Abstract
Bacterial vaginosis (BV) is a common condition of unknown etiology and has been linked to adverse reproductive and obstetric health outcomes. Prior dietary research on BV has focused on specific macro- and micronutrients, but not dietary indices. We assessed the relationship between BV and selected dietary indicators among a cohort of 1735 nonpregnant women ages 15–44 y from Birmingham, Alabama. Annual intake was assessed with the Block98 FFQ, and the glycemic index, glycemic load (GL), and Healthy Eating Index were calculated by the Block Dietary Data System. The Naturally Nutrient Rich (NNR) score was also calculated. Vaginal flora was evaluated using Nugent Gram-stain criteria. Crude OR and adjusted OR were determined by multinomial and logistic regression in cross-sectional and prospective analyses, respectively. Participants were predominantly African American (85.5%) aged 25.3 ± 6.8 y (mean ± SD). Per 10-unit increase, GL was positively (adjusted OR = 1.01, 95% CI = 1.00–1.03) and NNR was negatively (adjusted OR = 0.93, 95% CI = 0.88–0.99) associated with BV compared to normal vaginal flora. In prospective analyses, only GL was associated with BV progression (adjusted OR = 1.03, 95% CI = 1.00–1.05) and persistence (adjusted OR = 1.02, 95% CI = 1.01–1.04) after adjustment. Both GL and NNR were associated with greater BV prevalence and GL was associated with an increase in BV persistence and acquisition. These results suggest that diet composition may contribute to vaginal flora imbalances and be important for elucidating the etiology of BV.
Introduction
BV8 is the most frequent cause of vaginal complaints among reproductive-aged women and is characterized by an imbalance in the vaginal flora, with the replacement of Lactobacillus species by anaerobic bacteria and a corresponding increase in vaginal pH (>4.5) (1). In the United States, the prevalence of BV varies by ethnicity, with African American women having over 2 times higher prevalence of BV (51%) compared to white women (23%) (2). In addition, BV has considerable adverse reproductive and obstetric health consequences for women and their infants, including low birth weight, preterm birth, pelvic inflammatory disease, and HIV and other sexually transmitted infections (3). Despite the high prevalence of BV and adverse sequelae, its etiology remains poorly understood.
Although BV is related to sexual activity, its development may also depend on other nonsexual risk factors that make women more susceptible to vaginal flora imbalances. A growing body of evidence has emerged on the association between diet and nutritional status with BV (4–7), but the mechanisms remain unclear (8). Several studies have found associations between BV and poor micronutrient status, including vitamins A, C, E, and D and β-carotene (5, 7) as well as low dietary intakes of folate, calcium, and vitamin E (4). LSVF previously assessed the association between macronutrient intake and BV in 1520 predominately African American women 15–44 y of age and found an increased risk of BV among women with higher energy and total fat consumption (4). However, diets high in total energy and fat are often indicative of overall poorer diet quality; therefore, studying the relationship between individual nutrients and BV may be misleading due to the high correlations among nutrients within diets (9, 10).
In recent years, the field of nutrition has shifted toward examining dietary indices as opposed to single nutrients, because it has become recognized that nutrients are not consumed in isolation, that individuals consuming one health-promoting nutrient also tend to consume many others, and that the specific source of nutrients may be of importance (11, 12). In addition, dietary indices address the issue of multicollinearity of nutrients and may provide a better understanding of the relationship between diet and BV. A variety of dietary indices has been proposed, yet to date no one indicator has emerged as the gold standard. Indices that have shown utility in understanding the relationship of diet and disease outcomes include measures assessing the degree of conformance to dietary guidelines (13), measures of nutrient density, conceptualized as the concentration of micronutrients consumed per unit energy (14, 15), and measures based on differing physiological response to carbohydrate-containing foods (16, 17). Therefore, the objective of this paper was to examine the association between BV and several dietary indices among participants in the LSVF.
Participants and Methods
Study population and design.
Participants were enrolled in the LSVF between August 1999 and February 2002. Details of this study have been described previously (4, 18). Briefly, the study population consisted of nonpregnant women aged 15–44 y presenting for routine care at health clinics in Birmingham, Alabama. Women were ineligible to participate if they were immunocompromised or postmenopausal or had major gynecologic surgery, were receiving antibiotics or immunosuppressive drugs, planned to move within the next 12 mo, or were nonfluent in English. The study was approved by the institutional review boards of the Jefferson County Department Health, the University of Alabama at Birmingham, and the Eunice Kennedy Shriver National Institute of Child Health and Human Development.
Participants were assessed at baseline and followed quarterly for 1 y. At each visit, a pelvic exam and detailed in-person interview were conducted. Vaginal flora was evaluated blinded to clinical data using Nugent Gram-stain criteria and defined as normal (Nugent score: 0–3), intermediate (Nugent score: 4–6), or BV (Nugent score: 7–10) (19). At the second study visit, the 110-item Block98 FFQ was administered (20) by trained interviewers who did not know the Gram-stain results. For administrative reasons, the Block FFQ was discontinued after the first 2005 participants completed their second visit. Demographic and behavioral characteristics did not differ between participants who completed the Block FFQ and those who did not (4).
Dietary assessment.
The Block98 FFQ assessed annual intake and dietary data were analyzed by the Block Dietary Data System which uses the USDA Nutrient Database for Standard Reference for macro- and micronutrients (20). The dietary data were reanalyzed in 2010 for additional dietary variables, including mean daily GI, mean daily GL, and the HEI (13). Dietary data with serious errors (i.e. portions consumed were too few or many, identical serving sizes, etc.), as determined by the Block Dietary Data System, were excluded from further analyses (n = 265).
The GI ranks carbohydrate-containing foods according to their effect on blood glucose concentration and is expressed as the percent increase in blood sugar relative to an equivalent quantity of glucose (i.e. GI of glucose = 100) (21). GI values for this analysis were obtained from the Clinical Nutrition Research Center at the University of North Carolina (22) or based on published values (21).
Both GI and GL are measures of carbohydrate quality, with the latter also representing quantity of carbohydrate consumed and applied frequently in epidemiologic studies (16). To calculate mean daily GI and GL, carbohydrate, fiber, and GI values were assigned to all food items in the questionnaire. A mean daily GL was calculated by multiplying the GI value of each food by its available carbohydrate (minus fiber) and summed across daily intake of all foods. A mean daily GI was calculated by dividing the mean daily GL by the total available daily carbohydrate and multiplied by 100.
