Introduction
CHD is the most common birth defect, affecting nearly 1% of births in the United States. 1,Reference Hoffman and Kaplan2 As of 2010, two-thirds of the CHD population comprised adults, many of whom require follow-up with a cardiologist. Reference Marelli, Ionescu-Ittu, Mackie, Guo, Dendukuri and Kaouache3 With the advancement of medical and surgical management for CHD, the life expectancy for this chronic disease has lengthened, increasing the paediatric and adult patient populations, and therefore, their utilisation of the health care system. Reference Downing, Nembhard and Rose4,Reference Sachdeva, Valente and Armstrong5 In their population-level analysis, Downing et al. found the survival percentage to 35 years of age to be 95.3% for non-severe CHDs after the first year of life. Reference Downing, Nembhard and Rose4
There is known variation in outpatient care for mildly complex CHD. Reference Downing, Nembhard and Rose4,Reference Ziebell, Ghaleb, Anderson and Statile6–Reference Uzark8 In addition, care for CHD is expensive, for both hospital systems and families. Reference Chowdhury, Johnson and Baker-Smith9 The variation in practice patterns increases resource utilisation and the already heavy cost burden incurred by patients and families. Reference Ziebell, Ghaleb, Anderson and Statile6,Reference Weiland10,Reference Connor, Kline, Mott, Harris and Jenkins11 Ziebell et al. illustrated in their charge analysis that nearly $240,000 could have been saved if variation in practice was limited to the recommendations agreed upon by a local consensus. Reference Ziebell, Ghaleb, Anderson and Statile6 Prior implementation of standardised clinical practice algorithms improve care delivery and outcomes in outpatient care management cardiac conditions, Reference Afshar, Hogan and Conturie12–Reference Angoff, Kane and Giddins14 as well as CHD. Reference Porras, Brown and Rathod15,Reference Gongwer, Gauvreau, Huh, Sztam and Jenkins16 These algorithms display expected outcomes along a sequence of steps for a given diagnosis and as a result limit excess resource utilisation and reduce the cost burden incurred by the healthcare system. Reference Uzark8,Reference Khan, Siow and Lewis17
The American College of Cardiology Adult Congenital and Pediatric Cardiology Quality Working Group created a series of clinical practice algorithms for mildly complex CHD lesions to guide providers in their assessment of patients and prevent overuse of medical facilities. Reference Plummer, Parthiban, Sachdeva, Zaidi and Statile18–Reference Birnbaum, Hahn and Sheth22 The purpose of this study was to describe the current practice patterns of outpatient care among CHD patients relative to the American College of Cardiology clinical practice algorithms prior to their implementation at the Cincinnati Children’s Heart Institute. We aimed to characterise departures from these published algorithms and describe how these deviations impact resource utilisation. This study will serve as a baseline to inform future quality improvement initiatives to improve adherence to practice algorithms and decrease excess resource utilisation.
Materials and methods
Study design and population
We performed a retrospective chart review at Cincinnati Children’s Heart Institute. The local institutional review board reviewed and approved this study prior to data collection. The sample included an electronic medical record review of the most recent 100 outpatient encounters for patients with CHD for the following five lesions: atrial septal defect, patent ductus arteriosus, valvar pulmonary stenosis, aortic coarctation, and ventricular septal defect. A sample size of n = 100 encounters per lesion was selected because it provided a margin of error of 10% (e.g., 95% confidence interval of 40 to 60%) when the sample percentage adherent to the guidelines is 50%. This review included visits that (1) met the International Classification of Diseases, Tenth Revision diagnostic requirements and (2) were primarily for the congenital heart lesion. Because the purpose of the study was to track provider-level patterns, patients who met the inclusion criteria were not excluded if they had multiple encounters within the data collection period.
Encounters were separated into two categories: initial and follow-up. Initial encounters were recorded as the outpatient visit when or after the patient was diagnosed with a lesion, whereas follow-up encounters included all other visits to clinic. Patients were excluded if they were (1) pregnant, (2) had multiple cardiac lesions, or (3) were being seen for another cardiac condition.
Data collection
Study data were collected from the electronic medical record and managed using REDCap electronic data capture tools hosted at Cincinnati Children’s. Reference Harris, Taylor and Minor23,Reference Harris, Taylor, Thielke, Payne, Gonzalez and Conde24 The database was organised based upon the branching logic of the CHD algorithms. Demographic information was recorded for all encounters, prior to branching into the appropriate algorithm based on the patient’s lesion, age, and encounter type. Relevant testing data included electrocardiogram, echocardiogram, ambulatory blood pressure monitoring, MRI, and CT. The recorded scheduling data included the date of the last visit, the date of the current visit, and the recommendation made by the provider for timing of the next visit.
The requirements for departure from testing and scheduling recommendations were decided prior to the onset of data collection. We recorded relevant testing data within one month of the visit. If the recommended tests were completed, the encounter was marked as adhering to the algorithm and appropriate utilisation. If too many tests were completed with respect to the lesion algorithm, the encounter was marked as a departure and categorised as an overutilisation of resources, whereas if too few tests were completed, the encounter was marked as a departure and categorised as an underutilisation of resources.
