Jump to comment:

• Health economic burden that wounds impose on the National Health Service in the UK: request for further methodological information.
  Jo C Dumville, Sarah Rhodes, Ross Atkinson, Nicky Cullum, Gill Norman, Maggie Westby, Louise Hussey, Emily Petherick and Naseer Ahmad
  Published on: 20 June 2017
• Questions about Comorbidities
  Julian F. Guest
  Published on: 15 March 2016
• Questions about Comorbidities
  Marissa J. Carter
  Published on: 15 March 2016

• Published on: 20 June 2017

  Health economic burden that wounds impose on the National Health Service in the UK: request for further methodological information.

  • Jo C Dumville, Senior Lecturer in Applied Health Research University of Manchester
  • Other Contributors:
    • Sarah Rhodes, Statistician
    • Ross Atkinson, Research Fellow
    • Nicky Cullum, Head of Division of Nursing, Midwifery and Social Work
    • Gill Norman, Research Fellow
    • Maggie Westby, Research Fellow
    • Louise Hussey, Research Fellow
    • Emily Petherick, Lecturer in Epidemiology and Biostatistics
    • Naseer Ahmad, Consultant Vascular Surgeon

  Guest et al use routinely collected primary care data to estimate an annual prevalence of wounds in the UK of 4.5%.(1) They then report annual prevalence estimates for different wound sub-types, including venous leg ulcers. We have become aware of differences in the venous ulcer prevalence estimates of Guest et al (1) and our own estimates from THIN (2) and from surveys in Leeds (3) and Greater Manchester (data on file). We have calculated an annual UK prevalence of venous leg ulcers of 1.3 per 1000 (or 0.13%) compared with Guest’s estimate of 5.6 per 1000 (or 0.56%).

  Briefly our annual estimate was calculated by combining our 2011 point prevalence values (row B and C from Table 1: rounded to 0.3 per 1000. Please note since response format does not allow tables we have converted to text) with the 2006 UK annual incidence rate of venous leg ulcers estimate (row D Table 1), assuming that the annual incidence has not changed since 2006. Our value is very similar to the crude estimate of annual UK venous leg ulcer prevalence of 1.4 per 1000 people directly calculated by Petherick et al (2) using THIN and G(C)PRD (Row E: Table 1).

  We note that the annual prevalence estimate for venous leg ulcers of Guest et al (row A: Table 1) is four times higher than the others presented here. We also note that this is the most conservative estimate of the UK annual prevalence from Guest’s data given the large number of unspecified leg ulcers reported in this paper– a large...

  Show More

  Guest et al use routinely collected primary care data to estimate an annual prevalence of wounds in the UK of 4.5%.(1) They then report annual prevalence estimates for different wound sub-types, including venous leg ulcers. We have become aware of differences in the venous ulcer prevalence estimates of Guest et al (1) and our own estimates from THIN (2) and from surveys in Leeds (3) and Greater Manchester (data on file). We have calculated an annual UK prevalence of venous leg ulcers of 1.3 per 1000 (or 0.13%) compared with Guest’s estimate of 5.6 per 1000 (or 0.56%).

  Briefly our annual estimate was calculated by combining our 2011 point prevalence values (row B and C from Table 1: rounded to 0.3 per 1000. Please note since response format does not allow tables we have converted to text) with the 2006 UK annual incidence rate of venous leg ulcers estimate (row D Table 1), assuming that the annual incidence has not changed since 2006. Our value is very similar to the crude estimate of annual UK venous leg ulcer prevalence of 1.4 per 1000 people directly calculated by Petherick et al (2) using THIN and G(C)PRD (Row E: Table 1).

  We note that the annual prevalence estimate for venous leg ulcers of Guest et al (row A: Table 1) is four times higher than the others presented here. We also note that this is the most conservative estimate of the UK annual prevalence from Guest’s data given the large number of unspecified leg ulcers reported in this paper– a large proportion of which were likely to be venous leg ulcers.

  Table 1. Sources of data for calculating prevalence of venous leg ulcers in the UK (converted to text and available on request)

  Row A. Guest et al (1)
  Data source: THIN dataset 12 month period over 2012/13
  Denominator: Active adult (18 years and over) patients registered with GP
  Data type: Period prevalence (annual)
  Result for venous leg ulcers: 5.6 per 1000 persons (or 0.56%) Confidence intervals not presented

  Row B. Hall et al (3)
  Data source: 2011 multiservice complex wounds survey in Leeds, UK conducted over a 2 week period
  Denominator: Total population of Leeds (census data)
  Data type: Point prevalence
  Result for venous leg ulcers: 0.29 per 1000 persons (or 0.029%) (95% CI 0.25 to 0.33 per 1000)

  Row C. Publication pending
  Data source: 2015 NIHR CLAHRC Greater Manchester complex wound survey (data on file) conducted over 2 week period in community
  Denominator: Total relevant geographical population
  Data type: Point prevalence
  Result for venous leg ulcers: 0.32 per 1000 persons (or 0.032%) (95% CI 0.29 to 0.34 per 1000)

