Procedure Code Utilization for Vascular Access Device Placement in the Inpatient Setting: A Retrospective Analysis

Abstract

Vascular access (VA) is essential to inpatient care, and the documentation/coding practices for vascular access device (VAD) placement procedures remain unexplored. Accurate documentation may present benefits for patients, providers, and researchers. A retrospective analysis was performed in adult inpatients (2015 to 2020) using Cerner Real World Data™ to evaluate the utilization of CPT codes for VAD placement/replacement procedures. A total of 14,253,584 patient encounters were analyzed, 0.111 percent (n=15,833) of which received at least one VAD procedure code. Non-tunneled CVC procedures had the highest code rate (0.067 percent), while PIV/midline procedures were the least likely to be coded (0.004 percent). The annual proportion of code utilization increased from 10.9 percent in 2015 to 19.7 percent in 2020 (p<0.0001). Despite widespread use of VADs in the inpatient setting, the procedure coding rate was found to be remarkably low. Appropriate coding/documentation practices may ensure proper care by capturing VA-related patient history, and improve research quality and resource/staff allocation.

Introduction

Approximately 90 percent of all hospitalized patients require vascular access (VA).¹ Millions of vascular access devices (VADs) are placed in the US annually²,³ for the administration of medications, hydration fluids, and blood products as well as blood sampling and hemodynamic monitoring.⁴,⁵

Proper medical coding and documentation are crucial for multiple stakeholders, including patients, providers, and payers. Adequate coding enables collaboration and coordination among providers, which may lead to better patient care during subsequent visits.⁶,⁷ Current Procedural Terminology (CPT) codes allow documenting and accurate billing for medical procedures and professional services.⁸ Furthermore, CPT code data can be utilized for retrospective research since it constitutes a reliable and inexpensive resource for researchers.⁹,¹⁰

Despite the importance of appropriate coding, error rates in medical coding are substantial, and improper CPT code documentation is common.¹¹ Belanger et al. evaluated the utilization patterns of new CPT codes for advance care planning (ACP) visits and reported low code utilization by hospice and palliative medicine specialists, two-thirds of which did not use the relevant codes despite treating patients most likely needing ACP.¹²

VADs are both vital and common in clinical settings, thus the importance of capturing data on their placement procedures for improving patient care, clinical practice, and research. However, medical coding and documentation practices for VA procedures remain unexplored. This retrospective study aims to evaluate the utilization of CPT codes for VAD placement procedures in the inpatient setting.

Methods

Data Source

We conducted this retrospective, observational study using Cerner Real World Data™ (Cerner Corporation, Kansas City, Missouri), a de-identified, HIPPA-compliant, US database of electronic health records (EHR) from 101 clinical facilities and hospital systems. Our analysis of de-identified data was determined to be exempt from local institutional review board (IRB) review in advance by Western IRB (Puyallup, Washington).

Study Population

The study included adult patients who had a documented inpatient encounter between January 1, 2015, and December 31, 2020.

Outcomes and Analyses

We calculated the counts and proportions of inpatient encounters with at least one CPT code corresponding to placement or replacement of a VAD: non-tunneled central venous catheter (CVC; 36556, 36580); peripherally inserted central catheter (PICC; 36573, 36569, 37799, 36586); or midline or peripheral intravenous catheter (PIV; 36410).

Chi-squared and Wilcoxon-Mann-Whitney tests were used to characterize variations in hospital characteristics (academic versus non-academic institutions and bed size comparisons), patient demographics (age, sex, ethnicity, and census region), and Charlson Comorbidity Index (CCI) score. CCI was chosen as a method to capture patient comorbidities based on the diagnosis codes found in the dataset. To capture all relevant patient comorbidities, a longitudinal lookback that contained all diagnosis codes for the study encounters as well as all previous encounters was used to calculate the CCI score. All other characteristics were evaluated only for the study encounters. The Cochran-Armitage test for trends was performed to evaluate six-year trends in VAD procedure code use between 2015 and 2020. All statistical analyses were executed using PySpark 2.4.4. through the Cerner® HealtheDataLab platform/tool and SAS version 9.4 (SAS Institute Inc., Cary, North Carolina) with a significance level set α = 0.05.

Results

Patient Characteristics

A total of 14,253,584 unique patient encounters met the study inclusion criteria. Detailed demographic characteristics are provided in Table 1. Patients who received at least one CPT code for VAD placement/replacement were more likely to be older (62.7 versus 56.5 years, p<0.001), male (49.9 percent versus 42.9 percent, p<0.001), and to have a higher CCI score (3.7 versus 1.9, p<0.001). Most encounters with VAD procedure codes occurred in the Northeast (56.5 percent) US census region, while the South region had the lowest rate (3.9 percent).

Table 1.

