Disparities in Outpatient Rural Cholecystectomy Outcomes

Introduction

Rural surgeons and the care they provide is a critical resource to local communities.¹ Rural populations face social and geographical isolation that result in significant time and effort to access healthcare.² Obtaining care far away is less efficient, more costly, and more disruptive to patients and their families.³ Local hospitals, where rural surgeons work, are often a major, if not largest, employer in many small communities.^3,4

There are many issues facing rural surgery including workforce shortages^1–3,5 and a narrow profit margin that relies on surgeons, surgical procedures, and patient volumes for financial viability.^2–6 When patients bypass their local rural hospital and choose to receive care at another hospital, this places a valuable local resource under significant financial stress and risk of closure.^1,4,5 A recent study looking at predictors of hospital bypass for elective operations found that bypass was greatest for elective operations with the highest risk of intraoperative blood loss and post-operative complications such as coronary artery bypass graft and valve replacements.⁵ Although cholecystectomy had the lowest rate of bypass, more than 50% of patients still bypassed their local hospital.⁵ This bypass behavior is concerning and some authors have postulated it may secondary to perceptions of poor outcomes at rural hospitals.^2,4,5

Our aim was to characterize differences in elective cholecystectomy outcomes between rural and urban hospitals with the hypothesis that rural hospitals will have similar outcomes to urban hospitals in terms of laparoscopy utilization, complications, and patient discharge disposition.

Materials and methods

The Nationwide Ambulatory Surgery Sample (NASS) is part of the Healthcare Cost and Utilization Project (HCUP), which is the “largest collection of longitudinal hospital care data in the United States.”⁷ NASS is an ambulatory health care database that provides national estimates of major ambulatory surgery encounters at hospital owned facilities.⁸ The NASS was used to perform a retrospective cohort analysis of patients aged 20 and older undergoing cholecystectomies between 2016–2018 at urban and rural hospitals. Current Procedural Terminology (CPT) codes for open (47600) and laparoscopic (47562) cholecystectomy were used to identify patients in the database. The frequency of laparoscopic converted to open cholecystectomy was also queried (Z53.31).

Patient demographic data including sex, age, Elixhauser Comorbidity Measure (ECM), insurance status, and quartile of median household income were collected. Sex was classified as male or female. Patients were grouped by age (20–29, 30–39, 40–49, 50–59, 60–69, 70 and older). The ECM was chosen to compare comorbidities between the rural and urban cohorts as it identifies comorbidities that exist at the time of the patient encounter based on ICD-10 diagnoses.⁹ It captures long-standing comorbidities and stratifies severity for several diagnoses (e.g., diabetes complicated and diabetes uncomplicated). The comorbidities used in the ECM were chosen as they impact resource allocation and healthcare outcomes.⁹ Insurance status was categorized as private, public (Medicare and Medicaid), self-pay, other, or unknown. Median household income was stratified into quartiles based on patient’s ZIP code.

Indications for operation were grouped into five main diagnoses including ICD-10 code for cholelithiasis (K80.2, K80.8), cholecystitis (K80.0, K80.1, K81.0, K81.1, K81.2, K81.9), choledocholithiasis (K80.4, K80.5, K80.6, K80.7), cholangitis (K80.3), and gallstone pancreatitis (K85.1). Patient who did not fall into any of these diagnoses or ICD-10 codes were included in a separate category for “other” surgical indication.

Hospital characteristics collected included rural versus urban location, region, teaching status, and hospital bed size. Rural and urban hospital designations with NASS are based on the Core Based Statistical Area (CBSA) codes from the 2010 census.¹⁰ Urban hospitals were located within metropolitan counties with more than 50,000 people. Rural hospitals included micropolitan and non-core counties with 10,000 to 50,000 and less than 10,000 people, respectively. Regions within NASS are categorized as Northeast, Midwest, South, and West. Teaching status is classified as rural, urban teaching, and urban non-teaching. Due to the rarity of rural teaching hospitals, rural hospitals were not subdivided into rural teaching and rural non-teaching in NASS.¹⁰ Hospital size was determined to be small (< 100 beds), medium (100–299 beds), or large (300+ beds) as per NASS descriptors.¹¹

The primary exposure variable of interest is urban versus rural hospital location. Primary outcomes analyzed between urban and rural hospitals include surgical approach, complications, and patient disposition. Surgical approach for cholecystectomy operations included laparoscopic, laparoscopic converted to open, and open. A list of complications can be found in Appendix 1. Disposition included routine discharge (home or self-care, court/law enforcement), transfer to short term hospital (inpatient admission to the hospital where the operation was performed or discharge/transfer to another short-term hospital), transfer to other (includes skilled nursing facility, intermediate care facility, and another type of facility), home health care, against medical advice, deceased, and other.