The HEI is a measure of diet quality ranging from 0 to 100 that assesses conformance to federal dietary guidelines, with higher scores indicating greater conformance (13). The HEI was calculated using established methods of assigning scores based on adequacy of consumption of MyPyramid food groups as well as meeting guidelines for sodium, saturated fat, and cholesterol (23). An HEI score >80 implies a “good” diet, an HEI score between 51 and 80 implies a diet that “needs improvement,” and an HEI score <51 implies a “poor” diet (13).
An additional indicator of nutrient density of a diet was assessed using the NNR score (14). The NNR score was calculated from the FFQ as the unweighted arithmetic mean of percent daily values for 15 nutrients (protein, fiber, MUFA, vitamins A, C, D, E, and B-12, thiamin, riboflavin, folate, calcium, iron, zinc, and potassium) per 8370 kJ consumed.
Statistical methods.
The association between dietary indices and BV was assessed in cross-sectional and prospective analyses. Demographic and behavioral characteristics expressed as mean ± SD or n (%) across quartiles of dietary indices and correlations of dietary indices with total energy and fat intake were examined. Differences across quartiles were assessed using Kruskal-Wallis tests for continuous variables and chi-square tests for categorical variables. The association between dietary indices and the log odds of BV compared to normal was explored graphically using locally weighted least squares scatterplot smoothing techniques (24). Because intermediate vaginal flora represents a transitory state between normal and BV (25), cross-sectional analyses evaluated the association between different vaginal flora states (normal, intermediate, and BV) and dietary indices using multinomial logistic regression. Based on least squares scatterplot-smoothed curves, dietary indices were modeled as a continuous and a linear spline term for HEI at 70, which corresponds approximately to the 90th percentile of HEI in this population (there was no evidence of nonlinearity for other dietary indices). Models were adjusted for confounding factors chosen a priori, which included age, education, race, cigarette and alcohol use, number of sex partners since the previous visit, douching frequency, hormonal contraceptive use, and BMI. Interaction terms between the number of sex partners and dietary indices were assessed and were not significant. The high correlation of GL with total energy intake and total fat intake precluded adjustment for these variables in the model.
In the prospective analyses, BV status was classified into 4 groups based on BV status at the current visit (visit when FFQ was administered) and the subsequent follow-up visit only and defined as: 1) BV negative at both visits (non-BV maintenance); 2) BV negative at the current visit and positive at the next (BV progression); 3) BV positive at the current visit and negative at the next visit (BV resolution); and 4) BV positive at both visits (BV persistence). Logistic regression models were used to assess the association of dietary indices with BV progression, resolution, and persistence compared to maintenance of non-BV and BV resolution compared to persistence. Models were adjusted for the same confounding factors used in the cross-sectional analysis. A 2-sided P ≤ 0.05 was considered significant in all analyses. Data were analyzed using Stata/SE 10.1 for Windows (Stata).
Results
The sample consisted of predominantly African American (85.5%) women (n = 1735) aged 25.3 ± 6.8 y, education of 12.0 ± 1.8 y, and BMI of 29.7 ± 8.4 kg/m2. The majority of women had not used alcohol (47.3%) or smoked since the prior visit (73.4%). The GI, GL, HEI, and NNR were 55.1 ± 3.5, 191.7 ± 88.0, 54.2 ± 11.1, and 88.9 ± 19.1, respectively. The correlation between GI and GL was 0.12 and between HEI and NNR was 0.46. The correlation of GI, GL, HEI, and NNR with total energy intake was 0.06, 0.93, −0.18, and −0.25, and with total fat was 0.14, 0.78, −0.38, and −0.30, respectively. All correlations were significant (P < 0.001).
Quartiles of GI were associated with age, total energy intake, GL, HEI, NNR score, BMI, and alcohol use (Table 1). Quartiles of GL were associated with age, education, race, total energy intake, GI, NNR score, douching frequency, and vaginal flora (Table 1). Quartiles of HEI were associated with age, education, race, total energy intake, GI, and NNR score (Table 2). Quartiles of NNR were associated with age, education, race, total energy intake, GI, GL, HEI, smoking, number of sex partners, and vaginal flora (Table 2).
TABLE 1.
