If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Delivery Decisions Initiative, Ariadne Labs, Boston, MABeth Israel Deaconess Medical Center, Boston, MAOptumLabs Visiting Fellow, Cambridge, MAMaven Clinic, New York, NY
Despite the importance of prenatal care, quality measurement efforts have focused on the number of prenatal visits, or prenatal care adequacy, rather than the services received. It is unknown whether attending more prenatal visits is associated with receiving more guideline-based prenatal care services. The relationship between guideline-based prenatal care and patients’ clinical and sociodemographic characteristics has also not been studied.
Objective
This study aimed to measure the receipt of guideline-based prenatal care among pregnant patients and to describe the association between guideline-based prenatal care and the number of prenatal visits and other patient characteristics.
Study Design
This was a retrospective descriptive cohort study of 176,092 pregnancy episodes between 2016 and 2019. We used de-identified administrative claims data on commercial enrollees across the United States from the OptumLabs Data Warehouse. We identified the following 8 components of prenatal care that are universally recommended by the American College of Obstetricians and Gynecologists and other guideline-issuing organizations: testing for sexually transmitted infections, obstetric laboratory test panel, urine culture, urinalysis, anatomy scan ultrasound, oral glucose tolerance test, tetanus, diphtheria, and pertussis vaccine, and group B Streptococcus test. We measured the proportion of pregnant patients who received each of these guideline-based services at the appropriate gestational age. We measured the association between guideline-based services and the number of prenatal visits and prenatal care adequacy. We described variation of guideline-based care according to patient age, comorbidities, high deductible health plan enrollment, and their county’s rurality, health professional shortage area status, racial composition, median income, and educational attainment.
Results
The 176,092 pregnancy episodes were mostly among patients aged 25 to 34 years (63%) with few pregnancy comorbidities (81%) and living in urban areas (92%). Guideline-based care varied by service, from 51% receiving a timely urinalysis to 90% receiving an anatomy scan and 91% completing testing for sexually transmitted infections. Patients with at least 4 prenatal visits received, on average, 6 of the 8 guideline-based services. Guideline-based care did not increase with additional prenatal visits and varied by patient characteristics. Rates of tetanus, diphtheria, and pertussis vaccination were lower in counties with high proportions of minoritized populations, lower education, and lower income.
Conclusion
In this commercially insured population, receipt of guideline-based care was not universal, did not increase with the number of prenatal visits, and varied by patient- and area-level characteristics. Measuring guideline-based care is feasible and may capture quality of prenatal care better than visit count or adequacy alone.
Instead of tracking the services provided, quality measurement organizations and researchers describe prenatal care in terms of the timing of its initiation and the number of visits, together referred to as “adequacy” of care.
These 2 elements of care may capture important components of quality, but they fail to describe whether the patient received the recommended care such as laboratory screenings, psychosocial support, and education on childbirth and parenting.
If this finding is true of prenatal care, it may be the case that 2 pregnant patients with “adequate” prenatal care receive very different quality of care.
This distinction is relevant to the growing debate in the United States about the number and structure of prenatal visits.
As clinicians and policymakers rethink the models of prenatal care, it is important to understand the existing variation in the services provided, rather than just the quantity and timing of prenatal visits.
It is unknown whether a patient’s number of prenatal visits is positively associated with their receipt of guideline-based prenatal care services. We used national data to examine variation in the provision of 8 guideline-based services by a patient’s number of prenatal visits and other characteristics.
Key findings
Many patients do not receive these guideline-based services. There is little variation in the number of guideline-based services received among patients with 6 or more prenatal visits.
What does this add to what is known?
There is a significant gap in the provision of guideline-based prenatal care that is not well explained by the number of visits. Using number of visits alone to describe prenatal care quality masks important heterogeneity in the services received.
This study moves beyond adequacy to measure whether prenatal care is consistent with national clinical guidelines in a commercially insured population. We overcome barriers to such an analysis by using a national database of de-identified health insurance claims that includes prenatal services billed individually instead of as a global fee.
We describe variation in guideline-based care according to a patient’s number of prenatal visits, adequacy, and their clinical and demographic characteristics.
Materials and Methods
Data
This is a retrospective descriptive study of prenatal care utilization among pregnant patients with commercial insurance. We used de-identified administrative claims data from the OptumLabs Data Warehouse (OLDW), which include medical and pharmacy claims and enrollment records for commercial enrollees. The database contains longitudinal health information on enrollees, representing a mixture of ages and geographic regions across the United States.
US Preventive Services Task Force. Douglas K.O, Karina WD, HK, et al. Screening for asymptomatic bacteriuria in adults: us preventive services task force recommendation statement. JAMA 2019;322:1188–1194.
The study population included commercially insured enrollees who had a claim for delivery of a neonate between June 1, 2016, and July 1, 2019 (Appendix A provides billing codes used to identify delivery episodes). For each delivery episode, we identified a corresponding approximate date of last menstrual period (LMP) using International Classification of Diseases, 10th Revision, (ICD-10) codes (Appendix B and Appendix C).
Enrollees were required to have continuous coverage in a commercial plan with medical and pharmacy benefits starting 6 months before the estimated date of LMP; this criterion was necessary for accurate measurement of the obstetric comorbidity index (OCI).
All included patients were aged from 12 to 55 years at their approximate date of LMP and were not listed as male sex. We excluded patients with missing age, sex, or ZIP code data, which are indicators of incomplete or incorrect data capture in the OLDW (Appendix D describes the construction of the analytical cohort).
