JAN Forum: your views and letters


Response to: A method comparison study to assess the reliability of urine collection pads as a means of obtaining urine specimens from non-toilet-trained children for microbiological examination by M. Farrell, K. Devine, G. Lancaster and B. Judd (2002) Journal of Advanced Nursing 37, 387–393

We were interested in the study undertaken by Farrell and colleagues on the methodology comparison of collecting urine samples from children published in your journal recently (Farrell et al. 2002). The paper surprised us mainly because after addressing the shortcomings of all previous studies in this field, these authors undertook a study whose design failed to satisfy their own minimum criteria.

These authors were so concerned to ensure that the urine collections were contemporaneous that they decided to use both the bag and the pad collection techniques literally at the same time. In order to do this they had to modify the way they were done, that is they performed them incorrectly. All that was necessary was to collect a pad or bag sample correctly, and then repeat the collection with the other technique, ensuring that this sequence was randomized. That is indeed the method that we used when we tested approximately twice as many patients as Farrell used (Liaw et al. 2000). As the purpose of this study was to evaluate the extent of contamination of urine collections it would also be preferable for such a study to be on children thought unlikely to have an infection, as we did. Finally, it should be tested in an appropriate (often challenging) clinical environment. That is why we asked our parents to make these collections at home rather than them being undertaken in a variety of different clinical settings, as Farrell did.

We also note that these authors are very concerned about the precise white blood cell numbers in the urine samples. This seems misplaced. It is well established that urinary tract infection should be defined by the presence of minimum numbers of a single species of bacteria. The white cell numbers are poor markers of urinary tract infection. The numbers may fall extremely rapidly in some urine samples after collection, but not in others (Stansfeld 1962, Vickers et al. 1991), or may be significantly raised by febrile illnesses in the absence of a urine infection (Turner & Coulthard 1995), or from vaginal contamination. Children that are immunosuppressed or extremely sick may be simply unable to generate urinary white blood cells (Kumar et al. 1996).

We would suggest that any future urine collection methodology testing these authors undertake should include them applying the techniques correctly, ideally in a consistent and appropriate setting, and then evaluating bacterial culture only, as it is that alone which determines whether the child has a urine infection.

Response to Coulthard's critique

Mike Farrell BSc RGN RSCN PGCE
Lecturer/Practitioner, Alder Hey Royal Liverpool Children's NHS Trust, Liverpool, UK. E-mail:mfarrell@liv.ac.uk

Brian Judd FRCPCH
Consultant Paediatric Nephrologist, Alder Hey Royal Liverpool Children's NHS Trust, Liverpool, UK

Thank you for allowing us the opportunity to respond to the issues raised by Coulthard and Vernon with respect to their critique of our study.

Coulthard and Vernon are known proponents of the pad collection method, having undertaken several studies evaluating the effectiveness of this collection technique. These researchers are to be congratulated for the research they have undertaken and have demonstrated the pad collection method as a comfortable, convenient and less expensive way of collecting urine, something we would also support. Yet, despite these benefits, it would seem few centres have adopted this method of collection.

For the purpose of our study there was experimental modification of the collection technique, using an approach previously reported (Suri 1988), in order to obtain a concurrent time-paired matched sample. We did this because we believed it had advantages in reducing potential variance that might be associated with non-concurrent samples. We accept the use of a randomization sequence for the collection of urine specimens can be one way to counter variance, but we worked on the simple principle that the closer the match of the sample, in all aspects of it's collection, the more confident one could be in expecting equivalent results between the two methods.

Coulthard and Vernon dismiss the clinical significance of white blood cells (wbc) as a poor marker of urinary tract infection. However, in our setting, we believe that they are a valuable predictive marker and their absence (or reduction) can limit ability to confirm diagnosis, and we have no doubt that we are not alone in this view. We do not suggest that the diagnosis rests on wbc alone but it is the combination of this count and urinary bacteriology that is of importance. Consequently, anything that leads to an adverse manipulation of these results can affect diagnosis and possible treatment intervention.

We are unsure of the complaint made that we should have undertaken collection ‘in a consistent and appropriate setting’. Our samples were collected by one of two researchers, and were obtained from children attending the clinical setting. It would seem perfectly reasonable to establish the reliability of the technique in the clinical setting before we can advocate collection in different settings.

As we note in the report, this was a pilot study and we clearly state the limitations that such a study inevitably generates. Not unexpectedly, further work is required to confirm the findings we report. Perhaps one way that this might be done is to conduct a ‘within patient study’ where both the concurrent and non-concurrent urine collection methods are robustly compared.