1. Concept
The systematic review of the scientific literature (SRSE) is a methodology that allows, in the most objective form, to know, on one hand, that a certain procedure - being preventive, diagnostic, therapeutic or of rehabilitation – has demonstrated scientifically to have a positive effect on the course of the disease to which it is applied – that is to say, its efficacy - and, for other hand, the risk for the person - or for the population - that this procedure can have- this is to say, its safety profile-. Besides the efficacy and the safety, its effectiveness can also be known, that is to say the effect of a certain procedure in the medical habitual practice and not only in the ideal conditions determined by the controlled and random studies. Likewise, other aspects so or more important than the related ones to the effect of the procedure, as its efficiency or its equity, also can be studied by means of the SRSE. The last aim of the SRSE is to inform the processes of decision in health with the intention of contributing to the improvement of the quality of health care.

What is checked systematically is the conjoint of scientific knowledge that appears, that is publicized, and that is shown to the community in relation to the procedure or technology that is object of study.

Without linguistic subterfuges, the intention of the SRSE is to systematically check all scientific evidence related to a certain health problem, which could be related to a clinical condition, technology or binomial clinical technology - condition. Besides, according to what is detailed below, this method has the property of being explicit, since the methodological process is detailed on how it comes to the conclusions of the SRSE, precisely in order to question o criticize its results.

2. Methods
The SRSE consists of six phases or stages (Table 1):

1) scientific evidence search
2) classification
3) presentation
4) synthesis.
5) integration.
6) conclusions and recommendations

In the seventh step it is necessary to include the dissemination of the conclusions and recommendations so that the aim of the SRSE can be fulfilled: it’s to say, to improve the quality of the process of health care provission. Besides, the dissemination of the conclusions and recommendations must serve also in order to be reproduced (as it corresponds to any scientific process) and, therefore, to confirm the validity of the results; or, in order that they can be questioned correcting mistakes or contributing new knowledge. In this respect, it is necessary to say that once recommendations have been expressed, via the SRSE, these do not last stable in the time. Therefore, the expiration of the results, for the constant advance of knowledge, is one of the limitations of the SRSE and that its solution is found in the periodic updates. The advantage is that to update the review it is only necessary to do it from the year in which the conclusions of the SRSE were publicized without having to repeat the whole process of review from the beginning.

Nevertheless, lets start from the beginning. The first phase of the SRSE is the scientific evidence search, which can be performed directly or indirectly. The scientific evidence direct search refers to the search of specific scientific studies that have been designed with the same aim as that done through the SRSE – for example, the study of the efficacy of a certain medicament-. The term indirect refers to the information that comes from clinical records, or from administrative, epidemiological or economic databases and that have been gathered with a different intention to the purpose of the SRSE study. For the search of the direct evidence one comes to the diverse databases of bibliographical information and to the gray literature so as to locate all the scientific studies recounted for a concrete procedure or for a health problem. Once identified the scientific studies it is necessary to proceed to its classification according to the scientific quality, that is to say, the classification of the levels of scientific evidence. The criteria that determine this quality come from the field of the epidemiology and take into account especially the design of the study. In this respect, the classification scales of the scientific evidence (Table 2) are generic instruments that constitute a tool to help this classification according to the quality and have as the aim to guarantee the internal validity of the study. Several tables exist that propose levels of scientific evidence, published in the international literature, however, Foundation Ostomía, is inclined to use one of the most sophisticated, elaborated by Jovell and cols., in the Spanish context.

