Some tests are fairly accurate in identifying women who are at increased risk of premature delivery, while others have  proven ineffective. 

• Cervical length. The length of a woman's cervix is measured using vaginal ultrasound. Women with a shorter-than-average cervix and those whose cervix shortens on subsequent exams are at increased risk of premature delivery. This test is fairly accurate in determining which women are at lower risk of premature delivery.
• Fetal fibronectin. Fibronectin is a biological glue that helps attach the fetal sac to the uterine lining. It is normally seen in vaginal secretions up to 22 weeks of pregnancy, then not until one to three weeks before delivery. A swab is used to take a sample of vaginal secretions between 22 and 34 weeks of pregnancy. If fibronectin is seen, a woman appears to be at increased risk of premature labor. This test shows moderate success in predicting who will not deliver prematurely. In some cases, this test may be combined with a measurement of cervical length to increase accuracy.
• Salivary estriol. Because levels of this hormone in the saliva appear to increase just before labor, attempts have been made to measure it. Studies show that this test is not accurate.
• Home uterine monitoring. Women at high risk have been monitored for painless contractions in an attempt to diagnose premature labor early, when it was most treatable. A number of studies have shown that home uterine monitoring is not effective in preventing premature delivery.
  
  May 2007