The analysts faced the task of estimating how much crime patients commit before entering treatment, how much this is reduced by entering 'effective' treatment, and then attaching monetary values to this crime to estimate cost savings to society. These calculations were then extended to add in longer term savings in crime-related costs from the lasting recovery of patients who successfully complete treatment.
A key source was the DTORS study which sampled patients starting treatment for drug problems in England in 2007 and 2007. In the four weeks straddling On average 15 days before and 13 after according to the featured report. treatment entry, patients on average recalled having committed just over 10 money-raising crimes. The worst case, high-crime scenario In fact, things might have been even worse because the last four weeks might have represented an atypically low level of crime. For part of this time patients had already started treatment and (The Drug Treatment Outcomes Research Study (DTORS): baseline report. Appendices. http://www.homeoffice.gov.uk/rds/pdfs07/horr03append.pdf) some were constrained in other ways. While 24% said this had been an unusually high-crime period, 44% said the opposite. If the high-crime scenario actually underestimated pre-treatment crime, the effect would have been to underestimate the reduction in crime after starting treatment and the associated cost savings. assumed by the featured report was that this level of criminality typified the entire pre-treatment year.
Whether the DTORS four-week treatment entry period really was typical of the pre-treatment year could be gauged from the conviction records of 53,851 adults recorded by the national monitoring system as starting treatment in England in 2006/07. These were the subset of all 82,935 patients starting treatment who were either absent from or could reliably be matched to police records using a system which (anonymously) identifies the same individuals cropping up in criminal and treatment datasets. Overlapping substantially the crimes recorded by DTORS, the convictions concerned were for offences which prompt police to test for drugs, specifically drug offences involving opiates or cocaine and certain acquisitive crimes thought most commonly committed to finance drug purchases.
For each four-week period in the year leading up to treatment entry, the conviction tally for offences committed in each period was at first flat, but then rose to peak The pre-treatment peak in convicted crime nevertheless represented on average just 0.088 convictions per patient. If each actually committed the same number of crimes represented in DTORS, then each conviction represented on average about 116 crimes. in the period immediately before treatment started, largely due to more people being caught rather than the same people being caught more often. For many this conviction will have precipitated their treatment entry; for others, a deteriorating drug problem leading them to seek treatment on their own initiative was associated with committing more crimes, one or more of which resulted in a new conviction.
Whether the conviction record mirrored trends in the much larger total number of crimes was unclear. Assuming it did, the final four-week DTORS figures would have to be considered atypically high; instead of typifying the pre-treatment year, the estimate for that year would have to be reduced by 40% – the lower-crime scenario. These two scenarios translated in to either about 134 crimes per patient in the pre-treatment year or about 80. The primary analyses in the report assumed the truth was mid way at 107 crimes, but the figures were also analysed on the best and worst case assumptions.
In turn these estimates were applied to the 2010/11 caseload recorded by the national monitoring system as having remained in treatment for at least 12 weeks, or who left before then having 'successfully completed'. As judged by the treatment service from which the patient last exits, this means they are no longer seen as requiring structured drug treatment, and have left treatment (not just that service, but the system as a whole) no longer dependent on any illegal drug and not using opiates or crack cocaine. (1 2). These 191,129 patients were all but 7% of the full known caseload in that financial year.
The read-across to these patients from the DTORS study sample is imperfect, because many DTORS patients were not reflected in the monitoring system, and because many who might have been never made in to the DTORS sample (because they refused, their areas or services did not cooperate with the study, or their keyworkers never asked them to contact the researchers). How imperfect is unclear, but there is no reason to believe that the crime record of the DTORS sample did not broadly apply also to the 2010/11 caseload starting or restarting treatment. However, this assumption was also made for the bulk of patients who had continued in treatment from previous years, a bigger and more uncertain step.
The assumption was that had these patients not been in effective treatment during 2010/11, they would have offended in the same way as the DTORS patients before they entered treatment in 2006–2007. This key assumption could not be evidenced. In practice it envisages a situation where treatment had been unavailable for many years, so these patients could never have started treatment, or that suddenly in 2010/11 it was made unavailable to them. The first scenario implies such a different addiction policy and health service environment that other consequences would follow, notably that many of these addicts would already have died. The second implies that withdrawal of treatment sends patients back to where they started (perhaps many years before), leaving no legacy of persisting improvements, and that aging and other changes in their lives have not moderated their propensity to commit crime in order to maintain drug consumption. The document might have been on safer ground had it confined its analysis to the 74,028 patients who, like the DTORS patients, (re)started a new episode of treatment in 2011/12. Nevertheless, the absence of a basis for an alternative assumption makes this a defensible way to construct an estimate. Assuming the mid way estimate of how much crime each pre-treatment patient commits, these calculations led to an estimate that without treatment the 2010/11 (effective) treatment caseload would have committed over 20 million crimes. Of these, the most numerous based on the DTORS offending profile would have been shoplifting followed by handling stolen property and drug dealing.
