No full-text available
To read the full-text of this research,
you can request a copy directly from the author.
Estimating the number of people who inject drugs: A proposal to provide figures nationwide and its application to France
Abstract
Background:
This article provides an original method of estimating the prevalence of injecting drug users (IDU) in France.
Method:
In order to overcome legal restrictions hampering data linkage, the method relies on a single-source capture-recapture framework and makes use of individual data collected by a standardized survey conducted within treatment centres. Taking the hierarchical structure of data collection into consideration, a multilevel Zelterman regression model was used.
Results:
In 2014, last year IDU were thought to be 103 800 (85 300; 130 000), prevalence of 2.6‰ (2.1; 3.2), and last month IDU 86 000 (69 200; 110 400), prevalence of 2.2‰ (1.7; 2.7). Estimates have been broken down by gender, revealing a 3 to 1 ratio of male-to-female; and age, showing young adults less prone to regular use of injection. It is thought that one IDU out of three was seen in treatment centres in 2014.
Conclusion:
The number of last month IDU has remained stable when compared with the last available figures calculated in 2006. Nowadays IDU include heroin, buprenorphine, polydrug and stimulant users. This greater diversity ought to be taken into account by prevention policies, and underlines the need for perennial monitoring of IDU.
... The denominators used for rate calculations at the national and subnational level were based on population size (aged 18 to 64) estimates in 2014 produced by the French National Institute of Statistics and Economic Studies [29], which also provides estimates according to the country of birth, and estimates from the latest national surveys on sexual behaviour [30] and drug use [31]. Note that for the denominator of the incidence rate we subtracted the HIV prevalence [12] from the population size. ...
... Potential inaccurate adjustment could affect our estimates. Furthermore, as French surveillance system on newly diagnosed HIV cases was paper-based until recently, it was labour intensive and time-consuming to collect data, and thus, it generated MSM, men who have sex with men; PWID, persons who inject drugs; SSA, Sub-Saharan Africa; CI, confidence interval; IQR, interquartile range; IR, HIV incidence rates (per 10,000 inhabitants aged 18 to 64); UP, undiagnosed HIV prevalence rates (per 10,000 inhabitants aged 18 to 64); TI, time interval from HIV infection to diagnosis for individuals infected between 2011 and 2014; X, values could not be estimated due to insufficient HIV cases; xx, the 2006 French national survey on sexual behaviour [30] and the 2014 survey on drug use [31] did not provide estimates of the sizes of the MSM and PWID populations in overseas regions. ...
... More inclusive definitions, based on sexual activity or drug use over the last two or five years, would have led to higher population sizes and lower rates. Moreover, national surveys [30,31] provided estimates of the prevalence of MSM and PWID at the national and subnational level, but not according to the country of birth. We then assumed that the prevalence of MSM and PWID was similar among individuals born in France and those born abroad, but it remains unclear whether and how this assumption could have biased our results. ...
Introduction:
To close gaps in HIV prevention and care, knowledge about locations and populations most affected by HIV is essential. Here, we provide subnational and sub-population estimates of three key HIV epidemiological indicators, which have been unavailable for most settings.
Methods:
We used surveillance data on newly diagnosed HIV cases from 2004 to 2014 and back-calculation modelling to estimate in France, at national and subnational levels, by exposure group and country of birth: the numbers of new HIV infections, the times to diagnosis, the numbers of undiagnosed HIV infections. The denominators used for rate calculations at national and subnational levels were based on population size (aged 18 to 64) estimates produced by the French National Institute of Statistics and Economic Studies and the latest national surveys on sexual behaviour and drug use.
Results:
We estimated that, in 2014, national HIV incidence was 0.17‰ (95% confidence intervals (CI): 0.16 to 0.18) or 6607 (95% CI: 6057 to 7196) adults, undiagnosed HIV prevalence was 0.64‰ (95% CI: 0.57 to 0.70) or 24,197 (95% CI: 22,296 to 25,944) adults and median time to diagnosis over the 2011 to 2014 period was 3.3 years (interquartile range: 1.2 to 5.7). Three mainland regions, including the Paris region, out of the 27 French regions accounted for 56% of the total number of new and undiagnosed infections. Incidence and undiagnosed prevalence rates were 2- to 10-fold higher than the national rates in three overseas regions and in the Paris region (p-values < 0.001). Rates of incidence and undiagnosed prevalence were higher than the national rates for the following populations (p-values < 0.001): born-abroad men who have sex with men (MSM) (respectively, 108- and 78-fold), French-born MSM (62- and 44-fold), born-abroad persons who inject drugs (14- and 18-fold), sub-Saharan African-born heterosexuals (women 15- and 15-fold, men 11- and 13-fold). Importantly, affected populations varied from one region to another, and in regions apparently less impacted by HIV, some populations could be as impacted as those living in most impacted regions.
Conclusions:
In France, some regions and populations have been most impacted by HIV. Subnational and sub-population estimates of key indicators are not only essential to adapt, design implement and evaluate tailored HIV interventions in France, but also elsewhere where similar heterogeneity is likely to exist.
... Data from an internet survey conducted in Australia showed that 4.7% of the gay and bisexual men surveyed reported having injected drugs in the last 6 months (in the context of sex for 91.4% of them) (Bui et al., 2018), which is in line with our results. Even though injection in slamming contexts seems to concern a fraction of MSM, its prevalence in this population is still 3 to 4 times higher than the prevalence of drug injection observed in the general population in France (Janssen, 2017;Sarasa-Renedo, Barrio, Montanari, Guarita & De La Fuente, 2015). Recent estimates among the general population based on indirect statistical methods such as capture-recapture indicate a prevalence of injection at 0.26% (95% CI = 0.21-0.32) in France (Janssen, 2017) and 0.25% (95% CI = 0.22-0.3) in Europe (Sarasa-Renedo et al., 2015) (context of injection, sexual or not, was not specified). ...
... Even though injection in slamming contexts seems to concern a fraction of MSM, its prevalence in this population is still 3 to 4 times higher than the prevalence of drug injection observed in the general population in France (Janssen, 2017;Sarasa-Renedo, Barrio, Montanari, Guarita & De La Fuente, 2015). Recent estimates among the general population based on indirect statistical methods such as capture-recapture indicate a prevalence of injection at 0.26% (95% CI = 0.21-0.32) in France (Janssen, 2017) and 0.25% (95% CI = 0.22-0.3) in Europe (Sarasa-Renedo et al., 2015) (context of injection, sexual or not, was not specified). ...
Background
In the last decade, European cities saw the development of “slamming,” a practice related to chemsex that combines three elements: a sexual context, psychostimulant drug use, and injection practices. Epidemiological data on this practice is still sparse and media attention might have unintentionally distorted the size of this phenomenon. Therefore, we aimed to estimate the prevalence of men practicing slam and to identify factors associated with this practice.
