P U L M O N A RY R E H A B I L I TAT I O N
Impact of Pulmonary Rehabilitation on Hospitalizations for Chronic Obstructive Pulmonary Disease Among Members of an Integrated Health Care System Huong Q. Nguyen, PhD, RN; Annie Harrington, MD; In-Lu Amy Liu, MS; Janet S. Lee, MS; Michael K. Gould, MD, MS
■ PURPOSE: The evidence regarding the effects of pulmonary rehabilitation (PR) on health care resource use remains limited. This retrospective study evaluated the effects of PR on the primary outcome of all-cause hospitalizations and secondary outcomes of other health care use, exercise capacity, health-related quality of life (HRQOL), and body weight in patients with chronic obstructive pulmonary disease (COPD) in a large integrated health care system. ■ METHODS: The PR cohort included 558 patients with a COPD diagnosis, age ≥ 40 years, who were treated with a bronchodilator or steroid inhaler, participated in 1 of 13 PR programs between January 1, 2008, and August 1, 2013, and were continuously enrolled in the health plan ≥ 12 months prior to and after PR. Two non-PR control cohorts were assembled for comparison. Data were extracted from electronic health records. The 6-minute walk test and St. George’s Respiratory Questionnaire results were available for a subset. ■ RESULTS: The proportion of patients who were hospitalized 12 months post-PR was lower compared with the 12 months prior (37% vs 45%, P = .001) while emergency department use was not different (52% vs 54%). Patients who declined PR for logistical reasons had a 40% higher risk of hospitalization than PR participants (relative risk = 1.40, 95% CI: 0.96-2.06, P = .08). There were significant improvements in the 6-minute walk test distance (+43 m) and the St. George’s Respiratory Questionnaire total score (−9.6 points) but minimal changes in weight. ■ CONCLUSIONS: Our finding that participation in PR is associated with reductions in hospitalizations corroborates previous studies. A notable strength of this study is the capture of complete utilization data.
Pulmonary rehabilitation (PR) has been a guidelinerecommended therapy for patients with moderate to severe chronic obstructive pulmonary disease (COPD) since 2001 and was deemed a Medicare-covered benefit in 2010.1 There is unequivocal evidence that PR
K E Y
W O R D S
chronic obstructive pulmonary disease health care resource utilization hospitalizations pulmonary rehabilitation
Author Affiliations: Department of Research and Evaluation (Drs Nguyen and Gould and Mss Liu and Lee), and Pulmonary and Critical Care, Kaiser Permanente Southern California (Dr Harrington), Pasadena. A portion of these findings was presented in the abstract form at the 2014 American Thoracic Society International Conference in San Diego, California. No conflicts of interest were reported. Correspondence: Huong Q. Nguyen, PhD, RN, Department of Research and Evaluation, Kaiser Permanente Southern California, 100 S. Los Robles, Pasadena, CA 91101 ([email protected]). DOI: 10.1097/HCR.0000000000000128
improves symptoms, health-related quality of life (HRQOL), and exercise capacity in patients with stable COPD.1,2 However, data on the effects of PR on health care resource use from randomized clinical trials are more limited, partly because of the sample size
356 / Journal of Cardiopulmonary Rehabilitation and Prevention 2015;35:356-366
www.jcrpjournal.com
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
JCRP-D-14-00128_LR 356
12/08/15 3:59 PM
requirements needed for stable and precise estimates.3,4 The few published studies examining the association between PR participation and hospitalizations have necessarily relied on observational data.5-10 Moreover, most of these studies use a single-group pre- and postdesign, which potentially is limited by regression to the mean effects and secular trends. Finally, ascertainment of health care resource use has typically relied on patient self-report since the majority of studies are not embedded within integrated health systems that have complete capture of utilization data, and thus these data can be subjected to under- or overreporting. Therefore, the goal of this article is to extend the current evidence base on the effects of PR on the primary outcome of all-cause hospitalizations and secondary outcomes of other health care resource use, exercise capacity, HRQOL, and weight using data from patients with COPD who are members of a large integrated health delivery system in southern California.
