Results
`
`infonnation to comment on the comparability
`of the groups. All report differences at baseline.
`jacksou and colleagues attempted to deal with the
`differences by undertaking a logistic regression to
`establish that the case-mix was independent of
`major outcomesm Peterson and co-authors re-
`analysed the data using a narrow group of patients
`who had not had a previous revascularisation and
`restricting any outcomes to the target lesion”?
`This did not result in any change in the results.
`Palmer and co-authors did not deal with the
`
`baseline differences, except by establishing
`identical success and complication rates in
`the two g1oups.l$7
`
`Quality of the studies
`The quality of the studies is reported in the
`economic studies checklist (see appendix 14;
`page 14]). Six of the studies reported a sensitivity
`analysis, with explicit assumptions. All the studies
`have flaws. Only one study (BENESTENT II) was
`an RCT with costs and outcomes collected and
`
`reported simultaneously.27 The general pattern
`of quality for sources of effectiveness data (items
`8—10 on checklist; see pages I41 and 142} were
`good but the pattern for costs considerably poorer
`(items 16—19; see page 142).
`
`Source of cost data
`Nine of the studies based their costings on bottomr
`'27, Iii-[.137—
`up costing exercisesfl"MM—”5""32 and five of these
`used European data.
`”5"“ Five studies used
`UK prices"'"'m"m‘lf'” and in three studies there was
`insufficient information given to determine the
`source of the cost data.m'1"i"" Further detail is
`
`given in appendix 12 (page 137).
`
`Outcome measures
`
`A range ot‘ outcome measures have been reported:
`eventifree survival (EFS), cost per eventifree
`survivor (cost/EFS). cost per outcome avoided,
`incidence of major adverse coronary events, cost
`per quality adjusted life-year (QAIX). (RES in the
`clinical effectiveness review has been taken to be
`
`the reverse of total event rate.) Appendix 13
`(page 139) shows which studies have reported
`individual outcome measures.
`
`EFS includes the absence oi‘death, MI and
`revascularisation procedures. These outcomes
`were used in the three studies that used this
`
`measure to compare PTCA with stem ting. Each
`ofthese outcomes carries equal weight in the
`outcome measure, but all of the studies reported
`the individual event rates separately and found
`that the major difference was in the
`revascularisation rates.
`
`38
`
`\Nith the exception of the 'West Midlands DEC
`report,I the quality oflife data used in all the
`cost—utility analyses were derived from the paper
`by Cohen and colleagues (1994)."‘i Cohen and
`colleagues used data from Pliskin's study of
`patients with angina and made some assumptions
`about quality of life for three different degrees
`of severity of angina.
`
`Results of cost-effectiveness analysis
`The cost/EFS is largely the cost per
`revascularisation procedure averted (which is
`usually a repeat PTCA) although there are small
`proportions of patients with MI or deaths.
`There is concern about the meaning of cost/
`EFS when the main event being prevented is
`repeat PTCA which has mainly resource
`rather than health implications.
`
`The cost/EFS for stents ranges from 58% higher
`than I’TCA to 31% lower. Results from the four
`
`studies reporting this outcome are shown in
`Table 7. The differences are a function of
`differences both in costs and in the EF‘S rates.
`
`However, the majority contributor to lower
`costs/EFS in slent patients in recent studies
`appears to be a reduction in cost differential.
`
`The earliest report used data from BENESTENT I
`and there is a large (55%) additional cost of
`stenting compared with ”CAM This high cost
`is mainly due to the anticoagulatjon regimen
`used for BENESTENT I. The same study also
`used data from the BENESTENT II pilot
`(Phase IV) (approximately 2 years later) and
`compared the stenting results from this with the
`PTCA results of" BENESTENT I. This comparison
`results in an 18% lower cost/EFS. The main
`contributor to the low cost/EFS for stenting is
`the large (22%) difference in EFS rates between
`the two groups. As the ellectiveness data were
`not collected over the same time period, it is
`likely that factors other than the type of procedure
`affected the result. The cost difference between
`
`the stenting in the BENESTENT II pilot
`(Phase IV) and PTCA is much lower than for
`BENESTENT I and this difference is largely
`due to the change to an antiplatelet regimen.
`
`Schwicker and Ban: reported the largest differences
`in cost/EFSJNLHE’ Their effectiveness estimates were
`derived from a literature review up to 1996 with
`some input from experts. Although they used
`quality criteria for the inclusion of studies, they
`also included some non-randomised trials, which
`may account for the larger differences in BPS rates.
`They also had the longest follow—up period.
`
`Page 52
`
`Medtronic Exhibit 1814
`
`Page 52
`
`Medtronic Exhibit 1814
`
`

