Self-reported periodontitis and fecundability in a population of pregnancy planners

Abstract

STUDY QUESTION

Is a history of periodontitis among women associated with reduced fecundability?

SUMMARY ANSWER

A history of periodontitis, as assessed by three different self-reported measures, may be associated with reduced fecundability.

WHAT IS KNOWN ALREADY

Periodontitis is a chronic inflammatory condition affecting the hard and soft tissues surrounding the teeth. Few studies have evaluated the association between periodontitis and time to pregnancy, and findings are mixed. It is hypothesized that periodontitis may adversely affect time to pregnancy.

STUDY DESIGN, SIZE, DURATION

We conducted a prospective cohort study of 2764 female pregnancy planners residing in North America (March 2015–June 2020).

PARTICIPANTS/MATERIALS, SETTING, METHODS

Eligible participants had been attempting pregnancy for six or fewer menstrual cycles at enrollment and were not using fertility treatment. Women answered questions about their oral health. Pregnancy was ascertained via bi-monthly follow-up questionnaires. We used proportional probabilities regression models to estimate fecundability ratios (FRs) and 95% confidence intervals (CIs) for three different measures indicative of a history of periodontitis: ever diagnosed with periodontitis (N = 265), ever received treatment for periodontitis (N = 299), and ever had an adult tooth become loose on its own (N = 83). We adjusted for potential confounders and precision variables. Women at risk of misclassification of periodontitis diagnosis due to pregnancy-related gingivitis were reclassified in a sensitivity analysis.

MAIN RESULTS AND THE ROLE OF CHANCE

All three indices of periodontitis may be associated with reduced fecundability. FRs were 0.89 (95% CI 0.75–1.06) comparing women with and without a previous periodontitis diagnosis, 0.79 (95% CI 0.67–0.94) comparing women with and without previous periodontitis treatment, and 0.71 (95% CI 0.44–1.16) comparing women with and without a tooth that became loose. After reclassification of pregnancy-related gingivitis in the sensitivity analysis, the FR for periodontitis diagnosis was 0.83 (95% CI 0.68–1.00). Weaker FRs were observed among parous women as compared with nulliparous women for periodontitis diagnosis and tooth becoming loose, but not for periodontitis treatment.

LIMITATIONS, REASONS FOR CAUTION

Though we used validated self-report measures of periodontitis, clinical confirmation is the gold standard. These questions may be functioning as markers of different levels of periodontitis severity, but we were unable to measure disease severity in this population. Finally, we cannot eliminate the possibility of unmeasured confounding.

WIDER IMPLICATIONS OF THE FINDINGS

This is the first preconception prospective cohort study to evaluate the association between self-reported periodontitis and fecundability. Our results indicate that periodontitis may be associated with lower fecundability.

STUDY FUNDING/COMPETING INTEREST(S)

This work was partially funded by R01HD086742/Eunice Kennedy Shriver National Institute of Child Health and Human Development and R21HD072326/Eunice Kennedy Shriver National Institute of Child Health and Human Development. PRESTO has received in-kind donations from Swiss Precision Diagnostics, Sandstone Diagnostics, FertilityFriend.com, and Kindara.com for primary data collection. L.A.W. is a fibroid consultant for AbbVie, Inc. J.C.B., S.W., J.Y., K.J.R., E.E.H., and B.H. have no conflicts of interest to disclose.

TRIAL REGISTRATION NUMBER

N/A.

Introduction

Periodontitis is a chronic inflammatory disease caused by a response to bacteria that creep below the gum line. Over time, the inflammation results in the deterioration of the hard (alveolar bone) and soft (gingiva/gum) tissues surrounding the teeth, and ultimately can lead to tooth loss. Treatment for periodontitis, including removal of bacterial plaque, diseased tissue, and calculus, can mitigate disease effects and prevent tooth loss. Nationally representative studies estimate that 46% of US adults have periodontal disease, with 8.9% classified as severe disease (Eke et al., 2015).

There is evidence of associations between periodontal disease and other health outcomes (Cullinan and Seymour, 2013), including adverse pregnancy events like preeclampsia, preterm birth, and low birthweight (Daalderop et al., 2018). Clinical trials involving treatment of periodontitis during pregnancy, primarily in the second trimester, have shown little or no evidence for improvement of birth outcomes (Iheozor-Ejiofor et al., 2017). Thus, the association between periodontitis and adverse pregnancy outcomes may be explained by a mechanism that begins before pregnancy, such as oral bacteria entering the blood stream through the gums and infecting uterine tissue (Madianos et al., 2013; Hajishengallis 2015) or chronic systemic inflammation caused by chronic gum infection (Hajishengallis, 2015). Either of these pathways, especially chronic systemic inflammation, has the potential to affect fertility. For instance, chronic inflammation has been implicated in endometriosis-related infertility (Halis and Arici, 2004). Alternatively, periodontitis could be a marker for as-yet unidentified factor(s) that are a common cause of both periodontitis and infertility, for example, a susceptibility to chronic inflammation or an immunological problem (Loos and Van Dyke, 2020).