Outcomes
We measured “guideline-based care,” or timely receipt of specific prenatal services according to guidelines set by the American College of Obstetricians and Gynecologists, the Centers for Disease Control and Prevention, the US Preventive Services Task Force, and/or the International Society of Ultrasound in Obstetrics and Gynecology. Table 1 shows the 8 components of guideline-based care, the organization(s) that issued the corresponding recommendations, and their clinical purpose. These 8 components of care do not include all services that patients should receive prenatally; however, they are recommended for all pregnant patients and we can interpret the absence of any of this care as a departure from clinical guidelines. The relative value of the 8 guideline-based services, or their respective contributions to preventing maternal and infant morbidity and mortality, varies based on patient risk factors (eg, drug use, multiple sexual partners, preexisting hypertension, or diabetes).
Table 1Guideline-based prenatal care and recommending organizations
American Academy of Pediatrics. In: American College of Obstetricians and Gynecologists, ed. Guidelines for perinatal care. 7th ed. American Academy of Pediatrics; American College of Obstetricians and Gynecologists, Elk Grove Village, IL : Washington, DC; 2012.
American Academy of Pediatrics. In: American College of Obstetricians and Gynecologists, ed. Guidelines for perinatal care. 7th ed. American Academy of Pediatrics; American College of Obstetricians and Gynecologists, Elk Grove Village, IL : Washington, DC; 2012.
American Academy of Pediatrics. In: American College of Obstetricians and Gynecologists, ed. Guidelines for perinatal care. 7th ed. American Academy of Pediatrics; American College of Obstetricians and Gynecologists, Elk Grove Village, IL : Washington, DC; 2012.
American Academy of Pediatrics. In: American College of Obstetricians and Gynecologists, ed. Guidelines for perinatal care. 7th ed. American Academy of Pediatrics; American College of Obstetricians and Gynecologists, Elk Grove Village, IL : Washington, DC; 2012.
Concordance with guidelines for the second and third trimester services was only evaluated among pregnancies that reached the maximum recommended gestational age for each service (ie, 22 completed weeks for the anatomy scan, 28 completed weeks for gestational diabetes screening, 36 completed weeks for the TDAP vaccine, and 37 completed weeks for the group B Streptococcus test). This is because, for example, pregnancies lasting 30 weeks should not be expected to include a group B Streptococcus test at 35 to 37 weeks
American Academy of Pediatrics. In: American College of Obstetricians and Gynecologists, ed. Guidelines for perinatal care. 7th ed. American Academy of Pediatrics; American College of Obstetricians and Gynecologists, Elk Grove Village, IL : Washington, DC; 2012.
Concordance with guidelines for the second and third trimester services was only evaluated among pregnancies that reached the maximum recommended gestational age for each service (ie, 22 completed weeks for the anatomy scan, 28 completed weeks for gestational diabetes screening, 36 completed weeks for the TDAP vaccine, and 37 completed weeks for the group B Streptococcus test). This is because, for example, pregnancies lasting 30 weeks should not be expected to include a group B Streptococcus test at 35 to 37 weeks
American Academy of Pediatrics. In: American College of Obstetricians and Gynecologists, ed. Guidelines for perinatal care. 7th ed. American Academy of Pediatrics; American College of Obstetricians and Gynecologists, Elk Grove Village, IL : Washington, DC; 2012.
Concordance with guidelines for the second and third trimester services was only evaluated among pregnancies that reached the maximum recommended gestational age for each service (ie, 22 completed weeks for the anatomy scan, 28 completed weeks for gestational diabetes screening, 36 completed weeks for the TDAP vaccine, and 37 completed weeks for the group B Streptococcus test). This is because, for example, pregnancies lasting 30 weeks should not be expected to include a group B Streptococcus test at 35 to 37 weeks
Concordance with guidelines for the second and third trimester services was only evaluated among pregnancies that reached the maximum recommended gestational age for each service (ie, 22 completed weeks for the anatomy scan, 28 completed weeks for gestational diabetes screening, 36 completed weeks for the TDAP vaccine, and 37 completed weeks for the group B Streptococcus test). This is because, for example, pregnancies lasting 30 weeks should not be expected to include a group B Streptococcus test at 35 to 37 weeks
In July 2019 (after our study period), the group B Streptococcus screening guidelines were amended from recommending screening at between 35 to 37 weeks to recommending screening between 36 to 39 weeks’ gestation.37
American Academy of Pediatrics. In: American College of Obstetricians and Gynecologists, ed. Guidelines for perinatal care. 7th ed. American Academy of Pediatrics; American College of Obstetricians and Gynecologists, Elk Grove Village, IL : Washington, DC; 2012.
Prevent transmission to the newborn which can cause sepsis
ACOG, American College of Obstetricians and Gynecologists; CDC, Centers for Disease Control and Prevention; IgG, immunoglobulin G; ISUOG, International Society of Ultrasound in Obstetrics and Gynecology; RBC, red blood cell; Rh, rhesus; TDAP, tetanus, diphtheria, and pertussis; USPSTF, US Preventive Services Task Force.
Gourevitch et al. Variation in guideline-based prenatal care. Am J Obstet Gynecol 2022.
a Recommended timing of the service as specified by the organization’s guidelines
b Concordance with guidelines for the second and third trimester services was only evaluated among pregnancies that reached the maximum recommended gestational age for each service (ie, 22 completed weeks for the anatomy scan, 28 completed weeks for gestational diabetes screening, 36 completed weeks for the TDAP vaccine, and 37 completed weeks for the group B Streptococcus test). This is because, for example, pregnancies lasting 30 weeks should not be expected to include a group B Streptococcus test at 35 to 37 weeks
c In July 2019 (after our study period), the group B Streptococcus screening guidelines were amended from recommending screening at between 35 to 37 weeks to recommending screening between 36 to 39 weeks’ gestation.