Methods
We used data from the Prevagay 2015 bio-behavioral survey to estimate the prevalence of slamming practices. A time-location sampling was performed among gay-labeled venues in five French cites. Behavioral information was recorded using a self-administered questionnaire. The HIV and HCV serostatus were investigated using ELISA tests on dried blood spots. The factors associated with slamming were assessed using a multiple logistic regression. We applied a weighting mechanism to enhance the generalizability of the estimates.
Results
Among the 2646 men who have sex with men (MSM) included in our study, 3.1% reported slamming at least once during their lifetime (95% confidence interval (CI) = 2.2–4.3) and 1.6% (95% CI = 1–2.3) said they participated in a slamming session in the last 12 months. In the multivariate analysis, both HCV and HIV biological status were strongly associated with practicing “slam” in the last 12 months (OR = 13.37 (95% CI = 3.26–54.81) and 4.73 (95% CI = 1.58–14.44), respectively). Furthermore, a ten-point decrease in mental health scores was linked with the practice with an OR of 1.37 (95% CI = 1.08–1.73), indicating poorer mental health.
Conclusion
Even though slamming seems to involve a relatively small proportion of MSM, the vulnerability of this sub-group is high enough to justify setting up harm reduction measures and specific care. Training health professionals and creating services combining sexual health and drug dependence could be an effective response.
... Treatment centres are also requested to provide information on the type of contact each individual establishes with the structure: first demand, uninterrupted treatment follow-up, and readmission. Readmission is defined as no contact with the referral centre for at least 6 months and labelled as a new contact with the treatment centre, in which case individuals are assumed to be recorded twice (Galai, Safaeian, Boltin, & Celetano, 2003;Genberg et al., 2011;Janssen, 2018). ...
... This linear predictor can be approximated by a logistic regression. The method has been extrapolated to provide reliable estimates of substance users at the national level, using multilevel modelling (Janssen, 2016(Janssen, , 2018. The above-mentioned estimator is unbiased as long as heterogeneity is captured through the regression process, and is known to be robust in case of mild heterogeneity. ...
Background:
Despite several sources corroborating an expanding market and increased visibility and greater diversity in users' profiles, very little is known about the number of crack cocaine users in France.
Method:
The estimates rely on a single data source capture-recapture method. Annual data are extracted from treatment centres nationwide. To account for heterogeneity, we use an innovative zero-truncated geometric, regression-based estimator controlling for individual and centre characteristics. We use the well-known Zelterman estimator as a benchmark.
Results:
The number of crack cocaine users received in treatment centres increased dramatically, from 3388 in 2010 to 5143 in 2017 (+52%). The estimated number of crack cocaine users is believed to have tripled over the course of the same period (from 9775, 95% CI [8288-11530] to 28983 [24876-33766], respectively), with prevalences below 1‰, similar to other European countries. The coverage rate (observed number/estimated number of users) decreased in a similar fashion, indicating lower utilization. In particular, females and younger users are underestimated by data from treatment centres.
Conclusion:
The prevalence of crack cocaine use is fairly low but steadily increasing. The diversity in users' profiles is a challenge to prevention and public health policies that should expand their scope to a more inclusive perspective of what defines crack cocaine users. Our method overcomes several methodological issues (data sources, data linkage, heterogeneity) and can be easily applied to a wide range of settings.
... To date, very little is known about recent injection patterns at a population level and their potential evolution over time. Indeed, PWID continue to constitute an elusive, hard-to-reach population, despite recent efforts to provide a closer estimate of the number of people injecting drugs in France (Janssen 2018). A recent methodology developed to identify drugs injected by PWID analyzed used syringes and assessed the percentage of reuse in France, but information on patterns of drug use, in terms of specific practices, user characteristics, and standards of living, remains sparse (Néfau et al. 2015). ...
... Differences in age at first injection have been diminishing between the genders, suggesting increasingly converging behaviors between male and female users. Consistent with previous findings (Janssen 2018), the results reported show that a delayed age at first injection does not yield a decreased prevalence of injecting drugs among PWID accessing French harm reduction facilities, nor a reduction in daily injection. On the contrary, from 2008 to 2015, the prevalence of daily injection among substance users in harm reduction facilities increased significantly. ...
This exploratory analysis of time-series cross-sectional data provides insights on trends in age at first injection among people who inject drugs in France, and on associations with recent risky injecting behaviors. Data were collected from a national survey conducted in harm reduction facilities in five phases between 2006 and 2015. Standardized questionnaires collected information on demographics, substance use, and route of administration, as well as lifetime and past-month injection. Descriptive and multilevel models were applied to account for the hierarchical structure of the data. Prevalence of lifetime and past-month injection remained stable over time, while the prevalence of daily injection increased significantly. Mean age at first injection only appeared to increase for data collected after 2010. Gender differences in mean age at first injection decreased over time, suggesting the development of converging patterns of initiation independent of sex. After controlling for covariates, early initiation of injection was unrelated to daily injection or material sharing, and associated with the number of recently injected substances. Early initiation is likely a predictor of injected polysubstance use. Findings are relevant to the planning, implementation, and evaluation of prevention programs.
Background and aims:
Capture-recapture (CRC) analysis is recommended for estimating the prevalence of problem drug use or people who inject drugs (PWID). We aim to demonstrate how naïve application of CRC can lead to highly misleading results, and to suggest how the problems might be overcome.
Methods:
We present a case study of estimating the prevalence of PWID in Bristol, UK, applying CRC to lists in contact with three services. We assess: (i) sensitivity of results to different versions of the dominant (treatment) list: specifically, to inclusion of non-incident cases and of those who were referred directly from one of the other services; (ii) the impact of accounting for a novel covariate, housing instability; (iii) consistency of CRC estimates with drug-related mortality data. We then formally incorporate the drug-related mortality data and lower bounds for prevalence alongside the CRC, in a single coherent model.
Results:
Five of eleven models fitted the full data equally well but generated widely varying prevalence estimates, from 2740 (95% CI 2670, 2840) to 6890 (95% CI 3740, 17680). Results were highly sensitive to inclusion of non-incident cases, demonstrating the presence of considerable heterogeneity, and were sensitive to a lesser extent to inclusion of direct referrals. A reduced dataset including only incident cases and excluding referrals could be fitted by simpler models, and led to much greater consistency in estimates. Accounting for housing stability improved model fit considerably more than did the standard covariates of age and gender. External data provided validation of results and aided model selection, generating a final estimate of the number of PWID in Bristol in 2011 of 2770 (95% Cr-I 2570, 3110), or 0.9% (95% Cr-I 0.9, 1.0%) of the population aged 15-64 years.
Conclusions:
Steps can be taken to reduce bias in capture-recapture analysis, including consideration of data sources, reduction of lists to less heterogeneous sub-samples, use of covariates, and formal incorporation of external data. This article is protected by copyright. All rights reserved.