METHODS Cohort Definition This was a retrospective cohort study conducted at Kaiser Permanente Southern California. Human subjects approval was obtained from the Kaiser Permanente Southern California institutional review board. Analysis of administrative and claims data was conducted for patients with a COPD diagnosis (International Classification of Disease Ninth Revision [ICD-9] codes of 491, 492, and 496), age 40 years and older, treated with a bronchodilator or steroid inhaler, and participated in either 1 of our 6 internal PR programs or 7 externally contracted programs (n = 1081) from January 2008 through August 2013. A small percentage of patients (9%) participated in PR more than once. We included only the first PR episode for these patients. The structure and content of the 13 PR programs followed the Centers for Medicare & Medicaid Services guidelines11 and included core PR components such as individualized treatment plans, psychosocial assessments, self-care education and training, and supervised exercise.
Cohort 1 (PR Participants) The subsample of patients in the utilization analysis (n = 558) was composed of those who were continuously enrolled in the health plan for at least 12 months pre- and post-PR and who did not go on to have lung volume reduction surgery or lung transplant. Patients who attended any PR session were included in this cohort. www.jcrpjournal.com
Cohort 2 (Non-PR Controls) Patients with COPD who did not participate in PR were matched on a 2:1 ratio to Cohort 1 PR participants based on age, gender, and FEV1% (forced expiratory volume in the first second of expiration) predicted where available (n = 910) or Charlson Comorbidity Index (n = 204). If there were >2 non-PR controls available, we randomly selected 2 controls out of the available pool or included all controls if there were only 2 possible matches. The index dates were anchored to the date of the respective PR participant’s first PR session.
Cohort 3 (Referred but Declined Non-PR Controls) This cohort included patients who were referred to PR but had documentation in the records that they declined because of transportation barriers, scheduling conflicts, or competing responsibilities. We hypothesized that these patients would not differ appreciably on other unmeasured characteristics from PR participants and would serve as more appropriate controls compared with patients who decline for other reasons related to their health or motivation. The index date for comparison was based on the date they declined PR.
Measurement of Health Care Resources Use All-cause hospitalization was our primary outcome. Secondary outcomes included emergency department (ED) visits, COPD-related hospitalizations and ED visits, and primary and specialty care outpatient visits. Resource use was obtained from clinical, administrative, and claims data in the 12 months prior to the first PR session and in the 12 months after the final PR session.
Measurement of Exercise Capacity, HRQOL, and Body Mass Index Three out of the 6 Kaiser Permanente PR programs had routinely collected and electronically documented 6-minute walk test (6MWT) results and HRQOL data using the St. George’s Respiratory Questionnaire (SGRQ)12 before and after PR whereas the other 3 programs did not collect these outcomes or the data were inaccessible. We did not have access to these data for patients who participated in the external programs. Body weight was obtained from data collected as part of routine clinical care during outpatient visits; the average of all available weight data within 3 and 12 months after the last PR session was used for analysis. Patients in the PR cohort had on average 22 primary and specialty care outpatient visits where weights were measured. Several studies have shown
Impact of Pulmonary Rehabilitation on Hospitalizations for COPD / 357
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
JCRP-D-14-00128_LR 357
12/08/15 3:59 PM
that averaged weight measures derived from electronic medical records have validity and are responsive to changes in treatment.13,14
Covariates Sociodemographic characteristics were obtained from membership files, medication use from pharmacy databases, and oxygen use from durable medical equipment. Education and household incomes were geocoded on the basis of patients’ home addresses at the time of their participation in PR. Spirometry data were available on 69% of the sample. A Charlson Comorbidity Index was calculated on the basis of ICD-9 diagnosis codes from any inpatient or outpatient encounters. We also included the prevalence of heart failure, depression, and anxiety on the basis of ICD-9 codes since these conditions are common in COPD and not calculated in the Charlson Comorbidity Index.
Statistical Analysis Independent t tests and chi-square tests were used to compare baseline characteristics between PR participants versus 2 independent non-PR control cohorts. The McNemar test was used to compare withingroups changes in the percentage of patients who were hospitalized or had an ED visit. Paired t tests were used to compare within-groups changes in the mean number of hospitalizations, ED visits, and outpatient visits and changes in exercise capacity, HRQOL, and weight. Robust Poisson regression models were used to compare the proportion of patients with any cause hospitalizations between the PR cohort and the 2 non-PR control cohorts. Covariates were included in the models if they contributed substantially to model fit or were of clinical importance. Unadjusted generalized estimating equations were used to conduct exploratory subgroup analyses of changes in all-cause hospitalizations based on the median dose of exposure to PR (9 sessions), burden of chronic illness (Charlson Comorbidity Index < or ≥ 3), age < or ≥ 65 years, BMI (normal/underweight vs overweight/obese), and changes in exercise capacity and HRQOL. All significance levels were 2-sided with a P value < .05. Analyses were conducted using SAS 9.2 (SAS Institute Inc, Cary, NC).