Guideline-based care requires receiving the right services at the right time during pregnancy. Care was only considered guideline-based if the claim for that service occurred at the recommended pregnancy stage (when applicable) based on the difference between the date the service was provided and the patient’s approximate date of LMP (Appendix E provides results without the timeliness constraint and Appendix F provides a sensitivity analysis).
Prenatal care is often included in a global fee with the delivery hospitalization, which has limited the utility of claims data as a measure of prenatal services and visits.
However, claims for prenatal, delivery, and postpartum care in the OLDW are not bundled or billed under a global fee. Instead, each prenatal visit and service is individually billed. This data resource therefore represents a unique opportunity to overcome key limitations of claims data for measuring the number of prenatal visits and the services patients receive. We used ICD-10 codes, Current Procedural Terminology codes, Healthcare Common Procedure Coding System, and National Drug Codes to identify the 8 components of guideline-based care (Appendix G) and the number of prenatal visits (Appendix B and Appendix H provide details). We carefully selected the billing codes used to identify visits and services, testing the sensitivity of our definitions to exclusion and inclusion of relevant codes and relying on previous work where possible.
We measured the number of prenatal visits before delivery and the Adequacy of Prenatal Care Utilization (APNCU) index, which is also known as the Kotelchuck Index. The APNCU index describes the totality of a patient’s prenatal care as inadequate, intermediate, adequate, or adequate plus based on the timing of their prenatal care initiation and by comparing the number of prenatal visits they received with the number that they would have been recommended to receive based on their gestational age at initiation and at delivery.
We used enrollment and benefit design information in the OLDW to measure patient age at the approximate date of LMP and to capture high deductible health plan (HDHP) enrollment. The enrollment data included the patient’s ZIP code of residence, which we used to link to county characteristics from the Area Health Resources Files (AHRF). The AHRF data include US Department of Agriculture urban influence codes, which we used to identify whether the patient’s county is urban, nonurban urban-adjacent, or rural, consistent with previous work.
We also used the AHRF to identify whether the patient resided in a full or partial primary care health professional shortage area (HPSA).
To maintain de-identification of the OLDW data some individual-level demographic elements were not available for this analysis. We instead described race, income, and educational attainment at the county-level using episode-weighted percentiles of data from the AHRF. For race/ethnicity, we created three indicator variables capturing whether the patient resided in a county that was above the 75th percentile of counties with percent of the population that was (1) White non-Hispanic/Latino, (2) Black non-Hispanic/Latino, and (3) Hispanic/Latino. For income, we created a categorical variable with 3 levels corresponding to county terciles of median annual household income (low [<$56,732], middle [$56,732 to $70,461], or high [>$70,461]). For educational attainment, we created a categorical variable for whether the patient resided in a county in the lowest, middle, or highest tercile of the percentage of individuals aged 25 or older with less than a high school diploma (lowest attainment [>12.90% without diploma], middle attainment [9.21% to 12.90% without diploma], and highest attainment [<9.21% without diploma]).
Analyses
We report the overall levels of each guideline-based service across our analytical sample and describe how a patient’s number of guideline-based services received varies by their number of prenatal visits and their APNCU index category. We also measure variation in the number and type of guideline-based services across patient- and area-level characteristics.
All analyses were conducted at the delivery episode level. Descriptive analysis was performed with SAS version 9.4 (SAS Institute Inc, Cary, NC) and figures were created in R version 4.0.2 (R Core Team, 2020, Vienna, Austria); all analyses were independently replicated by a second data analyst. Because this study involved analysis of preexisting, de-identified data, it was deemed exempt from institutional review board approval.
We identified 329,101 pregnancy episodes with dates of delivery and LMP in our study time frame, of which 176,092 (54%) met our study inclusion criteria (Appendix D). The 176,092 delivery episodes were among 171,107 individual patients (3% of patients had >1 delivery episode during our study period). The pregnancies in our sample were mostly among patients between the ages of 25 and 34 years (63%) and with a few obstetrical comorbidities (81% had an OCI score of ≤2) (Table 2). Most patients were not in an HDHP (75%). Nearly all resided in an urban county (92%) and in a partial primary care HPSA (88%). The cohort was evenly distributed across the terciles of county racial composition, median income, and educational attainment.
A county is classified as having a “high” population of each racial or ethnic group if that county is above the 75th percentile (episode-weighted) of counties’ proportion of the population from that racial or ethnic group
The county median household income categories corresponded to terciles of episode counties’ household median income (low [<$56,732], middle [$56,732–$70,461], or high [>$70,461])
County educational attainment categories indicate whether the county is in the lowest, middle, or highest tercile of the percent of individuals ages ≥25 years with less than a high school diploma among the pregnancy episodes’ counties (lowest attainment [>12.90% without diploma], middle attainment [9.21%–12.9% without diploma] and highest attainment [<9.21% without diploma]).
Urban
162,064 (92.0)
Lowest education
54,986 (31.2)
Urban-adjacent
9329 (5.3)
Middle education
61,359 (34.8)
Rural
4699 (2.7)
Highest education
59,747 (33.9)
HDHP, high deductible health plan; HPSA, health professional shortage area.
