Editorials represent the opinions of the authors and not necessarily those of the BMJ or BMA

EDITORIALS

For the full versions of these articles see bmj.com

The multiple sclerosis risk sharing scheme

SOVEREIGN,ISM/SPL

Despite being flawed, has had unintended beneficial consequences

ANALYSIS, p 1282

Neil Scolding Burden professor of clinical neurosciences, Department of Neurology, Frenchay Hospital, Bristol BS16 1LE [email protected] Competing interests: The author has completed the Unified Competing Interest form at www.icmje.org/coi_disclosure. pdf (available on request from the corresponding author) and declares: (1) No financial support for the submitted work from anyone other than his employer; (2) Support for educational meetings from Biogen, Schering, Teva, and Serono; (3) No spouse, partner, or children with relationships with commercial entities that might have an interest in the submitted work; and (4) No non-financial interests that may be relevant to the submitted work. Provenance and peer review: Commissioned; not externally peer reviewed. Cite this as: BMJ 2010;340:c2882 doi: 10.1136/bmj.c2882

Four linked articles discuss the UK multiple sclerosis risk sharing scheme.1‑4 The scheme was set up in 2002 to allow patients access to certain disease modifying drugs (interfer‑ ons and glatiramer) after the National Institute for Health and Clinical Excellence (NICE) recommended that these drugs should not be used. No one could say that patients with multiple sclerosis have an easy time. We offer no cure. We cannot tell patients what causes the disease. We are poor at estimating prognosis. And when in the 1990s orthodox medicine did deliver treatments (three interferon beta preparations and copolymer-1), not only were they so expensive as to be virtually unavailable in most of the United Kingdom, but the profession could not—and still cannot—agree on their true efficacy. Now Raftery and McCabe and colleagues argue that either they should be withdrawn from the NHS altogether or have their price reduced to zero.3 4 The disease modifying treatments are essentially pro‑ phylactic, and like all drugs whose aim is to prevent events (relapses) that are unpredictable and relatively uncommon (the average relapse rate in multiple sclerosis is around one a year), their efficacy is difficult to assess even in large scale trials and impossible in the real world of clinical practice. If a treated patient is relapse-free for, say, two years, is that because of the efficacy of the drug or the natural course of the disease? No one knows. To make things even more difficult, the calculations assessing whether disease modifying treatments might be affordable in the NHS were predicated on the putative effects of preventing the accumulation of disability in multiple sclerosis, not on relapse rates. To many this seemed hazard‑ ous—firstly, because only a proportion of patients (who are unidentifiable) become progressively disabled; and secondly, because the relation between relapses and disability is more complex and more indirect than many imagined.5 Reduc‑ ing the frequency of relapses does not necessarily reduce the progression of disability. Indeed, studies with powerful monoclonal antibodies show that progression can continue unabated even when relapses are abolished completely.6 In 2002, NICE judged the cost effectiveness equation ­unfavourable and ruled against provision on the NHS. But the Department of Health brokered a deal with the drug com‑ panies in the same year—the risk sharing scheme—whereby disease modifying treatments would be available within the NHS under the conditions of a large study. Treated patients would be regularly assessed and disability measured and recorded. If the drugs were more effective than the NICE pre‑ dictions, and so achieved cost effectiveness, then all would be well. If not, there would need to be a financial reckoning— payback from the drug industry to the Department of Health or reduced drug costs—to achieve “affordability” post hoc.

BMJ | 12 june 2010 | Volume 340

The first substantial assessment of the accumulation of disability within the 5000 patients who were recruited and treated has now been performed.7 It found that patients taking disease modifying treatments have acquired more disability than an untreated historical control group. Cost effectiveness and Health Technology Assessment experts now call for the “deal” to be honoured and the risk sharing scheme discontinued—presumably meaning reversion to the original negative NICE judgment and so “their withdrawal from NHS practice”—or alternatively for the drug to be pro‑ vided free by the drug companies. These may not be canny suggestions, or realistic. The reac‑ tion by patients, patient groups, charities, and the medical profession would be unbearable for any government. The UK already has the lowest prescribing rate of disease modifying treatments of any of the developed nations. Nonetheless, would they be right in principle? One surprise is that the critical articles are notably free of that healthy scepticism and caution that are normally the health economists’ stock in trade.3 4 After all, this is just one study, unblinded, unrandomised, with an historical control group, using a measure of disability that Raftery and McCabe and colleagues say (rightly) is “not fit for purpose.” Would a positive recommendation for any new treatment be based on such evidence? If these authors really do accept the results at face value, then on safety grounds should they not be stri‑ dently alerting the world to the danger that disease modifying treatments accelerate progressive disability in multiple scle‑ rosis? (It is not, after all, impossible to imagine a mechanism for this—immune activation and inflammation are important for tissue repair,8 so reducing them might impair repair.) No one really seems fearful that this result might actually be true. But Raftery and McCabe and colleagues might at least call for a more intensive comparison of these data with any of the large drug company sponsored trials of interferon or copoly‑ mer-1 that have studied disability and been published since the scheme started. What of the benefits of the risk sharing scheme? It has spawned an extremely successful infrastructure of specialist multiple sclerosis care in the UK, including some 250 mul‑ tiple sclerosis nurses. All contribute to patient care immeas‑ urably beyond the administration of disease modifying treatments, and for all people with multiple sclerosis, not just the small treated minority. Specialist care in other complex neurological diseases has looked enviously at this multiple sclerosis revolution, and followed whenever ­possible. If the scheme—expensive and flawed, as Professor Ebers points out—turns out to have been no more than a clever wooden horse, then the army of multiple sclerosis healthcare special‑ ists it delivered may make it more than worthwhile. Also, the 1255

