Table 9Systematic Reviews Examining the Adverse Effects of Treatment (KQ5)

Drugs
Study, Year, Quality
Intervention;
Population
Total withdrawals;
Withdrawals due to adverse events
Adverse events:
Intervention group
Conclusions
α-glucosidase inhibitors
Van de Laar et al, 2005194
Fair
α-glucosidase inhibitors:
acarbose (30 studies); miglitol (7 studies); voglibose (1 study) + 3 studies combined various

DM2
NRAcarbose:
Any diabetes-related endpoint: RR 1.00 (0.81–1.23) vs placebo
Microvascular disease: RR 0.91 (0.61–1.35) vs placebo
Number of patients with side effects, odds ratio treated vs placebo; 3.37 (95% CI, 2.60 – 4.36)
NSD between acarbose and placebo with respect to morbidity and mortality
Van de Laar et al, 2007193
Good
α-glucosidase inhibitors:
acarbose (5 studies)

IGT and IFG
NRGastrointestinal (flatulence, diarrhea): OR 3.5 (2.7–4.4) vs placeboAcarbose causes significant gastrointestinal side effects compared to placebo
ACE inhibitors and ARBs
McDonald et al, 2005197
Good
ARBs

At risk for CV events
NRMI pooled effect: OR=0.94 (0.75 – 1.16)ARBs are not associated with an increased risk of MI when compared with placebo.
Strippoli et al, 2006,195
2004196
Fair
ACE inhibitor
ARBs
placebo

DM1: 20 studies
DM2: 23 studies
Mixed DM population: 6 studies
Total withdrawals:
0.2 to 1.0%

Withdrawal due to AEs: NR
ACE inhibitors, I vs C:
All-cause mortality (any dose) 12.3%; 12.7% (p>0.05)
CV mortality 5.8%; 5.9%(p=0.6)
Doubling of serum creatinine 3.0%; 4.3% (p=0.05)
End-stage kidney disease 0.85%; 1.5% (p=0.02)
Cough (vs placebo): 3.17 (2.29, 4.38)
Hyperkalemia: NSD vs placebo

ARBs:
All-cause mortality 13.7%; 15.6% (p=0.9)
Doubling of serum creatinine 15.1%; 21.5% (p=0.004)
End stage kidney disease 13.3%; 19.3% (p=0.001)
Cough (vs placebo): 4.93 (1.00, 24.35)
ACE inhibitors vs ARBs: Based on indirect analysis no significant differences for any outcome, including: all-cause mortality, end-stage renal disease, doubling of serum creatinine concentration, progression from microalbuminuria to macroalbuminuria or regression from microalbuminuria to normoalbuminuria.

ACE inhibitors or ARBs vs placebo: All-cause mortality: ACE inhibitors, but not ARBs, were associated with a significant reduction in all-cause mortality; end-stage renal disease and doubling of serum creatinine concentration: weak evidence of reduced risk with ACE inhibitor use with no significant difference in risk for ARBs; both ACE inhibitors and ARBs associated with significantly reduced risk of progression from microalbuminuria to macroalbuminuria and increased rate of regression from microalbumunuria to normoalbuminuria.
Velazquez-Armenta et al, 2007198
Fair
ARBs

Pregnancy
NR (case series)Favorable pregnancy outcomes: 57.8% (37 cases)

Unfavorable pregnancy outcomes (eg: abnormalities including limb and face deformations, enlarged kidneys, anuria, severe hypotension, etc): 42.2% (27 cases) ARBs in this group included valsartan, losartan, candesartan, and irbesartan

Duration of treatment during pregnancy among women who had adverse fetal outcomes was 26.3±10.5 weeks vs 17.3±11.6 weeks for those who had favorable outcomes (p=0.04)
Exposure to ARBs for a period longer than the first trimester of pregnancy appears to be associated with an increased risk of adverse fetal outcomes (p=0.04)
Verdecchia et al, 2005199
Fair
ARBs