Statistical analysis

Bivariate and multivariate analyses of survey-weighted data were conducted. The difference in demographics, indication for operation, hospital characteristics, and cholecystectomy outcomes between rural and urban hospitals was examined using a Rao-Scott Chi-square test (as shown in Table 1–4). Cholecystectomy outcomes between rural and urban hospitals were compared using bivariate and multivariate logistic regression. The multivariate logistic regression models included covariates such as sex, age group, ECM, insurance status, median household income, region, teaching status, and hospital bed size (as detailed in Table 5).

Table 1.

Patient Demographics

	Rural n*= 82,513		Urban n*= 422,968
	%**	95% CI	%**	95% CI	p-value
Sex
Male	28.8	28.3–29.4	26.1	28.8–26.3	< 0.01
Female	71.2	70.7–71.7	73.9	73.7–74.2
Age
20–29	17.3	16.8–17.7	14.4	14.2–14.7	< 0.01
30–39	19.8	19.3–20.2	19.6	19.4–19.8
40–49	18.0	17.7–18.4	19.3	19.1–19.4
50–59	18.7	18.3–19.1	19.4	19.3–19.6
60–69	15.1	14.7–15.5	15.9	15.7–16.1
70 and older	11.1	10.7–11.5	11.4	11.1–11.7
ECM ^	Mean	SD	Mean	SD
	0.9	0.03	1.1	0.01	< 0.01
Insurance	% **	95% CI	% **	95% CI
Private	41.2	40.1–42.4	35.9	35.2–36.7	< 0.01
Public	52.0	50.9–53.0	58.4	57.6–59.3
Self-Pay	4.1	3.6–4.5	3.0	2.6–3.4
Other	2.6	2.2–2.9	2.5	2.3–2.8
Unknown	0.2	0.1–0.2	0.1	0.1–0.2
Quartile of Median Household Income
0–25th percentile	46.7	46.5–52.9	22.2	21.2–23.3	< 0.01
26–50th percentile	38.6	36.0–41.2	27.0	26.0–27.9
51–75th percentile	10.8	9.4–12.3	28.1	27.3–29.0
76–100th percentile	0.9	0.6–1.1	22.6	21.3–23.9

^*weighted n

^**weighted column percent, CI- Confidence Interval

^{^}ECM Elixhauser Comorbidity Measure, SD- Standard Deviation

Table 4.

Cholecystectomy Outcomes

	Rural n*= 82,513		Urban n*= 422,968
	%**	95% CI	%**	95% CI	p-value
Laparoscopy
Yes	99.7	99.7–99.8	99.7	99.7–99.8	0.82
No	0.3	0.2–0.3	0.3	0.2–0.3
Complication
Yes	0.4	0.3–0.5	0.3	0.2–0.3	<.0001
No	99.6	99.5–99.7	99.7	99.7–99.8
Patient disposition (% of the total N)
Routine (91.4%)
Yes	94.7	93.1–96.2	92.2	92.0–94.4	0.16
No	5.3	3.8–6.9	6.8	5.6–8.0
Transfer to Short Term Hospital (0.09%)
Yes	0.3	0.2–0.3	0.1	0.05–0.07	<.0001
No	99.7	99.7–99.8	99.94	99.93–99.95
Transfer to Other Facility (0.38%)
Yes	0.2	0.1–0.2	0.1	0.12–0.14	0.01
No	99.8	99.8–99.9	99.9	99.86–99.88
Home Health Care (0.74%)
Yes	0.2	0.1–0.2	0.2	0.1–0.2	0.62
No	99.8	99.8–99.9	99.8	99.8–99.9
Against Medical Advice (0.02%)
Yes	0.02	0.01–0.03	0.02	0.02–0.03	0.67
No	99.98	99.97–99.99	99.98	99.97–99.98
Deceased (0.0094%)
Yes	≤ 10 cases#		≤ 10 cases#		0.21
No	99.99	99.98–100.00	99.998	99.996–99.999
Others (7.4%)
Yes	4.7	3.1–6.2	6.4	5.2–7.6	0.11
No	95.3	93.8–96.9	93.6	92.4–94.8

^*weighted n

^**weighted column percent, CI- Confidence Interval

^#The observed number of cases was ≤10 and has therefore been masked to protect privacy of individual(s)

Table 5.

