NY-CHAMP Severe Maternal Morbidity Examination

Descriptive Statistics and Data Exploration

Overall Tables

SMM

This table displays a list of all SMM comorbidities.

Characteristic	N = 3,923,3231
SMM_Comorbities
Acute myocardial infarction	120 (0.2%)
Acute renal failure	4,471 (6.4%)
Adult respiratory distress syndrome	2,567 (3.7%)
Amniotic fluid embolism	117 (0.2%)
Aortic aneurysm and dissection	128 (0.2%)
ARAE_CDC	165 (0.2%)
BLOOD_CDC	44,785 (64%)
Cardiac arrest/ventricular fibrillation	140 (0.2%)
Conversion of cardiac rhythm	103 (0.1%)
Disseminated intravascular coagulation	6,549 (9.3%)
Eclampsia	2,713 (3.9%)
EMBOLISM_CDC	1,125 (1.6%)
HEART_LEFT_CDC	1,488 (2.1%)
HEART_SURG_CDC	10 (<0.1%)
HYSTEREC_CDC	1,538 (2.2%)
SCD_CRISIS_CDC	477 (0.7%)
SEPSIS_CDC	1,502 (2.1%)
SHOCK_CDC	1,198 (1.7%)
STROKE_CDC	825 (1.2%)
TRACHEO_CDC	7 (<0.1%)
VENTI_CDC	117 (0.2%)
Unknown	3,853,178
1 n (%)

Twenty one SMM comorbidities are accounted for. There were a total of 70145 cases of SMM that were observed amongst 3,923,323 cases. Within those displaying SMM, 64% had SMM cases related to blood transfusion products, 9.3% had SMM cases related to disseminated intravascular coagulation (a type of severe blood clotting), and 6.4% had SMM cases related to acute renal failure.

Overall Summary Table

This table outputs the distribution of groups withing the overall population.

Characteristic	N = 3,923,3231
Types of Births
Vaginal Births	2,652,033 (68%)
C-Section	1,271,290 (32%)
Race
White	2,103,738 (58%)
Black	588,295 (16%)
Hispanic	682,453 (19%)
Asian or Pacific Islander	223,649 (6.2%)
Other	29,620 (0.8%)
Unknown	295,568
Payer Type
Private Insurance	2,025,771 (53%)
Medicare	25,796 (0.7%)
Medicaid	1,682,003 (44%)
Other	80,717 (2.1%)
Unknown	109,036
AGE	29.0 (25.0, 33.0)
Mortality	208 (<0.1%)
1 n (%); Median (Q1, Q3)

Descriptives of Age

Histogram: Distribution of Ages

Comorbidities by Age

This table displays how common each SMM is in each age group.

Characteristic	20s N = 1,856,1661	30s N = 1,753,5491	40s N = 134,9071	Under 20 N = 178,7011
SMM_Comorbities
Acute myocardial infarction	31 (0.1%)	77 (0.2%)	11 (0.3%)	1 (<0.1%)
Acute renal failure	1,725 (5.6%)	2,246 (7.2%)	333 (8.1%)	167 (4.1%)
Adult respiratory distress syndrome	1,016 (3.3%)	1,224 (3.9%)	204 (4.9%)	123 (3.0%)
Amniotic fluid embolism	36 (0.1%)	60 (0.2%)	16 (0.4%)	5 (0.1%)
Aortic aneurysm and dissection	37 (0.1%)	78 (0.2%)	12 (0.3%)	1 (<0.1%)
ARAE_CDC	85 (0.3%)	65 (0.2%)	6 (0.1%)	9 (0.2%)
BLOOD_CDC	20,380 (67%)	19,067 (61%)	2,416 (58%)	2,922 (72%)
Cardiac arrest/ventricular fibrillation	53 (0.2%)	64 (0.2%)	22 (0.5%)	1 (<0.1%)
Conversion of cardiac rhythm	40 (0.1%)	49 (0.2%)	12 (0.3%)	2 (<0.1%)
Disseminated intravascular coagulation	2,658 (8.7%)	3,264 (10%)	395 (9.6%)	232 (5.7%)
Eclampsia	1,336 (4.4%)	966 (3.1%)	92 (2.2%)	319 (7.9%)
EMBOLISM_CDC	486 (1.6%)	542 (1.7%)	58 (1.4%)	39 (1.0%)
HEART_LEFT_CDC	544 (1.8%)	741 (2.4%)	155 (3.7%)	48 (1.2%)
HEART_SURG_CDC	2 (<0.1%)	8 (<0.1%)	0 (0%)	0 (0%)
HYSTEREC_CDC	362 (1.2%)	982 (3.1%)	194 (4.7%)	0 (0%)
SCD_CRISIS_CDC	281 (0.9%)	159 (0.5%)	6 (0.1%)	31 (0.8%)
SEPSIS_CDC	687 (2.2%)	648 (2.1%)	70 (1.7%)	97 (2.4%)
SHOCK_CDC	440 (1.4%)	640 (2.0%)	83 (2.0%)	35 (0.9%)
STROKE_CDC	307 (1.0%)	450 (1.4%)	46 (1.1%)	22 (0.5%)
TRACHEO_CDC	4 (<0.1%)	3 (<0.1%)	0 (0%)	0 (0%)
VENTI_CDC	46 (0.2%)	62 (0.2%)	5 (0.1%)	4 (<0.1%)
Unknown	1,825,610	1,722,154	130,771	174,643
1 n (%)