The next and critical step was to estimate to what degree treatment entry reduced the number of convictions for each type of offence. As categorised by DTORS, these offences could roughly be matched to conviction data in police records. These records were obtained for the 13,819 patients continuously in treatment for two years from when they started treatment in 2006/07. By selecting for its crime reduction estimates the record of this atypical minority of stable and long-term patients, the analysts also selected an atypically high reduction estimate. This was then applied to all patients in the analysis, almost certainly over-estimating the crime reduction benefits of typical treatment for typical patients. Details below.
According to another report based on the national addiction treatment monitoring system, two years was the average time patients in treatment for opiate and/or crack problems in 2006/07 had continuously been in treatment. However, this average is derived from a population including patients who had been in treatment for many years. It is unusual for patients starting or re-entering treatment to then remain continuously in treatment for two years. The 13,819 patients who did so were about a quarter of the 53,851 in the same dataset who (re)started treatment in the same year. A different way of analysing the same records shows that of 38,421 patients starting treatment in 2008/09 who had not previously been in treatment, 4855 remained continuously in treatment two to three years later to the end of 2010/2011. Corresponding figures for the next one to two years for 2009/10 patients were 5708 of 32,646. In 2010/11, of all 204,473 patients in treatment, 72,611 had been continuously in prescribing-based treatments for at least two years. It seems likely that few patients will have been in non-prescribing treatments for this length of time, suggesting that in any one year, not much more than 36% of all patients have a record of this longevity and stability in treatment. The implication is that under 1 in 6 new patients go on to last two years in continuous treatment, as do about a quarter of those who are either new or restarting treatment, and that just over a third of all patients in any one year have already reached this milestone. Those who do stay this long have exceptionally good crime reduction records; among known recent offenders, nearly twice the overall average. This disparity will have been mitigated slightly by the featured study's selection of patients retained for at least 12 weeks or successfully completing treatment before that time, but will remain substantial.
Among this minority of long-term stable patients, reductions After eliminating offences committed in the 30 days before starting treatment in order to adjust for the possibility that these represented an atypical peak which would have evened out even without treatment. in the number of convictions from the two years before to the two years after (re)starting treatment in 2006/07 varied across offence types from 18% to 52%, with the 20% drop in shoplifting accounting for the greatest number of crimes. These reductions were applied across the board to the 191,129 patients who in 2010/11 were in 'effective' treatment, all but 7% of the full caseload in that financial year. The assumption was made that crime was reduced only while they were in treatment, 73% of the time during that year, so the reduction estimates were shrunk by just over a quarter. These conviction reduction estimates were then applied to the estimate of the number of crimes the patients would have committed had they not been in treatment, based primarily on data from DTORS patients starting or restarting treatment. The result based on the mid way estimate was that being in treatment on average prevented 26 of the 107 crimes each patient would otherwise have committed, totalling 4.9 million prevented crimes in 2010/11, of which nearly 3 million were handling stolen property, drug dealing, or shoplifting.
Applying Home Office cost-of-crime calculations to the different types of offences which could appropriately be costed Of the major omissions, drug dealing was essentially considered a 'victimless' crime, while handling stolen property was largely accounted for by the costs of the original theft. led to an estimate that as a result of these prevented crimes, society saved £5030 per patient in effective treatment, totalling £961 million in 2010/11, of which nearly a third was due to fewer violent thefts (robbery). Extrapolating Discounting the value slightly of the later cost savings, a conventional adjustment in economics. the same caseload and the same cost savings over the next three financial years led to the estimate of about £3600 million savings over the Spending Review 2010 period from 2011/12 to 2014/15. Given the nature of the offences, the bulk of these savings reflect the value of stolen property and money and real expenditure caused to business, individuals and services by the need to arm against crime and by its consequences. Only an assumed 20% reflects the physical and emotional harm caused to the victims of crime.
Because of the way the Home Office costs crime, these calculations effectively assume that stolen/defrauded money and goods were lost to society, rather than transferred (albeit illegally) from one member of society to another. These so-called 'transfer payments' will form the bulk of the half of the cost savings attributed to individuals and businesses, yet economists have argued that from the point of view of society as a whole, losses to victims are balanced by gains to the criminals and those who benefit financially from the consequences of their crimes, for example, in being able to buy stolen goods at reduced prices.
On who reaps the benefits of the averted crimes, the report says that half would accrue to individuals and businesses, over a quarter to criminal justice system and victim services, while the NHS would enjoy just 2% of the total savings, around £20 million in 2010/11.
In 2012/13 the budget for the treatment which reaps these estimated benefits will be around £500 million, posing an immediate policy problem, in that a major treatment provider, the NHS, is a very minor beneficiary in cost-savings terms. Later work will quantify health benefits additional to the crime reduction dividend, but in the DTORS study of addiction treatment in England in 2006 and 2007, health improvements were modest, meaning that one quality-adjusted life year was saved (assuming all the assumptions and figures are valid) at a cost of £90,620, considerably worse than the £30,000 yardstick commonly cited for cost-effective medical treatment. These figures calculated for a year after treatment starts may however look considerably better if the calculation can be extended over a lifetime and take in the reduced long-term death rate associated with treatment.