RESULTS
zation analysis (Cohort 1) and the non-PR control Cohorts 2 and 3 are summarized in Table 1. There were no appreciable differences in baseline characteristics between the overall PR sample and those included in Cohort 1, subsequently referred to as the PR participants. Pulmonary rehabilitation participants had a mean age of 70 years with comparable numbers of men and women, a majority of whom were white (84%), lived in neighborhoods with higher educational attainment and household incomes >$50 000/ year (62%), a large majority of whom were partnered (58%), receiving public health insurance (74%), and were overweight or obese (67%). Patients had a mean FEV1% predicted of 50 ± 20%, a mean comorbidity score of 3.0 ± 2.3, and a high prevalence of diabetes (21%), depression (21%), and heart failure (16%). Use of health care resources was high for PR participants with approximately half having at least 1 hospitalization or ED visit and a mean of 23 outpatient primary and specialty care visits in the 12 months before starting PR. Patients attended a mean of 13 PR sessions. There were significant differences in the following baseline characteristics between the PR cohort compared with the matched non-PR controls (Cohort 2): marital status, neighborhood income, race/ethnicity, FEV1% predicted, Charlson Comorbidity Index, use of inhalers, systemic corticosteroids, and oxygen and health care resource use. Patients who were referred to PR but declined (Cohort 3) were as expected, more similar to PR participants on most characteristics, but were slightly younger, had higher incomes, used oxygen less frequently and systemic steroids more frequently, and had lower health care resource use.
Effects of PR on Health Care Resources Use Within-Group Comparisons There was a significant decrease in the proportion of patients who had a hospitalization for any cause in the 12 months after starting PR compared with the 12 months before starting PR (45% vs 37%; P = .001). Similarly, there were significant reductions in COPDrelated hospitalizations (27%-18%, P < .001) and ED visits (30% vs 22%, P < .001). Primary care visits declined by 1 visit (P < .001) with a nonsignificant increase in specialty care visits (Table 2). For both non-PR control cohorts, there were increases in the proportion of patients with all-cause hospitalizations in the 12 months post-PR reference index date (Cohort 2) and decision to decline PR (Cohort 3), respectively.
Sample Description
Between-Group Comparisons
The baseline sociodemographic and disease characteristics of the total population of PR patients and the subsample of patients who were included in the utili-
A multivariate model comparing age, gender, and disease-burden in PR participants versus matched non-PR controls (Cohort 2) showed no significant
358 / Journal of Cardiopulmonary Rehabilitation and Prevention 2015;35:356-366
www.jcrpjournal.com
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
JCRP-D-14-00128_LR 358
12/08/15 3:59 PM
T a b l e 1 • Characteristics of Pulmonary Rehabilitation Participants and Nonparticipant Controlsa
Age, y
All PR Participantsb (n = 1081)
Cohort 1: PR Participants With 12-mo F/U (n = 558)
Cohort 2: Matched Non-PR Controls (n = 1114)
69.3 ± 8.5
70.7 ± 8.0
71.2 ± 7.9
Cohort 3: P Valuec P Valuec Declined PR Cohort Cohort Controls (n = 90) 1 vs 2 1 vs 3 68.5 ± 7.5
Gender Male
540 (50)
274 (49)
547 (49)
44 (49)
Female
541 (50)
284 (51)
567 (51)
46 (51)
Partner
605 (56)
325 (58)
599 (54)
51 (57)
No partner
396 (37)
201 (36)
418 (38)
28 (31)
80 (7)
32 (6)
97 (9)
11 (12)
Marital status
Missing Education High school or less
431 (40)
224 (40)
493 (44)
353 (33)
183 (33)
356 (32)
30 (33)
College degree or more
289 (27)
150 (27)
262 (24)
16 (18)
8 (1)
1 (0)
3 (0)
1 (1)
Household median income .99
.98
.05
.07
.20
.16
.001
.01
Impact of Pulmonary Rehabilitation on Hospitalizations for Chronic Obstructive Pulmonary Disease Among Members of an Integrated Health Care System Huong Q. Nguyen, PhD, RN; Annie Harrington, MD; In-Lu Amy Liu, MS; Janet S. Lee, MS; Michael K. Gould, MD, MS
