Hidden need for drug treatment services: measuring levels of problematic drug use in the North West of England

Transcription

1 Journal of Public Health Medicine Vol. 23, No. 4, pp Printed in Great Britain Hidden need for drug treatment services: measuring levels of problematic drug use in the North West of England Caryl Beynon, Mark A. Bellis, Tim Millar, Petra Meier, Rod Thomson and Kevin Mackway Jones Abstract Background In the North West of England, data on drug users are routinely collected from a variety of agencies including specialist treatment centres, police and probation services. However, the covert nature of drug use means that alone, these conventional monitoring systems cannot provide the epidemiology required to target and develop drug treatment and prevention initiatives. Methods Utilizing surveillance data and capture recapture techniques we estimate the rates of problematic drug users by age and sex in five North West health authorities and one local authority. Results Analyses show concentrations of problematic drug use in large metropolitan areas (Liverpool and Manchester) with levels as high as 34.5 and 36.5 per 1000 population (ages 15 44), respectively, and, for males, levels exceed 50 per 1000 in three authorities. Patterns of prevalence for those aged 25 and over differed from those in the younger age groups, with disproportionate levels of young users outside metropolitan areas. The proportion of young users already in treatment (21.3 per cent) was lower (older users, 35.3 per cent), with overall proportions in treatment varying between health authorities (range per cent). Conclusion With a multi-agency approach, established monitoring systems can be used to measure hidden populations of drug users. Estimates of the current populations of such users in the North West of England suggest that planned increases of people in treatment by 100 per cent would fail to accommodate even current level of problematic users. A holistic approach to new initiatives must ensure that the high level of relapse once drug users are discharged are reduced and that the needs of young users are addressed before prolonged treatment is required. Keywords; capture recapture, drug use, arrest referral, drug treatment Introduction The covert nature of illicit drug use creates particular problems for the planning and assessment of both drug treatment services and prevention initiatives. For instance, objectives of the British anti-drugs strategy 1 include diverting more drug users away from judicial systems into treatment-based alternatives and reducing the proportion of people under 25 reporting use of illegal drugs. 1 However, successfully meeting either objective requires epidemiological information on drug use beyond that currently available through conventional monitoring systems. 2 Consequently, initiatives to move individuals from judicial to treatment pathways [e.g. Arrest Referral (AR) 3 and Drug Treatment and Testing Orders (DTTO) 4 ] have begun but without suitable information on the changes in treatment capacity necessary to meet these new challenges. Equally, monitoring the actual proportion of young people (under 25) with drug problems requires measuring not only the number of young people in contact with services but also the unknown number using drugs throughout the community that potentially require treatment. Capture recapture (CR) analysis allows an estimation of the number of individuals within a given population in situations where it is often impossible to conduct a straightforward count. This approach was originally applied to animal studies 5 but more recently it has been used to estimate the size of human populations with conditions such as diabetes, 6 whooping cough, 7 tuberculosis 8 and drug misuse For human populations, the North West Public Health Observatory, Public Health Sector, School of Health and Human Sciences, Liverpool John Moores University, 70 Great Crosshall Street, Liverpool L3 2AB. Caryl Beynon, Research Associate Mark A. Bellis, Director, North West Public Health Observatory The Drug Misuse Research Unit, Kenyon House, School of Epidemiology and Health Sciences, University of Manchester, Bury New Road, Manchester M25 3BL. Tim Millar, Research Fellow Petra Meier, Database Manager Sefton Health Authority, Burlington House, Crosby Road North, Waterloo L22 0QB. Rod Thomson, Public Health Projects Co-ordinator Emergency Department, Manchester Royal Infirmary, Oxford Road, Manchester M13 9WL. Kevin Mackway Jones, Consultant in Emergency Medicine Address correspondence to Professor Mark A. Bellis. m.a.bellis@livjm.ac.uk Faculty of Public Health Medicine 2001