EDITORIALS

drugs prescribed will have prevented thousands of relapses, which the health economists might acknowledge. And as Pro‑ fessor Compston stresses, it leaves a platform for introducing new treatments and executing clinical research that is second to none in the world. Meanwhile, a treatment rate in the UK of around 10-15% of patients, compared with 55-70% in the United States and 40-50% in France and Germany, suggests that we may not have strayed excessively from a sound evi‑ dence base. As for the financial reckoning, the expiry of patents beck‑ ons, and substantially cheaper interferon beta preparations are already available and being used (such as Extavia). To add a political point to this lively mix of medicine and eco‑ nomics, the workings of the market may ultimately achieve what central planners have not.

1 2 3 4 5 6 7 8

Ebers GC. Outcome measures were flawed [commentary]. BMJ 2010;340:c2693. Compston A. Scheme has benefited patients [commentary]. BMJ 2010;340:c2707. McCabe C, Chilcott J, Claxton K, Tappenden P, Cooper C, Roberts J, et al. Continuing the multiple sclerosis risk sharing scheme is unjustified. BMJ 2010;340:c1786. Raftery J. Multiple sclerosis risk sharing scheme: a costly failure. BMJ 2010;340:c1672. Degenhardt A, Ramagopalan SV, Scalfari A, Ebers GC. Clinical prognostic factors in multiple sclerosis: a natural history review. Nat Rev Neurol 2009;5:672-82. Coles AJ, Wing MG, Molyneux P, Paolillo A, Davie CM, Hale G, et al. Monoclonal antibody treatment exposes three mechanisms underlying the clinical course of multiple sclerosis. Ann Neurol 1999;46:296-304. Boggild M, Palace J, Barton P, Ben-Shlomo Y, Bregenzer T, Dobson C, et al. Multiple sclerosis risk sharing scheme: two year results of clinical cohort study with historical comparator. BMJ 2009;339:b4677. Setzu A, Lathia JD, Zhao C, Wells K, Rao MS, Ffrench-Constant C, et al. Inflammation stimulates myelination by transplanted oligodendrocyte precursor cells. Glia 2006;54:297-303.

Conflicts of interest and pandemic flu

roche

WHO must act now to restore its credibility, and Europe should legislate

FEATURE, p 1274

Fiona Godlee editor in chief, BMJ, London WC1H 9JP [email protected] Competing interests: The author has completed the Unified Competing Interest form at www.icmje.org/coi_disclosure. pdf (available on request from the corresponding author) and declares: (1) No financial support for the submitted work from anyone other than her employer; (2) No financial relationships with commercial entities that might have an interest in the submitted work; (3) No spouse, partner, or children with relationships with commercial entities that might have an interest in the submitted work; (4) FG has written articles on the challenges faced by WHO, and on the influence of the drugs industry. She is in favour of a more assertive approach to conflict of interest and supports efforts to control the influence of the drugs industry on medical research, medical education, and health policy. Provenance and peer review: Commissioned; not externally peer reviewed. Cite this as: BMJ 2010;340:c2947 doi: 10.1136/bmj.c2947 1256

The world should of course be thankful that the 2009 influ‑ enza A/H1N1 pandemic proved such a damp squib. With so many fewer lives lost than had been predicted, it almost seems ungrateful to carp about the cost. But carp we must because the cost has been huge. Some countries—notably Poland—declined to join the panic buying of vaccines and antivirals triggered when the World Health Organization declared the pandemic a year ago this week. However, countries like France and the United Kingdom who have stockpiled drugs and vaccines are now busy unpicking vac‑ cine contracts, selling unused vaccine to other countries, and sitting on huge piles of unused oseltamivir. Meanwhile drug companies have banked vast profits—$7bn (£4.8bn; €5.7bn) to $10bn from vaccines alone according to invest‑ ment bank JP Morgan.1 Given the scale of public cost and private profit, it would seem important to know that WHO’s key decisions were free from commercial influence. An investigation by the BMJ and the Bureau of Investiga‑ tive Journalism, published this week, finds that this was far from the case.2 As reported by Deborah Cohen and Philip Carter, some of the experts advising WHO on the pandemic had declarable financial ties with drug companies that were producing antivirals and influenza vaccines. As an exam‑ ple, WHO’s guidance on the use of antivirals in a pandemic was authored by an influenza expert who at the same time was receiving payments from Roche, the manufacturer of oseltamivir (Tamiflu), for consultancy work and lecturing. Although most of the experts consulted by WHO made no secret of their industry ties in other settings, WHO itself has so far declined to explain to what extent it knew about these conflicts of interest or how it managed them. This lack of transparency is compounded by the exist‑ ence of a secret “emergency committee,” which advised the director general Margaret Chan on when to declare the pandemic—a decision that triggered costly pre-established vaccine contracts around the world. Curiously, the names of the 16 committee members are known only to people within WHO. Cohen and Carter’s findings resonate with those of other

