TY - JOUR
T1 - Predicting deseasonalised serum 25 hydroxy vitamin D concentrations in the D-Health Trial: An analysis using boosted regression trees
AU - Waterhouse, Mary
AU - Baxter, Catherine
AU - Duarte Romero, Briony
AU - McLeod, Donald S.A.
AU - English, Dallas R.
AU - Armstrong, Bruce K.
AU - Clarke, Michael W.
AU - Ebeling, Peter R.
AU - Hartel, Gunter
AU - Kimlin, Michael G.
AU - O'Connell, Rachel L.
AU - Pham, Hai
AU - Rodney Harris, Rachael M.
AU - van der Pols, Jolieke C.
AU - Venn, Alison J.
AU - Webb, Penelope M.
AU - Whiteman, David C.
AU - Neale, Rachel E.
N1 - Funding Information:
The D-Health Trial is funded by National Health and Medical Research Council (NHMRC) project grants (GNT1046681, GNT1120682). PM Webb and DC Whiteman are supported by fellowships from the NHMRC (GNT1173346, GNT1155413). DSA McLeod is supported by a Metro North Clinician Research Fellowship and a Queensland Advancing Clinical Research Fellowship. H Pham is supported by a University of Queensland PhD Scholarship. MW Clarke is affiliated to Metabolomics Australia, University of Western Australia, Perth, Western Australia, Australia. Measurement of serum 25(OH)D concentration was supported by infrastructure funding from the Western Australian State Government in partnership with the Australian Federal Government, through Bioplatforms Australia and the National Collaborative Research Infrastructure Strategy (NCRIS).
Publisher Copyright:
© 2021 Elsevier Inc.
PY - 2021/5
Y1 - 2021/5
N2 - Background: The D-Health Trial aims to determine whether monthly high-dose vitamin D supplementation can reduce the mortality rate and prevent cancer. We did not have adequate statistical power for subgroup analyses, so could not justify the high cost of collecting blood samples at baseline. To enable future exploratory analyses stratified by baseline vitamin D status, we developed models to predict baseline serum 25 hydroxy vitamin D [25(OH)D] concentration. Methods: We used data and serum 25(OH)D concentrations from participants who gave a blood sample during the trial for compliance monitoring and were randomised to placebo. Data were partitioned into training (80%) and validation (20%) datasets. Deseasonalised serum 25(OH)D concentrations were dichotomised using cut-points of 50, 60 and 75 nmol/L. We fitted boosted regression tree models, based on 13 predictors, and evaluated model performance using the validation data. Results: The training and validation datasets had 1788 (10.5% <50 nmol/L, 23.1% <60 nmol, 48.8 <75 nmol/L) and 447 (11.9% <50 nmol/L, 25.7% <60 nmol/L, and 49.2% <75 nmol/L) samples, respectively. Ambient UV radiation and total intake of vitamin D were the strongest predictors of ‘low’ serum 25(OH)D concentration. The area under the receiver operating characteristic curves were 0.71, 0.70, and 0.66 for cut-points of <50, <60 and <75 nmol/L respectively. Conclusions: We exploited compliance monitoring data to develop models to predict serum 25(OH)D concentration for D-Health participants at baseline. This approach may prove useful in other trial settings where there is an obstacle to exhaustive data collection.
AB - Background: The D-Health Trial aims to determine whether monthly high-dose vitamin D supplementation can reduce the mortality rate and prevent cancer. We did not have adequate statistical power for subgroup analyses, so could not justify the high cost of collecting blood samples at baseline. To enable future exploratory analyses stratified by baseline vitamin D status, we developed models to predict baseline serum 25 hydroxy vitamin D [25(OH)D] concentration. Methods: We used data and serum 25(OH)D concentrations from participants who gave a blood sample during the trial for compliance monitoring and were randomised to placebo. Data were partitioned into training (80%) and validation (20%) datasets. Deseasonalised serum 25(OH)D concentrations were dichotomised using cut-points of 50, 60 and 75 nmol/L. We fitted boosted regression tree models, based on 13 predictors, and evaluated model performance using the validation data. Results: The training and validation datasets had 1788 (10.5% <50 nmol/L, 23.1% <60 nmol, 48.8 <75 nmol/L) and 447 (11.9% <50 nmol/L, 25.7% <60 nmol/L, and 49.2% <75 nmol/L) samples, respectively. Ambient UV radiation and total intake of vitamin D were the strongest predictors of ‘low’ serum 25(OH)D concentration. The area under the receiver operating characteristic curves were 0.71, 0.70, and 0.66 for cut-points of <50, <60 and <75 nmol/L respectively. Conclusions: We exploited compliance monitoring data to develop models to predict serum 25(OH)D concentration for D-Health participants at baseline. This approach may prove useful in other trial settings where there is an obstacle to exhaustive data collection.
UR - http://www.scopus.com/inward/record.url?scp=85102384404&partnerID=8YFLogxK
U2 - 10.1016/j.cct.2021.106347
DO - 10.1016/j.cct.2021.106347
M3 - Article
C2 - 33684596
AN - SCOPUS:85102384404
SN - 1551-7144
VL - 104
SP - 1
EP - 11
JO - Contemporary Clinical Trials
JF - Contemporary Clinical Trials
M1 - 106347
ER -
