Highlights
-
•
Adults with CHD at tertiary centers have significant gaps in recommended ACHD care.
-
•
Nonsevere CHD, nonphysiologic class conditions, and age 40+ have higher care gaps.
-
•
Patients with gaps in specialist visits have higher gaps in recommended testing.
-
•
COVID-19 pandemic has worsened ACHD care gaps.
-
•
Enhancing interdisciplinary collaborations at tertiary centers can improve ACHD care.
ABSTRACT
Background
Guidelines recommend lifelong care with adult congenital heart disease (ACHD) specialists for adults with congenital heart disease (CHD). However, such gaps in visits at specialized ACHD centers have not been well-characterized from diverse US settings.
Methods
This retrospective study analyzed data from 12 centers in the national Patient-Centered Clinical Research Network. CHD conditions were classified using International Classification of Disease codes and a hierarchical algorithm. ACHD specialists were identified by investigators and encounter volumes. Data from the ‘Pre-COVID’ (2015-2019) and ‘COVID’ (2020-2022) periods were analyzed separately. Main outcome measures were: 1) Gaps in any ACHD specialist visit and recommended testing throughout the study period. 2) Gaps in recommended ACHD follow-up care.
Results
During pre-COVID ( N = 18,934) and COVID ( N = 22,453) periods, between 55.3%-55.8% were males, 27.2%-31.0% were 40+ years, 18.2%-19.6% had severe CHD, and 52.7%-55.0% had CHD physiologic class B-D conditions. Between 47.0%-54.5% had gaps in specialist visit and 13.0%-24.6% had gaps in all the testing. Patients with gaps in specialist visits were 6.33-9.44 times more likely to have gaps in testing. Gaps were more common among patients with moderate (adjusted odds ratio [AOR]: 2.61) and simple (AOR: 2.84) CHD, those aged 40+ (AOR: 1.53) and nonphysiologic class conditions (AOR 1.51). In both periods, 64.1%-71.5% of patients had gaps in follow-up care.
Conclusions
Three-quarters of adults with CHD experienced gaps in specialized ACHD care while receiving services at high volume comprehensive tertiary health centers. To address these gaps, interventions such as fostering physician collaboration within tertiary centers might be needed, and targeted to patients with less severe CHD, nonphysiologic class conditions, and those aged 40+ years.
Background
Advancements in care for pediatric congenital heart disease (CHD) has resulted in a rapidly growing population of adult CHD (ACHD). One driver of continued survival of adults with CHD is receiving lifelong regular care at specialized ACHD centers. Consequently, current American Heart Association/American College of Cardiology (AHA/ACC) guidelines recommend the frequency of ACHD specialist visits and testing specified by CHD anatomy and physiologic class. Surveys and single center data have reported that up to 85% adults with CHD do not receive ACHD specialist care. ,,, However, to date, there are no estimates of the burden of gaps in recommended ACHD specialist visits and testing at diverse United States (US) health centers, especially after considering the anatomic-physiologic classification of CHD. This limits the ability to design, advocate for, and implement strategies to reduce ACHD care gaps and improve outcomes.
Multiple factors have been identified in the literature to be associated with gaps in ACHD care. These include disease complexity, period during transition from pediatrics to adult care, distance from the specialist ACHD center, as well as socioeconomic factors such as race and rurality. ,,,, However, these data are either limited to non-US centers, single US centers, and/or use subjective data in the form of patient-reported surveys. Also, while certain patient characteristics for care gaps have been described, associations of disease characteristics such as physiologic class and non-CHD related conditions with gaps in care have not. A single-center study has further demonstrated the exacerbation of gaps in care by the COVID-19 pandemic in patients with Fontan palliation, but such data remain poorly described for other CHD lesions and do not include multiple centers.
Therefore, we used objective data from 12 tertiary centers encompassing a large, socioeconomically diverse cohort across multiple regions within the US. We determined the prevalence and predictors of gaps in guideline-recommended care in Adults with CHD with varying anatomical complexity, physiologic class, and medical comorbidities. The study period encompasses 5 years before (1/1/2015-12/31/2019) and 3 years from the start of the COVID-19 pandemic (1/2020-12/2022).