At risk for CV events
NRMI: OR 0.96 (95% CI, 0.84 – 1.10), p=0.57

CVD mortality: OR 0.91 (95% CI, 0.83 – 0.99), p=0.042
ARBs are not associated with an increased risk of MI when compared with placebo.
Aspirin
Berger et al, 2006201
Good
Aspirin

Primary prevention of cardiovascular events
NRBleeding in men: OR 1.72 (1.35 – 2.20; p<0.001)
Bleeding in women: OR 1.68 (1.13 – 2.52; p=0.01)
Stroke in men: OR 1.13 (0.96 – 1.33)
Stroke in women: OR 0.83 (0.70 – 0.97)
Ischemic stroke in men: OR 1.00 (0.72 – 1.41)
Ischemic stroke in women: OR 0.76 (0.63 – 0.93)
Hemorrhagic stroke in men: OR 1.69 (1.04 – 2.73)
Hemorrhagic stroke in women: OR 1.07 (0.42 – 2.69)
Reduced risk of CV events for men and women with aspirin use; significant increase in bleeding risk for both groups; NSD in CV or all-cause mortality
McQuaid et al, 2006200
Good
Aspirin or Clopidogrel

For cardiovascular prophylaxis
Aspirin:
All events: RR=1.16 (05% CI, 0.94 – 1.44)
GI events: RR=1.26 (0.94 – 1.70)
non-GI events: RR=0.84 (0.55 – 1.28)
Aspirin :
Major bleeding: RR=1.71 (95% CI, 1.41 – 2.08)
Major GI bleeding: RR=2.07 (1.61 – 2.66)
Intracranial bleeding: RR=1.65 (1.06 – 5.99)
Dyspepsia: RR=1.09 (0.97 – 1.22)
Diarrhea: RR=3.30 (1.42 – 7.66)
Constipation: RR=1.98 (1.14 – 3.44)
Rash: RR=0.77 (0.38 – 1.58)

769 patients need to be treated with aspirin to cause 1 additional major bleeding episode annually

No study compared clopidogrel with placebo
Low-dose aspirin associated with an increase in risk of major bleeding (~70%; NNT: 796) relative to placebo/no use
Compared to clopidogrel, aspirin associated with higher risk of GI bleeding (NNT 883 to prevent one major GI bleeding episode)
Beta-blockers
Wiysonge et al, 2007202
Good
Beta-blocker (not stratified; including atenolol, propranolol, oxeprenolol, metoprolol)
Placebo

Hypertension, >18 years
Total withdrawals NR

Withdrawals due to AEs I vs C
18.2% vs 8.6%; p=0.1
RR 2.34 (0.84–6.62)
I vs C:
Total mortality 5.0%; 5.2% (p=0.8)
CHD 3.5%; 3.7% (p=0.3)
Stroke 1.8%; 2.3% (p=0.02)
CV mortality 2.6%; 2.9% (p=0.4)
CV disease 5.7%; 6.4% (p=0.01)
No significant difference between beta-blockers and placebo in total mortality or CHD. Use of beta-blockers was associated with a significantly lower risk of stroke and CV events, relative to placebo.
Hypoglycemic agents
Bolen et al 2007203
Good
Various oral hypoglycemic agents:
pioglitazone
rosiglitazone
metformin
sulfonylureas
repaglinide
nateglinide
acarbose
placebo

DM2
Total withdrawals: I vs C
Pioglitazone: NR
Rosiglitazone 3.8–6.3% vs 2.7–12.0%
Metformin NR
Sulfonylurea 2.4% vs 7.9% (1 study)
Meglitinide (repaglinide or nateglinide) NR
Acarbose NR

