全文

Objective

The purpose of this study was to determine the use of contraceptive methods, which was defined by effectiveness, length of coverage, and their association with short interpregnancy intervals, when controlling for provider type and client demographics.

 

Study Design

We identified a cohort of 117,644 women from the 2008 California Birth Statistical Master file with second or higher order birth and at least 1 Medicaid (Family Planning, Access, Care, and Treatment [Family PACT] program or Medi-Cal) claim within 18 months after index birth. We explored the effect of contraceptive method provision on the odds of having an optimal interpregnancy interval and controlled for covariates.

 

Results

The average length of contraceptive coverage was 3.81 months (SD = 4.84). Most women received user-dependent hormonal contraceptives as their most effective contraceptive method (55%; n = 65,103 women) and one-third (33%; n = 39,090 women) had no contraceptive claim. Women who used long-acting reversible contraceptive methods had 3.89 times the odds and women who used user-dependent hormonal methods had 1.89 times the odds of achieving an optimal birth interval compared with women who used barrier methods only; women with no method had 0.66 times the odds. When user-dependent methods are considered, the odds of having an optimal birth interval increased for each additional month of contraceptive coverage by 8% (odds ratio, 1.08; 95% confidence interval, 1.08–1.09). Women who were seen by Family PACT or by both Family PACT and Medi-Cal providers had significantly higher odds of optimal birth intervals compared with women who were served by Medi-Cal only.

 

Conclusion

To achieve optimal birth spacing and ultimately to improve birth outcomes, attention should be given to contraceptive counseling and access to contraceptive methods in the postpartum period.

 

Key words

contraceptive coverage; contraceptive effectiveness; interpregnancy interval; Medicaid; postpartum contraception

 

 

In the United States, one-third of all repeat pregnancies are conceived within 18 months of the previous birth.1 These short interpregnancy intervals are associated with adverse maternal and child health outcomes, such as increased risk of preterm birth and infants with low birthweight.2, 3, 4 and 5 To address this public health problem, the US Department of Health and Human Services chose as one of its Healthy People 2020 objectives to reduce the proportion of pregnancies that were conceived within 18 months of a previous birth by 10% in 2020.1

 

Effective contraceptive method use after birth has the potential to achieve optimal interpregnancy intervals.6 and 7 However, methods that require regular refills such as oral contraceptives or contraceptive ring provide, on average, less coverage over a 12-month period,8 especially if women receive only a limited contraceptive supply per visit.9 Similarly, the effectiveness of barrier methods is influenced by the client's consistent and accurate use.

 

Minority and low-income women are more likely to have short birth intervals as a result of unintended pregnancies than are white or middle-class women.10, 11 and 12 Therefore, assessment of the access to and provision of contraceptive methods through publicly funded services for low-income women may help to guide interventions to reduce short interpregnancy intervals. Women who have a birth that is reimbursed by California's Medicaid program (Medi-Cal) usually are eligible to receive healthcare services that include contraceptive services, from either Medi-Cal or its Medicaid family planning expansion, the Family Planning, Access, Care, and Treatment (Family PACT) program. The Family PACT program provided reproductive health services to >1.8 million low-income, uninsured women and men, including adolescents, in 2012.13

 

Any Medi-Cal provider can enroll in the Family PACT program and get reimbursed on a fee-for-service basis. At enrollment, providers agree to adhere to program standards (such as making all Food and Drug Administration–approved contraceptive methods available to clients) and to provide comprehensive family planning counseling services.14 Family PACT program providers receive professional education and clinical support through clinical practice alerts, webinars, and skills-based training. Provider performance is monitored regularly through external evaluations, and the delivery of high quality of care has been documented.15 and 16

 

In a previously published analysis of the 2008 California Birth Statistical Master File (BSMF), we found that the provision of contraceptives within the first 90 days after delivery was associated significantly with optimal interpregnancy intervals of at least 18 months.17 In the current analysis, we evaluated for the effect of postpartum contraceptive method effectiveness and length of postpartum contraceptive coverage on short interpregnancy intervals, when the data were controlled for Medi-Cal provider type and client demographics.