Gourevitch et al. Variation in guideline-based prenatal care. Am J Obstet Gynecol 2022.
a The 176,092 delivery episodes were among 171,107 distinct patients
b A county is classified as having a “high” population of each racial or ethnic group if that county is above the 75th percentile (episode-weighted) of counties’ proportion of the population from that racial or ethnic group
c The county median household income categories corresponded to terciles of episode counties’ household median income (low [<$56,732], middle [$56,732–$70,461], or high [>$70,461])
d County educational attainment categories indicate whether the county is in the lowest, middle, or highest tercile of the percent of individuals ages ≥25 years with less than a high school diploma among the pregnancy episodes’ counties (lowest attainment [>12.90% without diploma], middle attainment [9.21%–12.9% without diploma] and highest attainment [<9.21% without diploma]).
Receipt of guideline-based prenatal care was not universal. As seen in Figure 1, approximately half of the cohort received a timely urinalysis (51%), and less than three-quarters had a timely obstetrical laboratory panel (61%) or received a tetanus, diphtheria, pertussis (TDAP) vaccination (63%). The components of guideline-based care that were most consistently received were an anatomy scan ultrasound (90%) and testing for sexually transmitted infections (STIs) (91%). Among those who did not undergo the full obstetrical laboratory panel, almost half did receive a timely complete blood count (CBC) (47%). Among those who did not receive the full STI testing panel, most were tested for syphilis (51%), but less than half were tested for hepatitis B (41%) or HIV (20%) (Appendix I).
Figure 1Receipt of guideline-based care in the full sample (N=176,092)
The figure displays the percentage of the full sample that received each component of guideline-based prenatal care at the recommended gestational age (timely guideline-based care). Table 1 provides additional details on the guideline-based services.
Gourevitch et al. Variation in guideline-based prenatal care. Am J Obstet Gynecol 2022.
The number of guideline-based services received did not have a strong association with the number of prenatal patient visits. Enrollees with at least 4 prenatal visits (97% of the sample) received, on average, 6 of the 8 timely guideline-based services. Figure 2 shows the distribution of the number of guideline-based services received by the binned number of prenatal visits (left) and by adequacy of prenatal care (right). Individuals who had <6 visits or inadequate prenatal care were less likely to receive all 8 of the guideline-based services. However, the distribution of the number of guideline-based services received is similar among patients with ≥6 visits and those with intermediate, adequate, or adequate plus care.
Figure 2Relationship between the number of guideline-based services received and number of prenatal visits (left) and the adequacy of prenatal care utilization index (right)
The proportion of patients who received at least 6 of the 8 timely guideline-based services varied by patient characteristics (Figure 3 and Appendix J). 60% of patients <25 years of age received at least 6 timely services compared with 71% of patients aged ≥25 years. We observed little difference in the rates of guideline-based care by HDHP enrollment (71% of HDHP enrollees vs 69% of non-HDHP enrollees received at least 6 timely services). Patients with higher-risk pregnancies were less likely to receive at least 6 timely services than patients with no obstetrical comorbidities (64% among patients with OCI ≥7 vs 71% among patients with OCI=0).
Figure 3Percentage of patients who received at least 6 out of the 8 guideline-based prenatal care services studied by patient characteristics
In addition, patients living in nonurban counties or counties that included a HPSA were less likely to receive at least 6 timely services than patients in urban counties or counties without a HPSA (64% in rural counties vs 70% in urban counties and 65% in full-HPSA counties vs 74% in non-HPSA counties). Patients living in counties with a high non-Hispanic White population were more likely to receive at least 6 timely services (72%) than patients living in counties with a high non-Hispanic Black (68%) or high Hispanic/Latino population (66%). Patients in counties with a lower median income or lower educational attainment were less likely to receive at least 6 timely services (65% among the lowest income tercile vs 74% among highest income tercile and 66% among lowest education tercile vs 74% among highest education tercile).
Of the 8 timely guideline-based services studied, TDAP vaccination showed the most pronounced variation by county-level demographics. Counties with the lowest educational attainment had a 29% lower rate of TDAP vaccination than counties with the highest educational attainment (55% vs 71%). Counties with the lowest median income had a 27% lower rate of TDAP vaccination than counties with the highest median income (56% vs 71%). TDAP vaccination rates also varied by county-level racial composition as follows: compared with counties with a high non-Hispanic White population, counties with a high non-Hispanic Black population had 13% lower rates of TDAP vaccination and counties with a high Hispanic population had 24% lower rates of TDAP vaccination (55% among high Hispanic counties, 60% among high non-Hispanic Black counties, and 68% among high non-Hispanic White counties).
Comment
Principal findings
Services provided during prenatal care are not always consistent with clinical guidelines. The relationship between the number of prenatal visits or prenatal care adequacy and guideline-based care is weak. Patients with very few prenatal care visits or inadequate care received fewer guideline-based services. However, there was little difference in receipt of these services among patients with 6 or more visits or among patients with intermediate, adequate, or adequate plus care.
Results
Younger patients were less likely to receive guideline-based care, which may be driven by an association between age and pregnancy intention. People with unintended pregnancies are more likely to be younger and to have inadequate prenatal care.
We find little variation in guideline-based care by deductible level, which may be because these preventive services are exempt from deductible spending or because patients anticipate surpassing their deductible during the year of a pregnancy. Patients with many comorbidities were less likely to receive guideline-based services; these cases may represent clinically appropriate deviations from guideline-based care, such as not screening for gestational diabetes in patients with preexisting diabetes. However, this finding and the observed variation across rurality and HPSA should be interpreted with caution given that only a small proportion of our sample is in the highest comorbidity, nonurban, or full-HPSA groups.