Age groups are categorized based on other papers that have chosen to group age by the groups above. To note, the BlOOD_CDC variable (Transfusion of blood products) commonly occurs in all age groups.

Univariate Test by Age

Here we use a univariate Chi-squared test

Characteristic	20s N = 1,856,1661	30s N = 1,753,5491	40s N = 134,9071	Under 20 N = 178,7011	p-value
Race
White	963,665 (56%)	999,761 (62%)	65,622 (53%)	74,690 (45%)
Black	318,973 (18%)	208,703 (13%)	20,276 (16%)	40,343 (24%)
Hispanic	346,209 (20%)	263,571 (16%)	25,841 (21%)	46,832 (28%)
Asian or Pacific Islander	79,384 (4.6%)	130,995 (8.1%)	10,715 (8.7%)	2,555 (1.5%)
Other	16,668 (1.0%)	9,714 (0.6%)	620 (0.5%)	2,618 (1.6%)
Unknown	131,267	140,805	11,833	11,663
Payer Types
Private Insurance	776,263 (43%)	1,131,603 (66%)	83,681 (63%)	34,224 (20%)
Medicare	11,123 (0.6%)	12,915 (0.8%)	1,366 (1.0%)	392 (0.2%)
Medicaid	977,306 (54%)	524,876 (31%)	43,185 (33%)	136,636 (78%)
Other	35,769 (2.0%)	37,623 (2.2%)	3,661 (2.8%)	3,664 (2.1%)
Unknown	55,705	46,532	3,014	3,785
Types of Births
Vaginal Births	1,328,736 (72%)	1,110,918 (63%)	69,142 (51%)	143,237 (80%)
C-Section	527,430 (28%)	642,631 (37%)	65,765 (49%)	35,464 (20%)
1 n (%)

Race, Payment method, and type of birth differs significantly between age groups.

Visual Descriptives of Race

Pay Method by Race

SMM Distribution by Race

Proportion of SMM per Racial Population

Maternal Mortality by Race

Proportion of Mortality per Racial Population

Delivery Type by Race

Univariate Tests by Race

This table displays how the factors of age, mortality, SMM, and birthing type varies by race.

Characteristic	White N = 2,103,7381	Black N = 588,2951	Hispanic N = 682,4531	Asian or Pacific Islander N = 223,6491	Other N = 29,6201	p-value
AGE	30.0 (25.0, 33.0)	28.0 (23.0, 32.0)	28.0 (24.0, 33.0)	31.0 (28.0, 35.0)	27.0 (23.0, 31.0)
Mortality	85 (<0.1%)	51 (<0.1%)	38 (<0.1%)	16 (<0.1%)	1 (<0.1%)
SMM	29,097 (1.4%)	16,767 (2.9%)	13,298 (1.9%)	4,202 (1.9%)	797 (2.7%)
Types of Births
Vaginal Births	1,447,506 (69%)	374,469 (64%)	456,007 (67%)	152,528 (68%)	20,641 (70%)
C-Section	656,232 (31%)	213,826 (36%)	226,446 (33%)	71,121 (32%)	8,979 (30%)
1 Median (Q1, Q3); n (%)

Based on the univariate Kruskal-Wallis (continuous v. categorical), age differs significantly between races. Mortality is missing a p-value because this group is too small. Based on the Chi-squared test (categorical v. categorical), SMM and types of birth differs significantly between races. Although only making up about 15% of the total population, black patients accounted for over 25% of mortality within the patient population. The hispanic patients accounted for about 17% of the total population and accounted for nearly 20% of mortality within the patient population. In contrast, their white counterparts made up over 50% of the population but accounted for less than 45% of mortality within the patient population.