TABLA 1

Stages of a systematic review of the scientific evidence (SRSE)
1. Search of the scientific evidence
	Direct: strategy of bibliographical search
	Indirect: strategy of local information search
2. Classification of the scientific evidence
	Generic instrument: scales of evaluation of the scientific evidence
	Specific instrument: quality of the individual studies
3. Presentation of the scientific evidence: tables or matrices of evidence
4. Synthesis of the scientific evidence: Meta-analysis
5. Integration of the scientific direct and indirect evidence
	Optimal Analysis
	Decision Analysis
	Economic Analysis
	Health Policy Analysis
	Social Impact Analysis
6.Production of recommendations based on scientific evidence
	Clinical practice
	Public health policy
	Clinical practice guidelines
	Clinical Protocols
Jovell AJ. The evaluation of Health technologies in the Primary Attention. MEDIFAM 1999-9:102-109

TABLA 2
Evaluation scale of the scientific evidence of the Agency d'Avaluació of Technology Medicates (AATM)

Levels	SE Quality	Type of study design	Conditions of scientific rigor*
I	Good	Meta-analysis of controlled clinical trials	Analysis of individuals patients informationMeta-regressionDifferent technological analysisAbsence of heterogeneity Quality of the studies
II		Controlled clinical trials of big sample	Evaluation of the statistical power Multicentred Quality of the study
III*	Good	Controlled clinical trials of smaller sample	Evaluation of the statistical powerQuality of the Multicentred studies
IV	Medium	Prospectives controlled non randomized studies	Evaluation of the statistical powerQuality of the Multicentred studies
V	Medium	Prospectives controlled non randomized studies	Historical Controls
VI*	Medium	Cohort Studies Case-control studies	Quality of Multicentred studies
VII*			Matching
VIII	Low	Clinical cases series not controlled: Descriptive studies: Disease follow-up, Disease alertness,
IX		Clinical Records Data basesExperts Committees Consensus Conferences Anecdotes or cases
Abbreviations: SE scientific Evidence.
* Quality of the study evaluated by means of protocols species and conditions of scientific rigor
Jovell AJ, Of Navarro - Rubio MD. Evaluation of the scientific evidence. Med Clin (Barc) 1995; 105:740-3.

What principally determines the validity of a study and hence that the results are relevant in the improvement of the health care, is its scientific design. It is for this reason that the existing scales allow doing judgments of value according to the internal validity of the design. Complementing these generic instruments, are the specific instruments that analyze the conditions of methodological rigor depending on the characteristics of the studies of the same design. In case of the controlled and random studies, Jadad's scale constitutes an example; in the case of the studies for the evaluation of the diagnostics tests, Irwing's criteria are another example. Nevertheless, on some occasions there is who considers the type of design to be fundamental without entering in other aspects of methodological rigor, such as the analysis of the design in it self or the data. Because of it, the scales of evaluation of the scientific evidence published, that appear in this chapter are the ones that have been considered most complete due not only to the type of design used for the classification but also other criteria of methodological rigor.

We have to take into account, that in these scales the internal validity of the study is especially considered, that is to say, the homogeneity of the design conditions and population of study. Special value is also given to the external validity or potential to generalize the results out of the population of study. Regardless, it must be said that it is a necessary condition, though not sufficient, to value the internal validity of a scientific study and, as such this it is the first one that must be examined for the quality evaluation of a study.

Once selected the studies depending on its internal validity those of equal design must be presented in a clear and concise form, for example by means of tables or matrices of evidence highlighting those aspects more relevant about the methodology and the results of the different studies. The aim is to identify if the possible variability in the results among individual studies is due to the nature of the health care procedure studied, to the characteristics of the population studied, or to the differences in the characteristics of the design. Sometimes, it is possible to observe in these tables of evidence that studies of the same quality produce results in different directions - positive versus negative. In these cases, but not only in these, it is useful to combine them to obtain a global quantitative measurement that summarizes the different results of the effect of the intervention on a certain clinical condition. This is, precisely, the intention of the fourth phase of the SRSE, the synthesis, which applies a technical procedure to obtain these global measurements that combines result of different studies; it is the called meta-analysis.