We find lower rates of TDAP vaccination in counties with a high proportion of non-Hispanic Black or Hispanic residents and in those with low educational attainment or median income. Although we cannot be sure that our patients are representative of their county demographics, other work has found lower TDAP vaccination rates among pregnant patients from minoritized populations, with lower educational attainment, or with incomes below the poverty level.
These disparities may reflect differences in the likelihood of the provider offering the vaccination or lower uptake rates of vaccinations among these vulnerable populations—both of which are linked to structural and interpersonal racism.
Our results suggest that many commercially insured pregnant patients are not receiving timely guideline-based services. Some patients who do not receive timely guideline-based care instead receive these services at another point in pregnancy (Appendix E). Still, these services were not universally applied in our sample despite being recommended for all pregnant patients.
Moreover, our results illustrate that the number of prenatal visits and adequacy do not sufficiently capture the quality of care. The measures of guideline-based care that we implement may be a suitable starting point for more meaningful measurements of prenatal care quality; these are universally recommended based on clinical evidence and are observable in health insurance claims data. Measuring the quality of care provided during the management of conditions like gestational diabetes or group B streptococcus infection, which are detected via the tests that we study here, is an important direction for future quality measurement efforts. In addition, understanding the quality and consistency of nonbillable physical examinations (eg, blood pressure monitoring, fundal height measurements) and the quality of the counseling and anticipatory guidance provided during prenatal visits is important for evaluating prenatal care quality.
These findings have implications for the ongoing work of redesigning prenatal care to be better suited to patients’ diverse medical and psychosocial needs.
As we move toward more flexible prenatal care models, the need for quality measures that go beyond visit number will only become more relevant.
Research implications
The primary value of high-quality prenatal care is improving maternal and neonatal outcomes. Future work should quantify the impact of guideline-based care on maternal and neonatal outcomes such as preterm delivery, birthweight, and infection. In addition, other measures of quality including patient-reported outcomes and satisfaction should continue to be developed and incorporated into quality measurement strategies.
There are many potential explanations for why a patient may not have received a component of guideline-based care. First, the provider may not submit a claim to the insurer. We use unbundled claims and find high rates of billing for these services across pregnancy (Appendix E), and therefore we do not believe that inconsistent billing practices are driving our results. Second, the patient may be unable to, or choose not to, receive the service, or the provider may feel that the service is not appropriate for that patient (eg, gestational diabetes testing for patients with preexisting diabetes). Patients may decline these services because of structural factors impeding access to care, like transportation or the ability to take time off from work, or hesitancy around blood draws or vaccinations.
Future work should examine the roles of patient and physician discretion in driving variation in guideline-based care.
In addition, measuring guideline-based care among patients covered by Medicaid or who churn across insurers before or during pregnancy is an important direction for future work.
Other work has shown that women covered by Medicaid may be less likely to receive preventive care; therefore, this analysis may overestimate the rates of guideline-based care across all pregnant people.
Our study adds to the literature on prenatal care by demonstrating that measuring only prenatal visit number, or adequacy, can mask important variation in guideline-based care. We used a large, nationwide sample rich in information on patients’ clinical and sociodemographic characteristics.
Our sample only included commercially insured women who were continuously enrolled in coverage during pregnancy; our results may therefore not be generalizable to pregnancies covered by Medicaid or to people who change insurers during pregnancy. We did not observe individual patient race, education or income, nor did we observe those variables at a geographic level smaller than county. Correlations between county-level demographics and receipt of guideline-based care should be interpreted with caution because the individuals in our sample may not be representative of the counties in which they live (and, inferring individual-level patterns from area-level estimates is subject to the ecologic fallacy).
We rely on claims data, which do not capture any services that were provided but not billed to the insurance company. Our lists of billing codes are based on previous analyses using claims when possible.
In addition, we use unbundled claims, which should limit the number of unbilled services relative to other databases of obstetrics claims.
Conclusion
Our work shows that receipt of guideline-based care is not universal, does not increase with the number of prenatal visits, and that care varies across patient characteristics. Future research and policy attention should be devoted to more meaningful measurements of prenatal care quality.
Algorithm to approximate date of last menstrual period
We developed a novel algorithm for approximating the date of the last menstrual period (LMP)—which marks the start of a pregnancy—using International Classification of Diseases (ICD)-10 codes in claims data. Identifying LMP is essential to understanding the gestational age at which each prenatal service was received. Because pregnancies vary in length, it is imprecise to identify the date of LMP as 40 weeks before delivery. Instead, our algorithm uses information contained in diagnosis and procedure codes to identify the date of LMP more precisely. We adapted this algorithm from previous studies that used ICD-9 codes to identify the date of LMP in claims data, and leveraged preliminary work by OptumLabs researchers to convert those algorithms to ICD-10 codes.
Phiri K, Clifford CR, Doherty M, Fan Y, Wang F, Seeger JD. Timing of routine prenatal tests relative to the last menstrual period estimated from an ICD-10 based algorithm. Poster presented at the: 35th International Conference on Pharmacoepidemiology and Therapeutic Risk Management; August 2019; Philadelphia, PA.
The algorithm takes advantage of new diagnosis codes included in the ICD-10 that capture gestational age at any clinical encounter during the pregnancy (the Z3A.XX codes). Appendix C provides the full algorithm.