investigations, most notably an inquiry by the Council of Europe, which reports this week and is extremely critical of WHO.1 It concludes that decision making around the influ‑ enza A/H1N1 crisis has been lacking in transparency. One of its chief protagonists is Paul Flynn, a UK member of parliament and a member of the council’s Parliamentary Assembly. He and others raised concerns last year about the lack of evidence to justify the scale of the international response to H1N1 (as also covered in the BMJ in December3), and the lack of transparency around the decision making process for declaring the pandemic.1 WHO’s response to these concerns has been disappointing. Although Margaret Chan has ordered an inquiry and WHO has stressed its commitment to transparency, her office has turned down requests to clear up concerns about poten‑ tial conflicts of interest.2 And at a hearing of the Council of Europe’s Parliamentary Assembly in January, WHO denied any industry influence on the scientific advice it received.1 Such a knee jerk defence before the facts were known may come to haunt the organisation. This response is also disappointing given WHO’s track record of standing up to industry. In the late 1970s WHO sparked two iconic clashes with multinational companies over the marketing of breast milk substitutes in the devel‑ oping world and the setting up of the Essential Drugs Pro‑ gramme.4 Both issues set WHO at loggerheads with the United States where these industries had major holdings. Partly in response to WHO’s position, America withdrew contributions to WHO’s budget. More recently, in 1999, when the forced disclosure of confi‑ dential tobacco industry documents alerted WHO to possible interference in its anti-tobacco activities, its then director gen‑ eral Gro Harlem Brundtland quickly set up an independent inquiry. She then published and press released its shock‑ ing findings—of an elaborate industry funded campaign to undermine WHO—without any attempt at interference or spin.5 The report recommended that all staff, consultants, temporary advisers, and members of expert committees should be required to declare their conflicts of interest, with BMJ | 12 june 2010 | Volume 340

EDITORIALS

well enforced penalties for those who failed to do so.6 As Cohen and Carter report, WHO subsequently published in 2003 new rules on managing conflicts of interest. These recommended that people with a conflict of interest should not be involved in the part of the discussion or the piece of work affected by that interest or, in certain circumstances, that they should not participate in the relevant discussion or work at all.7 WHO seems not to have followed its own rules for the decision making around the pandemic. WHO will not be the only body to come under scrutiny for its handling of the pandemic. The coming months will see a spate of reports, from the European Commission, the European Parliament, and from national bodies including the French Senate, and the UK’s Cabinet Office. This soul searching takes place against a backdrop of hardening atti‑ tudes to conflicts of interest around the world. Last year’s report from the Institute of Medicine8 has been followed by new guidance from groups such as the World Association of Medical ­Editors9 and the American College of Chest Phy‑ sicians,10 which stress that declaration alone is no longer enough. To quote the Institute of Medicine report, “Disclo‑ sure is the essential though limited first step in identifying and responding to conflicts of interest.” The big question is what to do about the conflicts. On the basis of our own investigation and those of ­others, the answer is now inescapable. As Barbara Mintzes says in Cohen and Carter’s report, “No one should be on a commit‑ tee developing guidelines if they have links to companies that either produce a product—vaccine or drug—or a medi‑ cal device or test for a disease.” The same, and more, must apply to committees making major decisions on public health. Where entirely independent experts are hard to find, experts who are involved with industry could be consulted but should be excluded from decision making. The United States has made important progress with its Sunshine Act and other legislation. European legislation on managing conflicts of interest is long overdue. As for WHO, its credibility has been badly damaged.

Response on bmj.com “Concerning the members of the Emergency Committee that advised WHO on the pandemic, including phase changes, the names will be released when the Committee finishes its work, as has always been intended.” Margaret Chan, directorgeneral, World Health Organization ЖЖTo submit a rapid response, go to any article on bmj.com and click “respond to this article.”

­ ecovery will be fastest if it publishes its own report without R delay or defensive comment; makes public the membership and conflicts of interest of its emergency committee; and develops, commits to, and monitors stricter rules of engage‑ ment with industry that keep commercial influence away from its decision making. In a briefing at the end of last year, a spokesperson for WHO said, “Given the discrepancy between what was expected [from the pandemic] and what has happened, a search for ulterior motives on the part of WHO and its scientific advisors is under‑ standable, though without justification.”11 The implication is that, had there been a huge death toll, the process behind WHO’s decision making would not have been subject to such scrutiny. This is almost certainly true. But it does not mean that we are wrong to ask hard questions. Neither does it make the answers we have found any less troubling. And nor does it remove from WHO the urgent need to restore its credibility and public trust before the next pandemic comes along. 1