Characteristics of participants by quartiles of dietary glycemic index (GI) and glycemic load (GL)12
| Quartiles of dietary GI |
Quartiles of dietary GL |
|||||||
| 1 (low) | 2 | 3 | 4 (high) | 1 (low) | 2 | 3 | 4 (high) | |
| N | 435 | 433 | 434 | 433 | 434 | 434 | 434 | 433 |
| Age, y | 25.6 ± 7.04 | 24.2 ± 5.98 | 25.5 ± 6.77 | 25.9 ± 7.07* | 26.3 ± 6.98 | 25.6 ± 7.06 | 25.3 ± 6.69 | 24.1 ± 6.07* |
| Education, n (%) | ||||||||
| <12 y | 126 (29.0) | 145 (33.3) | 147 (33.7) | 154 (36.0) | 110 (24.7) | 122 (28.5) | 163 (37.8) | 177 (41.0)* |
| 12 y | 188 (43.3) | 162 (37.7) | 166 (38.3) | 174 (40.4) | 174 (40.6) | 177 (41.1) | 174 (39.9) | 165 (38.2) |
| >12 y | 120 (27.7) | 125 (28.9) | 120 (17.9) | 103 (23.7) | 149 (34.6) | 132 (30.4) | 97 (22.4) | 90 (20.8) |
| Race, n (%) | ||||||||
| Black | 370 (85.7) | 370 (85.7) | 368 (84.8) | 374 (86.3) | 352 (81.1) | 359 (82.9) | 387 (89.2) | 384 (88.9)* |
| White/other | 65 (14.7) | 62 (14.4) | 66 (15.2) | 58 (13.7) | 82 (18.9) | 74 (17.1) | 47 (10.8) | 48 (11.1) |
| Energy intake, kJ/d | 11,800 ± 5980 | 12,900 ± 5940 | 12,500 ± 5100 | 12,700 ± 5400* | 6800 ± 1800 | 10,100 ± 1780 | 13,400 ± 2340 | 19,600 ± 4910* |
| GI | 50.6 ± 1.99 | 54.1 ± 0.70 | 56.4 ± 0.66 | 59.5 ± 1.66* | 54.3 ± 3.90 | 55.3 ± 3.47 | 55.3 ± 3.38 | 55.7 ± 3.15* |
| GL | 177 ± 88.8 | 194 ± 90.3 | 195 ± 79.7 | 201 ± 91.4* | 96.5 ± 21.8 | 152 ± 13.8 | 205 ± 17.6 | 313 ± 67.0* |
| HEI | 57.2 ± 11.8 | 54.4 ± 10.8 | 53.1 ± 10.9 | 52.1 ± 10.4* | 54.8 ± 11.6 | 54.5 ± 11.8 | 53.0 ± 11.4 | 54.4 ± 9.61 |
| NNR score | 97.8 ± 23.1 | 88.5 ± 16.3 | 85.6 ± 18.1 | 83.8 ± 14.8* | 96.4 ± 21.8 | 88.6 ± 17.7 | 87.5 ± 18.7 | 83.2 ± 15.3* |
| BMI, kg/m2 | 31.0 ± 8.42 | 29.6 ± 8.41 | 29.1 ± 8.62 | 29.1 ± 7.94* | 29.9 ± 7.79 | 30.0 ± 8.04 | 29.9 ± 8.72 | 29.0 ± 8.89* |
| Cigarette use,3n (%) | ||||||||
| No | 313 (71.9) | 308 (71.7) | 327 (76.6) | 316 (73.3) | 312 (72.6) | 317 (73.2) | 327 (76.4) | 308 (71.5) |
| Yes | 121 (28.1) | 123 (28.3) | 101 (23.4) | 115 (26.7) | 122 (27.4) | 115 (26.9) | 100 (23.7) | 123 (28.5) |
| Alcohol use,3n (%) | ||||||||
| Never | 177 (40.8) | 203 (47.1) | 205 (47.9) | 231 (53.7)* | 183 (41.9) | 212 (49.1) | 211 (49.4) | 210 (48.7) |
| ≤Once/mo | 100 (23.3) | 99 (23.0) | 113 (26.4) | 112 (26.0) | 126 (29.3) | 100 (23.2) | 97 (22.7) | 101 (23.4) |
| Once/wk – several times/mo | 89 (20.3) | 84 (19.3) | 71 (16.6) | 49 (12.1) | 84 (19.4) | 70 (16.2) | 69 (16.4) | 70 (16.2) |
| Several times/wk – every day | 68 (15.7) | 45 (10.7) | 39 (9.1) | 39 (8.9) | 41 (9.5) | 50 (11.6) | 50 (11.5) | 50 (11.6) |
| Vaginal flora, n (%) | ||||||||
| Normal | 153 (35.1) | 144 (33.4) | 142 (33.3) | 159 (37.6) | 180 (41.9) | 142 (33.0) | 147 (34.1) | 129 (30.3)* |
| Intermediate | 91 (21.0) | 94 (21.9) | 90 (21.1) | 104 (24.3) | 89 (20.1) | 93 (21.6) | 98 (23.0) | 99 (23.2) |
| BV | 189 (43.8) | 193 (45.0) | 195 (45.7) | 165 (38.1) | 163 (38.0) | 195 (45.4) | 186 (42.9) | 198 (46.5) |
| Number of sex partners,3n (%) | ||||||||
| 0 | 93 (21.7) | 83 (19.0) | 89 (20.6) | 100 (23.4) | 90 (20.3) | 93 (21.8) | 83 (19.7) | 99 (23.0) |
| 1 | 308 (70.7) | 310 (72.1) | 315 (73.8) | 311 (71.7) | 322 (74.4) | 312 (72.0) | 307 (71.7) | 303 (70.3) |
| 2 or more | 33 (7.6) | 38 (8.8) | 24 (5.6) | 20 (4.9) | 22 (5.3) | 27 (6.3) | 37 (8.7) | 29 (6.7) |
| Douching frequency,3n (%) | ||||||||
| Never | 243 (55.7) | 255 (59.5) | 236 (55.4) | 233 (53.8) | 251 (58.1) | 249 (57.4) | 235 (54.9) | 232 (54.0)* |
| <Once/wk | 139 (32.3) | 139 (32.1) | 133 (30.8) | 150 (35.0) | 147 (33.4) | 142 (33.1) | 139 (32.9) | 133 (30.9) |
| Once/wk | 37 (8.6) | 23 (5.4) | 43 (10.1) | 25 (5.8) | 32 (7.4) | 24 (5.6) | 34 (8.0) | 38 (8.8) |
| Several times/wk – every day | 14 (3.5) | 13 (3.0) | 16 (3.7) | 23 (5.3) | 4 (1.2) | 17 (3.9) | 18 (4.2) | 27 (6.3) |
| Hormonal contraception,3n (%) | ||||||||
| No | 207 (48.1) | 199 (46.1) | 199 (45.7) | 206 (47.7) | 197 (45.5) | 203 (47.1) | 195 (45.0) | 216 (50.0) |
| Yes | 227 (51.8) | 233 (53.9) | 234 (54.3) | 226 (52.3) | 236 (54.5) | 230 (52.9) | 238 (55.0) | 216 (50.0) |
Values are or means ± SD. Asterisks indicate P ≤ 0.05 based on Kruskal-Wallis tests for continuous variables and chi-square tests for categorical variables.
HEI, Healthy Eating Index; NNR, Naturally Nutrient Rich.
Since the last visit.
TABLE 2.