Once the effect of a certain procedure or intervention has been valued from gathered scientific studies on certain target populations (phases 1 to 4), other aspects so or more important than the efficacy and the effectiveness have to be taken in consideration. This way, in the fifth phase of the SRSE is necessary to integrate the whole scientific evidence checked and synthesized in the previous ones by means of designs imported from social sciences. It is at this stage where techniques such as the cost – effectiveness analysis, the decisions analysis, the ethical analysis, or the legal applications or the social impact analysis enter. These techniques allow allocating the results obtained to the society or health care context to formulate recommendations. Depending on the topic that is been studied, it is interesting to integrate the results in order that they can be applied to a clinical specific situation or to a concrete region, of Spain or Latin America, for example, at other times it will be a question of adapting the intervention or program to a level of concrete health care system, as could be primary health care.

Finally, all five phases previously described are the ones that guide the decision and allow to elaborate the conclusions and recommendations - phase 6 of the SRSE - if it is necessary or not to apply a treatment, a preventive strategy or to cover or not a certain presentation in the population of reference. These recommendations will be expressed depending on the degree of scientific evidence that has been identified and analyzed in the SRSE (Table 3). In this way, when the quality of the scientific evidence is good, one will conclude that there is suitable scientific evidence to recommend (or to dissuade) the adoption of a certain intervention, medicament or preventive strategy. And, in another end, if the quality of the scientific evidence is low, the results will not be conclusive and, therefore, one will affirm that insufficient scientific evidence exists to recommend (or to dissuade) of what has been object of analysis by means of the SRSE. These recommendations, which always will have to explain to what degree of scientific evidence they are based upon, are valid for the professional practice, constitute guides of clinical practice or recommendations for the politics on public health.

TABLE 3
Production of recommendations based on the scientific evidence

Quality of the SE	Designs	Magnitude of the recommendations
Good	Meta-analysis of controlled clinical trials Controlled clinical trials of big sample Controlled clinical trials of smaller sample (prevalent fewer disease)	Suitable scientific evidence exists to recommend (or to dissuade) the adoption
Medium	Controlled clinical trials of small sample (prevalent fewer disease) Cohort sutdiesCase - control studies	Certain scientific evidence exists to recommend (or to dissuade) its adoption
Bad	Clinical series- Cross studies - Conferences of consensus - Experts' committees- Anecdotes (about a case)	Insufficient scientific evidence exists to recommend (or to dissuade) its adoption
Abbreviators: SE = scientific evidence
Jovell AJ. The evaluation of sanitary technologies in the Primary healthcare. MEDIFAM 1999; 9:102-109

To conclude about the methodology we would like to add that SRSE gives place to two types of complementary interpretations. On one hand, some authors restrict SRSE to the results of the controlled and random studies and to its quantitative synthesis trough the meta-analysis, whereas, on the other hand, it has a wider interpretation. The latter is defined in this chapter and includes other types of epidemiological designs, together with the utilization of social sciences technologies, such as the economical or the political sanitary analysis.

3. Applications
After having detailed the SRSE methodology, the aim of what is described later is to explain all the phases of the systematic review already mentioned by means of examples of specific projects of evaluation. It is necessary to say that the presented methodology must be adapted to concrete situations of evaluation according to the type of problem or uncertainty that is being tried to analyze.

First we have described the phase of scientific evidence search. To be able to reach a considerable part of the aims of any project of evaluation, the query upon the direct evidence – either being a primary investigation or review - is indispensable.

Without entering in detail in the strategy of scientific evaluation search, it is necessary to emphasize that to answer a very concrete or specific questions of investigation like the ones raised on the effectiveness and safety of the treatment with trombolitics agents in established pulmonary tromboembolism or in a stroke or on the effectiveness and safety of the automatic implantable defibrillator versus other therapeutic alternatives, strategical searchers were designed that, unlike the previous ones, were orientated to select direct evidence but exclusively directed to identify studies of the level II - III on the scale in this chapter (Table 2), that is to say, controlled clinical trials with a big and small sample respectively.

Sometimes, the raised aims need the complementariness of indirect scientific evidence proceeding from another sources of information of minor quality or validity. For example, the scarce medical literature published or available on some specific surgical procedures, can raise the need to complement it with the information transmitted by hospital clinical records on the results obtained from patients submitted to surgery. Another examples of this complementariness are the consultations to the Minimal Basic Set of Information (MBSI). The MBSI, available in all the Spanish communities and in some Latin-American countries, includes administrative and clinical information on all hospital discharges.