`

`Health TechnologyAssessment 2000; Vol. 4: No. 23
`
`Both BENESTENT II and a study by Boston
`Scientific reported similar costs/RPS for PTCA
`and stenting.”"“" Both used the effectiveness data
`from BENESTENT II. Apart from the Boston
`Scientific study,”’“ all these studies used cost
`data from The Netherlands, which reduces the
`dilTerences between healthcare systems.
`
`Despite the above explaining variation. the general
`pattern revealed is a favourable or neutral impact
`on cost-effectiveness. This is particularly so when
`account is taken of the fact that the only cost-
`elI'ectiveness analysis showing markedly greater
`cost/EFS in the slenl group relative to the PTCA
`group is the oldest study which least reflects
`current practice.
`
`Difi'erencein costlEFSas
`
`%ofPTCA
`
`+38 —l8
`
`—29 —3
`
`+|2
`
`—2.9
`
`
`
`Cost-differi-CostlEFS
`
`
`
`DFI2|,000DFI22.000
`
`
`
`
`
`DFI29,000DFIl8.000
`
`
`
`PTCA
`
`Stents
`
`enceas36 ofFTCA
`
`+95
`
`+55
`
`
`
`DFI|5,208DFIl5.208
`
`
`
`
`
`DFI23.593DFIl6.663
`
`
`
`PTCA
`
`Stents
`
`
`
`DFIH.989 DFI27.27l
`
`DFIH.430 DFIH.697
`
`|.6
`
`+2.6
`
`
`
`DFI|2.479 DFIH.885
`
`
`
`DFI|2.8|2 DFI|5.|26
`
`
`
`
`
`DFI2|.073
`
`DFI2|.309
`
`+2.5
`
`
`
`DFII6.72?
`
`
`
`DFI|8.8|2
`
`£60|0
`
`£5840
`
`5S
`
`£4662
`
`£49I8
`
`Results of cost-utility analyses
`liable 8 shows the results of the studies reporting
`cost/QALY This also presents the ranges of
`cost/QAI.Y from the sensitivity analyses and the
`assumptions made in the models. Although the
`cost/QALY derived in the Wessex DEC studym
`is notably higher than in the other studies, the
`lower end of the sensitivity analysis is of a similar
`order as for the other results. Equally, the higher
`ranges oi'cost/QALY obtained from the studies
`by Guidant'“ and by Cohen and colleagues1mm
`are of a similar order to the Wessex DEC' result.
`The results are very sensitive to the assumptions
`used in the models, and the effectiveness and
`cost data used. In individual models the cost/
`QAIX was very sensitive to the restenosis rates
`and the costs of stenting. This was clearly
`demonstrated in a Inodel developed by Cohen
`and colleagues (1994).'54 The overall pattern
`suggests a cost/QALY difference between stents
`and PTCA of approximately £20,00{i—£30,00{L
`
`When comparing the costiutility results between
`studies other assumptions are important. The
`Wessex DEC assumed an equal mortality rate
`in the PTCA and stem groups and thus only
`included the difference in revaseularisation
`
`rates in their model)” The mortality rate after
`PTCA and stenting is approximately 1% at 1 year
`and thus it is a reasonable assumption to exclude
`deaths. When Guidantm excluded deaths from
`their model, the cost/QALY rose substantially.
`Although the West Midlands DEC also asstuned
`an equal death rate at 1 year, they included a
`higher mortality rate in the PTCA group at
`6 months hollow—up.I Boston Scientific'fio did
`not have a significantly different mortality rate
`at 1 year. The West Midlands DEC' used different
`quality ol'lil'e data for the diiierent grades of
`angina reported by BENESTENT II. This is in
`
`39
`
`
`
`
`
`
`
`Somefigureshavebeenrounded
`
`Costs
`
`EFSrate(96)
`
`Difference
`
`FTCA
`
`Stents
`
`IO 22
`
`I3 I4
`
`||
`
`5
`
`7D
`
`70
`
`76 68
`
`79
`
`78
`
`80 92
`
`89 82
`
`89
`
`84
`
`
`
`
`
`TABLE7FeaturesofstudiesreportingEFSratesandcosts
`
`Follow-up pefiod
`
`7months
`
`BENESTENT||pilot
`
`VanHoul:atal.”
`
`BENESTENT|
`
`Schwicker8cBanz'm”
`
`Study
`
`|year
`
`
`
`follow-upSVD3yearsfollow-up3yearsSVD|year
`
`
`
`|year |year
`
`
`
`BostonScientific'“
`
`
`
`BENESTENTll”
`
`
`
`
`
`SVD,singlevesselcoronarydisease
`
`
`
`
`
`Page 53
`
`Medtronic Exhibit 1814
`
`Page 53
`
`Medtronic Exhibit 1814
`
`

