Jump to comment:

• Re: Cross-sectional presentation of longitudinal data
  Alison M Kemp
  Published on: 22 March 2016
• Cross-sectional presentation of longitudinal data
  Rollo D Clifford
  Published on: 22 March 2016

• Published on: 22 March 2016

  Re: Cross-sectional presentation of longitudinal data

  • Alison M Kemp, Dr
  • Other Contributors:
    • Frank Dunstan, Diane Nuttall, M Hamilton, Peter Collins and Sabine Maguire

  We would like to thank Dr Clifford for his interest in our research. We do not agree with him that the title and abstract are misleading. The study was a longitudinal one and the results reflect that; for example we looked at the children over time and assessed the importance of within- child variation over time compared to between-child variation. It is very important to distinguish between a collection, that is a point...

  Show More

  We would like to thank Dr Clifford for his interest in our research. We do not agree with him that the title and abstract are misleading. The study was a longitudinal one and the results reflect that; for example we looked at the children over time and assessed the importance of within- child variation over time compared to between-child variation. It is very important to distinguish between a collection, that is a point in time at which data were collected, and the child on whom the data were collected, since a single child may have multiple collections. In summarising the data by collections we were reflecting the occurrence of bruising over the whole study.

  The situation most closely resembling the presentation of a child to a clinician is the first collection on that child. The prevalence of bruising at the first collection, that is the percentage of children who had at least one bruise, was 5.3%, (95% CI 2.6 - 10.5) in pre-mobile children, and only 1.3%, (95% CI 0.2 - 6.9) for those infants who are not yet rolling. The percentage of pre-mobile children who experienced a bruise at some time in up to 12 collections is necessarily much higher, but that does not estimate the prevalence. The situation is analogous to the occurrence of the common cold. At any one time during the year, the prevalence is quite low - probably below 10% - but most people have a cold at some time during a year.

  We agree that each case must be assessed on its own merit with a clear and detailed history taken for the cause of bruising, a careful examination and appropriate investigations to determine the cause of the injury; indeed that is the message of the concluding paragraph of the paper. Undertaking this type of work is indeed challenging and did require a high level of co-operation from our group of parents for which we are most grateful. The author of the letter is right to point out that a more intrusive approach would be unethical and impractical'.

  In future work we will compare patterns between the children in this study, those with bleeding disorders and those who have been confirmed as victims of child abuse to seek methods for discriminating between them.

  Conflict of Interest:

  None declared

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top
• Published on: 22 March 2016

  Cross-sectional presentation of longitudinal data

  • Rollo D Clifford, Consultant Paediatrician

  The authors of this study are to be congratulated on a unique and useful collection of data which, in the present climate, is increasingly difficult to achieve. Unfortunately the title is somewhat misleading and the abstract potentially open to mis-interpretation. Although parents collected diary data on their children during concurrent weeks, this is presented by the authors in a cross sectional, not longitudinal manner. Thus, wh...

  Show More

  The authors of this study are to be congratulated on a unique and useful collection of data which, in the present climate, is increasingly difficult to achieve. Unfortunately the title is somewhat misleading and the abstract potentially open to mis-interpretation. Although parents collected diary data on their children during concurrent weeks, this is presented by the authors in a cross sectional, not longitudinal manner. Thus, when the authors report just 6.7% of "collections" among premobile infants had at least one bruise, this refers to a single week of observation. Lost in the text of the results section is the information that 27% had a bruise over all weeks surveyed. The true proportion of infants with a bruise over the whole of the 0-11 month period must be significantly higher than this as parents did not necessarily provide data for the whole period. Equally although a bruise was recorded in just 2.2% of one week "collections" among infants unable to roll over, it seems likely that the true figure over the 17 weeks or so at that developmental stage is likely to be very much higher - perhaps as high as 10%! Although we should continue to ensure that health and care services maintain a high level of vigilance, it is important for paediatricians to assess each case on its merits looking for features such as clusters of bruises and bruises in areas typical of child abuse rather than responding to these and other data with a "one rule fits all" response thereby subjecting large numbers of infants to unnecessary and harmful investigation and parents to huge anxiety and intrusion. A further concern, and from my own experience from attempting to design a similar longitudinal approach, is that ethics committees and safeguarding boards are increasingly likely to make it very difficult to collect this type of essential longitudinal data without unacceptable intrusion.

  Conflict of Interest:

  None declared

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top