- Click here for explanation of classification scheme.

The aims of the present study were to:

• evaluate dietary intakes of antioxidant vitamins, zinc, and selenium and serum levels of zinc and selenium in women with different degrees of gestational hyperglycemia
• verify whether, after adjustment for multiple confounders, these micronutrients were independent predictors of gestational hyperglycemia.

• All women gave informed consent, all procedures were in accordance with the Helsinki Declaration as revised in 1983.

Pregnant women known to have diabetes mellitus or a disease affecting glucose metabolism, active infections, chronic illness, or medical treatments (drugs or vitamin supplements) were excluded. 

Recruitment

Cohort 1:

The study included all consecutive women screened from April 1999 - November 2000. All pregnant women attending the Department of Obstetrics and Gynecology of the University of Turin (Turin, Italy) are routinely screened with a 50-g oral glucose tolerance test (OGTT) at 24 - 28 wk of gestational age (as calculated by ultrasound examinations performed during the first trimester of gestation). All patients with a positive screening result (1-h serum glucose concentration >=7.8 M/L underwent a 3-h OGTT with 100 g glucose after 1 to 2 weeks. GDM was diagnosed when two serum glucose values were above the following levels: 5.3 mM/L after fasting, 10.0 mM/L at 1 h, 8.6 mM/L at 2 h, or 7.8 mM/L at 3 h according to established criteria . The category one abnormal value on the OGTT was defined when just one glucose value was higher than the noted cutoff levels.

Cohort 2:

To confirm the preliminary data obtained from the questionnaire analyses, serum levels of zinc and selenium were determined in a second cohort of women (24 to 30 wk of gestational age). This group included 71 consecutive hyperglycemic patients(42 with one abnormal value on the OGTT and 29 with GDM) in the same department from December 2000 to June 2001. All consecutive normoglycemic (n=123) during the first two weeks of this second period were enrolled as controls.

Design : Cohort

Blinding used :

Dietetic interviews were performed by 3 physicians who were trained in assessing habitual food patterns and blinded to the results of the OGTT; other physicians communicated the results of the OGTT to patients after the dietary interview. 

Intervention (if applicable):  Not applicable 

Statistical Analysis

• The distribution of dietary intakes of micronutrients was tested with the Kolmogorov-Smirmov test and was normal.
• Analysis of variance and chi-square test were used to compare means for continuous variables or frequencies for discrete variables
• Multiple logistic regression analyses based on the maximum likelihood method were performed to evaluate the independent contribution of dietary intakes and serum levels of zinc and selenium on gestational hyperglycemia after adjusting for for multiple variables and for total energy intake according to literature.
• Mutiple regression analyses were used to evaluate the association between birth weights of offspring of hyperglycemic women and maternal serum levels of trace elements after adjusting for maternal age, BMI, smoking status, and gestational age at birth.

Timing of Measurements Height was measured at the time of screening. Other timing of measurements were not specified.

Dependent Variables

• Age (y)
• Gestational age (wk)
• BMI (kg/m²) - Height was measured at the time of screening,and patirnts supplied information on their prepregnancy weight. Body Mass index was calculated as prepregnancy weight (kg) divided by height squared (m²). 
• Familial diabetes
• Zinc (mg/d) - serum zinc levels were analyzed by flame atomic absorption spectrometry (4110ZL spectometer, Perkin-Elmer); coefficients of variation were 4.8% to 5.3%.*
• Selenium (µ M/L) - serum selenium was determined by electrothermic atomic absorption spectrometry, Perkin-Elmer, Norwalk, CT,USA) with a graphic furnance and Zeemans background correction; coefficients of variation were 5.6% to 7.6%.*
• Total energy (kcal/d) - the reliability of the reported energy intake was assessed by calculating the ratio of estimated energy intake on predicted basal metabolic rates using age-and sex-specific formulas derived by Schofield et al. A value of 0.88 which represented the 97.55 confidence interval (CI), was the cut-off point for under-reporting. The 18 under-reporters (3.6%), according to the above formula, were equally distributed across the three categories of glucose tolerance. 

* Analytical quality of serum levels of selenium and zinc was checked by analyzing standard reference materials (normal human serum) at two concentration levels (UTAK B1, level trace elements: normal-range code 66816 and high range code 66815). Results showed that analytical procedures were reliable.

