Hypertension: antihypertensives for reduction of blood pressure, prevention of cardiovascular events and mortality

Several drug classes yielded similar results (Table DS3). Combining all agents, blood pressure was reduced by 9.4 mmHg systolic and 5.5 mmHg diastolic in the short term (SMDs 0.54 and 0.56 respectively). ^{Reference Law, Morris, Jordan and Wald15} In the long term all drug classes significantly reduced cardiovascular events, e.g. angiotensin-converting enzyme (ACE) inhibitors reduced such events from 18% to 14% (ARD 4%, RRR 22%). ¹⁶ A significant reduction of mortality has not been shown for all of them (Table DS3).

Fig. 1

Summary of effect sizes.

All effect sizes in online Tables DS3 and DS4 are presented except for duplicates (e.g. effect size on dichotomous response and continuous reduction of severity in schizophrenia). Online Fig. DS25 identifies which dot corresponds to which result, and Figs DS26–29 present the results of dichotomous outcomes as relative and absolute risk/responder differences. Data on older meta-analyses from Table DS1 are not included. Effect sizes of dichotomous outcomes were converted to standardised mean differences expressed as Hedges’ g. Effect sizes of general medicine medication are presented on the left-hand side (median 0.37, mean 0.45, 95% CI 0.37–0.53) and psychiatric drugs on the right-hand side (median 0.41, mean 0.49, 95% CI 0.41–0.57).

TABLE 1

Examples of the efficacy of general medicine drugs v. placebo (full version is given in online Table DS3)

						Participants with outcome %
Study/AMSTARFootnote a	Therapy	No. of studies	n	Duration, mean	Outcome (units)	Placebo	Drug	ARDFootnote b %	RRR/RR %	MD	SMDFootnote c
Hypertension
Effects on blood pressure
Law et al 15/5	Any antihypertensive	94	17 641	8 wk	RR systolic (mmHg)					9.4Footnote ***	0.56Footnote ***
		94	17 641	8 wk	RR diastolic (mmHg)					5.5Footnote ***	0.54Footnote ***
Long-term effects on cardiovascular events and mortality
BPLTTC16/5	ACE inhibitors	5	18 229	3.9 yr	Cardiovascular events	18.1	14.1	4 NI	22Footnote ***		0.16
Acute stroke
Wardlaw17/9	Thrombolysis	22	6 283	12-26 wk	Death/dependency	55.8	50.9	5Footnote **	9Footnote **		0.11
Prevention of cardiovascular disease and stroke
Baigent et al 21/5	Aspirin (primary prev.)	6	95 000	5.8 yr	Cardiovascular events	0.57/yr	0.51/yr	0.07/yr	12/yrFootnote ***		0.06
Baigent et al 21/5	Aspirin (secondary prev.)	16	17 000	NI	Cardiovascular events	8.2/yr	6.7/yr	1.5/yr	19/yrFootnote ***		0.12
Law et al 22/5	Statins	164	̃38 000	2–6 wk	LDL cholesterol (mmol/l)					1.54Footnote ***	NI
Baigent et al 23/6	Statins	14	90 056	5.0 yr	Cardiovascular events	17.8	14.1	4Footnote ***	21Footnote ***		0.15
Chronic heart failure
Flather et al 98/6	ACE inhibitors long term	5	12 763	2.9 yr	Mortality	26.8	23.0	4Footnote ***	15Footnote ***		0.11
Rheumatoid arthritis
Suarez-Almazor et al 99/10	Methotrexate	5	218	412 wk	Tender joints					NE	0.86Footnote ***
Migraine
McCrory & Gray31/8	Sumatriptan	8	2 221	2 h	Pain-free	8.5	29.5	20Footnote ***	220Footnote ***		0.41
Linde & Rossnagel33/9	Propanolol	4	205	13 wk	Response	30.9	52.3	35Footnote *	80Footnote *		0.49
Asthma
Sin et al 36/7	Corticosteroids	19	3 271	>12 wk	FEV1 (l)					0.33Footnote ***	0.56Footnote ***
		11	8 999	>12 wk	Exacerbations	NI	NI	NI	54Footnote *		NE
Chronic obstructive pulmonary disease
Yang et al 42/10	Inhaled corticosteroids	3	952	8–24 wk	FEV1 (l)					0.10Footnote ***	0.36Footnote ***
		4	2 063	>26 wk	Exacerbations					0.26Footnote ***	0.20Footnote ***
Diabetes
Saenz et al 44/11	Metformin	12	1 587	21.5 wk	Fasting glucose (mmol/l)					1.84Footnote ***	0.87Footnote ***
		1	753	10.7 yr	Mortality	21.7	14.6	7Footnote **	32Footnote **		0.27
Chronic hepatitis C
Myers et al 46/10	Interferon	8	409	426 wk	Virological response	1.0	38.3	35Footnote ***	1070Footnote ***		2.27
Reflux oesophagitis
Moayyedi et al 48/9	Proton pump inhibitors	5	645	8 wk	Remission	28.3	83.2	58Footnote ***	256Footnote ***		1.39
Ulcerative colitis
Sutherland & Macdonald50/9	5-ASA	4	892	8 wk	Remission	10.0	19.9	8Footnote ***	70Footnote **		0.44
Multiple sclerosis
Rice et al 53/10	Interferon	3	919	2.0 yr	Exacerbation	69.5	55.2	14Footnote ***	19Footnote ***		0.34
Breast cancer
EBCTCG56/4	Polychemotherapy	60	28 764	15.0 yr	Mortality age <50	42.4	32.4	10Footnote ***	24Footnote ***		0.24
Non-small cell lung cancer
Bria et al 57/8	Adjuvant chemotherapy	21	7 408	4.5 yr	Mortality	NI	NI	3 NI	9Footnote *		NE
Antibiotics for various diseases
Glasziou et al 60/10	Otitis media	10	2 791	2–7 d	With pain	22.2	16	6.2 NI	28Footnote ***		0.22
Falagas et al 61/8	Cystitis	4	1 062	3–17 d	Cure	25.7	61.8	36.1 NI	139Footnote ***		0.85