Methods
Setting and data source
This is an observational study that was approved by the University of Utah Institutional Review Board (IRB #00144050) and no informed consent was required. Retrospective data was extracted from 12 health systems participating in the National Patient-Centered Clinical Research Network (PCORnet). PCORnet is a network of participating health centers across the country that share data in a common database allowing access to a plethora of longitudinal patient information and standardizing the data to a common format. Data are grouped as demographics, vital status, insurance status, diagnoses, encounter and provider characteristics, primary care, specialty, and other service (outpatient, inpatient, and emergency room) utilization at institutions within Network Partners.
Identifying study cohort
A query program was developed for each participating site and health plan to run on their instance of the PCORnet Common Data Model (CDM), version 6.1. Patients were identified as having CHD by International Classification of Diseases, Ninth Revision (ICD-9) and Tenth Revision (ICD-10) codes using a modified version of a previously published algorithm that we have previously described (Supplementary 1). ,, Briefly, the CHD ICD code present only during pregnancy-related encounters was excluded to avoid assigning a fetal diagnosis of CHD to the mother. Patients were considered to have CHD if an ICD-9 or ICD-10 code for CHD was present for ≥1 inpatient encounter or ≥2 outpatient encounters at least 30 days apart. We excluded patients with nonspecific CHD ICD codes as well as those >45 years with bicuspid aortic valve but without associated codes indicating valve repair or replacement. ,, We also excluded those deceased and pregnancy-related encounters. Based on the 2018 AHA/ACC guidelines for the management of ACHD, we classified the remaining cohort into 3 anatomic categories as 1) Severe CHD: single ventricle, hypoplastic left heart syndrome, transposition of great arteries, and truncus arteriosus, 2) Moderate CHD: Tetralogy of Fallot, endocardial cushion defect, coarctation of the aorta, anomalies of the pulmonary artery/ valve, anomalies of the tricuspid valve, subaortic stenosis, anomalies of the aortic valve, and 3) Simple CHD: ventricular septal defect, patent ductus arteriosus or aortopulmonary window, anomalous pulmonary venous return.
For analytic purposes, the study period was categorized into the pre-COVID period (1/1/2015-12/31/2019) and the COVID period (1/1/2020-12/31/2022). We only included patients who were “part of the health system” and those ≥18 and <65 years of age at the start of each study period (Supplementary 1).
Identifying outpatient ACHD specialist visits and tests
PCORnet encounter variables of ambulatory visits and telehealth were used to identify outpatient office visits. If a patient had multiple outpatient encounters on the same date, it was considered 1 outpatient encounter. ACHD specialists were identified using a multistep iterative process that included ACHA’s provider directory, cardiology taxonomy codes, participating site Principal Investigators input, and ACHD encounter volumes (Supplementary 2). We used the diagnostic and procedural codes (Supplementary 3, Table S2) to identify if a patient completed the guideline-recommended testing for Adults with CHD. , These included: electrocardiogram (ECG), echocardiogram (Echo), computed tomography (CT)/ magnetic resonance imaging (MRI), ambulatory rhythm monitoring, and stress test.
Identifying covariates
Demographic variables such as age, sex, race, preferred language, insurance, and zip code were ascertained (Supplementary 4). Age was determined at the start of each study period. Race was categorized as Black or African American, White, and Others. Language was classified as English, Spanish and other. We determined insurance and 5 or 9-digit zip code using information in the patient’s clinical record that was closest to the middle of the study period. We used the zip code data to extract the rural residence, ,, neighborhood poverty, and distance from the primary ACHD center.
