Experience of Non-Fatal Heroin Overdose Among Heroin Users in Adelaide, Australia: Circumstances and Risk Perceptions

Abstract

Aims. To ascertain the prevalence and risk factors for non-fatal overdose among heroin users to assist in the development of an effective intervention.

Design. Cross-sectional design.

Setting. Community setting, principally metropolitan Adelaide.

Participants. Current heroin users (used heroin in the previous six months).

Measurements. A structured questionnaire including the Severity of Dependence Scale.

Finding. Of 218 current South Australian heroin users interviewed in 1996, 48% had experienced at least one non-fatal overdose their life-time (median: two overdoses), and 11 % had overdosed in the previous 6 months. At some time, 70% had been present at someone else's overdose (median: three overdoses). At the time of their own most recent overdose, 52% had been using central nervous system depressants in addition to heroin, principally benzodiazepines (33%) and/or alcohol (22%). The majority of overdoses occurred in a private home (81%) and in the presence of other people (88%). Unrealistic optimism regarding the risk of overdose was evident across the sample. Despite almost half the sample reporting having had an overdose, and the belief expressed by respondents that on average about 50% of regular heroin users would overdose during their life-time 73% had, during the previous 6 months, "rarely" or "never" worried about possibly overdosing. Optimism regarding the possibility of future overdose was reduced in those with recent experience of overdose in comparison to the rest of the sample. A targeted intervention aimed at the reduction of overdose among heroin users is outlined.

Introduction

A number of recent cohort studies conducted overseas have reported excess mortality ratios among heroin users compared to their non-heroin-using peers of between 10 and 22 to one (e.g. 10:1 in Rome, [1] 18:3 in Stockholm [2] and 22 in Glasgow [3]). Several European studies have shown that HIV and fatal overdose are the two major contributors to this excess mortality among heroin users. [1,4,5] In Australia, where HIV prevalence rates are relatively low, [6] fatal opioid overdose is the principal contributor to this excess mortality. The latest available data show that deaths due to opiate use among people aged 15-44 years increased from 79 in 1979 to 550 in 1995. The average age at death increased steadily from 24 to 30 years during the same period. These data represent a six-fold increase in the rate of opioid overdose mortality in Australia. [7]

While heroin overdose is often viewed as being a result of the quantity or strength of the heroin dose being higher than the individual's level of tolerance at the time of use, it remains difficult to test this assumption. Currently, samples of heroin for testing are obtained from both undercover purchases (street buys) and from those arrested for heroin possession (heroin seizures). However, there is no systematic collection and analysis of heroin across states, and therefore no readily accessible national or state database which would facilitate monitoring of heroin purity.

A recent study of all heroin-related deaths in NSW in 1992 provided the first detailed information on heroin-related deaths in an Australian state. [8] The majority of cases were males with a mean age at death of 30 years. Although over two-thirds (69%) of these heroin-related deaths had occurred in a home environment, an ambulance had been called in only 10% of cases while the subject was still alive. This suggested that potentially life-saving help was either delayed or not sought in most cases of overdose. Polydrug use was strongly implicated in these deaths: in only 29% of cases was a single drug (morphine) identified on toxicological analysis. Importantly, the presence of alcohol at autopsy was associated independently with lower blood morphine levels. [8] This is consistent with other work showing that the presence of ethanol markedly increases the likelihood of fatality when used in conjunction with other centrally acting drugs [9] including heroin.[10]

Polydrug use among heroin users is of particular concern, as central nervous system depressants, such as alcohol or benzodiazepines, may have a physiologically additive effect with heroin. For example, alcohol and heroin both produce respiratory depression when taken alone, and a combination of the two may increase the likelihood of this occurring. Moreover, the administration of other opiates with heroin may combine synergistically, further increasing the risk of respiratory depression.

Heroin overdose is not invariably fatal, and information on the extent of non-fatal overdose is accumulating. As part of the multi-centre Australian Study of HIV and Injecting Drug Use (ASHIDU), Loxley and colleagues reported that 53% of illicit drug users (including heroin) interviewed in 1994 had experienced a non-fatal overdose. Of this group 81% had, at some time, overdosed when using heroin. The same study identified regional differences in the experience of overdose among illicit drug users. Forty-four per cent of illicit drug users interviewed in Melbourne had overdosed, 47% in Adelaide, 58% in Perth and 65% in Sydney. Greater quantity or purity of the drug used was seen as the main reason for their most recent overdose by 44% of those who had ever overdosed. Of those who had ever overdosed, 82% had been using more than one drug on the last occasion. [6] Information regarding the prevalence of overdose among heroin users in the Australian Capital Territory came from a study by Bammer & Sengoz, who found that one-third reported overdosing on a median of two occasions in their life-time. [11]

