Factors Contributing to Delay in Specialist Care after Colorectal Cancer Diagnosis in Kentucky

Abstract

Introduction:

Abundant studies have associated colorectal cancer (CRC) treatment delay with advanced diagnosis and worse mortality. Delay in seeking specialist is a contributor to CRC treatment delay. The goal of this study is to investigate contributing factors to 14-days delay from diagnosis of CRC on colonoscopy to the first specialist visit in the state of Kentucky.

Methods:

The Kentucky Cancer Registry (KCR) database linked with health administrative claims data was queried to include adult patients diagnosed with stage I-IV CRC from January 2007 to December 2012. The dates of last colonoscopy and first specialist visit were identified through the claims. Bivariate and logistic regression analysis were performed to identify factors associated with delay to CRC specialist visit.

Results:

A total of 3,927 patients, from 100 hospitals in Kentucky were included. Approximately 19% of CRC patients visited a specialist more than 14 days after CRC detection on colonoscopy. Delay to Specialist (DTS) was found more likely in patients with Medicaid insurance (OR 3.1, p<0.0001), low and moderate education level (OR 1.4 and 1.3, respectively, p=0.0127), and stage I CRC (OR 1.5, p<0.0001). There was a higher percentage of DTS among Medicaid patients (44.0%) compared to Medicare (18.0%) and privately insured patients (18.8%).

Conclusion:

We identified Medicaid insurance, low education attainment and early stage CRC diagnosis as independent risk factors associated with 14-days delay in seeking specialist care after CRC detection on colonoscopy.

INTRODUCTION

Colorectal cancer (CRC) is the third leading cause of cancer related mortality in the United States ⁽¹⁾. The incidence of CRC in the United States is 145,600 patients with a mortality rate of 51,020 in 2018 ⁽¹⁾. According to the American Cancer Society, the state of Kentucky ranked the first in CRC incidence and fifth in mortality in the United States from year 2017 to 2019 ⁽²⁾. Nearly half (54) of 120 counties in eastern Kentucky belongs to the central Appalachia region, which historically carries a worse mortality rate (21.6 per 100,000) compared to its urban non-Appalachian counterpart (20.4 per 100,000)⁽³⁾. In the past two decades, Kentucky implemented several programs to successfully increase CRC screening, such as clinical module intervention to increase screening in rural Kentucky and the Affordable Care Act Medicaid expansion in 2016 to increase insurance coverage ^(4–9). These interventions helped to increase the CRC screening rate in Kentucky from 34.7% in 1999 to 70.1% in 2016 ^{(4, 6)}. Despite these efforts in increasing CRC screening, mortality remains high in Kentucky. This suggests further investigation in other barriers to care for CRC patients, such as prompt treatment delivery.

Delays in diagnosis and treatment initiation were found to be associated with more advanced stage and worse patient survival ^{(10, 11)}. Several European national healthcare services implemented fast track referral protocols for CRC patients ^(12–18). One recommendation established by the United Kingdom’s National Health Service was the 14-days referral rule, which recommended patients with symptoms detected by general practitioners to visit a specialist within 14 calendar days ⁽¹⁹⁾. Other institutions also utilized similar referral courtesy to have patients seen by a specialist and to initiate treatment within 14 days after CRC diagnosis ^{(14, 16)}. There is not a common consensus of standard referral waiting times in the United States currently; however, there is a general agreement that minimizing waiting time can reduce anxiety experienced by patients ^(20–22). At the University of Kentucky Markey Cancer Center, we utilize an expedited referral policy for newly diagnosed CRC patients to establish specialist care within 14 days ⁽²³⁾. We define delay to specialist (DTS) care as more than 14 days after CRC detection on colonoscopy.

