1. Trang chủ
  2. » Giáo Dục - Đào Tạo

The 2005 meta analysis of homeopathy the

9 3 0

Đang tải... (xem toàn văn)

THÔNG TIN TÀI LIỆU

Thông tin cơ bản

Định dạng
Số trang 9
Dung lượng 3,65 MB

Các công cụ chuyển đổi và chỉnh sửa cho tài liệu này

Nội dung

If we confine ourselves to the predefined hypotheses and the part of the analysis that is indeed comparative, the conclusion should be that quality of homeopathic trials is better than o

Trang 1

PROOF

ORIGINAL PAPER

The 2005 meta-analysis of homeopathy:

the importance of post-publication data

ALB Rutten1,*and CF Stolper2

1

Homeopathic physician, former general practitioner, Breda, The Netherlands

2

General practitioner, homeopathic physician, Machteldskamp 19, 8181 ZN Heerde, The Netherlands

Abstract: Background: There is a discrepancy between the outcome of a meta-analy-sis published in 1997 of 89 trials of homeopathy by Linde et al and an analymeta-analy-sis of 110 trials

by Shang et al published in 2005, these reached opposite conclusions Important data were not mentioned in Shang et al’s paper, but only provided subsequently

Questions: What was the outcome of Shang et al’s predefined hypotheses? Were the homeopathic and conventional trials comparable? Was subgroup selection justified?

The possible role of ineffective treatments Was conclusion about effect justified? Were essential data missing in the original article?

Methods: Analysis of post-publication data Re-extraction and analysis of 21 higher quality trials selected by Shang et al with sensitivity analysis for the influence of single indications Analysis of comparability Sensitivity analysis of influence of subjective choices, like quality of single indications and of cut-off values for ‘larger samples’

Results: Quality of trials of homeopathy was better than of conventional trials Regard-ing smaller trials, homeopathy accounted for 14 out of 83 and conventional medicine

2 out of 78 good quality trials with n < 100 There was selective inclusion of unpublished trials only for homeopathy Quality was assessed differently from previous analyses

Selecting subgroups on sample size and quality caused incomplete matching of homeop-athy and conventional trials Cut-off values for larger trials differed between homeophomeop-athy and conventional medicine without plausible reason Sensitivity analyses for the influence of heterogeneity and the cut-off value for ‘larger higher quality studies’ were missing Homeopathy is not effective for muscle soreness after long distance running,

OR = 1.30 (95% CI 0.96–1

was based was heterogeneous, comprising 8 trials on 8 different indications, and was not matched on indication with those of conventional medicine Essential data were missing in the original paper

Conclusion: Re-analysis of Shang’s post-publication data did not support the conclu-sion that homeopathy is a placebo effect The concluconclu-sion that homeopathy is and that conventional is not a placebo effect was not based on comparative analysis and not justified because of heterogeneity and lack of sensitivity analysis If we confine ourselves

to the predefined hypotheses and the part of the analysis that is indeed comparative, the conclusion should be that quality of homeopathic trials is better than of conventional trials, for all trials (p = 0.03) as well as for smaller trials (p = 0.003) Homeopathy (2008)

-, 1–9

Keywords: homeopathy; meta-analysis; comparative analysis; quality bias; selection bias; cut-off value; adverse effects

Q2

Introduction

The discussion about proof for homeopathy is in part,

a meta-discussion about proof Several meta-analyses of randomised controlled trials (RCT) – in 1991, 1997 and

*Corresponding author Lex Rutten, Aard 10, 4813 NN Breda, The

Netherlands Tel.: +31 (0) 765227340; Fax: +31 (0) 765227487.

E-mail: lexrtn@concepts.nl

Received 11 January 2008; revised 6 August 2008; accepted 11

September 2008

doi: 10.1016/j.homp.2008.09.008 , available online at http://www.sciencedirect.com

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142

Trang 2

PROOF

2000 – indicate a specific effect of homeopathy.1–3 Both

homeopathic and conventional meta-analyses have been

criticised.4–6Some authors suggest that there is no

differ-ence between proof for homeopathy and for conventional

methods.1,7 However, the implausibility of homeopathy’s

mechanism of action seems to have led to an amalgamation

of bias Sterne, Egger and Smith concluded that the role of

low quality in small studies was neglected in Linde’s

meta-analysis.8Commenting on the analysis of homeopathy by

Shang et al published in August 2005, and which referred

to the ‘small low quality study’ hypothesis, the editor of

the Lancet advised ‘‘doctors need to be bold and honest

with their patients about homeopathy’s lack of benefit’’.9

Vandenbroucke concluded that this meta-analysis showed

higher sensitivity for potential bias for homeopathic than

for allopathic trials.10

The Cochrane Handbook for Systematic Reviews states

‘‘Reliable conclusions can only be drawn from analyses

that are truly pre-specified before inspecting the trials’

results’’.11Such pre-specification is more difficult because

most homeopathy trials have been analysed in earlier

meta-analyses The Cochrane Handbook further

recom-mends ‘‘Meta-analysis should only be considered when

a group of trials is sufficiently homogeneous in terms of

participants, interventions and outcomes to provide a

mean-ingful summary’’ Pooling of results is thus questionable if

homeopathy works for some conditions and not for

others.12 Egger stated ‘‘If subgroup analyses are to be

done, they need to be as complete as possible and should

involve commonly defined subgroups and outcomes across

all the trials in the subgroup’’.13Meta-analysis is a

subjec-tive procedure, Boden warns that it can easily become

a weapon instead of a tool.14

The hypotheses predefined mentioned in the introduction

of Shang et al’s paper were: ‘‘Bias in conduct and reporting

of trials is a possible explanation for positive findings of

placebo-controlled trials of both homeopathy and allopathy

(conventional medicine)’’; and: ‘‘These biases are more

likely to affect small than large studies; the smaller a study,

the larger the treatment effect necessary for the results to be

statistically significant, whereas large studies are more

likely to be of high methodological quality and published

even if their results are negative’’