Withdrawals due to AEs: I vs C
Pioglitazone 1.1–3.0% vs 2.4–4.8%
Rosiglitazone 0.9%–7.4% vs 1.2–10.3%
Metformin 3.0–15.4% vs 0–17.2%
Sulfonylurea 0–14.3% vs 1.9–30.4%
Meglitinide (repaglinide or nateglinide) 1.5–7.6% vs 3.0–4.3%
Acarbose 2.5% vs 5.3% and 58.1% vs 44.8% (2 studies; rate varied widely)
I; C
Pioglitazone:
Hypoglycemia 0.6–11.0%; 0–11%
Edema 3.0–13.6%; 0–7.5%
CHF 3.6–14.0%; 0.6–16.0%
ALT elevations 0–6%; 0–6.0%
AST elevations 0–1%; 1%

Rosiglitazone:
Hypoglycemia 3.4–12%; 2.0–6.0%
Edema 6.0–18.0%; 3%
CHF 4.1–13.6%; 0–2.5%
ALT elevations 0–1.2%; 0–1.1%

Metformin:
Mortality (1 study) 0.3%; 0%
Hypoglycemia 1.3–13.4%; 0–10.3%

Sulfonylurea:
Hypoglycemia 0–17.7%; 0–1.2%
CHF 4.2%; 3.5% (1 study each)

Meglitinide (repaglinide or nateglinide):
Hypoglycemia 0–12.8%; 0–11.0%

Acarbose:
Hypoglycemia 9.7%; 10.3% (1 study each)
No clear conclusions regarding all-cause mortality associated with metformin + second generation sulfonylurea vs metformin and/or a second generation sulfonylurea could be drawn due to conflicting results and/or lack of evidence.

The effect of metformin + second generation sulfonylurea vs metformin or a second generation sulfonylurea on CV mortality was unclear; other oral diabetes medications lack adequate evidence to draw conclusions

No conclusions can be made regarding CV morbidity due to limited number of studies; pioglitazone+metformin associated with improved CV morbidity relative to placebo/diet
Gangji et al, 2007204
Good
Glyburide, other secretagogues, insulin

DM2
NR; loss to follow-up ranged from 0 to 37%Glyburide compared to other secretagogues
Hypoglycemia: RR 1.52 (1.21–1.92); compared to other sulfonylureas, RR 1.83 (1.35–2.49))
Cardiovascular risk: RR 0.84 (0.56–1.26)
Death: RR 0.87 (0.70–1.07)
Glyburide caused more hypoglycemia than other secretagogues and other sulfonylureas, but was not associated with increased risk of cardiovascular events or death.
Saenz et al, 2005205
Good
Metformin

DM2
NRMetformin; comparator
All-cause mortality: 0.51%; 0.0% (p=0.3)
Hypoglycemia: 2.7%; 0.5% (p=0.2)

No cases of lactic acidosis
Pooled data from trials of various active interventions, placebo and/or diet changes found no difference in rates of all-cause mortality or ischemic heart disease.
Salpeter et al 2003207, 2006206
Good
Metformin

DM2
NRFatal or non-fatal lactic acidosis: 0%
Estimated upper limit 95% confidence interval for incidence of lactic acidosis metformin vs non-metformin (cases/100,000 patient-years):
6.3 vs 7.8

No other AEs reported

Control group: 0% with various hypoglycemic agents as comparators
No evidence of an association between metformin use and lactic acidosis relative to other anti-hyperglycemic agents
Setter et al, 2003208
Poor
Metformin

DM2
Unable to tolerate as a result of prolonged adverse effects: <5%Episodes of severe hypoglycemia: ‘negligible’ (no other data)

Lactic acidosis: rate 8 cases/100,000 person-years (1 study)
Very limited data found that potentially fatal lactic acidosis can be associated with metformin use, although absolute risk is low.
Statins
Bonovas et al, 2007209
Fair
Pravastatin

Cardiovascular therapy for different ages
NRCancer risk: random-effects model (RR 1.06 (95% CI, 0.97 – 1.14))
Cancer risk as age increases: meta-regression, p=0.006
Possible association between pravastatin use and increased cancer risk in the elderly.
Findings need to be replicated.
Law et al, 2006211
Fair-Poor
Various statins