 

Materials and Methods

The data analysis was approved by the University of California, San Francisco, Committee on Human Research and the California Committee for the Protection of Human Subjects. We identified a total of 331,132 women who had second or higher order births among women from California's 2008 BSMF. The birth immediately before the 2008 birth is referred to as the “index birth.” Women who had multiple births, births that occurred before January 1, 2002, or index births that occurred outside California were excluded. Other exclusions were data inconsistencies, such as missing index birth dates, births with an interval of <30 days, or missing or improbable maternal age (<12 years old).

 

Of the remaining 230,850 women, we calculated the birth-to-conception interval between the date of the index birth and the conception date of the 2008 birth. Conception date was defined as the date of the last menses as recorded in the BSMF plus 9 days. To identify women who received contraceptive services from publicly funded programs, we applied a probabilistic linking method to match BSMF birth mother data with enrollment records for women with Medi-Cal or Family PACT program claims. The linking algorithm decided whether a pair of records from 2 disparate data files belongs to the same entity (person). 17, 18 and 19

 

We found 117,644 women who had at least 1 Family PACT program or Medi-Cal claim within 18 months after the index birth, which means that they were at least temporarily below the income threshold for publicly funded family planning programs. The remaining 113,206 women either received no services at all or services from a commercial health plan and were not included in the analysis.

 

Variables

This study explored whether contraceptive method provision within 18 months of the index birth and contraceptive coverage were associated with increased odds of an optimal interpregnancy interval. Contraceptive methods were categorized into tiers based on effectiveness20:

 

•

Tier 1: Long-acting reversible contraception (LARC)–implant and intrauterine contraceptives.

•

Tier 2: User-dependent hormonal–oral contraceptives, injection, patch, and ring.

•

Tier 3: Barrier method and supplies–condoms, diaphragm, and spermicides.

•

No method.

In cases of women who received multiple types of contraception that fell into different tiers, the most effective method that had been used after the index birth was defined as the “maximum tier” and used for descriptive and regression analyses.

 

Contraceptive coverage was defined to estimate the amount of contraceptive supply that a woman received. We calculated coverage for user-dependent hormonal methods and barrier methods using an algorithm that is based on the specified method and the quantity (for example, the number of pill packs or condoms distributed) for each of the pharmacy and onsite claims during the study period. If women switched methods, coverage was calculated on the aggregate of both methods without double counting periods of overlap. For this study, we focused on how contraceptive method provision affected the length of the interpregnancy interval, so the maximum length of coverage that was counted was up to 18 months from a woman's index birth. For LARC, unless a removal claim was found, we assigned the maximum length of coverage. Emergency contraception was not assigned any days of coverage. When a woman received multiple contraceptive methods in the same time period, we estimated contraceptive coverage based on the most effective method. The length of coverage was summed across service dates, from the first postpartum visit until the 18-month cutoff.

 

To identify Family PACT program providers, we used the provider enrollment status from the Medi-Cal Provider Master File, which contains provider information that was entered at the time of enrollment and is updated periodically. Medi-Cal providers who were not enrolled in the Family PACT program will be referred to as Medi-Cal–only providers.

 

Client demographics were determined from the mother's information that had been recorded in birth certificates. Demographic variables included education level (less than high school, high school and some college, or college graduate and higher); race/ethnicity (white, African American, Latina, Asian and Pacific Islander, Native American, or other/unknown); country of birth (United States or foreign-born); age at index birth (continuous variable); and parity (2 births or >2 births). Univariate analyses were performed on the demographic variables to examine the distribution of the cohort. Additionally, we examined the program in which women received postpartum services by provider enrollment status (enrolled in the Family PACT program only, Medi-Cal only, or both).