`

`Results
`
`TABLE 8 Anal‘ysis often—utility studies
`
`Study
`
`Key assumptions
`
`Difference in
`revascularisatiun
`rates (‘36)
`
`Costf
`Additional
`cast of stent QALY
`
`Range of costl
`QALY from
`sensitivity
`analysis
`
`Wessex DEC‘33 Patients with repeat FTCA had
`symptomatic restenosis with QOL
`valued at 0.8
`
`me
`
`£I43|
`
`£250,000
`
`£20,000—
`£772,000
`
`Waiting-time for revascularisation
`3 months
`
`Same procedural success rate in
`both groups
`
`Same survival rate in both groups
`PTCA if PTCA or stent
`
`West Midlands Different QOL data used for the
`DECI
`different grades of angina post
`PTCA and stent (data based on
`BENESTENT || results)
`
`5.6
`
`£9I9
`
`£23,000
`
`£I3,000—
`£53,000
`
`Average EUROQOL for post-PTCA
`patient with angina is 0.66I , and
`post-stent is 0.724
`
`Death rates at | year are the same,
`at 6 months for PTCA death rate
`= 0.5% and for stem = 0.2%
`
`One stent used per procedure
`
`Boston
`Scientificm
`
`Deaths: 0.2% more early deaths in
`PTCA group
`
`5.8
`
`(256”
`
`£3 |.500
`
`Approx.
`E I 5,00%
`£82,000
`
`Waiting-time for target-lesion
`revascularisation was 3 months
`
`Utility value with restenosis
`0.8 QALYs
`
`|.|7 scents used per procedure
`
`Cohen et at,
`I997 &
`I 999 I ‘17.H9
`
`55-year-old man with single
`vessel disease
`
`I6
`
`$900
`
`$33,700
`
`Ratenosis 3- 50% would require
`revascularisation
`
`Patients with restenosis would
`have a max. of 3 percutaneous
`revascularisation attempts
`before CABG
`
`‘lhis is the marginal cost afadjunctive stenting at 1 year, not the average price afa stem
`(201., quality of life
`
`40
`
`Costi'QALY
`increases to
`$200,000 for
`type A mid-right
`coronary
`stenosis, with
`abrupt closure
`rate of 336 and
`I'ESEEI‘IOSiS rate
`of 2¥SO%
`
`continued
`
`Page 54
`
`Medtronic Exhibit 1814
`
`Page 54
`
`Medtronic Exhibit 1814
`
`