When a group of injecting drug users in Sydney, Australia were surveyed recently, 68% were found to have overdosed during their heroin-using career, principally in conjunction with the use of other central nervous system depressants (e.g. alcohol, benzodiazepines and other opioids). [12] Only a small minority personally avoided the use of alcohol and other drugs when they used heroin. Given recent evidence regarding the possible effects of the concurrent consumption of multiple centrally acting drugs, these usage patterns and beliefs among heroin users have significant implications for the design of interventions aimed at reducing morbidity and mortality among this group. A similar study conducted in the United Kingdom found that 23% of a sample of heroin users had overdosed at least once in their life-time. While overdose in the UK sample was unrelated to gender, or to reported frequency or quantity of heroin use, those who had overdosed were found to be older and more dependent on heroin. Of those who reported having overdosed, 98% were injectors while only 2% smoked heroin. [13]

Heroin users are also frequent witnesses to overdose in other people, with the overwhelming majority (86%) of a Sydney sample indicating their presence at an overdose at some time. Witnesses in this study commonly attempted to revive the affected person themselves rather than call an ambulance, predominantly due to fear of police involvement. [14]

There is clear evidence that death from heroin-related overdose is rarely instantaneous. Moreover, the majority of overdoses, both fatal and non-fatal, occur in home surroundings and in the presence of other people. [8, 12] The opportunity for those present to correctly identify an overdose and to instigate timely, life-saving intervention therefore exists. Accurate information identifying the risk behaviours associated with such overdose events could inform the development of appropriate interventions aimed at reducing the morbidity and mortality associated with heroin use in the same way that harm reduction approaches have been successfully implemented for other health-risk behaviours.[15]

Assessment of risk judgements for a range of health-related risks has identified an "optimistic bias", or tendency to claim a lower personal risk of susceptibility to harm in comparison to one's peers. Weinstein has found that such optimism appears unrelated to age, sex, education or occupation. [16] Weinstein also identified a tendency for individuals to extrapolate from their past experience to estimate future vulnerability. That is, unrealistic optimism was associated with the belief that if the event or hazard had not yet occurred it was unlikely to do so in the future, although this optimism decreased with personal experience of a particular hazard. [17] It seems reasonable to assume therefore that heroin users who had overdosed would show less optimism regarding future overdoses in comparison to those who had not.

Concern regarding the incidence of heroin-related overdose prompted the development of an intervention to reduce the frequency of overdose among heroin users in South Australia. Regional differences in non-fatal overdose rates among illicit drug users have been identified, [6] and it is essential that any proposed intervention is based on accurate and current information regarding overdose risk factors in different localities. The major objectives of this study were therefore to investigate the circumstances, risk perceptions, risk behaviours and experience of non-fatal overdose among current South Australian heroin users (used heroin in the 6 months prior to interview). These data also provided a baseline against which the effect of a trial intervention could be measured.

Method

Subjects

The sample consisted of 218 current heroin users, (see Table 1). More than a quarter (29%) came from the eastern metropolitan region of Adelaide, which includes the city centre and some hill areas. Twenty-two per cent came from the southern metropolitan region, 21% came from the western metropolitan region and 16% came from the northern metropolitan region. A further 12% came from non-metropolitan areas (Australia Bureau of Statistics sub-divisions).

The mean age was 30 years (SD 6.9, range 14-45), with 55% of the sample being male. The mean length of formal school education was 10.8 years (SD 1.1, range 7-12). In terms of major demographic variables and mode of heroin administration, this sample was comparable to a sample of heroin users in Sydney. [12] However, the level of post-secondary education and the percentage currently in employment were both somewhat higher in the present sample.

Procedure

Subjects were recruited by means of a "snowballing" method, whereby individuals with knowledge of the community or contacts within the local networks of heroin users undertook "privileged access interviews" with heroin users known to them. [18-20] These methods have facilitated data collection and substantially increased knowledge of drug use patterns and group demographics, particularly among non-clinical samples or those drug users who do not access treatment services. [18] Interviewers for the present study were recruited via three major sources. First, from the Needle Exchange Program administered by the Drug and Alcohol Services Council (DASC); secondly, the Aids Council of South Australia (ACSA) and thirdly, the South Australian Voice for IV Education (SAVIVE), a group which has been active in the South Australian heroin user community for several years. With the cooperation of these organizations, individuals who were identified as having developed a variety of networks within the heroin user community, and who possessed the skills and competencies essential to accurate data gathering, were recruited as interviewers for the study.

