Why we need research: a short history of evidence-based psychiatry

1. Why we need research:a short history of evidence-based psychiatry John Geddes Oxford Clinical Trials Unit for Mental Illness

2. Conflict of Interest Funding from: MRC, ESRC, NIHR SMRI Trial drug supplies GSK Sanofi-Aventis

3. Pre-evidence-based medicine • Ackner and Oldham • NIMH • MRC • Prien et al

4. Long-term treatment: beyond lithium?

5. Lithium

9. Falling use of lithium in USA • Between 1992-5 and 1996-9* • Lithium fell from 51% to 30% • Valproate rose from 11% to 27% Trend has continued Limited evidence for valproate Increasing evidence for lithium *Blanco et al Am J Psych 2002 1005-1010

10. Pre-evidence-based medicine • Ackner and Oldham • NIMH • MRC • Prien et al

11. Aims • To determine if combination therapy with lithium plus divalproex is superior to monotherapy with either agent • To compare lithium and divalproex monotherapy • To establish proof of concept of large, simple randomised trials in psychiatry

12. lithium or divalproex or lithium + divalproex BALANCE Active run–in Up to 8 weeks lithium + divalproex Randomized Phase 2 years Time to first intervention for mood episode

13. Drug doses • Lithium – 0.4 to 1.0 mmol • Divalproex – target 1250mg, dose established during run-in

14. Sites • 4 countries (UK, US, France, Italy) • 90% UK • 41 sites

15. Characteristics of participants • 459 patients with Bipolar Disorder, type 1entered run-in • 129 withdrew before randomisation • 30% couldn’t tolerate drugs • 30% withdrew consent for various reasons • 330 patients randomised – 110 in each group • Equal men and women • Mean age 43 years • Median 2 hospital admissions (range 0-30) • 75% previous long-term drug therapy

16. Primary Outcome – New Treatment/Hospital Admission Li+Va vs Va HR 0.59 p=0.002 Li+Va vs Li HR 0.82 p=0.27 Li vs Va HR 0.71 p=0.05

17. Subgroup analyses No substantial differences between: • Men and women • Different age groups • Most recent episode • Illness stage/severity • Site

18. Sensitivity analyses Results unaffected when: • Events occurring in first 3 months post-randomisation excluded • Events occurring when no longer on allocated treatment excluded • Adjustment by minimisation factors

21. Time to Hospital Admission

22. Time From Randomization to First Use of Additional Medication

23. Time From Randomization to First Treatment for Mania

24. Time From Randomization to First Treatment for Depression

25. Time From Randomization to Stopping Allocated Treatment

26. Secondary outcomes No differences on: • Quality of life • Deliberate self harm/parasuicide/suicide • Global functioning • Adverse events including deaths

27. Conclusions • For people with bipolar disorder for whom long-term therapy is clinically indicated, combination therapy with lithium plus divalproex is more likely to prevent relapse than monotherapy • The relative advantage is most robust compared to divalproex monotherapy – risk reduced by about 40% • This relative benefit appears to be irrespective of patient or illness characteristics and is maintained for up to two years • Efficient, cost-effective trial designs are possible in psychiatry and can yield clinically useful results

28. Numbers Needed to Treat • Combination vs. valproate 7 • Combination vs. lithium 20 • Lithium vs. valproate 10

29. Clinical Scenarios • 56 year old female, 10 episodes, currently on lithium COMBINATION • 75 year old male, 5 episodes, on valproate COMBINATION • 36 year old male, 5 episodes, currently on valproate COMBINATION • 21 year old male, no previous long-term therapy, 2 episodes COMBINATION • 21 year old female, no previous long-term therapy 2 episodes LITHIUM

30. Methodology • Open design – potential for • Ascertainment bias • Performance bias • Active run-in: • Effects on generalisability • Drug dosing • Optimize vs clinically typical

31. Acknowledgements • Thanks to all participants in the study and the clinical and administrative staff in all four countries who assisted with the trial.

32. Writing Committee: Prof John R Geddes (Chief Investigator)*; Prof Guy M. Goodwin, Dr Jennifer Rendell (Trial Manager), Prof Jean-Michel Azorin (Chief Investigator, France), Dr Andrea Cipriani (Chief Investigator, Verona, Italy) Dr Michael J Ostacher (Chief Investigator, Massachusetts, USA), Prof Richard Morriss, Nicola Alder (Statistician), Ed Juszczak (Statistician) • Trial Steering Committee: Prof Shon Lewis (Chair), Prof John R Geddes (Chief Investigator); Prof Guy Goodwin, Professor Richard Morriss, Dr Jennifer Rendell (Trial Manager) • Trial Management Group: Prof John R Geddes (Chief Investigator)*; Prof Guy Goodwin, Dr Jennifer Rendell (Trial Manager), Jane Hainsworth, Emma Van der Gucht, Christine Healey, Will Stevens, Brigid Carter, Heather Young, Ed Juszczak • Clinical Trial Service Unit: Dr Christina Davies, Prof Richard Peto • Data Monitoring and Ethics Committee: Prof Thomas RE Barnes (Chair); Dr Vivienne Curtis; Dr Tony Johnson • Trial Pharmacy: Michael Marven

36. CEQUEL....... • Comparing lamotrigine plus quetiapine with quetiapine monotherapy • MRC funded • Currently recruiting • Join us!

40. With SPaRCLe data – All Relapse

41. With SPaRCLe data – Manic Relapse

42. With SPaRCLe data – Depressive Relapse

43. Lithium: prevention of suicide Cipriani, A, Pretty H, Hawton K, and Geddes JR. Am.J.Psychiatry 162 (10):1805-1819, 2005.

44. Primary Outcome – New Treatment/HospitalAdmission Li+Va vs Va HR 0.59 p=0.002 Li+Va vs Li HR 0.82 p=0.27 Li vs Va HR 0.71 p=0.05

45. ECT

46. Atypical antipsychotics

47. Conclusion • Scepticism is appropriate • We need replication • Multiples of trials