking. Where studies are available they have mainly assessed the use of services among those affected by dependent substance users—drug as well as alcohol users12,13—and have generally found that family members of dependent substance users utilise health practitioners more frequently or are more frequently hospitalised compared to families without a dependent substance user. A meta-analysis of 24 emergency department studies across 14 countries reported that the perpetrator was suspected to have been drinking in 52.5% of assaults presenting to emergency departments and in 23% in cases had definitely done so.14 A study in New Zealand by Connor et al15 showed that more than 62,000 physical assaults and 10,000 sexual assaults occur every year, which involve a perpetrator who has been drinking. Of these, 10,500 incidents required medical attention and 17,000 involved police.

In New Zealand, approximately 30% of police work is alcohol-related.16 The proportion who have been drinking is similar for those treated for injury in urban hospital emergency departments.3 However, the burden placed on such services related to another’s drinking has not been reported. There are data on those who seek help from health providers to reduce their alcohol use; primary healthcare physicians—general practitioners and counsellors—are those most commonly approached.17 However, again, how many of those affected by another’s drinking who seek help from these sources in New Zealand is not known.

An Australian study assessed service use among those with a heavy drinker in their life among a general population survey sample. This study reported proportions accessing help and the demographic factors and level of harm from others drinking that predicted accessing help in the general Australian population. This study reported that 13% of respondents had called the police because of someone else’s drinking and 4.5% had used a health related service in the previous 12 months. Key factors that predicted service use were the level of harm experienced from a drinker (as reported by the respondent), not having a partner and place of residence.18

Several factors may play a part in whether people seek help because of someone else’s drinking or how often they do so, such as how many heavy drinkers they have in their life or whether they cohabitate with a heavy drinker or not. As such, this study will predict the use of services due to someone else’s drinking among the general population based on an exposure to heavy drinkers index. This index has been used previously in work by Casswell et al 2011, and was created to examine the impact of exposure to heavy drinking and is based on an overall measure of numbers of heavy drinkers in the respondents’ lives and household. This index has previously appeared to capture aspects relevant to the respondents’ lives and/or households,7 and the value of using it in the current analyses is that it allows for cumulative effects of exposure to heavy drinkers, if any, to be estimated in a respondent’s life. Other factors which may affect service use due to others drinkers may be related to the person affected, such as demographic characteristics or their own consumption of alcohol.

This study, then, reports population estimates of service use because of someone else’s drinking in New Zealand, examines demographic predictors of such service use and investigates whether greater exposure to heavy drinkers relates to greater service use.

Methods

Data were collected using an in-house Computer Assisted Telephone Interviewing (CATI) System during 2008/9. Randomly generated landline phone numbers were generated to cover the whole country, and sampled in proportion to the usually resident population aged 12–80 years in the number’s area. Telephone coverage in New Zealand was fairly high in 2008: approximately 92% of households had landline telephones. Certain sectors of the population are under-represented among those with access to a landline telephone and these are Māori (the indigenous people of New Zealand), Pasifika and single-parent households.19,20All eligible people in the household were enumerated, and one respondent was randomly selected by computer algorithm. Once a phone line had been recognised as a residential line, at least 10 calls were made at different times of the day and days of the week to attempt to reach a respondent. A high level of quality control is ensured by means of interviewer training, ongoing quality checks and supervision to ensure consistency of data collection (for further details see Casswell et al, 200221).

The sample size was 3,068 and response rate 64%. This response rate was calculated using the formula: number of eligible responding/(the number of eligible responding + number of eligible non-responding + estimated numbers of eligible from the unknowns) x 100. Our method of calculating the estimated numbers of eligible from the unknowns is comparable to the AAOPR #3 response rate method (which uses the proportional allocation method to estimate the eligibles from the unknowns). Respondents were eligible if they were aged 12–80 years and had lived in New Zealand for at least 12 months.

The unweighted sample was reasonably representative of the New Zealand population aged 12–80 (Census 2006, see for example22,23,24). Weighting was applied to correct for respondent selection probabilities, to weight one of our area strata and to match the survey weights to New Zealand 2006 Census population distributions using Rim Weighting, for groups based on gender, age and ethnicity. Lastly, standardisation to match the weighted sample size back to the initial survey size was undertaken. Mean weight was 0.99 with standard deviation
of 0.56.