Descriptive Statistics Based on VAD Procedure Code Assignment

Inpatient Encounters (n=14,253,584)
Characteristics	Presence of a VAD procedure code (n = 15,833)	Absence of a VAD procedure code (n = 14,237,751)	p value*
Age (in years)
Mean ± SD	62.7 ± 16.4	56.5 ± 20.2	<0.001
Median [25th, 75th]	65 [53, 75]	59 [38, 73]
Min to Max	18 to 89	18 to 89
Sex
Male	7,898 (49.9%)	6,112,442 (42.9%)	<0.001
Female	7,902 (49.9%)	8,084,388 (56.8%)
Other	33 (0.2%)	40,921 (0.3%)
Ethnicity
White	11,134 (70.3%)	10,466,765 (73.5%)	<0.001
Black	3,327 (21.0%)	1,659,675 (11.7%)
Other**	1,372 (8.7%)	2,111,311 (14.8%)
CHARLSON Comorbidity Index
Mean ± SD	3.7 ± 3.1	1.9 ± 2.6	<0.001
Median [25th, 75th]	3 [1, 6]	1 [0, 3]
Min to Max	0 to 20	0 to 29
Hospital Bed Size
<200	468 (3.0%)	219,435 (1.5%)	<0.001
200-299	1,713 (10.8%)	206,442 (1.4%)
300-499	1,173 (7.4%)	1,216,796 (8.5%)
500-999	2,359 (14.9%)	3,921,692 (27.5%)
1000+	9,949 (62.8%)	8,591,035 (60.3%)
Unknown	171 (1.1%)	82,351 (0.6%)
Hospital Type
Academic	1,905 (12.0%)	978,299 (6.9%)	<0.001
Non-Academic	13,928 (88.0%)	13,259,452 (93.1%)
Region***
Northeast	8,949 (56.5%)	3,578,110 (25.1%)	<0.001
South	618 (3.9%)	2,997,084 (21.0%)
Midwest	3,980 (25.1%)	2,820,228 (19.8%)
West	2,286 (14.4%)	4,842,329 (34.1%)

Data are numbers (percentages) unless otherwise indicated.

^*Chi-square test for categorical variables and Wilcoxon-Mann-Whitney test for continuous variables were used.

^**Other includes all non-White and non-Black races (e.g., Asian, Hispanic, Native American).

^****** The first zipcode was used (0-1 = Northeast, 2, 3, 7 = South, 4-6 = Midwest, 8-9 = West).

VAD Procedure Code Utilization

The counts and proportions of inpatient encounters with at least one VAD procedure code by device type are presented in Table 2. Of the total 14,253,584 patient encounters, only 0.111 percent (n=15,833) received at least one VAD procedure code. Non-tunneled CVC procedures were the most likely to be coded (0.067 percent), while PIV/midline procedures had the lowest code rate (0.004 percent).

Table 2.

Counts and Percentages of Encounters with Procedure Codes by VAD Type

Inpatient Encounters (n=14,253,584)
Coded VAD Procedure	Counts	Percentage
Non-tunneled CVC	9,964	0.067%
PICC	5,769	0.040%
PIV/Midline	533	0.004%
Total (Any VAD)	15,833	0.111%

Device breakdown counts do not add up to the total count as the total count included encounters with one or more VAD procedure codes while the breakdown counts included encounters with a specific device procedure code. Since one patient can receive more than one device during a single encounter, the breakdown count is expected to be higher than the total count.

Code Utilization Trends 2015–2020

VAD procedure code utilization trend between 2015 and 2020 is illustrated in Figure 1. The annual proportion of encounters with VAD procedure codes consistently increased from 10.9 percent (2015) to 19.7 percent (2020), except for a minor drop to 18.8 percent in 2019. Overall, the number of reported encounters with VAD procedure codes increased by 80.6 percent between 2015 (n=1,724) and 2020 (n=3,114). The upward trend in VAD code utilization from 2015-2020 was statistically significant based on a Cochran-Armitage trend test (p<0.001).

Figure 1.

Six-Year Trends in the VAD Procedure Code Utilization

Discussion

This analysis has shown a remarkably low utilization of VAD procedure codes since only 0.111 percent of encounters included at least one relevant code despite the prevalent use of VADs in inpatient care. To our knowledge, this is the first study to explore the coding practices regarding VAD procedures. Our results are consistent with previous studies that evaluated CPT code use in other procedures/services and demonstrated low utilization levels.¹³,¹⁴ Several potential reasons may be considered regarding the underutilization of VA-related CPT codes. VAD placement/replacement procedures are commonly performed by nurses,¹⁵ and a professional fee is not reimbursed when the nurses complete these procedures, which may lead to a lack of incentive for documentation. Moreover, the process of assigning CPT codes is complex and labor-intensive, necessitating various resources, including specialized coding personnel.¹⁶ Nurses face a high burden of documentation workload and spend around 25-50 percent of their shifts documenting patient care.¹⁷ Considering the extended work hours, staff shortages, and demanding work conditions experienced by nurses,¹⁸ streamlined and simplified documentation processes should be implemented. This can partially be achieved by eliminating the need of documenting the same procedure multiple times (e.g., patient charts, electronic health records, reimbursement) in the absence of a unified platform.