ACE, angiotensin-converting enzyme; ARD, absolute response or risk difference; ASA, acetylsalicylic acid; BPLTTC, Blood Pressure Lowering Treatment Trialists' Collaboration; d, day; EBCTCG, Early Breast Cancer Trialists' Collaborative Group; FEV1, forced expiratory volume in 1 s; h, hours; LDL, low-density lipoprotein; MD, mean difference; NE, not estimable; NI, not indicated; NNT, number needed to treat; NS, not significant; prev., prevention; RRR/RR, response ratio/relative risk reductions; SMD, standardised mean difference; wk, week; yr, year.

a. AMSTAR quality score (range of possible values 0–11).

b. Positive values always mean superiority of drug.

c. Italics indicate mean estimated values.

^{Footnote *} P<0.05

^{Footnote **} P<0.01

^{Footnote ***} P<0.001. Results on absolute and relative risk/responder differences do not always exactly match with the formulae presented in the manuscript due to weighting processes in meta-analyses.

Acute ischaemic stroke: thrombolysis, aspirin and heparin for prevention of death or dependency

Thrombolysis reduced death or dependency from 56% to 51% (ARD 5%, RRR 9%), ^{Reference Wardlaw, Murray, Berge and Del Zoppo17} but when administered after 4.5 h mortality is increased by haemorrhages. ^{Reference Hacke, Kaste, Bluhmki, Brozman, Davalos and Guidetti18} Aspirin reduced death or dependency from 46% to 45%, ^{Reference Sandercock, Counsell, Gubitz and Tseng19} whereas heparin was ineffective. ^{Reference Sandercock, Counsell and Kamal20}

