Takaisin Tulosta

EBM Guidelines evidence summaries

Lisätietoa aiheesta
3.2.2008

The evidence summary database of EBM Guidelines

The evidence summary database of EBM Guidelines contains more than 3000 evidence summaries. New evidence summaries are produced continuously by the editorial team at a rate of several hundreds a year, and existing evidence summaries are updated on the basis of new evidence.

The original master documents of the evidence summaries are written and stored in XML format which also contains descriptive and administrative metadata. The publication formats are produced automatically from the XML originals.

Each evidence summary is an independent document that can be searched and linked to other documents. Each document in EBM Guidelines has a unique 8-digit alpha-numeric identifier that can be used in pointing to the document in an URL or used as a search term to retrieve the document. The identifier is shown at the end of each document by EBMG search and browser software.

The EBM Guidelines database is built on Oracle™, and the search program has been customized for the needs of health care professionals.

Selection of topics

We intend to cover clinically important evidence on diagnosis, screening, treatment and follow-up of all conditions encountered in primary and ambulatory care.

The guidelines and evidence summaries are intended for e.g. the following settings:

  • Unselected patients in primary care
  • Acute conditions treated by general physicians in emergency rooms
  • Managed care for chronic diseases
  • Treatment on community hospital wards by primary care physicians
  • Maternal and child health surveillance

The following settings are mostly excluded:

  • Management by subspecialists in secondary hospitals
  • Neonatology
  • Techniques in anaesthesia and surgery on specialist level
  • Interpretation of diagnostic procedures only performed and interpreted by subspecialists

However, subspecialist procedures and interventions are included when there is a possibility that primary care physicians have a role in referring patients to these investigations and treatments or counseling their patients on whether to undertake them.

The selection of topics is mainly evidence-driven at present. This means that we systematically search for high-quality evidence, appraise and grade its quality, and summarize it as briefly as possible. The following sources are systematically followed by our editorial team and evidence summaries are produced of all reviews and topics relevant to EBM Guidelines as described above:

  • Cochrane reviews
  • DARE abstracts by the Centre for Reviews and Dissemination
  • Clinical evidence
  • New England Journal of Medicine, JAMA, Lancet, BMJ

In addition, the following sources are followed regularly and used as supplementary sources for evidence summaries:

  • ACP Journal Club
  • BMJUpdates
  • InfoPOEMs
  • NHS EED and HTA databases

After having utilized these sources of evidence we specifically search evidence for decisive decision points in our guidelines not covered by the above-mentioned sources. We perform literature searches and use the guideline organizations in Finland (about 1000 experts) to appraise the evidence.

We prioritize Cochrane reviews and produce evidence summaries from them as soon as new reviews are published. We produce evidence summaries from other sources in batches, so that there may be some delay before new reviews or studies are included in the evidence summaries.

How the scope of an evidence summary is determined

The evidence summary basically gives an answer to a PICO question (patient group, intervention, control, outcome). In an ideal case, there is one intervention and a small number of patient-oriented outcomes.

As the evidence summaries are often based on one Cochrane review, their contents reflect the scope of that review. The following exceptions occur:

  • The results of several systematic reviews on the same topic and the latest studies not yet included in systematic rewievs are combined into one evidence summary. This is very common at present. The inclusion criteria for a Cochrane review are often very tight, and a systematic review cited by a DARE abstract may include also other study types than RCTs which may broaden the body of evidence.
  • The results of one systematic review assessing several interventions are divided into more than one evidence summary. This may be necessary e.g. with Cochrane reviews covering many different types of interventions for a certain condition.
  • Dividing results into more than one evidence summary according to outcome may be necessary in cases where there are several outcomes of interest, and the grade of evidence is different for these outcomes. However, the most common way of handling this situation is to express the different grades of evidence in the wording of the statement (see following paragraph).

We prefer to group interventions of the same class and of proven equal effectiveness into one evidence summary. Thus, our evidence summaries deal with drug classes such as ACE inhibitors, statins, or proton-pump inhibitors. In cases where systematic reviews are only available of single drugs, but not from the whole drug class, the evidence summary only deals with a single drug.

How the evidence summaries are structured

Heading

The heading is neutral – it does not contain a message of effectiveness. When an evidence summary is based on a Cochrane review or a DARE abstract, the heading is often the same as that of the cited review, or (quite often) an abbreviated version of it.

The grade (level) of evidence

The grades of evidence are determined according to the principles of the GRADE group «http://www.gradeworkinggroup.org»1 that divides the quality of evidence into high, moderate, low, or very low. The corresponding codes in EBMG are A, B, C and D, and the definitions (according to the GRADE group) are as follows:

High (A): Further research is very unlikely to change our confidence in the estimate of effect.

  • Several high-quality studies with consistent results
  • In special cases: one large, high-quality multi-centre trial

Moderate (B): Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.

  • One high-quality study
  • Several studies with some limitations

Low (C): Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.