■ PURPOSE: The evidence regarding the effects of pulmonary rehabilitation (PR) on health care resource use remains limited. This retrospective study evaluated the effects of PR on the primary outcome of all-cause hospitalizations and secondary outcomes of other health care use, exercise capacity, health-related quality of life (HRQOL), and body weight in patients with chronic obstructive pulmonary disease (COPD) in a large integrated health care system. ■ METHODS: The PR cohort included 558 patients with a COPD diagnosis, age ≥ 40 years, who were treated with a bronchodilator or steroid inhaler, participated in 1 of 13 PR programs between January 1, 2008, and August 1, 2013, and were continuously enrolled in the health plan ≥ 12 months prior to and after PR. Two non-PR control cohorts were assembled for comparison. Data were extracted from electronic health records. The 6-minute walk test and St. George’s Respiratory Questionnaire results were available for a subset. ■ RESULTS: The proportion of patients who were hospitalized 12 months post-PR was lower compared with the 12 months prior (37% vs 45%, P = .001) while emergency department use was not different (52% vs 54%). Patients who declined PR for logistical reasons had a 40% higher risk of hospitalization than PR participants (relative risk = 1.40, 95% CI: 0.96-2.06, P = .08). There were significant improvements in the 6-minute walk test distance (+43 m) and the St. George’s Respiratory Questionnaire total score (−9.6 points) but minimal changes in weight. ■ CONCLUSIONS: Our finding that participation in PR is associated with reductions in hospitalizations corroborates previous studies. A notable strength of this study is the capture of complete utilization data.
Pulmonary rehabilitation (PR) has been a guidelinerecommended therapy for patients with moderate to severe chronic obstructive pulmonary disease (COPD) since 2001 and was deemed a Medicare-covered benefit in 2010.1 There is unequivocal evidence that PR
K E Y
W O R D S
chronic obstructive pulmonary disease health care resource utilization hospitalizations pulmonary rehabilitation
Author Affiliations: Department of Research and Evaluation (Drs Nguyen and Gould and Mss Liu and Lee), and Pulmonary and Critical Care, Kaiser Permanente Southern California (Dr Harrington), Pasadena. A portion of these findings was presented in the abstract form at the 2014 American Thoracic Society International Conference in San Diego, California. No conflicts of interest were reported. Correspondence: Huong Q. Nguyen, PhD, RN, Department of Research and Evaluation, Kaiser Permanente Southern California, 100 S. Los Robles, Pasadena, CA 91101 ([email protected]). DOI: 10.1097/HCR.0000000000000128
improves symptoms, health-related quality of life (HRQOL), and exercise capacity in patients with stable COPD.1,2 However, data on the effects of PR on health care resource use from randomized clinical trials are more limited, partly because of the sample size
356 / Journal of Cardiopulmonary Rehabilitation and Prevention 2015;35:356-366
www.jcrpjournal.com
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
JCRP-D-14-00128_LR 356
12/08/15 3:59 PM
requirements needed for stable and precise estimates.3,4 The few published studies examining the association between PR participation and hospitalizations have necessarily relied on observational data.5-10 Moreover, most of these studies use a single-group pre- and postdesign, which potentially is limited by regression to the mean effects and secular trends. Finally, ascertainment of health care resource use has typically relied on patient self-report since the majority of studies are not embedded within integrated health systems that have complete capture of utilization data, and thus these data can be subjected to under- or overreporting. Therefore, the goal of this article is to extend the current evidence base on the effects of PR on the primary outcome of all-cause hospitalizations and secondary outcomes of other health care resource use, exercise capacity, HRQOL, and weight using data from patients with COPD who are members of a large integrated health delivery system in southern California.