A small number of studies have evaluated periodontitis and time to pregnancy (TTP) in women, including one study of Nigerian women (Nwhator et al., 2014), one study of Australian women (Hart et al., 2012), and one study of Finnish women (Paju et al., 2017). The results were mixed, but suggest an association between periodontitis and longer TTP. These studies also highlight the challenges in studying the reproductive health effects of periodontitis: specifically, effective confounding control and potential for misclassification of periodontitis if gum health is assessed during pregnancy. A recent review evaluating evidence for the association between periodontitis and female infertility reported that while periodontitis and infertility share many risk factors, including polycystic ovarian syndrome (PCOS), bacterial vaginosis, and obesity, evidence of a direct connection remains lacking (Machado et al., 2020).

We used data from a North American cohort of pregnancy planners to assess the association between self-reported indicators of preconception periodontal disease and TTP. Using detailed information collected on baseline and follow-up questionnaires, we were able to adjust for several covariates and to evaluate the potential for misclassification due to pregnancy-related gum inflammation.

Materials and methods

Study population

Pregnancy Study Online (PRESTO) is an ongoing internet-based prospective cohort study of pregnancy planners (detailed methodology described in Wise et al., 2015). Females were recruited via online advertising, posted flyers, and word of mouth. To be eligible, females must be aged 21–45 years, reside in the United States or Canada, and not be using contraception or fertility treatment. Upon enrollment, participants completed an online baseline questionnaire collecting information on demographics, behavior, reproductive and medical history and medication use. Ten days after enrollment, participants completed a web-based food frequency questionnaire designed and validated by the National Cancer Institute (Diet History Questionnaire II) (Subar et al., 2001). Every 8 weeks for 12 months or until reported conception, female participants completed online follow-up questionnaires that assessed behavioral and lifestyle factors that may have changed, as well as pregnancy status. Females who reported a conception completed additional questionnaires in early pregnancy, late pregnancy, and postpartum (6 months after delivery). Females were followed until reported conception, initiation of fertility treatment, cessation of pregnancy attempts, withdrawal, loss to follow-up, or 12 months without a censoring event, whichever came first. We restricted our analytic sample to those attempting pregnancy for six or fewer menstrual cycles at enrollment.

Questions related to oral health were added to the PRESTO questionnaires in April 2019. For females who had enrolled before April 2019 but were still participating, the dental health questions were added as supplemental questions to the next follow-up questionnaire (which included bi-monthly follow-up questionnaires, the early pregnancy questionnaire, late pregnancy questionnaire, and the postpartum questionnaire). For females who enrolled after April 2019, oral health questions were included on the baseline questionnaire only. Between April 2019 and June 2020, 1330 women completed the oral health questions on their baseline questionnaire. An additional 1434 females answered oral health questions on a follow-up questionnaire (range of enrollment dates: March 2015–April 2019), for a total study size of 2764.

Ethical approval

The study was approved by the Boston University Medical Campus Institutional Review Board and informed consent was obtained online from each participant.

Periodontal health exposures

We assessed the presence of periodontal disease via three clinically validated self-report measures.

1. “Has a dentist or dental hygienist ever told you that you had periodontal or gum disease?” (periodontitis diagnosis)

   (1a) If yes: Approximately how old were you when you were first told that you had periodontal or gum disease?

2. “Have you ever had treatment for gum disease such as scaling and root planing, sometimes called “deep cleaning?” (periodontitis treatment)

3. “Have you ever had any adult teeth become loose on their own, without resulting from an injury?” (tooth becoming loose)

Response options for each item were ‘Yes’, ‘No’, and, ‘I Don’t Know’. These questions were based on questionnaire items developed by the Centers for Disease Control and Prevention in collaboration with the American Academy of Periodontology to facilitate large-scale surveillance efforts for periodontitis (Eke et al., 2012). These questions have been validated against clinical measures of periodontitis, such as the probing depth of the gingival pocketing around the tooth and the level of detachment of the gingival tissue from around the tooth (Eke et al., 2013). They have also been validated against clinical case definitions of periodontitis in a broad range of populations, including women (LaMonte et al., 2014; Abbood et al., 2016; Heaton et al., 2017; Carra et al., 2018; Verhulst et al., 2019). We examined responses to each question independently, and considered those responding with ‘Yes’ to have the exposure of interest. Additionally, we created a variable with three exposure categories based on the items pertaining to diagnosis and treatment: (1) ‘Yes’ to both periodontitis diagnosis and treatment; (2) ‘Yes’ to one or the other; and (3) ‘No’ to both.

Outcome

The outcome of interest was TTP calculated using data from the baseline and follow-up questionnaires. At baseline, female participants reported the date of their last menstrual period (LMP), usual menstrual cycle length, and the number of menstrual periods they had since they began trying to conceive. At follow-up, women again reported LMP date and whether they had conceived since the previous questionnaire. Usual cycle length for women reporting irregular cycles was calculated using baseline LMP and consecutive LMPs reported on follow-up questionnaires. Total pregnancy attempt time was calculated based on the total discrete cycles at risk of pregnancy using the following formula: cycles of attempt at study entry + [(LMP date from most recent follow-up questionnaire−date of baseline questionnaire completion)/usual cycle length] + 1.

