Frequency of Pregnancy Testing among Adolescent Emergency Department Visits

Abstract

Objective

To estimate the frequency of pregnancy testing among adolescent emergency department (ED) patients and to determine factors associated with testing.

Methods

This was a retrospective cross-sectional study using the National Hospital Ambulatory Medical Care Survey (NHAMCS) from 2005-2009 of ED visits by females ages 14 to 21 years. We estimated the frequency of pregnancy testing among all visits, those for potential reproductive health complaints, and those associated with exposure to potentially teratogenic radiation. Multivariable logistic regression modeling was performed to calculate adjusted probabilities and odds ratios with 95% confidence intervals to evaluate factors associated with pregnancy testing by patient characteristics.

Results

We identified 11,531 visits, representing an estimated 41.0 million female adolescent ED visits. Of these, 20.9% (95% CI 19.3%, 22.5%) included pregnancy testing. Among visits for potential reproductive health complaints and those associated with exposure to potentially teratogenic radiation, 44.5% (95% CI 41.3%, 47.8%) and 36.7% (95% CI 32.5%, 40.9%), respectively, included pregnancy testing. Among the entire study population, we found statistically significant differences in pregnancy by testing (p<0.001 for all) by age, race/ethnicity, hospital admission, and geographic region.

Conclusions

A minority of female adolescent ED visits included pregnancy testing even if patients presented with potential reproductive health complaints or received exposure to ionizing radiation. Small but statistically significant differences in pregnancy testing rates were noted based on age, race/ethnicity, ED disposition, and geographic region. Future studies should focus on designing quality improvement interventions to increase pregnancy testing in adolescent ED patients, especially among those in whom pregnancy complications or the risk of potentially teratogenic radiation exposure is higher.

Introduction

Almost one million teenagers in the United States become pregnant annually,¹ of which more than 80% of pregnancies are unintended.² The United States continues to have one of the highest adolescent pregnancy rates of all industrialized countries,³ with underserved youth and those with poor access to health care dis proportionately represented.^{2, 4, 5} Young females have high rates of emergency department (ED) use, comprising 15% of all annual ED visits,⁶ and high rates of unintended pregnancy have been found among women utilizing EDs;⁵

The diagnosis or exclusion of pregnancy and its complications is critical during ED evaluation of many females of child-bearing age, but particularly among adolescent patients.⁷ Adolescents often present late in their first trimester for pregnancy diagnosis,⁸ and earlier pregnancy detection may help reduce some of the adverse sequelae associated with adolescent pregnancies, such as pre-eclampsia, prematurity, and fetal and maternal death.⁹ Furthermore, because some diagnostic imaging and therapies provided in the ED are teratogenic to a developing fetus, it is crucial to consider pregnancy.

Given the importance of detecting pregnancies during the ED evaluation of adolescent females, it is necessary to understand current pregnancy testing patterns. Although single center studies have found that the diagnosis of pregnancy in adolescents is often missed on initial ED evaluation,^{7, 10} national figures have yet to be estimated. Therefore, we sought to measure the frequency of pregnancy testing among a nationally representative sample of adolescent female ED visits and to determine factors associated pregnancy testing.

Methods

Study Design

We conducted a retrospective cross-sectional analysis of the National Hospital Ambulatory Medical Care Survey (NHAMCS) from 2005-2009. This study was determined to be exempt from formal review by our hospital's Institutional Review Board because of the use of de-identified, publicly available data.

Data Source and Study Population

The NHAMCS, a multi-stage national probability sample survey of hospital EDs (excluding federal, military, and Veterans Affairs hospitals) conducted by the Centers for Disease Control and Prevention, National Center for Health Statistics, is conducted during a randomly assigned 4-week data period annually. This design involves selection of geographic primary sampling units, hospitals within primary sampling units, EDs within hospitals, and patient visits within EDs, with the patient visit being the basic sampling unit. Each patient visit represents a larger number of visits. A weight is assigned to each observation and allows for the generation of nationally representative estimates.¹¹

The eligible study population included all ED visits by females between the ages of 14 and 21 years from 2005-2009. Sub-group analysis was also performed to evaluate the frequency of pregnancy testing among visits for potential reproductive health complaints as well as for visits with exposure to potentially teratogenic radiation. Potential reproductive health complaints were defined as visits with lower abdominal, pelvic, or genital pain; vaginal discharge or bleeding; or dysuria listed as one of the three reasons for visit collected by the NHAMCS for each patient visit. Furthermore, exposure to potentially teratogenic radiation was defined as patient visits associated with performance of a computed tomography scan (CT). NHAMCS does not distinguish type of radiographs performed and therefore, we were unable to determine whether patient visits were also associated with radiographs of the abdomen or pelvis.