Importance
Over the past several years, there have been a number of mainstream media reports that the abuse of heroin has migrated from low-income urban areas with large minority populations to more affluent suburban and rural areas with primarily white populations.Objective
To examine the veracity of these anecdotal reports and define the relationship between the abuse of prescription opioids and the abuse of heroin.Design, Setting, and Participants
Using a mixed-methods approach, we analyzed (1) data from an ongoing study that uses structured, self-administered surveys to gather retrospective data on past drug use patterns among patients entering substance abuse treatment programs across the country who received a primary (DSM-IV) diagnosis of heroin use/dependence (n = 2797) and (2) data from unstructured qualitative interviews with a subset of patients (n = 54) who completed the structured interview.Main Outcomes and Measures
In addition to data on population demographics and current residential location, we used cross-tabulations to assess prevalence rates as a function of the decade of the initiation of abuse for (1) first opioid used (prescription opioid or heroin), (2) sex, (3) race/ethnicity, and (4) age at first use. Respondents indicated in an open-ended format why they chose heroin as their primary drug and the interrelationship between their use of heroin and their use of prescription opioids.Results
Approximately 85% of treatment-seeking patients approached to complete the Survey of Key Informants’ Patients Program did so. Respondents who began using heroin in the 1960s were predominantly young men (82.8%; mean age, 16.5 years) whose first opioid of abuse was heroin (80%). However, more recent users were older (mean age, 22.9 years) men and women living in less urban areas (75.2%) who were introduced to opioids through prescription drugs (75.0%). Whites and nonwhites were equally represented in those initiating use prior to the 1980s, but nearly 90% of respondents who began use in the last decade were white. Although the “high” produced by heroin was described as a significant factor in its selection, it was often used because it was more readily accessible and much less expensive than prescription opioids.Conclusion and Relevance
Our data show that the demographic composition of heroin users entering treatment has shifted over the last 50 years such that heroin use has changed from an inner-city, minority-centered problem to one that has a more widespread geographical distribution, involving primarily white men and women in their late 20s living outside of large urban areas.
Background
Injection drug use provides an efficient mechanism for transmitting bloodborne viruses, including human immunodeficiency virus (HIV) and hepatitis C virus (HCV). Effective targeting of resources for prevention of HIV and HCV infection among persons who inject drugs (PWID) is based on knowledge of the population size and disparity in disease burden among PWID. This study estimated the number of PWID in the United States to calculate rates of HIV and HCV infection.
Methods
We conducted meta-analysis using data from 4 national probability surveys that measured lifetime (3 surveys) or past-year (3 surveys) injection drug use to estimate the proportion of the United States population that has injected drugs. We then applied these proportions to census data to produce population size estimates. To estimate the disease burden among PWID by calculating rates of disease we used lifetime population size estimates of PWID as denominators and estimates of HIV and HCV infection from national HIV surveillance and survey data, respectively, as numerators. We calculated rates of HIV among PWID by gender-, age-, and race/ethnicity.
Results
Lifetime PWID comprised 2.6% (95% confidence interval: 1.8%–3.3%) of the U.S. population aged 13 years or older, representing approximately 6,612,488 PWID (range: 4,583,188–8,641,788) in 2011. The population estimate of past-year PWID was 0.30% (95% confidence interval: 0.19 %–0.41%) or 774,434 PWID (range: 494,605–1,054,263). Among lifetime PWID, the 2011 HIV diagnosis rate was 55 per 100,000 PWID; the rate of persons living with a diagnosis of HIV infection in 2010 was 2,147 per 100,000 PWID; and the 2011 HCV infection rate was 43,126 per 100,000 PWID.
Conclusion
Estimates of the number of PWID and disease rates among PWID are important for program planning and addressing health inequities.
Capture-recapture methods, largely developed in ecology, are now commonly used in epidemiology to adjust for incomplete registries and to estimate the size of difficult-to-reach populations such as problem drug users. Overlapping lists of individuals in the target population, taken from administrative data sources, are considered analogous to overlapping "captures" of animals. Log-linear models, incorporating interaction terms to account for dependencies between sources, are used to predict the number of unobserved individuals and, hence, the total population size. A standard assumption to ensure parameter identifiability is that the highest-order interaction term is 0. We demonstrate that, when individuals are referred directly between sources, this assumption will often be violated, and the standard modeling approach may lead to seriously biased estimates. We refer to such individuals as having been "precaptured," rather than truly recaptured. Although sometimes an alternative identifiable log-linear model could accommodate the referral structure, this will not always be the case. Further, multiple plausible models may fit the data equally well but provide widely varying estimates of the population size. We demonstrate an alternative modeling approach, based on an interpretable parameterization and driven by careful consideration of the relationships between the sources, and we make recommendations for capture-recapture in practice.
In public health, demography and sociology, large-scale surveys often follow a hierarchical data structure as the surveys are based on multistage stratified cluster sampling. The appropriate approach to analyzing such survey data is therefore based on nested sources of variability which come from different levels of the hierarchy. When the variance of the residual errors is correlated between individual observations as a result of these nested structures, traditional logistic regression is inappropriate. We use the 2004 Bangladesh Demographic and Health Survey (BDHS) contraceptive binary data which is a multistage stratified cluster dataset. This dataset is used to exemplify all aspects of working with multilevel logistic regression models, including model conceptualization, model description, understanding of the structure of required multilevel data, estimation of the model via the statistical package MLwiN, comparison between different estimations, and investigation of the selected determinants of contraceptive use.
Using Bayesian capture-recapture analysis, we estimated the number of current injecting drug users (IDUs) in Scotland in 2006 from the cross-counts of 5670 IDUs listed on four data-sources: social enquiry reports (901 IDUs listed), hospital records (953), drug treatment agencies (3504), and recent Hepatitis C virus (HCV) diagnoses (827 listed as IDU-risk). Further, we accessed exact numbers of opiate-related drugs-related deaths (DRDs) in 2006 and 2007 to improve estimation of Scotland's DRD rates per 100 current IDUs. Using all four data-sources, and model-averaging of standard hierarchical log-linear models to allow for pairwise interactions between data-sources and/or demographic classifications, Scotland had an estimated 31700 IDUs in 2006 (95% credible interval: 24900-38700); but 25000 IDUs (95% CI: 20700-35000) by excluding recent HCV diagnoses whose IDU-risk can refer to past injecting. Only in the younger age-group (15-34 years) were Scotland's opiate-related DRD rates significantly lower for females than males. Older males' opiate-related DRD rate was 1.9 (1.24-2.40) per 100 current IDUs without or 1.3 (0.94-1.64) with inclusion of recent HCV diagnoses. If, indeed, Scotland had only 25000 current IDUs in 2006, with only 8200 of them aged 35+ years, the opiate-related DRD rate is higher among this older age group than has been appreciated hitherto. There is counter-balancing good news for the public health: the hitherto sharp increase in older current IDUs had stalled by 2006.