It is particularly challenging to deliver healthcare to residents of eastern Kentucky Appalachia due to lack of access, resources and low health literacy^(24–26). Prior studies reported that determinants such as, lack of presenting symptoms, refusal of care, and false-negative results accounted for DTS and initiation of treatment ^{(27, 28)}. However, we do not know the contributory factors to DTS. The purpose of this study is to identify factors associated with DTS, which will allow us to implement interventions to deliver prompt CRC care in Kentucky.

PATIENTS AND METHODS:

Patient selection and database

De-identified health claims linked Kentucky Cancer Registry (KCR) dataset was extracted for CRC patients diagnosed with stage I – IV between January 2007, to December 2012. Institutional Review Board approval with waiver for informed consent was obtained from the University of Kentucky’s Office of Research prior to data analysis. To ensure the data accuracy, only patients with at least 13 months of health insurance continuous enrollment (excluding the month of diagnosis, 6 months prior to cancer diagnosis to 6 months after) were included in the study. The final study analytical dataset included 3,927 CRC patients who met the following inclusion criteria: (1) adult patients age ≥ 20; (2) current Kentucky residents; (3) first invasive primary CRC diagnosis; (4) stage I – IV CRC diagnosis (4) receiving treatment within 6 months of diagnosis. According to AJCC guideline, stage 0 / in situ CRC patients diagnosed on colonoscopy and underwent endoscopic resection does not warrant an immediate follow up with specialist⁽²⁹⁾. Therefore, this group of patients was excluded from our study.

The KCR is a population-based registry and has been awarded the highest level of certification by the North American Association of Central Cancer Registries for an objective evaluation of completeness, accuracy and timeliness every year since 1997. The KCR is also part of the Surveillance, Epidemiology, and End Results (SEER) program, whose data have been widely used for cancer outcome research ⁽³⁰⁾. KCR performed a probabilistic data linkage linking KCR cancer cases with health administrative claim data from Medicaid, Kentucky state employee insurance database and private insurance groups, then consolidated the linked data with SEER Medicare data to create a claim-linked KCR dataset.

Definition of Delay to Specialist (DTS)

Delay time interval was calculated based on date of the latest colonoscopy prior to CRC diagnosis and the first specialist visit after colonoscopy. The date of colonoscopy was limited to within 6 months prior to CRC diagnosis and identified based on ICD-9 codes and HCPC codes (Table 1)⁽³¹⁾. The first specialist visit was defined as after the latest colonoscopy and prior to CRC treatment. If no specialist visit was obtained after 6 months of cancer diagnosis, the first treatment date was considered as the date of first specialist visit. The specialist was identified based on corresponding variables from the claim sources, including specialties such as medical oncologist, general surgeon, specialty surgeon, and radiation oncologist. Patients who had delay time interval greater than 14 days were considered the delay group, while within 14 days were considered the early group.

Table 1 –

CPT/HCPCS and ICD-9-CM codes

CPT/HCPCS and ICD-9 codes used to identify colonoscopy
CPT/HCPCS codesa
44388–44394	Under endoscopy, stomal
45378–45386, 45391,45392	Under endoscopy procedures on the rectum
G0105	Colorectal cancer screening; colonoscopy on individual at high risk
G0121	Colorectal cancer screening; colonoscopy on individual not meeting criteria for high risk
ICD-9-CMb
45.21	Transabdominal endoscopy of large intestine
45.22	Endoscopy of large intestine through artificial stoma
45.23	Colonoscopy
45.25	Closed endoscopic biopsy of large intestine
45.41	Excision of lesion or tissue or large intestine
45.42	Endoscopic polypectomy of large intestine
45.43	Endoscopic destruction of other lesion or tissue or large intestine
48.36	Endoscopic polypectomy of rectum

^aCPT/HCPCS – Current Procedural Terminology/Healthcare Common Procedure Coding System

^bICD-9-CM – International Classification of Disease, Ninth Revision, Clinical Modification