`

`Health TechnologyAssessment 2000; Vol. 4: No. 23
`
`Cost!
`Additional
`Difference in
`revascularisation cost of stent QALY
`rates (%)
`
`ID
`
`£I04|
`
`£68I2
`
`Range of costl
`QALY from
`sensitivity
`analysis
`
`£633—
`£360,000 (it
`impact of deaths
`and CABGS and
`longer waiting
`times ignored)
`
`TABLE 8 contd Analysis of cost—utility studies
`
`Study
`
`Key assumptions
`
`l‘SuidantI48
`
`No difference was assumed in
`dead'l rates from primary
`procedures. but the submission
`includes the eflects of higher total
`dead-Is from secondary and
`subsequent procedures in the
`absence of stems, due to higher
`rates of restonosis
`
`\Naiting-time for target-lesion
`revascularisation was 3 months
`
`2—year follow-up
`
`contrast to the other studies, whicl'i derived
`their utility values for angina l'rom (Iohen and
`colleagues (199%).”H Guidantm calculated the
`lowest cost/QALY. This was the lowest end of the
`range in their sensitivity analysis, and they took
`a 2—year perspective, unlike the other studies.
`
`Stents compared with CABG in
`multi-vessel disease
`The ARTS study70 and Schwicker and Banzm‘m
`looked at stents in comparison with CABG for
`multi-vessel disease. They both reported higher
`rates ol'liFS in patients following CABG. Schwicker
`and Banz report lower costs at 3 years follow—up in
`stem patients. and ARTS has similar findings for
`patients with two-vessel disease. Despite the lower
`effectiveness, steu ting may be a cost—ell'ective altern—
`ative to (IABG in patients with multi—vessel disease.
`
`Summary and implications of
`economic analysis
`Variation is a marked feature mall the health
`economic data reviewed. This variation was
`
`particularly apparent between different estimates
`of cost. cost—effectiveness or cost—utility. There was
`also a contrast between the general message about
`efficiency provided by cost-eli'ectiveness analyses.
`which presented elective stenting as efiicient and
`having relatively minimal resource consequences.
`and that presented by the cost—utility estimates,
`which in the range of £20,000—f30,000 would be
`close to an important threshold distinguishing
`efficient from inellicient.
`
`Although the interrelationship was only examined
`crudely, we believe that there ate clues to the
`source 01' this contradiction.
`
`From the analysis of cost information, hospital
`costs ol'stents remain higher than those of PTCA
`despite the falling costs of'stents — differential 01'
`approximately £1500 to £1800. The cost differ-
`ential between stems and PTCA falls when the
`
`wider costs {of'l'ollow—up and repeat revascular—
`isatiort procedures) are taken into account.
`Taking this into account would reduce the
`cost differential to about £900.
`
`This differential in costs is similar to those used
`
`in cost—utility calculations. However the cost
`differential used in the cost-eli'ectiveness analyses
`is much narrower. In contrast to estimates of"
`effectiveness used in all the health economic
`
`analyses, there is a marked difference in the
`costs used. The question arises as to which set
`of analyses uses the most accurate costs. This is
`particularly important because costing methods
`were rarely given in the studies reporting cost
`data. Thus. there was little indication of whether
`key factors likely to influence relative cost, such
`as the degree of use of bailout stenting or multiple
`use ofstents, were taken into account. Uniquely,
`McKenna and colleaguesm provided a bottom—
`up costing, but despite good methods, it is clear
`that current practice in these key respects could
`not be anticipated in 1997.
`
`We believe, therefore, that the observation that
`the cost-effectiveness analyses tended to be based
`on bottom-up costings, and cost—utility estimates
`tended to be based on ill—defined costs or prices,
`suggests that greater caution should be applied
`to the interpretation the cost/QAI.Y figures.
`This is particularly so as the utility values used to
`assess impact are underpinned by a limited amount
`
`4|
`
`Page 55
`
`Medtronic Exhibit 1814
`
`Page 55
`
`Medtronic Exhibit 1814
`
`