Covariates

Covariates were identified using a directed acyclic graph (DAG) based on published literature (Greenland et al., 1999; Moore et al., 2004; Salvi et al., 2008; Kavoussi et al., 2009; Hart et al., 2012; Nwhator et al., 2014; Paju et al., 2017) (Fig. 1). To assess the potential influence of each confounder, we used the change-in-estimate approach in which we evaluated the percent change between crude and adjusted effect estimates, adjusting for each covariate individually. Potential confounders included race/ethnicity (Hispanic/Latina, multi-racial or other race, Black non-Hispanic, Asian non-Hispanic, and White non-Hispanic), smoking (ever smokers, never smokers; participants who reported quitting ≥10 years before attempting conception were categorized as ‘never smokers’ because the smoking-associated risk of periodontitis is believed to decrease to that of non-smokers 9 years after cessation; Krall et al., 2006), history of diagnosed medical conditions (including bacterial vaginosis, type 2 diabetes, and endometriosis), age (<25, 25–29, 30–34, 35–39, and ≥40 years), weekly servings of sugar-sweetened soft drinks (0, 1, 2–6, and ≥7), marital status (married, unmarried), body mass index (BMI) (<18.5, 18.5–24, 25–29, and ≥30 kg/m²), markers of current socioeconomic status: female and male partner education (high school or less, some college, college, graduate school), annual household income (<$50 000, 50 000–99 000, 100 000–149 000, ≥150 000 USD), health insurance coverage (private from insurer, private from work, health insurance through government program, free clinic, out of pocket), and employment status (employed, unemployed), and markers of childhood socioeconomic status: participant’s maternal and paternal education level (high school or less, some college, college, graduate school). All potential confounders were included in the adjusted models.

Figure 1.

Directed acyclic graph with prespecified confounders. PCOS, polycystic ovarian syndrome; SES, socioeconomic status; U, unknown confounder. ^aRace/ethnicity, a social construct, was included as a proxy for a variety of potential exposures, including the impact of racism on an institutional, interpersonal, and internalized level (Jones, 2001).

Models were additionally adjusted for baseline variables associated with fecundability (i.e. precision variables), including use of prenatal or multi-vitamins, use of folic acid supplements, use of vitamin D supplements, frequency of intercourse (<1 time/month, 1 time/month, 2–3 times/month, 1 time/week, 2–3 times/week, 4–6 times/week, daily), doing something to improve conception chances (e.g. ovulation testing), and the Healthy Eating Index Score (continuous) (Guenther et al., 2014), a measure of diet quality derived from the Diet History Questionnaire II (Subar et al., 2001).

Missing data

We imputed missing values for covariates and pregnancy status using multiple imputation. We created five imputed datasets using the fully conditional specification method, analyzing each dataset separately, and combining coefficient and standard error estimates across the imputed datasets (Lee and Carlin, 2010). To reduce potential for selection bias from differential loss to follow-up, we assigned one cycle of follow-up to females with no follow-up data (N = 456, 17%) and then imputed their pregnancy status (yes vs. no) by multiple imputation. Ninety-one (3.3%) females reported ‘I don’t know’ with respect to periodontitis diagnosis, 93 (3.3%) reported ‘I don’t know’ with respect to periodontitis treatment, and 22 (1%) reported ‘I don’t know’ with respect to tooth becoming loose. ‘I don’t know’ values were also imputed using fully conditional specification multiple imputation. Among covariates with missing data, missingness ranged from 0.03% (intercourse frequency) to 6.17% (father’s education) except for Healthy Eating Index score, which was imputed for 26.7% of participants. These latter data were collected on a supplementary questionnaire with a lower response rate. For reference, we present the percentage missing for all covariates in Supplementary Table SI.

Analysis

We used proportional probabilities regression models to estimate fecundability ratios (FRs) and 95% confidence intervals (CI) for the association between each exposure variable (periodontitis diagnosis, periodontitis treatment, tooth becoming loose, and a three-tiered combination variable) and fecundability (Weinberg et al., 1989). A separate model was conducted for each exposure variable. The FR is the average per-cycle probability of conception in exposed relative to unexposed women. Because the model conditions on cycle at risk, it accounts for the natural decline in fertility over time. Thus, an FR <1 indicates a longer time to pregnancy and an FR > 1 indicates an accelerated time to pregnancy among those with self-reported periodontal disease indicators relative to those without. An FR of 1 indicates no association between the exposure and time to pregnancy. For the primary analysis, we calculated crude FRs, FRs adjusted only for age, and FRs adjusted for all potential confounders and additional precision variables (see Fig. 1). We stratified by parity to evaluate potential misclassification of pregnancy-related gingivitis as periodontitis. Additionally, we re-analyzed all fully-adjusted models excluding parity.

We also conducted two sensitivity analyses. First, we calculated the crude and adjusted FRs among those who were never smokers to assess possible residual confounding by smoking status. Second, we evaluated potential exposure misclassification among parous women with a periodontitis diagnosis. Hormonal changes during pregnancy can cause gum inflammation that is distinct from chronic periodontitis, but this distinction between a chronic condition and transient pregnancy gingivitis may not be appreciated by the patient or clearly communicated by the clinician during clinical assessments (Laine, 2002). In light of this, we compared the age at periodontitis diagnosis with the age at earliest live birth among diagnosed women. If the age of periodontitis diagnosis was younger than the age at their first live birth, women were considered to truly have periodontitis. If, however, the reported age of periodontitis diagnosis was the same or older than their first live birth, we only considered women as periodontitis cases if they also reported treatment. Periodontal treatment in the form of scaling and root planing is not commonly provided for pregnancy-related gingivitis (Silk et al., 2008; George et al., 2012). For this reason, those reporting diagnosis during or after the year of a pregnancy were considered exposed if they also reported receipt of periodontal treatment and unexposed if they did not. For this analysis, 38 participants were reclassified from exposed to unexposed for periodontitis diagnosis.