 

We constructed 2 logistic regression models to determine the relationship between contraceptive method receipt and interpregnancy intervals. With the use of multivariate logistic regression, model 1 examined the relationship between the effectiveness of contraceptive method by the maximum tier received and interpregnancy intervals. The main predictor, maximum tier, was defined as a categoric variable for which tier 3 (lowest effectiveness) was the reference group. Model 2 examined the relationship between the length of contraceptive coverage (continuous variable in months) and interpregnancy intervals. In model 2, we limited the regression analysis to users of maximum tiers 2 and 3 so that we could examine the effect of consistent use of user-dependent methods on birth interval length. Both models predicted the outcome of an optimal birth interval and controlled for provider enrollment in the Family PACT program, race/ethnicity, country of birth, education level, age at index birth, and parity. We used SAS software (version 9.2, PROC LOGISTIC; SAS Institute, Cary, NC) for all analyses.

 

Results

Sample characteristics

The cohort included 117,644 women with a Medi-Cal–reimbursed delivery who were seen at least once by a Medi-Cal–only or Family PACT program provider within 18 months after the delivery of their index birth. Of these women, 64% had optimal interpregnancy intervals (n = 74,883), and 36% (n = 42,761) had short birth intervals (Table 1).

  

      
      

 

Most of the women received user-dependent hormonal contraceptives as their most effective method (tier 2; n = 65,103; 55%). A much smaller percentage used barrier methods (tier 3; n = 8320; 7%), followed by LARC (tier 1; n = 5131; 4%). One-third of the women (n = 39,090; 33%) had no contraceptive claims.

 

In the entire cohort, the average length of contraceptive coverage was 3.81 ± 4.84 (SD) months. Most women (n = 60,894) initiated a tier 2 method as their first method after birth and tended to start using this method sooner after birth (n = 60,894; median = 1.9 months) than women who started a tier 1 method (n = 2801; median = 2.2 months) or tier 3 method (n = 14,859; median = 3.0 months). Approximately 40% of women who used tier 3 methods switched to a more effective method within the study period (Table 2).

 

      

 

Coverage varied greatly by the maximum tier received because women who received LARC (tier 1) had the highest mean contraceptive coverage (mean = 10.7 ± 5.96 months), followed by tier 2 methods (mean = 5.96 ± 4.60 months), and tier 3 methods (mean = 0.61 ± 0.45 months). Women who received tier 1 and 2 methods may have received contraception with up to 18 months of coverage; however, clients who received only tier 3 methods received a maximum of 6.85 months of contraception (Table 3)

 

      

 

More than one-half of the cohort (n = 65,975 women; 56%) was served by both the Family PACT program and Medi-Cal–only providers. Eleven percent of the women (n = 13,283) were served solely by the Family PACT program providers, and 33% of the women (n = 38,386) were served solely by Medi-Cal–only providers.

 

Most of the population was Latina (72%), followed by non-Hispanic white (14%). The remaining groups were relatively small and ranged from 7% African American and 5% Asian and Pacific Islander to <1% each of Native American or other/unknown. Although approximately one-half of the cohort (n = 59,462 women; 51%) was foreign born, this proportion varied widely among the racial and ethnic groups from 70% of Asian and Pacific Islander women and 62% of Latinas to <1% of Native American women. Among the groups, Asian and Pacific Islander women had the lowest proportion who used tier 2 methods (40%) and the highest proportion that used no method (53%; data not shown).

 

Forty-three percent of women had less than a high school diploma, and 52% had a high school degree or some college. Only 3% in the sample were college graduates. Nearly two-thirds of the women (64%) were 20-29 years old at the time of the index birth. Twenty-two percent of the women were <20 years old, and 14% of them were ≥30 years old. More than one-half of the sample (54%) had had >1 birth before 2008.

 

Relationship of predictor variables

When we assessed for colinearity, only 2 variable groups were correlated moderately. Older age at index birth had a positive association with higher parity (rs = 0.43; P < .0001), and higher education had a negative association with being foreign born (rs = –0.30; P < .0001). However, we observed a strong correlation between the maximum tier method received and the length of contraceptive coverage (rs = 0.81; P < .0001). We constructed 2 separate regression analyses with tier and contraceptive coverage as predictor variables. Contraceptive coverage for LARC methods does not depend on user behavior. Therefore, we limited the regression analysis for contraceptive coverage to the user-dependent tier 2 or 3 method.