Full ethical approval for this project was given by the Massey University Human Ethics Committee.

Measures

All measures were asked concerning the previous 12 months.

Heavy drinking associates: Respondents were asked: ‘are there any people in your life whom you consider to be a fairly heavy drinker or someone who drinks a lot?’ If they said “yes”, they were asked to think about the first ‘heavy drinker’ in their life and state their relationship to that person, and how much of the last 12 months they had lived in the same household as the person. The respondents were then asked to think about the next heavy drinker in their lives. Respondents could report up to 10 heavy drinkers.

Service use because of someone else’s drinking: Respondents were asked about use of services because of someone else’s drinking, including calling the police and using health services specifically; requiring medical treatment at a general practitioner (GP) or after hours doctor, at a hospital/emergency department or requiring counselling/professional advice. The number of times respondents used these services was asked about; response options ranged from never to daily.

Demographic variables: age, quadratic age (reflecting the non-linear relationship found), gender, ethnicity (European origin; Māori; Pasifika; Asian), marital status (married/partner; divorced; single), employment status (full time; part time; students; unemployed/sick; retired; parenting), educational achievement (University degree; postgraduate degree; professional certificate; diploma; trade/technical certificate; secondary certificate; non-secondary certificate) and income (no income; less than NZ$15,000; NZ$15,000–30,000; NZ$30,000–50,000; NZ$50,000–70,000; NZ$70,000 plus).

Respondent’s own drinking: was assessed using a within-location beverage-specific measure, which achieves a high coverage of population-level consumption.21 This obtains frequency and typical quantities consumed in a number of mutually exclusive locations.

Analysis

All statistical analysis was undertaken using SAS (Version 9.2) and significance was declared at p<0.05.

Index of exposure to heavy drinkers

An index of respondent’s exposure to heavy drinker(s) was derived in order to account for the cumulative effect of exposure where respondents had multiple drinkers in their lives and, if relevant, the time the heavy drinker lived in the household, as previous research has found that heavier drinkers can have greater impacts on others when they live in the same household.25 Weights were used only to categorise respondents; they were not used in the model itself. For each heavy drinker, weights were assigned 1: not/occasionally living in same household: 1.5: sometimes; 3: half of the time; 4.5: most of the time; 6: all of the time. Weights were summed across all heavy drinkers reported by the respondent and scores were categorised into three groups for analysis. Testing revealed that the weights showed consistency.

Level 0 = No heavy drinkers in life (n=2,173); Level 1 (Weight 1) (n=500); Level 2 = (Weight 1.1–3) (n=237); Level 3 = (Weight 3.1–6) (n=158). Due to lower numbers in Level 3, Level 2 and 3 were combined for analysis.

Analyses of service use

Descriptive analysis was undertaken to determine proportion of respondents who reported using each of the services in the past 12 months at least once. Additionally, the three health services asked about were combined to give an overall proportion of respondents using any of these (at least once). Logistic regression was conducted to predict respondent demographics against any service use in the past 12 months because of another’s drinking (yes/no). Respondent’s own consumption was also included (typical quantity in a drinking occasion and frequency). A proportional odds model for a univariate ordinal response was used to predict the relationship of exposure to heavy drinkers. Each model controlled for all demographics and the respondent’s own consumption (covariates).26,27

Results

Proportion of New Zealand population using types of services

Ten percent of New Zealanders reported having called the police at least once in the past 12 months because of someone else’s drinking, corresponding to 378,843 New Zealanders making at least one call to police (when converted to a proportion of the total population in 2008 aged 12–80 years). Almost 7% of the sample, representing 257,613 New Zealanders, reported requiring health services at least once for the same reason. Specifically, around 2% required medical treatment from a general practitioner or after-hours doctor, around 2% went to a hospital/emergency department and around 2.5% received counselling/professional advice because of someone else’s drinking (Table 1, first column).

Table 1: Prevalence of service use because of other’s drinking, and its prediction by index of exposure to heavy drinkers. 