Background:
To estimate the prevalence of problem drug use (PDU) and injecting drug use (IDU) in Luxembourg and analyze trends between 1997 and 2009. To assess the feasibility of prevalence estimations based on drug use surveillance systems.
Methods:
Serial multi-method PDU/IDU prevalence estimations based upon capture-recapture, Poisson regression, multiplier and back-calculation methods. Comparative analysis of methods and assessment of their robustness to variations of external factors.
Results:
National PDU and IDU prevalence rates were estimated at 6.16/1,000 (95% CI 4.62/1,000 to 7.81/1,000) and 5.68/1,000 (95% CI 4.53/1,000 to 6.85/1,000) inhabitants aged 15-64 years, respectively. Absolute prevalence and prevalence rates of PDU increased between 1997 and 2000 and declined from 2003 onwards, whereas IDU absolute prevalence and prevalence rates witnessed an increasing trend between 1997 and 2007.
Conclusions:
Drug use surveillance systems can be valuable instruments for the estimation and trend analysis of drug misuse prevalence given multiple methods are applied that rely on serial and representative data from different sources and different settings, control multiple counts and build upon standardized and sustained data collection routines. The described institutional contact indicator revealed to be a useful tool in the context of PDU/IDU prevalence estimations and thus contributes to enhancing evidence-based drug policy planning.
february 2001), Point and interval estimation of the population size using the truncated Poisson regression model, Utrecht. Summary A method to derive point and interval estimates for the total number of in-dividuals in a heterogenous Poisson population is presented. The method is based on the Horvitz-Thompson approach (Kendall and Stuart, 1991). The zero-truncated Poisson regression model is fitted and results are used to obtain point and interval estimates for the total number of individuals in the population. The method is assessed by performing a simulation experiment computing coverage probabilities of Horvitz-Thompson confidence intervals for cases with different sample sizes and Poisson parameters. We illustrate our method using capture-recapture data from the police registration system providing information on ille-gal immigrants in four large cities in the Netherlands.
Little information exists on the population prevalence or geographic distribution of injection drug users (IDUs) who are Hispanic in the USA. Here, we present yearly estimates of IDU population prevalence among Hispanic residents of the 96 most populated US metropolitan statistical areas (MSAs) for 1992-2002. First, yearly estimates of the proportion of IDUs who were Hispanic in each MSA were created by combining data on (1) IDUs receiving drug treatment services in Substance Abuse and Mental Health Services Administration (SAMHSA)'s Treatment Entry Data System, (2) IDUs being tested in the Centers for Disease Control and Prevention (CDC) HIV-Counseling and Testing System, and (3) incident AIDS diagnoses among IDUs, supplemented by (4) data on IDUs who were living with AIDS. Then, the resulting proportions were multiplied by published yearly estimates of the number of IDUs of all racial/ethnic groups in each MSA to produce Hispanic IDU population estimates. These were divided by Hispanic population data to produce population prevalence rates. Time trends were tested using mixed-effects regression models. Hispanic IDU prevalence declined significantly on average (1992 mean = 192, median = 133; 2002 mean = 144, median = 93; units are per 10,000 Hispanics aged 15-64). The highest prevalence rates across time tended to be in smaller northeastern MSAs. Comparing the last three study years to the first three, prevalence decreased in 82% of MSAs and increased in 18%. Comparisons with data on drug-related mortality and hepatitis C mortality supported the validity of the estimates. Generally, estimates of Hispanic IDU population prevalence were higher than published estimates for non-Hispanic White residents and lower than published estimates for non-Hispanic Black residents. Further analysis indicated that the proportion of IDUs that was Hispanic decreased in 52% and increased in 48% of MSAs between 2002 and 2007. The estimates resulting from this study can be used to investigate MSA-level social and economic factors that may have contributed to variations across MSAs and to help guide prevention program planning for Hispanic IDUs within MSAs. Future research should attempt to determine to what extent these trends are applicable to Hispanic national origin subgroups.
The objective of this study was to identify longitudinal patterns of injection drug use over 20 years in the AIDS Linked to the Intravenous Experience (ALIVE) Study, a community-based cohort of injection drug users (IDUs) in Baltimore, Maryland, with a focus on injection cessation. Starting in 1988, persons over 18 years of age with a history of injection drug use were recruited into the study. Participants provided information on their injection drug use semiannually through 2008. The analysis was restricted to 1,716 IDUs with at least 8 study visits. Finite mixture models were used to identify trajectories and predictors of injection patterns over time. The mean age of participants was 35 years; 75% were male, and 95% were African-American. Five distinct patterns were identified: 2 usage patterns (32% engaged in persistent injection and 16% had frequent relapse) and 3 cessation patterns (early cessation (19%), delayed cessation (16%), and late cessation (18%)). A history of drug treatment, no recent use of multiple substances, and less frequent injection distinguished the early cessation group from the other groups. This study demonstrated multiple trajectories of drug injection behaviors, with a substantial proportion of IDUs stopping injection over extended time frames. For maximum effectiveness, public health programs for IDUs should be long-term, comprehensive, and targeted toward individual patterns of use.
In this paper we consider the estimation of population size from one-source capture--recapture data, that is, a list in which individuals can potentially be found repeatedly and where the question is how many individuals are missed by the list. As a typical example, we provide data from a drug user study in Bangkok from 2001 where the list consists of drug users who repeatedly contact treatment institutions. Drug users with 1, 2, 3,... contacts occur, but drug users with zero contacts are not present, requiring the size of this group to be estimated. Statistically, these data can be considered as stemming from a zero-truncated count distribution. We revisit an estimator for the population size suggested by Zelterman that is known to be robust under potential unobserved heterogeneity. We demonstrate that the Zelterman estimator can be viewed as a maximum likelihood estimator for a locally truncated Poisson likelihood which is equivalent to a binomial likelihood. This result allows the extension of the Zelterman estimator by means of logistic regression to include observed heterogeneity in the form of covariates. We also review an estimator proposed by Chao and explain why we are not able to obtain similar results for this estimator. The Zelterman estimator is applied in two case studies, the first a drug user study from Bangkok, the second an illegal immigrant study in the Netherlands. Our results suggest the new estimator should be used, in particular, if substantial unobserved heterogeneity is present. Comment: Published in at http://dx.doi.org/10.1214/08-AOAS214 the Annals of Applied Statistics (http://www.imstat.org/aoas/) by the Institute of Mathematical Statistics (http://www.imstat.org)
This paper presents the zero-truncated negative binomial regression model to estimate the population size in the presence of a single registration file. The model is an alternative to the zero-truncated Poisson regression model and it may be useful if the data are overdispersed due to unobserved heterogeneity. Horvitz–Thompson point and interval estimates for the population size are derived, and the performance of these estimators is evaluated in a simulation study. To illustrate the model, the size of the population of opiate users in the city of Rotterdam is estimated. In comparison to the Poisson model, the zero-truncated negative binomial regression model fits these data better and yields a substantially higher population size estimate.