Characteristics of participants by quartiles of the Healthy Eating Index (HEI) and Naturally Nutrient Rich (NNR) score12
| Quartiles of HEI |
Quartiles of NNR |
|||||||
| 1 (low) | 2 | 3 | 4 (high) | 1 (low) | 2 | 3 | 4 (high) | |
| n | 479 | 448 | 427 | 381 | 434 | 434 | 434 | 433 |
| Age, y | 25.0 ± 6.70 | 24.5 ± 6.34 | 25.5 ± 6.80 | 26.4 ± 7.09* | 25.2 ± 6.40 | 24.6 ± 6.63 | 25.4 ± 7.00 | 26.0 ± 6.92* |
| Education, n (%) | ||||||||
| <12 y | 175 (36.0) | 168 (37.7) | 133 (31.7) | 96 (25.3)* | 165 (37.8) | 153 (35.0) | 145 (33.5) | 109 (25.7)* |
| 12 y | 201 (42.5) | 170 (38.1) | 176 (41.1) | 143 (37.6) | 181 (42.2) | 179 (41.7) | 154 (35.3) | 176 (40.5) |
| >12 y | 102 (21.5) | 108 (24.2) | 117 (27.2) | 141 (37.1) | 85 (20.0) | 102 (23.3) | 134 (31.2) | 147 (33.8) |
| Race, n (%) | ||||||||
| Black | 432 (90.2) | 400 (89.5) | 372 (87.1) | 278 (73.2)* | 364 (84.1) | 381 (87.6) | 383 (88.5) | 354 (81.9)* |
| White/other | 47 (9.8) | 47 (10.5) | 55 (12.9) | 102 (26.8) | 69 (15.9) | 53 (12.4) | 51 (11.5) | 78 (18.1) |
| Energy intake, kJ/d | 13,600 ± 5380 | 12,600 ± 5850 | 12,300 ± 5860 | 11,000 ± 4990 | 13,700 ± 5960 | 13,600 ± 5750 | 12,300 ± 5390 | 10,300 ± 4600 |
| GI | 55.9 ± 3.42 | 55.3 ± 3.47 | 55.2 ± 3.49 | 54.0 ± 3.45* | 56.1 ± 3.24 | 55.7 ± 3.19 | 55.2 ± 3.25 | 53.5 ± 4.80* |
| GL | 186 ± 79.1 | 194 ± 90.4 | 200 ± 93.9 | 187 ± 88.5 | 213 ± 93.2 | 205 ± 88.2 | 190 ± 84.8 | 159 ± 75.1* |
| HEI | 40.5 ± 5.41 | 51.7 ± 2.26 | 59.1 ± 2.09 | 69.0 ± 5.13* | 48.1 ± 9.50 | 51.2 ± 9.83 | 56.1 ± 9.85 | 61.4 ± 10.6* |
| NNR score | 80.0 ± 13.9 | 85.6 ± 17.7 | 89.8 ± 17.6 | 103 ± 19.8* | 67.9 ± 6.26 | 81.1 ± 2.92 | 92.3 ± 3.75 | 115 ± 15.1* |
| BMI, kg/m2 | 29.1 ± 8.04 | 29.5 ± 8.29 | 30.0 ± 8.31 | 30.3 ± 8.94 | 29.4 ± 8.36 | 29.6 ± 8.15 | 30.0 ± 8.28 | 29.8 ± 8.72 |
| Cigarette use,3n (%) | ||||||||
| No | 344 (72.9) | 327 (73.0) | 312 (73.5) | 281 (74.3) | 295 (68.4) | 309 (72.1) | 324 (75.4) | 336 (77.7)* |
| Yes | 132 (27.1) | 121 (27.0) | 110 (26.5) | 97 (25.7) | 138 (31.6) | 121 (27.9) | 107 (24.6) | 94 (22.3) |
| Alcohol use,3n (%) | ||||||||
| Never | 220 (46.2) | 213 (47.5) | 206 (48.8) | 177 (46.8) | 193 (44.3) | 200 (46.3) | 220 (51.3) | 203 (47.2) |
| ≤Once/mo | 117 (24.8) | 107 (23.9) | 103 (24.4) | 97 (25.7) | 108 (25.2) | 103 (24.0) | 96 (22.3) | 117 (27.2) |
| Once/wk – several times/mo | 85 (17.9) | 83 (18.5) | 66 (15.9) | 59 (15.6) | 74 (17.1) | 79 (18.4) | 72 (16.7) | 68 (16.1) |
| Several times/wk – every day | 54 (11.3) | 45 (10.0) | 47 (10.9) | 45 (11.9) | 58 (13.4) | 48 (11.4) | 43 (9.7) | 42 (9.5) |
| Vaginal flora, n (%) | ||||||||
| Normal | 150 (31.7) | 157 (35.6) | 143 (34.0) | 148 (39.0) | 134 (31.0) | 143 (33.7) | 141 (32.9) | 180 (41.7)* |
| Intermediate | 97 (19.9) | 104 (23.6) | 96 (23.0) | 82 (21.6) | 103 (23.8) | 99 (23.0) | 98 (22.7) | 79 (18.5) |
| BV | 230 (48.4) | 180 (40.8) | 182 (43.0) | 150 (39.5) | 195 (45.1) | 185 (43.3) | 189 (44.4) | 173 (39.8) |
| Number of sex partners,3n (%) | ||||||||
| 0 | 99 (20.4) | 96 (21.4) | 83 (20.1) | 87 (23.0) | 78 (17.8) | 97 (22.3) | 93 (21.6) | 97 (23.0)* |
| 1 | 336 (70.8) | 319 (71.2) | 317 (74.6) | 272 (72.0) | 316 (73.0) | 301 (70.0) | 311 (72.4) | 316 (73.0) |
| 2 or more | 41 (8.8) | 33 (7.4) | 22 (5.2) | 19 (5.0) | 39 (9.2) | 32 (7.7) | 27 (6.0) | 17 (4.0) |
| Douching frequency,3n (%) | ||||||||
| Never | 252 (53.4) | 253 (56.5) | 249 (58.5) | 213 (56.4) | 236 (54.5) | 219 (51.3) | 257 (59.8) | 255 (58.8) |
| <Once/wk | 162 (33.8) | 145 (32.4) | 127 (30.6) | 127 (33.6) | 144 (33.0) | 155 (35.9) | 124 (28.8) | 138 (32.6) |
| Once/wk | 43 (9.1) | 25 (5.6) | 33 (7.8) | 27 (7.1) | 32 (7.4) | 38 (8.9) | 31 (7.2) | 27 (6.3) |
| Several times/wk – every day | 17 (3.8) | 25 (5.6) | 13 (3.1) | 11 (2.9) | 21 (5.1) | 17 (4.0) | 18 (4.2) | 10 (2.3) |
| Hormonal contraception,3n (%) | ||||||||
| No | 246 (51.4) | 197 (44.1) | 190 (44.9) | 178 (46.8) | 214 (49.2) | 208 (47.7) | 187 (43.7) | 202 (47.1) |
| Yes | 233 (48.6) | 250 (55.9) | 235 (55.1) | 202 (53.2) | 219 (50.8) | 226 (52.3) | 246 (56.4) | 229 (52.9) |
Values are or means ± SD. Asterisks indicate P ≤ 0.05 based on Kruskal-Wallis tests for continuous variables and chi-square tests for categorical variables.