With respect to scheduling data, encounters were determined to depart from the recommendations based on buffer zones established prior to data collection. Cardiologists familiar with the Appropriate Use Criteria defined the buffer zones listed below based on the recommendations within the CHD algorithms. Reference Sachdeva, Valente and Armstrong5
-
• If the recommendation was a 1-month follow-up, schedule the follow-up visit between 2 and 6 weeks.
-
• If the recommendation was a 6-month follow-up, schedule the follow-up visit between 4 and 9 months.
-
• If the recommendation was to return to clinic between two and three years, schedule the follow-up visit between 18 months and 3.5 years.
-
• If the recommendation was to follow-up at a 1-year-old birthday, schedule the follow-up visit between 9 and 18 months old.
-
• If the recommendation was to return to clinic between 3 and 5 years of age, schedule the follow-up visit between 3 years old and before turning 6 years old.
If the visit was scheduled within the buffer zone, the encounter was marked as adhering to the algorithm and appropriate utilisation. If the visit was scheduled before the start of the buffer zone, the visit was marked as a departure and an overutilisation of resources (i.e., returning “too soon”). If the visit was scheduled after the end of the buffer zone, the encounter was marked as a departure and an underutilisation of resources (i.e., returning “too late”).
Key outcomes
The primary outcome of this study was the proportion of patients who departed from the published algorithm. Practice patterns for each patient encounter were classified as either adhering to or departing (i.e., yes/no) from the American College of Cardiology clinical practice algorithms. Reference Plummer, Parthiban, Sachdeva, Zaidi and Statile18–Reference Birnbaum, Hahn and Sheth22 Adherence to the algorithm was quantified in three domains: (1) timing of the current appointment (for follow-up visits), (2) testing obtained related to the visit, and (3) follow-up plan after the visit. Each of the lesion-specific algorithms defines the appropriate timing of the visit (for follow-up appointments) based on the duration from the prior visit, the appropriate testing for the visit, and the appropriate follow-up plan after the current visit. All clinical algorithms can be found in the attached supplementary materials (Supplementary Figure S1, Supplementary Figure S2, Supplementary Figure S3, Supplementary Figure S4, Supplementary Figure S5).
Analysis
All statistical analyses were performed using R Version 4.2.2. Patient demographics for each encounter (sex, lesion type, and lesion repair status), visit characteristics (encounter type), and clinical outcomes (adherence, utilisation) were described using frequencies with percents. Patient age was described using median and interquartile range. Descriptive analyses were conducted for all encounters and for encounter subtypes defined by lesion and by encounter type (e.g., initial and follow-up). Only descriptive statistics were used; no inferential statistics were planned for this analysis.
Results
The study included 500 encounters, 150 of which were new visits. Encounters occurred between April 2022 and December 2022 in the Heart Institute at Cincinnati Children’s Hospital Medical Center. Patients ranged from 3 days old to nearly 72 years, with a median age of 2.16 years (interquartile range 0.40–9.98 years), and they were cared for by 36 cardiologists. Table 1 describes the demographics of the study population.
Demographics information for whole sample by lesion type
Note: ASD = atrial septal defect; PDA = patent ductus arteriosus; PS = valvar pulmonary stenosis; Coarc = aortic coarctation; VSD = ventricular septal defect.
Encounter characteristics
Figure 1 shows the proportion of visits that adhered to each domain, broken down by the initial and follow-up visit cohorts. Overall, 68% (338) of all visits adhered to testing recommendations and 63% (314) of all visits adhered to the follow-up plan recommendations. There were disparities noted by encounter type. A total of 47% (71) of initial visits departed from follow-up plan recommendations set forth by the American College of Cardiology, whereas only 33% (115) of follow-up visits had the same result. However, when examining testing recommendations, only 12% (18) of initial visits departed and 41% (144) of follow-up. In the follow-up visit cohort, 55% (194) of visits adhered to the recommendations for timing of the current visit.
Summary of outcomes by encounter type.
Follow-up recommendation adherence
There are observable differences in the deviation patterns of follow-up plan recommendations by encounter type. There was a higher rate of adherence among follow-up visits, where 67% (235) adhered to follow-up plan recommendations, and only 53% (79) of initial visits adhered to follow-up plan recommendations. Among the departures, a total of 39% (58) of initial visits had indicated follow-up plans that propose follow-up too soon relative to the algorithm recommendation. For follow-up visits, only 23% (80) of encounters had the same result. A total of 10% (35) of providers at follow-up visits suggested that patients return to clinic after the recommended time, and this occurred in 9% (13) of initial visits.
