Improving the Economic Efficiency of Clinical Trials

1
Improving the Economic Efficiency of Clinical
Trials
2
Background

• The design of most clinical trials is dominated
  either by content specialists, usually
  clinicians, or, statisticians.
• Both groups do not necessarily consider the costs
  of the study in its design.

3
Different Perspectives

• Statisticians usually are interested in
  maximising the power of any study.
• Clinicians want to demonstrate something works.
• There should be a cost perspective.

4
Economists perspective

• There is an opportunity cost to undertaking a
  trial.
• Decisions about whether the trial should be
  placebo controlled or open the randomisation
  ratio the choice of comparator all have cost
  implications.
• The costs and benefits of these decisions need
  considering in a trials.

5
Costs in Trials

• Cost of treatments
• Research costs
• Participant costs
• Time costs - delayed trial completion results in
  costs.
• NEED to consider these costs to maximise trial
  design efficiency.

6
Economics and Trials

• Few trial designs are informed by economics.
• Usually economists are only involved in the
  design of the economic evaluation alongside the
  clinical trial rather than the trial design per
  se.

7
Placebos

• To use or not to use placebos
• Placebos blind participant, physicians and
  researchers, which reduces problems of bias
  associated with ascertainment, Hawthorne,
  preference and dilution effects.
• Use of placebos is not PRAGMATIC and can be
  costly.

8
Hidden Cost of Placebos

• Whilst placebos are inexpensive their use is NOT
• Trials that use placebos have to have a screening
  process to remove patients to whom active
  treatment is contra-indicated
• An open trial can avoid this by including them
  and not giving active treatment.
• This can lead to cost savings.

9
Including unsuitable participants

• Including participants within the active group
  and NOT giving them active treatment WILL lead to
  dilution effects.
• This could be dealt with by increasing the sample
  size to detect the smaller difference.
• This may be worthwhile IF significant cost
  savings and dilution effects small.

10
Unequal Randomisation

• Trial efficiency may be improved by allocating
  more participants to the less expensive treatment
  and more to the cheaper treatment.
• Statistical power can be maintained by increasing
  the sample size.

11
Optimum Randomisation Ratio

• The most efficient allocation ratio is calculated
  by the square root of the cost ratio of two
  treatments.
• If treatment A costs 4 X as much as treatment B
  then the optimum allocation ratio is 2.

12
Cost Savings of Unequal allocation
13
Cost Savings

• By using an unequal allocation ratio the trial
  saved about 1 million.
• Many studies do not have as dramatic cost
  difference but important savings can still be
  made.

14
Recruitment methods

• Different trial recruitment methods have
  different costs and effectiveness.
• Recruitment from clinics usually require the cost
  of a nurse or other clinician.
• Alternative may be database recruitment.

15
Database Recruitment

• Database recruitment is a potentially inexpensive
  method of recruiting to clinical trials.
• Members of a database (e.g. GP lists) are mailed
  to with trial information and asking them to take
  part.
• Relatively inexpensive clerical time can be used
  to complete most of the recruitment process.

16
Examples of two trials

• Both trials among people aged 70
• Both trials used calcium and vitamin D
  supplementation
• Both trials sought a fracture reduction
• One trial cost 1.1 million the other cost
  300,000.
• Difference was partly due to low cost design.

17
Trials

• MRC Record Trial 1.1 million factorial RCT of
  calcium with or without vitamin D for fracture
  prevention in women and men 70 attending a
  fracture clinic.
• Primary care calcium and D trial 300,000 RCT
  among women 70 with 1 risk factors for fracture.

18
RECORD Trial

• Started in 1998 aimed to recruit 6,500 women and
  men from fracture clinics
• Factorial and placebo design to test the
  hypothesis that supplementation with or without
  calcium or vitamin D prevented further fractures
• Powered to show a 20 reduction in fractures.

19
RECORD Recruitment

• Participants are recruited by research nurses.
  Potential participants have to be screened for
  eligibility, consent etc.
• This can take up to an hour per participant and
  often results in non-recruitment.
• This is a slow and EXPENSIVE process.

20
RECORD Recruitment

• Because RECORD is a placebo controlled trial ALL
  participants need to be screened to identify
  the few (1-2) who have contra-indications to the
  use of calcium supplements (e.g. recent history
  of kidney stones).

21
RECORD treatment delivery

• Because RECORD is placebo controlled ALL
  participants are mailed supplements every four
  months, which adds to cost of the trial.