Services

%

Exposure to heavy drinker (level 1) odds ratio

Exposure to heavy drinker (level 2) odds ratio

Linear contrast

p-value

Had to call police

10

1.41 (1.01, 1.98 )*

2.89 (2.12, 3.93 )*

<0.0001

Had to get medical treatment at GP or after-hours doctor

1.8

1.35 (0.64, 2.86 )

3.78 (2.08, 6.89 )*

<0.0001

Went to a hospital/emergency department

2.4

1.99 (1.02, 3.88)*

3.74 (2.09, 6.70)*

<0.0001

Got counselling/professional advice

2.6

2.08 (0.93, 4.67 )

8.53 (4.65, 15.65 )*

<0.0001 

Index of exposure to heavy drinkers

The estimates in Table 1 (2nd–4th columns) show that, while controlling for a range of demographic factors and respondents’ own consumption, lower exposure to heavy drinkers was not related to “getting medical treatment at a general practitioner or after-hours doctor” or for “getting counselling/professional advice (because of someone else’s drinking)”, but significant relationships were found for calling the police and going to a hospital/emergency department. Those with lower exposure were 1.4 times more likely to call the police or 1.9 times more likely to have gone to a hospital/emergency department than those with no heavy drinkers in their life because of someone else’s drinking.

Significant relationships were found for those exposed to higher levels of heavy drinkers for all variables investigated. Those with higher exposure to heavy drinkers were 2.9 times more likely to have called the police; 3.8 times more likely to have received treatment at a GP or after-hours doctor, 3.7 times more likely to have gone to a hospital or ED and 8.5 times more likely to have received counselling or professional advice because of someone else’s drinking (compared to those with no heavy drinkers in their life).

Respondents’ own characteristics predicting service use

Table 2 shows how respondents’ own characteristics predict whether or not they had used any service (yes/no) because of someone else’s drinking in the past 12 months. The findings show that older age, being Māori or Pasifika predicted using services because of someone else’s drinking. With regard to income, being in the middle income groups, relatively speaking, predicted service use. Respondent’s own drinking—including whether they consumed alcohol in the past 12 months—did not predict ever using a service because of someone else’s drinking. Living with a partner also did not predict ever using a service because of someone else’s drinking.

Table 2: Logistic regression: respondents’ own demographic characteristics and consumption predicting service use.  

Parameter

Odds ratios

LCI

UCI

P-value§

Age

1.06

1.01

1.12

0.020

Age in quadratic

0.99

0.998

1.00

0.003

Gender

Female

B

 

 

 

Male

0.92

 

 

0.494

Ethnicity

 

0.72

1.17

 

European

B

 

 

 

Asian

0.78

0.51

1.21

0.270

Māori

1.50

1.11

2.04

0.009

Pasifika

1.98

1.34

2.91

0.001

Marital status

Partner (married/de facto)

B

 

 

 

Single

0.94

0.70

1.27

0.690

Widowed/divorced/separated

1.29

0.85

1.95

0.230

Current employment status

Full-time employee

B

 

 

 

Part-time employee

0.99

0.66

1.47

0.954

Student

1.01

0.60

1.69

0.981

Unemployment or sick/on invalid

1.51

0.85

2.66

0.157

Retired

0.99

0.48

2.05

0.980

Parenting

1.19

0.72

1.95

0.495

Education level

Did not complete secondary school

0.94

0.59

1.48

0.775

Completed secondary school

1.10

0.75

1.61

0.626

Trade or technical certificate

1.17

0.76

1.79

0.473

Diploma

1.20

0.75

1.92

0.441

Professional qualification

0.69

0.32

1.48

0.338

University degree

B

 

 

 

Postgraduate degree

1.44

0.92

2.24

0.111

Personal income 

No income

1.26

0.65

2.44

0.495

<$15,001

1.50

0.82

2.72

0.185

$15,001–$30,000

2.11

1.29

3.44

0.003

$30,001–$50,000

1.61

1.02

2.53

0.039

$50,001–$70,000

1.54

0.98

2.43

0.061

$70,001+

B

 

 

 

Own drinking

Log of occasion quantity (ml)

1.04

0.88

1.23

0.631

Log of annual frequency

1.04

0.95

1.13

0.417

Drinker

No

B

 

 

 

Yes

0.93

0.45

1.90

0.833

B: reference category LCI: confidence interval—lower limit; UCI: confidence interval—upper limit.
§P-value at the 5% level of significance. 