GI, glycemic index; GL, glycemic load.
Since the last visit.
Cross-sectional analyses assessed the association between dietary indices as continuous variables (per 10-unit increase) and vaginal flora (Table 3). In crude analyses, GL was positively and NNR inversely associated with BV compared to normal vaginal flora. After adjustment, the associations for GL and NNR remained significant (P = 0.037 and 0.025, respectively). BV was not associated with HEI values <70; however, HEI values >70 were associated with a reduction in BV that was significant in crude analyses and borderline significant after adjustment (P = 0.08). Similar associations for dietary indices were found comparing intermediate to normal flora.
TABLE 3.
Crude OR (cOR) and adjusted OR (aOR) for the relationship between dietary indices and vaginal flora assessed by Nugent Gram-stain criteria1–3
| Intermediate flora (Nugent: 4–6) |
BV (Nugent: 7–10) |
|||||||
| Dietary index | cOR | 95% CI | aOR4 | 95% CI | cOR | 95% CI | aOR4 | 95% CI |
| Daily GI | 1.28 | 0.89–1.85 | 1.26 | 0.86–1.87 | 0.92 | 0.68–1.25 | 0.99 | 0.71–1.38 |
| Daily GL | 1.02 | 1.00–1.03 | 1.01 | 1.00–1.03 | 1.02 | 1.01–1.03 | 1.01 | 1.00–1.03 |
| HEI5 | ||||||||
| <70 | 1.01 | 0.89–1.15 | 1.04 | 0.91–1.20 | 0.94 | 0.84–1.05 | 0.99 | 0.88–1.11 |
| >706 | 0.44 | 0.19–1.02 | 0.49 | 0.21–1.15 | 0.38 | 0.19–0.78 | 0.48 | 0.23–1.04 |
| NNR score | 0.89 | 0.83–0.95 | 0.90 | 0.84–0.97 | 0.93 | 0.88–0.98 | 0.93 | 0.88–0.99 |
Change in odds of intermediate flora or BV compared to normal flora (Nugent: 0–3) per 10-unit increase in corresponding dietary index value.
= 1719 in crude analyses and n = 1644 in adjusted analyses.
aOR, adjusted OR; BV, bacterial vaginosis; cOR, crude OR; GI, glycemic index; GL, glycemic load; HEI, Healthy Eating Index; NNR, Naturally Nutrient Rich.
Models adjusted for age, race, education, cigarette smoking, alcohol use, BMI, douching frequency, hormonal contraceptive use, and number of sex partners.
Spline term with node at 70; OR correspond to the slopes < and > HEI = 70.
for the difference between slopes < and > HEI = 70 are 0.13 and 0.10 for crude and adjusted intermediate flora and 0.10 and 0.08 for crude and adjusted BV, respectively.
In prospective analyses, only GL was associated with BV progression and persistence (Table 4). The NNR score was significantly associated with BV progression in crude analyses but was borderline significant after adjustment (P = 0.11). At HEI scores ≥70, HEI was significantly associated with reduced BV persistence in crude analyses, but not after adjustment for confounding factors. All significant estimates showed similar directions of associations to cross-sectional analyses. There was no association between dietary indices and women who resolved compared to those who persisted with or without BV (data not shown).
TABLE 4.
Crude OR (cOR) and adjusted OR (aOR) for the relationship between dietary indices and bacterial vaginosis (BV) progression and persistence12
| BV progression3 |
BV persistence4 |
|||||||
| Dietary index | cOR | 95% CI | aOR5 | 95% CI | cOR | 95% CI | aOR5 | 95% CI |
| Daily GI | 0.98 | 0.60–1.61 | 1.03 | 0.61–1.75 | 0.82 | 0.58–1.16 | 0.99 | 0.67–1.46 |
| Daily GL | 1.03 | 1.01–1.05 | 1.03 | 1.00–1.05 | 1.02 | 1.01–1.04 | 1.02 | 1.01–1.04 |
| HEI6 | ||||||||
| <70 | 0.88 | 0.74–1.04 | 0.93 | 0.77–1.13 | 0.91 | 0.80–1.04 | 0.94 | 0.82–1.08 |
| >70 | 0.58 | 0.16–2.13 | 0.84 | 0.24–2.95 | 0.34 | 0.12–0.97 | 0.51 | 0.17–1.50 |
| NNR score | 0.90 | 0.82–0.99 | 0.92 | 0.83–1.02 | 0.95 | 0.89–1.01 | 0.94 | 0.87–1.01 |
Change in odds of BV progression or persistence compared to non-BV maintenance per 10-unit increase in corresponding dietary index value.
GI, glycemic index; GL, glycemic load; HEI, Healthy Eating Index; NNR, Naturally Nutrient Rich.
= 753 in crude analyses and n = 738 in adjusted analyses.
= 976 in crude analyses and n = 962 in adjusted analyses.
Models adjusted for age, race, education, cigarette smoking, alcohol use, BMI, douching frequency, hormonal contraceptive use, and number of sex partners.
Spline term with node at 70; OR correspond to the slopes < and > HEI = 70.
Discussion
In this study population of predominantly African American women, dietary indices of carbohydrate quality (GI and GL) and overall diet quality (HEI and NNR) were associated with each other and with other demographic factors, including age, education, and race. Mean scores for HEI were below and for GI and GL were above means found in other U.S.-based sample populations (63.9, 50.0, and 100.2, respectively) (23, 26), indicating a poorer diet among women in our study. A more healthful diet, as indicated by GL, NNR, and HEI indicating >70% conformance to dietary guidelines, was associated with a lower odds of BV compared to normal vaginal flora. However, only variation in GL remained significantly associated with BV in both cross-sectional and prospective analyses after adjustment for other covariates.