Multivariate Analysis of Cholecystectomy Outcomes- Urban Hospitals as Reference

Multivariate Results
	Estimated Odds Ratio*	95% Confidence Interval		p-value
Laparoscopy	1.93	0.73	5.13	0.19
Complications	1.39	1.11	1.75	<0.01
Patient Disposition
Routine	1.50	0.91	2.45	0.11
Transfer to Short Term	2.40	1.61	3.58	<0.01
Transfer to Other Facility	1.26	0.93	1.71	0.14
Home Health Care	1.37	0.97	1.94	0.07
Against Medical Advice	0.66	0.28	1.54	0.34
Deceased	3.23	0.10	100.0	0.51
Other	0.60	0.34	1.05	0.07

^*Estimated odds ratio adjusted for sex, age group, Elixhauser Comorbidity Measure, insurance status, median household income, hospital region, hospital bed size, hospital teaching status

A subgroup analysis of patients who experienced complications was performed using similar statistical methods (as provided in Table 6, Appendix 1). Complications were categorized within various domains, including infectious, intraoperative, cardiovascular, respiratory, renal, intestinal, shock, post-operative bleeding, and others. P-values < 0.05 were considered statistically significant, and SAS 9.4 (SAS Institute, Cary, NC, USA) was used for all statistical analyses.

Table 6.

Rate of Complications in Subgroup Analysis

	Rural n*= 326		Urban n*= 1,060
	%	95% CI	%	95% CI	p-value
Intraoperative	38.5	30.5–46.5	34.1	30.4–37.8	0.32
Cardiovascular	7.3	3.3–11.3	5.3	3.7–6.9	0.31
Respiratory	12.0	6.8–17.3	12.1	9.6–14.7	0.96
Shock	≤ 10 cases#		≤ 10 cases#		0.42
Renal	≤ 10 cases#		0	0	N/A
Intestinal	18.8	12.1–25.5	25.8	22.1–29.5	0.09
Post-operative Bleeding	9.5	4.9–14.1	6.8	5.0–8.6	0.23
Other	16.8	10.7–22.9	10.7	8.5–12.9	0.04

^*weighted n, CI- Confidence Interval

^#The observed number of cases was ≤10 and has therefore been masked to protect privacy of individual(s)

Results

Patients were predominantly female in both rural and urban hospitals with slightly fewer comorbidities in rural hospitals as seen by the lower Elixhauser Comorbidity Measure (ECM) (Table 1). And while there were some differences in insurance between rural and urban patients, a notable trend is the difference in median quartile household income. Over 85% of rural patients were in the bottom two quartiles, whereas over 50% of urban patients were in the top two quartiles.

In both rural and urban hospitals, the most common indication for cholecystectomy was cholecystitis followed by cholelithiasis (Table 2). In both rural and urban hospitals, the third most common diagnosis was “other” which included biliary dyskinesia, hydrops, perforation, and fistula. Cholangitis and gallstone pancreatitis were very rare indications for operation, likely reflecting how sick these patients are and the need for inpatient management.

Table 2.

Indication for Operation

	Rural n*= 82,513		Urban n*= 422,968
	%**	95% CI	%**	95% CI	p-value
Indication
Cholelithiasis	6.5	5.7–7.3	7.6	7.1–8.1	< 0.01
Cholecystitis	84.6	83.1–86.1	84.9	84.2–85.5
Choledocholithiasis	2.8	2.4–3.2	3.6	3.3–3.8
Cholangitis	≤ 10 cases#		0.0	0.01–0.02
Gallstone pancreatitis	0.3	0.2–0.3	0.4	0.3–0.4
Other	5.8	5.0–6.6	3.6	3.3–3.8

^*weighted n

^**weighted column percent, CI- Confidence Interval

^#The observed number of cases was ≤10 and has therefore been masked to protect privacy of individual(s)

Over 75% of rural hospitals were located in the south and Midwest, whereas just over 50% of urban hospitals were in these two regions (Table 3). The majority of urban hospitals were teaching (64.2%). Due to the rarity of rural teaching hospitals, rural hospitals were not further divided into teaching and non-teaching.¹⁰ Small hospitals predominated rural communities, whereas large hospitals were more common in the urban setting.