This article estimates the population prevalence of current injection drug users (IDUs) in 96 large US metropolitan areas to facilitate structural analyses of its predictors and sequelae and assesses the extent to which drug abuse treatment and human immunodeficiency virus (HIV) counseling and testing are made available to drug injectors in each metropolitan area. We estimated the total number of current IDUs in the United States and then allocated the large metropolitan area total among large metropolitan areas using four different multiplier methods. Mean values were used as best estimates, and their validity and limitations were assessed. Prevalence of drug injectors per 10,000 population varied from 19 to 173 (median 60; interquartile range 42-87). Proportions of drug injectors in treatment varied from 1.0% to 39.3% (median 8.6%); and the ratio of HIV counseling and testing events to the estimated number of IDUs varied from 0.013 to 0.285 (median 0.082). Despite limitations in the accuracy of these estimates, they can be used for structural analyses of the correlates and predictors of the population density of drug injectors in metropolitan areas and for assessing the extent of service delivery to drug injectors. Although service provision levels varied considerably, few if any metropolitan areas seemed to be providing adequate levels of services.
A number of studies suggest females may be more likely to engage in injection and sex risk behavior than males. Most data on gender differences come from industrialized countries, so data are needed in developing countries to determine how well gender differences generalize to these understudied regions.
Between 1999 and 2003, 2512 male and 672 female current injection drug users (IDUs) were surveyed in ten sites in developing countries around the world (Nairobi, Beijing, Hanoi, Kharkiv, Minsk, St. Petersburg, Bogotá, Gran Rosario, Rio, and Santos). The survey included a variety of questions about demographics, injecting practices and sexual behavior.
Females were more likely to engage in risk behaviors in the context of a sexual relationship with a primary partner while males were more likely to engage in risk behaviors in the context of close friendships and casual sexual relationships. After controlling for injection frequency, and years injecting, these gender differences were fairly consistent across sites.
Gender differences in risk depend on the relational contexts in which risk behaviors occur. The fact that female and male risk behavior often occurs in different relational contexts suggests that different kinds of prevention interventions which are sensitive to these contexts may be necessary.
In this paper we consider the estimation of population size from one-source capture–recapture data, that is, a list in which individuals can potentially be found repeatedly and where the question is how many individuals are missed by the list. As a typical example, we provide data from a drug user study in Bangkok from 2001 where the list consists of drug users who repeatedly contact treatment institutions. Drug users with 1, 2, 3, … contacts occur, but drug users with zero contacts are not present, requiring the size of this group to be estimated. Statistically, these data can be considered as stemming from a zero-truncated count distribution. We revisit an estimator for the population size suggested by Zelterman that is known to be robust under potential unobserved heterogeneity. We demonstrate that the Zelterman estimator can be viewed as a maximum likelihood estimator for a locally truncated Poisson likelihood which is equivalent to a binomial likelihood. This result allows the extension of the Zelterman estimator by means of logistic regression to include observed heterogeneity in the form of covariates. We also review an estimator proposed by Chao and explain why we are not able to obtain similar results for this estimator. The Zelterman estimator is applied in two case studies, the first a drug user study from Bangkok, the second an illegal immigrant study in the Netherlands. Our results suggest the new estimator should be used, in particular, if substantial unobserved heterogeneity is present
Background:
An updated prevalence estimate for heroin users in France is lacking. Accurate figures are needed to estimate the potential need for treatment.
Method:
This research relies on a capture-recapture framework and, for the first time, makes use of individual data collected by a standardized survey conducted within treatment centers. Taking the hierarchical structure of data collection into consideration, a multilevel Zelterman regression model was used, thus enabling estimates at national level.
Results:
The number of heroin users was estimated to be 76,000, 95% CI = [68,000; 85,000] in 2009 (prevalence of 1.9‰ [1.7-2.1]), 101,000 in 2011, 95% CI = [66,000-137,000] (prevalence of 2.5‰ [1.6-3.4]), and 107,000 in 2013, 95% CI = [85,000-124,000] (prevalence of 2.6‰ [2.1-3.1]), with a 4:1 ratio of male:female. Less than one heroin users out of three was seen in treatment centers in 2013.
Conclusions:
The rise in the number of heroin users is related to an increasing availability of heroin, increasing use in more remote areas, and providers targeting new types of users. The figures, however, should be completed with alternative estimates and enhance the need for a perennial monitoring.
Introduction and aims:
Field studies have indicated a recent increase in heroin availability and use in France, and yet very little is known about the mechanisms underlying heroin retail prices. This paper offers a first attempt at identifying the determinants of heroin pricing, to measure quantity discounts and assess the influence of purity on street prices, while controlling for a geographical effect.
Design and methods:
Data on heroin samples were collected during 2011 in seven urban areas of metropolitan France. Ordinary least squares regression was used to model the associations between price, quantity, purity and other independent variables.
Results:
Quantity remains the most influential variable on heroin pricing. We estimate that a 10% increase in the size of a transaction leads to a 2.3% decrease in the unit price. Assessed purity proved to be significant, although in modest proportion. Sociodemographic characteristics, such as gender, users' experience and relationships with dealers, proved to be insignificant. Heroin retail prices vary according to a geographical gradient related to the routes of entry and distribution.
Discussion and conclusions:
As a credence good, heroin retail prices in France are affected by more than simply the traditional supply and demand relationship. The results of this study also underline the limitations of a quantitative framework and should be complemented by further ethnographic research to obtain an in-depth understanding of local markets. Policies should be designed to better take local disparities into account.[Lahaie E, Janssen E, Cadet-Taïrou A. Determinants of heroin retail prices in metropolitan France:Discounts, purity and local markets. Drug Alcohol Rev 2015;00:000-000].
The objective of this study was to characterize longitudinal patterns of drug injection behavior for individuals and to identify their early determinants. Participants were 1,339 injection drug users recruited into the AIDS Link to Intravenous Experience (ALIVE) Study in Baltimore, Maryland, through community outreach efforts. The study was initiated in 1988, and follow-up continued through 2000, with semiannual visits. Patterns of self-reported drug injection (yes/no) were defined for each participant, based on the number of drug-use transitions. The effect of baseline factors was assessed using multinomial logistic regression models. Over the 12-year study period, four patterns were noted: 29% of participants remained persistent drug injectors, 20% ceased injection, 14% relapsed once, and 37% had multiple transitions. Persistent injectors had the shortest follow-up and the highest mortality. For persons who changed their behavior, 3.4 years elapsed before their first cessation attempt, on average. Factors differentiating the groups included history of incarceration, young age, participation in drug treatment programs, recent overdose, and commercial sex. The observed long-term injection patterns are consistent with the view of drug addiction as a chronic disease. This view emphasizes the need for prolonged efforts to sustain cessation and to prevent adverse health and social outcomes among injection drug users.