In order to provide the widest possible sample, a total of 11 interviewers were recruited and trained to conduct the subject interviews. Starting with their own established networks, these interviewers accessed heroin users from across socio-economic groupings and from different geographic regions within metropolitan Adelaide, the hills region and to a lesser extent, non-metropolitan centres within the state. Each interviewer conducted around five interviews per week until a total of 20 interviews each had been completed and an adequate coverage had been achieved. All subjects were volunteers and were compensated $20 for their time and any inconvenience caused in attending the interview and completing the questionnaire. Interviews were conducted during September/October, 1996. To be eligible for the study, subjects had to have used heroin in the previous 6 months. All interviews were conducted in a place of mutual safety and privacy negotiated between the interviewer and the subject. Each interview took between 30 and 45 minutes, and subjects were assured than any information they provided would be kept strictly confidential and anonymous.

Measures

A modified version of the questionnaire employed in the Sydney study [12] was developed for use in this study. The questionnaire elicited information on demographic characteristics, drug use history, level of heroin dependence, recognition of overdose signs, experience of heroin overdose in others, personal experience of overdose, use of preventative measures and risk perceptions and behaviours associated with heroin-related overdose. For clarity, overdose was distinguished from being "on the nod" and defined as: collapse; blue skin colour; difficulty breathing; loss of consciousness; inability to be woken or roused; or death associated with heroin use. Level of heroin dependence was assessed by the Severity of Dependence Scale (SDS), measured on an 0-15 point scale in which higher scores reflect greater dependence. [21]

Analyses

For continuous variables t-tests were employed. Group differences were analysed using one-way ANOVA with post-hoc analyses (Bonferroni-adjusted) and x2 tests. Categorical variables were analysed by X2 tests and the Mann-Whitney U-test. Where distributions were highly skewed, medians were reported and highly skewed continuous data analysed using the Mann-Whitney U-test. To determine which variables were independently associated with ever having overdosed, logistic regression analyses were conducted. The alpha level was set at 0.05 and confidence intervals of 95% were used. Regression analyses were conducted using SYSTAT. [22] SPSS for Windows was used for all other analyses. [23]

Results

The mean age of first heroin use was 19.9 years (SD 3.8, range 12-35 years). Mean length of heroin-using career was 9.6 years (SD 7, range < 1-29 years). The Severity of Dependence Scale showed a mean score for heroin dependence of 6.4 (SD 4.1, range 0-15), which was somewhat lower than the mean SDS score (i.e. 7.4) reported among the Sydney sample. [12] SDS scores greater than 6, generally considered indicative of severe dependence, were found for 48% of the South Australian sample. [13] SDS scores indicated higher levels of dependency on heroin for women (mean: 7.2 for women vs. 5.9 for men, t = 2.33, df = 215, p < 0.05).

Polydrug use among the sample was common (see Table 2). The median number of drug types used in the previous 6 months was five (range 1-13), and the median number of drug types injected in the previous 6 months was two (range 1-5). Apart from central nervous system depressants, 39% had used amphetamines in the 6 months prior to interview.

Methadone maintenance treatment

Subjects in methadone maintenance treatment (MMT) comprised 25% of the sample. These subjects were more dependent on heroin (as indicated by SDS score) in comparison to the remainder of the sample (median SDS score: 7 for MMT subjects vs. 6 for others, U = 3502, p < 0.05). Despite this greater dependency, MMT subjects had used fewer mean drug types in the previous 6 months, (MMT subjects had used a median of 4.5 vs. 5 drug types for NMM subjects, U = 3106 p < 0.001), and had fewer days of heroin use (although this difference narrowly missed out on statistical significance- median: 40 days for MMT subjects vs. 57 days for others, U = 3650, p < 0.052).

Experience of overdose

Almost half the total sample (48%), had overdosed on heroin in their life-time. The median life-time number of overdoses was two, and 11% had overdosed in the previous 6 months. There were no gender differences in the numbers of reported overdoses (see Table 3).

Of those who had ever experienced an overdose, 54% had been administered naloxone, 45% had been to a hospital because of a heroin-related overdose, while an ambulance had attended an overdose for 59%.

While methadone maintenance treatment (MMT) subjects reported similar rates of lifetime and recent overdoses, the length of time since their most recent overdose was greater (median: 2 years for MMT subjects vs. 1 year, U= 802, p < 0.01) in comparison to the remainder of the sample.

The prevalence of life-time experience of overdose increased with the length of heroin-using career. More than half (59%) those with more than 10 years of heroin use, 47% of those who had been using for 6-10 years, and 36% of those who had been using heroin for up to 5 years had overdosed (X2 = 9.22, df = 2, p < 0.01).