`

`Results
`
`of research. Futtlier, in the interpretation of
`cost/QAI.Y figures, although the health value of
`the main event avoided — need for repeat PTCA —
`is probably correctly attributed a relatively low
`health value. this does not take into account the
`potential value of' avoiding repeat ‘PTCA to the
`wider healthcare system. This may be particularly
`pertinent in the NHS where there is evidence of
`significant uttdetipt‘ovision of revascularisation
`procedures for severe 1H D. In a situation in which
`there is an imperative to increase revascularisation
`rates, and where it may take time to develop
`capacity [i.e. increased numbers of centres with
`trained stall willl the appropriate technical skills),
`the value of avoiding repeat PTCAs may not
`be truly reflected by its impact on individttal
`health alone.
`
`Although we tentatively [avour the picture of
`efficiency suggested by the cost—effectiveness
`analyses, some caution also needs to be exercised
`in interpreting these. We had concern about the
`meaning of cost/EFS, where the main event beingr
`
`prevented is repeat PTCA, which arguably has
`greater resource consequences than personal
`health consequences.
`
`On the basis of the above we conclude that there
`is evidence that initial costs to achieve a reduced
`
`late of repeat PTCA may be. largely off—set by the
`savings this brings about. However, the confidence
`with I.vhich this can be asserted would be greatly
`improved if the resource neutrality of coronary
`artery stents could be confirmed, using more
`rigorously derived cost data.
`
`Finally, two points should be noted: liistly, that,
`despite some infonnation on costs and a health
`economic analysis, conclusions concerning the
`efficiency of stenting relative to CABG are
`hampered by a lack offully published effectiveness
`data; secondly that. although eii'ectiveness data
`exist showing the relative benefit of stenting
`relative to PTCA in AMI, no relevant cost or
`health economic analyses were identified.
`again prohibiting conclusions.
`
`42
`
`Page 56
`
`Medtronic Exhibit 1814
`
`Page 56
`
`Medtronic Exhibit 1814
`
`

`

`Health TechnologyAssessment 2000; Vol. 4: No. 23
`
`Chapter 4
`
`Discussion and conclusions
`
`Results summary
`Stents versus PTCA for subacute IHD
`
`(Le. mainly angina and unstable angina)
`General
`It is important to remember that whatever
`llle results of the evidence examined, we have
`implicitly accepted that there is a role for stem ting
`in treating acute closure occurring during a PT(3A
`(bailout or rescue stenting). The evidence for
`this is mainly observational, but convincing. The
`main alternative in this situation. an emergency
`CABG. appears to have worse outcomes, and
`has major resource ilnplications.
`
`BCIS audit data suggest that increasing stent use
`has been associated with a reduction in emergency
`CABG. However other technological advances
`could also contribute to this change over time.
`Although not part of the elfectiveness review, two
`small trials provided little support for prolonged
`balloon perfusion balloon inflation as an
`alternative to bailout stenting.
`
`Finally the availability of bailout stenting does not
`obviate the need for recourse to emergency CABG.
`
`Effects and effectiveness
`The key points are shown in Box 6.
`
`Costs
`
`The key points are presented in Box 7.
`
`Cost-effectiveness and cost-utility
`The key points are presented in Box 8.
`
`Stents versus CABG for subacute IHD
`
`(i.e. mainly angina and unstable angina)
`General
`
`Understanding whether elective stenting is
`effective and cost-effective in the management of
`complex patterns of coronary artery occlusion, for
`which currently CABS is the preferred method of
`management, is critical to planning an appropriate
`balance of provision between the two main modes
`of coronary artery revascularisation — PTCA and
`CABC. The importance of this is compounded
`by the fact that the two sets of procedures are
`undertaken by different professional groups
`whose skills are not obviously transferable.
`
`Effects and effectiveness
`Seven randomised trials were identified (three
`with sufficient information to make some entry in
`our study characteristics table; four without such
`information, detailed in the table of excluded
`studies). Unfortunately, none of the trials have
`reported their results fully, although a number
`have completed recruitment. Currently. there is
`thus no rigorous evidence on the eli'ectiveness of
`stems relative to CABC. However it seems likely
`that such evidence may become available over
`the next 2 years.
`
`Cost:
`(lost data are available on both PTCA and (IABG.
`
`All the provisos concerning the available cost data
`mentioned above apply.
`
`Cost-efi'ectiveness and cost-utility
`One health economic analysis was identified.
`This is based on an ongoing trial, but clearly until
`confirmed and fully published effectiveness data
`are available. this analysis must be regarded
`as speculative.
`
`Stents versus PTCA for acute MI
`General
`
`In order to interpret research comparing elective
`sten ting and PTCA for acute MI, we have assumed
`that PCI is at least as effective and cost—effective as
`
`medical acute management of MI. Although we
`did not specifically review this evidence, this seems
`reasonably well established.
`
`Effects and effectiveness
`There are a good number of randomised trials.
`with more in progress. Unfilrtunately the results
`of those that have been completed are devalued
`by incomplete or poor reporting. Although we
`have not examined these studies in as much detail,
`most of the issues highlighted in the analysis of
`trials on eleclive stenting versus PTGA in subacule
`IHD seem to apply.
`
`0 The PTCA arms of most of the trials actually
`allow bailout or rescue stenting.
`0 What constitutes bailout sten ting in the PTCA
`alone trial arms varies, and does not only
`include stenting for acute closure, but also
`for suboptimal PT(LA results.
`
`43
`
`Page 57
`
`Medtronic Exhibit 1814
`
`Page 57
`
`Medtronic Exhibit 1814
`
`