Log-linear models are developed for capture-recapture experiments, and their advantages and disadvantages discussed. Ways in which they can be extended, sometimes with only partial success, to open populations, subpopulations, trap dependence, and long chains of recapture periods are presented. The use of residual patterns, and analysis of subsets of data, to identify behavioural patterns and acceptable models is emphasised and illustrated with two examples.
Injecting drug users (IDUs) have a direct social and economic effect yet can typically be regarded as a hidden population within a community. We estimate the size of the IDU population across the nine different Government Office regions of England in 2005–2006 by using capture–recapture methods with age (ranging from 15 to 64 years) and gender as covariate information. We consider a Bayesian model averaging approach using log‐linear models, where we can include explicit prior information within the analysis in relation to the total IDU population (elicited from the number of drug‐related deaths and injectors’ drug‐related death rates). Estimation at the regional level allows for regional heterogeneity with these regional estimates aggregated to obtain a posterior mean estimate for the number of England's IDUs of 195840 with 95% credible interval (181700, 210480). There is significant variation in the estimated regional prevalence of current IDUs per million of population aged 15–64 years, and in injecting drug‐related death rates across the gender × age cross‐classifications. The propensity of an IDU to be seen by at least one source also exhibits strong regional variability with London having the lowest propensity of being observed (posterior mean probability 0.21) and the South West the highest propensity (posterior mean 0.46).
Background and objectives:
Risk-taking behaviors differ among women and men injection drug users (IDU).
Goal:
To specify the nature of sexual and drug risk-taking among women IDU and ex-IDU and how it relates to partner characteristics.
Design:
A cross-sectional analysis of 324 HIV+ subjects enrolled into a prospective cohort study in Marseille, France.
Results:
Women, as compared with men, were considerably more likely to report nonuse of condoms with a main partner (31% versus 12%). They were more likely to shoot with a partner at last injection (39% versus 12%), but far less likely to sterilize used needles (4% versus 16%). Two thirds of both men and women reported consistent condom use with a seronegative partner, but only 47% of men and 23% of women reported the same with a seropositive partner. Among the women only, needle and syringe sharing was associated with consistent use of a condom.
Conclusions:
Women reported behaviors which protect their partners from STD infection more frequently than behaviors which protect themselves. Greater attention must be paid to sexual risk-taking among HIV+ women.
One way in which a capture-recapture experiment can be designed is to count each capture occasion as a separate sample, and tally the number of individuals caught once, twice, and so on. This paper discusses the estimators currently used for this type of experiment, and suggests some new ones. Each of the estimators is categorized by its derivation. Following this, their properties are discussed, particularly their bias and variance, with the aim of choosing one or two of the estimators which are the 'best' to use, especially in handling a heterogeneous population. It is found that the bias adjusted estimator of Chao (1989) is the best to use when the number of captures is relatively small, and that the estimator of Darroch & Ratcliff (1980) should be used otherwise.
Capture-recapture methodologies have been used to estimate the size of the hidden population of active offenders on the basis of the observed properties of the truncated distribution of arrested offenders. We use this approach to estimate the odds of arrest of marijuana, cocaine, crack, and heroin dealers and users in one Canadian province (Quebec). Findings indicate that risks of being arrested are much higher for sellers than for consumers and that this gap widens for the more harmful drugs. Findings also show, however, that vulnerability to arrest was significantly higher for marijuana users than for others users and that dealers in the smaller but more harmful drug markets (crack and heroin) manage to experience lower aggregate risks of arrest than cocaine or marijuana dealers.
How many kinds are there? Suppose that a population is partitioned into C classes. In many situations interest focuses not on estimation of the relative sizes of the classes, but on estimation of C itself. For example, biologists and ecologists may be interested in estimating the number of species in a population of plants or animals, numismatists may be concerned with estimating the number of dies used to produce an ancient coin issue, and linguists may be interested in estimating the size of an author's vocabulary. In this article we review the problem of statistical estimation of C. Many approaches have been proposed, some purely data-analytic and others based in sampling theory. In the latter case numerous variations have been considered. The population may be finite or infinite. In finite, samples may be taken with replacement (multinomial sampling) or without replacement (hypergeometric sampling), or by Bernoulli sampling; if infinite, sampling may be multinomial or Bernoulli, or the sample may be the result of random Poisson contributions of each class. Given a sampling model, one may approach estimation of C via a parametric or nonparametric formulation; in either case there may be frequentist and Bayesian procedures. We begin by discussing the existing literature on this problem (over 120 references), organizing it by sampling model, population specification, and philosophy of estimation. We find that (a) the problem is quite resistant to statistical solution, essentially because no matter how many classes have been observed, there may still be a large number of very small unobserved classes; (b) many closely related estimation procedures have been developed independently and have not yet been compared; (c) there is not as yet a globally preferable estimator of C, although for some models there is an acceptable estimator (for some not even this is true); and (d) there are promising directions for research to pursue; for example, it appears possible to exploit estimates of the "coverage" of the sample (the total proportion of the population represented by the observed classes) to improve the accuracy of estimators of the number of classes. Finally, we make specific recommendations for future research, regarding parametric estimation, coverage-based estimation, resampling methods, Poisson process representation of sampling models, and frequentist decision theory.
When fitting a generalized linear model—such as linear regression, logistic regression, or hierarchical linear modeling—analysts often wonder how to handle missing values of the dependent variable Y. If missing values have been filled in using multiple imputation, the usual advice is to use the imputed Y values in analysis. We show, however, that using imputed Ys can add needless noise to the estimates. Better estimates can usually be obtained using a modified strategy that we call multiple imputation, then deletion (MID). Under MID, all cases are used for imputation but, following imputation, cases with imputed Y values are excluded from the analysis. When there is something wrong with the imputed Y values, MID protects the estimates from the problematic imputations. And when the imputed Y values are acceptable, MID usually offers somewhat more efficient estimates than an ordinary MI strategy.
Public policy is often concerned with the size and characteristics of special populations that are difficult to reach in household
surveys. Chronic drug users, who often live outside conventional households, provide the illustration motivating this paper.
An alternative to household surveys is to question chronic drug users where they congregate—jails, treatment programs, and
shelters, for example. Using such opportunistic data for prevalence estimation raises difficult problems for statistical inference:
Study subjects who arrive at the collection points cannot be deemed a random sample of the general population. However, if
we could estimate the rates at which chronic drug users arrive at the collection points, then we could use those estimates
to weight the sample to represent the population. This paper presents a modified Poisson mixture model used to estimate the
stochastic process that accounts for how chronic drug users get arrested. It uses that model to estimate arrest rates for
38 counties using up to sixteen quarters of data from the Arrestee Drug Abuse Monitoring survey.
