How many Cochrane reviews are needed? Too many

PubMed Trend is a neat way to assess time treds in the Medline database. The number of Randomized controlled trials published per year continues to rise and has broken through the 33,000 barrier.

A 2003 analysis, of 989 Cochrane reviews which included information on a total of 9,778 studies, found the number of studies per review ranged from 0–136, with a median of 6 (interquartile range 3 to 12).

At the time of this analysis the average Cochrane review included 6 studies, and it was estimated that 45 000 Cochrane reviews would be needed to cover the existing 300 000 references.

By 2012 the cumulative total of trials in PubMed is 454 000: the number of reviews currently required is 75,700.

In terms of modelling the growth of trials a steady state will probably be arrived at by 2021, when approximately 50,000 trials will be added to the PubMed database each year.

The scary fact is, that at current rates we can expect as many trials to be published in the next ten years as have ever been published. Now that’s a lot of Cochrane reviews to be done.


  • carl heneghan

    September 30, 2014 at 11:23 am Reply

    Thought I’d post this blog from 2010, as it is relevant to the discussion:

    How many randomized trials are published each year?
    Carl Heneghan
    Last edited 17th March 2010

    I thought it was worth re-examining how many randomized controlled trials (RCTs) are published each year. To determine the number this number per year I used the Medline (Pubmed) Trend database. The page displays the number of entries (articles) in PubMed (Medline) published every year which conformed to the following search strategy (randomized and controlled and trial).

    The graph highlights the growth from 39 RCTs in 1965 to 26,017 in 2008. Some notable landmarks include the 1000 barrier being broken in 1976 and the 10,000 in 1994. Approximately every 10 years there is a doubling of the number of RCTs: 25,361 in 2007 versus 12,040 in 1997. At current rates we can expect to see 50,000 RCTs per year published by 2018-9.

    RCTs per year

    Breaking these numbers up by week you would have to read 500 RCTS per week in 2008 to cover the published RCTs in PubMed, the 100 barrier was broken in 1988. Per day it equates to 71 RCTs but I figured no-one submits and publishes on the weekend and if we consider only reading RCTS on weekdays then we have now have to read 100 RCTs Monday through Friday. To put this in perspective in 1980 you only had to read one RCT per day to cover all of the published RCTs. Today you have no chance of keeping up to date.

    RCTs per week and per day

    If you want to know how much is in PubMed in total. Well as of June 2007, approximately 1,000,000 items were archived and this is growing at 7% per year. Get reading quick!

  • This coupon website

    March 21, 2013 at 2:49 am Reply

    Greetings! This is my first comment here so
    I just wanted to give a quick shout out and tell you I genuinely
    enjoy reading through your articles. Can you suggest any other blogs/websites/forums that deal with the same topics?
    Thank you!

  • loredana

    November 12, 2012 at 9:12 pm Reply

    the problemi s not how many SR are needed to summarize the evidence of RCT, but to identifie the main relevant quenstion on specifi field and plan a specific SR

  • Jon Brassey

    November 4, 2012 at 7:10 pm Reply

    Surely, this suggests that the Cochrane model is not ‘fit for purpose’ given the current funding levels (or the funding levels we are likely to see in a lifetime).

    I would prefer to see more research carried out on the possible shortcuts that could be taken to prevent/reduce the need for a full-blown SR.

    Also, if we are only able to produce 1,000 SRs lets prioritise them better (the assumption is Cochrane has a robust prioritisation scheme – do they?) – based on clinical need and clinical uncertainty. Surely, if we find 10 RCTs all saying treatment X is better than treatment Y that is less of a priority for an SR than one with more varied results.

    We wordship at the alter of systematic reviews. Not sure that’s entirely healthy

Post a Comment