Table 3.

Hospital Characteristics

	Rural n*= 82,513		Urban n*= 422,968
	%**	95% CI	%**	95% CI	p-value
Region
Northeast	11.0	8.9–13.0	16.1	14.5–17.7	< 0.01
Midwest	31.6	28.5–34.7	20.2	18.5–21.8
South	47.3	43.5–51.0	36.5	34.2–38.8
West	10.2	8.1–12.2	27.3	25.0–29.6
Hospital Bed Size
Small	61.1	56.7–65.5	10.5	9.3–11.7	< 0.01
Medium	34.4	29.9–38.8	40.2	37.7–42.7
Large	4.5	1.9–7.2	49.3	46.8–51.9

^*weighted n

^**weighted column percent, CI- Confidence Interval

Over 99% of cholecystectomies were performed laparoscopically (Table 4). There was a slightly higher rate of complications in rural hospitals, 0.4% versus 0.3%. Over 90% of patients had routine discharges. There were higher rates of transfer to short term hospitals and transfer to other facilities after cholecystectomies performed at rural hospitals.

Using urban hospitals as the reference, no difference was detected in the rate of laparoscopy but more complications were observed the rural hospitals on multivariate analysis (Table 5). Although NASS was queried for laparoscopic converted to open operations (code Z53.31), there were no results recorded under this code. For patient disposition, there was a significantly higher rate of transfer to short term hospitals in rural hospitals. No difference was detected in routine discharge, transfer to other facilities, or mortality. The wide confidence interval for mortalities is secondary to the very few mortalities in both rural and urban hospitals.

On subgroup analysis of the patients with complications, there were similar trends in the demographics (Supplemental Table 1). Although there was a female predominance in both hospital settings, this was not statistically significant. No significant differences were observed in age or insurance status. Self-pay, other, and unknown insurance status had to be combined due to the rarity of these insurance types. Patients receiving care at rural hospitals had fewer comorbidities as seen by the ECM (mean (standard deviation), 1.4 (0.12) versus 1.7 (0.06), p<0.01). Substantially more rural patients were in the lowest two quartiles for median household income (85.9% versus 51.5%, p<0.01), which was similar to the adult cohort that included patients with and without complications. Although no significant difference in the indications for operation was detected, cholecystitis remained the most common indication (Supplemental Table 2). Due to the rarity of choledocholithiasis, gallstone pancreatitis, and cholangitis these three diagnoses were combined on the analysis.

Evaluating the complications in detail, there were no infectious complications in either the rural or urban cohort. The most common type of complication was intraoperative followed by intestinal then “other” in both rural and urban hospitals (Table 6). The only notable difference was that “other” complications occurred more commonly in the rural hospitals. This included a potpourri of complications including wound disruption and other unspecified complications (Appendix 1). Shock and renal complications were exceedingly rare, occurring in less than 10 patients in both rural and urban hospitals. To protect patient privacy, exact rates of these complications have been masked.

The small sample size further limited multivariate regression analysis with only intraoperative and intestinal complications having an adequate sample size. Urban hospitals were the reference with an adjusted estimated odds ratio of intraoperative complications of 1.31 (95%CI 0.79–2.18, p=0.29) and intestinal complications of 0.79 (95% CI 0.41–1.51, p=0.48) at rural hospitals. No significant difference in the rate of these complications was detected.

Evaluating outcomes in the subgroup with complications, laparoscopy was less frequently utilized in the rural cohort (Supplemental Table 3, 94.6% versus 98.2%, p=0.02). There was a significantly lower rate of routine discharge (82.2% versus 89.4%, p=0.03) and higher rate of discharge to a short-term hospital (10.6% versus 1.7%, p< 0.01) for patients undergoing cholecystectomy at rural hospitals. Discharge dispositions such as transfer to other facility, home health, against medical advice, and deceased were very rare.