2 HIDDEN NEED FOR DRUG TREATMENT SERVICES 287 technique utilizes multiple databases all potentially capable of capturing (i.e. recording) the same individuals with the particular condition in question. The pattern of overlap between databases can, using log linear modelling, 12 be used to estimate the additional number of individuals with a particular characteristic but who do not appear on any of the databases. Capture recapture is well suited to estimating drug-using populations as factors such as illegality and stigmatization of drugs often demand covert use. However, previous studies have often relied on isolated data surveys and therefore consistent and repeated estimates have not been possible. Here we use a variety of enhanced drug monitoring databases developed in the North West of England and data from existing surveys of problematic drug users to estimate the total number of problematic drug users, their age and sex characteristics in five health authorities (HA) and one local authority (LA). Furthermore, we estimate the proportion of problematic drug users already in treatment and the required growth in service capacity necessary to deliver treatment to all problematic users. Methods For each CR analysis a minimum of three independent data sources were required covering each geographical area. 13 Overlap (i.e. the same individual appearing on more than one database) could be calculated, as all data systems record initials, dates of birth, sex and some geographical information (usually health authority of residence or treatment). In all areas except Manchester the three sources of data used in the model were drug treatment agencies, police and probation. For Manchester, accident and emergency data replaced probation. Manchester HA and Bolton LA data cover the calendar year 1997, Liverpool HA 1998 and Sefton, St. Helens & Knowsley and Wirral HAs Choice of year was dictated by the most recent sets of data available and suitable for CR analysis. Data from drug treatment agencies included those from statutory and voluntary drug agencies such as Drug Dependency Clinics, Community Drugs Teams and Residential Rehabilitation Centres. 2,14,15 For Merseyside health authorities (Liverpool, Sefton, St Helens & Knowsley and Wirral) these data were provided by the outcomes and prevalence database. 15,16 This database annually records all drug users in contact with drug treatment services. For Manchester HA and Bolton LA treatment data were provided by the Regional Drug Misuse Database (DMD). 17 As part of a national monitoring system, the DMD records data on individuals presenting at services with new requests for treatment. For Merseyside health authorities, police datasets included only those individuals arrested for a drug crime; usually possession, possession with the intent to supply and supply. These data were taken from the Inter-Agency Drug Misuse Database (IAD); a multi-agency drug database recording drug users contacting health or judicial systems. 14,18 Bolton and Manchester police data were from police surgeons and included those arrested specifically for drug crimes as well as individuals arrested for other crimes but subsequently identified as having a drug problem. For Manchester and Bolton, these data were extracted from the DMD. 17 Probation data comprise individuals in contact with the Probation Service who have disclosed a drug problem to that service. In Merseyside data were extracted from the IAD; 14,18 in Manchester and Bolton they were provided through an independent research survey undertaken in Accident and Emergency figures included data recorded on any admission where the clinician considered drug misuse to be a factor. These data are from an unpublished Accident and Emergency research study at Manchester Royal Infirmary Emergency Department and cover only the Manchester area. In all cases individuals were considered to be problematic drug users only if they were reported by any Specialist Drug Service or were reported from any of the other data sources as users of heroin, methadone, cocaine or crack cocaine. Individuals reported for using only cannabis, ecstasy or amphetamine were not included. The majority (92.3 per cent) of individuals identified as problematic drug users were aged between 15 and 44. For the purposes of more robust analysis, and later to allow age-specific rates to be calculated, any individuals aged below 15 or aged 45 and over were also removed. Numbers included in the final data analyses are given in Table 1 categorized by age, sex, reporting agency type and geographical area of report. For CR, data were placed in a 2 n contingency table (where n is number of sources) for Poisson log linear modelling. Best fit between model and observed data is calculated by comparing observed overlaps between datasets with those generated by the model. Such comparisons utilize a 2 test (excluding the value generated by the model for individuals not observed in any datasets; i.e. the unobserved cell). We fit the model with the fewest possible parameters that is not significantly different (p 0.05) from observed values. Finally, we use the model to predict the value of the unobserved cell. Full mathematical details of estimating the size of a closed population have been published in detail elsewhere. 13 All analyses were undertaken using SPSS 20 and Generalized Linear Interactive Modelling (GLIM) 21 statistical software. Results Capture recapture estimates [ 95 per cent confidence intervals (CIs)] for all problematic drug users are presented in Table 2. Measures of closeness of fit between observed data and that generated by the models and any interaction parameters 13 between datasets are also given. Nearly all (27 of 30) models fitted observed data well (i.e. did not differ significantly), increasing confidence in the estimates of the hidden population. For those that differed statistically (Liverpool s overall and years and Sefton s female estimates) the difference only just reached significance (p 0.048, and 0.032, respectively).