Outcome Measures

Our primary outcome variable was performance of a pregnancy test. Covariates of interest included patient age, race/ethnicity, insurance status, disposition, geographic region, and care at a pediatric versus general ED. Patient age was categorized as 14-17 years and 18-21 years. With regards to insurance status, we compared private versus non-private insurance. Non-private insurance was comprised of the following categories: Medicare, Medicaid, Self-Pay, No charge/Charity. Geographic regions were defined as by the US Census Bureau, including the Northeast, South, Midwest, and West. EDs were categorized as pediatric if >75% of ED patient volume was comprised of patients <21 years old.

Data Analysis

We used descriptive statistics with appropriate survey weighting to calculate frequency of pregnancy testing. To identify factors that might be associated with pregnancy test performance, we first considered simple associations with these candidate factors and testing. Then we performed multivariate logistic regression to estimate associations with pregnancy test performance after adjusting for other factors. In this exploratory investigation that sought to adjust for confounding, variables with a p-value <0.10 in any of the bivariate analyses and variables which were a priori felt to have a potential relationship with pregnancy testing were retained in our multivariate model. A goodness of fit test applicable to survey data, the F-adjusted mean residual test,¹² was used to provide an estimate of model performance. Estimates were derived from the multivariate model including adjusted odds ratios (AOR) with 95% confidence intervals (95% CI) and standardized adjusted probabilities of receipt of pregnancy test. The survey sample suite of programs implemented in Stata 12.1 (College Station, TX; 2012) was used for regression analysis to account for the complex design of the NHAMCS survey. We estimated standardized probabilities of testing by means of predictive margins as implemented in the “margins” command in Stata.¹³

Results

During 2005-2009, 11,531 records were identified; representing an estimated 41.0 million female adolescent ED visits. Of these, 20.9% (95% CI 19.3%, 22.5%) or 8.6 million included a pregnancy test. Of the estimated 10.1 million ED visits for potential reproductive health complaints, only 44.5% (95% CI 41.3%, 47.8%) were associated with pregnancy testing. Additionally, among the estimated 3.7 million patient visits with exposure to potentially teratogenic radiation, only 36.7% (95% CI 32.5%, 40.9%) included pregnancy testing. Table 1 provides the frequency of pregnancy testing by age, race, insurance status, geographic region, and ED disposition.

Table 1.

Characteristics of adolescent female ED visits and frequency of pregnancy testing.

Demographic Characteristics		Actual # in Unweighted Sample	Weighted* Proportion of Visits	Weighted Proportion with Pregnancy Testing (95% CI)
Total Visits		11,531	41.0 million	20.9% (19.3, 22.5)
Age	Age 18-21 years	7,253	62.3%	22.6% (20.8, 24.4)
	Age 14-17 years	4,278	37.7%	18.1% (16.2, 20.1)
Race/Ethnicity	Non-Hispanic White	6,224	59.3%	19.6% (17.7, 21.5)
	Non-Hispanic Black/African American	3,151	23.9%	23.2% (21.0, 25.4)
	Hispanic	1,833	14.7%	20.9% (18.4, 23.5)
	Other	317	2.1%	27.2% (16.8, 37.7)
Insurance Status	Private	3,933	35.8%	19.5% (17.5, 21.6)
	Non-Private	7,598	64.2%	21.7% (19.9, 23.5)
Disposition	Hospitalized	591	4.0%	22.2% (17.7, 26.9)
	Discharged	10,940	96.0%	20.9% (19.3, 22.4)
Geographic Region	Northeast	2,858	17.9%	14.9% (12.1, 17.6)
	Midwest	2,493	23.7%	24.4% (21.0, 27.8)
	South	4,258	41.9%	22.2% (19.4, 24.9)
	West	1,922	16.4%	19.3% (16.4, 22.3)
ED type	General ED	11,117	96.8%	21.1 (19.5, 22.7)
	Pediatric ED	414	3.2%	15.6 (10.0, 21.2)

^*National population proportion estimates may not equal unweighted proportions because of survey sampling methodology.