Shang et al’s analysis was criticised because the authors

failed to include essential data to support their

conclu-sion.15–17Four months later the missing data were revealed

(www.ispm.ch) The missing data were 1 Excluded trials

2 The trials regarded as of higher quality 3 The trials

(8 homeopathy, 6 conventional medicine) that led to the

final conclusion

Questions

More or less the same set of homeopathy trials has been

re-analysed several times The contradiction between

Linde’s conclusion based on 89 trials, and Shang et al’s

conclusion, based on 110 trials seems odd Shang et al’s

analysis was presented as a comparative analysis matching

110 homeopathy trials with 110 conventional trials by

indications The conclusion was based on 8 homeopathy trials and 6 conventional trials

The post-publication data enabled us to reconstruct the analysis, although data were presented as graphs, not as raw numbers In our recent paper ‘The conclusions on the effectiveness of homeopathy highly depend on the set of analysed trials’ we re-analysed the data from the original articles, did sensitivity analyses and estimated the influence

of heterogeneity.18 The large amount of heterogeneity suggests that this factor was not considered at all We found

no reasonable explanation for the choice of cut-off value for

‘larger trials’

After these basic conclusions several questions remain:

1 What was the outcome of the pre-specified hypotheses?

2 Were the two methods comparable?

3 Was subgroup selection rectified?

4 What is the influence of ineffective treatments?

5 Was the final conclusion rectified?

6 Were essential data missing in the original article?

Methods

We analysed the subsequently disclosed data and inves-tigated which hypotheses were tested The ISPM website presented graphs, but no data about effect sizes and confidence intervals were given We reconstructed the odds ratios and confidence intervals of the 21 higher quality homeopathy studies from the original articles Data were processed and analysed with methods identical or equiva-lent to those of Shang et al’s analysis We checked the results with Shang et al’s data, then focused on these 21 higher quality studies because the conclusion was based

on larger higher quality studies For these trials a random effects meta-analysis was performed and the pooled odds ratio was estimated We estimated odds ratios and confi-dence intervals for some of the trials excluded by Shang

et al, but regarded as good quality by Linde et al We performed meta-analyses for other eligible sets of trials

We tested comparability and matching of trials We com-pared this analysis with referenced publications to check predefinition of hypotheses We assessed the influence of some subjective choices, like quality and cut-off values for sample size and performed sensitivity analysis to check for the influence of separate indications SAS/StatÒ, release 9.1 statistical software was used

Results

Shang et al presented their study and their conclusion as

a comparison of homeopathy and conventional medicine

To reconstruct their work we had to make several hypothe-ses that were not predefined by Shang et al, to arrive at their conclusions In this process comparibility of the homeo-pathic and conventional groups was lost

The predefined hypotheses The first predefined hypothesis (quality in homeopathy is worse than in conventional medicine) was falsified by

2

143

144

145

146

147

148

149

150

151

152

153

154

155

156

157

158

159

160

161

162

163

164

165

166

167

168

169

170

171

172

173

174

175

176

177

178

179

180

181

182

183

184

185

186

187

188

189

190

191

192

193

194

195

196

197

198

199

200

201

202

203

204

205

206

207

208

209

210

211

212

213

214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284

Trang 3

PROOF

Shang et al Median sample sizes were the same: 65.5 in

homeopathy, 65 in conventional medicine Effects of

home-opathy and conventional medicine were similar; 95% of the

odds ratios were from 0.12 to 1.65 for homeopathy and

from 0.13 to 1.52 for conventional medicine According

to Shang et al ‘‘Most odds ratios indicated a beneficial effect

of the intervention’’ In the homeopathy group (including

unpublished trials) 21 (19%) of the trials were of higher

quality, in the conventional group 9 (8%) Overall quality

in homeopathy studies was better than for conventional

medicine (p = 0.03)

Quality in small studies.Shang et al referred, for their

second predefined hypothesis, to Sterne, Egger and Smith

stating that quality bias is mainly influenced by quality of

small studies.8Effects of treatment could in their view truly

be larger in high quality smaller trials because of better

selection of patients On the other hand effects are

over-es-timated if quality is low In both cases we see asymmetry in

the funnel plot, but in the first case this does not indicate

bias and if larger trials with poorer patient selection then

indicate no effect the conclusion that the therapy is placebo

is not justified

Post-publication data showed which studies were

regarded as of higher quality We chosen < 100 as cut-off

value for smaller studies, Shang chosen < 98 for

homeopa-thy and n < 146 for conventional medicine (see below)

There were 14 homeopathy studies of higher quality out

of 83 trials (16.9%) withn < 100 There were two

conven-tional studies of higher quality out of 78 trials (2.6%) with

sample size <100.19,20 The hypothesis that low quality

small studies are therefore responsible for the positive

findings in homeopathy is mostly falsified (p = 0.003,

Fisher exact probability test) There is statistically

signifi-cant difference in quality of smaller studies in favour of

homeopathy

Since the quality of conventional and homeopathic

stud-ies was not comparable, comparison of effects of the two

methods was not valid The underlying hypothesis for

Shang et al’s analysis was that results cannot be compared

if quality is different As so much emphasis was laid on the

relation between quality and result we will nevertheless

continue with our observations concerning this relation,

although we did not compare effects of homeopathy and

conventional medicine

Comparability

The comparison of the two methods was somewhat

flawed by publication bias The 110 homeopathy trials

were matched on indication with 110 conventional trials

But all conventional trials were published as journal articles

while 16 (15%) of the trials in the homeopathy group were

unpublished According to Chan et al the odds of publishing

results in conventional medicine are greater if results were

significant (pooled odds ratio 2.4, 95% CI 1.4–4.0).21So,

in comparing effects homeopathy is disadvantaged by the

selective inclusion of unpublished trials But it also affects

comparison of quality Shang et al reported (in

post-publi-cation data) that none of the 16 unpublished homeopathy

trials were of higher quality The ratio of higher quality trials in published trials was 22% (21 out of 94) instead of 19% in the original paper