Cardiovascular disease: aspirin for primary and secondary prevention of cardiovascular events and mortality

In secondary prevention, low-dose aspirin reduced serious cardiovascular events per year from 8.2% to 6.7% (ARD 1.5%, RRR 19%) and vascular mortality per year from 4.1% to 3.7% (ARD 0.29%, RRR 9%, P = 0.05). ^{Reference Baigent, Blackwell, Collins, Emberson, Godwin and Peto21} In primary prevention, aspirin reduced the number of cardiovascular events per year from 0.57% to 0.51%, but there was no effect on mortality because the reduction of occlusive events was balanced by an increase in major bleeds (mortality per year: placebo 0.19%, drug 0.19%). ^{Reference Baigent, Blackwell, Collins, Emberson, Godwin and Peto21}

Hypercholesterolaemia: statins for reduction of cholesterol levels and prevention of cardiovascular disease and mortality

In the short term, statins reduced low-density lipoprotein (LDL) cholesterol by 1.54 mmol/l or 31%. ^{Reference Law, Wald and Rudnicka22} In the long term, cardiovascular events were reduced from 18% to 14% (primary and secondary prevention combined, ARD 4%, RRR 21%) and 5-year mortality from 9.7% to 8.5%. ^{Reference Baigent, Keech, Kearney, Blackwell, Buck and Pollicino23}

Chronic heart failure: various drugs for reduction of mortality

Angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, beta-blockers and diuretics respectively reduced mortality rates from 27% to 23% (ARD 4%, RRR 15%), from 18% to 11%, from 13% to 8% and from 12% to 3%. ^{Reference Garg and Yusuf24–Reference Faris, Flather, Purcell, Poole-Wilson and Coats27} Digitalis significantly reduced hospital admission (from 33% to 25%) but not mortality. ^{Reference Hood, Dans, Guyatt, Jaeschke and McMurray28}

Rheumatoid arthritis: antirheumatic drugs for the reduction of tender joints

Various immunosuppressants, corticosteroids and other agents reduced the number of tender joints with reasonably good SMDs between 0.33 and 1.33 (raw values for mean differences were not presented; Table DS3). ^{Reference Suarez-Almazor, Belseck, Shea, Homik, Wells and Tugwell29,Reference Blumenauer, Judd, Wells, Burls, Cranney and Hochberg30}

Acute migraine: effects of sumatriptan and aspirin on the number of patients pain-free after 2 h

Sumatriptan increased the percentage of patients pain-free after 2 h from 9% to 30% (ARD 20%, RR 220%) ^{Reference McCrory and Gray31} and intravenous aspirin increased it from 15% to 27%. ^{Reference Lampl, Voelker and Diener32}

Prophylaxis of migraine: effects of propanolol and anticonvulsants on responder rates and on the number of migraine attacks

Fifty-two per cent responded to propanolol prophylaxis and 31% to placebo (ARD 35%, RR 80%). ^{Reference Linde and Rossnagel33} Patients had approximately one migraine attack less (SMD 0.47). ^{Reference Linde and Rossnagel33} The results of anticonvulsants were similar. ^{Reference Mulleners and Chronicle34}

Chronic asthma: effects of inhaled corticosteroids and beta-2-agonists on forced expiratory volume and on asthma exacerbations

The first-line drugs for chronic, severe asthma are inhaled corticosteroids and beta-2-agonists (short-acting as needed, long-acting in patients with refractory disease). ³⁵ Inhaled corticosteroids increased forced expiratory volume in 1 s (FEV₁) by 330 ml (SMD 0.56). ^{Reference Sin, Man, Sharpe, Gan and Man36} The addition of long-acting beta-2-agonists improved FEV₁ by 190 ml (SMD 0.35), ^{Reference Ni, Greenstone, Danish, Magdolinos, Masse and Zhang37} but the reduction of asthma exacerbations found by some meta-analyses is controversial, ^{Reference Sin, Man, Sharpe, Gan and Man36,Reference Bateman, Nelson, Bousquet, Kral, Sutton and Ortega38} because another meta-analysis found more severe exacerbations. ^{Reference Salpeter, Buckley, Ormiston and Salpeter39}