Those prescribed statin treatment (details NR)
NRPeripheral neuropathy (OR from 4 cohort studies): 1.8 (1.1 – 3.4)

Rhabdomyolysis: Incidence per 100,000 person years Cohort studies:
Cervistatin: 46 (13 – 120)
Statins (without cervistatin): 1.6 and 6.5 (2 studies)
Gemfibrozil: 28 (6–81)

FDA Reporting System:
Cervistatin: 21 (19 – 25)
Statins (without cervistatin): 0.70 (0.62 – 0.79)

Mortality estimated at 10% of incidence

Treated minus placebo, Per 100,000 person years
Rhabdomyolysis: 1.6 (-2.4 – 5.5)
Myopathy: 5 (-17 – 27)
Minor muscle pain: 190 (-38 – 410)
Elevated Creatine kinase: 23 (-4 – 50)
Elevated ALT (single measure): 100 (64 – 140)
Elevated ALT (2 consecutive measures): 70 (50 – 90)
Despite high risk with cervistatin, incidence of rhabdomyolysis is low in patients taking simvastatin, lovastatin, atorvastatin, provastatin, or fluvastatin - estimated as 3 per 100,000 person-years. Myopathy attributable to these statins is also rare (11 per 100,000 person years). Most muscle symptoms in patients taking statins are not attributable to the statins.
McClure et al, 2007210
Good
Statins

Those prescribed statin treatment (details NR)
Discontinuation due to AEs: OR (95% CI)
Overall (w/o cervastatin) : OR 0.88 (0.84 – 0.93)
Lovastatin: 1.10 (0.98 – 1.24)
Pravastatin: 0.79 (0.74 – 0.84)
Simvastatin: 1.00 (0.89 – 1.11)
Fluvastatin: 0.93 (0.75 – 1.16)
Atorvastatin: 0.93 (0.75 – 1.14)
Rosuvastatin: 0.68 (0.26 – 1.77)
Cervastatin: 1.45 (0.98 – 2.16)
OR (95% CI)
Rhabdomyolysis (w/o cervistatin): 1.59 (0.54 – 4.70)
Myositis (w/o cervistatin): 2.56 (1.12 – 5.85)
Myositis (cervistatin): 3.36 (0.59 – 19.3)
Creatine kinase (w/o cervistatin): 1.11 (0.78 – 1.59)
Creatine kinase (cervistatin): 2.93 (1.08 – 7.92)
Myalgia (w/o cervistatin): 1.09 (0.97 – 1.23)
Myalgia (cervistatin): 1.74 (0.51 – 5.91)
Overall, discontinuation of statin therapy was no worse than placebo. Risks of muscle related AEs in agreement with known risks of statins; rates are much higher with ceruvistatin than other statins.
Silva et al, 2006212
Fair
Statins

Those prescribed statin treatment or placebo
NRRisk of any AE: OR 1.4 (1.09 – 1.80), p=0.008 vs placebo, NNH 197
Risk of clinical CV event: OR 0.74 (0.69 – 0.80), p<0.001, NNT = 27
Treating 1000 pts with statin would prevent 37 CV events, and 5 AEs would be observed.
Serious events (creatine kinase > 10× upper limit of rhabdomyolysis) are infrequent, NNH = 7428
Nonurgent AEs (myalgia and liver function tests) responsible for 2/3 of AEs reported in trials: 0.48 (0.25 – 0.7), NNH = 209

Rate of liver failure: 1 per 100,000 person years of statin use.