Identifying prenatal visits
We identified the first prenatal visit as the first claim after the patient’s estimated LMP with a diagnosis or procedure code for prenatal care, an evaluation and management (E&M) visit with an obstetrician-gynecologist, a provider-ordered pregnancy test in a setting other than the emergency department, an obstetrical ultrasound, or any of the obstetrical laboratory panel tests listed in Table 1 of the manuscript (Appendix G provides billing codes). If the patient receives most of their prenatal care from a primary care physician or neonatologist, we also count an E&M visit with that provider type as a prenatal visit.
We identify subsequent prenatal care visits in the same way with the caveat that claims for ultrasounds and laboratory tests do not count as a subsequent prenatal visit. Although these services should be taken as an indication that prenatal care has been initiated, they may be billed by ultrasound technicians or independent laboratories on dates when the patient did not have an in-person prenatal care visit.
Appendix H provides the list of billing codes used to identify prenatal visits.
Appendix C
Supplemental FigureAlgorithm for approximating date of LMP
2. Percentage symbol indicates that any code beginning with the text preceding the % should be captured. 3. Delivery date is defined as the date of admission for the inpatient stay for delivery 4. If an enrollee had >1 delivery during the study period (ie, had 1 or more deliveries that were subsequent to the first), we made the following adjustment to step 2: we looked for Z3A.% codes starting from the date of their subsequent delivery to either (1) 11 months before that delivery date, or (2) the discharge date of prior delivery, whichever was earlier.
ICD-10, International Classification of Diseases, Tenth Revision; LMP, last menstrual period.
Gourevitch et al. Variation in guideline-based prenatal care. Am J Obstet Gynecol 2022.
Three percent of individuals in the sample had 2 or more deliveries during the study period (the final sample consists of 176,092 episodes among 171,107 individuals).
Final analytical sample
LMP, last menstrual period.
Gourevitch et al. Variation in guideline-based prenatal care. Am J Obstet Gynecol 2022.
a The 384,649 episodes originally identified are those with an inpatient delivery date between June 1, 2016 and July 1, 2019
b Three percent of individuals in the sample had 2 or more deliveries during the study period (the final sample consists of 176,092 episodes among 171,107 individuals).
Supplemental Table 3Study outcomes by sample characteristics with and without timeliness constraint (%)
Full Sample
Obstetrical panel
STI panel
Urinalysis
Urine culture
Anatomy scan
Glucose test
TDAP vaccination
Group B Streptococcus test
Timely
Anytime
Anytime
Timely
Anytime
Timely
Anytime
Timely
Timely
Anytime
Timely
Anytime
Timely
Anytime
Full Sample
61
90
91
51
66
82
93
90
80
94
63
71
78
95
APNCU index
Inadequate
10
72
77
78
42
52
72
82
74
66
80
51
57
68
85
Intermediate
30
73
92
93
45
58
84
94
91
82
95
64
72
80
96
Adequate
26
57
92
94
49
66
83
95
92
83
96
67
75
80
97
Adequate Plus
34
50
91
93
60
78
81
94
93
79
94
62
71
77
96
Number of prenatal visits
0–5
14
77
82
83
42
51
77
85
80
71
84
54
61
71
87
6–8
31
72
92
93
45
59
83
94
91
82
95
64
72
80
96
9–11
23
57
91
93
50
67
82
95
92
82
96
66
74
80
97
12–14
14
52
91
93
57
74
82
94
92
81
95
64
72
78
96
≥15
18
42
91
92
63
81
81
95
94
79
94
62
71
76
96
Age (y)
12–24
13
61
87
88
58
76
76
91
83
70
87
49
58
72
91
25–29
25
65
90
91
52
67
82
93
90
81
94
62
70
78
95
30–34
38
62
91
93
49
63
83
94
92
82
95
67
75
79
96
35–55
24
55
90
92
49
64
82
93
92
80
94
65
73
78
95
HDHP
HDHP
25
64
91
91
48
62
82
93
91
80
94
65
73
79
95
No HDHP
75
60
90
91
52
67
81
93
90
80
94
62
70
77
95
Obstetric comorbidity index
0
42
65
90
91
50
63
82
93
89
81
94
63
71
79
96
1–2
39
60
90
92
50
65
82
93
91
80
94
63
71
77
95
3–6
15
55
90
91
53
71
81
93
91
77
93
62
70
76
95
≥7
4
49
90
91
56
77
80
93
92
72
90
60
68
74
93
Rurality
Urban
92
61
90
92
50
65
82
93
90
80
94
63
71
78
95
Urban-adjacent
5
63
88
88
56
73
77
91
88
77
92
58
67
77
94
Rural
3
63
87
87
57
74
74
88
87
74
90
58
67
74
92
Health professional shortage area
Non-HPSA county
9
62
91
92
50
64
85
95
92
82
95
68
76
80
96
Partial-HPSA county
88
61
90
91
51
66
81
93
90
80
94
63
71
78
95
Full-HPSA county
3
64
88
91
55
72
78
91
86
75
91
53
62
74
93
County racial composition
High non-Hispanic White
25
64
89
90
49
65
81
93
92
80
94
68
76
80
95
High non-Hispanic Black
24
60
90
92
48
63
81
93
89
79
93
60
69
77
95
High Hispanic or Latino
25
61
91
92
55
70
81
93
88
79
93
55
63
74
94
County median income
Lowest income
33
62
90
91
52
68
79
92
88
78
93
56
64
75
94
Middle income
33
62
91
90
51
66
80
92
90
79
94
62
70
78
95
Highest income
33
59
90
93
49
63
86
95
92
82
95
71
78
80
96
County educational attainment
Lowest education
31
60
89
92
57
72
79
92
88
78
93
55
64
74
94
Middle education
35
61
91
91
49
65
81
93
90
80
94
61
69
78
95
Highest education
34
62
90
91
47
62
84
94
92
81
95
71
79
81
96
Note: “Timely” indicates that the guideline-based service was received at the recommended gestational age (Table 1). “Anytime” indicates that the service was received anytime between the patient’s approximate date of last menstrual period and the day before their delivery episode. There is no timely column for STI panel because we did not impose a gestational age requirement for this service. There is no anytime column for anatomy scan because it cannot be considered at the 18 to 22 weeks’ gestation anatomy scan if it is outside of that recommended gestational age range.