Circumstances of most recent overdose

At the time of their most recent overdose, only 16% of those who had ever overdosed were in methadone treatment, one subject had just left a therapeutic community and 12% had been discharged from prison within the previous 2 weeks. The majority of those who had ever overdosed (81%) had done so in a private home on the most recent occasion. Only 12% of those who had ever overdosed reported being alone at the time of their most recent overdose.

Risk factors for overdose

Overdose risk factors were examined for three exclusive sample groups: those who had overdosed in the previous 6 months; those who had experienced an overdose, but not in the previous 6 months; and those who had never overdosed. One-way ANOVA with post-hoc analyses (Bonferroni-adjusted) showed that those whose overdose was more than 6 months ago were older (mean age 32 years vs. 28 years for those who had overdosed in the previous 6 months, and those who had never overdosed, F = 8.41, df=2,215, p<0.001). As indicated by SDS score, those who had never overdosed had lower levels of dependence on heroin (mean 5.5) in comparison with those whose overdose was more than 6 months ago (mean 7.5), and those who had overdosed in the previous 6 months, (mean 7.7, F= 7.36, df = 2,214, p< 0.001). No differences were identified between these groups in terms of the frequency of trial tasting a new batch of heroin, using heroin alone, or using with the door locked in the past 6 months.

Of those who had overdosed in the previous 6 months, 32% reported drinking alcohol "every time" or "often" in conjunction with heroin in the past 6 months. In comparison, 12% of those whose overdose was more than 6 months ago, and only 5% of those who had never overdosed reported drinking alcohol "every time" or "often" with heroin in the past 6 months (x2 = 15.17, df 2, p < 0.001). While similar numbers in both overdose groups (16% of those who had overdosed in the previous 6 months, and 15% of those whose overdose was more than 6 months ago) reported consuming benzodiazepines "every time" or "often" in conjunction with heroin in the past 6 months, only 4% of those who had never overdosed reported doing so (x2 = 7.33, df = 2, p < 0.05).

Of those subjects who had ever experienced an overdose, almost two-thirds (62%) had been using other drugs, principally other central nervous system depressants (52% used alcohol, benzodiazepines, methadone and/or other opiates, with some overlap in usage) in addition to heroin at the time of their most recent overdose. Benzodiazepines had been used by one-third (33%) and alcohol by over one-fifth (22%). Use of opiates other than heroin was relatively low among this sample (e.g. methadone-8%; other opiates 1%).

Presence at heroin-related overdoses

A majority of the sample (70%) had been present at someone else's overdose on a median of three (range 1-50) occasions. Of those who had ever been present at another's overdose, 41% had been present at an overdose in the past 6 months on at least one occasion. Recognition of acute overdose signs was high. The three most commonly nominated signs of overdose- cyanosis (66%), depressed level of consciousness (60%) and depressed respirations (58%)- are, together with pupil constriction, characteristic of acute narcosis. Less well recognized was the gradual descent into central nervous system depression characterized by snoring or gurgling breathing in someone who is asleep (15%).

On the most recent occasion that an overdose was witnessed, the most common initial response by those who had been present at an overdose was to check the level of consciousness (35%; a further 13% as a subsequent action). The next most common initial response was to check the breathing and/or pulse (16%; 39% as a subsequent response). Of those who had been present at an overdose only 9% rang an ambulance as the initial response, although a further 36% rang an ambulance as a subsequent action. More than a quarter placed the affected person in the coma position (9% as an initial response and 18% as a subsequent response) while almost two-fifths applied a resuscitation procedure (either mouth-to-mouth or cardio-pulmonary resuscitation) as an initial (8%) or as a subsequent action (31%).

Of those who had ever been present at an overdose, 40% had been delayed or stopped from getting help at the most recent witnessed overdose. A fear of police involvement was indicated by the overwhelming majority of this group (80%). The next most common reason for delay, being stopped by another person, was nominated by only 5%, while no subject nominated the cost of an ambulance or intoxication as a initial delaying factor in getting help. Concerns regarding outstanding warrants, (31%) and fear of manslaughter charges, (33%) predominated among the secondary reasons nominated for a delay in help-seeking at overdose. There were no gender differences among those delayed from getting help at the time of their most recent witnessed overdose. However, among those who were
delayed, more men cited a fear of police involve-ment (93% of men vs. 69% of women, X2 = 5.7, df = 1, p < 0.05).

Overdose risk perceptions

There was a striking contrast between personal perceptions of overdose risk and of other's chances of overdose. While 62% of the total sample thought that the chances of a regular heroin user in Adelaide overdosing in the future was "likely" or "very likely", only 20% thought their own chances of overdose were similar.

