7th December 2012: Validating the Difficult Intravenous Access Clinical Prediction Rule

 

071212

Lots of discussion about this already on Twitter – let’s take a look!071212

Where can I find this paper?

http://www.ncbi.nlm.nih.gov/pubmed/23187988

What is this paper about (what is the research question)?

How good is the Difficult Intravenous Access (DIVA) clinical prediction rule at identifying paediatric patients likely to require more than one attempt at intravenous cannulation.

Summary of the Paper

Design: prospective cohort study

Objective: validation of the DIVA score with derivation of test characteristics.

Test of interest: DIVA, a 4-part weighted clinical prediction rule. A score of 4 or more was thought to predict patients “50% more likely than the mean success rate to have a failed IV placement on the first attempt”

Success was placement of an intravenous cannula through which saline could be flushed without compromising the vein.

Primary outcome: test characteristics

Population: children aged 0-14 years requiring IV placement presenting to the ED of Mayo General Hospital in Ireland

  • Inclusion: children requiring IV placement
  • Exclusion: children requiring emergency treatment

Results: 500 children were recruited, 133 were younger than one year.

Mean failure rate was 22% – the authors state that 110 had a failed IV placement on first attempt (this doesn’t quite add up – see below).

151/500 had a DIVA score ≥ 4, of whom 57 (37.7%) had a failed IV placement on first attempt.

349/500 had a DIVA score <4, of whom 55 (15.8%) had a failed IV placement on first attempt.

For cutoff ≥4: sensitivity was 75.9%, specificity 51.8%, PPV 84.4%, NPV 37.7%.

Authors’ Conclusions:

This study validated the DIVA CPR in a population similar to that originally studied. Its acceptability and usefulness needs to be further accessed by health professionals involved in IV placement.

On the study design

The great thing about this study is that it appears highly generalisable – at least on the surface. All children who required IV access were eligible for inclusion between ages 0-14years and they were only excluded if IV was required as an emergency. The circumstances seem to reflect standard ED practice; topical local anaesthetic was used (although we don’t know which one) and doctors were limited in their attempts at cannulation before seeking help.

There are a few methodological issues though. Although we know a good amount about the population from which the sample was obtained, we don’t know how it was obtained – was it consecutive in a time frame? What time frame? Was it a convenience sample? Were these children selected at random? These factors might all affect the results – and we need to know them.

There are also some ethical issues – while there was a clear policy for seeking help after three failed attempts, it is unclear whether a second doctor was any more experienced/skilled at cannulation than the first. It would be nice to think that a senior doctor’s assistance was being enlisted. The study does have ethical  approval, but I wonder  whether this reflects standard rather than ideal practice.

We don’t know which local  anaesthetic was used; was there consistency here? Again, this might confound the results, or affect their external validity.

What were the results and what does this mean?

So, the results don’t add up, which is the first problem. Even if we take the authors’ word for the sums being right, the test characteristics are pretty useless. It’s easy to imagine that a ROC curve would have a pretty small area under the curve.

We also don’t have confidence intervals for the test characteristics. These are vital for gaining a true feeling for the usefulness of the data. There’s a great youtube resource explaining the importance of confidence intervals here.

What can we take from this paper into clinical practice?

Lots of questions! Much of the conversation on twitter asked whether this was even a relevant clinical question; @damian_roland wondered whether those who didn’t recognise the challenge of cannulation in infants – especially ex-prems – should be attempting cannulation.

There were also questions about what we should do with children for whom DIVA predicts difficult cannulation. Should we forego all cannulation attempts? Should we go straight for IO access? As much as I advocate timely IO as an approach for anyone with difficult access, the numbers don’t seem good enough to draw that sort of conclusion – but we do need to understand what we would want to  do with the findings before investing much more time in further validation of the rule.

What’s interesting is that in the paper children for whom failure was absolute went on to have enteral  rehydration by the NG route. There  is a cultural hesitancy around NG but it could certainly be argued that where it is possible it is  more physiological than intravenous hydration – so shouldn’t we be trying this before repeated IV attempts instead of afterwards? It also raises the question – who is deciding that these children need IV access? Is that also coming  from  doctors with one month of paediatric experience?

Lastly, we can reflect on the sites used for cannulation with greatest success. In my experience,  local anaesthetic is often applied by nursing staff to the antecubital fossa, particularly in the ED, whereas my preference is the dorsum of the hand. In the paper, the back of  the hand was used in 78.7% of successful cannulations. It would be interesting to  look at relationships between success  and site, considering where local anaesthetic  had been placed and whether doctor preference was expressed in advance of this.

More questions to ask

  • What do we want to use this data for in a clinical  context?
  • Has this paper simply highlighted a need to consider NG route for rehydration before IV?

Read the Twitter conversation here

Follow us on twitter: @PEMLit

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s