Jump to comment:

• Response to Comments on the analyses and the generalizability of findings from the Economic Evaluation of SIFT
  Tracy E Roberts
  Published on: 25 October 2021
• Comments on the analyses and the generalizability of findings from the Economic Evaluation of SIFT
  Manoj Kumar
  Published on: 25 October 2021

• Published on: 25 October 2021

  Response to Comments on the analyses and the generalizability of findings from the Economic Evaluation of SIFT

  • Tracy E Roberts, Professor of Health Economics Health Economics Unit, Institute of Applied Health Research, University of Birmingham, Birmingham, UK

  We thank the authors for the comments on the Economic Evaluation of SIFT (1) and we are grateful for the opportunity to respond to their comments.
  Taking each of the authors’ points in the order in which they are presented:
  1. In relation to the first point about the loss to follow up and the exclusion of such patients from the analysis, we point out that we used complete case analysis and accounted for the missing patients following best practice using a multiple imputation analysis which is provided in the supplementary materials. We state the following in the paper:

  “Mean total costs for all infants, adjusting for missing data using multiple imputation, are found in the online supplementary table S3. When the missing values were accounted for, faster feed increments remain more costly in comparison to slower feed increments but at a slightly higher level (£378 more) per infant, reflecting the high level of uncertainty in the difference in costs, especially with regard to the healthcare resource use after discharge estimated by the multiple imputation” (last paragraph of methods))

  2. In relation to the authors second concern, whilst death was slightly higher in the slower feeds arm during initial hospital stay there are two important points in response to this. First, we clarify that by definition economic analysis is not an exercise in accountancy where death is assumed to incur a zero cost, because economic evaluation focuses on costs and ou...

  Show More

  We thank the authors for the comments on the Economic Evaluation of SIFT (1) and we are grateful for the opportunity to respond to their comments.
  Taking each of the authors’ points in the order in which they are presented:
  1. In relation to the first point about the loss to follow up and the exclusion of such patients from the analysis, we point out that we used complete case analysis and accounted for the missing patients following best practice using a multiple imputation analysis which is provided in the supplementary materials. We state the following in the paper:

  “Mean total costs for all infants, adjusting for missing data using multiple imputation, are found in the online supplementary table S3. When the missing values were accounted for, faster feed increments remain more costly in comparison to slower feed increments but at a slightly higher level (£378 more) per infant, reflecting the high level of uncertainty in the difference in costs, especially with regard to the healthcare resource use after discharge estimated by the multiple imputation” (last paragraph of methods))

  2. In relation to the authors second concern, whilst death was slightly higher in the slower feeds arm during initial hospital stay there are two important points in response to this. First, we clarify that by definition economic analysis is not an exercise in accountancy where death is assumed to incur a zero cost, because economic evaluation focuses on costs and outcomes together in a ratio. Thus, death causes a severe penalty in the denominator of the analysis as fewer survivors in the denominator means the cost effectiveness is less favorable ( eg. Dividing an integer by a zero gives a result tending to infinity). Second, death alone was not the primary outcome. The primary outcome was neurodevelopmental disability at aged 2, since surviving with severe disability will incur substantial cost.

  3. With respect to point 3, we refer to the response to point one - in accordance with best practice multiple imputation was carried out to account for missing data.

  4. We do not exclude the parental nutrition costs - these are presented in the first line of Table 1 and are supported by the reference of Walter et al. (see reference 28) in the paper (1)

  5. In terms of generalizability, whilst the transferability of healthcare resource use and costs will be affected by differences in clinical practice and relative prices across countries, varying costs alone may not be sufficient (2).

  Finally, whilst our paper concluded that SIFT would not be deemed cost- effective, that result was not driven by the costs alone. The clinical paper showed no statistical difference in the primary outcome of survival without moderate or severe disability (3). The recommended approach to economic evaluation is not to use statistical significance but to estimate and quantify the uncertainty that is implied using simulation techniques. Following these techniques the analysis suggested that the Sift intervention of faster feeds had potential to be harmful as reported in the paper. To assume that the results of the paper are driven by cost is to misunderstand health economics and the definition of cost-effectiveness which is not based on cost alone but a ratio of the difference in costs divided by the difference in effects.

  Yours sincerely

  Tracy Roberts On behalf of all co authors
  Sift authorship

  References:

  1. Tahir W, Monahan M, Dorling J, et al. Economic evaluation alongside the Speed of Increasing milk Feeds Trial (SIFT). Arch Dis Child Fetal Neonatal Ed. 2020 Nov;105(6):587-592. doi: 10.1136/archdischild-2019-318346.
  2. Drummond M, Barbieri M, Cook J, Glick HA, Lis J, Malik F, et al. Transferability of economic evaluations across jurisdictions: ISPOR Good Research Practices Task Force report. Value in health. 2009; 12 (4):409–18. https://doi.org/10.1111/j.1524-4733.2008.00489.x PMID: 19900249).
  3. Dorling J, Abbott J, Berrington J, et al; SIFT Investigators Group. Controlled Trial of Two Incremental Milk-Feeding Rates in Preterm Infants. N Engl J Med. 2019 Oct 10;381(15):1434-1443. doi: 10.1056/NEJMoa1816654.