Independent Variables

• Age (y)
• BMI (kg/m²)
• Familial diabetes (%)*
• Nulliparious (%)
• Smoking (%)
• Physical Activity (%): Light;Moderate;Vigorous
• Total energy (kcal/d) - See Dietary Data Section
• Serum Glucose - measured with the glucose oxidase method (Glucose-Analyser II, Beckman, Fullerton, CA, USA).
• Zinc (mg/d) 
• Selenium (µg/d)
• Vitamin A (µg/d)
• Vitamin C (mg/d- measured in first cohort only)

Dietary Data:

Detailed dietetic interviews were performed with all patients by means of a semiquantitative food-frequency questionnaire adapted from Willett and Coll (1985) with the following considerations:

• subjects were European
• searched for diet influences on the occurence of gestational hyperglycemia that was closely related to a short-lasting event pregnancy, and different from chronic conditions
• an evaluation of dietary intake in the previous years was meaningless because women frequently cahnge their dietary habits in pregnancy (more "healthy habits, avoidance of some foods, etc).

An "ad hoc" questionnaire, designed by an expert dietitian, included a list of foods (60 items) most frequently consumed in northen Italy. Commonly consumed portions were specified.

Dietetic interviews were performed by 3 physicians who were trained in assessing habitual food patterns and blinded to the results of the OGTT; other physicians communicated the results of the OGTT to patients after the dietary interviews. The procedures utilized were as follow:

• photographs were used to compare different portions for each item
• patients were asked to indicate how often a day and how many days a week they consumed every specific food during the past week
• data on cooking methods were also collected.

The food list completed by the physicians were entered into a software program (Food Meter, Medimatica s.r.I., Martinsicuro, Teramo, Italy,1990; modified in 1998 in the University of Turin according to food composition tables by the Italian National Institute of Nutrition). Subgroups of foods within the same category and with similar composition were collapsed in single items. Every item carried the weighted mean of macronutrients or minerals and vitamins. The database contained all foods consumed by the subjects.

 

Initial N: The first cohort included 504 pregnant women(210 with hyperglycemia and 294 with normoglycemia). A second cohort of  71 hyperglycemic and 123 normoglycemic women,with a mean age and body mass index similar to those in the first cohort.

Attrition (final N):  as above

Age: See Table 1

Ethnicity: All subjects identified as European

Other relevant demographics:

Anthropometrics

Location:  University of Turin,Turin,Italy

In the two cohorts, patients with one abnormal value on the OGTT and GDM were older and had significantly higher BMI values before pregnancy (Table I).

Table 1: Clinical characteristics and nutritional data obtained by dietary assessment in the first cohort and characteristics of the second cohort of patients

	Normoglycemia	one abnormal value on the OGTT	GDM	P
First cohort
Subjects	294	84	126
Age (y)	31.8±4.4	33.0±4.9	33.0±4.8	0.02
BMI (kg/m2)	23.6±4.6+	26.0±5.5♦	25.4±5.3	<0.0001
Familial diabetes (%)*	28+	33	41	0.04
Nulliparous(%)	62	57	55	NS
Smoking (%)	27	25	33	NS
Physical Activity (%)
Light	21	21	24	NS
Moderate	65	66	67	NS
Vigorous	14	13	9	NS
Total energy(kcal/d)	2035±627	2002±568	1889±490	NS
Zinc (mg/d)	9.4±2.8+	8.7±2.5	8.5±2.4	0.007
Selenium(µg/d)	33.7±2.8+	31.2±10.0	30.5±10.6	0.01
Vitamin A(µg/d)	890.0±459.3	837.7±399.4	857.0±417.5	NS
Vitamin E(mg/d)	12.9±27.1	17.6±52.6	11.7±8.3	NS
Vitamin C(mg/d)	127.9±55.7	117.6±52.3	125.4±60.4	NS
Second cohort
Subjects	123	42	29
Age (y)	31.1±3.8+	32.7±3.9	33.5±3.9	0.002
BMI (kg/m2)	23.1±4.8	25.1±5.3	25.4±6.0	0.02
Familial diabetes (%)*	29	33	38	NS
Nulliparous(%)	61	74	58	NS
Smoking (%)	24	27	27	NS
Physical activity (%)
Light	22	20	23	NS
Moderate	64	65	63	NS
Vigorous	13	10	12	NS

BMI, body mass index; GDM, gestational diabetes; NS, not significant; OGTT, oral glucose tolerance test

*First-degree relative.

