Sampling Methodology
60 The Final Report discloses that the Committee had regard to one class of services provided by Dr Carrick, namely MBS item 30487 provided in association with one or more gastrointestinal endoscopic procedural items. It examined this class of services by employing a sample of the services included in the class as authorised by s 106K(1) of the Act. Dr Carrick contends that the sampling methodology used by the Committee for this purpose was not a sampling methodology specified in a determination made by the Minister under s 106K(3).
61 Section 106K relevantly provides:
'(1) The Committee may, in respect of conduct in connection with rendering or initiating the services included in a particular class of the referred services, have regard only to a sample of the services included in the class.
(2) If the Committee finds that conduct in connection with rendering or initiating all, or a proportion, of the services included in the sample constituted engaging in inappropriate practice, then, the conduct of the person under review, in connection with rendering or initiating all, or that proportion, as the case may be, of the services included in the class from which the sample is chosen, is taken, for the purposes of this Part, to have constituted engaging in inappropriate practice.
(3) The Minister may make written determinations specifying the content and form of sampling methodologies that may be used by Committees for the purposes of subsection (1).
…'
62 The sampling methodology purportedly used by the Committee was the methodology specified by the Health Insurance (Professional Services Review - Sampling Methodology) Determination 2000 (No 1) ('the Determination').
63 Section 6 of the Determination provides:
'In having regard, under subsection 106K(1) of the Act, only to a sample of the services included in a particular class of referred services, a Committee must ensure that the sample (the preliminary random sample) is a random sample.'
64 Section 7(1) of the Determination is concerned with the appropriate sample size. It provides:
'The preliminary random sample must be of an appropriate sample size to enable prediction with 95% confidence that the percentage of services constituting inappropriate practice worked out from the sample is within +10% of the actual percentage of inappropriate services rendered or initiated in the class of referred services sampled, at a hypothesised incidence of inappropriate practice of 50% of services rendered or initiated (the incidence requiring the largest sample).'
65 Section 8 of the Determination provides:
'In making a finding based on statistical sampling, the Committee must:
(a) examine a sample, preferably of 30 or more services … (the exploratory sample), randomly drawn from the preliminary random sample; and
(b) determine whether or not each of those services constitutes inappropriate practice.'
66 As I understand Dr Carrick's submissions concerning the sampling methodology adopted by the Committee, she does not contend that the Committee breached any express requirement of the Determination. Rather she argues that it is necessarily implicit in the Determination that the Committee was required:
'to find for itself and inspect a fresh pool of randomly selected medical services to examine, instead of examining a set of 30 from the 40 provided to them by the Director.'
Again, as I understand it, this is a re-run of an argument unsuccessfully advanced before Tamberlin J in Phan [2007] FCA 269.
67 The evidence demonstrates that on 2 July 2002 the Director requested the Commission to select a random sample of 98 patient services from the total population of MBS item 30487 services rendered by Dr Carrick for the period 1 July 2000 ‑ 30 June 2001. It is accepted that the Commission used a computer program to create a report which listed the names of 98 patients who had received a MBS item 30487 service from Dr Carrick randomly selected from the total population as required. The report discloses that it was produced on 29 July 2002 by the Analysis & Evaluation Section of the Program Review Division of the Commission.
68 It is not disputed that the preliminary sample size of 98 patient services met the requirements of s 7 of the Determination. Nor is it disputed that before providing the list of randomly selected patient services to the Director, the Commission used the same computer program to randomly assign a new random number to each of the services in the preliminary random sample. It then re‑sorted the services in the order corresponding to the new random numbers assigned to each service. The list of randomly selected services provided by the Commission to the Director was the list re‑sorted in that order.
69 The exploratory sample which the Committee examined consisted of the first 30 services on the list of randomly selected services provided by the Commission to the Director. I accept the evidence of Professor Desmond Nicholls, which was not challenged, that these first 30 services were a sample selected randomly from the preliminary random sample as required by s 8 of the Determination.
70 The Director, by a written notice given under s 89B(2) of the Act, required Dr Carrick to produce to him original patient records for the first 40 patients listed on the random sample of 98 patient services provided to the Director by the Commission. The Director engaged Dr Heap to review the 40 medical records provided by Dr Carrick. Dr Heap reported that he did not believe that it was justified for Dr Carrick to charge any of the 40 patients for an MBS item 30487 service. The Director also obtained data which showed that Australia wide 7,800 claims were made for MBS item 30487 services during the referral period and that Dr Carrick provided 4,073 of these services. The next highest provider claimed 876 MBS item 30487 services.