A family of estimators is proposed for the relative frequency of the unobservable zero class in a truncated Poisson distribution. Large sample behavior is described by asymptotic expansion and robust properties are discussed in the presence of compound distributions. Two applications are presented involving the estimation of the number of unobserved individuals in complex capture-recapture experiments.
Injecting drug use is an important risk factor for transmission of viral hepatitis, but detailed, transparent estimates of the scale of the issue do not exist. We estimated national, regional, and global prevalence and population size for hepatitis C virus (HCV) and hepatitis B virus (HBV) in injecting drug users (IDUs).
We systematically searched for data for HBV and HCV in IDUs in peer-reviewed databases (Medline, Embase, and PsycINFO), grey literature, conference abstracts, and online resources, and made a widely distributed call for additional data. From 4386 peer-reviewed and 1019 grey literature sources, we reviewed 1125 sources in full. We extracted studies into a customised database and graded them according to their methods. We included serological reports of HCV antibodies (anti-HCV), HBV antibodies (anti-HBc), or HBV surface antigen (HBsAg) in studies of IDUs with more than 40 participants (<100% HIV-positive) and sampling frames that did not exclude participants on the basis of age or sex. With endorsed decision rules, we calculated prevalence estimates with anti-HCV and anti-HBc as proxies for exposure and HBsAg as proxy for current infection. We combined these estimates with IDU population sizes to calculate the number of IDUs with positive HBV or HCV statuses.
We located eligible reports with data for prevalence of anti-HCV in IDUs for 77 countries; midpoint prevalence estimates suggested 60-80% of IDUs had anti-HCV in 25 countries and more than 80% of IDUs did so in 12 countries. About 10.0 million (range 6.0-15.2) IDUs worldwide might be anti-HCV positive. China (1.6 million), USA (1.5 million), and Russia (1.3 million) had the largest such populations. We identified eligible HBsAg reports for 59 countries, with midpoint prevalence estimates of 5-10% in 21 countries and more than 10% in ten countries. Worldwide, we estimate 6.4 million IDUs are anti-HBc positive (2.3-9.7 million), and 1.2 million (0.3-2.7 million) are HBsAg positive.
More IDUs have anti-HCV than HIV infection, and viral hepatitis poses a key challenge to public health. Variation in the coverage and quality of existing research creates uncertainty around estimates. Improved and more complete data and reporting are needed to estimate the scale of the issue, which will inform efforts to prevent and treat HCV and HBV in IDUs.
WHO and US National Institutes of Health (NIDA R01 DA018609).
Previous reviews have examined the existence of HIV prevention, treatment, and care services for injecting drug users (IDUs) worldwide, but they did not quantify the scale of coverage. We undertook a systematic review to estimate national, regional, and global coverage of HIV services in IDUs.
We did a systematic search of peer-reviewed (Medline, BioMed Central), internet, and grey-literature databases for data published in 2004 or later. A multistage process of data requests and verification was undertaken, involving UN agencies and national experts. National data were obtained for the extent of provision of the following core interventions for IDUs: needle and syringe programmes (NSPs), opioid substitution therapy (OST) and other drug treatment, HIV testing and counselling, antiretroviral therapy (ART), and condom programmes. We calculated national, regional, and global coverage of NSPs, OST, and ART on the basis of available estimates of IDU population sizes.
By 2009, NSPs had been implemented in 82 countries and OST in 70 countries; both interventions were available in 66 countries. Regional and national coverage varied substantially. Australasia (202 needle-syringes per IDU per year) had by far the greatest rate of needle-syringe distribution; Latin America and the Caribbean (0.3 needle-syringes per IDU per year), Middle East and north Africa (0.5 needle-syringes per IDU per year), and sub-Saharan Africa (0.1 needle-syringes per IDU per year) had the lowest rates. OST coverage varied from less than or equal to one recipient per 100 IDUs in central Asia, Latin America, and sub-Saharan Africa, to very high levels in western Europe (61 recipients per 100 IDUs). The number of IDUs receiving ART varied from less than one per 100 HIV-positive IDUs (Chile, Kenya, Pakistan, Russia, and Uzbekistan) to more than 100 per 100 HIV-positive IDUs in six European countries. Worldwide, an estimated two needle-syringes (range 1-4) were distributed per IDU per month, there were eight recipients (6-12) of OST per 100 IDUs, and four IDUs (range 2-18) received ART per 100 HIV-positive IDUs.
Worldwide coverage of HIV prevention, treatment, and care services in IDU populations is very low. There is an urgent need to improve coverage of these services in this at-risk population.
UN Office on Drugs and Crime; Australian National Drug and Alcohol Research Centre, University of New South Wales; and Australian National Health and Medical Research Council.
The problem of selecting one of a number of models of different dimensions is treated by finding its Bayes solution, and evaluating the leading terms of its asymptotic expansion. These terms are a valid large-sample criterion beyond the Bayesian context, since they do not depend on the a priori distribution.
Injecting drug use is an increasingly important cause of HIV transmission in most countries worldwide. Our aim was to determine the prevalence of injecting drug use among individuals aged 15-64 years, and of HIV among people who inject drugs.
We did a systematic search of peer-reviewed (Medline, EmBase, and PubMed/BioMed Central), internet, and grey literature databases; and data requests were made to UN agencies and international experts. 11 022 documents were reviewed, graded, and catalogued by the Reference Group to the UN on HIV and Injecting Drug Use.
Injecting drug use was identified in 148 countries; data for the extent of injecting drug use was absent for many countries in Africa, the Middle East, and Latin America. The presence of HIV infection among injectors had been reported in 120 of these countries. Prevalence estimates of injecting drug use could be ascertained for 61 countries, containing 77% of the world's total population aged 15-64 years. Extrapolated estimates suggest that 15.9 million (range 11.0-21.2 million) people might inject drugs worldwide; the largest numbers of injectors were found in China, the USA, and Russia, where mid-estimates of HIV prevalence among injectors were 12%, 16%, and 37%, respectively. HIV prevalence among injecting drug users was 20-40% in five countries and over 40% in nine. We estimate that, worldwide, about 3.0 million (range 0.8-6.6 million) people who inject drugs might be HIV positive.
The number of countries in which the injection of drugs has been reported has increased over the last decade. The high prevalence of HIV among many populations of injecting drug users represents a substantial global health challenge. However, existing data are far from adequate, in both quality and quantity, particularly in view of the increasing importance of injecting drug use as a mode of HIV transmission in many regions.
This paper examines the feasibility of using information on treatment admissions to infer the size of a heroin-using population. Several variants of capture-recapture and truncated Poisson models which have been used for this purpose are assessed here, and are generally found to be unsuited to the task at hand. A cohort model is developed which is consistent with the dynamics of drug use and help-seeking behavior, and which would allow estimation of the size of the treatment susceptible or "inbound" population, but its information requirements are difficult to meet. The analysis suggests that capture-recapture and truncated Poisson models should be evaluated in terms of their potential for reformulation and application within the present context, or alternatively, that other methods be developed for this purpose.