 

Regression analysis

The 2 logistic regression models estimated the odds of women who achieved optimal interpregnancy intervals by contraception received (by tier and by length of coverage) while controlling for other factors that potentially were associated with interpregnancy intervals (provider enrollment, race/ethnicity, nativity, education level, age at index birth, and parity; Table 4).

 

      
      

 

Model 1 focused on the association between the effectiveness of a contraceptive method and optimal birth interval with control for covariates. The main predictor was highly significant (P < .0001). Women who used tier 1 methods had 3.89 times the odds (odds ratio [OR], 3.89; 95% confidence interval [CI], 3.55–4.26) of achieving an optimal birth interval compared with women who used tier 3 methods. Women who used tier 2 methods experienced a slightly lower, but significant, effect with 1.89 times the odds (OR, 1.89; 95% CI, 1.80–1.98), and women who used no contraceptive method had 0.66 times the odds of obtaining an optimal birth interval compared with women who used tier 3 methods (OR, 0.66; 95% CI, 0.63–0.69).

 

Model 2 examined the effect of length of contraceptive coverage of women who used tier 2 or 3 method on the outcome of an optimal birth interval when the data were controlled for covariates. For each additional month of contraceptive coverage, the odds of having an optimal birth interval increased by 8% (OR, 1.08; 95% CI, 1.08–1.09). Both models held the same general trends with regard to the predictor variables.

 

Women who were seen by a Family PACT program provider (model 1: OR, 1.67; 95% CI, 1.60–1.75; model 2: OR, 1.90; 95% CI, 1.78–2.04) or by both the Family PACT program and Medi-Cal–only providers (model 1: OR, 1.25; 95% CI, 1.22–1.29; model 2: OR, 1.38; 95% CI, 1.33–1.43) had significantly higher odds of achieving an optimal birth interval compared with women who were served solely by Medi-Cal–only providers.

 

Women with higher levels of education had significantly greater odds of optimal intervals. Compared with women with ≤12th grade education, women who completed high school had approximately 1.2 times the odds and college graduates had approximately 1.5-1.6 times the odds of reaching an optimal interpregnancy interval.

 

In both models, Latina women had lower odds of optimal intervals (model 1: OR, 0.95; 95% CI, 0.91–0.98; model 2: OR, 0.92; 95% CI, 0.87–0.97); African-American women had higher odds (model 1: OR, 1.15; 95% CI, 1.09–1.22; model 2: OR, 1.27; 95% CI, 1.17–1.38), relative to white women. In both models, the odds of achieving optimal intervals among women with other/unknown ethnicity compared with white women were not significant. Asian and Pacific Islander women had significantly lower odds than white women in model 1 (OR, 0.86; 95% CI, 0.81–0.92), although there was no significant difference in model 2. In contrast, Native American women had significantly lower odds in model 2 (OR, 0.74; 95% CI, 0.59–0.93) but no significant difference in model 1. In both models, foreign-born women had 1.3 times the odds of having an optimal interpregnancy interval compared with women who were born in the United States.

 

The influence of age at index birth on interpregnancy intervals was minor but significant, which indicates that the odds of having an optimal interval decreased by 1% for each additional year of age. Women who had >2 previous births had 1.03 times the odds in model 1 (OR, 1.03; 95% CI, 1.01–1.06) and 1.07 times the odds in model 2 (OR, 1.07; 95% CI, 1.03–1.11) of an optimal interval, compared with women who had only 2 births.

 

Comment

This study highlights the importance of the provision of effective contraception at the time of postpartum follow-up evaluation as a key strategy for the achievement of optimal interpregnancy intervals. Long-acting reversible contraception has the ability to protect from unintended pregnancy consistently for a long period of time and is recommended as a preferred contraceptive option.21 Data on how to promote method initiation after birth are mixed.22 and 23 A study that used patient navigators to provide personalized contraceptive assistance did not show a significant impact on getting women interested in the use of intrauterine contraceptives and implants in the postpartum24 period; however, other studies suggest that high-quality client-centered contraceptive counseling during the healthcare visit may facilitate the uptake of highly effective methods.25 Policies and hospital protocols that facilitate immediate postpartum insertion of LARC have become more common and promise ultimately to contribute to better contraceptive coverage after birth.