We did not further investigate possible selection bias by excluding trials, but we were surprised by the exclusion of Wiesenauer’s trial on chronic polyarthritis.22 This was

a larger trial (n = 176), of good quality according to Linde, with positive results.2 This trial would have contributed positively to the outcome of the larger higher quality trials

Shang excluded this trial because no matching trial could be found

Subgroups were selected on quality This selection further influenced matching on indication, and therefore comparability The homeopathy group contained 21 ‘higher quality’ studies, the conventional group 9 At this point only

4 homeopathy studies were matched on indication by con-ventional studies (19%) From this point onward Shang’s study consisted in fact of two incomparable meta-analyses

of effects, one about homeopathy, one about conventional medicine

Differences in effect between methods can no longer be evaluated if the matching is disrupted This can be shown

by comparing results for muscle soreness The post-publi-cation data show that neither homeopathy nor conventional medicine is effective for this indication, seeFigure 1 But the homeopathy studies are of higher quality while the conventional studies are not This difference was of fundamental importance in the subset that led to the final conclusion

The indication ‘muscle soreness’ has the largest influence

on the results of homeopathy and on the comparison between homeopathy and conventional medicine because four homeopathy studies were classified as higher quality against none for conventional medicine One of the home-opathy trials was also large and therefore higher in the funnel plot This trial inclines the funnel plot to the right (towards OR = 1.0), while the smaller trials for this indica-tion inclined it to the left because the pivot point is above these trials There is a strong influence of chance in such

a limited number of indications

We did not consider clinical relevance, but one could wonder about the inclusion of treatments that may not be used because of serious adverse effects Shang et al men-tioned in the discussion that a limitation of their study was its disregard of adverse effects They highly valued larger studies as a measure of quality and extrapolated effects towards the largest studies This extrapolation is questionable if the largest studies involve treatments that are not available because of serious adverse effects In

a larger trial of higher quality on weight loss homeopathy had no effect.23 The matched conventional study showed

a considerable positive effect of Dexfenfluramine,24 but Dexfenfluramine for weight loss was withdrawn by the American Food and Drug Administration in 1997 because

of serious cardiac complications.25 Two other larger stud-ies, Deladumone (androgen–estrogen) in breastfeeding and Piroxicam for soft tissue injury suffered from the same problem.26,27These two treatments were also with-drawn because of adverse effects.28,29There might be other

3 285

286

287

288

289

290

291

292

293

294

295

296

297

298

299

300

301

302

303

304

305

306

307

308

309

310

311

312

313

314

315

316

317

318

319

320

321

322

323

324

325

326

327

328

329

330

331

332

333

334

335

336

337

338

339

340

341

342

343

344

345

346

347

348

349

350

351

352

353

354

355

356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426

Trang 4

PROOF

treatments which are hard to compare because of safety,

such as Tamoxifen for pre-menstrual syndrome.30

Possible bias in subgroup selection

Shang et al’s conclusion was based on comparison of

‘larger higher quality trials’ Possible pitfalls here are

in-comparability, heterogeneity and subjective criteria for

quality and sample size

The subjectivity of interpreting quality is demonstrated

by differences between authors of meta-analyses The

following studies were not classified as good quality by

Shang et al, although they are among the quality top 10

of Linde’s meta-analysis2: de Lange-de Klerk,31 Reilly

1986,32Hofmeyr,33Reilly 1994,34seeTable 1

If we add these four studies to Shang et al’s pooled Odd

Ratio (OR) of 25 trials becomes 0.74 (95% CI: 0.59–0.94)

Why should these studies not be valued as of higher

qual-ity? Schultz et al showed that inadequate concealment of

treatment allocation is the most important quality factor,

as-sociated with 41% (95%CI: 27–52%) exaggeration of

ef-fect.35 Other quality factors have less effect; sequence

generation 15%, (95%CI: 12–19%), double blinding 17%

(95%CI: 4–29%) The Jadad score for quality used by

Linde, does not consider allocation concealment These 4

trials in Linde’s meta-analysis had maximum Jadad scores,

and as far as we can tell also had adequate allocation

con-cealment We cannot estimate the influence of effect of

choices regarding quality because some new trials were

published after Linde’s study and some of Linde’s lower quality trials were regarded as of higher quality But, in combination with an unclear definition of ‘larger’ sample size, this subjectivity in defining quality opens a variety

of possible subgroups which could be considered ‘larger higher quality’ trials

Pooled odds ratio of all higher quality studies We reconstructed the ORs and confidence intervals of the 21 higher quality studies selected by Shang et al 36–55The pooled OR using random effects analysis for all 21 higher quality studies in homeopathy is 0.76 (95% CI 0.59– 0.99), which is not compatible with the placebo hypothe-sis.18

Cut-off value for sample size.Cut-off values for sample size were not mentioned or explained in Shang el al’s analysis Why were eight homeopathy trials compared with six conventional trials? Was this choice predefined

or post-hoc? Post-publication data showed that cut-off values for larger higher quality studies differed between

Figure 1 The effects of homeopathy and conventional medicine on ‘muscle soreness’ compared The other trials in the group ‘Musculo-skeletal complaints’ are disregarded The four studies concerning muscle soreness for both methods are indicated by author names.