`

`Discussion and conclusions
`
`BOX 6 Stems versus PTCA for subaeute IHD: key points on effects and effectiveness
`
`There is a good volume of randomised trials, with many more in progress, Unfortunately the results of those
`that have been completed are in many cases devalued by incomplete or poor reporting.
`
`Interpretation of the available published trials is complicated by considerable clinical heterogeneity manifested
`by important differences in:
`— II-[D sub—types investigated
`— slanting strategies used
`— anticoagulation strategies used.
`
`The PTCA arms of most ofthe trials actually allow use of slenls when acute closure occurs during the
`angioplasty procedure {bailout stenting). Thus it is inaccurate to interpret the results of the trials as the ilnpact
`of stents versus no stents.
`
`Further, the definition ofwltat constitutes bailout stenling varies. In some trials, stenting occurring in the
`control arm appears to have been undertaken not just for acute closure but also for sub-optimal PTCA results.
`
`Thus, effectively trials compare treatment lmckages Comprising:
`— the PCI
`
`— rules for and patient preference for crossover
`— antithmmhotic therapy.
`
`
`
`There is a consistent difference between treatment and control groups other than use of stents, especially in the
`use of more intensive antithrombotic therapy. This could account for some of the difference in observed
`outcome, currently wholly attributed to slent use alone.
`
`Aside from the quality of reporting, the quality of trial Conduct also needs to he taken into account.
`Randomisation processes were often inadequately reported or sub-optimal. Further, steps to increase the
`objectivity of outcome assessment, although difficult, were rarely attempted. This is important to maintain
`validity, as in the absence of blinding there is clear risk of decisions to rte—intervene being heavily influenced by
`whether a patient was allocated to elective stenting or P’TCA alone.
`
`Although the above points introduce important sources of uncertainty, the following effects appear to have
`been established:
`
`— stents decrease total event rates (generally consisting of death, MI and need for re-intervention [either
`repeat I’TCA or CABGD; the summaly OR from the meta-analysis is 0.68 (95% (II. 0.59 to 0.78}
`the main component of this decrease is reduced numbers of repeat PTCAs'. the summary OR is 0.57 (95%
`Cl, 0.48 to 0.69)
`because of the relative rarity of events, it is impossible to be categorical about whether there is any impact on
`deaths, M15 and CABGs
`
`it is impossible to be categorical about the effect on being angina-free because relatively few trials have
`measured this outcome.
`
`This pattern exists whether outcomes are examined in the medium term (4—1 1 months) or the long—term
`(1—5 years).
`
`The general consistency of the results, with the possible exception of the effect on angina status, suggests that
`the [tracked clinical heterogeneity noted may not he as important in assessing the effectiveness of elective
`stenting as it might at [inst appear.
`
`Although not conclusive, there is no obvious evidence of publication bias.
`
`There is insufficient evidence to draw conclusions on whether prm'isional stenting (observing initial PTCA
`result, and only inserting a stent if deterioration in the initial result occurs} is an effective or cost-effective
`strategy relative to routine insertion of slenls.
`
`There is insufficient evidence to draw conclusions on use of stems in small coronary arteries (where the lumen
`
`44
`
`of the coronary artery is < 3 mm).
`
`Page 58
`
`Medtronic Exhibit 1814
`
`Page 58
`
`Medtronic Exhibit 1814
`
`