Flynn P. Social, Health and Family Affairs Committee. Parliamentary Assembly of the Council of Europe. The handling of the H1N1 pandemic: more transparency needed. 2010. http://assembly.coe.int/ CommitteeDocs/2010/20100329_MemorandumPandemie_E.pdf. 2 Cohen D, Carter P. WHO and the pandemic flu “conspiracies.” BMJ 2010;340:c2912. 3 Godlee F. We want raw data, now [Editor’s Choice]. BMJ 2009;339:b5405. 4 Godlee F. WHO in retreat: is it losing its influence? BMJ 1994;309:1491-5. 5 Godlee F. WHO faces up to its tobacco links. BMJ 2000;321:314-5. 6 WHO. Committee of Experts on Tobacco Industry Documents. Tobacco company strategies to undermine tobacco control activities at the World Health Organization. 2000. www.who.int/tobacco/resources/ publications/general/who_inquiry/en/index.htmlhttp://www.who.int/ home/reports.html. 7 WHO. Guidelines for WHO guidelines. 2003. http://whqlibdoc.who.int/ hq/2003/EIP_GPE_EQC_2003_1.pdf. 8 Institute of Medicine. Conflict of interest in medical research, education, and practice. 2009. www.iom.edu/Reports/2009/Conflict-of-Interest-inMedical-Research-Education-and-Practice.aspx. 9 World Association of Medical Editors. WAME statement on conflict of interest in peer-reviewed medical journals. 2009. www.wame.org/conflictof-interest-in-peer-reviewed-medical-journals. 10 Guyatt G, Akl EA, Hirsh J, Kearon C, Crowther M, Gutterman D, et al. The vexing problem of guidelines and conflict of interest: a potential solution. Ann Intern Med 2010;152:738-41. 11 WHO. Pandemic (H1N1) 2009 briefing note 19. 2009. www.who.int/csr/ disease/swineflu/notes/briefing_20091203/en/index.html.

Improving immunisation coverage in rural India

who

Incentives help, but not nearly enough

RESEARCH, p 1291 Cite this as: BMJ 2010;340:c2553 doi: 10.1136/bmj.c2553

Despite decades of rhetoric about improving health and two decades of economic growth, vaccination rates in India remain low. As in Ethiopia, Burkina Faso, and Afghanistan, measles vaccination rates in India are around 70%, and only 44% of children aged 1-2 years are fully immunised.1 Low vaccination rates have been alternately blamed on insuf‑ ficient public funds, poor implementation of vaccination programmes, and a general apathy towards the health of the poor. Yet, we have remarkably little evidence to help us separate problems with implementation of vaccination pro‑ grammes from design flaws that restrict take-up. Banerjee and colleagues’ linked cluster randomised trial brings together time tested methods from public health (randomised trials) with the latest thinking in economics on incentives and human behaviour to examine fundamental problems of design in the delivery of ­vaccinations.2

BMJ | 12 june 2010 | Volume 340

The authors compared two interventions in a region where vaccination rates are low. In the first intervention, vaccination camps were held in villages on a monthly basis. The second intervention also established camps, but the researchers pro‑ vided households a small food incentive (lentils worth $1; £0.66; €0.78) for every vaccination and a slightly larger incen‑ tive for children who completed the full package (plates, worth just under $2). In the control villages with no interventions, 6% (95% confidence interval 3% to 9%) of children aged 1-3 years had received the basic package of vaccinations in the end point survey. This increased to 18% (11% to 23%) in vil‑ lages that received the first intervention and to 39% (30% to 47%) in those that received the second intervention. The rela‑ tive risk of being immunised was 3.09 (1.96 to 4.21) for the first intervention versus the control and 2.16 (1.54 to 2.78) for the second intervention versus the first intervention. 1257

EDITORIALS

Jishnu Das senior economist, World Bank Main Complex, 1818 H St NW, Washington DC, USA [email protected] Competing interests: The author has completed the Unified Competing Interest form at www.icmje.org/coi_disclosure. pdf (available on request from the corresponding author) and declares: (1) No financial support for the submitted work; (2) No relationship in the past three years with companies that might have an interest in the submitted work; (3) No spouse or children with financial relationships that are relevant to the submitted work; (4) No non-financial interests that may be relevant to the submitted work Provenance and peer review: Commissioned; not externally peer reviewed.

The difference between the “camp” villages and the “camp plus incentives” villages emerged after the first two vaccina‑ tions. Households therefore seem not to be averse to immuni‑ sations but unable to maintain visits to the vaccination camps over time. The small food incentives “nudged” households into returning for repeat vaccinations, leading to a large differ‑ ence in the proportion of fully immunised children at the end of the trial, with a relative risk of vaccination of 2.16 between the first and second interventions. Because the fixed cost of setting up and manning a vaccination camp is much higher than the cost of the vaccine, it was half as expensive to fully immunise a child in the “incentive” villages relative to the “camp only” villages—$28 v $56. One interpretation of these results is that small food incen‑ tives double vaccination rates and lower immunisation costs. Given that the camp model is a standard component of the Indian government’s vaccination strategy, such food incen‑ tives should be immediately incorporated into the delivery of vaccinations because they increase vaccination rates while decreasing costs. In cost-benefit calculations, the camp with incentives approach generates an infinite return relative to the camp only strategy. But another more pessimistic view is also possible. The best implemented camp and incentives model, held in a region with low population resistance to vaccination; an established relationship with the implementing organisation; and enor‑ mous mobilisation, with health workers visiting households to educate them about vaccinations and inform them about camps, only increased the proportion of children immunised with a basic package to 39%—far short of what is needed to achieve herd immunity. Further increases using this approach face two obstacles. Firstly, increasing the size of the incentives is unlikely to have greater impact; studies of similar programmes suggest that the first dollar has the largest effect.3 4 Secondly, vaccination rates drop off so dramatically with distance that this approach will work only if camps are held regularly in every village. It is a tall order for the government to replicate the gold standard conditions that this study operated under, unless vaccination