N = trial size Source www.ispm.ch

Table 1 The four best studies according to Linde et al, arranged

by sample size First author Indication Sample size OR 95% CI of OR

de Lange-de Klerk 31 Upper respiratory

tract infection

170 0.85 0.47–1.53 Reilly32 Pollinosis 144 0.43 0.22–0.85 Hofmeyr 33 Childbirth 122 1.03 0.40–2.64 Reilly34 Asthma 24 0.08 0.02–0.40

4

427

428

429

430

431

432

433

434

435

436

437

438

439

440

441

442

443

444

445

446

447

448

449

450

451

452

453

454

455

456

457

458

459

460

461

462

463

464

465

466

467

468

469

470

471

472

473

474

475

476

477

478

479

480

481

482

483

484

485

486

487

488

489

490

491

492

493

494

495

496

497

498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568

Trang 5

PROOF

the two groups In the homeopathy group the cut-off value

wasn = 98, including eight trials (38% of the higher quality

trials) The cut-off value for larger conventional studies in

this analysis wasn = 146, including six trials (66% of the

higher quality trials) These cut-off values were

consider-ably above the median sample size of 65 There were 31

homeopathy trials larger than the homeopathy cut-off value

and 24 conventional trials larger than the conventional

cut-off value We can think of no criterion that could be

common to the two cut-off values This suggests that this

choice was post-hoc

Effect of larger higher quality trials Shang et al

decided that, based on this subset, homeopathy is a placebo

response The studies that constitute the evidence for the

conclusion of the authors are listed inTable 2

The two sets of trials are incomparable and

heteroge-neous with a pooled OR = 0.88 (95% CI: 0.65–1.19) for

homeopathy Only two homeopathy studies (ref.38,40) could

be matched with conventional studies (56–58) The

homeop-athy group consisted of 8 trials on 8 different indications

Egger warned ‘‘Opinions will often diverge on the correct

method for performing a particular meta-analysis The

robustness of the findings to different assumptions should

therefore always be examined in a thorough sensitivity

anal-ysis’’.59Our sensitivity analysis showed that if Vickers´ trial

on muscle soreness is omitted from the eight largest higher

quality homeopathy trials the overall odds ratio reduces

from 0.88 to 0.80, but remains statistically not significant

(95% CI: 0.61–1.05)

Ineffective unusual treatment

Sensitivity analysis of the higher quality studies showed

one indication with four studies: homeopathic Arnica for

muscle soreness after long distance running.37,49,50,52The

pooled effect of those studies was in favour of placebo,

OR = 1.30 (95% CI: 0.96–1.76) As treatment of healthy individuals is very rare in homeopathic practice this outcome has low external validity to judge the effect of homeopathy as a method The fact that conventional medi-cine is also ineffective for this indication (seeFigure 1) is omitted due to disrupting of matching on indication

The final conclusion The final conclusion that homeopathy is a placebo response (and conventional medicine is not) was flawed

on several grounds:

1 Homeopathy and conventional trials were not comparable

2 Heterogeneity disallows conclusions about effect

3 Sensitivity analysis was missing

4 The cut-off value for larger trials was decisive

How essential were the missing data?

In theBox 1we summarise the conclusions that could only be drawn from post-publication data These data pro-vided all answers to the questions we mentioned above except one: the fact that overall quality was better in homeopathy trials

Another possible outcome Often subjective choices must be made in meta-analy-ses We evaluated the influence of some such choices in this case: the indication ‘muscle soreness’, the cut-off value and the interpretation of quality We did not con-sider exclusion of trials, publication bias, quality bias or other possible bias Table 3andFigure 2show the influ-ence of cut-off values after exclusion of the trials on mus-cle soreness In Table 3some pooled OR and confidence intervals for ‘larger trials’ are given, with and without Linde’s trials discarded by Shang et al If we choose the overall median sample size (n = 65) as cut-off we disre-gard half of all trials This seems a reasonable cut-off value for larger studies

Figure 2shows the cumulated-pooled OR (including the Linde trials omitted by Shang et al) if we increase step by step the number of included higher quality studies, starting with the largest two

Adding the four ‘Linde trials’ does not change effects, but shifts the most unfavourable cut-off value from the 7th to the 10th trial Discarding the indication ‘muscle sore-ness’ lowers the pooled OR from 0.88 to 0.80 Depending

on the choice of cut-off value the OR varies between 0.72 and 0.80 Cochrane reviews are typically based on 8–10 studies60and are homogenous as to indication Linde con-cluded that there was insufficient evidence for only single condition.2But Shang et al’s 110 trials included 8 of home-opathy for acute upper respiratory tract infection, with no evidence of quality bias and a considerable effect size,

OR = 0.36, 95% CI: 0.26–0.50 For muscle soreness after marathon running homeopathic Arnica is clearly not effective

Table 2 Larger higher quality studies, according to Shang et al

Indication Homeopathy Conventional

medicine Diarrhoea Jacobs 40

N = 116

Kaplan 56

N = 256 Treatment of

influenza

Papp.38N = 334 Nicholson.57

N = 319

de Flora.38

N = 248 Prevention of

influenza

Rottey.36

N = 501 Plantar warts Labrecque.39

N = 162 Weight loss Schmidt.23

N = 208 Muscle

soreness

Vickers 37

N = 400 Headaches Walach 42

N = 98 Sinusitis Weiser 41

N = 104

Post operative

infection

Crowley.