Even if the savings do not accrue to the health service, the featured report calculates that society as a whole would be the loser if health and other funders withdrew support from addiction treatment.
In 2012/13 it estimates that £2604 will be spent per patient in effective treatment. To illustrate the consequences of disinvestment, the report assumes a £1 million cut resulting in 384 fewer patients, who then return to typical pre-treatment levels of crime, totalling 9863 offences each year at a cost to society of £1.8 million – a net loss of £0.8 million and apparently a false economy.
However, cuts could be made in ways which might not have this impact, notably by focusing treatment on the relatively few high-crime offenders who account for most of the crimes before treatment and most of the post-treatment entry reduction (1 2). This would risk patients who are excluded from treatment later becoming high-crime offenders, but might still look an attractive strategy, especially if services are paid in line with their crime reduction results. It is also the case that in 2010 the National Treatment Agency for Substance Misuse itself highlighted that in recent years less has been spent per patient in addiction treatment while, it claimed, outcomes have actually improved, attributed to improved productivity. If this is the case, it remains to be shown why a reduction in funding could not be catered for by further increasing productivity.
For Britain the report broke new ground in attempting to cost the benefits of treatment not over the next year or two, but over the addiction careers of the patients. The main added element to the calculations is an estimate of how much longer patients would have remained dependent on drugs and incurring social costs if they had not successfully completed treatment. In England patients typically use drugs for eight years before being recorded as starting treatment for the first time, and those who successfully complete typically do so after three years. Assuming they have recovered from their addiction, an average drugtaking career of 20 years would have been reduced by nine years.
However, not all do recover; some relapse. Relapse is likely to lead to reappearance in treatment and/or in criminal justice nets intended to identify problem drug users. According to an analysis of patients leaving treatment for drug problems in 2005/06, over the next four years, 57% who left having successfully completed avoided being officially recorded as problematic users of illegal drugs, assumed to be the recovery rate. The featured report extrapolates these trends to estimate the ten-year recovery rate at 49%. With minor adjustments, each year this proportion of the caseload is added to a cumulating total of former addicts now in recovery who it is assumed would not be were it not for treatment. The total drops after the nine years when the patients would it is assumed have recovered even without treatment.
These former patients are assumed not just to have reduced their crime tally, but to no longer be committing any drug-related crimes. Effectively this amounts to the assumption that not only are they fully recovered, but also that all the crimes (of the kind analysed by the featured report) they committed before treatment were caused by their drug use, almost certainly not the case. However, this overestimation of benefit will be countered by the neglect of the degree of crime reduction contributed by successful completers who reappear in the records, and patients who leave without successfully completing treatment. Of the latter, over the next few years 43% do not reappear in treatment or criminal justice records and may be reliably recovered, a proportion not very far behind that of the successful completers.
One otherwise major source of overestimation could be corrected for: the possibility that patients who go on to successfully complete treatment and then seem to stay out of drug-related trouble for at least ten years, were even before they started treatment less involved in crime. An analysis conducted to clarify this point found this was very much the case for patients leaving treatment in 2006/07. Those who later reappeared in the records had over the eight years before starting treatment been convicted on average just over seven times. But for patients who did not reappear, the corresponding figure was just over one. It confirmed that patients who leave and 'stay left' were even before starting treatment much less criminally active (or possibly less successful criminals and more likely to get caught). In turn this means that applying the average pre-treatment crime tally to them would greatly over-estimate its reduction, because they had less scope to cut down from what was in any event a low level of crime. After adjusting for this, the estimate was that after leaving, successful completers who avoid reappearing in the relevant databases do not commit 34 crimes they would have committed without treatment. As before, attaching Home Office cost-of-crime estimates to each type of the constituent offences yielded the cost-savings estimates in the report.
This adjustment raises the issue of whether these patients would have experienced the average 20 years addiction span without treatment. Conceivably along with a much less active (or more successful) crime career than average before treatment, might also come higher than average resources and stability and lower than average marginalisation and immersion in crime, meaning they (like some non-treatment addict groups) would have ended their dependence much sooner.
The 20 years average is itself questionable. The analysts faced the impossibility of knowing how long the addiction career would have been of addicts who seek and engage in treatment if that treatment was unavailable. Their solution was to offer estimates based not just on 20 years, but down to the point where treatment did not shorten the average career at all, just mitigate its consequences. Nevertheless, the baseline assumption of 20 years and a nine-year shortening provides the headline figures and is presumed to be the most evidence-based scenario. But the only documentary source presents this figure not as the span of a non-treatment career, but of a treatment career, one derived from studies of people who at least sought and mainly engaged with treatment – certainly populations to whom treatment was available and who generally availed themselves. The same source says that on the basis of the evidence, "it is not possible to conclude that 'treatment as normal' has a significant impact on the longevity of addiction" and argues that treatment – especially in the form of opiate substitute prescribing – may elongate careers.
Comment on these background notes Return to main entry