This work was aimed at identifying differences in HIV and HCV risk behaviors among intravenous drug users (IDUs) according to the gender and their determinants.
IDUs over 18 years, having had sexual intercourse and able to answer the questionnaire were interviewed in 10 drug abuse treatment centers or social institutions using a questionnaire adapted from the one used in the survey of sexual lifestyle in France.
Over 612 eligible IDUs, 595 completed the questionnaire (women: 29%);37% had multiple partners with no difference according to the gender; 14% of the women and 7% of the men reported trading sex. More women reported inconsistent condom use (46% vs. 55%) and inconsistent clean equipment use (65% vs. 73%). A younger age, independently associated to sharing equipment (men: OR = 0.94; 95% CI = 0.90-0.99; women: OR = 0.92; 95% CI = 0.85-0.99) and inconsistent HIV serology testing, independently associated to inconsistent condom use (men: OR = 3.36; 95% CI = 2.02-5.60; women: OR = 10.72; 95% CI = 3.18-36.18), were the only risk markers common to both genders. For women, being HIV negative increased the risk of inconsistent condom use. Low educational level increased the risk of inconsistent clean equipment use. No risk marker among those analyzed was associated to having had multiple partners. For men, low socioeconomic status markers, a steady sexual partner or not living in couple and educational level were associated with sexual risk behaviors (inconsistent condom use and having multiple partners); low socioeconomic status markers were associated with injecting risk behaviors (inconsistent clean equipment use and sharing).
Women were more likely to have both sexual and injecting risk behaviors; their significant risk markers are less numerous thus harm reduction could be more difficult than for men.
Capture-recapture methods are used to estimate the incidence of a disease, using a multiple-source registry. Usually, log-linear
methods are used to estimate population size, assuming that not all sources of notification are dependent. Where there are
categorical covariates, a stratified analysis can be performed. The multinomial logit model has occasionally been used. In
this paper, the authors compare log-linear and logit models with and without covariates, and use simulated data to compare
estimates from different models. The crude estimate of population size is biased when the sources are not independent. Analyses
adjusting for covariates produce less biased estimates. In the absence of covariates, or where all covariates are categorical,
the log-linear model and the logit model are equivalent. The log-linear model cannot include continuous variables. To minimize
potential bias in estimating incidence, covariates should be included in the design and analysis of multiple-source disease
registries. Am J Epidemiol 1999;149:392–400.
To evaluate the extent to which the introduction (February 1996) of ambulatory prescriptions of buprenorphine for drug maintenance treatment (DMT) has been associated with its intravenous illicit use by French injecting drug users (IDUs).
Cross-sectional survey (September 1997), using self-administered questionnaires, in a sample of IDUs recruited at 32 pharmacies, four needle exchange programmes and three syringe vending machines.
Thirty-nine sites where IDUs have access to sterile syringes in the city of Marseille (South-Eastern France).
Sample of IDUs attending community pharmacies, vending machines and needle exchange programs to obtain equipment.
To compare characteristics of IDUs who declared that they only injected buprenorphine in the prior 6 months versus the rest of the sample.
Among the 343 respondents (response rate = 70.7%), 33.8% were polydrug users who occasionally injected buprenorphine in parallel to heroin and/or cocaine, while 23.9% only injected buprenorphine in the previous 6 months. IDUs in this latter group were younger, injected more frequently, and were more frequently on buprenorphine DMT, but they were less likely to be HIV-infected and to declare HIV-related injecting risky behaviours.
Substantial risk of injecting misuse is associated with large-scale diffusion of buprenorphine DMT. A more stringent regulation for medical dispensation of buprenorphine than the current French general freedom of prescription for all physicians, including general practiioners in ambulatory care, may be necessary in other countries which are considering the diffusion of buprenorphine DMT.
National and local monitoring of policies on illicit drug use requires information on the number of problematic drug users in a country. This article reports the findings from a study that estimated the number of problematic and injecting drug users for all Drug Action Teams (DATs) in England for 2001.
The Multiple Indicator Method (MIM) is a statistical technique for using aggregated data to estimate numbers of drug users across a large number of areas. The MIM was used to combine eight indicators available for all DATs, with prevalence estimates available from a small number of DATs. The indicators were drug possession and supply offences, arrest referrals, people recorded in drug treatment databases, methadone prescriptions, drug-related hospital episodes, drug-related deaths and DATs' Townsend score. The latter is a measure of material deprivation. A three-stage process involved, (i) factor analysis of the drug indicators, (ii) regression linking factor scores to known prevalence estimates and (iii) imputation of estimates to all other DATs.
Factor analysis yielded two statistically significant factors underlying the drug indicators in 150 DATs in England. The estimated prevalence rate of problematic drug use in the DATs varied from 0.2 to 1.5 per cent of the population. The estimated average number of problematic drug users per DAT was 1943 (standard deviation = 1300). The estimated average number of injecting drug users per DAT was 627 (standard deviation = 572). The estimates for England in 2001 were 287,670 (population rate = 0.64 per cent) problem drug users, and 93,185 (population rate = 0.23 per cent) injecting drug users.
Although the model cannot take account of specific local factors, the results are likely to be accurate in areas that do not have these idiosyncrasies. The estimated prevalence figures provide a basis for all DATs to assess their contact rates with problematic and injecting drug users.
The history of the development of statistical hypothesis testing in time series analysis is reviewed briefly and it is pointed out that the hypothesis testing procedure is not adequately defined as the procedure for statistical model identification. The classical maximum likelihood estimation procedure is reviewed and a new estimate minimum information theoretical criterion (AIC) estimate (MAICE) which is designed for the purpose of statistical identification is introduced. When there are several competing models the MAICE is defined by the model and the maximum likelihood estimates of the parameters which give the minimum of AIC defined by AIC = (-2)log-(maximum likelihood) + 2(number of independently adjusted parameters within the model). MAICE provides a versatile procedure for statistical model identification which is free from the ambiguities inherent in the application of conventional hypothesis testing procedure. The practical utility of MAICE in time series analysis is demonstrated with some numerical examples.
Prevalence of problem drug use in France: estimates for 2006 [Prévalence de l'usage problématique de drogues en France: estimations 2006]
- Costes
Patients welcomed in treatment centres. Facts in 2014 and evolution since 2007 [Les personnes accueillies dans les CSAPA - Situation en 2014 et évolution depuis 2007]
- Palle
Harm reduction facilities’ client profiles and practices in 2012 [Profils et pratiques des usagers des CAARUD en 2012]
- Cadet-Taïrou
Prevalence of problem drug use and injecting drug use in Luxembourg: a longitudinal and methodological review
- Origer
Multilevel logistic regression analysis applied to binary contraceptive prevalence data
- Kahn