When in the search of information that is identified as insufficient scientific evidence, so much in relation to quantity as to its quality, there exists the option to design specific, complementary investigations.

The following phase is that of presentation of the scientific evidence. The exhaustion in this presentation will depend on the aims defined in the studies. In the Table 4 there is an example of table of evidence of some of the studies through observation of case-control type published in the medical literature on the physical inactivity as a risk factor in future hip fractures in adults, measurements by means of the decrease of the osseous mass. The results of this table of evidence indicate the difference presence in the characteristics of the design that could explain the differences observed in the results of the studies. The type of comparison used on having analyzed the risk factor, the large part in women, a major in the middle ages and the adjustment factors used could explain the variations observed in the effect of the physical inactivity on hip fracture.

After the presentation of the results, the following phase is the classification of the scientific evidence according to the quality of the design.

In occasions, the fact that the procedure or intervention evaluated is still in the first stages of clinical application, reduces the possibilities of this classification to a sole level since only clinical series exist (level VIII, lowers quality)

TABLE 4
Table of evidence of case - control studies on the physical inactivity as a possible risk factor of hip fracture in adults.

Study and year	Size (cases)	Age	Sex	Type of comparison	RR/OR (95% IC) adjustment factors
Cooper C., 1988	240	>50	Men and women	Productive activities such as housework or the garden (1-2 vs. 5)	3,3 (1,9-5,7) BMI, tobacco, alcohol, AVC and corticoids
Michel BA, et al, 1993	227	54±0,4	879 women and 231 men	Activity (activity decreased by 1 hour/day)	1,05(1,00-1,10) Diagnostic of osteoporosis, duration of corticoid treatment, disability, sex and duration of the disease
Coupland C, et al.,1993	197	>50	85,3% mujeres	Productive activities housework (1-4 hours vs. > 4 hours 7 week)	2,0 (0,6-6,6) Age, sex, IMC, tobacco, alcohol, dependence on daily activities
Cumning RG, et al, 1994	209	>65	Men and women	Physical sports activity (at the age of 20, none vs. medium activity)	0,83 (0,5-1,43) age and sex
Jaglal S, 1995	331	55-80	Women	Activity in the free time (inactive vs. very active.)	2,44(1,43 4,17)
Johnell  O, et al, 1995	2.086	78,1±9,4	Women	Physical and leisure activity at any age (none vs. some)	1,32 (1,44 1,52) centre, age and IMC

On the other hand, on having tried to classify studies that analyze diagnostic tests, it puts of manifest a limitation of this scale. In applications of this type, where the results of diagnostic tests are normally compared with standards of reference, the studies through observation as that of cohorts or that of case-control studies represent the best designs to consider. In this way, on certain occasions, we will have to consider that certain studies can only be placed in a certain level of the scale of the scientific evidence. However, the maximum level of the quality obtained, does not depend only on the scientific existing evidence but on the aim of the SRSE. That is to say, that on occasions when the quality of the scientific evidence available is low-ranking it will not allow obtaining conclusions on its adoption based on a suitable level of scientific evidence. Or what is the same thing, studies of good quality according to criteria of internal validity will be needed in order to express recommendations based on the scientific evidence.

On the other hand, in the case of diagnostic tests, provided that the aim of these does not allow the design of a controlled clinical trial (levels I - II), the studies of the following level (III) already will allow to conclude that suitable scientific evidence exists to recommend them if the values of sensibility and specificity are acceptable and, above all, if the predictive values are the suitable ones.

Nevertheless, it has already been commented that, in certain evaluations, such as studies in diagnostic tests or risk factors, the controlled clinical trials are not the standard design of election and, nevertheless, also it is necessary to obtain a global quantitative measurement that synthesizes the results of the existing studies. In this respect and considering its limitations, meta-analysis studies have also been done through observation that they have allowed to come to global measurements of sensibility and specification.