Chronic obstructive pulmonary disease: effects of various agents on FEV₁ and on disease exacerbations

Guidelines recommend anticholinergics, beta-2-agonists and inhaled corticosteroids. ^{Reference Vogelmeier, Buhl, Criée, Gillissen, Kardos and Köhler40} The anticholinergic tiotropium improved FEV₁ by 200 ml (SMD 0.99). ^{Reference Barr, Bourbeau, Camargo and Ram41} It reduced exacerbations from 31% to 23% (ARD 5%, RRR 17%). ^{Reference Barr, Bourbeau, Camargo and Ram41} Inhaled corticosteroids improved FEV_l by 100 ml (SMD 0.36) and the number of exacerbations per patient and year by 0.26 (SMD 0.20). ^{Reference Yang, Fong, Sim, Black and Lasserson42} The data on long-acting beta-2-agonists are equivocal. They reduced exacerbations (e.g. Salpeter et al), ^{Reference Salpeter, Buckley and Salpeter43} but one systematic review found them to increase respiratory deaths. ^{Reference Salpeter, Buckley and Salpeter43}

Type 2 diabetes: various antidiabetics for reduction of HbA_1c and mortality

Metformin reduced HbA_1c by 1% (SMD 0.97) and α-glucosidase inhibitors reduced it by 0.8% (SMD 0.64). ^{Reference Saenz, Fernandez-Esteban, Mataix, Ausejo, Roque and Moher44,Reference Van de Laar, Lucassen, Akkermans, Van de Lisdonk, Rutten and van Weel45} In the long term, metformin reduced the death rate from 22% to 15% (ARD 7%, RRR 32%), ^{Reference Saenz, Fernandez-Esteban, Mataix, Ausejo, Roque and Moher44} but α-glucosidase inhibitors have not been shown to change the death rate. ^{Reference Van de Laar, Lucassen, Akkermans, Van de Lisdonk, Rutten and van Weel45}

Hepatitis C: effects of interferon and ribavirin on virological response/morbidity and mortality

Interferon increased the number of participants with no detectable virus at treatment end (virological response) from 1% to 38% (ARD 35%, RR 1070%). ^{Reference Myers, Regimbeau, Thevenot, Leroy, Mathurin and Opolon46} Ribavirin was only efficacious in combination with interferon. ^{Reference Brok, Gluud and Gluud47}

Reflux oesophagitis: proton pump inhibitors for clinical remission and relapse prevention

Proton pump inhibitors are highly effective in acute treatment (response: placebo 28%, drug 83%, ARD 58%, RR 256%), ^{Reference Moayyedi, Santana, Khan, Preston and Donnellan48} and also in maintenance treatment (relapse: placebo 75%, drug 36%). ^{Reference Donnellan, Sharma, Preston and Moayyedi49}

Ulcerative colitis: 5-aminosalicylic acid for clinical remission and maintenance of remission

Five-aminosalicylic acid (5-ASA) increased remission from 10% with placebo to 20% (ARD 8%, RR 70%), ^{Reference Sutherland and Macdonald50} and maintenance of remission from 37% to 53% (ARD 18%, RR 50%). ^{Reference Sutherland and Macdonald51}

Multiple sclerosis: corticosteroids for treatment of acute episodes and interferon for prevention of exacerbations

Acute treatment with corticosteroids increased the proportion of responders from 28% with placebo to 68% (ARD 41%, RR 140%). ^{Reference Filippini, Brusaferri, Sibley, Citterio, Ciucci and Midgard52} In the first 2 years, prevention with interferon beta reduced exacerbations from 70% to 55% (ARD 14%, RRR 19%). ^{Reference Rice, Incorvaia, Munari, Ebers, Polman and D'Amico53}