`

`Health TechnologyAssessment 2000; Vol. 4: No. 23
`
`BOX 7 Stems versus PTCA for subacute IHD: key points on costs
`
`There is a considerable amount of recent, routine and published cost data.
`
`“Whether considering the procedure costs, the hospital costs or the wider costs of stents relative to I‘TCA, there
`is uncertainty. manifest by wide variation.
`
`Some of this variation is likely to be due to rusting method, although it is difficult to substantiate this owing
`to poor reporting of the method by which costs or prices were derived. We have placed greatest reliance on
`explicit methods, which in practice meant weighting more highly bottomiup or microcosting exercises.
`
`It is unclear to what extent the following potentially very influential factors on cost have been taken into
`account:
`
`— established use of stents in routine P’I‘CA practice, particularly for bailout stenting
`
`— trends towards using multiple stents.
`
`Failure to take account of the first of the above would have a tendency to overestimate the cost differential:
`failure to take account of the second would have a tendency to underestimate the cost differential.
`
`With these provisos. there is a cost differential, stems costing more than P’TCA. The cost differential is smaller
`when wider costs are taken into account.
`
`9 There is a considerable volume of recent published health economic analyses, relating effectiveness and costs in:
`
`
`
`— mist—effectiveness analyses, particularly expressing cost/EFS
`
`— cost—utility analyses. expressed as cost/QALY.
`
`On appraisal, all analyses examined had important weaknesses.
`
`The overall pattern from cost-effectiveness analyses is that cost/EFS is less for elective stenting than PTCA,
`particularly in more recent analyses. [11 these the increased initial costs of stems are almost completely offset by
`savings resulting from reduced need for rcvascularisation.
`
`Although [ht-rt- was some concern about the interpn-latiun of the measure cust/EFS, when: the main event
`being prevented is repeat PTCA, the implication is that use of'stents, at least in the context of the trials on
`which the cost—effectiveness analyses were based, could be cost—neutral.
`
`The overall pattern from cost—utility analyses is less easy to discern, there being much wider variation, hilt
`marginal crnt/QALY in the region of £2ll,(HHl—30,000 are typical.
`
`Thus the cost—utility analyses appear less encouraging, partly reflecting the apparently low perceived personal
`health value of requiring a repeat PTGA after the Initial procedure. However, there is very little evidence in the
`literature on the impact ofslents on quality oflife.
`
`The view ofthe general efficiency of elective stunting thus seems to be dependent on the type of analysis used.
`Based on a limited exploration of the data we believe that this difference could arise from general differences
`in cost differential between stems and P'TCA. The cost-effectiveness analyses tend to use bottom-up costing; the
`cost—utilityr analyses tend simply to use prices. We believe the latter method of costing is less likely to take into
`account important factors influencing cost,
`
`A further important issue relevant to the interpretation of cost/QAIX figures, is that although the health value
`of the main event avoided — need for repeat PTCA — is correctly attributed a relatively low health value, this
`does not take into account the potential value of avoiding repeat l‘TlCLA to the wider healthcare system. This
`may be particularly pertinent in the NHS where there is evidence of significant, under-provision (if
`revascularisation procedures for severe IHD. In a situation where there is an imperative to increase
`revascularisalion rates, and where it may take time to develop capacity (i.e. increased numbers of' staff with the
`appropriate staff with the appropriate technical skills}, the value ofavoiding repeat PTCAs may not be truly
`reflected by its impact on individual health alone.
`
`45
`
`Page 59
`
`Medtronic Exhibit 1814
`
`Page 59
`
`Medtronic Exhibit 1814
`
`