We also evaluated the possibility of an etiologically relevant exposure window, using the reclassified dataset created in the second sensitivity analysis, described above. For this analysis, we calculated the years between reported age of diagnosis of periodontitis and age at study enrollment. Those who reported an age of diagnosis younger than 11 years old (N = 4) were excluded. While it is possible for periodontitis diagnoses to occur prior to age 11, they are uncommon and likely due to other underlying systemic issues (American Academy of Periodontology – Research, Science, and Therapy Committee, 2008). We evaluated the association between years since diagnosis and fecundability, using both a continuous exposure variable and a binary variable (≤4 years/>4 years between diagnosis and enrollment), adjusting for all potential confounders and precision variables.

Following the guidance of the American Statistical Association, we interpreted our results based on the magnitude, precision, and possible biases in our effect estimates, as opposed to relying on statistical significance (Wasserstein and Lazar, 2016).

All statistical analyses were conducted using SAS version 9.4 (SAS Institute, Cary, NC).

Results

The final study population included 2764 females, who contributed a total of 1658 pregnancies over 9558 menstrual cycles of observation. Selected baseline characteristics are presented in Table I, stratified by exposure status. The majority had at least a college degree and an annual income over $50 000. The majority of women identified as White, non-Hispanic. The prevalence of periodontitis as measured via three self-report items was as follows: 9.6% for periodontitis diagnosis, 10.8% for periodontitis treatment, and 3.0% for a tooth becoming loose.

Table I.

Baseline characteristics by periodontitis exposures among 2764 PRESTO participants, 2015–2020.

	Periodontitis diagnosis		Periodontitis treatment		Tooth becoming loose
Characteristica	Yes (N = 265)	No (N = 2499)	Yes (N = 299)	No (N = 2465)	Yes (N = 83)	No (N = 2681)
	Mean (SD)
Age, years mean (SD)	30.9 (3.9)	29.8 (3.9)	30.8 (3.8)	29.8 (3.9)	30.8 (3.9)	29.9 (3.9)
	Age-standardized %
Self-reported oral health, %
Excellent, very good, or good	71.1	90.1	79.8	89.3	26.8	90.4
Fair or poor	28.9	9.9	20.2	10.7	73.2	9.7
Parity, births %
≥1	35.3	33.2	34.4	33.7	60.8	32.6
0	64.7	66.8	65.6	66.3	39.2	67.4
Ever smoking, %
Yes	20.8	14.9	21.8	14.8	46.2	14.6
Participant education, %
≤High school	3.2	4.5	5.7	4.3	17.9	4.0
Some college	22.2	19.2	21.8	19.2	49.8	18.5
College degree	35.1	37.1	39.8	36.6	16.0	37.6
Graduate school	39.5	39.2	32.7	39.8	14.9	39.8
Income, USD %
<$50,000	16.9	15.7	18.0	15.7	52.2	14.7
$50,000-99,000	38.9	36.4	42.9	35.9	32.5	36.7
$100,000-149,000	25.6	29.4	24.5	29.5	7.6	29.7
≥$150,000	18.6	18.5	14.6	19.0	6.2	18.9
Race/Ethnicity, %
Hispanic/Latina	6.2	5.8	10.4	5.2	8.3	5.7
Multi-racial or other race	5.2	3.6	4.1	3.8	3.1	3.7
Black, Non-Hispanic	2.6	2.1	3.3	2.0	4.5	2.1
Asian, Non-Hispanic	1.8	1.9	3.3	1.7	1.0	1.9
White, Non-Hispanic	84.2	86.6	78.8	87.4	81.6	86.6
BMI, kg/m2 %
<18.5	2.7	1.8	2.5	1.8	3.3	1.8
18.5-24	40.0	43.2	37.9	43.6	22.5	43.6
25-29	23.2	24.9	24.4	24.8	18.2	24.9
≥30	34.2	30.2	35.2	29.8	54.6	29.8
Endometriosis, %	5.5	2.9	3.3	3.0	4.6	3.0
Diabetes, %	2.1	1.0	2.8	0.9	5.9	0.9
Bacterial Vaginosis (ever diagnosed), %	20.9	14.2	18.5	14.4	20.0	14.6
Polycystic ovarian syndrome, %	7.5	8.4	7.9	8.5	16.3	8.1
Health insurance, %
Private from insurer	6.2	7.3	8.0	7.2	6.1	7.1
Private from work	81.7	77.9	79.7	77.9	42.8	79.3
Government program	22.3	26.3	28.8	25.8	43.7	25.4
Free clinic	5.7	5.7	9.7	5.4	11.8	5.5
Out of pocket	1.3	1.6	0.4	1.7	10.0	1.3
Currently employed	87.2	87.8	83.9	88.0	65.8	88.4
Married to partner	87.0	89.6	88.4	89.4	77.0	89.6
Sugar-sweetened soft drink consumption, servings/week, %
0	53.9	63.3	57.4	63.2	40.7	63.1
1	18.0	16.3	20.3	16.0	16.3	16.5
2-6	19.6	15.3	15.4	15.7	27.0	15.3
≥7	8.5	5.1	6.9	5.2	14.6	5.1
Male partner education, %
<College	41.7	44.9	44.1	44.6	61.6	44.0
College or greater	58.3	55.1	55.9	55.4	36.7	56.0
Mother’s education, %
<College	55.0	56.5	64.1	55.4	67.3	55.8
College or greater	42.9	40.8	34.3	41.8	24.2	41.6
Father’s education, %
<College	56.4	51.5	58.8	51.0	60.8	51.8
College or greater	38.5	42.5	36.8	42.9	28.4	42.3
Prenatal or multivitamin use, %	79.8	80.6	81.5	80.4	62.2	81.0
Folic acid supplementation, %	10.7	11.6	15.4	11.3	11.3	11.5
Vitamin D supplementation, %	13.3	17.5	15.5	17.4	3.5	17.5
Frequency of intercourse, times/week %
≤1	44.4	41.4	42.4	41.4	37.2	41.6
2-3	40.2	44.6	42.4	44.5	39.9	44.4
≥4	15.7	14.1	15.2	14.2	21.4	14.0
Doing anything to improve chances of conceiving, %	77.9	80.0	81.9	79.6	78.0	79.7

