REVIEW
JNEPHROL 2007; 20 (suppl 12): S1-S3
www.sin-italy.org/jnonline – www.jnephrol.com
The impact of meta-analyses on clinical practice:
the benefits
Alberto Morganti
ABSTRACT
Meta-analyses are frequently criticized because in most
cases they are compiled from quite heterogeneous studies. In spite of this limitation meta-analyses are increasingly published because in many areas of clinical research the results of individual studies are devoid of statistical power and end up with conflicting results. Metaanalyses, if performed with a rigorous and exhaustive
search of all accountable information on a specific topic,
have the potential of overcoming the drawbacks of single studies and, in addition, of adjusting for publication
bias and interstudy variability. These strengths of metaanalyses can be exploited to provide conclusive answers
on diagnostic and therapeutic issues being debated,
which in turn may help guide doctors toward more rational decisions.
Key words: Benefits, Clinical trial, Hypertension, Meta-
analysis
INTRODUCTION
In discussing the benefits of meta-analyses it is only fair to
acknowledge in the first place that the scientific value of this
method of a posteriori analysis is far from being unanimously
recognized. Indeed meta-analyses are frequently blamed for
“lumping together apples and oranges,” alluding to the heterogeneity of the individual studies used for their compilation,
while their authors are sometimes accused of exploiting the
work of other, more creative investigators. Yet, despite these
harsh criticisms, meta-analyses are increasingly appreciated
by readers of scientific journals, as can be seen from the fact
that, according to a Medline search, the number of those
Chair of Internal Medicine and Hypertension Center,
Ospedale San Giuseppe Fatebenefratelli - Milano Cuore;
Center of Clinical Physiology and Hypertension,
IRCCS Ospedale Maggiore, University of Milan, Milan - Italy
published rose steadily during the years 2000 to 2005. The
most likely reason for the increasing success of meta-analyses is that in some specific areas the results of single trials
may not provide sufficient evidence, or their results are even
conflicting. Thus meta-analyses are often performed to overcome the limits of individual trials and to generate a consensus on issues under debate. Therefore, in this view, they may
indeed represent a useful support to clinical research. Herein,
I will briefly summarize the strengths of meta-analyses and
present a few examples of some recent ones which, in my
opinion, have contributed to solve some unsettled questions
in the diagnosis and treatment of hypertensive patients.
DEFINITION AND POTENTIALITIES
OF META-ANALYSES
Most pitfalls of meta-analyses derive from the inaccuracy
and incompleteness of data collection; these limitations
could be overcome if meta-analyses were carried out
according to the precise and comprehensive definition
recently proposed by Schork (1): “A meta-analysis is a
review of all the literature on a specific topic which includes
statistical accountability of interstudy variability; a quantitative rigorous investigation of published summary information
from an exhaustive list of studies relevant to a medical
issue.” When the requirements outlined in this definition are
satisfied, the meta-analysis may indeed become a powerful
tool because the combination of numerous studies enhances
their statistical power; in addition, meta-analyses can adjust
for publication bias and for the characteristics of various
studies and provide a complete and easy instrument for
review and comparison. A list of the topics for which the
meta-analytic approach may be particularly useful is shown
in Table I.
S1
Morganti: Benefits of meta-analyses
EXAMPLES OF BENEFIT OF META-ANALYSES
IN HYPERTENSION
The guidelines on the diagnosis and treatment of high
blood pressure recently published by the European
Society of Cardiology and the European Society of
Hypertension (2) represent a clear demonstration of the
impact that a meta-analysis may have on a major issue of
public health. Indeed the information gathered by the
Prospective Studies Collaboration (3) from 61 observational studies of blood pressure and mortality encompassing more than 1 million adults has allowed us to establish,
within each decade of age from 40 to 89 years, a strong
and direct relationship between usual blood pressure values and overall cardiovascular mortality, without any evidence of a threshold up to at least 115/75 mm Hg.
According to the evidence provided by this huge amount
of data the classification of hypertension stages was
revisited and the threshold of the “optimal” blood pressure
values lowered to 120/80 mm Hg. It is obvious that the
medical, social and economical consequences deriving
from the recommendations generated by this meta-analysis will be immense.
Another issue being debated in hypertension is whether
the protection from specific complications of high blood
pressure (cardiovascular vs. cerebrovascular) is entirely
attributable to the degree of reduction of blood pressure
or rather to the ancillary properties of some antihypertensive agents. In this respect Verdecchia et al (4) recently
reviewed 28 trials comparing either angiotensin-converting enzyme inhibitors (ACEIs) or calcium channel blockers
(CCBs) vs. diuretics, beta-blockers or placebo for a total
of 117,122 patients; overall there were 9,509 incident
TABLE I
CLINICAL TOPICS FOR WHICH META-ANALYSIS MAY BE
USEFUL
• Provide solid evidence for preparing guidelines
• Justify the use of complex methodologies
• Assess the validity and relevance of small mechanistic
studies
• Quantify the benefits deriving from specific treatment
approaches
• Answer doubts raised by incomplete and subjective analysis
of individual trials
• Guide toward selection of the best drugs
S2
cases of coronary heart disease and 5,971 cases of
stroke. The potential source of the significant heterogeneity between trials was investigated by the metaregression
technique. This meta-analysis has clearly shown that all
active treatments are better than placebo for protecting
patients from hypertension complications and, also more
important, that ACEIs are superior to CCBs for prevention
of coronary heart disease, whereas the opposite is true for
the prevention of stroke. These findings will be very relevant for guiding the therapeutic decisions of clinicians in
everyday life.
