RESULTS
From February 2008 through April 2013, 109 consenting women were randomized, 57 in the exercise group and 52 in the usual care group (Figure 1). One woman was excluded because of diabetes type 1 (randomised in error) and another was lost to follow-up (was not followed in our unit and delivered in another hospital), leaving 107 women for the analysis.
Baseline characteristics are summarised in Table 1. There were more obese women (BMI 30 kg/m2 or more) randomized in the intervention group and more nulliparous women in the usual care group. In the intervention group, six women (11%) never attended the exercise sessions and two women (4%) had minimal participation (one session only). We considered as compliant to the intervention 35 women (64%), 32 because they attended more than 50% of the scheduled sessions between randomisation and delivery and 3 because they did not, but reached on average more than 5000 steps per day during the same period. Average duration of the exercise sessions was 43 (SD 5) minutes, the average heart rate during the exercise sessions was 114 (SD 12) bpm, and the average peak heart rate was 142 (SD 17) bpm.
The incidence of insulin prescription was not significantly different between groups: 31 women (56%) in the exercise group compared with 24 (46%) in the control group (relative risk 1.22, 95% CI 0.84 to 1.78; P=0.39) (Table 2). Days between randomization and start of insulin treatment were not significantly different: 16 (SD 13) and 17 (SD 15) in the exercise and usual care group, respectively (P= 0.62). The means of the maximal dose of insulin were similar in the two groups.
Between 26 and 32 weeks, mean postprandial glucose was 9.4 (SD 1.8) mmol/L in the exercise group, compared to 8.9 (SD 2.0) mmol/L in the usual care group, (P=0.22). The number of postprandial glucose values per week above 8.0 mmol/L was similar (4.1, SD 3.4 and 3.8, SD 3.1; P= 0.77). Between 32 and 36 weeks, mean postprandial glucose was 9.0 (SD 2.1) mmol/L in the exercise group, compared to 8.9 (SD 1.8) mmol/L in the usual care group, (P=0.75). The number of postprandial glucose values above 8.0 mmol/L was also similar (3.9, SD 3.4 and 4.1, SD 3.4; P=0.84).
Secondary outcomes, including gestational weight gain, caesarean section, maternal complications, did not differ significantly between groups. Neonatal outcomes were also similar between groups.