Monday, October 27, 2008

Moving to page views

Historically, TRIP has used number of searches as a sign of usage/success. However, we have begun to question the merit of this for a number of reasons:

  • The search statistics also included third-party sites that use TRIP via web-services, so users never actually come to the site (the 3rd party site sends a search string to TRIP and we return the results which the site integrates into their pages).
  • The number of searches doesn't take into account how people interact with the site. For instance a user searching on TRIP might look at 2 categories (e.g. EB synopses and guidelines) and that counts as 3 page views (first results page, EB synopses and guidelines). We feel this is much more meaningful.
  • On a more pragmatic basis, our advertising income is based on page views not searches!

In reality this means that in September we had 740,697 page views while we were searched 1,138,699 times. The discrepancy can be explained in part due to 3rd-party access via our web-services, but also because our page views are based on Google Analytics, which can give a conservative view of actual page views (as it requires things like java to be activated to work!). Irrespective of potential discrepancies we still feel the page views are more meaningful.

So, goodbye to search numbers and hello to page views....

Sunday, October 26, 2008

Speed 2

A boring extention of the last post, but the effect of the speed increases in TRIP are continuing to be impressive. Comparing the last 4 full days (22nd-25th Oct) with the same 4 days of the previous week (15-18th Oct) has seen the following:

  • Visits: +21%
  • Pageviews: +92%
  • Bounce rate: -41%

It's hardly surprising, but the increase in speed means people are more prepared to hang around and explore the site. I imagine the overall satisfaction with the site will increase making it more likely that people will come back.

The speed increase and its impact has given us all a real boost.

Thursday, October 23, 2008

Speed: an update

This may be stating the obvious, but people seem to really like fast websites! It's only been 36 hours of a 'fast' TRIP, but some interesting figures already (compared with the weekdays of last week):

  • 24% increase in visits.
  • 30% increase in page views.
  • 27% reduction in bounce rate (proportion of people visiting the site and leaving without exploring).

As it happens, the number of visits is the highest one-day figure we've ever had.

Wednesday, October 22, 2008

Speed: finally a breakthrough

I'm almost scared to admit it, but I think we've finally made a significant breakthrough on the speed of TRIP. For what seems like forever search speeds have been between 10-20 seconds. Today, most of my searches have been under 2 seconds with the odd one hitting 5 seconds.

We've commented here before about improvements in speed (most recently in July 2008), but this time I think it's sustainable. As well as optimising various sections of code we've increased our search capacity by nearly 600%.

What we have noticed in the past has been an increase in usage as soon as our speed increases. However, given the improvements the system is much more resilient to future usage increases.

TRIP is a complex beast made up of new code built upon old code. Our recent investigations have taught us a lot about the system and allowed us even to plan for future hardware/software developments.

We're pleased, I hope you are too.

Friday, October 17, 2008

Experience as evidence

I heard an interview on today’s Today programme on Radio 4 with Sir Michael Rawlins, the chairman of NICE. In the interview he highlighted the need for post-research evaluation to ensure the research promise of an intervention is realised.

This notion of experience as a form of evidence has been of interest to me for many years, ever since I was approached by a GP who was using bupropion in smoking cessation. His perspective was that bupropion was pretty rubbish for his patients and wanted to understand why the discrepancy with the evidence (which stated it was effective and evidence-based).

My own perspective is that the experience of ‘coal face’ clinicians is crucial in addressing some of the biases seen in clinical research. In a very nicely funded and implemented randomised controlled trial (RCT) you may very well exclude patients with co-morbidities – alas, in the ‘real world’ clinicians don’t have that luxury. Take an intervention such as cognitive behaviour therapy (CBT) for depression, those carrying out the trial will use an experienced CBTer as opposed to a ‘coal face’ clinician who may have basic training in the matter. The bottom line is that conditions in a RCT cannot be the same as those of the ‘coal face’ – that should surely be a worry.

Currently, the ‘evidence’ used in EBM is overwhelmingly devoted to the hard evidence obtained from trials such as a RCT. I’m not suggesting that RCTs are not useful, far from it. However, I’m increasingly of the opinion that harnessing the experience of clinicians is vital to supplement the evidence found in clinical trials. Experience can be a form of evidence.
Interestingly, in my previous post highlighting 16 years of EBM (that linked to this JAMA article) the JAMA article had the following passage (from their 1992 article):

“A new paradigm for medical practice is emerging. Evidence-based medicine de-emphasizes intuition, unsystematic clinical experience, and pathophysiologic rationale as sufficient grounds for clinical decision making and stresses the examination of evidence from clinical research. Evidence- based medicine requires new skills of the physician, including efficient literature searching and the application of formal rules of evidence evaluating the clinical literature.”