N = 273

5 569

570

571

572

573

574

575

576

577

578

579

580

581

582

583

584

585

586

587

588

589

590

591

592

593

594

595

596

597

598

599

600

601

602

603

604

605

606

607

608

609

610

611

612

613

614

615

616

617

618

619

620

621

622

623

624

625

626

627

628

629

630

631

632

633

634

635

636

637

638

639

640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710

Trang 6

PROOF

Discussion

We found indications that Shang et al’s hypothesis and

hence its conclusion was sensitive to subjective choices

and the influence of one indication and that the subsets on

which the conclusions were based were not comparable

The missing data were of crucial importance, exploring

these data seriously undermines the conclusion that

home-opathy is a placebo response

We calculated a number of possible pooled odds ratios

for the effect of homeopathy as a method, excluding one

indication for which homeopathy is ineffective, to show

the isolated position of Shang et al’s hypothesis Taking

all pooled odds ratios indicates an effect, but in some cases

the confidence interval includes 1.0, depending on the

definition of ‘larger trial’ The quality of the whole set

(p = 0.03) and quality in small studies (p =0.003, Fisher

exact test) are better in homeopathy than in conventional

medicine The placebo hypothesis is also falsified if only

higher quality studies are considered The comparison of

homeopathy and conventional medicine was flawed by

the inclusion of unpublished trials only in the homeopathy group and possibly by excluding trials The conclusion of Shang et al is based on one subgroup of 8 trials on 8 differ-ent indications, not on a comparative analysis Our sensitiv-ity analysis showed one indication and a specific cut-off value play a decisive role in the final conclusion Our addition of Linde’s best trials did not alter Shang’s overall results, but increased the number of (larger) higher quality trials, and the effect of the eight largest higher quality trials became significant (OR = 0.73; 95%CI: 0.59–0.91)

Small effects can be clinically relevant In a meta-analy-sis of statin treatment and the occurrence of haemorrhagic stroke Vergouwen et al found an effect of OR = 0.88 This is the same OR as in Shang’s final subset.61 In the case of statins the 95% confidence interval is below 1.0 because the pooled sample size is large Insufficient sample size and heterogeneity could easily lead to type II error (false negative) if OR = 0.88

Shang et al compared homeopathy and conventional medicine in terms of quality and effect, but originally matched trials on indication Quality and effect are

Table 3 The influence of cut-off values for ‘larger studies’, excluding ‘muscle soreness’, with or without Linde’s best studies, using random

effects analysis

Largest higher quality trials,

Shang’s quality criteria,

without muscle soreness

6 trials, cut-off n = 104 OR = 0.73 (0.59–0.91) 0.0051

7 trials, cut-off n = 98 OR = 0.80 (0.61–1.05) 0.1087

8 trials, cut-off n = 81 OR = 0.75 (0.58–0.96) 0.0246

13 trials, cut-off overall median OR = 0.66 (0.49–0.89) 0.0058 Largest higher quality trials,

without muscle soreness, + 3

of Linde’s best quality studies

8 trials, cut-off n = 116 OR = 0.73 (0.54–0.98) 0.0336

9 trials, cut-off n = 104 OR = 0.72 (0.56–0.94) 0.0158

10 trials, cut-off n = 98 OR = 0.78 (0.59–1.03) 0.0776

11 trials, cut-off n = 81 OR = 0.75 (0.57–0.98) 0.0340

16 trials, cut-off overall median OR = 0.78 (0.57–0.97) 0.0273

Box 1 Conclusions that could only follow from the subsequently disclosed data

 Comparison of quality and effect is flawed by inclusion of unpublished trials only for homeopathy Restraining to

published trials, quality was higher in 22% instead of 19% as mentioned by Shang

 Judgement of quality is different from other analyses in at least four trials

 The predefined hypothesis that positive results of homeopathy could be explained by quality bias in smaller trials was

falsified (p = 0.003)

 If only higher quality trials are considered, the placebo hypothesis for homeopathy is falsified, OR = 0.76 (95%

CI 0.59–0.99)

 The final conclusion was not based on comparative analysis, there was no matching on indication between homeopathy

and conventional medicine

 The conclusive subgroup analysis was not rectified because of heterogeneity, it considered 8 trials for 8 different

indications

 Cut-off values for larger trials were unexplainably different for homeopathy (n = 98) and conventional medicine

(n = 146) This suggests post-hoc hypothesizing

 Sensitivity analysis showed that one indication and the chosen cut-off value for larger trials explained the final

conclusion of statistically non-significant effect

 At least one larger higher quality homeopathy trial with positive result was excluded on unclear grounds

 Comparative extrapolation of effects was questionable because of publication bias, selection bias, difference in quality

and sample size and difference in safety

6

711

712

713

714

715

716

717

718

719

720

721

722

723

724

725

726

727

728

729

730

731

732

733

734

735

736

737

738

739

740

741

742

743

744

745

746

747

748

749

750

751

752

753

754

755

756

757

758

759

760

761

762

763

764

765

766

767

768

769

770

771

772

773

774

775

776

777

778

779

780

781

782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852

Trang 7

PROOF

interrelated; better quality trials show less effect

Compar-ing effects when quality is not matched is thus questionable

They subsequently selected subgroups matched for quality,

but that disrupts matching by indication In the end the

conclusion about the effect of homeopathy was based on

meta-analysis of a selection of trials Discarding trials for

some indication because no comparable trials could be

found causes selection bias This selection could have had

more influence on the final conclusion (positive or negative)