Our findings of a lower likelihood of BV among those consuming more healthful diets are consistent with other studies that report an association between nutritional status and BV (4–7). Unlike those studies, which were limited to assessing individual macro- or micronutrients, our study evaluated the effect of overall diet and carbohydrate quality on BV prevalence, progression, and persistence. Dietary indices have utility in epidemiologic studies and in demonstrating lowered risk of mortality and morbidity (11, 16). With the exception of vitamin D, which can be obtained from nondietary sources (i.e. UV-B exposure), intakes of nutrients are often correlated. Thus, dietary indices may be a more useful method of evaluation, because they address multicollinearity of nutrients and provide a useful target for nutritional interventions that capture the complexity of the human diet.
These results may be important for elucidating the etiology of BV. In particular, GL, but not GI, was consistently associated with BV prevalence, progression, and persistence. Both the source and amount of carbohydrate can influence an individual’s glycemic response to a food or meal (27). Chronic exposure to postprandial rises in glucose may have adverse health effects and have been linked to diabetes, coronary heart disease, and cancer (28–35). Continual exposures to postprandial hyperglycemia may trigger oxidative damage through a reduction in plasma antioxidant defenses and increased inflammation due to free radical production (36, 37). It is also known that women with diabetes, especially those with poor glycemic control, are more likely to be affected by genital tract infections (38, 39). We could not distinguish the independent effects of GL and total energy or fat intake on BV given the high correlation between these variables in this study. Nevertheless, the biologic mechanism may be similar, given that even large amounts of low-GI foods may result in high spikes in postprandial glucose (40). Thus, it is plausible that chronic exposure to high-GL, energy-dense diets may affect host response to bacterial colonization and, in particular, the pathogenesis of BV through oxidative stress and impaired immune function.
Alternatively, diet may influence the microbial population of mucosal surfaces in the gastrointestinal and reproductive tracts. Studies conducted in mice suggest dietary factors, in particular high-fat diets, alter the composition of the gut microbiome (41). In turn, bacterial colonization of the gut may function as a reservoir for vaginal microbiota. A study by Antonio et al. (42) found concordance between Lactobacillus species in the rectum and vagina, suggesting that the rectum may serve as a potential source for vaginal colonization. The authors speculate that adherence properties of specific bacteria in the rectum may determine what species adhere to vaginal epithelium.
There are several strengths of this study. Studies have validated the original Block FFQ in several populations, including ones similar to this study population (43, 44), and the updated Block98 FFQ used in this study (45, 46) but have not been specifically validated for the indices used in this study. Our outcome variable, the Nugent score, is currently the gold standard for assessment of vaginal flora in research settings (47). As previously mentioned, dietary indices have the advantage of accounting for multicollinearity and complexity of dietary nutrients. Finally, the association of BV with different dietary indicators was evaluated both cross-sectionally and prospectively after adjustment for several demographic and lifestyle factors among a large population of women with consistent results across different levels of vaginal flora disruption (i.e. intermediate and BV) and over time (BV progression and persistence).
Our ability to generalize these findings to other populations is limited given the geographic location and the majority of women in this sample were African American. In addition, there was little variability in participants’ diets compared to other U.S. populations (23, 26). Unmeasured confounding may also be present if women who consume a particular diet also exhibit other health-related behaviors that increase their risk of BV, such as riskier sex, personal hygiene, or physical activity. To account for some of these factors, we adjusted for number of sex partners and douching frequency and found no change in inference. We controlled for BMI, but physical activity was not measured in this study and may also confound the association between BV and GL. We are not aware of any studies that have assessed the relationship between physical activity and BV and think this is an important area for future study.
Finally, it is generally recognized that associations between dietary indices and disease may be sensitive to slight differences in how those indices are calculated (11). The merits of the GI concept have been debated, without consensus, since its introduction, particularly owing to technical limitations in its calculation (48). A criticism of the GI is that the values can be imprecise, because they are significantly altered by factors including processing, ripeness, and cooking methods and that it does not account for the effect of carbohydrate quantity on the glycemic response. GL addresses this latter limitation by quantifying the overall glycemic effect of the diet but is still limited by imprecision from its calculation based on the GI. Output from the Block98 FFQ did not enable us to calculate an updated HEI (HEI-2005) in our study. In addition, the most currently proposed nutrient density measure is the Nutrient Rich Foods index (49); however, there are numerous nutrient density measures with no clear gold standard. Calculation of the Nutrient Rich Foods index required information on added sugars, which was not provided in the Block98 FFQ analysis. Therefore, we calculated another measure of nutrient density, the NNR, to assess this concept in our study (14). Despite these limitations, the consistency of associations across several dietary indices and vaginal flora measures demonstrates the robustness of our findings.
Our results suggest that BV is associated with dietary indices. Because this study included primarily African American women in the South, epidemiologic studies should be conducted to explore whether this relationship is consistent in other populations. Future studies investigating dietary and biochemical indicators of nutritional status, physical activity, and the role of diet in changes to gut and vaginal microbiome may further elucidate biologic mechanisms for these findings.
Acknowledgments
W.A., M.K., Y.N., and J.S. designed research; W.A., Y.N., and J.S. conducted research; M.E.T. analyzed data; M.E.T., M.K., A.R., T.N., Y.N., W.A., and J.S. wrote the paper; and M.E.T. had primary responsibility for final content. All authors read and approved the final manuscript.
Footnotes
Supported by the Intramural Research Program of the NIH, Eunice Kennedy Shriver National Institute of Child Health and Human Development and contract N01-HD-8-3293.
Abbreviations used: BV, bacterial vaginosis; GI, glycemic index; GL, glycemic load; HEI, Healthy Eating Index; LSVF, Longitudinal Study of Vaginal Flora; NNR, Naturally Nutrient Rich.