This table displays how payer type varies by race.

Characteristic	White N = 2,103,7381	Black N = 588,2951	Hispanic N = 682,4531	Asian or Pacific Islander N = 223,6491	Other N = 29,6201	p-value
Payer Type
Private Insurance	1,342,838 (66%)	186,572 (32%)	195,599 (29%)	141,669 (65%)	9,007 (32%)
Medicare	12,731 (0.6%)	7,020 (1.2%)	3,797 (0.6%)	528 (0.2%)	213 (0.7%)
Medicaid	653,139 (32%)	370,080 (64%)	439,925 (66%)	72,044 (33%)	18,908 (66%)
Other	28,248 (1.4%)	12,012 (2.1%)	27,448 (4.1%)	4,162 (1.9%)	376 (1.3%)
Unknown	66,782	12,611	15,684	5,246	1,116
1 n (%)

Based on the univariate Chi-squared test, Payment method differs significantly between races.

Question 1: How do rates of SMM and maternal mortality vary by race/ethnicity, controlling for income, payer type, and comorbidities?

Univariate Test by SMM

This table displays how the factors of age, mortality, SMM, and birthing type varies by race.

Characteristic	SMM No N = 3,853,1781	SMM Yes N = 70,1451	p-value
Race
White	2,074,641 (58%)	29,097 (45%)
Black	571,528 (16%)	16,767 (26%)
Hispanic	669,155 (19%)	13,298 (21%)
Asian or Pacific Islander	219,447 (6.2%)	4,202 (6.5%)
Other	28,823 (0.8%)	797 (1.2%)
Unknown	289,584	5,984
Payer Type
Private Insurance	1,995,346 (53%)	30,425 (45%)
Medicare	24,757 (0.7%)	1,039 (1.5%)
Medicaid	1,646,646 (44%)	35,357 (52%)
Other	79,226 (2.1%)	1,491 (2.2%)
Unknown	107,203	1,833
AGE	29.0 (25.0, 33.0)	30.0 (25.0, 34.0)
Types of Births
Vaginal Births	2,624,332 (68%)	27,701 (39%)
C-Section	1,228,846 (32%)	42,444 (61%)
1 n (%); Median (Q1, Q3)

Based on the univariate Chi-squared test, all categories differs significantly between races.

Logistic Regression Model

Here we ran a logistic regression model with SMM as the outcome containing the following predictors: race, age, birthing type, and payer type.

Characteristic	OR	95% CI	p-value
Race_Labels
White	—	—
Black	1.82	1.78, 1.86	<0.001
Hispanic	1.27	1.25, 1.30	<0.001
Asian or Pacific Islander	1.35	1.30, 1.39	<0.001
Other	1.79	1.66, 1.93	<0.001
AGE	1.01	1.01, 1.01	<0.001
VGNL_ALL_Labels
Vaginal Births	—	—
C-Section	3.20	3.14, 3.25	<0.001
Pay_Labels
Private Insurance	—	—
Medicare	2.28	2.13, 2.43	<0.001
Medicaid	1.31	1.29, 1.33	<0.001
Other	1.19	1.13, 1.26	<0.001
Abbreviations: CI = Confidence Interval, OR = Odds Ratio

For age, the odds ratio of 1.01 indicates that there is a slight correlation between an increase in age indicating an increased chance of displaying SMM. With a one unit increase in age, there is a 1% increase for experiencing SMM.

For delivery type, vaginal births have an odds ratio of 0.31, indicating a decrease of 69% for experiencing SMM.

For racial groups, black patients displayed an odds ratio of 1.35, indicating an increased chance of experiencing SMM by 35%. For hispanic patients, the displayed odds ratio of 0.95 indicates a 5% decrease for experiencing SMM. White patients displayed an odds ratio of 0.74, which indicates a 26% decrease for experiencing SMM.