3 288 JOURNAL OF PUBLIC HEALTH MEDICINE Table 1 Total number, age range and sex breakdown of problematic drug users used in capture recapture models by geographical area and reporting agencies (numbers, with percentages given in parentheses) Sex Age Geographical area Year Total Male Female Mean age Bolton LA (80.4) 256 (19.6) 632 (48.4) 675 (51.6) 26 Manchester HA (80.4) 544 (19.6) 838 (30.5) 1909 (69.5) 28 Liverpool HA (69.8) 787 (30.2) 467 (17.9) 2140 (82.1) 28 Sefton HA (70.9) 381 (29.1) 196 (15) 1114 (85) 31 St. Helens & Knowsley HA (80.4) 346 (19.6) 407 (23.1) 1354 (76.9) 29 Wirral HA (74.8) 690 (25.2) 361 (13.2) 2374 (86.8) 31 Sex Age Reporting agency type Male Female Mean age Drug treatment agency 4459 (68.5) 2047 (31.5) 1214 (18.7) 5292 (81.3) 29 Probation 1697 (81.8) 378 (18.2) 733 (35.3) 1342 (64.7) 28 Police 534 (83.6) 105 (16.4) 243 (38) 396 (62) 28 Accident & Emergency 206 (70.1) 88 (29.9) 93 (34.4) 177 (65.6) 28 LA, Local Authority; HA, Health Authority. Table 2 Estimated problematic drug use in each geographical area St Helens & Bolton LA Manchester HA Liverpool HA Sefton HA Knowsley HA Wirral HA All Estimate % CIs 1739, , , , , , 4766 Fit (p)* Model 1,2 3 1,2,3 1 3,2 1 3,2 1,2,3 1,3 2 Male Estimate % CIs 1357, , , , , , 3789 Fit (p) Model 1,2 3 1,2,3 1 3,2 1 3,2 1 3,2 1,2 3 Female Estimate % CIs 324, , , , , , 1235 Fit (p) Model 1,2 3 1,2,3 1 3,2 1 3,2 1,2,3 1, years Estimate % CIs 840, , , , , , 1203 Fit (p) Model 1,2 3 1,2,3 1 3,2 1 3,2 1,2,3 1, years Estimate % CIs 847, , , , , , 3763 Fit (p) Model 1,2 3 1,2,3 1 3,2 1 3,2 1,2,3 1,2 3 LA, Local Authority; HA, Health Authority. *p values 0.05 indicate that data from the log linear, capture recapture model are a good fit to known levels of overlap between data sources. Model refers to any interactions between datasets included in the model design. For instance, 1,2,3 indicates that the best-fit model was generated assuming independence between all three datasets whereas 1 2,3 indicates that best fit was obtained by including an interactive term to allow for dependence between datasets 1 and 2. In all areas 1 indicates drug treatment centres and 2 indicates probation; 3 indicates police in all areas except Manchester, where 3 indicates accident and emergency.