In the passage they highlight that EBM de-emphasises, amongst other factors, the ‘unsystematic clinical experience’. While I completely agree with that sentiment, it opens the door for the emphasis of SYSTEMATIC clinical experience.

So, who’s for the systematic collection of clinical experience?

Sunday, October 12, 2008

The return of innovation

Following on from my previous blog post is another article (click here) on web 2.0, this time on ReadWriteWeb, which examines the effects of the economic downturn. But my interest in this article is the following passage on innovation:

"During boom times, companies direct development and occupy great talent with at best evolutionary improvements over the state of the art. Companies are great chasers of new things, but aren't great at making new things. A recession means technologists cease to be paid vast amounts to duplicate the work of others."

The above feels right to me. Over the years I've noticed that nothing focuses my mind than bad times, I suppose it brings out the survival instinct.

When I was studying for my PhD on innovation I was always intrigued by the distinction between evolutionary changes/innovation and radical changes/innovation. The sociologists tended to highlight that evolutionary changes came from those integrated into a particular system. The argument being that those people tended to have been trained in a particular system and think in a similar way, making radical change difficult.

For truly radical changes what is needed are people not integrated into a particular system, in other words outsiders. They haven't been indoctrinated by 'the system' and are better able to think for themselves and think 'outside the box'.

If the above analysis is correct we're in for some interesting times ahead. Time for everyone to get their thinking caps on.

Saturday, October 11, 2008

The end of web 2.0

One of my favourite blogs is TechCrunch and the main guy there is Mike Arrington and he's written about the end of web 2.0 - An Ignoble But Much Needed End To Web 2.0, Marked By A Party In Cyprus.

Web 2.0 has been associated with blogs, social networks, wikis and generally user generated content. I doubt it's dead, but I'm guessing that's not what Mike meant.

My view is that user-generated content is no longer new, its an integral part of the web and therefore why should it be seen as distinct? I'm guessing that we've got the situation where you had the 'old' (static web pages) then web 2.0 (user-generated content) and it's now settling down to just being one web, which will mean the old sites will simply absorb sections of web 2.0 into their domains.

So the days of a 'separate' web 2.0 may well me dead or dying, roll on the next innovation.

Sunday, October 05, 2008

The TRIP database. A valuable source of literature on oral health

Why do I like this article?

Save you all clicking to view the abstract, here it is:

Locating the desired evidenced-based literature via the internet is usually time-consuming and not particularly straight-forward. Websites offering evidence-based information should be accessible, efficient and user friendly. Many of these websites, however, only make their own content available, making it necessary to change websites in order to get a complete picture of the required information.The so-called website of ''Turning Research Into Practice'' is a positive exception.This website makes it possible to search the whole range of evidence-based publications on a certain subject with 1 search action, is straight-forward and practical to use, and the presentation is exceptionally good.

Answers to Questions Posed During Daily Patient Care...

...Are More Likely to Be Answered by UpToDate Than PubMed has just been published in the Journal of Medical Internet Research. This is proving an interesting journal. Due to the wide coverage I find many articles of little interest, but the occasional gem appears.

Perhaps unsurprisingly the authors conclude:

"Specialists and residents in internal medicine generally use less than 5 minutes to answer patient-related questions in daily care. More questions are answered using UpToDate than PubMed on all major medical topics."

If you asked 100 information specialists to predict the outcome, I'd be surprised if many would have thought PubMed would do better!

The results were more interesting:

"We analyzed 1305 patient-related questions sent to PubMed and/or UpToDate between October 1, 2005 and March 31, 2007 using our portal. A complete answer was found in 594/1125 (53%) questions sent to PubMed or UpToDate. A partial or full answer was obtained in 729/883 (83%) UpToDate searches and 152/242 (63%) PubMed searches (P < .001). UpToDate answered more questions than PubMed on all major medical topics, but a significant difference was detected only when the question was related to etiology (P < .001) or therapy (P = .002). Time to answer was 241 seconds (SD 24) for UpToDate and 291 seconds (SD 7) for PubMed."

I'd be interested to know if clinicians were satisfied with partial answers. The authors state "Whether an answer is partial or complete is a subjective qualification. We therefore combined partial and full answers when determining significance of our findings" - am I the only one thinking that's inappropriate? At least give us the results separately as well as combined and let us make draw the conclusions.

Another issue, not restricted to this paper but all papers on Q&A, is the assumption that the clinician knows that they've answered the question. They may well have found an answer, but whose to say it's right?

Anyway, if anyone reads the article, I'd be interested to hear what you make of it.