Effect of injectable and oral contraceptives on glucose and insulin levels

Abstract

Objective

To estimate the effect of using two methods of hormonal contraceptives (depot medroxyprogesterone acetate (DMPA) or oral contraceptives (OC) containing 20 micrograms ethinyl estradiol and 0.15 mg desogestrel) on serum glucose and insulin levels as well as predictors of any observed changes.

Methods

Fasting glucose and insulin levels were measured on 703 white, black, and Hispanic women using DMPA, OC, or nonhormonal (NH) birth control at baseline and every 6 months thereafter for 3 years. Participants also completed questionnaires containing demographic and behavioral measures every 6 months. Mixed model regression analyses were used to estimate changes over time in glucose and insulin levels by method, along with their predictors.

Results

DMPA, but not OC, users experienced slightly greater increases in glucose and insulin as compared with NH users (p <.001). Among DMPA users, a small, but steady increase in serum glucose (2 mg/dl at 6 months to 3 mg/dl at 30 months) was observed throughout the first 30 months but leveled off after that. In contrast, serum insulin showed an upward (3 units at 6 months to 4 units at 18 months) trend for the first 18 months of DMPA use and then remained almost flat thereafter. Elevation of insulin and glucose levels was slightly more pronounced in obese and overweight DMPA users than those who were normal weight.

Conclusion

Use of DMPA, but not very low dose OC containing desogestrel, can lead to slightly higher fasting glucose and insulin levels.

Introduction

Depo medroxyprogesterone acetate (DMPA) is a highly effective, long acting method of birth control. Use of this contraceptive by young women, however, has been associated with a number of adverse effects, such as irregular menstrual bleeding (1–3) and loss of bone mineral density (4–11). It has also been suggested that DMPA may adversely affect serum glucose and insulin levels. For example, Fahmy observed that 20 Egyptian women who used DMPA for 12 months exhibited increases in their serum glucose levels both in the fasting state and during an oral glucose tolerance test (12). Additional studies have similarly shown an adverse effect of DMPA on glucose and insulin levels (12–19). However, others have not shown any effect (20, 21). Although a number of these prior studies suggest that DMPA has an adverse effect on glucose-insulin metabolism, the results of all these investigations must be interpreted with caution as they were based on small sample sizes (12, 14, 15, 17, 19–21), were cross sectional (13, 18) or had short follow up intervals (12, 15–17, 19–21). In addition, most did not include a control group for comparison (12, 14–17, 19–21). Finally, none took into account the potential effects of obesity, which has been shown to affect insulin and glucose levels.

Similarly, it is not clear whether use of very low dose oral contraceptives (OC) containing desogestrel adversely affect insulin-glucose metabolism. This third generation progestin has been used in newer birth control pills because it is less androgenic and thus should have less effect on carbohydrate metabolism than previously marketed oral contraceptives. However, most published studies which have evaluated this progestin when used in an OC containing only 20 micrograms EE are based on small sample sizes or did not include a control group (22–28). As a result, a recent Cochrane review of interventional studies related to hormonal contraception stated that results are conflicting with regards to desogestrel’s effects on carbohydrate metabolism (29).

The purpose of this study was to estimate the effects of DMPA as well as an OC containing 20 micrograms EE and 0.15 mg desogestrel over 3 years on fasting insulin and glucose levels using longitudinal data with a control group.

Methods

As part of a larger study, 805 non-Hispanic black, non-Hispanic white, and Hispanic women between 16 and 33 years of age were recruited between October 9, 2001, and September 14, 2004. The methods for the larger study are reported in detail elsewhere (5). Briefly, recruitment was conducted to achieve a sample that was balanced by age group (16–24 years and 25–33 years), race (Black, White, Hispanic) and contraceptive method: nonhormonal (NH), oral contraceptives (OCs), and depot medroxyprogesterone acetate injections (DMPA). Of the 805 women who signed a consent form for the larger study, 5 withdrew prior to completing their first visit and 97 were excluded due to abnormal laboratory or bone scan results. Thus, 703 women were invited to participate in the longitudinal study. Those excluded (n=102) did not differ from women included in the longitudinal study (n=703) on age, marital status, parity, or education (all P>.05). Written, informed consent was obtained from all participants; parental consent was obtained for participants <18 years of age. All procedures were approved by the Institutional Review Board at the University of Texas Medical Branch at Galveston.