Person years for any event/serious event:
Placebo: 181/48
Statin therapy in associated with greater odds of AEs compared with placebo, but with there is also substantial clinical benefit. Similar rates of serious AEs was observed between statins and placebo.
Thiazolidinediones
Norris et al, 2006213
Good
Pioglitazone (pio) 7.5–45 mg qd
Rosiglitazone (rosi) 4–12 mg qd

DM2, pre-DM, the metabolic syndrome
Total withdrawals, I v C (placebo):
pio: 7.0–33.0% v 2.4–20.0%; pooled RD v placebo -1.0% (-3.0 – 1.0%)
rosi: 0–27.0% v 0–38.4%; pooled RD v placebo -3.0% (-9.0 – 2.0%)

Withdrawals due to AEs, I v C (placebo):
pio: 4.8% v 4.5%; pooled RD 0% (-2.0 – 2.0%)
rosi: 4.9% v 7.2%; pooled RD v placebo -2% (-4% – -1%)
Thiazolidinedione; placebo
Pioglitazone:
Cardiac-related events: 3.6%; 6.3%
CHF: 11.0%; 8.0% (p<0.05)
Peripheral edema: 0–22.0%; 0–16.0%
Abnormal LFT: 0.77%–2.4%; 1.3%
Hypoglycemia: 0–28.0%; 0–20.0%

Rosiglitazone
Peripheral edema: 4.1–6.6%; 1.6% (p<0.05 (4mg bid dose only, rosiglitazone rate 6.6%)
Abnormal LFT: 0–0.6%; 0.0%
Total withdrawals and withdrawals due to AEs were similar in each of the rosi, pio, and placebo groups.
The incidence of edema was significantly greater in both rosi and pio, than placebo.
The risk difference for hypoglycemic events between placebo and each of rosi and pio was not significant.
Weight gain was greater with both rosi and pio compared to placebo.
Richter et al, 2007215
Fair
Pioglitazone

RCTs in adults with DM2 and trial duration ≥ 24w
Total withdrawals: NR% drop-outs due to AEs; similar between pio and comparatorsDecrease in A1c: consistent in 6 studies which examined this outcome compared to : range 0.5 to 0.75 g/dl
Body weight: increased in 15 studies compared to various comparators: up to 3.9 kg
Hypoglycemic episodes (%): somewhat lower rates with pio than various active controls
Edema: relative risk pio vs various other comparators: 2.86 (95% CI, 2.14 – 2.52)
Pioglitaone appears to decrease A1c, increase body weight, and increase edema compared to various other active therapies or placebo.
Richter et al, 2007214
Fair
Rosiglitazone

RCTs in adults with DM2 and trial duration ≥ 24 weeks
Total withdrawals: NR
Withdrawals due to AEs: I 2.7 to 11.6%, C: 2.0 to 14.9% (no pooled estimates available; no between-group-values available)
Edema: OR 2.27 (95% CI, 1.83 – 2.81)

Fractures, CVD events, CHF, PVD, mortality: data reported from the ADOPT trial only

Severe hypoglycemic episodes: I 0–5.4%, C 0–2.9%; no pooled data and no statistics
Rates of edema are increased with rosiglitazone compared with various other drugs or placebo. The ADOPT trial suggests that fractures rates in women may be increased.
Singh et al, 2007216
Fair
Rosiglitazone

RCTs in DM2 or IGT and trial duration ≥ 12 months
NRRelative risk 95% CI) rosiglitazone vs comparator:
MI: 1.42 (1.06 – 1.91)
Heart failure: 2.09 (1.52 – 2.88)
CV mortality: 0.90 (0.63 – 1.26)
Rosiglitazone use for 12 or more months increases the risk of MI and heart failure, without significantly increasing the risk of CV mortality.
Weight loss drugs
Li et al, 2005218
Good
Sibutramine, phentermine, diethylpropion, orlistat, fluoxetine, bupropion, topiramate, sertraline, zonisamide