APNCU, Adequacy of Prenatal Care Utilization; HDHP, High deductible health plan; HPSA, Health professional shortage area; STI, sexually transmitted infection; TDAP, tetanus, diphtheria, and pertussis.
Gourevitch et al. Variation in guideline-based prenatal care. Am J Obstet Gynecol 2022.
The outcomes included in the sensitivity analysis were limited to those with timeliness definitions depended on the claims-based algorithm used to estimate the date of LMP. The STI panel was not included because we did not impose a gestational age requirement for this service. The anatomy scan was not included because it cannot be considered at a 18–22 week anatomy scan if it is outside of that recommended gestational age range. The obstetrical laboratory panel was not included because it is considered timely if it is billed any time before the patient’s second prenatal visit; this definition does not depend on the gestational age at the time of the visit
For the primary definitions of timely care, urinalysis and urine culture were considered timely if they occurred either in the first 90 days of gestation (first trimester), within 38 days of the first prenatal visit, or within 8 days of the second prenatal visit. The sensitivity analysis only altered the 90 days of gestation criterion because it is the only component dependent on the claims-based algorithm for determining date of LMP.
For the primary definitions of timely care, urinalysis and urine culture were considered timely if they occurred either in the first 90 days of gestation (first trimester), within 38 days of the first prenatal visit, or within 8 days of the second prenatal visit. The sensitivity analysis only altered the 90 days of gestation criterion because it is the only component dependent on the claims-based algorithm for determining date of LMP.
0–90 d
82
0–95 d
83
Glucose test
24 wk+0 d to 28 wk+0 d
80
23 wk+2 d to 28 wk+5 d
87
TDAP vaccination
27 wk+0 d to 36 wk+0 d
63
26 wk+2 d to 36 wk+5 d
65
Group B Streptococcus test
35 wk+0 d to 37 wk+0 d
78
34 wk+2 d to 37 wk+5 d
88
LMP, last menstrual period; STI, sexually transmitted infection; TDAP, tetanus, diphtheria, and pertussis.
Gourevitch et al. Variation in guideline-based prenatal care. Am J Obstet Gynecol 2022.
a The outcomes included in the sensitivity analysis were limited to those with timeliness definitions depended on the claims-based algorithm used to estimate the date of LMP. The STI panel was not included because we did not impose a gestational age requirement for this service. The anatomy scan was not included because it cannot be considered at a 18–22 week anatomy scan if it is outside of that recommended gestational age range. The obstetrical laboratory panel was not included because it is considered timely if it is billed any time before the patient’s second prenatal visit; this definition does not depend on the gestational age at the time of the visit
b For the primary definitions of timely care, urinalysis and urine culture were considered timely if they occurred either in the first 90 days of gestation (first trimester), within 38 days of the first prenatal visit, or within 8 days of the second prenatal visit. The sensitivity analysis only altered the 90 days of gestation criterion because it is the only component dependent on the claims-based algorithm for determining date of LMP.
CBC, complete blood count; CPT, current procedural terminology; HCPCS, Healthcare Common Procedure Coding System; ICD-10, International Classification of Diseases, Tenth Revision; NDC, National Drug Codes; RBC, red blood cells; STI, sexually transmitted infections; TDAP, tetanus, diphtheria, and pertussis.
Gourevitch et al. Variation in guideline-based prenatal care. Am J Obstet Gynecol 2022.
Provider must be an OB/GYN or neonatologist. If a primary care provider was the primary prenatal care provider (provided most of the prenatal care visits), then codes billed by primary care providers were counted
CPT, current procedural terminology; ER, emergency room; E&M, evaluation and management; HCPCS, Healthcare Common Procedure Coding System; ICD-10, International Classification of Diseases, Tenth Revision; OB/GYN, obstetrician or gynecologits.
Gourevitch et al. Variation in guideline-based prenatal care. Am J Obstet Gynecol 2022.
Supplemental Table 7Receipt of individual obstetric laboratory panel components among episodes without all components (N = 17,447)
Obstetrical laboratory panel
Component
Received panel component, timely
Received panel component, anytime
n
%
n
%
D (Rh) type
4220
24
6513
37
RBC antibody screen
3918
22
6819
39
CBC
8116
47
14993
86
Rubella avidity test (IgG)
4272
24
6282
36
There were 17,447 pregnancies (10%) that did not receive all the obstetrical panel components over the course of their pregnancy. The table shows receipt of each component of the obstetrical laboratory panel, with and without the timeliness criteria, among those episodes.
Supplemental Table 8Receipt of individual obstetric laboratory panel components among episodes without all timely components (N = 51,309)
Component
Received panel component, timely
n
%
D (Rh) type
7323
14
RBC antibody screen
3669
7
CBC
10021
20
Rubella avidity test (IgG)
2114
4
There were 51,309 pregnancies (29%) that did receive all the obstetrical panel components during pregnancy but did not receive at least 1 of those components in a timely way. The table shows receipt of each component of the obstetrical laboratory panel, with the timeliness criteria (because all satisfied the anytime criteria), among those pregnancies.