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top
• Published on: 25 October 2021

  Comments on the analyses and the generalizability of findings from the Economic Evaluation of SIFT

  • Manoj Kumar, Staff Neonatologist Department of Pediatrics, University of Alberta, Edmonton, Canada

  I read with interest the economic evaluation of Speed of Increasing milk Feeds Trial (SIFT) in preterm infants presented by Tahir and colleagues.(1) While the clinical findings from the SIFT had shown short-term benefits such as lesser TPN days with faster feed increments, and was equivocal for the composite primary outcome of death and disability measured at 24 months, this analysis recommends against faster feed increments based on the cost-effectiveness analyses. The average total costs is shown to be marginally higher for subjects in this arm, with a mean difference of £267 (0.25%). I highlight below many issues that probably affect the conclusions, and the generalizability of the findings, of this economic evaluation.

  First, the trial enrolled 1394 and 1399 patients in the two study arms. However, the cost data of initial hospitalization is presented for 1224 and 1246 patients in these arms. It seems that 170 and 153 patients from the two trial arms were lost to follow-up after the initial hospitalization;(2) however, the data for these subjects should not be excluded while calculating the average initial hospitalization costs per subject.

  Second, more deaths during the initial hospitalization were reported in the slower increment arm and those probably lowered the average cost for this arm. It is well known that the hospitalization costs for very preterm infants that die during the neonatal period are substantially lower than those of the survivors....

  Show More

  I read with interest the economic evaluation of Speed of Increasing milk Feeds Trial (SIFT) in preterm infants presented by Tahir and colleagues.(1) While the clinical findings from the SIFT had shown short-term benefits such as lesser TPN days with faster feed increments, and was equivocal for the composite primary outcome of death and disability measured at 24 months, this analysis recommends against faster feed increments based on the cost-effectiveness analyses. The average total costs is shown to be marginally higher for subjects in this arm, with a mean difference of £267 (0.25%). I highlight below many issues that probably affect the conclusions, and the generalizability of the findings, of this economic evaluation.

  First, the trial enrolled 1394 and 1399 patients in the two study arms. However, the cost data of initial hospitalization is presented for 1224 and 1246 patients in these arms. It seems that 170 and 153 patients from the two trial arms were lost to follow-up after the initial hospitalization;(2) however, the data for these subjects should not be excluded while calculating the average initial hospitalization costs per subject.

  Second, more deaths during the initial hospitalization were reported in the slower increment arm and those probably lowered the average cost for this arm. It is well known that the hospitalization costs for very preterm infants that die during the neonatal period are substantially lower than those of the survivors.(3)

  Third, health services costs for 2 years after the initial hospitalization was collected via a patient questionnaire, completed by 842 and 873 parents for the two trial arms. The missing data on a large percentage of study subjects could have resulted in biased estimates for these costs.

  Fourth, unit cost of resource items presented in table 1, omit the parenteral nutrition costs that tend to be substantial.(4) Since the trial showed less parenteral nutrition use in the faster arm, the inclusion of these costs (as planned in the protocol), could have had a significant impact on the analyses.

  Also, this economic evaluation lacks generalizability, in terms of its applicability to other jurisdictions, as no sensitivity analyses were conducted for a range of cost values for salient resource items that vary significantly across countries with similar economic development. For example, the costs listed for community paediatrician’s visit (£407 per visit) and donor breast milk (£335/litre) are 4 to 6 times higher than the typical costs for these items in the North America.

  References
  1. Tahir W, Monahan M, Dorling J, et al. Economic evaluation alongside the Speed of Increasing milk Feeds Trial (SIFT). Arch Dis Child Fetal Neonatal Ed. 2020 Nov;105(6):587-592. doi: 10.1136/archdischild-2019-318346.

  2. Dorling J, Abbott J, Berrington J, et al; SIFT Investigators Group. Controlled Trial of Two Incremental Milk-Feeding Rates in Preterm Infants. N Engl J Med. 2019 Oct 10;381(15):1434-1443. doi: 10.1056/NEJMoa1816654.

  3. Phibbs CS, Schmitt SK, Cooper M, et al. Birth Hospitalization Costs and Days of Care for Mothers and Neonates in California, 2009-2011. J Pediatr. 2019 Jan;204:118-125.e14. doi: 10.1016/j.jpeds.2018.08.041.

  4. Walter E, Liu FX, Maton P, et al. Cost analysis of neonatal and pediatric parenteral nutrition in Europe: a multi-country study. Eur J Clin Nutr. 2012 May;66(5):639-44. doi: 10.1038/ejcn.2011.225.

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top