Data variables

The following demographic variables were included in our analysis: age at diagnosis, race, gender, education attainment, poverty status, rural status, Appalachian status, disability status, insurance type, hospital type and distance to hospital. Education attainment and poverty status was determined by high school completion rate and percentage of population below poverty income within the 2000 Census tract. Education attainment was categorized into three levels based on the tertiles of corresponding distributions: low education group (high school completion rate less than 67.9%); moderate education group (high school completion rate between 67.9% to 79.2%), and high education group (high school completion rate of 79.2% or higher). High poverty index group reside in area with 18.9% or above population below poverty income, while low poverty index patients live in area with less than 9.7% population below poverty income. Appalachian status was determined by the Appalachia Regional Commission ⁽³²⁾. The patient’s disability status was defined using the corresponding variables from claim sources. For Medicaid patients, long term disability coverage was considered as disabled. The type of hospital was categorized as large academic hospital (University of Louisville and University of Kentucky), large non-academic hospital (reporting 100 ≥ cancer cases per year), small non-academic hospital (reporting < 100 cancer cases per year), and out-of-state hospital. Since the institute of the specialist was de-identified, we estimated the distance from the registered residence to the hospital where patients received treatment to calculate the great circle distance, proposed by NAACCR ⁽³³⁾. If the hospital where patients received treatment was not available, the hospital where claims were generated was used to calculate great circle distance. Socioeconomic status was categorized based on education, income level, poverty percentage, occupation or social grouping, and other defining matrix (i.e Jarman index, Townsend deprivation score, Elley-Irving Scale, Nam-Powers score) ^(34–37).

The following clinical variables were also included: stage, tumor grade, and Charlson comorbidity Index (CCI). Stage and tumor grade were defined using American Joint Committee of Cancer (AJCC) staging manual⁽³⁸⁾. CCI was calculated using the modified version for claim data and categorized into four categories (0, 1, 2 and 3+) ⁽³⁹⁾.

Statistical analysis

A descriptive analysis of the follow-up interval, demographic and clinical factors was performed. We used χ² tests to examine associations between DTS and variables described above in the bivariate analysis. Logistic regressions were fitted to identify significant factors associated with DTS while controlling for other covariates. A sensitivity analysis was performed with and without missing and unknown variables. Goodness of fit was also tested. All analyses were performed using SAS Statistical software version 9.4 (SAS Institute, Inc., Cary, North Carolina, USA). All statistical tests were two sided with a P-value ≤ 0.05 to identify statistical significance. The EQUATOR/STROBE checklist was used to guide organization of the epidemiologic study⁽⁴⁰⁾.

RESULTS

Patient demographics, disease characteristics and functional status: Stage I CRC was independently associated with DTS

A total of 3,927 patients were included from years January 2007 to December 2012. Two (2) large academic hospitals, 38 large non-academic hospitals and 60 small non-academic hospitals were included. Approximately 19.0% of newly diagnosed CRC patients (n=753) experienced DTS. Among patients who experienced DTS, the median was 23 days with an interquartile range of 17 days. The bivariate analyses demonstrated that younger patients, both age group 20–49 year-old and 50–64 year-old (22.8% and 23.8%, respectively) have higher percentage experiencing DTS compared to elderly patients age ≥ 75 years old (15.9%, p<0.0001). There was no statistical difference in terms of gender and racial groups (Caucasians, African-Americans, and others) (Table 2).

Table 2 –

Bivariate analysis of patient demographics, disease burden factors, and socioeconomic status/rurality on 14-days delay in seeking specialist after colorectal cancer diagnosis on colonoscopy.