71 Having concluded that Dr Carrick may have engaged in inappropriate practice as defined by s 82 of the Act, the Director decided to make the Adjudicative Referral. It appears that the Director provided to the Committee the names of the first 40 patients listed on the random sample of 98 patient services provided to him by the Commission together with the patient records for those patients as provided by Dr Carrick. That is, the Director provided to the Committee the same medical records that he had earlier provided to Dr Heap.
72 The Committee considered Dr Carrick's conduct in respect of the first 30 services on the list of 40 services that had been considered by Dr Heap. At [19] of the Final Report the Committee stated that it:
'considered Dr Carrick's conduct in respect of the 30 services contained in the final random sample (drawn from a preliminary random sample of 98 taken from a total of 4062 MBS item 30487 services provided in association with one or more gastrointestinal endoscopic procedural items by Dr Carrick during the review period) which were examined during the hearing.' (footnote omitted)
73 The Committee found that Dr Carrick's conduct would be unacceptable to the general body of medical practitioners in connection with 28 of the 30 MBS item 30487 services examined by it. It concluded that in this circumstance the Determination authorised it to extrapolate its findings to the overall class of services in the manner prescribed by s 11 of the Determination.
74 The first error claimed to affect the methodology adopted by the Committee was that it failed to 'ensure' that the preliminary random sample was a random sample as required by s 6 of the Determination. Dr Carrick submitted that s 6 required the Committee itself to undertake a random sampling procedure.
75 A submission to the same effect was 'rejected out of hand' by Edmonds J in Mathews v Health Insurance Commission (2006) 90 ALD 49 at [40]. It is appropriate for me to follow the approach adopted by his Honour unless I am satisfied that his Honour was in error. I am not so satisfied. Indeed, in my respectful view, his Honour was plainly correct.
76 The sampling methodology prescribed by the Determination assumes an understanding of, and facility with, statistics. It is unlikely that the Minister intended that Committee members, rather than appropriately qualified statisticians, should undertake important statistical procedures. The obligation on a Committee to 'ensure' that the preliminary random sample is a random sample may, in my view, be met by the Committee satisfying itself that the preliminary random sample has been generated as a random sample by an appropriately qualified person. Nothing in the evidence establishes that the Committee did not take that step in this case.
77 I am not persuaded that it was necessary for the Committee to be given the full preliminary sample in the order of its original random selection, as argued by Dr Maxwell Stevenson, an expert statistician whose affidavit evidence was adduced by Dr Carrick. The responsibility of the Committee under s 106K and the Determination was not to engage in 'sound audit practice', to use Dr Stevenson's expression, but rather to ensure that s 6 of the Determination was complied with.
78 In any event, I see no reason to conclude that the Determination, considered in the context of the Act and s 106K in particular, discloses an intention that, where a preliminary random sample is in fact random, the findings of a Committee should be rendered invalid because the Committee itself failed to 'ensure' that the preliminary random sample was random (Project Blue Sky Inc v Australian Broadcasting Authority (1998) 194 CLR 355).
79 The second error claimed to affect the methodology adopted by the Committee was that the preliminary random sample was not a random sample in the hands of the Committee because it had already been the subject of consideration by the Director. This argument was rejected by Tamberlin J in Phan [2007] FCA 269 at [52]. His Honour there observed:
'In my view, the sample used in the present case was in accordance with the requirements of the Act. The evidence is that the Commission drew a preliminary random sample and that the Director required the production of records which he examined. He concluded, for the reasons given by him in his Report, that it would be appropriate for him to refer the matter to a Committee. In examining the material and completing his Report, the Director did not decide that there was inappropriate conduct. That determination was left for the Committee to make following a hearing and having regard to the evidence adduced at that hearing. It is not correct to suggest that the Director had found that the samples had in fact given rise to inappropriate conduct. The fact that the Director examined and considered the records and forwarded them to the Committee does not affect their randomness when they were considered afresh by the Committee. The sample items retained their character as random samples and were not skewed or biased so that the Committee's decision should be considered invalid.' (emphasis in original)
80 Again, I should follow the approach adopted by his Honour unless I am satisfied that his Honour was in error. I am not so satisfied. Indeed, in my respectful view, his Honour was correct.