than our re-analysis of 21 trials could detect The

insuffi-cient matching on indication in the final subsets of 8

home-opathy and 6 conventional trials did not allow any

comparison of effects We assumed matching on indication

when we tested the hypothesis of quality bias in smaller

trials The subgroups of 78 (homeopathy) and 83

(conven-tional medicine) smaller trials are largely but not fully

(82%) matched on indication

Shang et al made the choice to disregard safety This

decreased the relevance of the comparison of effects of

homeopathy and conventional medicine Some

conven-tional treatments in this analysis are not available because

of serious adverse effects

We also performed meta-regression analysis on the 21

good quality trials and found asymmetry in the funnel

plot If we extrapolate the odds-ratios by meta-regression

we see no difference between homeopathy and placebo at

extreme sample numbers We think that this is irrelevant

to this discussion Sterne, Egger and Smith stated that

asym-metry in good quality trials is not caused by bias but by

stronger effects in smaller trials.8This could be interpreted

as proof that the asymmetry in the set of 110 homeopathy

trials is not caused by bias However, from mathematical

statistics it is well known that such meta-regressions are

imprecise, especially when the number of observations is

small Asymmetry of homeopathy trials and conventional

trials cannot be compared, because there is a significant

dif-ference in the number of smaller good quality trials between

homeopathy and conventional medicine Different size in

matched trials also plays a role in asymmetry of the funnel

plot: both homeopathy and conventional medicine are

inef-fective for muscle soreness, but homeopathy is higher in the

funnel plot because the trials are larger The influence of this indication on asymmetry is opposite for homeopathy and conventional medicine One could also question the role

of drugs with strong effects but with serious adverse effects

on asymmetry Three conventional treatments, which have been withdrawn because of serious adverse effects, had large effect sizes and small standard errors and therefore considerable positive influence on the asymmetry and the extrapolated effect of the funnel plot of conventional med-icine Difference in publication bias has also influence on the position of the funnel plot

We did not investigate the influence of subjective choices

on the OR of conventional trials We think that there are methodological objections against comparing effects in this analysis Moreover, the clinical relevance of such

a comparison is low Homeopathy is mostly used after con-ventional medicine failed, so the indication for use is differ-ent Homeopathy is highly valued for its safety The scientific relevance of Shang’s comparative analysis lies

in the comparison of quality Quality of trials was an impor-tant issue in the discussion about proof and implausibility

The fact that the pooled effect of homeopathy excluding the indication ‘muscle soreness’ is positive does not mean that homeopathy is effective for all other indications If the trials on influenza or Jacobs’ trials on diarrhoea are excluded results become statistically not significant As our re-analysis is post-hoc we cannot draw conclusions regarding efficacy

The clinical relevance of trials is not considered in this analysis, but doctors must be interested in Shang’s finding that eight trials showed a substantial effect of homeopathy

in acute upper respiratory tract infections (OR = 0.36, 95% CI: 0.26–0.50), without indications of bias

Conclusion

A review of data provided after publication of Shang et al’s analysis did not support the conclusion that homeopa-thy is a placebo effect There was intermingling of compar-ison of quality and comparcompar-ison of effects, and thus matching was lost The comparison of effects was also flawed by subjective choices and heterogeneity The result

in the subgroup from which the conclusion was drawn was further influenced by the choice of cut-off value for ‘larger’

trials If we confine ourselves to the predefined hypotheses and the part of this analysis that is consistent with the comparative design, the only legitimate conclusion is that quality of homeopathy trials is better than of conventional trials, for all trials (p = 0.03) as well as for smaller trials withn < 100 (p = 0.003)

Acknowledgements

We thank Rainer Lu¨dtke for assisting with the statistics

References

1 Kleijnen J, Knipschild P, ter Riet G Clinical trials of homeopathy.Q3

BMJ 1991; 302: 316–323.

Figure 2 Indicates that all but 3 of 20 possible cut-off values lead

to a significant effect if we consider all higher quality trials The

ex-ceptions are n = 144 (including 6 trials), n = 122 (7 trials) and

n = 98 (10 trials) If the cut-off point is below n = 98 the pooled

results of higher quality studies is consistently statistically

significant.

7 853

854

855

856

857

858

859

860

861

862

863

864

865

866

867

868

869

870

871

872

873

874

875

876

877

878

879

880

881

882

883

884

885

886

887

888

889

890

891

892

893

894

895

896

897

898

899

900

901

902

903

904

905

906

907

908

909

910

911

912

913

914

915

916

917

918

919

920

921

922

923

924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962

Trang 8

PROOF

2 Linde K, Clausius N, Ramirez G, et al Are the clinical effects of

homeopathy placebo effects? A meta-analysis of placebo-controlled

trials Lancet 1997; 350: 834–843.

3 Cucherat M, Haugh M, Gooch M, Boissel J Evidence of clinical

efficacy of homeopathy - A meta-analysis of clinical trials Eur J

Clin Pharmacol 2000; 56: 27–33.

4 Morrison B, Lilford RJ, Ernst E Methodological rigour and results

of clinical trials of homeopathic medicines Perfusion 2000; 13:

132–138.

5 Ernst E A systematic review of systematic reviews of homeopathy.

J Clin Pharmacol 2002; 54: 577–582.

6 Ezzo J, Bausell B, Moerman DE, Berman B, Hadhazy V Reviewing

the reviews How strong is the evidence? How clear are the

conclu-sions? Int J Technol Assess Health Care 2001 Fall; 17(4): 457–466.

7 Vandenbroucke JP Medical journals and the shaping of medical

knowledge Lancet 1998; 352: 2001–2006.

8 Sterne JAC, Egger M, Smith GD Investigating and dealing with

publication and other biases in meta-analysis BMJ 2001; 323:

101–105.