Despite the high prevalence of peripheral access in inpatient care, CPT codes for PIV/midline placement had the lowest utilization rate. In the US, around 300 million PIVs are placed annually,¹⁹,²⁰ and based on their common occurrence, peripheral catheter placements may be considered minor procedures, which could partially explain poor documentation practices. On the other hand, the relatively higher utilization of central line procedure codes may be due to their use for longer-term therapies, which provide ample time and opportunities for documentation. Furthermore, virtually all CVCs are inserted by physicians, who are eligible for professional fees and likely have more incentive to utilize corresponding CPT codes.

Our study demonstrated a significant increase in VAD procedure code utilization from 2015 to 2020, suggesting improved awareness in VA-related medical coding among clinicians. This increase was most notable from 2015 to 2018. The slight decrease in code utilization in 2019 coincides with the American Medical Association revision and issuance of new guidance for the CPT codes for multiple VADs (e.g., PICCS, midlines),²¹ which may have temporarily affected coding practices.

Our research also revealed significant variability among patient/institutional characteristics and their association with code utilization. While some of these differences were straightforward (e.g., higher code use in sicker patients or academic institutions), others point to more complex factors. It is not surprising that patients with multiple comorbidities (higher CCI score) were more likely to receive a VAD procedure code. Sicker patients often stay hospitalized for extended periods and may be more likely to receive attention/care from healthcare providers, which could lead to more accurate reporting. We also identified differences in age, sex, and ethnicity regarding VAD code utilization. Previous studies suggested that patient care might differ based on demographic factors and demonstrated such disparities based on gender.²² Our findings showed that younger or female inpatients were less likely to receive a VAD procedure code. Furthermore, while the proportion of black patients who received a VAD code (21 percent) was higher than those without codes (8.7 percent), the direction of data was the opposite for other patients of color (8.7 percent; 14.8 percent). These differences might stem from disparities in quality of care, which was documented in the literature.²³,²⁴ While the underlying reasons for these differences should not be overlooked, our study was not designed to capture the related root causes.

Accurate medical coding and documentation are critical to coordinate patient care and foster collaboration among healthcare providers, which may improve patient care during subsequent encounters.²⁵,²⁶ VA is an essential part of patient care, and VA-related procedures are associated with complications such as phlebitis, thrombosis, and infection.²⁷,²⁸ Therefore, it is valuable for clinicians to be aware of the patients’ VA-related medical history, especially for optimal VAD selection when treating patients with difficult intravenous access since multiple insertion attempts may result in diminished clinical and economic outcomes as well as poor patient experience.²⁹ Moreover, proper medical coding is essential to produce accurate data for high-quality retrospective research.³⁰ Finally, appropriate documentation could inform unit-specific resource and personnel needs and may lead to improvements in inventory management and medical staff allocation.³¹

Limitations

This study presents several limitations, including intrinsic limitations of any retrospective research reporting descriptive analyses. Our study evaluated data from 101 US hospitals, and its findings may not be generalizable. Further research is necessary to understand the impact of low CPT code utilization on patient care and outcomes. Additional research may also be conducted to evaluate other forms of documentation (e.g., patient charts, clinician notes), which was not within the scope of our study. Worth noting, even if different forms of documentation are available, coding is an essential component of appropriate and complete documentation³²,³³ considering its wide range of uses, including reporting, increased accuracy/efficiency, claims processing, guideline development, communication among providers, and retrospective research.³⁴,³⁵

Conclusion

Despite the widespread use of VADs, our study revealed a strikingly low use of VAD procedure codes in the inpatient setting. Accurate coding of VA procedures is particularly important to capture VA-related history and to ensure high-quality research and accurate resource/staff allocation. Simplified/streamlined coding processes should be considered, and more research is needed to evaluate the impacts of low code utilization.

Data Availability Statement

Cerner Real-World Data TM is a national, de-identified, person-centric data set that can be acquired from Cerner Corporation.

Acknowledgements

The authors would like to acknowledge the following from Boston Strategic Partners Inc. (supported by BD) for their contributions to the study: Dr. Halit Yapici and Dr. Nicholas Antaki for their research and editorial support, and Mr. Seungyoung Hwang for statistical assistance.

Author Biographies

Erica DeBoer (Erica.DeBoer@sanfordhealth.org) is the chief nursing officer at Sanford Health.

Kimberly Alsbrooks (Kim.Alsbrooks@bd.com) is the senior director of medical affairs at Becton, Dickinson and Company (BD).