Although respondents estimated that, on average, 50% of regular heroin users would overdose during their heroin-using career, the majority of respondents were unconcerned about the possibility of a personal overdose in the future. In the previous 6 months only 7% of subjects had worried "very often" or "often", 20% had worried "sometimes" while 73% had "rarely" or "never" worried about possibly overdosing.

Those who had never overdosed underestimated the number of times, on average, regular heroin users in Adelaide would overdose in their life-time (median, two times) compared to those with recent experience of overdose (median, 2.8 times) and those whose overdose was more than 6 months ago (median, three times, X2 = 17.20, df =2, p < 0.001).

Compared to the other two groups, the recent overdose group tended to inflate the likelihood of a future overdose for an average, regular heroin user in Adelaide (92% likely or very likely to overdose vs. 58% of those whose overdose was more than 6 months ago, and those who had never overdosed, x2 = 10.96, df=2, p<0.01). Similarly, in terms of the likelihood of a future personal overdose, those with recent experience were less optimistic in regard to the likelihood of a future personal overdose in comparison to the other two groups (56% likely or very likely to overdose in the future vs. 16% for those whose overdose was more than 6 months ago, and 14% for those who had never overdosed X 2 = 23.43, df = 2, p < 0.001). There were no differences between the groups in terms of "worry" about the possibility of a future overdose.

When the perceived cause of their most recent personal overdose was investigated, 49% of those who had ever overdosed in their life-time implicated the quantity or strength of the heroin used at the time. Only 15% nominated the concomitant use of benzodiazepines and 8% the use of alcohol as causative in their most recent overdose. Similarly, 44% of the total sample felt that more or stronger heroin was causative in other's overdoses while only 9% nominated the concomitant use of benzodiazepines and 5% the use of alcohol as causative in other's overdose experiences.

Prevention strategies

Measures to avoid overdosing when they used heroin were taken by 97% of the total sample. Not using more than they knew they could tolerate (43%), the use of a test dose of a new batch of heroin (38%), avoidance of alcohol when using heroin (33%) and avoiding the use of benzodiazepines (25%) were the most common strategies used. Although 38% of the total sample had nominated having a trial taste of a new batch of heroin as a preventive measure in heroin overdose, only 17% of the total sample had done so "every time" or "often" in the previous 6 months. Over half (53%) "rarely" or "never" used heroin alone in the previous 6 months while 22% had used heroin with the door locked "every time" or "often" in the previous 6 months. Over half (59%) the total sample had drank alcohol while 40% took benzodiazepines with heroin in the previous 6 months.

Predictors of overdose

To identify those variables which were independently associated with ever having overdosed, logistic regressions were performed. Age, gender, length of heroin-using career, the total number of different drug types ever used, frequency of heroin use, frequency of alcohol use, frequency of benzodiazepine use and SDS score were used in a series of logistic regressions. Frequency of alcohol use (coded as weeks used in the previous 6 months), length of heroin-using career (coded in years), SDS scores and the total number of different drug types ever used were significantly related to having overdosed (see Table 4).

The regression equation was significant (X2 = 41.84, df = 4, p < 0.001), and had a good fit (Hosmer-Lemeshow X2 = 9.03, p < 0.34, note that higher p values indicate better goodness-of-fit). These results indicate that after controlling for the effects of other variables in the model, each additional drug type ever used increased the odds of ever having overdosed by 26%. Each additional point on the SDS (indicating higher levels of heroin dependence) increased the odds of ever having overdosed by 12%, and each extra week of alcohol consumption in the previous 6 months increased the odds of having ever overdosed by 5%.

Discussion

Results for this South Australian sample were largely consistent with those from the Sydney studies, [12,14] although the rate of life-time nonfatal overdose in the South Australian sample was lower by 20%. The 48% overdose rate found in the present sample was comparable to that found in a South Australian sample of illicit drug users interviewed as part of the ASHIDU study in 1994. [24] Almost half the present sample of current heroin users had experienced an overdose during their life-time, with no gender difference in overdose rates. Consistent with earlier studies, the vast majority of the most recent overdose episodes occurred in a private home and in the presence of other people, thus the opportunity for intervention was potentially present. Additionally, the majority of subjects who had overdosed were not in treatment at the time of their most recent overdose, supporting earlier work suggesting a protective effect for methadone maintenance treatment. [8]

As in the Sydney sample, the proportion of respondents who had experience an overdose increased with age and length of heroin-using career. [12] The overall mean SDS score for the South Australian sample was about one point lower than the mean for the Sydney sample. Given that each additional point on the SDS (indicating higher levels of heroin dependence) increased the odds of ever having overdosed by 12%, this factor may go some way towards explaining the lower overdose rate in the South Australian sample. This difference may in turn be related to differences in recruitment methods and characteristics between the South Australian and Sydney groups.