9 Shang A, Huwiler-Mu¨ntener K, Nartey L, et al Are the clinical

ef-fects of homeopathy placebo efef-fects? Comparative study of

placebo-controlled trials of homeopathy and allopathy Lancet

2005; 366: 726–732.

10 Vandenbroucke JP Homeopathy and the growth of truth Lancet

2005; 366: 691–692.

11 Higgins JPT, Green S (eds) Cochrane Handbook for Systematic

Re-views of Interventions 4.2.5 The Cochrane Library 2005;(Issue 3).

Chichester, UK: John Wiley & Sons, Ltd [updated May 2005].

12 Jonas WB, Kaptchuk TJ, Linde K A critical overview of

homeopa-thy Ann Intern Med 2003; 138: 393–399.

13 Egger M, Smith GD, Altman DG Systematic reviews in health care:

meta-analysis in context 2nd edition BMJ Books, 2001.

14 Boden WE Meta-analysis in clinical trials reporting: has a tool

become a weapon? Am J Cardiol 1992; 69: 681–686.

15 Walach H, Jonas W, Lewith G Letter to the editor Lancet 2005;

366: 2081.

16 Linde K, Jonas WB Letter to the editor Lancet 2005; 366:

2081–2082.

17 Fisher P, Berman B, Davidson J, Reilly D, Thompson T, et al Letter

to the editor Lancet 2005; 366: 2082–2083.

18 Lu¨dtke R, Rutten ALB The conclusions on the effectiveness of

ho-meopathy highly depend on the set of analysed trials J Clin Epid (To

be filled in).

19 Gade J, Thorn P Paraghurt for patients with irritable bowel

syndrome: a controlled clinical investigation from general practice.

Scand J Prim Health Care 1989; 7: 23–26.

20 Humphrey RG, Bartfield MC, Carlan SJ, O’Brien WF, O’Leary TD,

Triana T Sulindac to prevent recurrent preterm labor: a randomized

controlled trial Obstet Gynecol 2001; 98: 555–562.

21 Chan AW, Hjobartsson A, Haahr MT, Gotzsche PC, Altman DG.

Emperical evidence for selective reporting of outcomes in

random-ized trials: comparison of protocols to published articles JAMA

2004; 291: 2457–2465.

22 Wiesenauer M, Gaus W Wirksamkeitsnachweis eines

Homo¨opathi-kums bei chronischer Polyarthritis Eine randomisierte

Doppelblindstu-die bei niedergelassenen A ¨ rzten Aktuelle Rheumatologie 1991; 16: 1–9.

23 Schmidt JM, Ostermayr B Does a homeopathic ultramolecular

dilution of Thyroidinum 30CH affect the decrease of body weight

re-duction in fasting patients? - A randomised Placebo-controlled

dou-ble-blind trial Homeopathy 2002; 91(4): 197–206.

24 Enzi G, Crepaldi G, Inelmen EM, Bruni R, Baggio B Efficacy and

safety of dexfenfluramine in obese patients: amulticenter study Clin

Neuropharmacology 1995; 12: S173–S178.

25 Available from: <http://www.fda.gov/CDER//news/phen/

fenphenpr81597.htm

26 Louviere RL, Upton RT Evaluation of Deladumone OB in the

suppres-sion of postpartum lactation Am J Obstet Gynecol 1975; 121: 641–642.

27 Lacey PH, Dodd GD, Shannon DJ A double blind, placebo con-trolled study of piroxicam in the management of acute muscoloske-letal disorders Eur J Rheum Inflam 1984; 7: 95–104.

28 Available from: <http://www.fda.gov/ohrms/dockets/98fr/ 102998b.pdf>

29 EMEA Press release London: European Medicines Agency Rec-ommends Restricted Use for Piroxicam, 25 June 2007.

30 Grio R, Cellura A, Geranio R, Porpiglia M, Piacentino R Efficacia clinica del tamoxifene nel trattamento della mastodynia premenstru-ale Min Ginecol 1998; 50: 101–103.

31 de Lange-de Klerk ESM Effects of homeopathic medicines on chil-dren with recurrent upper respiratory tract infections BMJ 1994; 309: 1329–1332.

32 Reilly DT, Taylor MA, McSharry C, Aitchison T Is homeopathy

a placebo response? Controlled trial of homeopathic potency with pollen in hay fever as model Lancet 1986;881–886.

33 Hofmeyr GJ, Picconi V, Blauhof P Postpartum homeopathic Arnica montana: a potency-finding pilot study Br J Clin Pract 1990; 44: 619–621.

34 Reilly D, Taylor MA, Beattie NG, et al Is evidence for homeopathy reproducible? Lancet 1994; 344: 1601–1606.

35 Schulz KF, Chalmers I, Hayes RJ, Altman D Emperical evidence of bias Dimensions of methodological quality associated with esti-mates of treatment effects in controlled trials JAMA 1995; 273: 408–412.

36 Rottey EED, Verleye GB, Liagre RLP Het effect van een homeopa-thische bereiding van micro-organismen bij de preventie von griep-symtomen - Een gerandomiseerd dubbel-blind onderzoek in de huisartspraktijk Tijdschr IntegGeneeskunde 1995; 11: 54–58. Q7

37 Vickers AJ, Fisher P, Wyllie SE, Rees R Homeopathic Arnica 30X

Is Ineffective for Muscle Soreness After Long-Distance Running - A randomized, double-blind, Placebo-controlled trial Clin J Pain 1998; 14(3): 227–231.

38 Papp R, Schuback G, Beck E, et al Oscillococcinum in patients with influenza-like syndroms: A placebo controlled double-blind evalua-tion Brit Homeopath J 1998; 87(2): 69–76.