Following counseling on the different types of contraception available and their efficacies, women were allowed to select one of three types of birth control: 245 chose OC (0.15 mg desogestrel + 20 micrograms ethinyl estradiol taken for 21 days, followed by 2 days of placebo and 5 days of 10 micrograms ethinyl estradiol); 240 chose DMPA; and 218 chose NH method. Contraception was dispensed every 3 months. All participants received free well-woman care and contraception during the study as well as monetary compensation. Those who did not return for scheduled visits were reminded by phone and certified letters.

At each visit, weight was measured with a digital scale accurate to the nearest 0.1kg while women were wearing light indoor clothing. Height was measured using a wall-mounted stadiometer (Heightronic, Snoqualmie, WA) accurate to the nearest 0.001m. BMI was calculated as weight (kg)/height(m²). To obtain estimates of dietary intake, a registered dietician conducted a 24-h recall interview with each participant Using food models to estimate portion sizes. Nutrient calculations were performed using the Nutrition Data System for Research (NDS-R) software (Nutrition Coordinating Center, University of Minnesota, Minneapolis, MN, Food and Nutrient Database, Version 33, released July 2002). Physical activity was assessed using the Exercise Module of the Behavioral Risk Factor Surveillance System Questionnaire (30). This measure lists 56 common activities for which women were asked to report the frequency and duration of up to two activities performed during the past month.

Serum glucose and insulin were measured at baseline and every 6 months thereafter. All blood samples were collected between 7:00 a.m. and 10:00 a.m. after an overnight fast. Serum insulin assays were performed on the Access Immunoassay system® (Beckman Coulter Inc., Brea, CA) using the ultrasensitive insulin chemiluminescent immunoassay and the manufacturer’s reagents and calibrators. Serum glucose assays were performed using the VITROS 5,1 FS Chemistry System (Ortho-Clinical Diagnostics, Raritan, NJ) using the manufacturer’s reagents and calibrators. Serum glucose was measured using the VITROS GLU slides, along with the VITROS chemistry products calibrator kit 1. Each slide analyzed 10 microliters of serum using a multilayered analytical element coated on a polyester support with oxidative coupling producing a dye. The intensity of the dye was measured by reflective light. Coefficients of variation for serum insulin and glucose were 4.0% and 1.5%, respectively, using a single lot of reagents calibrated weekly. Samples were assayed in batches at the University of Texas Medical Branch laboratory after the study was completed between March and November of 2009. Participants were not screened for diabetes mellitus at the beginning of the study as this is not part of our protocol for contraceptive care. When conducting the assays at the conclusion of the study, four participants were noted to have a fasting glucose level of 126 mg/dl or higher at baseline indicative of diabetes. Therefore, these four women (3 DMPA users and 1 OC user) were excluded from these analyses on glucose and insulin levels.