Discussion

The findings of this study are the first to show the extent of service use because of others’ drinking in New Zealand. In 2008, when the survey was conducted, an estimated 378,843 (or 10%) of New Zealanders made at least one call to the police, and 257,613 (or 6.8%) required a health-related service because of someone else’s drinking. These population estimates found in this study are in line with those found in Australia, which is New Zealand’s nearest neighbour and which had a relatively similar level of per capita consumption in 2008 (10.32 litres absolute alcohol in Australia compared to 9.5 litres in New Zealand).1,28 In Australia in 2008, 13% of the population had to call the police at least once in a 12-month period, and 4.5% used a health-related service because of someone else’s drinking.18

The index we created to examine the impact of exposure to heavy drinking provided an overall measure of numbers of heavy drinkers and co-habitation and appeared to capture aspects relevant to the respondents’ lives. This was evidenced by the relationships found in the data, which generally showed that the extent of exposure to heavy drinkers in respondents’ lives was related to increased odds of services being utilised because of someone else’s drinking.

In this study, those with greater exposure to heavy drinkers (which included cohabitation as one factor) had increased odds of service use. In some cases the odds were relatively large, including for “had to get medical treatment at a general practitioner or after-hours doctor” or “an emergency department/hospital”, where respondents were almost four times more likely to have done so compared to those who had no exposure to heavy drinkers in their lives, “received counselling/professional advice”, where those exposed to heavy drinkers were over eight times more likely to have received counselling/professional advice. These findings are consistent with the wider literature showing that co-habitation with a heavy drinker is associated with greater impact.7,25

With respect to the respondent’s own demographic characteristics: older age, being Māori or Pasifika, and having a higher income predicted using each of the services because of someone else’s drinking. Living with a partner did not predict using a service because of someone else’s drinking, even though those most exposed to heavy drinkers, as measured by the exposure index that included cohabitation as one aspect, had increased odds of frequency of using each of the services. This could mean that exposure to a greater number of heavy drinkers was more important than the cohabitation aspect of the exposure index. Another possibility is that those harmed by another drinker who is their partner may be less likely to report this.18

There is little provision of services directly for family members of others affected by a heavy drinker in New Zealand. Some specialised treatment and harm reduction services are family-inclusive, but there is generally a lack of assessment or intervention for those affected by the drinker.29 Further, since most heavy drinkers do not receive treatment, only a small proportion of family members and significant others affected are likely reached through specialised treatment services. There are 12-step fellowships for those affected by the heavy drinking of another, eg, Al-Anon and helplines (eg, Alcohol Drug Helpline/Youthline). These were not, however, asked about in the current study.

Exposure to heavy drinkers is related to increased service use by those affected, and this contributes to the cost of running police and health services (which dominate public spending in New Zealand).8 The cost of alcohol’s harm to others to services remains largely hidden, however, as these data are not routinely collected or, if are collected, to the best of our knowledge these have not been utilised to estimate the costs to services in New Zealand. Routine measurement that documents the numbers of those that seek or receive help at services because of the effects of another’s drinking, and the service provided, would allow costings in terms of dollars spent by services, and this information would contribute to the policy debate on the extent of alcohol’s harm to others and could be taken into account in the alcohol policy making process, ie, translation into practical policy, service delivery and to inform interventions. The study has several limitations. The survey design was cross-sectional, which is a limitation in terms of drawing conclusions about causality. The measure of heavy drinkers was limited to the respondents’ self-reports. Not all factors known to be associated with service use could be controlled for. Survey data usually suffer from under-representation of the members of the community most affected due to non-response biases.30

Conclusion

There are considerable numbers of New Zealanders requiring intervention from police or health-related services due to the effects of someone else’s drinking. Heavy drinkers place increased burden on police and health-related services in New Zealand, not only because of directly attributable effects but because they impact others. Routine measurement of the numbers of those that seek or receive help at services because of the effects of another’s drinking, and of the type of service provided, would provide useful data to contribute to the policy debate in the future.

Summary

The findings of this study are the first to show the extent of service use because of others’ drinking among the general population in New Zealand. There are considerable numbers of New Zealanders requiring intervention from police or health-related services due to the effects of someone else’s drinking. Heavy drinkers place increased burden on police and health-related services, not only because of directly attributable effects but because they impact others.