Supplemental Table 9Receipt of individual sexually transmitted infection tests among episodes without all three tests (N = 15,160)
Component
Received panel component
n
%
Syphilis test
7751
51
HIV test
3028
20
Hepatitis B test
6218
41
There were 15,160 pregnancies that did not receive all the components of the STI testing panel (9%). The table shows receipt of each component of the STI testing panel among those episodes.
STI, sexually transmitted infections.
Gourevitch et al. Variation in guideline-based prenatal care. Am J Obstet Gynecol 2022.
Supplemental Table 10Percentage of patients receiving at least 6 of 8 guideline-based services by patient characteristics, both timely and anytime
Characteristics
Percentage of patients receiving at least 6 of the 8 guideline-based services
Timely
Anytime
Percentage
95% CI
Percentage
95% CI
Full sample
70
93
APNCU
Inadequate
58
(57–58)
76
(76–77)
Intermediate
73
(73–74)
94
(94–94)
Adequate
71
(71–72)
96
(95–96)
Adequate plus
69
(68–69)
95
(95–95)
Number of prenatal visits
0–5
64
(63–65)
80
(80–81)
6–8
73
(72–73)
94
(94–94)
9–11
71
(70–71)
95
(95–96)
12–14
70
(69–71)
95
(95–96)
≥15
67
(66–68)
95
(95–96)
Age (y)
12–24
60
(59–61)
88
(87–88)
25–29
71
(70–71)
93
(93–93)
30–34
73
(72–73)
94
(94–94)
35–55
68
(68–69)
93
(93–93)
HDHP
HDHP
71
(71–72)
93
(93–93)
No HDHP
69
(69–69)
93
(92–93)
Obstetric comorbidity index
0
71
(71–71)
92
(92–93)
1–2
69
(69–70)
93
(93–93)
3–6
67
(66–68)
93
(93–94)
≥7
64
(61–66)
93
(92–94)
Rurality
Urban
70
(70–70)
93
(93–93)
Urban-adjacent
66
(65–67)
90
(90–91)
Rural
64
(62–65)
89
(88–90)
Health professional shortage area
Non-HPSA county
74
(73–75)
94
(93–94)
Partial-HPSA county
69
(69–70)
93
(93–93)
Full-HPSA county
65
(63–66)
91
(90–92)
County racial composition
High White Non-Hispanic
72
(71–72)
92
(92–92)
High Black Non-Hispanic
68
(67–68)
92
(92–93)
High Hispanic/Latino
66
(66–67)
92
(92–93)
County median income
Lowest income
65
(65–66)
91
(91–92)
Middle income
69
(69–70)
92
(92–93)
Highest income
74
(74–75)
94
(94–95)
County educational attainment
Lowest education
66
(66–67)
92
(92–92)
Middle education
69
(68–69)
93
(92–93)
Highest education
74
(73–74)
93
(93–94)
Percentages and 95% confidence intervals are predicted probabilities calculated from logistic regressions (regression of an indicator for receiving at least 6 of the 8 guideline-based services on each patient characteristic).
APNCU, Adequacy of Prenatal Care Utilization; CI, confidence interval; HDHP, High deductible health plan; HPSA, Health professional shortage area.
Gourevitch et al. Variation in guideline-based prenatal care. Am J Obstet Gynecol 2022.
US Preventive Services Task Force. Douglas K.O, Karina WD, HK, et al. Screening for asymptomatic bacteriuria in adults: us preventive services task force recommendation statement. JAMA 2019;322:1188–1194.
Phiri K, Clifford CR, Doherty M, Fan Y, Wang F, Seeger JD. Timing of routine prenatal tests relative to the last menstrual period estimated from an ICD-10 based algorithm. Poster presented at the: 35th International Conference on Pharmacoepidemiology and Therapeutic Risk Management; August 2019; Philadelphia, PA.
American Academy of Pediatrics. In: American College of Obstetricians and Gynecologists, ed. Guidelines for perinatal care. 7th ed. American Academy of Pediatrics; American College of Obstetricians and Gynecologists, Elk Grove Village, IL : Washington, DC; 2012.
N.T.S. reports receiving grants as the principal investigator from Harvard Pilgrim Healthcare, the George Kaiser Family Foundation , Peterson Center on Healthcare, Rx Foundation, CRICO, the Rita and Alex Hillman Foundation, Yellow Chair Foundation; Gordon and Betty Moore Foundation , and Merch for Mothers. N.T.S. also reports receiving royalties from McGraw-Hill and serving on the boards of March for Moms, Costs of Care, Maven Clinic, Diana Clinic, and the National Institutes of Health Office of Research on Women’s Health. The other authors have no conflict of interest to report.
This study was supported by a Health Data for Action grant from the Robert Wood Johnson Foundation . The funder was not involved in the study design, data analysis or interpretation, writing of the manuscript, or manuscript submission.
Cite this article as: Gourevitch RA, Natwick T, Chaisson CE, et al. Variation in guideline-based prenatal care in a commercially insured population. Am J Obstet Gynecol 2022;226:413.e1-19.
01088-7&id=gr1.jpg){kind=link}
01088-7&id=gr2.jpg){kind=link}
01088-7&id=gr3.jpg){kind=link}
01088-7&id=fx1.jpg){kind=link}