Leveraging a new ICD-10 diagnosis code to characterize hospitalized patients with primary sclerosing cholangitis

Research Letter

Primary sclerosing cholangitis (PSC) is a progressive cholestatic liver disease characterized by inflammation and destruction of the biliary tree, often leading to cirrhosis requiring liver transplantation.^{1, 2} PSC is rare, affecting ~1 to 16 per 100,000 in the U.S.,¹ and is typically considered a disease of young White males.^{1, 3} However, with changing demographics of US population in terms of age, gender, and race/ethnicity, it is unknown whether demographics of patients with PSC has also shifted.

Large administrative datasets can be useful for studying uncommon diseases, but prior studies of PSC have been limited by poor validity of diagnostic codes.² Previously, an International Classification of Diseases (ICD) code was used to identify several different cholangitic diseases, including PSC and diseases such as acute cholangitis, making accurate identification of PSC patients in administrative data challenging. In 2018, a PSC-specific ICD-10 code was introduced for more accurate identification of PSC patients in administrative datasets. We leveraged this new ICD-10 code in a contemporary cohort of hospitalized PSC patients to preliminarily characterize this population’s demographics and associate them with clinical characteristics and hospitalization outcomes.

We used de-identified data from the Vizient® Clinical Data Base (CDB), a performance improvement database including hospital encounters from >800 US health systems, including 95% of academic medical centers, as well as affiliated and community hospitals. Hospital encounters for patients ≥18 years old admitted with a principal CMS-defined medical diagnosis (i.e. not surgical, obstetric, etc.) 10/1/2018–1/1/2020 (N=479,601) were included in our cohort; those with either a principal or secondary ICD-10 code for PSC (K83.01) were included in the present study (n=944). Charlson Comorbidity Index (CCI) was calculated and categorized to assess comorbidity.⁴ Reason for admission was categorized based on principal and secondary ICD-10 codes. ICD-10 Procedure Coding System (PCS) codes identified inpatient procedures of interest. Hospitalization outcomes included hospital length of stay (LOS) and in-hospital mortality. Demographic and clinical characteristics were analyzed using Chi-squared analysis and Wilcoxon rank sum analysis. We used logistic regression to associate demographic characteristics with outcomes. A p-value<0.01 was considered statistically significant. Data were analyzed using Stata/MP 16.1 statistical software (College Station, TX).

We identified 944 patients hospitalized with PSC; median age was 46.9 years (IQR 34.7–62.9) with unimodal distribution. 60% had a history of IBD, cirrhosis was present in 48%, and 5% had a history of liver transplant. Most common reasons for admission were cirrhosis-related (28%) and biliary obstruction/infection (25%). Inpatient mortality rate was 3% and median LOS was 4 days (IQR 2–6). 22% of our cohort was ≥65 years; age ≥65 was associated with reduced rates of concurrent IBD (50% vs 62%, p=0.002) and IBD-related admissions (OR 0.98, 95%CI 0.97–0.99, p<0.001), but increased rate of sepsis (24% vs 14%, p=0.001). Other reasons for admission were similar by age.

Race- and gender-stratified analyses were conducted to further characterize demographics of hospitalized PSC patients. The racial distribution was: 76% White, 16% Black, 2% Asian, and 6% Other race. Black patients were significantly younger than non-Black patients, with a median age of 34.8 years (IQR 27.1–47.8) vs 48.2 years (IQR 37.8–64.7), p<0.001 (Figure 1A). There was no significant difference in rates of IBD or cirrhosis by race. Reasons for admission did not differ by race or gender. Endoscopic retrograde cholangiopancreatography (ERCP) was performed twice as often in White compared to Black patients (16.1% vs 8.2%, p=0.020).

Figure 1A.

Age and race/ethnicity distribution among patients hospitalized with PSC

Forty percent of patients were female. There was no significant difference in age by gender (p=0.66), though there was a trend towards higher proportion of younger patients being female (Figure 1B). Men were less likely to have severe CCI than women (57% vs 49%, p=0.05. There was no significant difference in ERCP rates by gender or age. On univariable and multivariable logistic regression, none of our demographic characteristics of interest (age, race, gender) were associated with inpatient mortality or LOS.

Figure 1B.

Age and gender distribution among patients hospitalized with PSC

Prior PSC epidemiological studies were limited due to difficulties accurately identifying PSC patients using large administrative datasets.² The new PSC-specific ICD-10 code released in 2018 enabled us to address this major weakness. We observed in this US-based preliminary cohort of inpatients that PSC patient demographics differ with age. Notably, hospitalized Black PSC patients with PSC are significantly younger than hospitalized non-Black PSC patients. Despite this, we did not observe any demographic differences in hospitalization outcomes.

With respect to age, we did not observe a bimodal age distribution as demonstrated in some previous studies.⁵ Most prior U.S. studies have focused on Caucasian patients given the classically strong White predominance of PSC.^{6, 7} Recent data suggest that Black (versus White) PSC patients may have unique clinical features, including lower rates of IBD and a higher proportion of women,⁸ and those of African Caribbean heritage may experience more PSC-related clinical events.⁹ In addition to these differences, our data suggest that compared to White patients, Black patients hospitalized with PSC tend to be younger, possibly suggesting younger age at diagnosis, more severe disease, or more frequent rates of hospitalization among Black patients due to lack of access to outpatient care. Our preliminary data also reveal disparities in rates of ERCP among Black patients, as has been previously shown in patients hospitalized with acute cholangitis.^{10, 11}

Whereas the male-to-female ratio has classically been reported as 2:1,¹ our cohort demonstrated a slightly higher proportion of women, suggesting an increased hospitalization rate among women with PSC, explained by differences in comorbidities or disease severity. While this study is unique in that it is the first to use a large administrative dataset since the introduction of a unique PSC ICD-10 code, we acknowledge the following limitations. First, provider or hospital implementation differences of this code may have introduced bias into our cohort. Specifically, it is possible that ongoing use of previous coding are still used to identify PSC-related hospitalizations, which would result in an underestimation of this cohort. Second, we were not able to fully characterize severity of PSC due to the lack of clinical granularity in this dataset. Finally, as a study limited to hospitalized patients with principal medical diagnosis of PSC, the differences we observed are specific to this population and may not represent all patients (i.e. hospitalized with surgical/obstetric diagnoses or outpatients).

We have leveraged a new ICD-10 code for PSC to analyze contemporary administrative admission data to preliminarily characterize hospitalized patients with PSC, highlighting possible differences in racial/ethnic and gender distribution by age. As use of this code increases with time, future studies should explore reasons for these differences and implications for inpatient/outpatient PSC management. Additional validation of this ICD-10 code is needed to leverage administrative data to further explore PSC demographics, identify disparities, and optimize management and outcomes among all PSC patients.

Abbreviations:

PSC

primary sclerosing cholangitis

CDB/RM

Clinical Data Base/Resource Manager

ICD-10

International Classification of diseases, 10^th Revision- Clinical Modification

PCS

Procedure Coding System

LOS

length of stay

ERCP

endoscopic retrograde cholangiopancreatography

IQR

interquartile range

OR

odds ratio