  Row D. Petherick et al (2)
  Data source: THIN dataset we extracted year 2006 data from graph in paper
  Denominator: All patients aged 20 years and over registered with GP used to calculate annual person years at risk
  Data type: Crude annual incidence density rates
  Result for venous leg ulcers: 1 per 1000 (or 0.1%) person years

  Row E. Petherick et al (2)
  Data source: THIN dataset we extracted year 2006 data from graph in paper
  Denominator: All patients aged 20 years and over registered with GP used to calculate annual population at risk
  Data type: Crude annual venous leg ulcer period prevalence
  Result for venous leg ulcers: 1.4 per 1000 persons (or 0.14%)

  We have interrogated the available data in order to seek an explanation of these diverging prevalence estimates and our aim here is to encourage more discussion and ideally methodological clarification. We acknowledge that there are a number of reasons why the estimates presented here may differ, indeed we have explored the impact on our data of considering an adult-only population but this does not change our estimates greatly. We believe a crucial issue may be the different denominators used in calculations. Guest et al (1) used ‘active’ patients only whereas Petherick et al (2) used the total number of patients registered on THIN and Hall et al (3) used the total geographical population. We have undertaken further analysis to try to resolve these differences (not presented here) but our understanding would be aided hugely by further methodological insights from Guest et al.

  In summary, information which would be valuable to aid transparency includes:

  • The Read Codes that were used to identify the wounds – we note recent developments that allow clinical codes to be added to a repository for transparency; (4)
  • Information on how people whose surgical wounds were defined as healed after 4 weeks were identified;
  • The exact number of wounds that estimates are based on, and 95% CIs around estimates;
  • How the issue of multiple wounds was accommodated: we note that in our point prevalence survey in Leeds (3) 37% of patients were reported to be receiving treatment for more than one complex wound;
  • Details about the denominator used, in particular what is meant by the term ‘active patients’

  We obviously all have the shared goal of ensuring that the most accurate figures for the prevalence of complex/chronic wounds are publicly available to ensure accurate decision making at every level of health care.

  References

  1. Guest JF, Ayoub N, McIlwraith T, et al. Health economic burden that wounds impose on the National Health Service in the UK. BMJ Open 2015;5:e009283. doi: 10.1136/bmjopen-2015-009283.

  2. Petherick E, Pickett K, Cullum N. Can different primary care databases produce comparable estimates of burden of disease: results of a study exploring venous leg ulceration. Family Practice 2015; 32: 374-380.

  3. Hall J, Buckley HL, Lamb KA, Stubbs N et al. Point prevalence of complex wounds in a defined United Kingdom population. Wound Repair and Regeneration 2014;22:694-700.

  4. Springate DA, Kontopantelis E, Ashcroft DM et al. ClinicalCodes: an online clinical codes repository to improve the validity and reproducibility of research using electronic medical records. PLoSOne 2014;9:e99825. doi: 10.1371/journal.pone.0099825.

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top
• Published on: 15 March 2016

  Questions about Comorbidities

  • Julian F. Guest, Director

  There was never any intention to match the wound patients with their matched controls on the basis of their comorbidities, since differences in comorbidities between the groups is an outcome we wanted to measure. The patients' records did not describe the severity of their comorbidities in all cases. Furthermore, it would have been difficult to disentangle resource use for the comorbidities from that associated with wound...

  Show More

  There was never any intention to match the wound patients with their matched controls on the basis of their comorbidities, since differences in comorbidities between the groups is an outcome we wanted to measure. The patients' records did not describe the severity of their comorbidities in all cases. Furthermore, it would have been difficult to disentangle resource use for the comorbidities from that associated with wound care in all instances. It would also have proved very challenging to identify which resources were used to manage individual comorbidities as many patients in both groups had two or more comorbidities. Hence, we employed the two methods described in the article to separate the cost of comorbidities from that of wound care. We recognise this is methodologically limited and we state in the article that the possibility of resource use associated with managing a comorbidity being conflated with that of wound management cannot be excluded. Previous experience working with the THIN database indicated that the population size of 1,000 patients per group would be sufficient to address the study's objectives.

  Conflict of Interest:

  None declared

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top
• Published on: 15 March 2016

  Questions about Comorbidities

  • Marissa J. Carter, President

  Guest et al are to be congratulated on a well-conducted study providing further insight into the economic burden of chronic wounds in the UK.

  I assume that random sampling was chosen because of the considerable work required just to assess the economic resources used in the 1,000 patients sampled in each group. The authors then chose to employ sensitivity analysis in regard to the number of comorbidities rather...

  Show More

  Guest et al are to be congratulated on a well-conducted study providing further insight into the economic burden of chronic wounds in the UK.

  I assume that random sampling was chosen because of the considerable work required just to assess the economic resources used in the 1,000 patients sampled in each group. The authors then chose to employ sensitivity analysis in regard to the number of comorbidities rather than other approaches. This begs some questions. For example, did the authors considered weighting comorbidities in terms of severity and likely resources consumed? Did the authors attempt to determine the variance of the random sampling technique--for example by resampling 100 times and determining the 95% confidence intervals of the percentages of the listed comorbidities? Last, did the author settle on 1,000 as a reasonable number or was some prior research employed to choose this number over (for example) 500 or even 5,000?

  Conflict of Interest:

  None declared

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top