Surgical Approach as a Mediator of Rural–Urban Disparity after Colon Cancer Resection

Abstract

BACKGROUND:

Rural populations experience higher colon cancer (CC) mortality than urban populations, and rural patients may have more complications after resection. Reoperations due to complications after CC resection among rural and urban patients are not clear, and factors mediating disparities in rural–urban postoperative outcomes have not been examined.

STUDY DESIGN:

Using the Surveillance, Epidemiology, and End Results Program–Medicare database, individuals with CC who underwent surgical resection were compared via multivariable logistic regression to determine the association of rurality with postoperative complications and reoperations at 30, 60, and 90 days. A 4-way effect decomposition (mediation analysis) assessed how surgical approach (open vs minimally invasive surgery [MIS]) mediated the relationship between rurality and complication risk. Survival was assessed at 30 days, 90 days, and 1 year.

RESULTS:

Among 33,183 patients, rural patients (n = 5,938; 18%) had lower Charlson Comorbidity Indices (p < 0.0001) and presented less emergently (p = 0.0307); stage did not differ from urban patients. Rural patients underwent more postdischarge reoperations (p = 0.0007) and had higher odds of surgical site infection (SSI; odds ratio [OR] 1.31, 95% CI 1.13 to 1.52) and anastomotic leak (OR 1.19, 95% CI 1.07 to 1.33). Rural individuals were less likely to undergo MIS (OR 0.67, 95% CI 0.63 to 0.72); MIS was associated with lower odds of SSI (OR 0.58, 95% CI 0.51 to 0.67) and leak (OR 0.57, 95% CI 0.52 to 0.63). Surgical approach mediated 29% of the overall complication risk, 18% of the SSI risk, and 28% of the anastomotic leak risk for rural patients. Finally, survival was lower among rural patients (30 days: p = 0.0313; 90 days: p = 0.0195; 1 year: p = 0.0063).

CONCLUSIONS:

Surgical approach mediated one-third of the complication risk experienced by rural patients, who had higher odds of complications, more postdischarge reoperations, and lower odds of accessing MIS. Expanding rural MIS colectomy practices may be one way to achieve more geographically equitable outcomes.

Colon cancer (CC) is the second leading cause of cancer-related death in the US, with a mortality rate of approximately 16 per 100,000 individuals.¹ Fortunately, CC survival has increased during the past 50 years along with advances in screening and treatment.² Despite improving survival, rural populations experience >20% higher CC mortality compared with urban populations, which has been hypothesized to be due to lower rates of screening, later stage at presentation, and higher rates of individual-level CC risk factors, such as smoking, obesity, and alcohol use among rural populations.^3,4 Provider shortages and closures of rural hospitals in recent years have further worsened disparities in access to CC prevention and treatment for individuals living in rural areas.^5–8 Given the known rural–urban disparities in CC mortality, identifying modifiable disparities along the disease course is important to target interventions.

Surgical resection is a mainstay of treatment, and although some literature suggests that rural patients access colon resection at similar rates to urban patients, rural–urban disparities in oncologic and survival outcomes persist.⁹ Specifically, rural patients are less likely to undergo adequate lymphadenectomy and more likely to present emergently.¹⁰ Rural and urban populations both experience postoperative morbidity and complications, including pneumonia, MI, sepsis, and anastomotic leaks (ALs), which are associated with increased mortality, readmissions, loss of preoperative independence, and healthcare costs.^11,12 Surgical complications such as deep surgical site infections (SSIs) and ALs may result in delays to chemotherapy, which is also associated with higher mortality.^13,14 SSI and AL are the most common indications for reoperation after colorectal surgery, but reoperations may have long-term clinical implications.^9,15,16 However, literature comparing SSI and AL outcomes between rural and urban patients is conflicting, and reoperations related to surgical complications have not been studied among rural and urban patients.^9,17 Additionally, mediators of rural–urban disparities in postoperative outcomes remain unexplored.

We sought to identify factors associated with rurality that may explain rural–urban postoperative disparities. We aimed to examine (1) medical and surgical postoperative complications among rural and urban patients after CC resection, (2) complication-related reoperations, and (3) mediation by surgical approach (open vs minimally invasive) on rural–urban complication disparities. We hypothesized that rural patients would undergo more reoperations and that the surgical approach would mediate the higher complication risk among rural populations.

METHODS

Study design, setting, and data source

We used the 2014 to 2019 Surveillance, Epidemiology, and End Results (SEER)–Medicare linked database, which contains claims data from 2013 to 2020 for Medicare beneficiaries with cancer.^18,19 SEER captures demographic and clinical data, and the linkage with Medicare claims data allows for evaluation of a wide range of services rendered to patients with cancer diagnoses. The Boston University IRB deemed this study exempt from review as data were retrospectively obtained.