Risk factors for overdose

In common with other samples of Australian heroin users, administration of heroin was almost exclusively by intravenous injection, a known risk factor for heroin overdose. [21] The substantial minority who had used heroin while alone and with the door locked was of major concern. Using heroin under conditions which renders potential assistance delayed or absent greatly increases the risk of morbidity or mortality in heroin users. Moreover, substantial numbers also reported the use of alcohol and/or benzodiazepines in conjunction with heroin in the previous 6 months. Consistent with the Sydney study, over half the sample had been using other central nervous system depressants, principally alcohol and benzodiazepines in addition to heroin at the time of their most recent overdose. The finding that almost two-thirds of those who had ever overdosed had been using other drugs on the most recent occasion confirms that heroin overdose, rather than being a unitary phenomenon solely related to the amount or strength of the heroin used, commonly occurs in conjunction with other drugs, and particularly other central nervous system depressants.

Consistent with previous work, [12] a substantial minority had last overdosed following a period of abstinence from heroin. In one case the subject had just left a therapeutic community, while the rest of this group had been discharged from prison within the 2 weeks prior to their most recent overdose. Pre-release education of prisoners regarding the effects of abstinence on tolerance, and the dangers of polydrug use, may prevent some of these overdoses occurring. Given the efficacy of methadone treatment in reducing overdose among heroin users, the initiation of pre-release methadone programmes may reduce the risk of overdose in this vulnerable group.

Those with a history of recent overdose and those whose overdose was more than 6 months prior to interview had higher levels of dependence on heroin and had consumed benzodiazepines in conjunction with heroin on more occasions in the previous 6 months in comparison to the rest of the sample. Importantly, those with recent experience of overdose had consumed alcohol in conjunction with heroin on more occasions in the previous 6 months. Similarly, greater dependence on heroin, increased polydrug use, particularly greater frequency of alcohol use were identified as independent predictors of past overdose among this sample.

Risk perceptions

The "optimistic bias" identified by Weinstein was evident across the present sample. [16] There was a striking contrast between personal perceptions of overdose risk and the perception of other heroin user's chances of overdose. While almost two-thirds of subjects thought that the chances of a regular heroin user in Adelaide overdosing in the future was likely, only one-fifth thought their own chances of overdose were comparable. Similarly, despite almost half the sample reporting an overdose, almost three-quarter of respondents had rarely or never worried about the possibility of overdosing during the previous 6 months. This apparent lack of concern is contrasted with respondents' own estimations that 50% of regular heroin users would overdose during their heroin-using careers. It would seem that despite an apparent awareness of the strong possibility of overdose, respondents did not personalize the risk of experiencing an overdose.

Consistent with Weinstein's findings, optimism decreased with personal experience of overdose. In comparison to those who had never overdosed, those with recent and life-time experience of overdose tended to inflate the estimated number of times that an average, regular heroin user in Adelaide would experience an overdose in their life-time. Similarly, those with recent experience of overdose inflated the likelihood of a future overdose for themselves and for other heroin users in comparison to the other two groups. It may be that the salience and immediacy of having recently experienced an overdose raises the level of concern regarding future overdose in others but that this apparent concern diminishes over time.

Presence at other's overdose

While respondents were familiar with the major signs of acute narcosis, fewer subjects recognized the less dramatic, slow descent into unconsciousness characterized by snoring or gurgling breathing in someone who is asleep. This finding suggests that a narcosis which develops slowly is less likely to be recognized by others present and may therefore be more likely to proceed to a fatal outcome. Witnesses to an overdose may be unaware of the significance of this sign, especially if the individual had exhibited similar signs on previous occasions without subsequent ill effects. Moreover, witnesses to overdose events may be unaware of the total number of different drugs taken on any particular occasion or of their synergistic and/or additive effects. It is crucial that users of heroin and their associates are familiar with the dangers of combining other central nervous system depressants (especially alcohol) with heroin, as well as the need for immediate medical attention in cases of suspected opiate overdose.

While respondents were familiar with overdose in others, in only 9% of cases was an ambulance called as an initial response, an even lower figure than that found in the Sydney sample. In order to reduce the morbidity and mortality associated with heroin-related overdose, ways of increasing the rate of ambulance calls to overdoses should be investigated. Despite a reluctance to call ambulances, witnesses to overdose showed a high level of awareness of the acute signs of heroin overdose as well as the appropriate first-aid techniques. This awareness of the role of vital signs in diagnosing the level of consciousness and use of life-saving techniques should be augmented, and the continued education of heroin users in the use of life-saving techniques such as expired air resuscitation encouraged.