Abstract

Aim

To report population estimates of service use because of someone else’s drinking in New Zealand, investigate whether greater exposure to heavy drinkers relates to greater service use and examine demographic predictors of such service use.

Method

A general population survey of respondents aged 12–80 years was conducted in New Zealand. The sample size was 3,068 and response rate 64%. Respondents’ use of police and health-related services because of someone else’s drinking were measured along with self-reports of heavy drinkers in their lives, demographic variables and own drinking.

Results

Ten percent of New Zealanders reported having called the police at least once in the past 12 months because of someone else’s drinking—corresponding to 378,843 New Zealanders making at least one call to police. Almost 7% of the sample, representing 257,613 New Zealanders, reported requiring health-related services at least once for the same reason.

Conclusion

There are considerable numbers of New Zealanders requiring intervention from police or health-related services due to the effects of someone else’s drinking. Further, increased exposure to heavy drinkers among respondents predicted increased service use. Heavy drinkers place increased burden on police and health-related services, not only because of directly attributable effects but because they impact others.

Author Information

Taisia Huckle, Senior Researcher, SHORE & Whariki Research Centre, College of Health, Massey University, Auckland; Khoon Wong, Biostatistician, Centre for Public Health Research, Massey University, Wellington; Karl Parker, Statistician, SHORE & Whariki Research Centre, College of Health, Massey University, Auckland; Professor Sally Casswell, Director, SHORE & Whariki Research Centre, College of Health, Massey University, Auckland.

Acknowledgements

Funded by the Health Research Council of New Zealand.

Correspondence

Dr Taisia Huckle, SHORE & Whariki Research Centre, PO Box 6137, Wellesley Street, Auckland 1141.

Correspondence Email

t.huckle@massey.ac.nz

Competing Interests

Nil.