	Percentage of Patients Experiencing 14-days Delay to Specialist
Variable (s)	Early (n)	%	Delay (n)	%	p-value
Total	3174	80.83	753	19.17
Patient demographics
Age (years old)					<0.0001
20–49	169	77.17	50	22.83
50–64	563	76.18	176	23.82
65–74	1149	80.29	282	19.71
75+	1293	84.07	245	15.93
Race					0.4407
Caucasian	2988	80.87	707	19.13
African-American	166	81.37	38	18.63
Other	20	71.43	8	28.57
Sex					0.8865
Male	1601	80.74	382	19.26
Female	1573	80.92	371	19.08
Disease characteristics and functional status
Stage					<0.0001
Stage I	930	75.98	294	24.02
Stage II	918	85.47	156	14.53
Stage III	848	83.96	162	16.04
Stage IV	338	82.24	73	17.76
Stage Unknown	140	67.31	68	32.69
Grade					<0.0001
Well differentiated	136	73.51	49	26.49
Moderately- differentiated	2270	81.80	505	18.20
Poorly differentiated	339	83.09	69	16.91
Un-Differentiated	211	86.83	32	13.17
Grade Unknown	218	68.99	98	31.01
CCIa					0.3047
0	1690	81.64	380	18.36
1	811	79.98	203	20.02
2	354	81.57	80	18.43
3+	319	78.00	90	22.00
Disable Status					0.0022
Not Disabled	2475	81.98	544	18.02
Disable	377	78.22	105	21.78
Unknown	322	75.59	104	24.41
Socioeconomic status
Education Attainmentb					<0.0001
Low	1024	77.63	295	22.37
Moderate	1041	79.77	264	20.23
High	1109	85.11	194	14.89
Poverty levelc					<0.0001
Low	1108	84.77	199	15.23
Moderate	1043	80.05	260	19.95
High	1023	77.68	294	22.32
Appalachia Statusd					<0.0001
Non-Appalachia	2292	82.45	488	17.55
Appalachia	882	76.90	265	23.10
Healthcare access
Insurance Status					<0.0001
Private Insured	601	81.22	139	18.78
Medicare	2484	82.03	544	17.97
Medicaid	89	55.97	70	44.03
Metro Statuse					<0.0001
Rural	1460	78.24	406	21.76
Metropolitan	1714	83.16	347	16.84
Hospital typesf					<0.0001
Academic Hospital	100	62.89	59	37.11
Large Non-Academic	2521	82.85	522	17.15
Small Non-Academic	553	76.28	172	23.72
Great Circle Distance To Hospitalg					0.0001
<10 Miles	1651	82.18	358	17.82
10–50 Miles	1188	81.37	272	18.63
>50 Miles	169	74.78	57	25.22
Unknown/Out of State	166	71.55	66	28.45

^aCCI, Charlson Comorbidity Index

^bEducation attainment is categorized as a range of patients who completed high school: low (75.8–84.3%), moderate (84.4%−88.0%) and high (88.1–91.8%)

^cPoverty level is categorized into percentage below poverty income: high (above 18.7%), moderate (9.7–18.7%) and low (less than 9.7%)

^dAppalachia status is defined by 54 counties in Eastern Kentucky belong to Appalachian Regional Consortium

^eMetropolitan and rurality is defined based on Urban-Rural Continuum codes with the values of 1–3 as urban and 4–9 as rural

^fLarge hospitals are defined as treating >100 cancer cases per year; Academic hospitals include University of Kentucky and University of Louisville

^gGreat circle distance to hospital is the distance between patient residence to treatment hospital; if treatment hospital is not available, claim reporting hospital is used instead.

We found that patients diagnosed with stage I CRC (24.0%, p<0.0001) were more likely to experience DTS compared to later stage CRC (14.5% for stage II, 16.0% for stage III, and 17.8% for stage IV, and 32.7% for unknown stage, accordingly, p<0.0001). Similarly, tumor grade, which was another pathological characteristic used to categorize disease burden, was also associated with DTS. The well-differentiated group was found to have the highest percentage (26.5%) compared poorly differentiated and undifferentiated groups (16.9% and 13.2%, respectively, p<0.0001). We also investigated patient functional status, such as disability and CCI, and their association with DTS. The disabled group was found to have a slightly higher percentage in DTS (21.8%) compared to the non-disabled group in the bivariate analysis (18.0%, p = 0.0022). On the other hand, CCI was not statistically significant in contributing to DTS (Table 2).