Literature Cited
- 1.Pirotta M, Fethers KA, Bradshaw CS. Bacterial vaginosis: more questions than answers. Aust Fam Physician. 2009;38:394–7 [PubMed] [Google Scholar]
- 2.Koumans EH, Sternberg M, Bruce C, McQuillan G, Kendrick J, Sutton M, Markowitz LE. The prevalence of bacterial vaginosis in the United States, 2001–2004; associations with symptoms, sexual behaviors, and reproductive health. Sex Transm Dis. 2007;34:864–9 [DOI] [PubMed] [Google Scholar]
- 3.Koumans EH, Kendrick JS. Preventing adverse sequelae of bacterial vaginosis: a public health program and research agenda. Sex Transm Dis. 2001;28:292–7 [DOI] [PubMed] [Google Scholar]
- 4.Neggers YH, Nansel TR, Andrews WW, Schwebke JR, Yu KF, Goldenberg RL, Klebanoff MA. Dietary intake of selected nutrients affects bacterial vaginosis in women. J Nutr. 2007;137:2128–33 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Tohill BC, Heilig CM, Klein RS, Rompalo A, Cu-Uvin S, Piwoz EG, Jamieson DJ, Duerr A. Nutritional biomarkers associated with gynecological conditions among US women with or at risk of HIV infection. Am J Clin Nutr. 2007;85:1327–34 [DOI] [PubMed] [Google Scholar]
- 6.Verstraelen H, Delanghe J, Roelens K, Blot S, Claeys G, Temmerman M. Subclinical iron deficiency is a strong predictor of bacterial vaginosis in early pregnancy. BMC Infect Dis. 2005;5:55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Bodnar LM, Krohn MA, Simhan HN. Maternal vitamin D deficiency is associated with bacterial vaginosis in the first trimester of pregnancy. J Nutr. 2009;139:1157–61 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Ahluwalia N, Grandjean H. Nutrition, an under-recognized factor in bacterial vaginosis. J Nutr. 2007;137:1997–8 [DOI] [PubMed] [Google Scholar]
- 9.Subar AF, Ziegler RG, Patterson BH, Ursin G, Graubard B. US dietary patterns associated with fat intake: the 1987 National Health Interview Survey. Am J Public Health. 1994;84:359–66 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Kant AK. Consumption of energy-dense, nutrient-poor foods by adult Americans: nutritional and health implications. The third National Health and Nutrition Examination Survey, 1988–1994. Am J Clin Nutr. 2000;72:929–36 [DOI] [PubMed] [Google Scholar]
- 11.Wirt A, Collins CE. Diet quality: what is it and does it matter? Public Health Nutr. 2009;12:2473–92 [DOI] [PubMed] [Google Scholar]
- 12.Kant AK. Indexes of overall diet quality: a review. J Am Diet Assoc. 1996;96:785–91 [DOI] [PubMed] [Google Scholar]
- 13.Basiotis PP, Carlson A, Gerrior SA, Juan WY, Lino M. The Healthy Eating Index: 1999–2000. USDA, Center for Nutrition Policy and Promotion; 2002 [Google Scholar]
- 14.Drewnowski A, Fulgoni V III. Nutrient profiling of foods: creating a nutrient-rich food index. Nutr Rev. 2008;66:23–39 [DOI] [PubMed] [Google Scholar]
- 15.Drewnowski A. Concept of a nutritious food: toward a nutrient density score. Am J Clin Nutr. 2005;82:721–32 [DOI] [PubMed] [Google Scholar]
- 16.Jenkins DJ, Kendall CW, Augustin LS, Franceschi S, Hamidi M, Marchie A, Jenkins AL, Axelsen M. Glycemic index: overview of implications in health and disease. Am J Clin Nutr. 2002;76:S266–73 [DOI] [PubMed] [Google Scholar]
- 17.Ludwig DS. The glycemic index: physiological mechanisms relating to obesity, diabetes, and cardiovascular disease. JAMA. 2002;287:2414–23 [DOI] [PubMed] [Google Scholar]
- 18.Klebanoff MA, Schwebke JR, Zhang J, Nansel TR, Yu KF, Andrews WW. Vulvovaginal symptoms in women with bacterial vaginosis. Obstet Gynecol. 2004;104:267–72 [DOI] [PubMed] [Google Scholar]
- 19.Nugent RP, Krohn MA, Hillier SL. Reliability of diagnosing bacterial vaginosis is improved by a standardized method of gram stain interpretation. J Clin Microbiol. 1991;29:297–301 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Questionnaires and Screeners-Assessment and Analysis Services-NutritionQuest [Internet]. NutritionQuest; 2009 [cited 2011 Apr 8]. Available from: www.nutritionquest.com/assessment/list-of-questionnaires-and-screeners
- 21.Foster-Powell K, Holt SH, Brand-Miller JC. International table of glycemic index and glycemic load values: 2002. Am J Clin Nutr. 2002;76:5–56 [DOI] [PubMed] [Google Scholar]
- 22.UNC School of Public Health-Diet Physical Activity and Body Composition [Internet] The University of North Carolina at Chapel Hill, Gillings School of Global Public Health, Nutrition Obesity Research Center; [updated 2011 March 14; cited 2011 Apr 8]. Available from: www.sph.unc.edu/cnrc/core_b.html
- 23.Kennedy ET, Ohls J, Carlson S, Fleming K. The Healthy Eating Index: design and applications. J Am Diet Assoc. 1995;95:1103–8 [DOI] [PubMed] [Google Scholar]
- 24.Cleveland WS, Devlin SJ. Locally weighted regression: an approach to regression-analysis by local fitting. J Am Stat Assoc. 1988;83:596–610 [Google Scholar]
- 25.Thoma ME, Gray RH, Kiwanuka N, Aluma S, Wang MC, Sewankambo N, Wawer MJ. The short-term variability of bacterial vaginosis diagnosed by Nugent Gram stain criteria among sexually active women in Rakai, Uganda. Sex Transm Dis. 2011;38:111–6 [DOI] [PubMed] [Google Scholar]