Parkinson's disease: effects of levodopa on disease symptoms

There was no systematic review of the standard treatments levodopa or dopamine agonists with data compared with placebo. We parenthetically note that the National Institute for Health and Clinical Excellence (NICE) guideline based its recommendation on a pivotal 42-week trial in which levodopa produced 7 points more improvement in the Unified Parkinson's Disease Rating Scale total score than placebo (SMD 0.93), ⁵⁴ but also a 7% stronger decline of striatal dopamine transporter density (SMD —0.44), suggesting a possible acceleration of nigrostriatal dopamine nerve terminal loss. ^{Reference Fahn, Oakes, Shoulson, Kieburtz, Rudolph and Lang55}

Breast and lung cancer: polychemotherapy for reduction of mortality

Breast cancer is the most frequent neoplasm in women and lung cancer is the leading cause of cancer death. Polychemotherapy reduced the 15-year breast cancer mortality in younger women (<50 years) from 42% to 32% (ARD 10%, RRR 24%) but in older women only from 50% to 47%. ⁵⁶ Tamoxifen added to polychemotherapy reduced the 15-year mortality in oestrogen receptor-positive patients from 35% to 26%. ⁵⁶ In the study by Bria et al, adjuvant chemotherapy led to a small reduction of 5-year lung-cancer mortality (ARD 3%, RRR 9%), ^{Reference Bria, Gralla, Raftopoulos, Cuppone, Milella and Sperduti57} confirming a landmark previous meta-analysis. ⁵⁸

Infectious diseases: antibiotics for rhinosinusitis, otitis media, uncomplicated cystitis and prophylaxis of wound infection after surgery

The effects of antibiotics depend on the infection. We did not find meta-analyses on severe infections such as pneumonia or on antivirals (monotherapy v. placebo) for HIV. A meta-analysis concluded against their general use in rhinosinusitis owing to small effect size (response: placebo 57%, drug 64%, ARD 7%, RRR 13%). ^{Reference Young, De Sutter, Merenstein, van Essen, Kaiser and Varonen59} The use of antibiotics in otitis media is debated, as within 2–7 days 78% of patients recovered spontaneously compared with 84% taking antibiotics (ARD 6%, RR 28%). ^{Reference Glasziou, Del Mar, Sanders and Hayem60} In contrast, the efficacy in uncomplicated cystitis (response: placebo 26%, drug 62%) and for the prophylaxis of wound infections after major operations (infections: placebo 39%, antibiotics 10%) was clear. ^{Reference Falagas, Kotsantis, Vouloumanou and Rafailidis61,Reference Nelson, Glenny and Song62}

Schizophrenia: antipsychotics for reduction of overall symptoms and relapse prevention

In acute treatment, second-generation antipsychotics increased the percentage responding from 24% with placebo to 41% (ARD 18%, RR 70%), and reduced the Brief Psychiatric Rating Scale/Positive and Negative Syndrome Scale total score by 9/10 points (SMD 0.51). ^{Reference Leucht, Arbter, Engel, Kissling and Davis4} Antipsychotic maintenance treatment reduced relapse rates from 57% to 22% within approximately 10 months (ARD 38%, RRR 65%). ^{Reference Leucht, Komossa, Heres, Kissling and Davis63}

Bipolar disorder: mood stabilisers for acute mania, antidepressants for depression and mood stabilisers for relapse prevention