39 Labrecque M, Audet D, Latulippe L, Drouin J Homeopathic treat-ment of plantar warts Can Med Assoc J 1992; 146(10): 1749–1753.

40 Jacobs J, Jime´nez LM, Malthouse S, et al Homeopathic Treatment

of Acute Childhood Diarrhea - Results from a Clinical Trial in Nepal.

J Alternat Complement Med 2000; 6(2): 131–139.

41 Weiser M, Clasen BPE Randomisierte plazebokontrollierte Doppel-blindstudie zur Untersuchung der klinischen Wirksamkeit der homo¨-opathischen Euphorbium compositum-Nasentropfen S bei chronischer Sinusitis Forsch Komplementarmed 1994; 1: 251–259.

42 Walach H, Haeusler W, Lowes T, et al Classical homeopathic treat-ment of chronic headaches Cephalalgia 1997; 17: 119–126.

43 Jacobs J, Jime´nez LM, Gloyd SS, Gale JL, Crothers D Treatment of Acute Childhood Diarrhea With Homoeopathic Medicine: A Ran-domized Clinical Trial in Nicaragua Pediatrics 1994; 93(5): 719–725.

44 Jacobs J, Springer DA, Crothers D Homeopathic treatment of acute otitis media in children: a preliminary randomized placebo-controlled trial Pediatr Infect Dis J 2001;177–183.

45 Hart O, Mullee MA, Lewith G, Miller J Double-blind, placebo-con-trolled, randomized clinical trial of homoeopathic arnica C30 for pain and infection after total abdominal hysterectomy J R Soc Med 1997; 90: 73–78.

46 Wiesenauer M, Ha¨ussler S, Gaus W Pollinosis-Therapie mit Galphi-mia glauca MMW Fortschr Med 1983; 101(17): 811–814.

47 Zell J, Connert WD, Mau J, Feuerstake G Behandlung von akuten Sprunggelenksdistorsionen - Doppelblindstudie zum Wirksamkeits-nachweis eines homo¨opathischen Salbenpra¨parats MMW Fortschr Med 1988; 106(5): 96–100.

48 Bo¨hmer D, Ambrus P Behandlung von Sportverletzungen mit Trau-meel-Salbe - Kontrollierte Doppelblindstudie Biol Medizin 1992; 21(4): 260–268.

8

963

964

965

966

967

968

969

970

971

972

973

974

975

976

977

978

979

980

981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051

Trang 9

PROOF

49 Vickers AJ, Fisher P, Smith C, Wyllie SE, Lewith GT

Homoeopa-thy for delayed onset muscle soreness - A randomised double blind

placebo controlled trial Brit J Sports Med 1997; 31: 304–307.

50 Jawara N, Lewith GT, Vickers Aj, Mullee MA, Smith C

Homoeo-pathic Arnica and Rhus toxicodendron for delayed onset muscle

soreness - A pilot for a randomized, double-blind,

placebo-con-trolled trial Brit Hom J 1997; 86(1): 10–15.

51 Chapman EH, Weintraub RJ, Milburn MA Homeopathic Treatment of

Mild Traumatic Brain Injury - A Randomised, Double-Blind,

Placebo-Controlled Clinical Trial J Head Trauma Rehabil 1999; 14: 521–542.

52 Tveiten D, Bruset S, Borchgrevink CF, Norseth J Effects of the

homeopathic remedy Arnica D30 on marathon runners: A

random-ized, double-blind study during the 1995 Oslo Marathon

Comple-ment Ther Med 1998; 6(2): 71–74.

53 Stevinson C, Devaraj VS, Fountain-Barber A homeopathic arnica

for prevention of pain and bruising: randomized placebo-controlled

trial in hand surgery J R Soc Med 2003; 96: 60–65.

54 Lepaisant C Essai the´rapeutique en home´opathie: traitement des

tensions mammaires et mastodynies du syndrome pre´menstruel.

Rev Fr Gyne´col Obste´t 1995; 90(2): 94–97.

55 Chapman EH, Angelica J, Spitalny G, Strauss M Results of a study

of the homoeopathic treatment of PMS J Am Inst Hom 1994; 87:

14–21.

56 Kaplan MA, Prior MJ, McKonly KI, Du Pont HL, Temple AR, Nelson EB A multicenter randomized controlled trial of a liquid lo-peramide product versus placebo in the treatment of acute diarrhea in children Clin Pediatr 1999; 38: 579–591.

57 Nicholson KG, Aoki FY, Osterhaus ADME, et al Efficacy and safety of oseltamivir in treatment of acute influenza: a randomised controlled trial Lancet 2000; 355: 1845–1850.

58 de Flora S, Grassi C, Carati L Attenuation of influenza-like symp-tomatology and improvement of cell-mediated immunity with long-term N-acetylcysteine treatment Eur Respir J 1997; 10:

1535–1541.

59 Egger M, Smith GD, Philips AN Meta-analysis: principles and procedures BMJ 1997; 315: 1533–1537.

60 Shang A, et al Author’s reply Lancet 2005; 366: 2083–2085.

61 Vergouwen MD, de Haan RJ, Vermeulen M, Roos YB Statin treatment of hemorrhagic stroke in patients with a history of cerebro-vascular disease Stroke 2008; 39: 497–502.

9 1052

1053

1054

1055

1056

1057

1058

1059

1060

1061

1062

1063

1064

1065

1066

1067

1068

1069

1070

1071

1072

1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094

Ngày đăng: 05/01/2022, 17:03

TÀI LIỆU CÙNG NGƯỜI DÙNG

TÀI LIỆU LIÊN QUAN

🧩 Sản phẩm bạn có thể quan tâm