57 Therefore, at rest, the claim is consistent with the results of the study.
58 As the study was conducted in an acceptable manner, I am confident that the study may be relied upon to substantiate the finding.
59 For completeness, I note that there is a reference in the abstract of the Merry Study to "pain relief". I consider that, as there is no baseline pain measurement in the Merry Study, the term pain "relief" is inappropriate, as this term has a technical meaning in the field of clinical trials which measure pain. For the Merry Study to draw a conclusion regarding pain "relief", it would need to have included either a specific question relating to pain relief, or measured a pain intensity difference from baseline, both of which would need the patient to be experiencing pain prior to starting study treatment. However, I consider that the Merry Study does show that Maxigesic provides a larger reduction in pain intensity as set out in its conclusions.
238 In her report entitled "Review of affidavit of Professor Christie sworn 20 October 2017", Dr Solterbeck said, relevantly:
47 At paragraph 98, Professor Christie states that patients who required rescue analgesia "should have been deemed a treatment failure and excluded from the study altogether" with the result that their data would not be analysed. I disagree.
48 The appropriate method for dealing with patients in clinical analgesic trials who require additional analgesia is a contentious issue, and is the subject of considerable discussion in the present statistical literature. The ICH is currently preparing an addendum to the E9 Guidelines, the Addendum on Estimands and Sensitivity Analysis in Clinical Trials to the Guideline on Statistical Principles for Clinical Trials (ICH E9 R1).
49 The approach which Professor Christie suggests (excluding all data from patients who require rescue analgesia) introduces an unacceptable risk of bias into a study. This is because it would exclude all patients who require additional analgesia from the study, leaving only those patients for whom the relevant treatments were a success. Consider the use of a placebo - under Professor Christie's approach, all those patients for whom the placebo was ineffective would be excluded, leaving the appearance of a potentially very effective placebo.
50 The ICH E9 Guidelines state that in superiority trials the appropriate approach to analysing a study's data set is the "intention to treat (ITT) or "full analysis set (FAS) approach, as it "avoids over-optimistic estimates of efficacy" (page 24). ICH E9 R1 expressly notes (page 7) that it may be appropriate to include results observed after a patient has taken rescue medication. The ITT approach involves analysing all data recorded by patients participating in the trial, regardless of whether these patients complete the trial or experience unusual events. In this way, the ITT approach more closely mirrors the range of events that may happen in the "real world", and a treatment that is shown to be superior by this approach is likely to demonstrate this result when applied more generally to patients.
51 Indeed, this approach is usually regarded as conservative, as it tends to under-rate the efficacy of the treatment being examined. In this way, it introduces minimal bias into the reported results. This is because patients on the less effective treatment (in this case, reasonably predicted to be paracetamol and ibuprofen) will have greater recourse to rescue medication, and accordingly will experience greater reductions in pain levels than they otherwise would, which will under-estimate the efficacy of the more effective treatment. I note that in the Merry Study potentially more patients in the monotherapy arms required rescue medication than patients in the combination arm (as set out in Table 4), however we can only fully assess the impact of rescue medication if we know the amount of rescue medication taken as well as the number of patients (as only the number of patients is reported In Table 4).
52 The decision in the Merry Study to include results from those patients requiring rescue medication is therefore an acceptable approach.
239 In a report entitled "Review of affidavit of Professor Christie sworn 21 November 2017, at paras 8 to 16, Dr Solterbeck expressed additional views concerning the recording of results after use of rescue medication. In particular, Dr S Solterbeck stated:
(1) The Merry study adopted her preferred approach for dealing with data after the use of rescue medication.
(2) Where all data from patients has been included in the primary analysis, the data concerning the use of rescue medication provides a useful insight into the efficacy of the treatments.
(3) All available data suggests that the use of rescue medication did not bias the results of the Merry study.
240 In re-examination, Dr Solterbeck gave the following evidence:
A. … [W]e're doing an ITT analysis, as I've just described. The ITT analysis is about what answers did I get for the patients, irrespective of whether they took the study medication and, here, whether they took rescue medication. And the things that make me comfortable about that is I don't just take that on face value, I actually went and looked and asked myself the question is there similar usage of rescue medication across the three groups? Because if there was a lot more use in the Maxigesic group, for argument's sake, then there would be an issue. But, here, there is similar usage across the three groups, so, therefore, that's not introducing any particular bias in the answer that I'm getting. I also went and asked, okay, how much information do I have to get the AUC - area under the curve? Because, for some patients, I only had a few pain scores recorded over a short timeframe, and for others I had much more. Then, again, that might be introducing a bias that I'm not comfortable with, so I asked for the data, and that was this data sheet that I've just been taken through and looking at the various patients. And on the basis of looking at that data sheet, I, then, was quite comfortable that for most patients, actually, they had done a good job of recording their pain, and there was, certainly, no difference between the three different groups about how many scores went into the AUC and the timeframe over which those values were recorded. So given those two pieces of information, I am comfortable that the study is robust and that, within that study, I can say that Maxigesic was superior to the individual components.
Q. It was put to you, Dr Solterbeck, that the results of the Merry study were unreliable because the high use of rescue medication had contaminated the study; do you recall ---?
A. I do recall that.
Q. You said you disagreed; that's right, isn't it?
A. That's correct.
Q. Can you please explain, as briefly as you can, why you disagree?
A. Yes. All right. So the reason why I disagree - because the use of rescue medication was similar across the three groups, we've not introduced any bias in answering the question "was Maxigesic better than ibuprofen or paracetamol alone?" If I wanted to get a pure answer of exactly what could Maxigesic do with nothing else happening for the patient, then I don't get that answer, but that's not the question that has been asked. What they're looking for here is how effective is Maxigesic as a treatment policy - as a strategy - compared with a strategy with paracetamol alone or a strategy with ibuprofen alone, and so given that that's the question that they asked, accepting that, in clinical practice, patients aren't perfect and they are going to need to go and do other things, they're not going to take the doses exactly every six hours but, every way that we've looked at these data, we're not introducing any bias because the use of rescue was similar, so I feel "contamination" is not necessarily the right word. It's not - we're not getting the pure answer exactly - what are these treatments doing for pain - but that's not what we want to know. We want to know, in this treatment policy, is Maxigesic better and the study is telling us, yes, it is.
Q. And this treatment policy which includes the taking of rescue medication as required?
A. That's correct.
(emphasis added)
241 In my view, this evidence is inconsistent with a conclusion that the Merry study provides a scientific foundation for representation (4), because that representation is not qualified by reference to the treatment policy applied in the Merry study. Accordingly, I understood this evidence to have qualified the evidence that Dr Solterbeck had earlier given, referred to at [237] above.