Testing recommendation adherence
Testing utilisation data for this sample by lesion and encounter type is shown in Table 2. Most lesion and encounter types demonstrated a higher proportion of adherence to testing recommendations, rather than departure. However, 70% (61) of the follow-up visits for coarctation departed from testing recommendations, with 98.3% (60) of these visits not recommending enough testing. In contrast, 62% (8) of coarctation initial visits adhered to testing recommendations, re-emphasising the disparity between initial and follow-up adherence patterns. Other lesion specific information can be found in Table 2.
Testing utilisation by lesion and encounter type
Note: ASD = atrial septal defect; PDA = patent ductus arteriosus; PS = valvar pulmonary stenosis; Coarc = aortic coarctation; VSD = ventricular septal defect.
For encounters that departed from algorithm-specific recommendations, utilisation trends varied based upon the visit type. Higher rates of underutilisation were observed in initial visits for patients with atrial septal defect (100%; 4), patent ductus arteriosus (75%; 3), coarctation (80%; 4), and ventricular septal defect (67%; 2), where the provider did not perform enough tests (e.g. echocardiogram, ambulatory blood pressure monitoring, or no testing at all). Follow-up visits exhibited a higher proportion of overutilisation in patients who had atrial septal defect (79%; 15), patent ductus arteriosus (84%; 26), and ventricular septal defect (95%; 20), where the provider administered too many tests relative to algorithm recommendations. Any valvar pulmonary stenosis visit that departed from the algorithm (initial: 14% (2), follow-up: 14% (12)) saw fewer tests than the algorithm recommended, regardless of appointment type.
Discussion
This study illustrates variation in outpatient care delivery for patients with simple CHD at a single centre. There was variation in testing, follow-up recommendations, and return to clinic across several types of CHD. These findings are similar to other studies that have showed variation in care across the spectrum of management of CHD. Reference Ziebell, Ghaleb, Anderson and Statile6,Reference Burstein, Rossi and Jacobs7 This study also attempted to quantify the over- or underutilisation of care for the population.
Care delivered in this study demonstrated both over- and underutilisation of clinical time and testing. Overutilisation of testing, and the resultant over-treatment, is a well-studied phenomenon in adult and paediatric medicine. Reference Zhi, Ding, Theisen-Toupal, Whelan and Arnaout25–Reference Muskens, Kool, van Dulmen and Westert28 The causal factors of overutilisation are complex; however, there are roots in the patient-provider interaction and communication strategies. Reference Kool, van Dulmen, Patey and Grimshaw29 This study demonstrated a significant overutilisation of both clinical time—as measured by early follow-up visits—and testing for patients returning for follow-up visits when compared to published guidelines. Not only does overutilisation add unnecessary cost to the system, but overuse of clinical resources and testing reduces space for patients who appropriately need to be seen. In 2020, the American Heart Association issued an advocacy statement for increasing access to care for patients with CHD in rural and medically underserved areas, especially to help identify CHD lesions early in development. Reference Chowdhury, Johnson and Baker-Smith9 Work on eliminating overutilisation or unnecessary care will be important for improving access generally.
There was underutilisation of testing in a subset of our population. Healthcare services are critical for populations with CHD, as these patients require stable follow-up care for optimal outcomes. Reference Chowdhury, Johnson and Baker-Smith9 Specifically, there was a large majority of coarctation patients that did not receive enough recommended testing. This was primarily due to the departure from the algorithm recommendation for ambulatory blood pressure monitoring in patients ages 1 to 18. Reference Makhoul, Markush and Baker-Smith21 In May 2022, the American Heart Association released a statement recommending the standard use of ambulatory blood pressure monitoring in paediatric patients, which was a marked departure from previous recommendations. Reference Flynn, Urbina and Brady30 Studies on implementation science indicate guideline uptake is facilitated by leadership support and financial incentives which may end up being important as we work on improving compliance with these new guidelines. Reference Shanbhag, Graham and Harlos31
There are limitations to this study due to its design and analysis. This study is retrospective, which hinders any ability to extract data that is not already recorded in the electronic medical record. Additionally, the study period was before most CHD algorithms were published, and none had been distributed by Cincinnati Children’s Hospital leadership. However, this does not necessarily prevent individual cardiologists from knowledge of the algorithms. Finally, because this study was conducted at a single centre, these patients do not represent the entire population of patients with CHD. We also did not look at how these deviations impacted the outcomes of care, and without this information, it is hard to understand how returning too late or not testing enough affects patients and their families.
This data represent the current state of practice at Cincinnati Children’s prior to the implementation of the American College of Cardiology algorithms. Clear variations in care exist across these mildly complex lesions, increasing the cost of care delivery for all parties. Implementation of these clinical practice algorithms will impact over- and underutilisation of care delivery for these patients.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1047951125001246.
Acknowledgements
We would like to acknowledge all members of Cincinnati Children’s Hospital, especially the Heart Institute and the Office of Population Health, who helped with this study.
Financial support
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Competing interests
None.
Ethical standards
Ethical approval was waived by the local institutional review board of Cincinnati Children’s Hospital Medical Center in view of the retrospective nature of the study, and all the procedures and tests performed were part of the routine care.
Open access