Those prescribed obesity management treatment
NRPooled OR (95% CI):
Orlistat:
Diarrhea 54.85 (44.88 – 67.48)
Flatulence 3.72 (3.16 – 4.39)
Bloating, abdominal pain, dyspepsia 1.55 (1.18 – 2.06)
Headache 1.18 (0.68 – 2.05)
Fluoxetine:
Nervousness, sweating tremors 7.85 (3.87 – 17.63)
Nausea, vomiting 3.27 (1.94 – 5.67)
Fatigue, asthenia, hypersomnia, somnolence 2.83 (1.82 – 4.45)
Insomnia 2.19 (1.10 – 4.58)
Diarrhea 1.86 (1.10 – 3.23)
Urticaria, pruritus, rash 1.67 (0.53 – 5.65)
Headache 1.35 (0.91 – 2.03)
Bupropion:
Dry mouth 3.26 (1.71 – 6.64)
Diarrhea 1.37 (0.52 – 4.01)
Constipation 1.31 (0.72 – 2.44)
Upper respiratory problems 1.22 (0.88 – 1.69)
Topriamate:
Paresthesia 20.18 (13.99 – 29.67)
Taste perversion 11.14 (2.80 – 23.57)
Central nervous system effects 3.97 (2.90 – 5.49)
Constipation 3.96 (1.77 – 9.77)
Dry mouth 3.13 (1.59 – 6.55)
Upper abdominal symptoms 1.76 (1.27 – 2.47)
Fatigue 1.36 (1.03 – 1.80)
Upper respiratory problems 1.32 (0.87 – 1.99)
Sibutramine: Effects on BP varied; A1c and fasting blood glucose decreased; heart rate was consistently elevated by 4 beats per minute.

Orlistate was associated with diarrhea, abdominal pain, and dyspepsia; it was unclear if these improved over time.

Fluoxitine: nervousness, sweating, tremors, nausea and vomiting, and insomnia increased significantly compared with placebo.

There were few studies with long-term adverse effects data.
Norris et al, 2005217
Good
Fluoxetine, orlistat, sibutramine
Placebo

DM2
Total withdrawals NR

Withdrawals due to AEs fluoxetine v control:
1–9% v 0–2%
orlistat v control:
0.3–22% v 0.5–28%
sibutramine:
3–7% v 0%(1 study)
Data based on 1 study (no pooled data available)

Orlistat; placebo
Hypoglycemia: 7–17%; 3–10.0%
GI events: 65–80%; 27–62%

Fluoxitine; placebo
Tremor: 5–15%; 0–3%
Somnolence: 11–22%; 4–7%
Sweating: 28%; 11%

Sibutramine; placebo
Palpitations 41%; 29%
Dry mouth: 38%; 223%
Gastrointestinal adverse effects were common with orlistate; tremor, somnolence, and sweating with fluoxetine; and palpitations with sibutramine.

Abbreviations: ACE, Angiotensin-converting enzyme; ADOPT, A Diabetes Outcomes Progression Trial; AE, adverse event; ALT, Alanine aminotransferase; ARBs, Angiotensin II Receptor Blockers; AST, Aspartate aminotransferase; bid, twice daily; C, control group; CHD, coronary heart disease; CHF, congestive heart failure; CI, confidence interval; CV, cardiovascular; CVD, cardiovascular disease; DM, diabetes; DM1, type 1 diabetes; DM2, type 2 diabetes; FDA, Food and Drug Administration; GI, gastrointestional; I, intervention group; IFG, impaired fasting glucose; IGT, impaired glucose tolerance; LFT, liver function tests; MI, myocardial infarction; NNH, number needed to harm; NNT, number needed to treat; NR, not reported; NSD, no significant difference; OR, odds ratio; pio, pioglitazone; PVD, peripheral vascular disease; qd, daily; RCT, randomized controlled trial; RD, risk difference; rosi, rosiglitazone; RR, relative risk; y, years.

From: Summary Tables

Cover of Screening for Type 2 Diabetes Mellitus: Update of 2003 Systematic Evidence Review for the U.S. Preventive Services Task Force
Screening for Type 2 Diabetes Mellitus: Update of 2003 Systematic Evidence Review for the U.S. Preventive Services Task Force [Internet].
Evidence Syntheses, No. 61.
Norris SL, Kansagara D, Bougatsos C, et al.

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.