Various antimania agents increased the percentage responding from approximately 30% with placebo to approximately 50% within 3 weeks (response to lithium 52% v. placebo 34%, ARD 17%, RR 50%; ^{Reference Storosum, Wohlfarth, Schene, Elferink, van Zwieten and van den Brink64} response to valproate 47% v. placebo 21%, ARD 27%, RR 150%; ^{Reference Smith, Cornelius, Warnock, Tacchi and Taylor65} response to carbamazepine 51% v. placebo 26%, ARD 25%, RRR 100%; ^{Reference Smith, Cornelius, Warnock, Tacchi and Taylor65} response to antipsychotics 50% v. placebo 31%, ARD 20%, RR 60%). ^{Reference Scherk, Pajonk and Leucht66} In bipolar depression, antidepressants increased the response rate from 34% to 58% (ARD 34%, RR 130%). ^{Reference Gijsman, Geddes, Rendell, Nolen and Goodwin67} In maintenance treatment, lithium reduced relapse rates from 81% to 36% (ARD 53%, RRR 51%), ^{Reference Davis, Janicak and Hogan68} or from 61% to 40% after excluding studies in which lithium was suddenly discontinued (ARD 24%, RRR 35%). ^{Reference Geddes, Burgess, Hawton, Jamison and Goodwin69}

Major depressive disorder: antidepressants for acute depression and relapse prevention

The absolute responder differences in recent meta-analyses of various selective serotonin reuptake inhibitors (SSRIs) (or tricyclic antidepressants used as an active comparator in SSRI v. placebo studies) ^{Reference Storosum, Elferink, van Zwieten, van den Brink, Gersons and van Strik70} v. placebo in major depressive disorder were 10–15% (Table DS4). For example, paroxetine increased the percentage responding from 42% to 53% (ARD 10%, RR 20%) and reduced the Hamilton Rating Scale for Depression score by 3 points (SMD 0.32). ^{Reference Barbui, Furukawa and Cipriani71} These studies were currently primarily conducted in out-patients with less severe disorder (e.g. 90% of the sample were out-patients in the meta-analysis by Barbui et al). ^{Reference Barbui, Furukawa and Cipriani71}

Maintenance treatment with any antidepressant reduced the relapse rate from 41% with placebo to 18% (ARD 23%, RRR 58%), ^{Reference Geddes, Carney, Davies, Furukawa, Kupfer and Frank72} consistent with another meta-analysis restricted to new antidepressants (placebo 48% v. drug 26%, ARD 22%, RRR 44%). ^{Reference Hansen, Gartlehner, Lohr, Gaynes and Carey73}

Obsessive–compulsive disorder: effects of SSRIs on responder rates and overall symptoms

Selective serotonin reuptake inhibitors increased the proportion of patients responding in the acute phase from 23% to 43% (ARD 20%, RRR 84%). ^{Reference Soomro, Altman, Rajagopal and Oakley-Browne74} These drugs reduced the mean Yale–Brown Obsessive Compulsive Scale score by 3.2 points (SMD 0.44). ^{Reference Soomro, Altman, Rajagopal and Oakley-Browne74}

Panic disorder: tricyclic antidepressants, SSRIs and benzodiazepines for anxiety symptoms

The mean SMDs (raw differences in rating scale scores or responder rates were not indicated) of tricyclic antidepressants, SSRIs and benzodiazepines in acute treatment were 0.40–0.41. ^{Reference Mitte75}

Alzheimer's disease: cholinesterase inhibitors for prevention of cognitive decline

Within 6 months, cholinesterase inhibitors increased the percentage of participants unchanged or improved from 17% to 24% (ARD 7%, RRR 43%). ^{Reference Birks76} The cognitive subscore of the Alzheimer's Disease Assessment Scale was better by 2 points (SMD 0.41). ^{Reference Birks76}

Attention-deficit hyperactivity disorder: effects of various drugs on symptoms

Methylphenidate (SMD 0.78), amphetamines (SMD 1.00) and atomoxetine (SMD 0.64) showed robust effect sizes in overall reduction of attention-deficit hyperactivity disorder symptoms (raw differences in rating scale scores or responder rates were not indicated). ^{Reference Schachter, Pham, King, Langford and Moher77–Reference Cheng, Chen, Ko and Ng79}