The finding that two-fifths of subjects who had been present at an overdose had been delayed or stopped from getting help at that time was of major concern. For the overwhelming majority of these subjects, and especially among males, fear of police involvement was the most common reason for delay. Other reasons, such as the negative attitudes of medical staff or being stopped by another person, were nominated by very few respondents. Not one subject nominated the cost of an ambulance, or concern at the reaction of the affected person, or their own intoxication as a initial delaying factor in getting help. Negotiations to identify the circumstances in which police presence is essential at overdose events are currently under way in several states, including South Australia.

The protective effect of methadone maintenance treatment was suggested by the finding that, although those in methadone treatment were more dependent, they had used fewer mean drug types in the previous 6 months. Although this group had experienced as many overdoses as other heroin users, there had been a greater length of time since the most recent overdose.

Conclusions and recommendations

The results of the present study point to the need for an educational campaign, targeted specifically at heroin users and based upon identified risk factors, particularly the concomitant use of other centrally acting drugs. Heroin users should be informed of the dangers of "drug cocktails" or the mixing of other drugs with heroin, especially other central nervous system depressants such as alcohol, benzodiazepines and other opiates. Moreover, information regarding the dangers of using heroin under conditions where help would be delayed or unavailable, such as using heroin while alone, or segregated from others behind a locked door, should be an important part of any intervention to reduce the morbidity and mortality associated with heroin use.

Of major concern is the present reluctance of heroin users to call ambulances. The evidence suggests that there is time to intervene, and that most fatal and non-fatal overdoses occur in a home environment where others are present. [8,12] Additionally, interventions aimed at addressing the fears which prevent the calling of ambulances immediately the signs of narcosis are recognized should be addressed. To this end, liaison with ambulance services and police to support the safe calling of ambulances should also form an important component of any planned intervention. Importantly, heroin users should be taught to recognize the signs and symptoms of overdose, not only the dramatic "drop" immediately following heroin use, but the slow and gradual depression of the central nervous system whereby deterioration may occur over several hours. Snoring or gurgling breathing in someone who has recently used heroin should act as a warning of impending unconsciousness and possible death. A process of focus group testing involving members of the target community would be a useful first step in the development of an appropriate intervention to reduce the harm associated with overdose among heroin users. This process of community consultation would identify both the best vehicles to carry the campaign messages, and the messages which would be most productive in encouraging changes in the direction of safer drug use behaviours.

Rapid access to the opiate antagonist naloxone may prevent an overdose event becoming a fatality. Naloxone has a long established use in heroin overdose. Because of its safety and ability to rapidly reverse the effects of opiates, naloxone is widely used for suspected opiate toxicity and coma of undetermined aetiology. Given the proven efficacy and safety of this drug, the supply of naloxone to heroin users should be considered, particularly to those heroin users at higher risk of overdose, eg following discharge from prison. The potential for abuse of naloxone is negligible. Naloxone has no reinforcing properties and is strongly antagonistic to opiates. It rapidly produces a markedly unpleasant withdrawal syndrome in heroin users and is therefore unlikely to be abused. [25]

In conclusion, overdose among heroin users is largely preventable. Reduction of the current incidence of fatal and non-fatal overdose could best be achieved by means of a targeted campaign aimed at educating heroin users regarding appropriate responses to overdose in others as well as ways of avoiding personal overdose. Particular emphasis should be placed on avoiding the concomitant use of other centrally acting drugs particularly alcohol. The identification of newly released prisoners as a group at risk of overdose points to the need for increased education and an expansion of treatment options, including pre-release methadone for this group. In addition to the modification of environmental influences on risk behaviour, consideration should be given to an extension of the present range of substitution treatment options seen to be preventative in heroin-related overdose. Finally, further research on the relationship between the purity of street-level heroin and overdose morbidity and mortality may identify causal relationships between these factors.

Acknowledgements

Ibis research was funded by a Commonwealth Department of Health and Family Services Research into Drug Abuse Grant (RIDAG) and the Drug and Alcohol Services Council of South Australia (DASC). The authors would like to thank the South Australian Voice for Intravenous Education (SAVIVE) for their support of this project and all members of the interviewer team for carrying out subject recruitment and data collection. The views expressed in this paper do not necessarily represent the views of the Drug and Alcohol Services Council of South Australia.

References

1. PERUCCI, C.; A., DAVOLI, M.; RAPITII, E.; ABENI, D. D. & FORSTIERI, F. (1991) "Mortality of intravenous drug users in Rome: a cohort study," American Journal of Public Health, 81, 1307-1310.