Patients paying with Medicare had an odds ratio of 1.74, indicating an increased chance of 74% for experiencing SMM. Those paying through private insurance displayed an odds ratio of 0.74, indicating a decrease for experiencing SMM by 24%. Patients paying out of pocket or by other means displayed an odds ratio of 0.91, indicating a decreased chance of experiencing SMM by 9%.

Question 2: Do Black-White and Hispanic-White county-level inequality ratios (education, unemployment, incarceration) predict SMM outcomes at the zip code or patient level?

To address this question, we approached the data by first looking at inequality ratios on the zipcode level. At the zipcode level, we built a model for each inequality ratio. In addition, we built three 2x2 models that exam the education, incarceration, and unemployment domains. [Note: For all models at the zipcode level, we used a linear regression since we made the SMM outcome variable a mean value.] Next, we evaluated SMM outcomes at the patient level.

Linear Regression for each Inequality Ratio on Zipcode Level

Below 6 linear regression models are made for the following inequality ratios at the zipcode level: 1) Black to White inequality ratio for education 2) Black to White inequality ratio for incarceration 3) Black to White inequality ratio for unemployment 4) Hispanic to White inequality ration for education 5) Hispanic to White inequality ration for incarceration 6) Hispanic to White inequality ration for unemployment

Black to White Inequality Ratio for Education

Here we ran a linear regression model with SMM as the outcome containing the mean Black to White Inequality Ratio for Education as the predictor.

Characteristic	Beta	95% CI	p-value
mean_EduIneqBW	1.1	0.35, 1.9	0.005
Abbreviation: CI = Confidence Interval

For a one unit increase in the inequality ratio, the expected number of SMM events increases by 1.1 patients per 1000 patients.

Black to White Inequality Ratio for Incarceration

Here we ran a linear regression model with SMM as the outcome containing the mean Black to White Inequality Ratio for Incarceration as the predictor.

Characteristic	Beta	95% CI	p-value
mean_JailIneqBW	-0.07	-0.28, 0.14	0.5
Abbreviation: CI = Confidence Interval

This inequality ratio is not an significant predictor for SMM on the zipcode level.

Black to White Inequality Ratio for Unemployment

Here we ran a linear regression model with SMM as the outcome containing the mean Black to White Inequality Ratio for Unemployment as the predictor.

Characteristic	Beta	95% CI	p-value
mean_UnemIneqBW	0.29	-1.1, 1.7	0.7
Abbreviation: CI = Confidence Interval

This inequality ratio is not an significant predictor for SMM on the zipcode level.

Hispanic to White Inequality Ratio for Education

Here we ran a linear regression model with SMM as the outcome containing the mean Hispanic to White Inequality Ratio for Education as the predictor.

Characteristic	Beta	95% CI	p-value
mean_EduIneqHW	0.48	-0.03, 0.99	0.065
Abbreviation: CI = Confidence Interval

This inequality ratio is not an significant predictor for SMM on the zipcode level.

Hispanic to White Inequality Ratio for Incarceration

Here we ran a linear regression model with SMM as the outcome containing the mean Hispanic to White Inequality Ratio for Incarceration as the predictor.

Characteristic	Beta	95% CI	p-value
mean_JailIneqHW	0.16	-0.46, 0.78	0.6
Abbreviation: CI = Confidence Interval

This inequality ratio is not an significant predictor for SMM on the zipcode level.

Hispanic to White Inequality Ratio for Unemployment

Here we ran a linear regression model with SMM as the outcome containing the mean Hispanic to White Inequality Ratio for Unemployment as the predictor.

Characteristic	Beta	95% CI	p-value
mean_UnemIneqHW	0.23	-1.7, 2.2	0.8
Abbreviation: CI = Confidence Interval

This inequality ratio is not an significant predictor for SMM on the zipcode level.

Overall, these results show that the black to white inequality ratio for education is the only significant independent predictor of SMM.

Linear Regression for Inequality Ratios by education, incarceration, and unemployment on the Zipcode Level

Below 3 linear regression models are made for the inequality ratios at the zipcode level by education, incarceration, unemployment: 1) The education model includes the black to white and hispanic to white education inequality ratio 2) The incarceration model includes the black to white and hispanic to white jail inequality ratio 3) The unemployment model includes the black to white and hispanic to white unemployment inequality ratio

Education Inequality Ratio Model

Here we ran a linear regression model with SMM as the outcome containing the mean Black to White and Hispanic to White Inequality Ratios for Education as the predictors.

