As the government searches for models of incentivising specific outcomes in addiction treatment. Deirdre Boyd picks the brains of Keith Humphreys and Tom McLellan on research to inform policy, as well as advice on “payment by results”.

Print-friendly version:
Download AddictionToday128-Payment by Results

Until recently, care for substance-use disorders was omitted from most discussions of how to increase healthcare effectiveness. Across the Atlantic, the GW Bush administration initiated outcome improvement efforts in the early 2000s, and the Obama Administration expanded on this. Here in the UK, the coalition government is seeking models of incentivising particular outcomes in addiction treatment

Professors Tom McLellan of the University of Pennsylvania and Keith Humphreys of the University of Stanford served with the White House Office to draw up the latest US drug policy last year (see cover story of Addiction Today May/June 2010). They have thus been able to review what research reveals about improving outcomes of care for substance-use disorders – regardless of whether this activity is termed “payment by results”, “value-based purchasing”, “continuous quality improvement”, “performance contracting” or other buzzwords. So we have picked their brains for guidance to policy makers seeking to apply the findings of PbR research and pilot PbR projects across the UK this year.

WHAT ARE WE MEASURING?

What is substance-use disorder treatment supposed to deliver? If we do not know, we cannot know what results we are supposed to measure.

For example, the UK’s National Treatment Agency, headed up by a former probation officer, has emphasised that a major purpose of addiction treatment is to reduce crime. The Betty Ford Center in the US believes that the major purpose of addiction treatment is “recovery”: a voluntarily maintained lifestyle comprising abstinence, subjective wellbeing and civic responsibility. This theme has been adopted by the coalition government’s first drug strategy, released in December.

“Employers and unions might both view the primary purpose of treatment as restoring people to their work function and avoiding industrial accidents,” Humphreys adds.

“Some harm-reduction activists argue that the major purpose of treatment is to help people use drugs in a less risky fashion. The family of a violent addicted patient might wish for treatment to remove their immediate risk of danger.

“The drug- or alcohol-dependent person could want any number of things, from a dramatic change in life to the restoration of self-respect, to building healthy relationships with friends and family to gaining/regaining employment, or even just to feel well and content.”

HOW IMPARTIAL IS “EVIDENCE”?

Some scientists believe there should be a strictly ‘evidence-based policy’ in which decisions about treatment would be left to putatively impartial academics rather than being determined through politico-economic processes. “But doing research does not necessarily confer superiority or even average competence in governing,” Humphreys warns.

Addiction Today agrees in that, for too long in the UK, too many ‘scientists’ have cherrycked snippets of research which suit their purposes and ignored or misinterpreted research findings which do not suit them, particularly when it concerns pharmaceutical interests – witness the £1billion wasted on unused swine flu vaccine a year or so ago. The concept of critical thinking requires that we assess what is said and written, not take words as fact on the basis of speakers’ seeming authority.

For example, NTORS – the UK’s most expensive research into the effects of drug treatment – mixed up the outcomes of detoxification with those from residential rehabilitation. Whatever the motives, it meant that rehab outcomes were artificially lowered when averaged down to detox outcomes from statutory providers unfamiliar with drug-free outcomes. And now some of those researchers, and the politicians who heed them, say there is “no evidence that rehab works”(!).

“A second critical concept is that any policy aiming to improve the effectiveness of a system can have direct effects (usually assessed by ensuing evaluation studies) and indirect effects (that often are not),” Humphreys explained.

“A case in point is the US Veterans Health Administration which instituted a national system of measuring and incentivising particular clinical practices in a large integrated healthcare system (Francis, in press; Jha et al, 2003). A London School of Economics scholar noted that the VHA approach created a ‘league table competition’, in which more providers put more effort into becoming the best at a range of clinical practices, even ones not incentivised (Oliver, 2007, 2008).

“But changes can also include ‘gaming’, in which organisations make efforts to hide poor performance or overstate adequate or good performance (Bevan & Hamblin, 2009).”

Similar guidelines for projects in different areas can also have different results depending on the leadership in each. Leaders who convey a compelling vision for change to staff and actively engage staff and others in the change process are more likely to succeed in outcomes improvement efforts (Gollop et al 2004; Ham, Kipping, & McLeod 2003). When such leadership is lacking, initiatives can falter or fail.

PROCESS vs OUTCOME STRATEGIES

McLellan and Humphreys divide strategies into two broad classes. The first class, “process-focused quality improvement strategies”, aims to manipulate aspects of the care system to yield better patient outcomes. There is no justification for investing resources if processes do not do this (Bevan & Hamblin 2009).

The second class of strategies focuses on patients and their outcomes more directly, without specifying which clinical or organisational practices attain them. This would include, for example, paying for care on the basis of how much improvement patients make, or creating publicly-available ranking systems of success rates.