METHODS Cohort Definition This was a retrospective cohort study conducted at Kaiser Permanente Southern California. Human subjects approval was obtained from the Kaiser Permanente Southern California institutional review board. Analysis of administrative and claims data was conducted for patients with a COPD diagnosis (International Classification of Disease Ninth Revision [ICD-9] codes of 491, 492, and 496), age 40 years and older, treated with a bronchodilator or steroid inhaler, and participated in either 1 of our 6 internal PR programs or 7 externally contracted programs (n = 1081) from January 2008 through August 2013. A small percentage of patients (9%) participated in PR more than once. We included only the first PR episode for these patients. The structure and content of the 13 PR programs followed the Centers for Medicare & Medicaid Services guidelines11 and included core PR components such as individualized treatment plans, psychosocial assessments, self-care education and training, and supervised exercise.
Cohort 1 (PR Participants) The subsample of patients in the utilization analysis (n = 558) was composed of those who were continuously enrolled in the health plan for at least 12 months pre- and post-PR and who did not go on to have lung volume reduction surgery or lung transplant. Patients who attended any PR session were included in this cohort. www.jcrpjournal.com
Cohort 2 (Non-PR Controls) Patients with COPD who did not participate in PR were matched on a 2:1 ratio to Cohort 1 PR participants based on age, gender, and FEV1% (forced expiratory volume in the first second of expiration) predicted where available (n = 910) or Charlson Comorbidity Index (n = 204). If there were >2 non-PR controls available, we randomly selected 2 controls out of the available pool or included all controls if there were only 2 possible matches. The index dates were anchored to the date of the respective PR participant’s first PR session.
Cohort 3 (Referred but Declined Non-PR Controls) This cohort included patients who were referred to PR but had documentation in the records that they declined because of transportation barriers, scheduling conflicts, or competing responsibilities. We hypothesized that these patients would not differ appreciably on other unmeasured characteristics from PR participants and would serve as more appropriate controls compared with patients who decline for other reasons related to their health or motivation. The index date for comparison was based on the date they declined PR.
Measurement of Health Care Resources Use All-cause hospitalization was our primary outcome. Secondary outcomes included emergency department (ED) visits, COPD-related hospitalizations and ED visits, and primary and specialty care outpatient visits. Resource use was obtained from clinical, administrative, and claims data in the 12 months prior to the first PR session and in the 12 months after the final PR session.
Measurement of Exercise Capacity, HRQOL, and Body Mass Index Three out of the 6 Kaiser Permanente PR programs had routinely collected and electronically documented 6-minute walk test (6MWT) results and HRQOL data using the St. George’s Respiratory Questionnaire (SGRQ)12 before and after PR whereas the other 3 programs did not collect these outcomes or the data were inaccessible. We did not have access to these data for patients who participated in the external programs. Body weight was obtained from data collected as part of routine clinical care during outpatient visits; the average of all available weight data within 3 and 12 months after the last PR session was used for analysis. Patients in the PR cohort had on average 22 primary and specialty care outpatient visits where weights were measured. Several studies have shown
Impact of Pulmonary Rehabilitation on Hospitalizations for COPD / 357
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
JCRP-D-14-00128_LR 357
12/08/15 3:59 PM
that averaged weight measures derived from electronic medical records have validity and are responsive to changes in treatment.13,14
Covariates Sociodemographic characteristics were obtained from membership files, medication use from pharmacy databases, and oxygen use from durable medical equipment. Education and household incomes were geocoded on the basis of patients’ home addresses at the time of their participation in PR. Spirometry data were available on 69% of the sample. A Charlson Comorbidity Index was calculated on the basis of ICD-9 diagnosis codes from any inpatient or outpatient encounters. We also included the prevalence of heart failure, depression, and anxiety on the basis of ICD-9 codes since these conditions are common in COPD and not calculated in the Charlson Comorbidity Index.
Statistical Analysis Independent t tests and chi-square tests were used to compare baseline characteristics between PR participants versus 2 independent non-PR control cohorts. The McNemar test was used to compare withingroups changes in the percentage of patients who were hospitalized or had an ED visit. Paired t tests were used to compare within-groups changes in the mean number of hospitalizations, ED visits, and outpatient visits and changes in exercise capacity, HRQOL, and weight. Robust Poisson regression models were used to compare the proportion of patients with any cause hospitalizations between the PR cohort and the 2 non-PR control cohorts. Covariates were included in the models if they contributed substantially to model fit or were of clinical importance. Unadjusted generalized estimating equations were used to conduct exploratory subgroup analyses of changes in all-cause hospitalizations based on the median dose of exposure to PR (9 sessions), burden of chronic illness (Charlson Comorbidity Index < or ≥ 3), age < or ≥ 65 years, BMI (normal/underweight vs overweight/obese), and changes in exercise capacity and HRQOL. All significance levels were 2-sided with a P value < .05. Analyses were conducted using SAS 9.2 (SAS Institute Inc, Cary, NC).