Characteristic	Beta	95% CI	p-value
mean_EduIneqBW	1.4	0.16, 2.6	0.027
mean_EduIneqHW	-0.20	-1.0, 0.59	0.6
Abbreviation: CI = Confidence Interval

Given the hispanic education inequality ratio, for a one unit increase in the black/white education inequality ratio, the expected number of SMM events increases by 1.1 patients per 1000 patients. Also, when controlling for the black/white inequality ratio for education, the hispanic ratio is not significant (note that when we look at the hispanic/white education ratio as the only predictor in the “Hispanic to White Inequality Ratio for Education” model, it is almost significant).

Incarceration Inequality Ratio Model

Here we ran a linear regression model with SMM as the outcome containing the mean Black to White and Hispanic to White Inequality Ratios for Incarceration as the predictors.

Characteristic	Beta	95% CI	p-value
mean_JailIneqBW	-0.09	-0.32, 0.13	0.4
mean_JailIneqHW	0.24	-0.41, 0.89	0.5
Abbreviation: CI = Confidence Interval

Neither the black/white incarceration inequality ration nor the hispanic/white inequality ratio is significant in predicting SMM per 1000 patients.

Unemployment Inequality Ratio Model

Here we ran a linear regression model with SMM as the outcome containing the mean Black to White and Hispanic to White Inequality Ratios for Unemployment as the predictors.

Characteristic	Beta	95% CI	p-value
mean_UnemIneqBW	0.28	-1.1, 1.6	0.7
mean_UnemIneqHW	0.20	-1.8, 2.2	0.8
Abbreviation: CI = Confidence Interval

Neither the black/white unemployment inequality ration nor the hispanic/white unemployment ratio is significant in predicting SMM per 1000 patients.

Overall, it seems that education is meaningful while incarceration and unemployment ratios do not seem to matter in SMM outcomes

Logistic Regression for each Inequality Ratio at Patient Level

Black to White Inequality Ratio on Patient Level for Education

Here we ran a logistic regression model with SMM as the outcome containing the following predictors: Black to White Education Inequality Ratio for Education, age, race, birthing type, and payer type.

Characteristic	OR	95% CI	p-value
EDU_LOW_BWR_CNTY	1.03	1.03, 1.03	<0.001
AGE	1.02	1.02, 1.02	<0.001
Race_Labels
White	—	—
Black	1.96	1.92, 2.00	<0.001
Hispanic	1.31	1.28, 1.34	<0.001
Asian or Pacific Islander	1.30	1.26, 1.34	<0.001
Other	1.83	1.70, 1.97	<0.001
Pay_Labels
Private Insurance	—	—
Medicare	2.46	2.30, 2.62	<0.001
Medicaid	1.32	1.29, 1.34	<0.001
Other	1.15	1.09, 1.21	<0.001
Abbreviations: CI = Confidence Interval, OR = Odds Ratio

When controlling for age, delivery type, race, and method of payment, it was found that county-level education inequalities were still significant in influencing black individuals’ and white individuals’ SMM outcome.

Black to White Inequality Ratio on Patient Level for Incarceration

Here we ran a logistic regression model with SMM as the outcome containing the following predictors: Black to White Education Inequality Ratio for Incarceration, age, race, birthing type, and payer type.

Characteristic	OR	95% CI	p-value
JAIL_BWR_CNTY	1.01	1.01, 1.01	<0.001
AGE	1.02	1.02, 1.02	<0.001
Race_Labels
White	—	—
Black	1.89	1.85, 1.93	<0.001
Hispanic	1.27	1.24, 1.30	<0.001
Asian or Pacific Islander	1.30	1.25, 1.34	<0.001
Other	1.83	1.70, 1.98	<0.001
Pay_Labels
Private Insurance	—	—
Medicare	2.51	2.35, 2.69	<0.001
Medicaid	1.31	1.29, 1.34	<0.001
Other	1.15	1.09, 1.22	<0.001
Abbreviations: CI = Confidence Interval, OR = Odds Ratio

When controlling for age, delivery type, race, and method of payment, it was found that county-level incarceration inequalities were still significant in influencing black individuals’ and white individuals’ SMM outcome.