2. ENGSTROM, A.; ADAMSSON, C. M.; ALLEBECK, P. & RYDBERG, W. (1991) "Mortality in patients with substance abuse: a follow-up in Stockholm County, 1973-1984," International Journal of the Addictions, 26, 91-106.

3. FRISCHER, M.; GOLDBERG, D.; RAHMAN, M. & .BERNEY, L. (1997) "Mortality and survival among
a cohort of drug injectors in Glasgow, 1982- 1994," Addiction, 92, 419-427.

4. PUSCHEL, K. (1993) "Drug-related death-an update," Forensic Science International, 62, 121-128.

5. RISSER, D. & SCHNEIDER, B. (1994) "Drug-related deaths between 1985 and 1992 examined at the Institute of Forensic Medicine in Vienna, Austria," Addiction, 89, 851-857.

6. LOXLEY, W.; CARRUTHERS, S. & BEVAN, J. (1995) In The Same Vein, First Report of the Australian Study of HIV and Injecting Drug Use (ASHIDU) (Perth, National Centre for Research Into the Prevention of Drug Abuse, Curtin University of Technology, Perth, Western Australia).

7. HALL, W. & DARKE, S. D. (1997) Trends in Opiate Overdose Deaths in Australia 1979-1995 (Sydney, National Drug and Alcohol Research Centre).

8. ZADOR, D.; SUNJIC, S. & DARKE, S. (1996) "Heroin-related deaths in New South Wales, 1992: toxicological findings and circumstances," Medical Journal of Australia, 164, 204-207.

9. KING, L. (1982) "Effect of ethanol on drug levels in blood in fatal cases," Medicine, Science and the Law, 22, 232-234.

10. LEVINE, B.; GREEN, B. & SMIALEK, J. (1995) "The role of ethanol in heroin deaths," Journal of Forensic Sciences, 40, 808-810.

11. BAMMER, G. & SENGOZ, A. (1994) "Non-fatal heroin overdoses," Medical Journal of Australia, 161, 572-573.

12. DARKE, S.; ROSS, J. & HALL, W. (1996) "Overdose among heroin users in Sydney, Australia: 1. Prevalence and correlates of non-fatal overdose," Addiction, 9, 405-411.

13. GOSSOP, M.; GRIFFITHS, P.; POWIS, B.; WILLIAMSON, S. & STRANG, J. (1996) "Frequency of non-fatal heroin overdose: survey of heroin users recruited in non-clinical settings," British Medical Journal, 313, 402.

14. DARKE, S.; ROSS, J. & HALL, W. (1996) "Overdose among heroin users in Sydney, Australia: II. Responses to overdose," Addiction, 9, 413-417.

15. WARD, J.; MATTICK, R. & HALL, W. (1992) Key Issues in Methadone Maintenance Treatment (Sydney, University of New South Wales Press).

16. WEINSTEIN, N. D. (1987) "Unrealistic optimism about susceptibility to health problems: Conclusions from a community-wide sample," Journal of Behavioural Medicine, 10, 481-500.

17. WEINSTEIN, N. D. (1989) "Effects of personal experience on self-protective behaviour," Psychological Bulletin, 105, 31-50.

18. GRIFFITHS, P.; GOSSOP, M.; POWIS, B. & STRANG, J. (1993) "Reaching hidden populations of drug users by privileged access interviewers: Methodological and practical issues," Addiction, 90, 607-614.

19. BIERNACKI, P. (1986) Pathways from Heroin Addiction: recovery without treatment (Philadelphia, Temple University Press).

20. POWER, R.; HUNTER, G. & WARD, J. (1996) "Guidelines for peer intervention," Druglink, 18-19.

21. GOSSOP, M.; DARKE, S.; GRIFFITHS, P.; POWIS, B.; HALL, W. & STRANG, J. (1995) "The Severity of Dependence Scale (SDS) in English and Australian samples of heroin, cocaine and amphetamine users," Addiction, 90, 607-614.

22. WILKINSON, L. (1990) SYSTAT: The system for statistics (Evanston II, SYSTAT Inc).

23. NORUSIS, M. J. (1993) SPSS for Windows, release 6.0. (Chicago, II, SPSS Inc).

24. LOXLEY, W.; McDONALD, C. & MARSH, A. (1992) The Australian National AIDS and Injecting Drug Use Study: Perth 1990 (Perth, National Centre for Research into the Prevention of Drug Abuse, Curtin University of Technology).

25. STRANG, J.; DARKE, S.; HALL, W.; FARRELL, M. & ALI, R. (1996) "Heroin overdose: the case for take-home naloxone," British Medical Journal, 312, 1435.