After controlling for all significant variables, the multivariable logistic regression showed that stage I CRC was the only disease characteristic factor independently associated with DTS. Patients with Stage I disease were more likely to experience delay compared to stage IV patients (Odds Ratio (OR) 1.5, p<0.0001). Younger age, tumor grade, CCI and disability status were no longer statistically significant in the multivariable analysis (Table 3).

Table 3 –

Multivariable analysis of independent variables associated with 14-days delay in seeking specialist after colorectal cancer diagnosis on colonoscopy.

Factors Associated with 14-days Delay to Specialist from Colorectal Cancer Diagnosis on Colonoscopy
	Logistic Regression Analysis
Independent Variable (s)	ORa	95% CIb	p-value
Age (years)			0.0506
20–49	1.27	0.86–1.90
50–64	1.37	1.05–1.79
65–74	1.27	1.05–1.54
> 75	Refc
Stage			<0.0001
Stage I	1.53	1.14–2.04
Stage II	0.81	0.59–1.10
Stage III	0.89	0.66–1.21
Stage IV	Ref
Stage Unknown	2.08	1.41–3.08
CCId			0.3083
Score 0	Ref
Score 1	1.09	0.90–1.33
Score 2	1.02	0.78–1.35
Score 3+	1.28	0.98–1.68
Insurance status			<0.0001
Privately Insured	Ref
Medicare	1.03	0.80–1.34
Medicaid	3.09	2.12–4.50
Education attainmente			0.0127
Low	1.38	1.08–1.75
Moderate	1.33	1.08–1.65
High	Ref
Appalachian statusf			0.2293
Non-Appalachia	Ref
Appalachia	1.14	0.92–1.40

^aOR, odds ratio

^bCI, confidence interval

^cRef, reference

^dCCI, Charlson Comorbidity Index

^eEducation attainment is categorized as a range of patients who completed high school: low (75.8–84.3%), moderate (84.4%–88.0%) and high (88.191.8%)

^fAppalachia status is defined by 54 counties in Eastern Kentucky belong to Appalachian Regional Consortium.

Socioeconomic factors: low education attainment contributed independently to DTS

Several socioeconomic risk factors found to be associated with DTS included low education attainment, high poverty level and Appalachia status from bivariate analysis. First of all, patients with low and moderate education attainment subgroups were found to have higher percentage of delay (22.4% and 20.0%, respectively, p<0.0001) compared to high education attainment subgroup (14.9%) in the bivariate analysis. As expected, high poverty index was also found to be associated with DTS (22.3%, p<0.0001). More specifically pertinent to our Kentucky population, we found that Appalachian status had a higher percentage of DTS compared to non-Appalachian subgroup (23.1% compared to 17.6%, p<0.0001) (Table 2). After controlling for all variables, we continued to observe that patients with low and moderate education attainments were more likely to experience delay compared to those patients with a high education attainment (OR 1.4 for low education and OR 1.3 for moderate education, respectively, p=0.0127) from multivariable analysis. Appalachia status, which was a proxy for low socioeconomic status, was no longer statistically significant (Table 3).

Healthcare access: Medicaid insurance status was strongly indicative of DTS

We further investigated factors impacting healthcare access, which included insurance status, rurality, hospital types and great circle distance to hospital. Most significantly, we found that there was an association between insurance type and delayed care. A further analysis revealed that our study population was comprised of a predominant 3,028 Medicare patients (77%), 740 privately insured patients (19%) and only 159 Medicaid patients (4%). The Medicaid insurance subgroup had a remarkably higher percentage of patients experiencing delay (44.0%) compared to either Medicare subgroup or privately insured subgroup from the bivariate analysis. (18.0% and 18.8%, respectively, p<0.0001) (Figure 1). Medicaid insurance status remained as a strong indicative factor contributing to DTS from multivariable analysis. These patients were more likely to experience delay compared to privately insured patients (OR 3.1, 95% CI 2.1–4.5, p<0.0001) (Table 3).