In addition to anatomic classification, the AHA/ACC guidelines categorize CHD complexity based on the severity of its hemodynamic consequences (such as NYHA Class, cyanosis, or end-organ dysfunction secondary to CHD) into physiological classes A to D. Class A include patients with no symptoms or signs or no residual hemodynamic abnormality while classes B, C, and D include patients with mild, moderate or severe symptoms or functional limitations respectively. Since there are no ICD codes for some of the variables that are used to describe a physiological classification (e.g. NYHA class), we used ICD codes to identify medical conditions that might be associated with the CHD physiologic class. For this, we used a combination of Elixhauser and nonelixhauser comorbidity measures, and procedural codes for cardiac valve replacements (Supplementary 4, Table S3). ,,, We, then, classified these conditions into the following: (1) Conditions likely associated with CHD physiologic class B-D (hereby referred as ‘Physiologic class B-D conditions’): renal disease, chronic pulmonary disease, liver disease, arrhythmia, congestive heart failure, pulmonary vascular disease, pulmonary hypertension, any valve replacement, infective endocarditis, Eisenmenger syndrome or cyanosis; (2) Conditions less likely associated with any CHD physiologic class (hereby referred to as ‘Nonphysiologic class conditions’): hypertension, hypercholesterolemia, coronary artery disease, peripheral arterial disease, stroke, diabetes mellitus, obesity, neurologic disorder, thyroid disorders, peptic ulcer, AIDS, any tumor, rheumatoid arthritis/collagen vascular disease, coagulopathy, weight loss, fluid and electrolyte disorders, anemia, substance abuse, mental health disorder, psychoses, and deep venous thrombosis/ pulmonary embolism; and (3) Physiologic class A conditions: Since ACC/AHA defines physiologic class A as CHD patients with no significant physiologic limitations, we categorized those with no physiologic class B-D conditions as physiologic class A category.
Defining gaps in care
For this study, we defined gaps in any ACHD specialist visit and gaps in recommended testing as lack of a single outpatient visit with an ACHD specialist or ACHD-related testing, respectively, throughout the study duration (2015-2019 for pre-COVID and 2020-2022 for COVID). We also defined “gaps in ACHD follow-up care” using the ACC/AHA recommended follow-up frequency for various CHD anatomic-physiologic class as follows: >1 year gap in ACHD specialist visit for severe physiologic class B-D patients or >2 years gap for severe physiologic class A and moderate or simple physiologic class B-D patients, or > 3 years gap for moderate physiologic class A. We only included patients who were ‘part of the health system’ for at least a year longer than their respective recommended care interval. For example, we only included severe physiologic class B-D patients if they were part of the health system for at least 2 years, severe physiologic class A if they were part of the health system for at least 3 years, and so on. We excluded simple physiologic class A patients since their recommended follow-up interval ranges up to 5 years, limiting our ability to have a 6-year follow-up period required for these patients during the pre-COVID period. Similarly, we excluded simple physiologic class A and moderate physiologic class A patients in the COVID period since it only spanned 3 years.
Statistical analysis
We present continuous variables as median (25th, 75th percentile) or mean (± standard deviation [SD]), and categorical variables as frequency (percentages). We used chi-square, student t-test or Wilcoxon rank sum test as appropriate. We used logistic mixed models to determine the unadjusted and adjusted association between gaps in ACHD specialist visit and gaps in testing, with health center as the random effect. Similarly, we conducted random-effects logistic regression to estimate odds ratio of CHD complexity and age on gaps in ACHD specialist visit. We conducted within stratum-analyses by CHD severity and/or physiologic class as appropriate. All models were adjusted for the covariates, health system, and number of years within the health system, as appropriate. For all adjusted analyses, we accounted for the missing data in the demographic variables using multiple imputation. Based on the maximum fraction of missing information of all models, we imputed 37 datasets for pre-COVID and 25 for COVID period datasets. We used fully conditional specification method, and for imputation model we used discriminant function for categorical variables and predictive mean matching for continuous variables. All analyses were conducted for the pre-COVID and COVID periods separately using SAS version 9.4. A 2-sided t-test with P <.05 was considered significant.
Results
Study population
After applying the inclusion/exclusion criteria, 18,934 CHD patients (18.2% severe, 65.8% moderate, and 16.0% simple CHD), were included in the analysis for the pre-COVID and 22,453 for the COVID period ( Figure 1 ). Patients with severe disease were younger, more likely to be uninsured or self-insured, live in rural areas, travel longer distance to ACHD centers, and have physiologic class B-D status ( Table 1 ). Similar findings were observed in the COVID period (Supplementary 5, Table S4).