campaigns are meant to combat high absence rates among rural nurses and public sector doctors in primary healthcare centres by inducing a greater presence of health worker in periodic vaccination camps.5 How food incentives would work even when doctors and nurses don’t go to work (so no home visits or timely regular camps) remains an open question. Under this second view, the study is crucial precisely because the effect is small. Even if the government, through some remarkable transformation, could get its doctors and nurses to work, camps with food incentives would increase vaccination rates to 38% at most and cost $27 for each fully immunised child. While part of the problem may be poor implementation by the government, these numbers suggest a fundamental design flaw in the entire camp based approach to vaccinations. This study is then a wake-up call for more experimentation and evaluation on the fundamental design of vaccination pro‑ grammes. While public health specialists have been actively engaged in such efforts around the world, economists can bring to the table an extensive understanding of incentives and behaviour.6 Combining recent advances in behavioural economics with a long established tradition in public health on experimentation and evaluation, as this study shows, could rapidly and dramatically change existing models of vaccina‑ tion delivery. Ultimately, it could also ensure that two million more Indian children live to celebrate their 5th birthday. 1 2

3 4 5 6

World Health Organization. Immunization surveillance, assessment and monitoring. www.who.int/immunization_monitoring/data/data_subject/ en/index.html. Banerjee AV, Duflo E, Glennerster R, Kothari D. Improving immunisation coverage in rural India: clustered randomised controlled evaluation of immunisation campaigns with and without incentives. BMJ 2010;340:c2220. Thornton R. The demand for and impact of learning HIV status: evidence from a field experiment. Am Econ Rev 2008;98:1829-63. Filmer D, Schady N. Are there diminishing returns to transfer size in conditional cash transfers? Policy Research Working Paper Series 4999. World Bank; 2009. Chaudhury N, Hammer J, Kremer M, Muralidharan K, Rogers FH. Missing in action: teacher and health worker absence in developing countries. J Econ Perspect 2006;20:91-116. Shea B, Neil A, David H. Increasing the demand for childhood vaccination in developing countries: a systematic review. BMC Int Health Hum Rights 2009;9(suppl I):S5.

High frequency oscillation in acute lung injury and ARDS Definitive evidence about efficacy and quality of life is awaited rESEARCH, p 1290 Cite this as: BMJ 2010;340:c2315 doi: 10.1136/bmj.c2315

1258

Acute lung injury is a life threatening form of acute respira‑ tory failure with a variety of pulmonary causes (such as pneu‑ monia) or extrapulmonary causes (such as sepsis). Acute respiratory distress syndrome (ARDS) is the most severely hypoxaemic form of acute lung injury. Both acute lung injury and ARDS are common,1 and they are associated with sub‑ stantial short term mortality,2 with some survivors developing long term morbidity.3 Mechanical ventilation (with limited tidal volumes and airway pressures) is the cornerstone of treatment for acute lung injury, because no effective drug treatments are avail‑ able.4 However, mechanical ventilation itself can potentiate lung injury through a variety of mechanisms (for example, overdistension of healthy alveoli and repetitive alveolar open‑ ing or collapsing). High frequency oscillation is an alternative

ventilatory mode that incorporates the main tenets of “lung protection” to mitigate iatrogenic lung injury—namely, sus‑ tained recruitment of collapsed lung using a relatively con‑ stant airway pressure, combined with the use of very small, often sub-dead space, tidal volumes.5 In the linked study, Sud and colleagues report a meta-anal‑ ysis of the effect of high frequency oscillation on physiologi‑ cal and clinical outcomes in patients with acute lung injury.6 The pooled analysis from eight randomised controlled trials found that this method improved short term oxygenation, significantly reduced mortality (relative risk 0.77, 95% con‑ fidence interval 0.61 to 0.98), and reduced treatment failure (for example, refractory hypoxaemia and pneumothorax; 0.67, 0.46 to 0.99). Although methodologically sound, the meta-analysis is BMJ | 12 june 2010 | Volume 340