Figure 1 –

Percentage of patients experiencing 14-days Delay to Specialist among various insurance types

Next, we examined rurality as a factor and found that patients living in rural area had a higher percentage of delay compared to metropolitan residents (21.8% compared to 16.8%, p<0.0001). We also evaluated hospital type and great circle distance to hospital as two other surrogates to capture our population. Our data also suggested that patients seeking care in academic hospital and small non-academic hospitals had higher percentage of delay (37.1% and 23.7%, respectively) compare to large non-academic hospitals (17.2%, p<0.0001). Majority of patients (n=3,469, 88%) were able to receive treatment within 50 miles radius of a hospital. However, patients who traveled more than 50 miles to hospital and out-of-the-state had a higher percentage of delay within their subgroups (25.2% and 28.5%, respectively, p=0.0004) (Table 2). Nevertheless, these healthcare access factors, namely rurality, hospital type, and great circle distance to hospital, were not significant in the multivariable analysis (Table 3).

Discussion:

Our study found multiple contributing factors and identified the population at risk to experience 14-days delay in seeking specialist care after CRC diagnosis on colonoscopy. We found that stage I CRC disease, low education attainment, and Medicaid insurance status were three independent factors associated with worse DTS. These factors were particularly relevant in the eastern Kentucky Appalachian population and could potentially explain the poor cancer care outcomes.

First, our study found that patients with stage I disease were more likely to experience delays in seeking specialty care compared to patients with stage IV metastatic CRC. This finding is consistent with current literature. Previous studies suggested that locally advanced and metastatic disease were generally more symptomatic, such as rectal bleeding and weight loss ^{(27, 41, 42)}. Therefore, patients with asymptomatic early stage disease may not seek medical attention promptly after CRC diagnosis ^{(43, 44)}. One study found that nearly 90% of CRC were diagnosed after development of symptoms and subsequently had more advanced disease, compared to asymptomatic patients⁽⁴⁵⁾. The eastern Kentucky Appalachian population, which was economically distressed^{(26, 46)}, were more likely to prioritize urgent personal and financial needs over seeking medical care for asymptomatic disease⁽²⁵⁾. In addition, the Appalachians historically had lower health care utilization and lack of follow-up with referrals, which were influenced by a culture of personal privacy and pride⁽²⁵⁾. Therefore, our eastern Kentucky Appalachian patients, though not an independent contributing factor in the multivariable analysis, were particularly at risk to experience DTS, especially among asymptomatic patients.

Next, we examined socioeconomic factors and found that patients from low and moderate education attainment groups were more likely to experience delay compared to high attainment patients. Our result was well supported and consistent with existing studies that lower education level was directly associated with healthcare delay and clinical outcome ^{(34, 44, 47)}. Lower education attainment rendered patients at risk for poor healthcare literacy, which had direct impact on clinical outcomes found in multiple existing studies^(48–50). One study found that adults with inadequate or limited health literacy were more likely to report avoiding doctor’s visits and information about disease, and to have a more fatalistic attitudes towards cancer⁽⁵¹⁾. While most of the available studies examined diagnosis delays and recapitulated the importance of screening and preventive measures in various cancer types^(51–54), our study provided knowledge regarding delays in seeking a specialist after diagnosis. We observed low education attainment contributed independently to DTS from multivariable analysis. The eastern Kentucky Appalachian population was characterized by lower education ascertainment and poor healthcare literacy compared to other parts of the country ^{(24, 55)}. Our population exhibited similar health-seeking behaviors and attitudes, such as under-utilization of healthcare and lack of initiative to seek prompt healthcare, comparing to non-Appalachians with low education and inadequate healthcare literacy⁽²⁵⁾. This finding further suggested that promoting patient education and healthcare literacy could shorten DTS, thus improve promptness of cancer care among eastern Kentucky Appalachians.