Black to White Inequality Ratio on Patient Level for Unemployement

Here we ran a logistic regression model with SMM as the outcome containing the following predictors: Black to White Education Inequality Ratio for Unemployment, age, race, birthing type, and payer type.

Characteristic	OR	95% CI	p-value
UNEMP_BWR_CNTY	1.03	1.02, 1.04	<0.001
AGE	1.02	1.02, 1.02	<0.001
Race_Labels
White	—	—
Black	1.96	1.92, 2.00	<0.001
Hispanic	1.32	1.29, 1.35	<0.001
Asian or Pacific Islander	1.32	1.28, 1.36	<0.001
Other	1.81	1.68, 1.95	<0.001
Pay_Labels
Private Insurance	—	—
Medicare	2.44	2.29, 2.60	<0.001
Medicaid	1.31	1.29, 1.33	<0.001
Other	1.14	1.07, 1.20	<0.001
Abbreviations: CI = Confidence Interval, OR = Odds Ratio

When controlling for age, delivery type, race, and method of payment, it was found that county-level employment inequalities were still significant in influencing black individuals’ and white individuals’ SMM outcome.

Hispanic to White Inequality Ratio on Patient Level for Education

Here we ran a logistic regression model with SMM as the outcome containing the following predictors: Hispanic to White Education Inequality Ratio for Education, age, race, birthing type, and payer type.

Characteristic	OR	95% CI	p-value
EDU_LOW_HWR_CNTY	1.01	1.01, 1.01	<0.001
AGE	1.02	1.02, 1.02	<0.001
Race_Labels
White	—	—
Black	1.96	1.92, 2.00	<0.001
Hispanic	1.32	1.29, 1.34	<0.001
Asian or Pacific Islander	1.31	1.26, 1.35	<0.001
Other	1.82	1.68, 1.95	<0.001
Pay_Labels
Private Insurance	—	—
Medicare	2.46	2.30, 2.62	<0.001
Medicaid	1.32	1.29, 1.34	<0.001
Other	1.15	1.08, 1.21	<0.001
Abbreviations: CI = Confidence Interval, OR = Odds Ratio

When controlling for age, delivery type, race, and method of payment, it was found that county-level education inequalities were still significant in influencing hispanic individuals’ and white individuals’ SMM outcome.

Hispanic to White Inequality Ratio on Patient Level for Incarceration

Here we ran a logistic regression model with SMM as the outcome containing the following predictors: Hispanic to White Education Inequality Ratio for Incarceration, age, race, birthing type, and payer type.

Characteristic	OR	95% CI	p-value
JAIL_HWR_CNTY	1.02	1.02, 1.03	<0.001
AGE	1.02	1.02, 1.02	<0.001
Race_Labels
White	—	—
Black	1.88	1.84, 1.93	<0.001
Hispanic	1.28	1.25, 1.31	<0.001
Asian or Pacific Islander	1.32	1.27, 1.36	<0.001
Other	1.83	1.70, 1.98	<0.001
Pay_Labels
Private Insurance	—	—
Medicare	2.50	2.34, 2.68	<0.001
Medicaid	1.31	1.29, 1.34	<0.001
Other	1.15	1.08, 1.21	<0.001
Abbreviations: CI = Confidence Interval, OR = Odds Ratio

When controlling for age, delivery type, race, and method of payment, it was found that county-level incarceration inequalities were still significant in influencing hispanic individuals’ and white individuals’ SMM outcome.

Hispanic to White Inequality Ratio on Patient Level for Unemployement

Here we ran a logistic regression model with SMM as the outcome containing the following predictors: Hispanic to White Education Inequality Ratio for Unemployment, age, race, birthing type, and payer type.

Characteristic	OR	95% CI	p-value
UNEMP_HWR_CNTY	1.0	0.98, 1.01	0.5
AGE	1.02	1.02, 1.02	<0.001
Race_Labels
White	—	—
Black	1.97	1.93, 2.01	<0.001
Hispanic	1.32	1.30, 1.35	<0.001
Asian or Pacific Islander	1.32	1.28, 1.37	<0.001
Other	1.81	1.68, 1.95	<0.001
Pay_Labels
Private Insurance	—	—
Medicare	2.44	2.29, 2.60	<0.001
Medicaid	1.31	1.29, 1.33	<0.001
Other	1.14	1.07, 1.20	<0.001
Abbreviations: CI = Confidence Interval, OR = Odds Ratio

When controlling for age, delivery type, race, and method of payment, it was found that county-level employment inequalities were not significant in influencing hispanic individuals’ and white individuals’ SMM outcome.