In bivariate analyses, we found associations of pregnancy testing by age category (p<0.001), racial/ethnic group (p=0.003), insurance status (p=0.04), and geographic region (p<0.001), and ED type (0.09). In our subgroup analysis of visits for potential reproductive health complaints, analyses revealed statistically significant associations of pregnancy testing with racial/ethnic group (p=0.03), ED disposition (p<0.001), geographic region (p<0.001), ED type (p=0.04). In our subgroup analysis of visits with exposure to potentially teratogenic radiation, pregnancy testing was associated with ED disposition (p=0.04) only.

Table 2 provides the standardized probabilities for pregnancy testing among all patient visits and the two subgroups, patient visits for potential reproductive health complaints and patient visits associated with receipt of potentially teratogenic radiation. Table 3 provides the adjusted odds ratios for pregnancy testing by variable. Among all patient visits, we found that visits made by patients 14-17 years old (AOR 0.79, 95% CI 0.70, 0.89) compared to 18-21 year olds and Non-Hispanic White patients (AOR 0.85, 95% CI 0.73, 0.98) compared to Non-Hispanic Black/African American were less likely to undergo pregnancy testing. Furthermore, we also found geographic differences in that the Northeast (AOR 0.53, 95% CI 0.40, 0.71) and Western states (AOR 0.72, 95% CI 0.56, 0.92) were less likely to perform pregnancy testing when compared to the Midwest region. Insurance status, ED disposition, and ED type were not associated with pregnancy testing in the multivariate model. Among patient visits for potential reproductive health complaints, visits made by Hispanic patients (AOR 0.71, 95% CI 0.52, 0.98) as compared to visits by Non-Hispanic Black patients and visits resulting in hospital admission (AOR 0.35, 95% CI 0.23, 0.54) were less likely to be associated with pregnancy testing. With regards to regional differences, visits in the Northeast were less likely to be associated with pregnancy testing as compared to the Midwest (AOR 0.56, 05% CI 0.38, 0.82). Among visits associated with exposure to potentially teratogenic radiation, the only factor associated with pregnancy testing was ED disposition, with visits resulting in hospital admission (AOR 0.52, 95% CI 0.27, 0.97) less likely to be associated with pregnancy testing than visits resulting in ED discharge.

Table 2. Standardized Probabilities of Pregnancy Testing.

Variable		All Visits (20.9% Tested)	Potential Reproductive Health Visits (44.5% Tested)	Visits with Exposure to Potentially Teratogenic Radiation (36.7% Tested)
Age category	Age 18-21	0.22 (0.21, 0.24)	0.46 (0.41, 0.51)	0.38 (0.33, 0.43)
	Age 14-17	0.19 (0.17, 0.20)	0.44 (0.40, 0.47)	0.36 (0.30, 0.42)
Race/Ethnicity	White	0.20 (0.18, 0.22)	0.44 (0.40, 0.49)	0.38 (0.32, 0.43)
	Black/ African American	0.22 (0.20, 0.24)	0.47 (0.43, 0.52)	0.34 (0.26, 0.43)
	Hispanic	0.23 (0.20, 0.25)	0.40 (0.34, 0.46)	0.37 (0.27, 0.48)
Insurance status	Private	0.20 (0.18, 0.22)	0.44 (0.40, 0.49)	0.36 (0.30, 0.42)
	Non-private	0.21 (0.20, 0.23)	0.45 (0.41, 0.48)	0.38 (0.33, 0.43)
ED disposition	Admit	0.23 (0.18, 0.27)	0.23 (0.15, 0.31)	0.24 (0.13, 0.35)
	Discharge	0.21 (0.19, 0.22)	0.46 (0.42, 0.49)	0.38 (0.33, 0.42)
Geographic Region	Northeast	0.15 (0.12, 0.18)	0.36 (0.29, 0.43)	0.35 (0.25, 0.45)
	Midwest	0.25 (0.21, 0.28)	0.50 (0.44, 0.56)	0.44 (0.37, 0.52)
	South	0.22 (0.19, 0.25)	0.45 (0.40, 0.51)	0.36 (0.29, 0.42)
	West	0.19 (0.16, 0.22)	0.44 (0.37, 0.51)	0.33 (0.22, 0.44)
ED type	General ED	0.21 (0.19, 0.23)	0.45 (0.42, 0.48)	0.38 (0.33, 0.42)
	Pediatric ED	0.15 (0.09, 0.21)	0.26 (0.09, 0.42)	0.19 (0.02, 0.39)

Model Goodness of Fit: F-test=0.379, p=0.945

Table 3. Adjusted Odds Ratios of Pregnancy Testing.