Process-focused Strategies…

Research proves substantial deficits in the quality of substance use disorder care in the US and the UK (Best & Day 2009; McLellan, Carise & Kleber 2003). These problems include an excess of paperwork, insufficient time spent with patients, demoralised staff, a lack of medically trained staff and associated services, and dysfunctional organisational dynamics. Another oft-noted problem is that care has scant research basis.

“Long-term outcomes of addicted people who receive services cannot be equated simply with the quality of those services,” Humphreys says. “As chronic conditions with powerful behavioural and environmental components, substance-use disorders are deeply influenced by life context (Moos & Finney 1983). A patient can receive high-quality care but return to an environment (eg, working in a pub, living on skid row) that results in a poor long-term outcome. Or a patient might receive low-quality care but have a major life event (eg, becoming a mother, marrying someone in recovery, landing a dream job) that produces a good long-term outcome.”

Increase credentialing. It seems logical that raising credentialling requirements for staff would improve quality of treatment and thus patient outcomes. But matters are not so simple, as Humphreys and McLellan explain in a paper to be published later this year: new credentialling policies are a weak lever for improving outcomes of substance-use disorder treatment systems.

Measure/incentivise evidence-based clinical practices. “In the US VHA, particular clinical practices were selected by national management and included in ongoing monitoring throughout the system. Facility directors, chiefs of staff and other leaders were rewarded for augmenting use of such practices, with financial and/or professional incentives,” Humphreys comments. “Clinical practices included screening all primary care patients for drinking problems, recommending smoking cessation to psychiatric patients, and retaining alcohol- and drug-dependent patients in specialty treatment for 90 days (Humphreys, Harris, & Kivlahan 2009). Other health systems also monitor patients who engage early in care, and are retained in care.

“Such incentives dramatically improved hitting process-of-care targets. The same has been found in, for example, Delaware, which financially rewarded treatment programmes that reduced the number of patients dropping out of care quickly (McLellan, Kemp, Brooks, & Carise 2008). Yet careful research shows that these measures bear at best a weak relationship to patient outcomes (Harris et al, 2007, 2009, 2010).”

A recent clinical trial also found that giving providers substantial feedback on their patients’ perceptions of the therapeutic alliance, satisfaction with care and level of current substance use had no effect on clinical performance (Crits-Cristoph et al 2010). This confirms other findings that, if there is no risk of reputational damage, information on performance is generally disregarded (Bevan & Hamblin 2009). So performance on measured clinical practices is best made public.

Improve managerial & business practices. Here, experts in business practices help treatment programmes to improve their management skills, knowledge and capacity. Examples include teaching how to pilot new procedures and make prompt use of data on their impact, engage with customers to discover their needs, reduce paperwork and bureaucratic processes, improve financial management, and create a professional and educational organisation culture.

The highest-profile efforts of this sort are the Network for the Improvement of Addiction Treatment and a follow-on project, Advancing Recovery (McCarty, Gustafson et al 2007; Capoccia et al 2010). Treatment programmes applied competitively to join these projects. One improvement strategy is the ‘walkthrough’ in which managers try to access care in their own programmes from the patient’s point of view. For many, this was an eye-opening experience in poor organisational practice (eg, phones not being answered, messages being lost, unfriendly assessment staff) which helped explain low rates of treatment entry and retention. Other management practices taught include analysis and allocation of funding and better development of a business case when seeking new funding.

“Benefits stayed after the intensive phase of organisational consultation ended, critical if this is to be a sustainable strategy,” Humphreys summarised. “But there was a competitive entry process, meaning that better-organised and more-motivated programmes with stronger leadership were over-represented. Also, although both initiatives are trying to link their implemented changes in care processes to patient outcomes, this has not yet occurred.”

Embedding substance-use care in a higher quality care network. In the Affordable Care Act of 2010 and the US National Drug Control Strategy 2010, the Obama administration pursued a policy of medicalising the care of substance-use disorders (Humphreys & McLellan, in press), as we see in the NHS. One reason was to improve quality, as US substance-use disorder care is in a clinical environment of low resources and low quality, and in a financial environment that pays for care with little consideration of effectiveness.

Through more funding for screening and brief intervention in primary-care, the US hopes to bring in medically trained individuals, financial incentives for quality, an electronic health record to track and monitor patients, and inspections and monitoring. Co-location should also help patients needing medical services (eg, psychiatric or infectious disease) to get them.

“But some advocates fear that the mainstream medical care system will take the resources for substance-use disorder care, then not provide those services,” Humphreys warns. Care integration could also lead to some addicts receiving care from providers with little knowledge of addictive disorders (ie, most doctors). “Integration thus is a promising idea in search of rigorous evidence rather than something assumed to be effective.”