^aAll characteristics except age are standardized to the age distribution of the cohort at baseline.

Periodontitis diagnosis

Adjustment for all potential confounders and precision variables identified in the DAG resulted in an adjusted FR of 0.89 (95% CI 0.75–1.06) (Table II). When we reclassified participants who were potentially misclassified due to pregnancy gingivitis, the adjusted FR was 0.83 (95% CI 0.68–1.00) (Table II). Stratification by parity suggested some effect modification: fully-adjusted effect estimates were 0.98 (95% CI 0.74–1.30) among parous women and 0.83 (95% CI 0.66–1.06) among nulliparous women (Table III). The covariates with the greatest changes in estimate (suggesting the most influential confounders) were age and marital status (Supplementary Table SII).

Table II.

Association between periodontitis exposures and fecundability.

		N	Number of cycles	Number of pregnancies	Crude FR (95% CI)	FR adjusted ONLY for age (95% CI)	FR fully adjusteda (95% CI)
Periodontitis diagnosis or treatment	Both	157	665	84	0.77 (0.62-0.95)	0.79 (0.64-0.98)	0.80 (0.63-1.00)
	One	250	917	141	0.85 (0.72-1.01)	0.86 (0.72-1.02)	0.88 (0.74-1.05)
	Neither	2357	7976	1433	Ref	Ref	Ref
Periodontitis diagnosis or treatment after reclassificationb	Both	155	663	83	0.76 (0.61-0.94)	0.78 (0.63-0.97)	0.80 (0.63-1.00)
	One	216	797	114	0.79 (0.65-0.95)	0.79 (0.66-0.95)	0.81 (0.67-0.99)
	Neither	2393	8098	1461	Ref	Ref	Ref
Periodontitis diagnosis	Yes	265	1064	152	0.85 (0.72-1.00)	0.87 (0.73-1.02)	0.89 (0.75-1.06)
	No	2499	8494	1506	Ref	Ref	Ref
Periodontitis diagnosis after reclassificationb	Yes	227	940	123	0.78 (0.65-0.94)	0.80 (0.66-0.96)	0.83 (0.68-1.00)
	No	2537	8618	1535	Ref	Ref	Ref
Periodontitis treatment	Yes	299	1183	157	0.77 (0.66-0.90)	0.79 (0.68-0.92)	0.79 (0.67-0.94)
	No	2465	8375	1501	Ref	Ref	Ref
Tooth becoming loose	Yes	83	286	35	0.67 (0.46-0.99)	0.69 (0.46-1.01)	0.71 (0.44-1.16)
	No	2681	9272	1623	Ref	Ref	Ref

FR, fecundability ratio.

^aFully-adjusted models included race/ethnicity, history of bacterial vaginosis, diabetes, endometriosis, polycystic ovarian syndrome, age, education, annual income, BMI in kg/m², insurance coverage, weekly servings of sugar-sweetened soft drinks, marital status, and current employment, partner’s education level education, and maternal and paternal education level education, use of a prenatal or multi-vitamin at baseline, use of folic acid supplementation at baseline, use of vitamin D supplementation at baseline, frequency of intercourse at baseline, doing anything to improve conception chances at baseline, healthy eating index score, parity, smoking.

^bReclassifies 38 women reporting a previous diagnosis of periodontitis as unexposed because their diagnosis occurred at the same age of, or after, a pregnancy, and they did not report treatment for periodontitis.

Table III.

Association between periodontitis exposures and fecundability, stratified by parity.