EDITORIALS

limited by the small number and size of the studies and their qualitative heterogeneity, including the lack of pressure lim‑ ited ventilation and volume limited ventilation in most control patients. These limitations temper the conclusions that can be offered to clinicians on the use of high frequency oscillation in these patients—the evidence is encouraging but does not support routine adoption at this time. Thus, this meta-analysis may be more useful in providing a glimpse into opportunities for future research rather than as a synthesis of high quality evidence that can inform clinicians on best practice. So how might the results of this study inform the design of future clini‑ cal trials of this technique in patients with acute lung injury? Firstly, an efficacy trial of this technique should study patients who are most likely to respond to this intervention. Similar to the use of higher positive end expiratory pressure (PEEP) and prone positioning in patients with acute lung injury,7 8 the benefit of high frequency oscillation over con‑ ventional mechanical ventilation in the current study was seen in patients with greater hypoxaemia (those with ARDS). Thus, such a trial should exclusively enrol patients with severe hypoxaemia and exclude those with milder oxygena‑ tion defects, where the risks of high frequency oscillation may outweigh any potential benefits. Secondly, the pooled treatment effect of high frequency oscillation on mortality from this meta-analysis may serve as a starting point for sample size considerations in future trials. To detect a treatment effect of this size (a 23% relative risk reduction), a trial would need to enrol more than 1100 patients (assuming a baseline mortality of 36%,2 with 80% power and two sided α of 0.05), which would be the largest evaluation of any ventilatory strategy in patients with acute lung injury to date. However, the pooled effect is probably an overestimate owing to bias in favour of high frequency oscil‑ lation in trials that did not use pressure and volume limited ventilation in the control group; this suggests that the number of study participants needed for an adequately powered study would be even higher. Finally, survivors of acute lung injury often have sub‑ stantially and persistently impaired physical and mental health, which reduces quality of life,9 but no randomised controlled trials of interventions during acute lung injury have assessed their effect on subsequent quality of life.10 Patients receiving high frequency oscillation may be at risk for poor quality of life because large doses of sedatives and analgesics are needed. Thus, any benefits of this treatment

Eddy Fan instructor, Division of Pulmonary and Critical Care Medicine, Johns Hopkins University, Baltimore, MD 21287, USA [email protected] Gordon D Rubenfeld professor, Interdepartmental Division of Critical Care Medicine, University of Toronto, Toronto, Canada M4N 3M5 Competing interests: The authors have completed the Unified Competing Interest form at www.icmje.org/coi_disclosure. pdf (available on request from the corresponding author) and declare: (1) No financial support for the submitted work from anyone other than their employer; (2) No financial relationships with commercial entities that might have an interest in the submitted work; (3) No spouses, partners, or children with relationships with commercial entities that might have an interest in the submitted work; (4) No non-financial interests that may be relevant to the submitted work. Provenance and peer review: Commissioned; not externally peer reviewed.

may be offset by an increased incidence of acute brain dys‑ function (delirium or coma) from heavy sedation, which may lead to poorer short term outcomes (such as mortality and length of stay)11 and predispose patients to neurocognitive and ­neuropsychological morbidity.12 Fortunately, more definitive answers may be forthcoming from two ongoing randomised controlled trials: the Oscilla‑ tion for Acute Respiratory Distress Syndrome Treated Early (OSCILLATE) trial (ISRCTN87124254; planned N=1200), and the High Frequency Oscillation in ARDS (OSCAR) trial (ISRCTN10416500; planned N=1006). In addition to physio‑ logical and mortality outcomes, quality of life will be assessed in survivors from both studies, as well as cognitive function at one year after randomisation in the OSCAR trial. With this information, clinicians will be better equipped to choose the ventilation strategy that will enhance survival in patients with acute lung injury, as well as to inform patients and their ­caregivers about what that survival may entail long after they have left the intensive care unit. 1

Rubenfeld GD, Caldwell E, Peabody E, Weaver J, Martin DP, Neff M, et al. Incidence and outcomes after acute lung injury. N Engl J Med 2005;353:1685-93. 2 Phua J, Badia JR, Adhikari NK, Friedrich JO, Fowler RA, Singh JM, et al. Has mortality from acute respiratory distress syndrome decreased over time? A systematic review. Am J Respir Crit Care Med 2009;179:220-7. 3 Herridge MS, Cheung AM, Tansey CM, Matte-Martyn A, Diaz-Granados N, Al-Saidi F, et al. One-year outcomes in survivors of the acute respiratory distress syndrome. N Engl J Med 2003;348:683-93. 4 Fan E, Needham DM, Stewart TE. Ventilatory management of acute lung injury and acute respiratory distress syndome. JAMA 2005;294:2889-96. 5 Fan E, Stewart TE. New modalities of mechanical ventilation: highfrequency oscillatory ventilation and airway pressure release ventilation. Clin Chest Med 2006;27:615-25. 6 Sud S, Sud M, Friedrich JO, Meade MO, Ferguson ND, Wunsch H, et al. High frequency oscillation in acute lung injury and acute respiratory distress syndrome (ARDS): systematic review and meta-analysis. BMJ 2010;340:c2327. 7 Briel M, Meade M, Mercat A, Brower RG, Talmor D, Walter SD, et al. Higher vs lower positive end-expiratory pressure in patients with acute lung injury and acute respiratory distress syndrome: systematic review and metaanalysis. JAMA 2010;303:865-73. 8 Sud S, Friedrich JO, Taccone P, Polli F, Adhikari NK, Latini R, et al. Prone ventilation reduces mortality in patients with acute respiratory failure and severe hypoxemia: systematic review and meta-analysis. Intensive Care Med 2010;36:585-99. 9 Dowdy DW, Eid MP, Dennison CR, Mendez-Tellez PA, Herridge MS, Guallar E, et al. Quality of life after acute respiratory distress syndrome: a metaanalysis. Intensive Care Med 2006;32:1115-24. 10 Spragg RG, Bernard GR, Checkley W, Curtis JR, Gajic O, Guyatt G, et al. Beyond mortality: future clinical research in acute lung injury. An NHLBI workshop report. Am J Respir Crit Care Med 2010; http://ajrccm. atsjournals.org/cgi/reprint/201001-0024WSv1. 11 Ouimet S, Kavanagh BP, Gottfried SB, Skrobik Y. Incidence, risk factors and consequences of ICU delirium. Intensive Care Med 2007;33:66-73. 12 Griffiths RD, Jones C. Delirium, cognitive dysfunction and posttraumatic stress disorder. Curr Opin Anaesthesiol 2007;20:124-9.