Question 3: How do urban vs rural disparities contribute to differences in SMM, maternal mortality, and C-section rates?

How Urban versus Rural Disparities contribute to differences in SMM

Here we ran a logistic regression model with SMM as the outcome containing the following predictors: rurality, race, age, birth type, and payer type.

##  num [1:3923323] 3 3 2 3 3 4 2 3 3 4 ...

Characteristic	OR	95% CI	p-value
rurality_label
Suburban	—	—
Rural	1.16	1.12, 1.20	<0.001
Urban	1.06	1.04, 1.08	<0.001
AGE	1.01	1.01, 1.01	<0.001
Race_Labels
White	—	—
Black	1.83	1.79, 1.87	<0.001
Hispanic	1.28	1.25, 1.31	<0.001
Asian or Pacific Islander	1.35	1.31, 1.40	<0.001
Other	1.71	1.56, 1.87	<0.001
VGNL_ALL_Labels
Vaginal Births	—	—
C-Section	3.24	3.18, 3.29	<0.001
Pay_Labels
Private Insurance	—	—
Medicare	2.28	2.12, 2.44	<0.001
Medicaid	1.31	1.29, 1.34	<0.001
Other	1.20	1.13, 1.28	<0.001
Abbreviations: CI = Confidence Interval, OR = Odds Ratio

When examining the impact of one’s rurality status on SMM, it was determined that when using suburban status as the reference group, being from a rural population made one 21% more likely to experience SMM, and being from an urban population made one 4% more likely to experience SMM.

How Urban versus Rural Disparities contribute to differences in maternal mortality

Here we ran a logistic regression model with Mortality as the outcome containing the following predictors: rurality, race, age, birth type, and payer type.

Characteristic	OR	95% CI	p-value
rurality_label
Suburban	—	—
Rural	1.90	1.02, 3.33	0.032
Urban	0.87	0.62, 1.25	0.4
AGE	1.09	1.07, 1.12	<0.001
VGNL_ALL_Labels
Vaginal Births	—	—
C-Section	3.40	2.49, 4.70	<0.001
Race_Labels
White	—	—
Black	1.74	1.17, 2.56	0.006
Hispanic	1.27	0.83, 1.92	0.3
Asian or Pacific Islander	1.82	1.01, 3.08	0.034
Other	1.03	0.06, 4.67	>0.9
Pay_Labels
Private Insurance	—	—
Medicare	8.12	3.57, 16.1	<0.001
Medicaid	2.16	1.54, 3.05	<0.001
Other	1.66	0.50, 4.06	0.3
Abbreviations: CI = Confidence Interval, OR = Odds Ratio

When examining the impact of one’s rurality status on maternal mortality, it was determined that when using suburban status as the reference group, being from a rural population made one 68% more likely to experience mortality, and being from an urban population made one 20% less likely to experience mortality.

How Urban versus Rural Disparities contribute to differences in Csection rates

Here we ran a logistic regression model with C-Section as the outcome containing the following predictors: rurality, race, age, and payer type.

Characteristic	OR	95% CI	p-value
rurality_label
Suburban	—	—
Rural	1.22	1.20, 1.23	<0.001
Urban	1.00	0.99, 1.00	0.6
AGE	1.06	1.06, 1.06	<0.001
Race_Labels
White	—	—
Black	1.38	1.37, 1.39	<0.001
Hispanic	1.20	1.19, 1.20	<0.001
Asian or Pacific Islander	0.97	0.96, 0.98	<0.001
Other	1.04	1.01, 1.07	0.009
Pay_Labels
Private Insurance	—	—
Medicare	1.33	1.30, 1.37	<0.001
Medicaid	1.00	1.00, 1.01	0.074
Other	0.84	0.82, 0.85	<0.001
Abbreviations: CI = Confidence Interval, OR = Odds Ratio

When examining the impact of one’s rurality status on likelihood to have a c-section, it was determined that when using suburban status as the reference group, being from a rural population made one 20% more likely to have a c-section, and being from an urban population displayed no difference in likelihood of having a c-section.