Meta-analyses are also useful to provide clinicians with
reliable quantification of the effects of some consumed
substances or of some nonpharmacological treatments
on blood pressure. As to the first issue, a large number of
people drink coffee on a daily basis, and coffee consumption is known to influence blood pressure, but little
is known about the long-term effects of caffeine ingestion. Noordzij et al (5) addressed this issue reviewing a
total of 66 randomized controlled studies with a total of
1,010 participants; the meta-analysis was performed
using a random-effects model and examining possible
publication bias. By this approach, authors were able to
establish that coffee-drinking significantly increases both
systolic and diastolic blood pressure, by 2.0 and 0.7 mm
Hg, respectively.
Individual studies on the effect of weight loss have generated inconsistent results because of the interstudy differences in baseline blood pressure and body weight. Neter
et al (6) carried out a meta-analysis of 25 randomized
studies with a total of 4,874 participants and were able to
determine that the loss of 1 kg is associated with a reduction of systolic and diastolic pressure of about 1 mm Hg.
These results are particularly relevant for providing a scientific basis for weight loss as an important component of
a nonpharmacological blood pressure control program.
Modern medicine is facing a generalized restraint on economical resources, and doctors are more and more often
requested to justify the use of the sophisticated but
expensive procedures that technology has made possible.
For instance, the evaluation of blood pressure with modern devices which measure it throughout the 24 hours
(ambulatory blood pressure monitoring, ABPM) has, in
theory, several advantages over conventional, office measurements. This may be particularly true for evaluating the
efficacy of antihypertensive treatment, but until recently, it
was unclear whether, and to what extent, the changes in
blood pressure estimated with the 2 methods differed.
Mancia and Parati (7), using a Medline search, identified
JNEPHROL 2007; 20 (suppl 12): S1-S3
44 papers which satisfied the rigorous predetermined
inclusion criteria of their meta-analysis and found that
treatment-induced reduction in blood pressure is smaller
for the 24-hour ABPM than for the office systolic and diastolic blood pressure measurements, with an average ratio
ranging from 0.67 to 0.75. These findings are important for
the interpretation of interventional trials in which the effect
of antihypertensive treatment is based solely on office
blood pressure and provide a justification for a more
extensive use of ABPM.
Finally, meta-analysis may contribute to dismantling the
diffidence and concern raised in patients and doctors by
incomplete or distorted interpretation of individual trials.
This was the case for the editorial by Verma and Strauss
recently published in the BMJ (8) where the authors, on
the basis of a partial analysis of some selected trials, concluded that angiotensin receptor blockers (ARBs) may
increase the rate of myocardial infarction in hypertensive
patients. However, 2 comprehensive and updated metaanalyses (9, 10) taking into account all major randomized
trials in which ARBs were compared with conventional
treatments failed to find any significant difference in both
fatal and nonfatal myocardial infarction; and similar negative results were obtained when considering only trials in
which ARBs were compared with ACEI-based treatment.
REFERENCES
1. Schork MA. Publication bias and meta analysis. J
Hypertens 2003; 21: 243-5.
2. European Society of Hypertension (ESH) / European Society
of Cardiology (ESC) guidelines for the management of arterial hypertension. J Hypertens 2003; 21: 1011-53.
3. Prospective Studies Collaboration. Age-specific relevance
of usual blood pressure to vascular mortality: a meta-analysis of individual data for one million adults in 61 prospective
studies. Lancet 2002; 360: 1903-13.
4. Verdecchia P, Reboldi G, Angeli F, Gattobigio R, Bentivoglio
M, Thijs L, Staessen JA Porcellati C. Angiotensin converting
enzyme inhibitors and calcium channel blockers for coronary heart disease and stroke prevention. Hypertension
2005; 46: 386-92.
5. Noordzij M, Uiterwaal CS, Arends LR, Kok FJ, Grobbee DE,
Geleijnse JM. Blood pressure response to chronic intake of
coffee and caffeine: a meta-analysis of randomized con-
CONCLUSIONS
Admittedly meta-analyses lack originality and may suffer
from the heterogeneity existing among individual studies.
However, when compiled by competent and unbiased
researchers, meta-analyses may provide rational and synthetic information useful for guiding diagnostic and therapeutic decisions.
Conflict of interest statement: None declared.
Address for correspondence:
Prof. Alberto Morganti
Cattedra di Medicina Interna
U.O. di Medicina Interna
Centro Ipertensione Arteriosa
Ospedale San Giuseppe
Via San Vittore, 12
I-20123 Milano, Italy
alberto.morganti@unimi.it
trolled trials. J Hypertens 2005; 23: 921-8.
6. Neter JE, Stam BE, Kok FJ, Grobbee DE, Geleijnse JM.
Influence of weight reduction on blood pressure: a metaanalysis of randomised controlled trials. Hypertension 2003;
42: 878-84.
7. Mancia G, Parati G. Office compared with ambulatory blood
pressure in assessing response to antihypertensive treatment: a meta-analysis. J Hypertens 2004; 22: 435-45.
8. Verma S, Strauss M. Angiotensin receptor blockers and
myocardial infarction. BMJ 2004; 329: 1248-9.
9. Volpe M, Mancia G, Trimarco B. Angiotensin II receptor
blockers and myocardial infarction: deeds and misdeeds. J
Hypertens 2005; 23: 2113-8.
10. Verdecchia P, Angeli F, Gattobigio R, Reboldi GP. Do
angiotensin II receptor blockers increase the risk of myocardial infarction? Eur Heart J 2005; 26: 2381-6.
© Società Italiana di Nefrologia
S3