Statistical Analysis

One-way analysis of variance with Bonferroni correction for continuous variables and chi-square test for categorical variables were performed to compare the three contraceptive groups at baseline. We used longitudinal analyses to compare changes in blood glucose and serum insulin for each contraceptive method, along with their predictors over time. To accommodate the repeated measurements, the data were modeled with the use of a mixed effects regression procedure (xtmixed module; Stata Corporation, College Station, TX), which allowed us to obtain regression coefficients for various predictors while adjusting for the estimated errors for the repeated measurements. This class of model also allows inclusion of time-dependent covariates and accommodates subjects with incomplete data because of variation in number and spacing in observations over the period of follow-up, which frequently occurs in longitudinal studies. The primary outcomes were serum glucose and insulin. To examine the overall effect of method, race, and time, our models included contraceptive method (OC/DMPA/NH), race/ethnicity, and duration of contraceptive use (time) as main effects after adjusting for other covariates. Interaction terms (method × race/ethnicity; method × time) were then included in the model. Age, parity, previous use of pills and DMPA, lifestyle variables (smoking, alcohol use, and physical exercise), and socioeconomic variables were also examined and retained if P<.2. Separate mixed models were constructed to examine the effect (adjusted by total calories) of daily intake of protein, fat, and carbohydrates (based on 12-month follow-up data) on glucose and insulin levels. All analyses were performed using Stata 11 (Stata Corporation, College Station, TX).

Results

The mean age of the entire sample was 24 ± 5 years. Twenty-nine percent of the sample (n = 200) was non-Hispanic black, 36% (n = 256) were Hispanic (predominately Mexican American), and 35% (n = 247) were non-Hispanic white. The number of women in each racial/ethnic group and in each age category (16–24 y and 25–33 y) did not significantly differ by contraceptive method (Table 1). Furthermore, there were no significant differences between contraceptive groups in baseline height, weight, body mass index, total fat mass, percent body fat, age at menarche, previous use of birth control pills, and serum insulin and glucose levels (Table 1). NH users did have a higher mean parity, OC users were less likely to have used DMPA in the past, and DMPA users were more likely to smoke. As previously reported, follow-up data were available on 424, 290, and 182 women at 12, 24, and 36 months, respectively, (5) after excluding the four subjects with an elevated baseline glucose level).

Table 1.

Sample characteristics according to contraceptive selected at baseline

Characteristic	OC (n = 245)	DMPA (n = 240)	NH (n = 218)
Age, %
16–24 y	54.3	56.7	45.9
25–33 y	45.7	43.3	54.1
Race, %
Black	29.8	30.0	25.2
White	33.5	34.2	38.1
Hispanic	36.7	35.8	36.7
Current smoker, %	23.3§	36.3†	22.0
Weight-bearing exercise >120 min/wk, %	43.2	36.3	37.5
Mean (SD)
Weight, kg	73.3 (17.7)	71.8 (19.2)	73.2 (18.6)
BMI, kg/m2	27.9 (6.4)	27.2 (6.9)	28.3 (7.0)
Fat Mass, kg	27.0 (11.2)	25.6 (12.1)	27.2 (11.7)
Fat Mass, % of total	36.4 (7.4)	34.8 (8.4)	36.7 (7.5)
Age at menarche, y	12.2 (1.5)	12.5 (1.7)	12.2 (1.6)
Previous use of pill (months)	21.9 (31.8)	16.9 (29.6)	17.7 (28.6)
Previous use of DMPA injection (#)†	1.4 (3.5)‡	3.5 (6.5)§	2.6 (5.2)
Alcohol use, gm/day	1.6 (8.7)	1.2 (6.1)	2.3 (14.5)
Parity	0.9 (1.1)‡	1.1 (1.2)†	1.6 (1.5)
Blood glucose	84.7 (7.2)	85.8 (7.9)	84.3 (7.5)
Serum insulin	7.8 (4.9)	7.6 (4.5)	7.8 (5.0)

OC = oral contraceptive; DMPA = depot medroxyprogesterone acetate; NH = nonhormonal contraception; BMI = body mass index; SD = standard deviation One-way analysis of variance with Bonferroni correction was used for continuous variables and chi-square tests were used for categorical variables. To identify specific pairwise differences for categorical variables, we created separate 2×2 tables for each of the pairs and used chi square tests. To adjust for multiple comparison, P<.017 (.05/3) was used to indicate the statistical significance between any two contraceptive groups.