Bivariate and multivariate regression analyses for outcomes were frequently limited by the small sample size and number of variables in the cohort of patients that had complications. The trend of less laparoscopy utilization at rural hospitals held true on bivariable analysis (OR 0.32, 95%CI 0.12–0.87, p=0.03), but the multivariate regression analysis could not be completed. Although routine discharge was less common from rural hospitals in bivariate analysis (OR 0.55, 95%CI 0.31–0.96, p=0.03), this trend did not hold true on multivariate analysis (OR 1.5, 95%CI 0.91–2.46, p=0.70). It was not possible to complete multivariate regression analysis for the other discharge dispositions, but general trends held true on bivariate analysis. Transfer to a short-term hospital was more frequent at rural hospitals (OR 6.74, 95%CI 3.16–14.4, p< 0.01), and there was no significant difference detected in transfer to other facility (OR 1.17, 95%CI 0.22–6.11, p=0.85), home health care (OR 1.19, 95%CI 0.14–9.92, p=0.87), leaving against medical advice (OR 1.60, 95%CI 0.14–17.9, p=0.70), and “other” discharge dispositions (OR 0.77, 95%CI 0.21–2.28, p=0.55).

Discussion

Patients were predominantly female in both rural and urban hospitals, which reflects the natural pathophysiology of biliary disease and findings in other studies.^12,13 Although the Nationwide Ambulatory Surgery Sample tracks ambulatory encounters, the most common indication for operation was cholecystitis and not cholelithiasis. This is likely secondary to billing influences on charting. The infrequency of cholangitis and gallstone pancreatitis in both rural and urban settings is consistent with the acuity of these disease processes that usually require inpatient admission.

Several studies examining rural populations, operations, and bypass behavior have noted that older patients more commonly seek care at their local, rural hospital.^4,5,14–17 This was reflected in a study using data from the Nationwide Inpatient Sample, where patients undergoing laparoscopic cholecystectomy at rural hospitals were older compared to urban hospitals (56.5 v 54.3 years, p< 0.01).¹⁸ Our data did not reflect this trend with 26.2% of patients having operations at rural hospitals and 27.3% of patients at urban hospitals being aged 60 and older. Older patients receiving care at urban hospitals may be a combination of a number of factors and confounded by case complexity and comorbidities.

We also found that patient’s receiving care at rural hospitals had fewer comorbidities. This trend has been seen in patients receiving emergency general surgery care at rural hospitals and critical access hospitals.^19–21 Similar to a study by Huntington et al. that showed a lower Charlson Comorbidity Index for patients undergoing laparoscopic cholecystectomy at rural hospitals,¹⁸ our results may be statistically significantly but unlikely to be clinically significant given the small difference in ECM. As older patients with increased comorbidities are more likely to be transferred to an urban hospital for emergency general surgery, ^22,23 another hypothesis is that patients with increased complexity may choose to receive elective care at an urban hospital and bypass their local hospitals.^4,13,14

We found a significant higher rate of transfer to short term hospitals when cholecystectomies were performed at rural hospitals. This is likely secondary to a number of reasons, particularly the resources available at the rural hospitals including specialists that can provide consultation recommendations and short term follow up as an outpatient. This hypothesis is consistent with a study evaluating acute care surgery transfers in a rural state that found transferred patients were older with more comorbidities and required more resources.²³

A recent study by de Jager et al.²⁴ showed that rural hospitals did not have an association between income quartile and post-operative adverse events such as mortality and complications. This was felt to be secondary to decreased healthcare segregation in rural hospitals. Although a substantially higher proportion of emergency general surgery at rural hospitals fell in the bottom two income quartiles in Chaudhary’s study,¹⁹ the differences in morbidity and mortality were not thought to be clinically significant. An interesting finding in Carbonell’s study¹³ is that as median income decreased, the surgical morbidity of laparoscopic cholecystectomy also decreased and there was no change in mortality. Although a substantial higher percentage of the rural patients fell in the lowest two median household income quartiles in our study, the difference in complications is clinically debatable (0.4% versus 0.3%) and fatalities were exceptionally rare.