Variable		All Visits (20.9% Tested)	Potential Reproductive Health Visits (44.5% Tested)	Visits with Exposure to Potentially Teratogenic Radiation (36.7% Tested)
Age Category	Age 18-21	Ref	Ref	Ref
	Age 14-17	0.79 (0.70, 0.89)	1.1 (0.88, 1.4)	0.93 (0.68, 1.3)
Race/Ethnicity	Black/ African American	Ref	Ref	Ref
	White	0.85 (0.73, 0.98)	0.88 (0.68, 1.1)	1.2 (0.71, 1.9)
	Hispanic	0.99 (0.84, 1.2)	0.71 (0.52, 0.98)	1.1 (0.64, 2.1)
Insurance Status	Non-Private	Ref	Ref	Ref
	Private	0.94 (0.82, 1.1)	0.98 (0.79, 1.2)	0.94 (0.67, 1.3)
ED Disposition	Discharge	Ref	Ref	Ref
	Admission	1.1 (0.85, 1.4)	0.35 (0.23, 0.54)	0.52 (0.27, 0.97)
Geographic Region	Midwest	Ref	Ref	Ref
	Northeast	0.53 (0.40, 0.71)	0.56 (0.38, 0.82)	0.67 (0.39, 1.1)
	South	0.87 (0.68, 1.1)	0.83 (0.59, 1.1)	0.70 (0.46, 1.1)
	West	0.72 (0.56, 0.92)	0.79 (0.53, 1.2)	0.63 (0.35, 1.1)
ED Type	General ED	Ref	Ref	Ref
	Pediatric ED	0.66 (0.42, 1.0)	0.42 (0.17, 1.0)	0.37 (0.09, 1.5)

Model Goodness of Fit: F-test=0.379, p=0.945

Discussion

In this nationally representative cross-sectional analysis of ED visits by adolescent females, we found low rates of pregnancy testing overall, even among patients presenting with potential reproductive health complaints or among those with exposure to potentially teratogenic radiation. The diagnosis of pregnancy and its potential complications are of utmost importance in the ED evaluation of any woman of child-bearing age. Emergency physicians play an important role in the timely recognition and referral of pregnancy among teenagers, especially because EDs are disproportionately utilized by young women.¹⁴ Furthermore, Todd and colleagues found that one-third of adolescent ED females are at risk for unintended pregnancy.⁵ As an ED visit might be an adolescent patient's only access to care, failure to diagnose pregnancy could result in serious delay of receipt of obstetric services, thereby increasing the risk for preventable morbidity.

Complications of pregnancy must be considered in any woman of childbearing age with abdominal pain or vaginal bleeding, and as such, pregnancy testing for patients presenting with these complaints is considered standard of care in management algorithms. Although pregnant adolescents can present to the ED with a variety of nonspecific complaints,¹⁵ abdominal pain and vaginal bleeding are among the most common reasons for female adolescent ED visits.¹⁴ In our study, we found that even among patients who presented with potential reproductive health complaints, fewer than half were tested for pregnancy. This finding is concerning given previous studies demonstrating the high frequency (40%) of presentations with abdominal or genitourinary complaints among adolescents eventually diagnosed as pregnant and whose pregnancy was initially missed in the ED.⁷ Givens and colleagues found that among adolescents diagnosed with pregnancy in the ED, 80% presented with gastrointestinal or genitourinary complaints, but only 8% of the patients requested a pregnancy test or mentioned the possibility of pregnancy to the triage nurse.¹⁶ Furthermore, adolescents who are at risk of pregnancy might also be in denial about their risk; 10% of pregnant teenagers denied sexually activity.¹⁶ Additionally, Ramoska and colleagues found that 11.5% of female ED patients who stated that there was no chance they could be pregnant were found to have positive pregnancy tests, and over 15% of patients who did not think they were pregnant were found to have positive pregnancy tests.¹⁷ Therefore, the treating provider should suspect pregnancy when female adolescent patients present with these complaints.