Study population. ICD = international classification of diseases; CHD = congenital heart disease.
Table 1
Cohort characteristics (2015-2019).
| Variables | All patients ( n = 18,934) |
Simple CHD
( n = 3,031) |
Moderate CHD ( n = 12,457) | Severe CHD ( n = 3,446) | P – value |
|---|---|---|---|---|---|
| Age category, years | |||||
| 18-21 | 3,410 (18.0%) | 493 (16.3%) | 2,137 (17.2%) | 780 (22.6%) |
<.0001 |
| 22-26 | 3,447 (18.2%) | 497 (16.4%) | 2,135 (17.1%) | 815 (23.7%) | |
| 27-40 | 6,200 (32.7%) | 862 (28.4%) | 4,022 (32.3%) | 1,316 (38.2%) | |
| 40+ | 5,877 (31.0%) | 1,179 (38.9%) | 4,163 (33.4%) | 535 (15.5%) | |
| Median (Q1, Q3) age, years | 32 (23, 44) | 34 (24, 49) | 33 (23, 45) | 27 (22, 35) | <.0001 |
| Sex: Male | 10,470 (55.3%) | 1,373 (45.3%) | 7,143 (57.4%) | 1,954 (56.7%) | <.0001 |
| Race | |||||
| Black or African American | 1,492 (9.4%) | 385 (15.7%) | 819 (7.8%) | 288 (9.6%) | <.0001 |
| White | 13,438 (84.3%) | 1,818 (74.1%) | 9,062 (86.3%) | 2,558 (85.5%) | |
| Other | 1,010 (6.3%) | 249 (10.2%) | 616 (5.9%) | 145 (4.8%) | |
| Missing | 2994 | 579 | 1960 | 455 | |
| Hispanic | 3,004 (17.7%) | 568 (21.2%) | 1,995 (18.0%) | 441 (13.8%) | <.0001 |
| Missing | 1958 | 348 | 1361 | 249 | |
| Preferred language | |||||
| English | 11,108 (97.2%) | 1,571 (95.0%) | 7,081 (97.7%) | 2,456 (97.5%) |
<.0001 |
| Spanish | 213 (1.9%) | 50 (3.0%) | 122 (1.7%) | 41 (1.6%) | |
| Other | 103 (0.9%) | 32 (1.9%) | 48 (0.7%) | 23 (0.9%) | |
| Missing | 7510 | 1378 | 5206 | 926 | |
| Insurance status | |||||
| Government/ public | 3,695 (29.7%) | 672 (33.4%) | 2,328 (28.4%) | 695 (30.9%) | <.0001 |
| Private | 6,760 (54.2%) | 1,054 (52.3%) | 4,579 (55.9%) | 1,127 (50.1%) | |
| Self/ uninsured/ others | 2,007 (16.1%) | 288 (14.3%) | 1,291 (15.7%) | 428 (19.0%) | |
| Missing | 6,472 | 1,017 | 4,259 | 1,196 | |
| Rural residence | 1,276 (7.1%) | 179 (6.2%) | 808 (6.8%) | 289 (8.8%) | <.0001 |
| Missing | 853 | 137 | 560 | 156 | |
| High neighborhood poverty | 3,942 (22.0%) | 726 (25.3%) | 2,523 (21.4%) | 693 (21.3%) | <.0001 |
| Missing | 1003 | 165 | 653 | 185 | |
| Distance from primary health center ≥20 miles | 9,336 (51.6%) | 1,324 (45.7%) | 6,002 (50.4%) | 2,010 (61.1%) | <.0001 |
| Missing | 852 | 137 | 559 | 156 | |
| Medical conditions | |||||
| Physiologic B-D conditions | 10,407 (55.0%) | 1,509 (49.8%) | 6,569 (52.7%) | 2,329 (67.6%) | <.0001 |
| Nonphysiologic conditions | 11,848 (62.6%) | 1,938 (63.9%) | 7,969 (64.0%) | 1,941 (56.3%) | <.0001 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