`

`Discussion and conclusions
`
`' Randomisation processes were often
`inadequately reported or sub-optimal, and
`steps to reduce the bias introduced by the
`dilliculty of blinding to treatment allocation
`was rarely attempted.
`
`Similarly. although the above points introduce
`uncertainty, the following effects appear to have
`been established.
`
`U Elective sten ting decreases total event rates
`(genelally consisting of death, MI and need for
`re—intervention [either repeat PTCA or (JABGD.
`The summary OR from the meta—analysis is 0.39
`(95% Cl, 0.28 to 0.54).
`' The main component of this decrease is
`reduced numbers of repeat I’TCAs. The
`summary OR is 0.44 (95% CI, 0.25 to 0.74).
`I Because of the relative rarity of events, it is
`inlpossible to be categorical about whether
`there is any impact on deaths, Mls and CABCs.
`It is impossible to be categorical about the effect
`on being angina—tree because relatively few trials
`have measured this outcome, although one
`large trial found a significant difference in
`favour of stems)?“
`
`'
`
`Costs
`
`No cost data specific to the use of stents or PTCAs
`in the context of acute M] were identified.
`
`Cost-efiéctiveness and cost—utility
`Similarly, no health economic evaluations of the
`use of PTCA in comparison with stems in the
`context of acute MI were identified. The absence
`
`of such information is critical because of the major
`structural and resource implications of widespread
`use of either PTCA or stenting immediately
`after MI.
`
`Potential methodological
`strengths and weaknesses of
`the technology assessment
`Strengths
`We identify the following methodological features
`as being particularly robust:
`
`' a series of clearly defined questions
`' a comprehensive search strategy incorporating
`both published and partially published material
`I duplicate application ofinclusion and
`exclusion criteria
`
`' detailed assessment of included study quality
`0 duplicate data abstraction
`' use of meta—analysis to amplify the assessment of
`
`46
`
`patterns of results across several trials assessing
`the same intervention.
`
`Potential weaknesses
`In systematic reviews. publication bias is always a
`potential problem, and although the compre-
`hensive search strategy is a defence against this
`and the funnel plot showed no obvious evidence
`ofpublication bias, the possibility of it can never
`be completely excluded. Related to this is the
`major constraint of the lack of complete inform-
`ation on finished trials. The response to requests
`for further information from lead authors was
`
`poor but understandable given the relatively short
`time-scales involved. Collecting missing outcome
`data could be important for two reasons:
`
`I
`
`0
`
`it might allow more definitive conclusions
`on rarer outcomes like deaths, MI and
`repeat CABG
`it might provide reassurance that there is
`no selective reporting (Le. reporting only
`outcomes that show the intervention in its
`
`most favourable light).
`
`Ideally it would have been useful to explore
`completely the influence of different variables
`on the pattern of effectiveness results using meta-
`regression. However, although available time
`was a limiting factor. so too was availability of
`complete data, which as indicated above was
`outside our control.
`
`In the review of economic evaluations, quality of
`available cost data was a major limitation. Without
`clear methods it is impossible to assess the degree
`to which important costs have. or have not been
`included. Not undertaking our own de novo
`modelling of costs and effects might also be con-
`strued as a limitation, but our own view was that
`in the time available we could not overcome a
`
`major short—coining of the cost—utility estimates
`(in particular. poor assessment of costs using
`micro-costing techniques). Finally, as for the
`effectiveness data, additional efforts to explore
`the differences between the various economic
`evaluations identified could have increased the
`
`certainty of some of our conclusions on the
`general efficiency ofelecu've stenting.
`
`Important issues not addressed by this
`health technology assessment
`Key issues that this assessment did not encompass
`include the following.
`
`0 The evidence base for use of stents for bailout
`
`stenting.
`
`Page 60
`
`Medtronic Exhibit 1814
`
`Page 60
`
`Medtronic Exhibit 1814
`
`

`

`* The relative effectiveness of different stent types.
`O The effectiveness of PTCA + stents in those
`
`patients for whom the risk li'om FTCA and/or
`(JABG is currently perceived to be too great.
`These patients can currently only be offered
`medical therapy, which in the specific situation
`is unlikely to be offering complete relief of
`symptoms attributable to IHD.
`- The evidence base for newer technologies
`(e.g. laser and minimally invasive CABS}.
`However, although possible in theory, we are
`not convinced that it is possible to predict
`how stenting will relate to developing
`technologies, particularly whether it will
`be superseded, and if so when.
`0 The impact on PC] of dillerent anti—thrombotic
`regimens, particularly glycoprotein lib/Illa
`inhibitors. The assessment also did not address
`the issue of whether the newer antiithrombotic
`
`regimens added to PTCA alone without use
`ofstents may achieve some of the benefit
`currently attributed wholly to stent use.
`
`Conclusions
`
`'
`
`In subac