Three Statistics Multiple Choice Questions

NNIPS-II was a randomized study of 15,987 infants born alive to 43,559 women who received Vitamin A, Beta-carotene, or a placebo prior to and during pregnancy. (Investigators: Drs. Keith West, Joanne Katz, Parul Christian et al..) Infants were followed for up to 180 days after birth, until death (the event (outcome) being studied),
dropout or study's end (180 days). We are using a representative subset of 10,295 live births from the original data set to investigate factors associated with increased risk of infant mortality in this population. Of the 10,295 in the sample, 1,342 died within 180 days and 8,952 were censored.

A Cox-proportional hazards regression was used to investigate the relationship between infant survival, vitamin treatment, and infant and mother characteristics. Specifically, a Cox multiple regression model was fit relating infant survival to vitamin treatment, infant sex, and whether the mother had previous pregnancies. The results of the multiple regression appear in the following table:

PREDICTOR ADJUSTED HAZARD RATIO 95% Confidence Interval
Vitamin Treatment
Placebo Reference
Beta Carotene 0.97 (0.80, 1.18)
Vitamin A 1.04 (0.86, 1.25)

Mother Experienced Night Blindness
No reference
Yes 1.50 (1.14, 1.84)

Mother Has Other Children
No reference
0.69 (0.58, 0.82)

12. Based on the regression results, give an estimated adjusted hazard ratio (adjusted for treatment group and night blindness) of infant death for infants whose mothers had other children prior to pregnancy relative to those who were first children (ie: mother had not previously given birth to a child)

a. -0.37
b. 0.69
c. 1.45
d. 0.31

13. The study was a large, randomized study conducted to determine predictors of infant survival, and assess the relationship between infant survival and beta carotene. The study was designed with high power (95%) to detect a clinically significant relationship between increased survival and beta-carotene, if beta carotene was really associated with increased survival. (if true hazard ratio for beta-carotene compared to placebo was at least 0.80 indicating a 20% reduction in risk of death). Ninety-five percent power means:

a. The probability is 95% that the study finds a statistically significant association between survival and beta-carotene at the population level.
b. The probability is 95% that the study finds a statistically significant association between survival and beta-carotene if beta carotene reduces the population risk of death by at least 20% compared to placebo.
c. The probability is 95% of obtaining a sample result as or more extreme (unlikely) than was observed by chance alone.
d. The probability is 5% that the study finds a statistically significant association between survival and beta-carotene if beta carotene reduces the population risk of death by at least 20% compared to placebo.

14. No statistically significant relationship was detected between increased infant survival and beta-carotene (ie: p >> .05). Given this result, and the information from question 13, which of the following conclusions is most likely?

a.The study was not conducted properly.
b.Beta-carotene is not associated with increased infant survival.
c.The study should be conducted again with a larger sample size.
d.The study should be repeated until a statistically significant result is
found.

See attached file for full problem description.