Which tool is best for colorectal cancer screening? Flexible sigmoidoscopy shows promise, but randomised data are needed

dr larpent/grehgep/spl

Cite this as: BMJ 2010;340:c2831 doi: 10.1136/bmj.c2831

Colorectal cancer screening programmes have now been introduced or are about to be launched in many European countries, Australia, and New Zealand. In the United States, screening for colorectal cancer has long been actively pro‑ moted by the medical community and patient organisa‑ tions. Evidence from randomised controlled trials has been an absolute requirement for introducing new techniques in many fields of medicine, and screening is no exception.1 Until recently, the only screening tool that has been proved to be

BMJ | 12 june 2010 | Volume 340

effective in reducing mortality from colorectal cancer is faecal occult blood testing.2 The results of the UK Flexible Sigmoidoscopy Screen‑ ing Trial were published recently.3 This multicentre ran‑ domised controlled trial investigated the effect of one off sigmoidoscopy screening on the incidence of and mortal‑ ity from ­colorectal cancer. More than 170 000 people aged 55-64 years were randomised to flexible sigmoidoscopy screening or no screening. The screening took place at 14 1259

EDITORIALS

Michael Bretthauer head, Centre for Colorectal Cancer Screening, The Cancer Registry of Norway, 0304 Oslo, Norway michael.bretthauer@ rikshospitalet.no Competing interests: The author has completed the Unified Competing Interest form at www.icmje.org/coi_disclosure. pdf (available on request from the corresponding author) and declares: (1) No financial support for the submitted work from anyone other than his employer; (2) No financial relationships with commercial entities that might have an interest in the submitted work; (3) No spouse, partner, or children with relationships with commercial entities that might have an interest in the submitted work; (4) The following nonfinancial interests that may be relevant to the submitted work: the author is principal investigator of the Nordic-European Initiative on Colorectal Cancer (NordICC), an ongoing randomised trial on colonoscopy screening, and coinvestigator of the Norwegian Colorectal Cancer Prevention trial, an ongoing trial on flexible sigmoidoscopy screening. Provenance and peer review: Commissioned; not externally peer reviewed. Cite this as: BMJ 2010;340:c2831 doi: 10.1136/bmj.c2831

centres ­throughout the United Kingdom between 1994 and 1999, and 71% of people who were invited attended the screening. In the intention to treat analysis, after 11 years of follow-up people invited to screening had a significantly reduced incidence of colorectal cancer (absolute difference 35 cases; hazard ratio 0.77, 95% confidence interval 0.70 to 0.84) and mortality from colorectal cancer (absolute differ‑ ence 14 deaths; 0.69, 0.59 to 0.82). In people who actually attended the screening (per protocol analysis) the incidence was reduced further (absolute difference 49 cases of color‑ ectal cancer; 0.67, 0.60 to 0.76), as was mortality (absolute difference 19 deaths from colorectal cancer; 0.57, 0.45 to 0.72). The effect was apparent in both men and women. The UK trial illustrates the value of long term publicly funded medical research. The study was designed in the early 1990s, and the main results are available almost 20 years later. Many people argue that medicine is developing so rapidly that a trial of this duration would be outdated by the time the results are available. This landmark study shows that this is a false assumption. It is important that large funding organisations like the NHS, the European Union, and others support long term clinical trials that tackle important health problems beyond the often short term scope of industry funded medical research. Some countries already recommended endoscopic screen‑ ing when the UK trial was in its recruitment phase. Endoscopic screening has been adopted for many years in the US despite the lack of randomised trials. During the past couple of years, such screening has been viewed more critically because of the lack of evidence for efficacy and effectiveness in randomised trials.4 5 This clearly indicates that high quality randomised trials must underpin any cancer screening programme. The Norwegian equivalent of the UK trial (NORCCAP, Norwegian Colorectal Cancer Prevention), published in the BMJ in 2009, found no significant effect on the incidence of colorectal cancer.6 Mortality from colorectal cancer was not significantly different in the intention to screen analysis but was significantly reduced in people who attended screening. The most likely reason for the lack of effect on incidence is the short follow-up of seven years, compared with 11 years in the UK trial.6 Reassuringly, the data on mortality in the two trials are comparable. Yearly hazard rates are useful for understanding the dynamic process that occurs over time after a screening intervention. The yearly hazard for the incidence of distal colorectal cancer in the UK trial shows a sustained reduction in the incidence after screening over the 11 year follow-up.3 Colorectal cancer screening guidelines usually recommend flexible sigmoidoscopy with a five year screening interval.7 In light of the UK trial, longer screening intervals should be recommended. Further follow-up of the trial would provide