 

In regression analyses, controlling for Medi-Cal provider type and client demographics, we found that the higher the effectiveness of the contraceptive, the stronger the odds of an optimal interpregnancy interval compared with tier 3 methods. Similarly, increased length of contraceptive coverage significantly increased the odds of having an optimal interpregnancy interval. Current national indicators that measure postpartum quality include an indicator to determine postpartum visit rates.26 However, it is crucial to monitor contraceptive coverage of women who use tier 2 and 3 methods after delivery. Health programs must enable the ongoing use of effective contraceptive methods through 18 months after birth.

 

Clients who were seen by a Family PACT program provider, independent of whether the contraceptive claim was reimbursed through Medi-Cal or the Family PACT program, had significantly higher odds of having optimal interpregnancy intervals than if they were seen only by a Medi-Cal provider. This finding suggests that clear expectations to adhere to evidence-based clinical standards combined with provider training and support were successful in increasing the effective use of contraception to prevent unintended pregnancies.27 With the expansion of publicly funded programs as part of healthcare reform, these types of provider support and performance expectations should be maintained to ensure high-quality reproductive healthcare. For example, performance standards can be included in managed care contracts and in funding requirements of safety net providers.

 

Client demographics were also associated with optimal interpregnancy intervals. As in previous studies, higher education and being foreign born were associated significantly positively with optimal interpregnancy intervals.11 and 28 Most foreign-born women in this study were Latinas. When we controlled the data for country of birth, Latinas were less likely to have optimal interpregnancy intervals, which suggests that Latinas who were born in the United States may be in need of special attention in the postpartum period. One explanation of these intergroup differences may be different breastfeeding rates. Foreign-born Latinas are more likely to breastfeed and to breastfeed longer than Latinas who were born in the United States, which may provide additional protection from unintended pregnancy.29, 30, 31 and 32 Other minority groups, such as Asian and Pacific Islander women and Native American women were also less likely to achieve optimal interpregnancy intervals when we controlled for covariates. The exception was African American women who were significantly more likely to achieve optimal interpregnancy intervals. It may be that enabling factors (such as being seen by a Family PACT program provider or having a higher education) have a stronger influence on the contraceptive choices and behaviors in this group. As a result of the relatively small number of Asian and Pacific Islander women and the high percentage of those women not using contraception, the data were unable to provide a significant odds ratio that relates to months of contraceptive coverage.

 

Women who have higher parity tend to be older. We observed that, with increasing age, women had lower odds of having optimal interpregnancy intervals and that women with at least 1 child at the index birth tended to have higher odds of optimal interpregnancy intervals. We could not determine women's pregnancy intentions and therefore do not know to what extent the lack of the use of contraception was associated with the desire to get pregnant again. Women who are older and who wish to have another child may be more concerned about their ability to conceive and therefore may not want to wait 18 months to get pregnant. In contrast, births to women who initiate childbearing at <30 years old are significantly less likely to have shorter interpregnancy intervals than births to women ≥30 years old at first birth.33 Women with higher parity were observed to be more likely to use contraception after the index birth, which may be explained with a stronger desire to space their children.

 

The use of a large population-based sample enabled us to control for important covariates. It also avoided sampling and recall bias and loss of follow up. However, because we had access only to publicly funded data, the variables that influence contraceptive use are limited to low-income women. Contraceptives that were purchased over the counter or through other payer sources were not measured, which led to an underestimation of contraceptive coverage for women who rely only on barrier methods. Furthermore, our analysis of administrative records did not assess client attitudes on family planning practices, breastfeeding practices, expectations for family size, and birth spacing.

 

Access to and use of family planning services are critical to the achievement of a longer birth interval. The positive association of optimal birth intervals that corresponds to the method tier demonstrates the advantage of using methods with longer duration and lower rates of contraceptive failure. In the provision of user-dependent methods, an extended length of coverage also facilitates longer interpregnancy intervals. In order to achieve optimal birth spacing and ultimately to improve birth outcomes, attention should be given to the assurance of client-centered contraceptive counseling and access to contraceptive methods in the postpartum period.