The high rate of laparoscopic utilization likely reflects these were planned elective operations for benign biliary disease and the near ubiquitous use of laparoscopy today. There was no difference detected in utilization of laparoscopy between rural and urban hospitals. As laparoscopy was becoming more widely adopted in the 1990’s to early 2000’s, several studies investigated the adoption rate between rural and urban hospitals. A study using HCUP data from 2000, found that urban hospitals had decreased utilization of laparoscopy,¹³ and a separate study comparing rural and urban hospitals in Kentucky found that urban surgeons converted to open more frequently.²⁵ Another study utilizing national HCUP data found no difference in laparoscopy utilization for both elective and urgent cholecystectomies.¹⁶ More contemporary studies evaluating rural surgeon caseloads have consistently shown that laparoscopic cholecystectomy is one of the most common operations performed,^6,22 and open cholecystectomy was performed less than 1% of the time.⁶

Multiple previous studies have found no difference in surgical complications or mortality in rural versus urban hospitals.^{13,16,18,21,25} All of these studies evaluated inpatient operations, which can make direct comparisons to outpatient, elective operations challenging. One study looking at a several emergency general surgery operations found rural hospitals had lower odds of major complications, but higher odds of in-hospital mortality.¹⁹ This study did not stratify rates of complications by the type of operation performed and emergency general surgery patients are often sicker than patients who are eligible for elective operations. The overall mortality rate was very low in our study- weighted n of less than 10 in rural and urban hospitals. Our study has similar limitations to Chaudhary’s study¹⁹ in the aspect that due to the large sample size, even small differences in outcomes are statistically significant but may not be clinically significant.

It is challenging to reconcile the multivariate analysis detecting a significantly higher rate of complications in rural hospitals, but the only organ system where a significant difference in complications was detected is “other.” The complications captured in the “other” category include many that did not fit within a specified organ system and are described as “unspecified” and “not elsewhere classified” (Appendix 1). These complications are non-specific and unrelated, making it difficult to draw meaningful conclusions. This may be a limitation secondary to the relatively small weighted sample size and that several organ systems had less than ten cases.

On further analysis of the complications, it is not surprising that no infectious complications were recorded. For an ambulatory encounter, it would be highly unusual for signs of infection to show so early after surgical intervention. There is sparse data regarding the specific rates of intraoperative complications between rural and urban hospitals. A study using the Nationwide Inpatient Sample found no difference in intraoperative complications for endometrial cancer,²⁶ which reflects our findings for ambulatory laparoscopic cholecystectomies. This same study also showed a decrease ileus rate in rural hospitals, and no difference in rates of bowel obstruction or Clostridium difficile infection.²⁶

There are several limitations to this study secondary to the study population and structure of the NASS. This study only looked at ambulatory surgery encounters. Although the most common diagnosis was cholecystitis, these findings may not be applicable to patients with cholecystitis requiring an urgent, inpatient operation. It is possible there is selection bias in terms of which patients are well enough to undergo an elective, outpatient cholecystectomy and the data may be skewed by surgeons less willing to perform cholecystectomies in the ambulatory setting. The NASS is not designed to capture readmissions as it is structured based on the ambulatory surgery encounter. It does not have a mechanism to track patients for emergency department visits, readmissions at the same hospital, or readmissions to another facility. This is an important area for future research as many complications from cholecystectomy present in a delayed fashion. Although the data is older, it is too soon to evaluate any impact the COVID pandemic had on outcomes in the NASS as data is only available through 2021. This will be an important avenue for further research as the data becomes released.

Further limitations to this study are similar to other large administrative database studies. The accuracy of the information is dependent on correct data entry. It is not possible to verify the data or obtain individual level data. As the study is retrospective, it is possible to identify associations but not causation. Due to the large sample size, a number of findings are statistically significant but are unlikely clinically significant. The NASS is based on hospital owned facilities for ambulatory encounters, so independently or private owned surgical centers were not included in the analysis. Although we attempted to track laparoscopic converted to open procedures, it was not possible to identify this in the database.

Conclusions

Patients undergoing cholecystectomy at rural hospitals had lower median household income and fewer comorbidities. These patients had statistically higher transfer rates to short term hospitals and complications, but the clinical significance of these findings is likely confounded by the large sample size as overall rates were low (<0.5%). There was no difference in routine discharge for these elective operations. Laparoscopic cholecystectomies can be safely completed in the ambulatory rural setting with appropriate consideration for patient comorbidities and anticipation of operative complexity.