81 The report of Dr Stevenson, parts of which I received by way of submission (O 10 r 1(2)(j) of the Federal Court Rules), assumed a requirement for the conclusions arrived at by the Committee to be 'truly independent of those of the Director'. Dr Carrick argued that were this not the case, the conclusions would otherwise be 'unfair'. It is, of course, the case that a Professional Services Review Committee is obliged to make its own finding in respect of the conduct of the person under review; that is, a finding as to whether the conduct constituted engaging in inappropriate practice. However, as I understand the report of Dr Stevenson, he takes the view that, at least where a Committee places reliance on s 106K of the Act, the findings of the Committee must be arrived at by reference to samples of services not previously considered by the Director. Additionally Dr Stevenson places reliance on the notion of 'data snooping' in suggesting that the preliminary random sample did not satisfy the requirements of s 7 of the Determination. Data snooping can occur when a given set of data is used more than once for the purpose of inference.
82 The question of whether the findings of the Committee were required to be arrived at by reference to samples of services not previously considered by the Director is to be answered by reference to the Act. References to sound audit practice and notions of fairness are of only limited assistance in this regard. Section 93(6) of the Act provides that where the Director makes an adjudicative referral the Director must prepare and attach to the adjudicative referral:
'a written report to the Committee, in respect of the services to which the referral relates, giving the reasons why the Director thinks that conduct by the person under review in connection with rendering or initiating the services may have constituted engaging in inappropriate practice' .
This statutory requirement suggests against any legislative intention that the conclusions arrived at by the Committee are to be arrived at by reference to evidence independent of that considered by the Director.
83 It is also significant in this regard that the Director is not authorised to make a finding that a person has engaged in inappropriate practice. Rather the role of the Director under Part VAA of the Act is as outlined in [12] above. A function of the Director under the scheme established by Part VAA is to ensure that only matters where the Committee could reasonably find that the person under review has engaged in inappropriate practice are referred to a Professional Services Review Committee.
84 I conclude that the conclusions arrived at by the Committee are not required to be 'truly independent of those of the Director' in the way assumed by Dr Stevenson.
85 I accept the evidence of Professor Nicholls that the notion of data snooping has no application in the context of the sampling methodology prescribed by the Determination. Even if it did, the Committee did not use any data more than once. The Committee was not required by the Determination to examine any sample other than the exploratory sample of 30 services. As the Act does not, in my view, require the conclusions of the Director to be arrived at by reference to evidence independent of that considered by the Committee, the fact that the Director had earlier given consideration to the same 30 services as part of his sample of 40 services is, I conclude, irrelevant.
86 The third and fourth errors claimed to affect the methodology adopted by the Committee are related. The third alleged error was that:
'The Committee should have not accepted or taken into account the various medical records of the applicant that the Director had delivered to it with or as part of the Adjudicative Referral as it adversely affected the sampling determination process under the Act and the Sampling Determination itself and it affected the independence and partiality of the Committee as a whole, afflicting the Committee with an apprehension of bias in undertaking the sampling determination process'.
87 The fourth alleged error was that the exploratory sample of 30 patient services considered by the Committee were not, in the hands of the Committee, a sample 'randomly drawn from the preliminary random sample' as required by s 8(a) of the Determination because they formed part of the pool of 40 patient services examined by Dr Heap as a consultant to the Director.
88 The submissions in support of both the third and fourth claimed errors assumed a requirement for the conclusions of the Committee to be arrived at independently of those of the Director in the sense of not being supported by reference to the same sample of patient services. For the reasons given above, I do not accept that the Act imposes such a requirement.
89 The final error claimed to affect the methodology adopted by the Committee was that the Committee failed to examine a random sample drawn from the preliminary random sample as required by s 8 of the Determination because the preliminary random sample was not before the Committee. The above proposition involves a non sequitur. Nothing in s 8 requires the Committee itself to draw the exploratory sample from the preliminary random sample. Section 8 requires the Committee to 'examine a sample … (the exploratory sample), randomly drawn from the preliminary random sample'. If the submissions based on the Director's previous review of the random sample of 40 patient services are put to one side, nothing in the evidence suggests that the exploratory sample examined by the Committee was not randomly drawn from the preliminary random sample. As indicated above, I do not consider that the Director's previous consideration of the 40 patient services affects this position.