+p<0.01, patients with normoglycemia versus those with GDM.

♦ p<0.01, patients with normoglycemia versus those with one abnormal value on the OGTT.

In multiple logistic regression analysis, zinc and selenium dietary intakes were negatively associated with gestational hyperglycemia (one abnormal value on the OGTT and GDM; odds ratio=.089, 95% CI=0.82 to 0.96, P=0.002 for each increment in milligrams of daily zinc intake; odds ratio=0.97, 95%  CI=0.95 to 0.99, P=0.003 for each increment in micrograms of daily selenium intake) after adjusting for age, gestational age, familial diabetes, BMI before pregnancy, and total caloric intake (Table 2).

Table 2: Variables associated with gestational hyperglycemia* in multiple logistic regression

 

Variables	OR	95% CI	P
First cohort
Age (y)	1.09	1.04-1.14	0.0008
Gestational Age (wk)	0.93	0.83-1.03	NS
BMI (kg/m2)	1.12	1.06-1.17	<0.0001
Familial diabetes♦	1.23	0.78-1.94	NS
Zinc (mg/d)	0.89	0.82-0.96	0.002
Selenium (µg/d)	0.97	0.95-0.99	0.003
Total energy (kcal/d)	0.99	0.98-1.02	NS
Second cohort
Age (y)	1.20	1.09-1.33	0.0003
Gestational age (wk)	1.06	0.97-1.15	NS
BMI (kg/m2)	1.10	1.02-1.18	0.02
Familial diabetes♦	1.06	0.48-2.34	NS
Zinc (µ M/L)	0.93	0.91-0.96	<0.0001
Selenium (µ M/L)	0.92	0.8-0.95	<0.0001

BMI, body mass index; CI, confidence interval; NS, not significant; OR, odds ratio

*Gestational diabetes and one abnormal value of the oral glucose tolerance test.

♦ First degree relatives.

Other Findings:

Patients with gestational hyperglycemia showed significantly lower serum levels of zinc and selenium (normoglycemia, one abnormal value of the OGTT, and GDM, respectively: 15.7±2.5µ M/l, P<0.0001 for serum zinc; 10.8±1.8 µM/L, 9.7±1.4µ M/L, and 8.8±1.3µ M/L, P<0.0001 for serum selenium).

 

 

The data suggested a significant inverse association of dietary intakes and serum levels of zinc and selenium with gestational hyperglycemia.

Relaibility of data:

• Total energy (kcal/d) - the reliability of the reported energy intake was assessed by calculating the ratio of estimated energy intake on predicted basal metabolic rates using age-and sex-specific formulas derived by Schofield et al. A value of 0.88 which represented the 97.55 confidence interval (CI),was the cut-off point for under-reporting. The 18 under-reporters (3.6%), according to the above formula, were equally distributed across the three categories of glucose tolerance. 
• Analytical quality of serum levels of selenium and zinc was checked by analyzing standard reference materials (normal human serum) at two concentration levels (UTAK B1, level trace elements: normal-range code 66816 and high range code 66815). Results showed that analytical procedures were reliable.

Limitations:

• Assessing intakes of microelements by a food frequency questionnaire does not represent a very sensitive method . The principle sources of zinc for the cohorts were meat, milk, cheese, and whole grains.
• Dietary intake and serum levels of micronutrients were obtained from different cohorts. The inverse association between serum levels of selenium and zinc and gestational hyperglycemia obtained in the second group strengthened the data obtained in the first one by questionnaire analysis (lower dietary intake).

University/Hospital:

University of Turin

The limitations and critique of the study, as stated by the authors appear to be very appropriate.

Analytical longitudinal surveys refer to what epidemiologists term prospective or cohort studies. A Cohort Study is a study in which patients who presently have a certain condition and/or receive a particular treatment are followed over time and compared with another group who are not affected by the condition under investigation. Studies of this kind provide a better opportunity than one time cross sectional studies to examine whther certain behaviors do in fact lead to (or cause) the disease.