  • One or more studies with severe limitations

Very low (D): Any estimate of effect is very uncertain.

  • Expert opinion
  • No direct research evidence
  • One or more studies with very severe limitations

Evidence statement

A PICO statement that describes the most important research finding(s) for the topic of the evidence summary, and gives a verbal description of the grade of evidence that corresponds to the letter code. The verbal expression of the main outcome is always reported first and its phrasing corresponds to the evidence code. Example (evd05438 Chest-compression only by bystanders for cardiac arrest):

Level of evidence = B

Cardiac-only resuscitation by bystanders appears to be more effective than conventional cardio-pulmonary resuscitation in patients with out-of-hospital cardiac arrest

Sometimes other outcomes of interest are mentioned in the statement. If the grade of evidence of these other outcomes is different from that of the main outcome, this is expressed in the wording. Example:

Level of evidence = A

Pre-operative radiotherapy followed by surgery is more effective than surgery alone for preventing local recurrence of rectal cancer, and may also improve survival.

In this example, the grade of evidence is A for local recurrence, and C for survival (which is indicated by the word may – see Table 1).

Summary of results

The most important outcomes are reported, with numerical information and confidence intervals. The summary is usually shorter than that of the original review or study abstract. Important adverse effects are reported.

Concepts like relative risk, odds ratio, weighted mean difference, confidence intervals are given in abbreviated form (RR, OR, WMD, CI) in the summary.

Comment

The factors downgrading or upgrading the quality of evidence are reported on the basis of the GRADE principles. Important comments on the applicability of the results may also be included. Example (evd05438 Chest-compression only by bystanders for cardiac arrest «Cardiac-only resuscitation by bystanders appears to be more effective than conventional cardio-pulmonary resuscitation in patients with out-of-hospital cardiac arrest.»B):

Comment: The quality of evidence is upgraded by strong association. The results are not applicable to patients who have primary respiratory arrest (e.g. drowning or suffocation).

The two situations where a comment on downgrading or upgrading is not needed are the following.

  • The grade of evidence for RCTs is by default high (A). A comment on downgrading is added only if the given grade of evidence is lower.
  • The grade of evidence for observational studies (cohort studies, case-control studies) is by default low (C). A comment on upgrading is added only if the given grade of evidence is higher.

References

The references are Vancouver style. At present the references of DARE abstracts have links to CRD website. In the future, all references will have links to PubMed abstracts.

How the evidence summaries are indexed

All evidence summaries are indexed by hand. There are two indexing fields:

  • The browsing class indicates the specialities that deal with the topic of the evidence summary. One evidence summary can be indexed into several browsing classes.
  • The search terms describe the contents of the evidence summary.

The search terms are mainly MeSH terms with U.S. spelling, but other terms are also used in cases where a suitable MeSH term does not exist..The synonyms of the search terms are also included in our vocabulary, so that e.g. “herpes zoster” and “shingles” should give the same search result. The synonym vocabulary of EBMG will be updated and considerably enhanced during the next few months. The Unified Medical Language System (UMLS) by the National Library of Medicine is used for building the synonym vocabulary. Abbreviations are also included in the synonym vocabulary.

The search terms indexed by hand are assigned a category “key” or “non-key”. In the search result, the guidelines and evidence summaries containing the given search term as a key term are shown first, and the ones containing the given search term as a non-key term are shown lower in the search result. The third category of search terms are “text words”. They are indexed automatically by the database software. The text word hits are shown at the end of the search result.

Update markings

When an evidence summary is created or updated, the date is changed to correspond the latest time of the update. The dates of the evidence summaries are shown at the beginning of each document (exception: the older evidence summaries produced in the beginning of the 2000’s do not have dates. The dates will be added during the updating process so that all evidence summaries will have dates by the end of 2007.

There are two categories of updates.

  • A major update means that the contents of the evidence summary or guideline have been changed. The level of evidence may either remain the same or change during a major update.
  • A minor update means that there is a minor change that does not affect the clinical decisions of the user. Guidelines have minor updates quite often, evidence summaries only rarely.

A document with a major update during the last 6 months is indicated by a red dot in front of the heading in the search result. A document with a minor update within the last 6 months is indicated by a blue dot. At present our software is being updated, and some update markings may indicate that the document has been updated although there are no significant changes. This problem will be corrected soon.

Updated text within a guideline or evidence summary is shown in red colour.

How the evidence summaries are linked to guidelines

The evidence summaries are linked to guidelines at the exact places where the interventions are mentioned whenever possible.

In cases where the evidence summary deals with a condition discussed in the guideline, but the guideline does not contain a recommendation that matches the evidence summary, the evidence summary statement is included in a Related resources article attached to the guideline. The Related resources articles are a new development, and they are not included in all EBM Guidelines topics yet. Their predecessor is a paragraph named Related evidence at the end of the guideline.

Each evidence summary contains link(s) to all guidelines to which the evidence summary is linked.