Pre-Term Disparities and Intervention Grant


Valerie Montgomery-Rice MD Chair, Ob/Gyn, Meharry
Rani Lewis MD OB-GYN, Meharry
Pat Temple MD Pediatrics, VU and Nurses for Newborns
Melanie Lutenbacher VU School of Nursing
Kelly Bennett MD VU Ob/Gyn
Connie Graves MD VU Ob/Gyn
Diana Marver PhD VU Medical Admin, Director of Research & Research Training, M-V Alliance
Paulette Tempro  
Luis Litonjua  
Frank Harrell PhD VU Biostatistics


Wednesdays 7:30am-9:00am, Meharry-Vanderbilt Alliance office, 1919 Charlotte Ave (corner of 20th & Charlotte, behind Baptist Hospital), Suite 300.

E-mail list:,,,,,,,, diana.marver@Vanderbilt.Edu,,,, f.harrell@Vanderbilt.Edu

Meeting Minutes

Discussion and Questions

Study Design

The clinical trial has a 2x2x3 factorial design in which all possible combinations of treatments are randomized. The treatment combinations are dental (SOC (standard of care) vs. deep cleaning), Nursing for Newborns vs SOC, and progesterone (none, vaginal, or IM). Mothers with no periodontal disease are not randomized to standard dental care vs. deep cleaning but are part of a sub-factorial design that is 2x3 (NN +/- x progesterone none/vaginal/IM). Subjects will be accrued so that there are equal numbers of subjects with and without periodontal disease. Sample sizes are determined so that the comparison of the two smallest groups (SOC vs. deep cleaning in mothers with periodontal disease) has sufficient precision. Comparisons involving the other two dimensions of therapy will have greater precision than this. The effect of Nursing for Newborns has twice the precision, and comparisons of two of the three progesterone groups will have 4/3 the precision.

Response Variables

  1. Time to gestation in days (primary)
  2. Stress score (secondary)
  3. Serum progesterone level at 31 weeks

Analysis Plan

The impact of dental therapy will be assessed in the subgroup of subjects with baseline periodontal disease. For the other two factors, effects will be assessed by pooling over the two disease groups. Comparisons involve contrasting two or three treatment groups at a time, adjusting for the main effects of the other treatments and baseline disease status using additive regression models.

For the primary response, the Cox proportional hazards model will be used. For the stress score, ordinary multiple regression will be used. For progesterone level, ordinary regression will be used adjusting for baseline progesterone as a covariate.
  • Study Design Schematic:
DirectedGraphPlugin_1.png diagram

Obtain high resolution diagram here
Topic attachments
I Attachment Action Size Date Who Comment
plan.pdfpdf plan.pdf manage 5.2 K 18 May 2004 - 17:46 FrankHarrell High-resolution Study Design Schematic
Topic revision: r8 - 19 Oct 2013, WikiGuest

This site is powered by FoswikiCopyright © 2013-2017 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Vanderbilt Biostatistics Wiki? Send feedback