		N	Number of cycles	Number of pregnancies	Crude FR (95% CI)	FR adjusted only for age (95% CI)	FR fully adjusteda (95% CI)
Parous (N = 924)
Periodontitis diagnosis or treatment	Both	53	216	32	0.78 (0.56-1.08)	0.78 (0.56-1.08)	0.81 (0.54-1.22)
	One	89	292	58	0.92 (0.70-1.21)	0.93 (0.70-1.22)	0.95 (0.69-1.30)
	Neither	782	2512	510	Ref	Ref	Ref
Periodontitis diagnosis	Yes	97	363	67	0.94 (0.73-1.21)	0.93 (0.74-1.16)	0.98 (0.74-1.30)
	No	827	2657	533	Ref	Ref	Ref
Periodontitis treatment	Yes	98	361	55	0.75 (0.58-0.97)	0.76 (0.58-0.98)	0.78 (0.57-1.07)
	No	826	2659	545	Ref	Ref	Ref
Tooth becoming loose	Yes	52	152	26	0.81 (0.52-1.25)	0.82 (0.53-1.27)	0.89 (0.50-1.58)
	No	872	2868	574	Ref	Ref	Ref
Nulliparous (N = 1840)
Periodontitis diagnosis or treatment	Both	104	449	52	0.76 (0.58-0.99)	0.80 (0.61-1.05)	0.79 (0.59-1.06)
	One	161	625	83	0.81 (0.66-1.00)	0.82 (0.67-1.02)	0.82 (0.66-1.03)
	Neither	1575	5464	923	Ref	Ref	Ref
Periodontitis diagnosis	Yes	168	701	85	0.79 (0.63-0.99)	0.82 (0.66-1.03)	0.83 (0.66-1.06)
	No	1672	5837	973	Ref	Ref	Ref
Periodontitis treatment	Yes	201	822	102	0.79 (0.65-0.95)	0.81 (0.67-0.99)	0.80 (0.63-1.01)
	No	1639	5716	956	Ref	Ref	Ref
Tooth becoming loose	Yes	31	134	9	0.39 (0.19-0.83)	0.40 (0.19-0.84)	0.51 (0.22-1.18)
	No	1809	6404	1049	Ref	Ref	Ref

FR, fecundability ratio.

^aFully-adjusted models included race/ethnicity, history of bacterial vaginosis, diabetes, endometriosis, polycystic ovarian syndrome, age, education, annual income, BMI in kg/m², insurance coverage, weekly servings of sugar-sweetened soft drinks, marital status, and current employment, partner’s education level education, and maternal and paternal education level education, use of a prenatal or multi-vitamin at baseline, use of folic acid supplementation at baseline, use of vitamin D supplementation at baseline, frequency of intercourse at baseline, doing anything to improve conception chances at baseline, healthy eating index score, smoking.

The average age at periodontitis diagnosis was 25.1 years (standard deviation 5.9), and the average number of years between diagnosis and age at baseline was 5.7 (standard deviation 5.3). Compared with participants who were diagnosed ≤4 years before baseline, those who were diagnosed >4 years before baseline had an FR of 0.78 (95% CI 0.43–1.40) after adjustment for all covariates (Table IV). An additional year between periodontitis diagnosis and baseline was associated with a fully-adjusted FR of 0.96 (95% CI 0.91–1.01).

Table IV.

Association between years between gum disease diagnosis and enrollment age and fecundability, among those with a gum disease diagnosis.

	Na	Number of cycles	Number of pregnancies	Crude FR (95% CI)	FR adjusted only for age (95% CI)	FR fully adjustedb (95%CI)
Continuous measure of years between gum disease diagnosis and age at enrollment	204	833	110	0.97 (0.94-1.01)	0.98 (0.94-1.01)	0.96 (0.91-1.01)
Categorical measure of years between gum disease diagnosis and age at enrollment
>4 years since diagnosis	101	417	52	0.83 (0.57-1.20)	0.87 (0.60-1.27)	0.82 (0.45-1.50)
≤4 years since diagnosis	103	416	58	Ref	Ref	Ref

^aAnalysis excluded 19 women who reported a periodontitis diagnosis but no age of diagnosis and 4 women who reported an age of diagnosis < 11 years old. Thirty-eight women who reported periodontitis diagnosis were reclassified as not having a periodontitis diagnosis due to the likelihood of misclassification due to pregnancy-related gingivitis.

^bFully-adjusted models included race/ethnicity, history of bacterial vaginosis, diabetes, endometriosis, polycystic ovarian syndrome, age, education, annual income, BMI in kg/m², insurance coverage, weekly servings of sugar-sweetened soft drinks, marital status, and current employment, partner’s education level education, and maternal and paternal education level education, use of a prenatal or multi-vitamin at baseline, use of folic acid supplementation at baseline, use of vitamin D supplementation at baseline, frequency of intercourse at baseline, doing anything to improve conception chances at baseline, healthy eating index score, parity.

Periodontitis treatment

The adjusted FR for periodontitis treatment (yes vs. no) was 0.79 (95% CI 0.67–0.94). After stratifying by parity, the fully-adjusted estimates were 0.78 (95% CI 0.57–1.07) among parous women and 0.80 (95% CI 0.63–1.01) among nulliparous women.

The change in estimate criterion revealed that the most influential covariates were age and type 2 diabetes (Supplementary Table SII).

Both periodontitis diagnosis and periodontitis treatment

The adjusted FR was 0.88 (95% CI 0.74–1.05) for either periodontitis diagnosis or periodontitis treatment but not both and 0.80 (95% CI 0.63–1.00) for both diagnosis and treatment, compared with neither (Table II).

Tooth becoming loose

The adjusted FR for tooth becoming loose (yes vs. no) was 0.71 (95% CI 0.44–1.16). There appeared to be some effect modification by parity after stratification: estimates were 0.89 (95% CI 0.50–1.58) among parous and 0.51 (95% CI 0.22–1.18) among nulliparous women, though the estimates were imprecise.

