We then created three categorical variables, each of which had 3 levels that were coded as: [losses] 0 losses (referent), 1 loss, and 2+ losses; [injuries] 0 injuries (referent), 1-2 injuries, and 3+ injuries; [summary] 0 losses/injuries (referent), 1-2 losses/injuries, and 3-7 losses/injuries.
The authors need to change their analysis, and incorporate number of medications as a categorical variable with categories: zero (referent), 1, 2, 3 and 4 or more.
Three pictures of each magnification were randomly taken for each sample (50,100 and 200 times referent to 50x, 100x and 200x respectively).
A dose-response relationship was seen; with 0 cows/1,000 m as the referent, ≤1 cow/1,000 m was associated with a 30% decrease, 1.1 2 cows/1,000 m with 40% decrease, and >2 cows/1,000 m with 43% decrease in kala-azar risk.
Children who were at risk for overweight or were obese at baseline and consumed 1 to <2 drinks/day, 2 to <3 drinks/day and ⩾3 drinks/day had a significantly higher risk to become or remain overweight compared with the referent (<1 drink/day).
Birth order was examined using two different categorical schemes: 1 (referent) versus ≥2; and 1 (referent), 2 to 4, and ≥5.
This makes sense because the referent 30-day period for drinking was different.
Maternal age was classified into three categories: <25 years old (referent), 25 to 29 years old, and ≥30 years old.
This scheme divides participants into three groups of low (referent, 22%), medium (56%) and high (22%) community poverty rates.
To account for secular trends in breast cancer incidence, all models included variables for birth cohort (before 1920 as a referent, 1920 1934, and ≥ 1935).
There was no clear association between cumulative years of smoking and colorectal cancer; the multivariate-adjusted ORs for 0, 15-29 15-29 and ≥30 years of smoking were 1.00 (referent), 0.73 (0.92-1.05), 0.92 (0.65-1.29) and 1.02 (0.69-1.49), respectively.