Finally, we evaluated healthcare access factors and identified Medicaid insurance status as an independent factor associated with DTS. Our results showed that there was a higher percentage of patient experiencing delay in the Medicaid group compared to the others. Kentucky was one of the 33 states participating in Medicaid expansion in 2016. There was 1.35 million Medicaid patients in 120 counties of Kentucky as of May 2019, which was 30.0% of the state population ⁽⁵⁶⁾. The Medicaid enrollment rate was higher compared to the 22.2% overall US enrollment rate ⁽⁵⁷⁾. Although previous studies have investigated the impact of Medicaid status on treatment delay, a gap in the literature exists to identify an association between Medicaid insurance and timeliness in seeking specialist care ^{(58, 59)}. One likely explanation for worse DTS in the Medicaid population was the variation in distribution of resources (i.e. medical staff, treatment facilities, disparities in access), with less specialty care located in areas high with Medicaid patients⁽²⁶⁾. In addition, prior studies suggested that specialists were less likely to accept Medicaid insurance compared to general practitioners and thus, an important barrier to access for underserved population ^(60–63). With a high Medicaid enrollment in the state of Kentucky, our population is especially at risk of experiencing DTS from the healthcare access perspective.

The eastern Kentucky Appalachian patients are particularly vulnerable to experience DTS. This culturally and geographically isolated Appalachian population historically present with more advanced disease and less likely to seek medical attentions ^{(25, 26)}. Our study also found a higher percentage of Appalachian patients in the delay group compared to the early group in the bivariate analysis, though not a statistically significant factor in multivariable analysis. The eastern Kentucky Appalachian population encompasses multiple characteristics that contributed to DTS. These factors that characterize the eastern Kentucky Appalachian population, such as Medicaid insurance status and poor education attainment, could have exerted a stronger influence, thus reducing the impact of Appalachian status statistically. This is likely the reason that we did not directly observe Appalachian status being statistically significant in the multivariable analysis. A similar result may pertain to the lack of significance found in delay associated with non-academic small hospital, greater circle distance to hospital, and rurality.

Limitations:

There were several limitations to our study. First, this was a retrospective study from a large population-based database. Retrospective data were subject to multiple confounding factors and prone to selection bias. We were not able to conduct causation analysis due to the nature of retrospective study. Secondly, our data were extracted from a claim-based KCR database and uninsured patients were unfortunately excluded from the study population. Our study was biased in terms of this exclusion criteria, and we anticipated that uninsured patients experienced worse delay compared to insured patients included in our study. Therefore, DTS could have been worse among the Kentucky population compared to our study. In addition to the factors discussed in our study, we did not include healthcare literacy, specialty availability and acceptance due to unavailability of data from the claim-based registry. These confounding factors could have had an impact on DTS, and unfortunately, could not be assessed. Lastly, our study population was predominantly (93%) Caucasians⁽³²⁾, which was representative with Kentucky demographics. This relative homogeneity did not allow us to observe difference among racial groups, which was suggested by previous studies. Results should be interpreted with caution when applied in a different population setting. Despite these limitations, our study found multiple factors independently contributed to delay in seeking specialist that were supported by literature. In regard to missing and unknown data in the variables collected in this study, we performed sensitivity analysis with and within missing data. This did not result in any significant discrepancy in the conclusion of this study.

Conclusion:

In summary, this study identifies stage I CRC diagnosis, low education attainment and Medicaid insurance status as independent factors contributing to delay to specialist after CRC diagnosis on colonoscopy. The eastern Kentucky Appalachian population, which encompasses several risk factors, is particularly vulnerable. Future studies and interventions should be directed towards reducing healthcare disparities and promoting patient education in the state of Kentucky.