- 26.Hu Y, Block G, Sternfeld B, Sowers M. Dietary glycemic load, glycemic index, and associated factors in a multiethnic cohort of midlife women. J Am Coll Nutr. 2009;28:636–47 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Sheard NF, Clark NG, Brand-Miller JC, Franz MJ, Pi-Sunyer FX, Mayer-Davis E, Kulkarni K, Geil P. Dietary carbohydrate (amount and type) in the prevention and management of diabetes: a statement by the American Diabetes Association. Diabetes Care. 2004;27:2266–71 [DOI] [PubMed] [Google Scholar]
- 28.Salmeron J, Ascherio A, Rimm EB, Colditz GA, Spiegelman D, Jenkins DJ, Stampfer MJ, Wing AL, Willett WC. Dietary fiber, glycemic load, and risk of NIDDM in men. Diabetes Care. 1997;20:545–50 [DOI] [PubMed] [Google Scholar]
- 29.Salmeron J, Manson JE, Stampfer MJ, Colditz GA, Wing AL, Willett WC. Dietary fiber, glycemic load, and risk of non-insulin-dependent diabetes mellitus in women. JAMA. 1997;277:472–7 [DOI] [PubMed] [Google Scholar]
- 30.Liu S, Willett WC, Stampfer MJ, Hu FB, Franz M, Sampson L, Hennekens CH, Manson JE. A prospective study of dietary glycemic load, carbohydrate intake, and risk of coronary heart disease in US women. Am J Clin Nutr. 2000;71:1455–61 [DOI] [PubMed] [Google Scholar]
- 31.Franceschi S, Dal ML, Augustin L, Negri E, Parpinel M, Boyle P, Jenkins DJ, La VC. Dietary glycemic load and colorectal cancer risk. Ann Oncol. 2001;12:173–8 [DOI] [PubMed] [Google Scholar]
- 32.Augustin LS, Galeone C, Dal ML, Pelucchi C, Ramazzotti V, Jenkins DJ, Montella M, Talamini R, Negri E, Franceschi S, et al. Glycemic index, glycemic load and risk of prostate cancer. Int J Cancer. 2004;112:446–50 [DOI] [PubMed] [Google Scholar]
- 33.Augustin LS, Gallus S, Bosetti C, Levi F, Negri E, Franceschi S, Dal ML, Jenkins DJ, Kendall CW, La Vecchia C. Glycemic index and glycemic load in endometrial cancer. Int J Cancer. 2003;105:404–7 [DOI] [PubMed] [Google Scholar]
- 34.Augustin LS, Polesel J, Bosetti C, Kendall CW, La VC, Parpinel M, Conti E, Montella M, Franceschi S, Jenkins DJ, et al. Dietary glycemic index, glycemic load and ovarian cancer risk: a case-control study in Italy. Ann Oncol. 2003;14:78–84 [DOI] [PubMed] [Google Scholar]
- 35.Augustin LS, Dal ML, La VC, Parpinel M, Negri E, Vaccarella S, Kendall CW, Jenkins DJ, Francesch S. Dietary glycemic index and glycemic load, and breast cancer risk: a case-control study. Ann Oncol. 2001;12:1533–8 [DOI] [PubMed] [Google Scholar]
- 36.Hu Y, Block G, Norkus EP, Morrow JD, Dietrich M, Hudes M. Relations of glycemic index and glycemic load with plasma oxidative stress markers. Am J Clin Nutr. 2006;84:70–6 [DOI] [PubMed] [Google Scholar]
- 37.Ceriello A, Bortolotti N, Crescentini A, Motz E, Lizzio S, Russo A, Ezsol Z, Tonutti L, Taboga C. Antioxidant defences are reduced during the oral glucose tolerance test in normal and non-insulin-dependent diabetic subjects. Eur J Clin Invest. 1998;28:329–33 [DOI] [PubMed] [Google Scholar]
- 38.Donders GG. Lower genital tract infections in diabetic women. Curr Infect Dis Rep. 2002;4:536–9 [DOI] [PubMed] [Google Scholar]
- 39.Rayfield EJ, Ault MJ, Keusch GT, Brothers MJ, Nechemias C, Smith H. Infection and diabetes: the case for glucose control. Am J Med. 1982;72:439–50 [DOI] [PubMed] [Google Scholar]
- 40.O'Keefe JH, Gheewala NM, O'Keefe JO. Dietary strategies for improving post-prandial glucose, lipids, inflammation, and cardiovascular health. J Am Coll Cardiol. 2008;51:249–55 [DOI] [PubMed] [Google Scholar]
- 41.Tilg H. Obesity, metabolic syndrome, and microbiota: multiple interactions. J Clin Gastroenterol. 2010;44 Suppl 1:S16–8 [DOI] [PubMed] [Google Scholar]
- 42.Antonio MA, Rabe LK, Hillier SL. Colonization of the rectum by Lactobacillus species and decreased risk of bacterial vaginosis. J Infect Dis. 2005;192:394–8 [DOI] [PubMed] [Google Scholar]
- 43.Coates RJ, Eley JW, Block G, Gunter EW, Sowell AL, Grossman C, Greenberg RS. An evaluation of a food frequency questionnaire for assessing dietary intake of specific carotenoids and vitamin E among low-income black women. Am J Epidemiol. 1991;134:658–71 [DOI] [PubMed] [Google Scholar]
- 44.Neggers YH, Bindon JR, Dressler WW. The relationship between zinc and copper status and lipid levels in African-Americans. Biol Trace Elem Res. 2001;79:1–13 [DOI] [PubMed] [Google Scholar]
- 45.Boucher B, Cotterchio M, Kreiger N, Nadalin V, Block T, Block G. Validity and reliability of the Block98 food-frequency questionnaire in a sample of Canadian women. Public Health Nutr. 2006;9:84–93 [DOI] [PubMed] [Google Scholar]
- 46.Johnson BA, Herring AH, Ibrahim JG, Siega-Riz AM. Structured measurement error in nutritional epidemiology: applications in the Pregnancy, Infection, and Nutrition (PIN) Study. J Am Stat Assoc. 2007;102:856–66 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Koumans EH, Markowitz LE, Hogan V. Indications for therapy and treatment recommendations for bacterial vaginosis in nonpregnant and pregnant women: a synthesis of data. Clin Infect Dis. 2002;35:S152–72 [DOI] [PubMed] [Google Scholar]
- 48.Ludwig DS, Eckel RH. The glycemic index at 20 y. Am J Clin Nutr. 2002;76:S264–5 [DOI] [PubMed] [Google Scholar]
- 49.Drewnowski A. Defining nutrient density: development and validation of the nutrient rich foods index. J Am Coll Nutr. 2009;28:S421–6 [DOI] [PubMed] [Google Scholar]