4 HIDDEN NEED FOR DRUG TREATMENT SERVICES 289 Estimated total numbers of problematic drug users range from 1983 (95 per cent CIs 1739, 2254) in Bolton to 7220 (95 per cent CIs 6104, 8654) in Manchester. Female to male sex ratios differed between raw data (Table 1) and CR estimates (Table 2) (1:3.2 and 1:3.7, respectively; , p 0.001). Furthermore, between geographical areas the female to male ratios generated by CR varied considerably, with a range from 1:3.2 in Liverpool to 1:6.0 in St. Helens & Knowsley. To correct for differences in sizes and population structures across different geographical areas, rates of problem drug use per 1000 population were calculated (see Table 3). For all drug users (aged 15 44) rates per 1000 ranged from 17.8 per 1000 (95 per cent CIs 15.6, 20.2) in Bolton to 36.5 per 1000 (95 per cent CIs 30.9, 43.8) in Manchester. Males (aged 15 44) provided the highest rates, ranging from 27.2 per 1000 (95 per cent CIs 23.8, 31.4) in Bolton to 55.6 per 1000 (95 per cent CIs 45.8, 68.7) in Manchester. Consistently across all geographical areas, CR rates for drug use among females (overall 12.9 per 1000 females) were considerably lower than the rate among males (overall 45.7 per 1000 males). The Merseyside and Cheshire Prevalence and Outcomes database 15,16 records all individuals currently in treatment and therefore, for the Merseyside health authorities only, it was possible to assess the proportion of the estimated total number of problematic drug users already receiving treatment (Figure 1). Overall 32 per cent (5479/17 096) of problematic drug users were in treatment, ranging from 26.2 per cent in St. Helens & Knowsley to 46.5 per cent in Wirral. Discussion Developing appropriate drug services, targeting prevention strategies as well as monitoring effectiveness all require information on drug users not just in contact with services, but Table 3 Estimated rates of problematic drug users per 1000 of population (with 95 per cent confidence intervals given in parentheses) Problematic users per 1000 of population Total population Area (15 44 years) All Females Males years years Bolton LA (15.6, 20.2) 7.5 (6.0, 9.9) 27.2 (23.8, 31.4) 31.9 (27.3, 38.0) 12.5 (10.5, 15.1) Manchester HA (30.9, 43.8) 15.8 (11.6, 22.8) 55.6 (45.8, 68.7) 32.9 (24.7, 45.5) 36.5 (29.8, 45.5) Liverpool HA (23.9, 54.5) 16.8 (9.8, 38.1) 52.2 (33.4, 92.2) 19.1 (8.0, 92.3) 42.7 (28.9, 69.6) Sefton HA (18.3, 26.0) 10.0 (7.9, 13.4) 32.6 (26.5, 41.3) 19.6 (12.2, 36.2) 22.4 (14.4, 27.1) St. Helens & Knowsley HA (20.4, 28.0) 7.2 (6.0, 9.0) 42.1 (33.8, 53.0) 22.3 (16.6, 31.3) 23.3 (19.6, 28.2) Wirral HA (29.6, 36.9) 15.4 (12.9, 19.0) 51.3 (45.2, 58.8) 23.3 (17.3, 33.3) 35.9 (32.2, 40.4) LA, Local Authority; HA, Health Authority. Percentage reported in treatment All users Female Male Liverpool Sefton St. Helens & Knowsley Geographical location Wirral Figure 1 Percentage of estimated total of problematic drug users seen in specialist drug treatment programme.

5 290 JOURNAL OF PUBLIC HEALTH MEDICINE throughout the wider community. Capture recapture (CR) offers one method of monitoring this population and provides important perspectives on the demography of hidden drug users, penetration of users into treatment services and consequently, the needs for further service and prevention development. The analyses presented here demonstrate that CR can be implemented as part of routine monitoring. This approach makes a number of assumptions. First, the population is assumed to be closed (i.e. with no substantial changes in the population under study during the CR period). Although deaths and migration will have occurred during our study periods, surveillance indicates that levels are low 15 and consequently, unlikely to significantly affect estimates. Second, CR assumes data sources are independent of one another. At the time of this study, referral of drug users between judicial and health agencies was not common practice. Furthermore, by using log linear modelling we have been able to control for dependences between data sources by including appropriate interaction terms in the model (see Table 2). Finally, all members of the population should have the same probability of being captured. Stratifying CR analyses by either age or sex did not alter estimates for problematic drug use (Table 2), suggesting consistent capture (at least by sex and age) between databases. More importantly, however, despite the use of sometimes very different data sources for different areas (compare Manchester and Liverpool), the CR results appear consistent with, for instance, 36.5 and 34.5 problematic drug users per 1000 in Manchester and Liverpool, respectively. Other estimates in the United Kingdom include 3.1 per cent problem drug users (age 15 49) in Lamberth, Southwark and Lewisham HA in 1992 and, for problem opiate users in Newham HA, 3.3 per cent in A previous study in Liverpool 10 estimated problematic drug use in the age band at 16.9 per 1000 in This compares with a rate of 19.1 per 1000 (ages 15 24) identified in this study, suggesting (for Liverpool) a rise in levels of problematic drug use. The size of CR study undertaken here has allowed us to highlight a number of hitherto hidden demographic features of drug use. Thus, large metropolitan areas (here Liverpool and Manchester) appear to conceal disproportionate numbers of problematic drug use (Table 2). Furthermore, analyses by age group suggest that higher rates of drug use in metropolitan areas are due primarily to higher number of older users. Thus, although Liverpool has higher rates of drug use in the age band than surrounding areas (Sefton, St Helens & Knowsley, Wirral) the difference is not apparent in the younger age band (Table 3). Comparisons between Manchester and its neighbour (Bolton) show a similar pattern. This may result from much older epidemics of drug use around larger metropolitan areas. Equally, however, migration of established drug users into larger metropolitan areas later in their drug use career could also contribute to this age distribution. Further research would be needed to identify the relative strength of these effects. Overall, younger drug users were also disproportionately represented in the hidden population (71.1 per cent unknown and 64.4 per cent unknown 25 44, , p 0.001). Thus, unadjusted data from monitoring systems overestimate the age of all problematic users and underestimate the size of the younger population. Furthermore, young users are under-represented to an even greater extent in drug treatment services. Only 21.3 per cent of the estimated problematic users under 25 were in treatment compared with 35.3 per cent of those aged years (Merseyside areas, , p 0.001). Such results highlight the need for CR-adjusted estimates when considering the impact of problematic use on younger age groups and assessing the effectiveness of anti-drugs strategies 1 in recruiting young users into treatment. Population rates were always lower in females than males (Table 3). Higher levels of problematic drug use in males have often been reported in the United Kingdom 15 and elsewhere (Australia 22 and Canada 23 ). Previously, however, it has been difficult to differentiate possible sex differences in sampling or ease of service access from underlying differences in actual levels of problem use. This study identified higher female to male sex ratios in the hidden compared with recorded populations (female to male, 1:4.2 and 1:3.2, respectively; , p 0.001) and a greater proportion of female problematic drug users (47.1 per cent females compared with 27.3 per cent males; Merseyside only) already in contact with specialist drug treatment services ( , p 0.001). Far from being reluctant to contact treatment agencies, results here suggest that female drug users may actually seek assistance more readily than their male counterparts; as is often the case with other health-related conditions. 24 Results indicate that only a third of problematic drug users between 25 and 44 years are in contact with specialist drug services and that this falls to less than a quarter in those under 25. Recent evidence suggests that within 2 years of being discharged drug free from services many of those individuals will be back in treatment. 15 Treatment services, often already with lengthy waiting lists, require substantial growth to meet hidden demand from established users, new users and those relapsing. Without such growth, new initiatives (e.g. AR 3 and DTTO 4 ) designed to bring more individuals into treatment cannot function optimally. In the United Kingdom, drug services have been required to increase the number of problematic drug users in treatment by 66 per cent by 2005 and 100 per cent by Analyses in this study indicate that even an immediate 100 per cent increase in number of users in treatment would (for Merseyside) leave approximately one-third of problematic users outside of service contact. However, the health benefits 26 and economics 27 of continuous growth in treatment services should be weighed against alternatives including more holistic packages for those already in treatment and broader prevention initiatives. A stronger focus on users physical and social environment 28 (compared with maintenance therapy) may improve success in attaining and maintaining a drug-free status and therefore reduce pressures for service growth. Furthermore,