22
Primary Care Trial

• Trial aimed to recruit 2855 women to demonstrate
  a 33 reduction in fracture incidence.
• Trial was an open design with controls
  receiving an information leaflet about
  osteoporosis.

23
Primary Care Recruitment

• All women registered with GPs taking part in the
  study are mailed a risk factor questionnaire,
  consent form and information.
• Response rate for participation is 7.5.
• Relatively inexpensive method of identifying
  potential participants.

24
Primary Care Recruitment

• Participants that are eligible are randomised.
• Those allocated to the intervention group are
  seen by the practice nurse and given calcium and
  D IF there are no contra-indications. If there
  are contra-indications trial participation
  continues without the supplement.

25
Dilution Effects?

• Including participants who cant start therapy
  WILL dilute the effect size.
• Small dilution is not a problem particularly if
  resources saved can be used to increase the
  sample size.

26
Primary Care Allocation Ratio

• Women in intervention group would be seen by
  nurse costing 30 per woman in nurse time plus
  recruitment cost (10) 40 in total.
• Women in control group estimated cost about 10.
• Efficient allocation was estimated at about 21.

27
Primary Care Trial Flow Chart
28
Primary Care Trial COST Flow Chart
29
RECORD Recruitment

• Trialists underestimated the numbers of eligible
  participants.
• This lead to recruitment difficulties, research
  nurses had to screen many ineligible people.

30
RECORD Solutions

• Close poor recruitment centres (some money wasted
  on minimum contracts)
• Trial needed 12 month recruitment extension, more
  funds, and did not reach target - got 5272 NOT
  6500.

31
RECORD - Cost implications

• To meet recruitment target extra funds were
  required - increased cost.
• Recruitment time was extended - time cost of
  uncertainty.

32
Primary Care Recruitment

• Trialists overestimated numbers who would take
  part (10).
• Early indications only gave a 5 recruitment
  rate.
• To meet recruitment target numbers of people
  mailed out to was doubled.

33
Primary Care Cost implications

• To increase mail out increased cost per recruited
  subject.
• This narrowed the cost differential between
  intervention and control as recruitment cost was
  a constant cost per participant.

34
Trial Budgetary Implications

• Main fixed cost of trial - trial co-ordinators
  salaries.
• Variable costs included postage costs and 30
  treatment cost.
• The need to increase mail-out expenditure led
  to the following scenarios.

35
Recruitment with fixed budget

• Increase allocation to control using saved money
  to increase mail out.
• Disadvantages will increase workload for local
  trial co-ordinators in terms of data entry and
  data management.
• On the plateau of the power curve need to
  recruit substantially more to make up loss of
  power.

36
Recruitment with small increased budget

• If budget could be increased to account for
  increased mail out costs we could maintain 21
  allocation ratio and meet target.
• Disadvantage the allocation ratio is now
  suboptimal for the overall trial budget as cost
  ratio has declined - will be getting less power
  than is optimum.

37
Recruitment with slightly larger budget increase

• If increased mail out was funded AND extra funds
  for optimal recruitment ratio (32) extra
  statistical power can be obtained for relatively
  small increased cost.
• Disadvantage - need more money.

38
Solution

• Pharmaceutical company agreed to increase funds
  for mail out and allow optimal allocation ratio
  (32) AND cover for recruitment overshoot (3400
  instead of 2855).
• Detectable difference fell from 33 to 30.

39
Allocation Ratio - lessons

• A fixed allocation ratio is unlikely to be
  correct through the lifetime of a trial.
• Should plan for an adaptive allocation and change
  ratio during recruitment if cost ratio changes.
• Budget planning should probably start with an
  inefficiently high allocation (e.g. 32 or 11)
  ratio and adapt downwards (e.g. 21) as trial
  proceeds if necessary.

40
Database Recruitment

• Recruitment from some sort of database is likely
  to be more cost effective than from clinics.
• Need to make costs as variable as possible, that
  is as closely related to recruitment (e.g. a per
  patient cost).

41
Summary of Trials Design
42
Trial Time Lines
43
Conclusions

• Looking closely at the design of a trial can lead
  to important cost savings.
• This can lead to trial designs that are faster
  and less expensive than normal designs.
• WARNING - grant referees do NOT understand
  unequal allocation and some see it as
  UNSCIENTIFIC.