The change in estimate criterion revealed that the most influential covariates were BMI and type of health insurance (Supplementary Table SII).

All analyses conducted among never smokers were not appreciably different from the overall analyses (Table V). Similarly, repeating the analyses without adjusting for parity did not appreciably change the associations.

Table V.

Association between periodontitis exposures and fecundability, restricted to never smokers (N = 2,336).

		N	Number of cycles	Number of pregnancies	Crude FR (95% CI)	FR adjusted only for age (95% CI)	FR fully adjusteda (95% CI)
Periodontitis diagnosis	Yes	209	820	123	0.87 (0.73-1.03)	0.89 (0.74-1.06)	0.91 (0.76-1.09)
	No	2127	7336	1315	Ref	Ref	Ref
Periodontitis diagnosis after reclassificationb	Yes	183	751	101	0.78 (0.64-0.95)	0.80 (0.66-0.97)	0.83 (0.68-1.02)
	No	2153	7405	1337	Ref	Ref	Ref
Periodontitis treatment	Yes	237	916	128	0.79 (0.66-0.93)	0.81 (0.68-0.95)	0.83 (0.68-0.99)
	No	2099	7240	1310	Ref	Ref	Ref
Tooth becoming loose	Yes	46	176	22	0.67 (0.42-1.08)	0.70 (0.43-1.12)	0.72 (0.42-1.23)
	No	2290	7980	1416	Ref	Ref	Ref

FR, fecundability ratio.

^aFully-adjusted models included race/ethnicity, history of bacterial vaginosis, diabetes, endometriosis, polycystic ovarian syndrome, age, education, annual income, BMI in kg/m², insurance coverage, weekly servings of sugar-sweetened soft drinks, marital status, and current employment, partner’s education level education, and maternal and paternal education level education, use of a prenatal or multi-vitamin at baseline, use of folic acid supplementation at baseline, use of vitamin D supplementation at baseline, frequency of intercourse at baseline, doing anything to improve conception chances at baseline, healthy eating index score, parity.

^bReclassifies 38 women reporting a previous diagnosis of periodontitis as unexposed because their diagnosis occurred at the same age of, or after, a pregnancy, and they did not report treatment for periodontitis.

Discussion

We observed possible associations between three different self-reported measures of periodontitis and reduced fecundability. The strength of the point estimates did not change appreciably after adjustment for pre-specified potential confounders and precision variables, as well as after restriction of the study population to never smokers. There was evidence of effect measure modification by parity, with the observed FRs being stronger among nulliparous women for periodontitis diagnosis and tooth becoming loose, but not periodontitis treatment. The observed FRs were stronger among those reporting both a diagnosis of and treatment for periodontitis compared with those reporting only periodontitis diagnosis. Additionally, the observed FRs increased in strength from periodontitis diagnosis to periodontitis treatment to tooth becoming loose. Taken together, these results could suggest a possible dose–response relationship, as the strength of the point estimates increased with markers of more severe disease.

Previous studies evaluating the association between periodontitis and TTP have reported mixed results, though they generally suggest a positive association between periodontitis (or risk factors for periodontitis) and TTP greater than 12 months. In a retrospective Australian cohort of 1956 women, Hart et al. (2012) reported an association between periodontitis and infertility among non-White women, the majority of whom were classified as ‘Asian’, but found no such association among White women. In a retrospective Nigerian cohort of 128 women, Nwhator et al. (2014) found an association between periodontitis and infertility. Both studies performed clinical periodontitis assessments during the second trimester of pregnancy and ascertained TTP retrospectively. Evaluation of periodontitis during pregnancy may have resulted in misclassification of pregnancy-related gingivitis as periodontitis. The authors also evaluated infertility as a binary variable (≥12 vs. <12 months to conceive). Dichotomization of fertility may not detect more subtle decrements in fertility that can be captured using fecundability, which estimates the probability of conception on a per-cycle basis. In a prospective study of 256 Finnish women, Paju et al. (2017) performed preconception clinical assessments of periodontal status and found an association between saliva concentrations of Porphyromonas gingivalis, the primary bacteria associated with development of periodontitis, and infertility, but no association between preconception clinically-confirmed periodontitis and longer TTP.

We observed that effect estimates for periodontitis diagnosis and tooth becoming loose were attenuated among parous women compared with nulliparous women. Stronger FRs observed among nulliparous women for periodontitis diagnosis and fecundability may reflect reduced misclassification of pregnancy-related gingivitis as periodontitis. Increases in progesterone during pregnancy can cause the gingival tissue to swell and become sensitive such that bleeding may occur while flossing or brushing. The condition, referred to as pregnancy gingivitis, typically occurs between the second and the eighth month of pregnancy and resolves once progesterone levels return to normal (Laine, 2002). Thus, women who reported a periodontitis diagnosis and previous live birth may erroneously report having had periodontal disease. Indeed, when we reclassified women susceptible to such misclassification, the association strengthened. These findings highlight the salience of pregnancy history when using self-reported measures of periodontitis. Reports of a tooth becoming loose were the least prevalent of the three exposure measures, but generated the strongest FR overall. It is important to note that in its most severe form, periodontitis results in the destruction of both the hard (e.g. alveolar bone) and soft tissues that support the tooth, leading to mobile or loose teeth. Thus, the stronger FRs for reports of a loose tooth could reflect greater periodontitis severity, with reports of diagnosis only reflecting the mildest form of the condition. However, these results could simply reflect differences in the specificity of the measures, with ‘tooth loose’ having the highest specificity of the three exposure variables assessed. Finally, the observed effect modification by parity for a tooth becoming loose could be due to chance.