Patient-focused Strategies…

A common criticism in the US and the UK is that process is overmanaged by central government regardless of whether it leads to valued outcome (Schorr 1997). Research shows how processes often did not translate into better patient outcomes, or even delivered negative effects (Wachter, Flanders, Free & Pronovost 2008). So let's turn to patient-focused strategies.

Rewarding Providers for post-treatment outcomes. In this, providers are incentivised to produce specific long-term patient outcomes.

Just before christmas, the NTA posted a web invitation to tender for six payment-by-result pilots by 20 January – a brief timescale, given the holiday period, and shorter for providers not alerted to it. The Ministry for Justice is expected to invite tenders for nine pilots, aiming for payment by results to be applied to all providers by 2015.

They are not actually pilots – which implies they are a test with a finite life – but rather ‘pathfinders’ in that they will never be completed but amended until they work. During the process, the idea is that others will watch, learn and imitate.  The NTA/Public Health will set up a new working party to control their pilots.

There are myriad problems, not least that most bidders do not know how to deliver recovery and it is unlikely the working party will.

But the proposal incurring most opprobrium is to pay addiction treatment programmes a percentage of reimbursement (eg, 50%) at intake, then pay the rest (50%) only if post-treatment follow-up shows that the patient is abstinent – which could be two years later. We reiterate the earlier point about setting outcome measurement points too late after treatment to affect it or be reliably linked. This could also lead to demoralisation and lack of effort by clinical providers, as they are held accountable for things over which they have little control.

Providers might also “cream” patients – ie, not admit people who seem unlikely to have good long-term outcomes. In parliament recently, NTA CEO Paul Hayes described NTA patient placement criteria as “triaging” people.

Also, the cost of locating and assessing patients after treatment can be high. “When this work was assigned to clinicians, it led to poor follow-up rates, less time spent treating current patients, and poor data,” Humphreys revealed. “It is better if an independent outcomes monitoring team follows up random samples of patients from programmes. This requires an economic commitment to sustain (Tiet, Byrnes, Barnett & Finney 2006).”

Rewarding providers for in-treatment performance. A different purchasing approach is to reward programmes for outcomes attained during treatment, as trialled in Maine. Programmes could get more funds by increasing the proportion of patients who achieved various outcomes by the end of their programme – eg, abstinent the past 30 days of treatment, significant reduction in substance use since treatment intake.

“Performance appeared to improve (Commons, McGuire & Riordan 1997), but later analysis showed that programmes began treating fewer severely troubled clients after the contract was in place, implying that apparent gains were illusory (Shen 2003),” Humphreys said. “Further, the Maine system was entirely based on self-report, and the incentive for a patient to not be candid could clearly be magnified if the programme would be rewarded by overstating ‘success’.”  So performance contracts should reward outcomes that are case-mix-adjusted, and outcome assessment should be biological (eg, urinalysis).

“However, of the outcome improvement approaches, we view purchasing in-treatment outcomes as among the most promising. We need more research to establish the strength of the relationship between in-treatment outcomes and longer-term ones,” Humphreys concluded.

Rewarding patients for particular outcomes. Evidence show that people with substance-use disorders respond to incentives. In the criminal-justice system, for example, programmes that put probationers and drink-driving offenders in jail for a day immediately due to a positive drug/alcohol test produce dramatic decreases in substance use (Caulkins & DuPont 2010; Hawken & Kleiman 2009). In healthcare, contingency management programmes produce substantial behavioural changes when they reward abstinence or other outcomes with money, the chance of a prize or privileges during treatment (Lussier et al 2006; Prendergast et al 2006).

But these behaviour changes often deliquesce when the reward structure is removed. So policies of this sort could inconsistently affect long-term outcomes, but be useful in early treatment.

Making the patient a customer with vouchers. During the GW Bush administration, people early in the recovery process were given vouchers with which they could buy services they thought would further their recovery. Examples included community college classes, transportation to work or to self-help group meetings, transitional housing, dental care, work training and clothes for job interviews. The services were provided by organisations certified by the state and, as in any market, there was pressure to serve the customer well because he or she could always take the voucher somewhere else.

A study of over 7,000 substance-use disorder outpatients in the State of Washington found that those with vouchers stayed in treatment longer and were more likely to be employed than patients who did not get a voucher (Krupski et al 2009). So the US plans a large budget increase for this (Humphreys & McLellan, in press).

Among the questions being asked in the UK is how to ensure that a patient will not, for example, use the voucher to flee an organisation that imposes necessary constraints (eg, not tolerating substance use in treatment) and go to a provider that will tolerate anything as long as it gets paid. The state must certify which providers are eligible for vouchers rather than  surrender care.

As part of the policy pilots contemplated by the new coalition government, it seems reasonable to include a pilot test of vouchers in the UK.

Deirdre Boyd is editor of Addiction Today and CEO of the Addiction Recovery Foundation