Characteristics of commonly used colorectal cancer screening tests Test Faecal occult blood testing Flexible sigmoidoscopy Colonoscopy

CRC incidence CRC mortality Screening reduction* reduction* interval None 15% Short (yearly or biennially) 23% 27-31% Long (every 5-10 years) Unknown Unknown Long (at least 10 years)

*Figures for intention to screen analyses observed in randomised trials.

1260

Invasiveness and preparation None Invasive; enema bowel cleansing Invasive; oral bowel cleansing

insight into the incidence beyond 11 years. However, this may be challenging because of the introduction of the NHS bowel cancer screening programme, which may lead to gradual contamination of the control group. The observed effect on colorectal cancer incidence and mortality in the UK trial resulted from an effect on cancers in the distal colon. Incidence and mortality for proximal cancers were not reduced.3 This can be explained by the high thresh‑ old for follow-up of distal screen detected lesions using colon‑ oscopy in the UK trial. The three other large scale randomised trials still in progress (in Italy, Norway, and the US) have lower thresholds for colonoscopy.6 8 9 Thus, in these trials, an effect of screening may be expected also for the proximal colon. The table shows efficacies for the three most commonly used CRC screening tests. Flexible sigmoidoscopy is more effective in reducing colorectal cancer mortality than faecal occult blood testing and has a profound effect on incidence, which faecal occult blood testing does not. Another advan‑ tage of flexible sigmoidoscopy is the long interval between screening. An obvious disadvantage is the invasiveness and the need for bowel cleansing. No data from randomised ­trials on colonoscopy screening are available yet, but two randomised trials are in progress.10 11 The UK trial provides valid and robust evidence for the effi‑ cacy of flexible sigmoidoscopy screening. The effectiveness of such screening in the general population is still uncertain, however, because the UK trial excluded people who did not explicitly express their wish to be randomised. The NORCCAP trial is the only study of flexible sigmoidoscopy screening that is truly population based and will provide an estimate for effectiveness after 10 years of follow-up in 2013.6 How‑ ever, compliance with screening and preferences for differ‑ ent screening tests differ between populations. Therefore, flexible sigmoidoscopy should be introduced into existing screening programmes in a randomised fashion, enabling head to head comparison with the standard screening test used. This is the only way to obtain valid effectiveness data in particular populations. 1

Bretthauer M. The capsule and colorectal cancer screening—the heart of the matter. N Engl J Med 2009;361:300-1. 2 Hewitson P, Glasziou P, Watson E, Towler B, Irwig L. Cochrane systematic review of colorectal cancer screening using the fecal occult blood test (hemoccult): an update. Am J Gastroenterol 2008;103:1541-9. 3 Atkin WS, Edwards R, Kralj-Hans I, Wooldrage K, Hart AR, Northover JM, et al. Once-only flexible sigmoidoscopy screening in prevention of colorectal cancer: a multicentre randomised controlled trial. Lancet 2010;375:162433. 4 Ransohoff DF. How much does colonoscopy reduce colon cancer mortality? Ann Intern Med 2009;150:50-2. 5 Imperiale TF, Ransohoff DF. Understanding differences in the guidelines for colorectal cancer screening. Gastroenterology 2010;138:1642-7. 6 Hoff G, Grotmol T, Skovlund E, Bretthauer M; for the Norwegian colorectal cancer prevention group. Risk of colorectal cancer seven years after flexible sigmoidoscopy screening: randomised trial. BMJ 2009;338;b1846. 7 US Preventive Services Task Force. Screening for colorectal cancer: US Preventive Services Task Force recommendation statement. Ann Intern Med 2008;149:627-37. 8 Segnan N, Senore C, Andreoni B, Aste H, Bonelli L, Crosta C, et al; SCORE Working Group. Baseline findings of the Italian multicentre randomised controlled trial of “once-only sigmoidoscopy.” J Natl Cancer Inst 2002;94:1763-72. 9 Weissfeld J, Schoen R, Pinsky P, Bresalier RS, Church T, Yurgalevitch S, et al; PLCO Project Team. Flexible sigmoidoscopy in the PLCO cancer screening trial: results from the baseline screening examination of a randomized trial. J Natl Cancer Inst 2005;97:989-97. 10 Northern-European Initiative on Colorectal Cancer (NordICC). ClinicalTrials. gov. http://clinicaltrials.gov/ct2/show/NCT00883792. 11 Colorectal Cancer Screening in Average-Risk Population: Immunochemical Fecal Occult Blood Testing Versus Colonoscopy. ClinicalTrials.gov. www. clinicaltrials.gov/ct2/show/NCT00906997 BMJ | 12 june 2010 | Volume 340