References

  1. Statistics New Zealand. Hot off the Press - Alcohol Available for Consumption: Year ended December 2015. Wellington: 2015. Available at: http://www.stats.govt.nz/browse_for_stats/industry_sectors/alcohol_and_tobacco_availability/Alcohol-available-for-consumption_HOTPYeDec15.aspx (accessed 15 August 2016).
  2. Connor J, Kydd R, Shield K, et al. Alcohol-attributable burden of disease and injury in New Zealand: 2004 and 2007. Wellington: Health Promotion Agency, 2013. http://www.alcohol.org.nz/sites/default/files/research-publications/pdfs/Attributable%20fractions.pdf
  3. Humphrey G, Casswell S, Han D. Alcohol and Injury among attendees at a New Zealand emergency department. New Zealand Medical Journal 2003; 116:298–306.
  4. New Zealand Law Commission. Alcohol In Our Lives: Curbing the Harm. Wellington, 2010. (Law Commission report; no. 114).
  5. Huckle T, You RQ, Casswell S. Increases in quantities consumed in drinking occasions in New Zealand 1995–2004. Drug Alcohol Rev 2011; 30:366–71.
  6. Huckle T, You RQ, Casswell S. Socio-economic status predicts drinking patterns but not alcohol-related consequences independently. Addiction 2010; 105:1192–202.
  7. Casswell S, You RQ, Huckle T. Alcohol’s harm to others: reduced wellbeing and health status for those with heavy drinkers in their lives. Addiction 2011; 106:1087–94.
  8. Connor J, Casswell S. Alcohol-related harm to others in New Zealand: evidence of the burden and gaps in knowledge. NZ Med J 2012; 125 http://www.nzma.org.nz/journal/read-the-journal/all-issues/2010-2019/2012/vol-125-no-1360/article-connor
  9. Livingston M, Wilkinson C, Laslett A. Impact of Heavy Drinkers on Others’ Health and Well-Being. J Stud Alcohol Drug 2010; 71:778–85.
  10. Anderson P, Møller L, Galea G. eds. Alcohol in the European Union: Consumption, harm and policy approaches. Copenhagen: World Health Organization Regional Office for Europe; 2012.
  11. Moan I, Storvoll E, Sundin E, et al. Experienced Harm from Other People’s Drinking: A Comparison of Northern European Countries. Subst Abuse 2015; 9s2.
  12. Raitasalo K, Holmila M, Autti-Rämö I, et al. Hospitalisations and out-of-home placements of children of substance-abusing mothers: a register-based cohort study. Drug Alcohol Rev 2014; 34:38–45.
  13. Svenson L, Forster D, Woodhead S, et al. Individuals with a chemical-dependent family member. Does their health care use increase? Can Fam Physician 1995; 41:1488–93.
  14. Cherpitel C, Ye Y, Bond J, et al. Attribution of alcohol to violence-related injury: self and others’ drinking in the event. J Stud Alcohol Drug 2012; 73:277–84.
  15. Connor J, Casswell S, You R. Alcohol-related harm to others: a survey of physical and sexual assault in New Zealand. NZ Med J 2009; 122(1303): Available from: http://www.nzma.org.nz/__data/assets/pdf_file/0011/17786/Vol-122-No-1303-25-September-2009.pdf
  16. New Zealand Police. Alcohol - law and penalties. 2016. Available at: http://www.police.govt.nz/advice/drugs-and-alcohol/alcohol-laws-and-penalties (accessed 15 August 2016).
  17. Ministry of Health. Alcohol Use 2012/13: New Zealand Health Survey. Wellington: Ministry of Health, 2015.
  18. Mugavin J, Livingston M, Laslett A. Seeking help because of others’ drinking. Drug Alcohol Rev 2014; 33:161–8.
  19. Wyllie A, Black S, Zhang J, et al. Sample frame bias in telephone-based research in New Zealand. NZ Statistician 1994; 29:40–53.
  20. Statistics New Zealand. Household Economic Survey (Year ended 30 June 2004) - Hot Off The Press. Wellington: 2004. (25 November). Available at: http://www.stats.govt.nz/browse_for_stats/people_and_communities/Households/HouseholdEconomicSurvey_HOTPYeJun04.aspx (accessed 10 October 2009).
  21. Casswell S, Huckle T, Pledger M. Survey data need not underestimate alcohol consumption. Alcohol Clin Exp Res 2002; 26:1561–7.
  22. Statistics New Zealand. QuikStats About Culture and Identity - Pacific Peoples. 2007. Available at: http://www.statistics.govt.nz/Census/2006CensusHomePage/QuickStats/quickstats-about-a-subject/culture-and-identity/pacific-peoples.aspx (accessed 22 July 2010).
  23. Statistics New Zealand. Quikstats about education and training - 2006 census. 2007. Available at: http://www.statistics.govt.nz/Census/2006CensusHomePage/QuickStats/quickstats-about-a-subject/education-and-training.aspx (accessed 2 February 2011).
  24. Statistics New Zealand. Quikstats about incomes - 2006 census. 2007. Available at: http://www.statistics.govt.nz/Census/2006CensusHomePage/QuickStats/quickstats-about-a-subject/incomes.aspx (accessed 2 February 2011).
  25. Johansson P, Jarl J, Eriksson A, et al. The Social Costs of Alcohol in Sweden. Stockholm: SoRAD Social Research on Alcohol and Drugs, Stockholm University, 2006. (SoRAD - Forskningsrapport 36 - 2006).
  26. Gameroff M. Using the Proportional Odds Model for Health-Related Outcomes. SUGI 30 Proceedings, Philadelphia, Pennsylvania, April 10–13 2005: Available from: http://www2.sas.com/proceedings/sugi30/205-30.pdf (accessed Archived by WebCite® at http://www.webcitation.org/5unyKxafH).
  27. Stokes M, Davis C, Koch GS. Partial proportional odds model. Categorical Data Analysis Using the SAS System. 2nd ed. ed. Cary, NC: SAS Publishing, BBU Press and John Wiley Sons Inc.; 2000: 533–41.
  28. Australian Bureau of Statistics. Apparent Consumption of Alcohol, Australia, 2008–09. 2010. (27 May). Available at: http://www.abs.gov.au/AUSSTATS/abs@.nsf/allprimarymainfeatures/EEB3774B4963BE15CA2578A3001146C5?opendocument#Total (accessed 15 August 2016).
  29. Adams P. Fragmented Intimacy: Addiction in a Social World. New York: Springer; 2008.
  30. Kypri K, Samaranayaka A, Connor J, et al. Non-response bias in a webbased health behaviour survey of New Zealand university students. Prev Med 2011; 53:274–7.