^†Significant difference was found between DMPA and nonhormonal contraception after Bonferroni adjustment

^‡Significant difference was found between OC and nonhormonal contraception after Bonferroni adjustment

Over 3 years, DMPA users experienced increases in levels of glucose and insulin which were greater than those experienced by NH and OC users (P< .05; Table 2 & Figure 1). These patterns differed somewhat by the type of assay. Glucose steadily increased during the first 30 months of DMPA use, although the greatest increase occurred during the first 6 months (2 mg/dL at 6 months to 3 mg/dL at 30 months). Serum insulin increased about 2 units during the first 6 months of DMPA use and an additional one unit during next 12 months, but then leveled off. In contrast, NH and OC users exhibited very little change in their glucose and insulin levels over the 36 months. Both groups experienced about 1 mg/dl increase in their mean glucose level over 36 months. With regards to insulin levels, NH users experienced less than a one unit increase during the 3 years of follow up while OC users had about a 2 unit increase. Very few participants had abnormal (>27 IU/ml) levels of serum insulin (NH 3, DMPA 1, OC 3) at baseline. Overall, 6 DMPA users developed an abnormally high level of insulin after 6 months of use and 9 participants had an abnormal insulin level after 18 months. However, in the majority of these cases, the serum insulin level was noted to be normal at a subsequent visit. Only two DMPA users (as compared to none of the NH users) were noted to have a normal baseline insulin level which became elevated within 6 months and remained high at subsequent visits.

Table 2.

Estimated changes in blood glucose and serum insulin across 36 months by contraceptive method^a,^b

	NH	DMPA	OC	P value
Blood glucose
Baseline	84.31	85.80	84.67	NH vs. OC: =.570
6 mo	84.08	87.25	84.10	NH vs DMPA: =<.001
12 mo	84.40	87.40	84.28	OC vs. DMPA: =<.001
18 mo	84.94	87.86	84.64
24 mo	85.17	88.43	85.10
30 mo	85.62	88.52	85.17
36 mo	85.29	88.16	85.48
Serum Insulin
Baseline	7.85	7.61	7.79	NH vs. OC: =.136
6 mo	8.54	10.24	8.87	NH vs DMPA: <.001
12 mo	8.75	10.83	8.90	OC vs. DMPA: =.012
18 mo	8.43	11.10	8.92
24 mo	8.40	10.79	9.18
30 mo	8.79	10.41	9.10
36 mo	8.75	11.08	9.61

OC = oral contraceptive; DMPA = depot medroxyprogesterone acetate; NH = nonhormonal contraception.

^aAdjusted by baseline value of glucose and insulin, weight status (time varying variable; normal weight (BMI: <25 kg/m²), overweight (BMI: 25–29.9 kg/m²), and obese (BMI: 30 or above kg/m²)), age (16–24 years vs. 25–33 years), race/ethnicity (black, white, and Hispanic), parity, smoking status (current smoker vs not), and previous use of DMPA.

^bMixed-model regression analyses were used for the adjustment

Figure 1.

Modeled mean change of blood glucose and serum insulin from baseline across 36 months by contraceptive method. Empty circles: nonhormonal contraceptives; X shape: oral contraceptives; solid triangles: depot medroxyprogesterone acetate. Solid lines represent the estimated mean changes, and dotted lines represent the unadjusted values.

BMI affected both glucose and insulin levels irrespective of the contraceptive method used. DMPA users exhibited a significantly greater increase than NH users in glucose and insulin levels with each increase in BMI category (Fig 2). Obese DMPA users had an average increase of 4 units of glucose and 5 units of insulin over 36 months. The respective increase observed in glucose and insulin levels among overweight and normal weight DMPA users were 3 and 3, and 2 and 2 units. On the other hand, changes in OC and NH users were varied between 0–2 units in different weight categories over 36 months. No racial differences were observed with regard to the effect of contraception on glucose and insulin levels. With regard to age, 25–33 years old had lower insulin levels (1.4 units lower) compared to 16–24 years old irrespective of contraceptive method over 36 months. Baseline values also affected results with women with higher levels of glucose and insulin at baseline more likely to demonstrate higher values after 36 months of contraceptive use. Overall, there was a positive trend of glucose, but not insulin levels, over 36 months. Changes in percentage of calories obtained in the diet from protein, fat, and carbohydrates per day did not predict any changes in glucose or insulin levels. However, total caloric intake was positively associated with insulin levels.