Excluding pregnancy is also essential before ordering radiologic examinations that can potentially introduce teratogenic radiation to a developing fetus. We found that among patients visits associated with exposure to potentially teratogenic radiation (e.g. those undergoing CT scan), less than 40% were evaluated for pregnancy. Furthermore, our finding might be overestimating the true frequency of pregnancy testing among patients exposed to potentially teratogenic radiation due to the differential nature of radiologic imaging coding in the data abstraction forms used by the NHAMCS.

We identified certain patient demographics associated with receipt of a pregnancy test. For example, we found that visits by younger adolescents were less likely to be associated with pregnancy testing. However, pregnancy risk must be considered even in younger patients, as almost 30% of 9^th grade female students reported being sexually active.¹⁸ Furthermore, we found racial differences in pregnancy testing rates, as visits made by White patients were less likely to be associated with pregnancy testing than visits by Non-Hispanic Black/African American or Hispanic patients. Although pregnancy rates are higher among Black/African American and Hispanic adolescents compared to White adolescents, pregnancy rates among White adolescents are not negligible.¹ Therefore, performance of pregnancy testing should not be influenced by a patient's racial or ethnic background. Similarly, geographic differences were found in the performance of pregnancy testing, as the Northeast region was the least likely to perform pregnancy testing. The reason for such a geographic discrepancy in pregnancy testing rates is unclear and unrelated to teenage pregnancy rates, which are lowest in the Midwest and highest in the South.¹⁹ Among ED visits for potential reproductive health complaints and exposure to potentially teratogenic radiation, patients who were hospitalized were less likely to undergo pregnancy testing. This may be because they had other identified causes for their symptoms, and consequently, ruling out pregnancy was not considered. However, the noted differences in pregnancy testing by these mentioned factors in our study were small. Therefore, it appears that the decision to perform pregnancy tests may not be largely driven by patient demographics, disposition, or geographic location as pregnancy testing rates were universally low.

Our study has several important limitations. First, pregnancy testing may have been under-coded in the overall survey. However, the data miscoding rate among NHAMCS surveys is less than 1%¹¹ and beginning in 2005, data abstractors were specifically trained on pregnancy test data abstraction and given a list of acronyms, including “HCG”. Furthermore, to validate our findings, we also estimated pregnancy testing rates using the Nationwide Emergency Department Sample (NEDS) of the Healthcare Cost and Utilization Project, a family of health care databases sponsored by the Agency for Healthcare Research and Quality,²⁰ and found similar rates of pregnancy testing (unpublished data, available on request). Although we found relatively low rates of pregnancy testing, what the optimal rate should be is difficult to determine. In some cases, patients and physicians may have already been aware of pregnancy or may have been referred from a clinic where testing was already performed and thus, repeat or confirmatory testing, may not have been necessary. For some patients with potential reproductive health complaints, an alternative diagnosis to pregnancy that accounts for symptoms may be or may become readily apparent. Furthermore, provider knowledge of sexual activity or contraceptive use is unknown. However, other studies have found positive pregnancy tests even among patients who denied sexual activity.^{16, 17} Moreover, given the nature by which radiographs were coded in NHAMCS, we were unable to determine which visits had exposure to abdominal or pelvic radiographs. Therefore, our results do not account for visits with such radiation exposure and we may have overestimated pregnancy testing frequency among those exposed to potentially teratogenic radiation. Additionally, we cannot determine the timing of pregnancy testing among the patients who had a radiologic examination. As such, it is possible that patients were exposed to potentially teratogenic radiation prior to receipt of a pregnancy test, making our finding of a 37% frequency of pregnancy testing among this population an overestimate.

Conclusion

In summary, we found that a minority of adolescent females presenting to the ED had pregnancy testing, even when presenting with potential reproductive health complaints or when receiving potentially teratogenic radiologic examinations. Differences in pregnancy testing were noted based on age, race, geographic region, and ED disposition. Future studies should focus on designing quality improvement interventions to increase pregnancy testing in adolescent ED patients, especially among those in whom pregnancy complications or the risk of potentially teratogenic radiation exposure is higher.