6 HIDDEN NEED FOR DRUG TREATMENT SERVICES 291 efforts to engage younger age groups in treatment may prove successful in reducing problematic incidence by engaging people in treatment before problematic use is well established. The new national drug misuse database, due to be launched for England and Wales in 2001, 29 is broadly based on the enhanced surveillance systems developed in Merseyside 15 and used in these analyses. Consequently, from 2002 routine CR analyses using health and judicial data to identify proportions of individuals in treatment could be possible across England. Paradoxically, as referrals from the judicial system (i.e. AR and DTTO) account for more individuals in treatment, independence between data samples will be compromised and CR analysis made more difficult. However, with the inclusion of more partners in drug use surveillance and intelligence (e.g. social services, prisons and accident and emergency units) such anomalies can be more effectively corrected. 13,14 Thus, greater multiagency collaboration appears to provide the potential both for better monitoring of problematic drug use and for more effective remedies that address both the causes of use and the behavioural and physiological problems associated with addiction. Acknowledgements The authors would like to thank the DMD reference group and the wide range of agencies that continue to provide drug use information to the regional monitoring units. Thanks also go to Jim McVeigh, Diana Leighton and in particular Penny Cook for their assistance, and to two anonymous referees for their constructive comments. References 1 The Stationery Office. Tackling drugs to build a better Britain. The government s 10-year strategy for tackling drug misuse. Cm London: The Stationery Office, University of Manchester and Liverpool John Moores University. Drug misuse in the north west of England Manchester: University of Manchester; Liverpool: Liverpool John Moores University, Home Office. Arrest referral scheme. Home Office Circular 30/2000. London: The Stationery Office, The Crime and Disorder Act. The Drug Testing and Treatment Order. Sections London: The Stationery Office, Begon, M. Investigating animal abundance. In: Investigating animal abundance. London: Edward Arnold, Ismail AA, Beeching NJ, Gill, GV, Bellis, MA. Capture recaptureadjusted prevalence rates of type 2 diabetes are related to social deprivation. Q J Med 1999; 92: Devine MJ, Bellis MA, Tocque K, Syed Q. Whooping cough surveillance in the North West of England. Commun Dis Publ Hlth 1998; 1: Tocque K, Bellis MA, Beeching NJ, Davies PDO. Capture recapture as a method of determining the completeness of tuberculosis notifications. Commun Dis Publ Hlth 2001; 4(2): Bloor M, Leyland A, Barnard M, McKeganey N. Estimating hidden populations: a new method of calculating the prevalence of drug-injecting and non-injecting female street prostitution. Br J Addict 1991; 86: Squire NF, Beeching NJ, Schlect BMJ, Ruten SM. An estimate of the prevalence of drug misuse in Liverpool and a spatial analysis of known addiction. J Publ Hlth Med 1995; 17: Hickman M, Cox S, Harvey J, et al. Estimating the prevalence of problem drug use in inner-london: a discussion of three capture recapture studies. Addiction 1999; 94(11): Cormack RM. Log linear models for capture recapture. Biometrics 1989; 45: Bishop Y, Fienberg SE, Holland PW. Discrete multivariate analysis. Cambridge, MA: MIT Press, 1995: Hardi LV, Hounsome J, Bellis, MA. Merseyside inter agency drug misuse database. Liverpool: Liverpool John Moores University, Birtles RL, Bellis MA. Drug services in Merseyside and Cheshire 1999 prevalence and outcomes. Liverpool: Liverpool John Moores University, Birtles RL, Bellis MA. Drug services in Merseyside and Cheshire 1998 prevalence and outcomes. Liverpool: Liverpool John Moores University, University of Manchester and Liverpool John Moores University. Drug misuse in the North West of England Manchester: University of Manchester; Liverpool: Liverpool John Moores University, Hardi LV, Hounsome J, Bellis MA. Merseyside inter agency drug misuse database. Liverpool: Liverpool John Moores University, Millar T, McFarlane S. Drug misuse monitoring in the Probation Service: combining health and CJS information about drug misuse. Manchester: University of Manchester Drug Misuse Research Unit, Norussis M. SPSS for Windows: base system user s guide: release 6.0. Chicago, IL: SPSS Inc., Francis B, Green M, Payne C, eds. The GLIM system release 4 manual. Oxford: Oxford University Press, Lynskey M, White V, Hill D, Letcher T, Hall W. Prevalence of illicit drug use among youth: results from the Australian School Students Alcohol and Drugs Survey. Aust NZ J Publ Hlth 1999; 23(5): DeWit DJ, Offord DR, Wong M. Patterns of onset and cessation of drug use over the early part of the life course. Health Educ Behav 1997; 24(6): Green CA, Pope CR. Gender, psychosocial factors and the use of medical services: a longitudinal analysis. Social Sci Med 1999; 48(10): United Kingdom Anti-drugs Co-ordinator. Second national plan. London: Central Office of Information, Gossop M, Marsden J, Stewart D. NTORS at one year. The national treatment outcome research study. Changes in substance use, health and criminal behaviour one year after intake. London: Department of Health, Edmunds M, May T, Hearnden I, Hough M. Arrest referral: emerging lessons from research. Home Office paper 23. London: Home Office, Advisory Council on the Misuse of Drugs. Drug misuse and the environment: a report by the Advisory Council on the Misuse of Drugs. London: The Stationery Office, Donmall M, Hickman M, Glavas R. New DMD. Specifications for the new national drug treatment information system. Manchester: University of Manchester, Accepted on 25 June 2001