Figure 2.

Modeled mean change of blood glucose and serum insulin from baseline across 36 months by contraceptive method and body mass index status (A), blood glucose – normal weight, (B) blood glucose – overweight (C), blood glucose – obese, (D) serum insulin –normal weight (E) serum insulin –overweight, and (F) serum insulin –obese. Empty circles: nonhormonal contraceptives; X shape: oral contraceptives; solid triangles: depot medroxyprogesterone acetate.

Discussion

We observed that DMPA use resulted in slightly higher levels of both glucose and insulin. Furthermore, the magnitude of this increase in insulin and glucose levels was higher among obese DMPA users as compared to obese NH and OC users. Although we did observe an increase, both glucose and insulin levels remained within a normal range for all DMPA users who exhibited normal levels at baseline. Furthermore, the magnitude of increase in fasting glucose levels we observed is actually less than reported in some prior studies. For example, in a study of 20 DMPA users, Fahmy et al (12) observed an increase of 10 units of fasting glucose within 12 months of DMPA use while Vermeulen et al (17) observed 7 units within 7 months of use in 20 women. Variation in serum glucose levels between studies may be due to differences in sample sizes, baseline characteristics of the participants, study duration, or assay methods.

The mechanism by which DMPA causes an increase in insulin and glucose level has yet to be elucidated. One possible mechanism is a compensation for increased insulin resistance. Tuttle et al (31) reported increased free fatty acids after a glucose challenge among medroxyprogesterone users, a condition usually associated with insulin resistance (32). A direct stimulation of pancreatic B-cells was also speculated. Other possible mechanisms for DMPA-associated elevation of insulin-glucose are related to weight gain associated with DMPA or the glucocorticoid-like activity of progestogen (21).

In contrast to prior reports on third generation oral contraceptive pills, we observed no differences in fasting insulin and glucose levels between nonhormonal contraception and OCs after 36 months of use. In fact, third generation birth control pills containing norgestimate, desogestrel or gestodene were designed to minimize adverse effects associated with the androgenic activity of oral contraceptives, including unfavorable effects on lipid and carbohydrate metabolism. Since their introduction, a number of studies have confirmed that their effects on carbohydrate metabolism are less than those observed with first and second generation birth control pills (29, 33). This study confirms that this formula does not cause significant changes in fasting insulin or glucose levels.

This study has several limitations. First, we did not randomly assign women to a contraceptive method because the three types under study have different efficacies and randomization could have led to unintended pregnancies. Second, we did not have data on glucose or insulin levels during an oral glucose tolerance test, which would have given us more complete picture of the effect of these contraceptives on glucose metabolism. Finally, we studied only one formulation of OC, so our findings cannot be generalized to other types of birth control pills with different amounts of estrogen or other progestins. Together, these limitations could impact the overall generalizability of our findings, and selection bias cannot be ruled out.

It is important to remember that the reason women use DMPA and OC is to avoid pregnancy. When counseling women about contraception, the provider and the patient must weigh potential adverse effects of each method against the risk of unintended pregnancy. Overall, we conclude that women receiving DMPA may experience mild increases in fasting insulin and glucose levels which are probably not clinically meaningful and those receiving very low dose oral contraceptives with desogestrel will not experience any change in fasting levels. Thus, data from this longitudinal study are overall reassuring about the effects of these two contraceptives on insulin-glucose metabolism.