RESULTS
zation analysis (Cohort 1) and the non-PR control Cohorts 2 and 3 are summarized in Table 1. There were no appreciable differences in baseline characteristics between the overall PR sample and those included in Cohort 1, subsequently referred to as the PR participants. Pulmonary rehabilitation participants had a mean age of 70 years with comparable numbers of men and women, a majority of whom were white (84%), lived in neighborhoods with higher educational attainment and household incomes >$50 000/ year (62%), a large majority of whom were partnered (58%), receiving public health insurance (74%), and were overweight or obese (67%). Patients had a mean FEV1% predicted of 50 ± 20%, a mean comorbidity score of 3.0 ± 2.3, and a high prevalence of diabetes (21%), depression (21%), and heart failure (16%). Use of health care resources was high for PR participants with approximately half having at least 1 hospitalization or ED visit and a mean of 23 outpatient primary and specialty care visits in the 12 months before starting PR. Patients attended a mean of 13 PR sessions. There were significant differences in the following baseline characteristics between the PR cohort compared with the matched non-PR controls (Cohort 2): marital status, neighborhood income, race/ethnicity, FEV1% predicted, Charlson Comorbidity Index, use of inhalers, systemic corticosteroids, and oxygen and health care resource use. Patients who were referred to PR but declined (Cohort 3) were as expected, more similar to PR participants on most characteristics, but were slightly younger, had higher incomes, used oxygen less frequently and systemic steroids more frequently, and had lower health care resource use.
Effects of PR on Health Care Resources Use Within-Group Comparisons There was a significant decrease in the proportion of patients who had a hospitalization for any cause in the 12 months after starting PR compared with the 12 months before starting PR (45% vs 37%; P = .001). Similarly, there were significant reductions in COPDrelated hospitalizations (27%-18%, P < .001) and ED visits (30% vs 22%, P < .001). Primary care visits declined by 1 visit (P < .001) with a nonsignificant increase in specialty care visits (Table 2). For both non-PR control cohorts, there were increases in the proportion of patients with all-cause hospitalizations in the 12 months post-PR reference index date (Cohort 2) and decision to decline PR (Cohort 3), respectively.
Sample Description
Between-Group Comparisons
The baseline sociodemographic and disease characteristics of the total population of PR patients and the subsample of patients who were included in the utili-
A multivariate model comparing age, gender, and disease-burden in PR participants versus matched non-PR controls (Cohort 2) showed no significant
358 / Journal of Cardiopulmonary Rehabilitation and Prevention 2015;35:356-366
www.jcrpjournal.com
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
JCRP-D-14-00128_LR 358
12/08/15 3:59 PM
T a b l e 1 • Characteristics of Pulmonary Rehabilitation Participants and Nonparticipant Controlsa
Age, y
All PR Participantsb (n = 1081)
Cohort 1: PR Participants With 12-mo F/U (n = 558)
Cohort 2: Matched Non-PR Controls (n = 1114)
69.3 ± 8.5
70.7 ± 8.0
71.2 ± 7.9
Cohort 3: P Valuec P Valuec Declined PR Cohort Cohort Controls (n = 90) 1 vs 2 1 vs 3 68.5 ± 7.5
Gender Male
540 (50)
274 (49)
547 (49)
44 (49)
Female
541 (50)
284 (51)
567 (51)
46 (51)
Partner
605 (56)
325 (58)
599 (54)
51 (57)
No partner
396 (37)
201 (36)
418 (38)
28 (31)
80 (7)
32 (6)
97 (9)
11 (12)
Marital status
Missing Education High school or less
431 (40)
224 (40)
493 (44)
353 (33)
183 (33)
356 (32)
30 (33)
College degree or more
289 (27)
150 (27)
262 (24)
16 (18)
8 (1)
1 (0)
3 (0)
1 (1)
Household median income .99
.98
.05
.07
.20
.16
.001
.01