Our analysis of time since periodontitis diagnosis raises the possibility that the potential association between periodontitis and fecundability may be influenced by duration of disease. However, we cannot infer causality from the results of this study. Generally, greater time since diagnosis was associated with longer TTP. It is possible that the systemic impact of periodontitis is cumulative and worsens over time, which would explain why intervention trials to treat periodontitis in the second trimester do not improve adverse birth outcomes (Iheozor-Ejiofor et al., 2017). It is also possible, however, that earlier diagnosis is an indicator of more severe disease at the time of enrollment or perhaps reflects a genetic or biologic susceptibility to chronic inflammation.

Our study had several important limitations. First, we used self-reported measures to assess a history of periodontal disease. These questions have been validated in a variety of ways and in several different populations (LaMonte et al., 2014; Abbood et al., 2016; Heaton et al., 2017; Carra et al., 2018; Verhulst et al., 2019) and our selected questions generally have good predictive qualities for clinically confirmed periodontitis when included in models that account for population demographics. Specifically, the sensitivity and specificity of models that included our questions were generally around 80% (Carra et al., 2018; Verhulst et al., 2019). However, clinical assessment remains the gold standard. Additionally, as with all observational studies that rely on self-reported data, the present study may be susceptible to residual confounding. However, previous evaluations of the validity of selected self-reported variables in the PRESTO study have demonstrated high validity (Wise et al., 2015). Second, our periodontitis measures also limited our ability to glean more details about the duration and progression of periodontitis in our population. Like many chronic diseases, periodontitis can be managed, particularly in the earliest stages, but can also progress over time (Tonetti et al., 2018; Loos and Van Dyke, 2020). As periodontitis increases in severity, interventions can range from management with oral hygiene practices to more intensive treatment, such as scaling and root planing. Particularly given our finding that fecundability decreased as time since periodontitis diagnosis increased, future research may benefit from collecting more detailed information about periodontitis infection and management/treatment history. Third, because of the rollout of the oral health questionnaire, some women in our sample answered the periodontitis history questions after baseline. We compared the prevalence of each exposure between those who answered dental health questions at baseline versus after baseline and found that the prevalence was equivalent for periodontitis treatment and tooth becoming loose, though there was a slightly higher prevalence of periodontitis diagnosis among those who filled out the dental questions after baseline. However, after we reclassified women at risk of misclassification due to pregnancy gingivitis in sensitivity analyses, this discrepancy was eliminated. Fourth, our study does not provide any mechanistic insights and cannot determine whether the observed FRs reflect a causal relationship. Periodontitis may be a marker for underlying susceptibility to chronic inflammation, which has been found to be a common cause of periodontitis and fertility issues (Loos and Van Dyke, 2020). The possibility of an unknown common cause between periodontitis and other chronic health conditions has been postulated previously in the context of other chronic diseases (Meurman et al., 2004). Finally, we cannot comment on the potential role of alcohol consumption. The evidence base for the relationship between alcohol consumption and periodontal disease is generally weak. Observed associations with periodontal disease are predominantly limited to individuals reporting at least daily consumption (Tezal et al., 2001, 2004; Pitiphat et al., 2003; Hach et al., 2015). The evidence for an association between alcohol consumption and fecundability is mixed, especially for light to moderate consumption (Fan et al., 2017). Data on self-reported alcohol use in the 4 weeks prior to enrollment of our sample revealed that the prevalence of at least daily consumption is 10%, with heavier consumption much lower (3%). These prevalences are too low to introduce meaningful confounding, and so alcohol consumption was not considered in our analyses.

While these findings indicate that periodontitis may be associated with reduced fecundability, we would caution against interpretations of causality given the stated limitations of self-reported data. However, given the high prevalence of periodontitis in the general population and the fact that it can often be mitigated with treatment, periodontitis warrants careful evaluation as a potential modifiable risk factor for reduced fecundability.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

Authors’ roles

The parent study was designed by L.A.W., K.J.R., and E.E.H. B.H. designed and implemented the questions on oral health. Data collection for the present analysis was overseen by L.A.W. and B.H. Statistical analysis was performed by J.C.B. with assistance from S.W. and J.Y., with oversight from B.H., L.A.W., and K.J.R. All co-authors contributed to the interpretation of the data. J.C.B. drafted the manuscript and all co-authors reviewed for intellectual content and approved the final version.

Funding

This work was partially funded by R01HD086742/Eunice Kennedy Shriver National Institute of Child Health and Human Development and R21HD072326/Eunice Kennedy Shriver National Institute of Child Health and Human Development.

Conflict of interest

PRESTO has received in-kind donations from Swiss Precision Diagnostics, Sandstone Diagnostics, FertilityFriend.com, and Kindara.com for primary data collection. L.A.W. is a fibroid consultant for AbbVie, Inc. J.C.B., S.W., J.Y., K.J.R., E.E.H., and B.H. have no conflicts of interest to disclose.