Study participants

We included individuals with a diagnosis of CC as their first primary cancer, who underwent surgical resection between 1 month before and 6 months after diagnosis. Only those aged 66 years and older with continuous Medicare Parts A and B enrollment and without HMO insurance were included. We excluded those who underwent nonoperative management or surgery more than 6 months from diagnosis, those without a diagnosis date or a diagnosis date before January 2014, patients with a date of birth or death differing by more than 3 months between SEER and Medicare data, diagnoses from death certificates or autopsies, and those missing acuity of presentation. We excluded individuals missing acuity of presentation because emergent presentations have been associated with worse postoperative outcomes, and we included acuity as a covariate in multivariable models. Patient demographic data included age, sex, race (non-Hispanic White, non-Hispanic Black, Hispanic, Asian, or other), region of residence (Northeast, Midwest, South, or West), and Charlson Comorbidity Index (CCI), based on diagnoses made 1 year before diagnosis. Patients with no claims for CCI diagnoses within 1 year before diagnosis were classified as having a CCI of 0. Clinical and oncologic data included acuity of presentation (admission code type: elective, urgent or emergent), type of colectomy (open, minimally invasive), stage (see Supplemental Digital Content 1 at https://links.lww.com/JACS/A521 for definitions), and colonic location of cancer (right, transverse, left, and overlapping sites/unspecified combined). Patients with cancer of overlapping sites of the colon or unspecified sites were combined into 1 group for analysis, given low numbers in each group. Hospital characteristics included National Cancer Institute designation (comprehensive and clinical cancer centers grouped together) and status as an academic or community hospital.

Surgical resection

Surgical resections were identified using the ICD-9 and ICD-10 procedure codes obtained from the Medicare Provider Analysis and Review inpatient files.²⁰ We did not include individuals who only had carrier claims (National Claims History [NCH]). Patients with procedure codes for open or laparoscopic colectomy within the timeframe of 1 month before diagnosis until 6 months after diagnosis were included (Supplemental Digital Content 1, https://links.lww.com/JACS/A521). Because robotic colectomy does not have a separate ICD-9/−10 code and is coded as laparoscopic colectomy, we defined minimally invasive surgery (MIS) by ICD-9/−10 codes for laparoscopic colectomy (Supplemental Digital Content 1), https://links.lww.com/JACS/A521). If patients had multiple surgical claims within the defined timeframe, the most recent surgical resection was selected as the index operation.

Exposure

Rurality was defined by the 2013 US Department of Agriculture’s Rural–Urban Continuum Codes (RUCC), which classifies counties on a scale from 1 to 9, with 1 being the most urban and 9 being the most rural.²¹ RUCC classification is based on population size and adjacency to urban areas. Based on established rural–urban categorizations, we categorized RUCC 1 to 3 as urban, and RUCC 4 to 9 as rural.^2,22

Outcomes

The primary outcomes were any postoperative complication, AL, and SSI at 30, 60, and 90 days postoperatively. Postoperative complications included both medical and surgical complications: surgical complications included postoperative hemorrhage, SSI, AL, and “other,” which included diagnosis codes for unspecified surgical complications. AL was defined as an anastomotic breakdown, abdominopelvic abscess, or intestinal fistula.^23,24 Postoperative medical complications included pneumonia, pulmonary failure, MI, deep venous thrombosis/pulmonary embolus, and renal failure. The composite outcome of any postoperative complication included all of the above-mentioned complications.

Secondary outcomes included reoperations for surgical complications and survival. Inpatient reoperations were defined as any reexploration in the operating room, diverting ileostomy or colostomy, hemorrhage control procedure, drain placement, or abdominal washout within 30 days from index operation (Medicare Provider Analysis and Review) or within 30 days from discharge (NCH and outpatient files). In a subset analysis of postdischarge reoperations (NCH and outpatient files), drain placements were excluded a priori because these procedures typically occur in the inpatient setting. Survival was calculated as the number of days from the date of surgery until date of death or the last available date from the dataset for those who survived beyond our study window.

Statistical analysis

Data were analyzed from October 2024 to January 2025. We evaluated categorical variables with Pearson chi-square tests and continuous variables with the Wilcoxon rank-sum tests. A 2-tailed p value of 0.05 was considered statistically significant and SAS software version 9.4 (Cary, NC) was used. Multivariable logistic regression models were used to evaluate odds of (1) any complication, (2) SSI, and (3) AL; covariates included age (older than 75 years vs younger than 75 years), sex, race (White, non-White), stage (0 to III vs IV), rurality (urban, rural), region (Midwest, South, West vs Northeast), acuity of presentation (emergent, nonemergent), CCI (continuous), surgical approach (MIS vs open), hospital type (academic vs community), and site of resection (transverse, left, overlapping sites vs right). Odds of undergoing MIS were also evaluated using a multivariable model with the same covariates as above (except surgical approach). We examined collinearity between all variables using Eigenvalues, condition indices, and the proportion of variation; there was no indication of collinearity in our models. Logistic regression models excluded patients with unknown stage (n = 490). Survival rates at 30 days, 60 days, 90 days, and 1 year were modeled using Kaplan–Meier curves.

Effect decomposition

To examine the relationship between rurality (exposure), surgical approach (mediator), complications (outcome), and other covariates (age, sex, race, stage, acuity, and comorbidities), we performed a mediation analysis with hypothesized causal relationships depicted in Figure 1. Four-way mediation analysis (or 4-way effect decomposition) is used to untangle the causal relationships between exposures and outcomes and has been applied in studies examining clinical and health services outcomes.^25–28 For binary outcomes, such as for any complication, AL, and SSI, decomposition results are calculated as proportions of the relative risk experienced by rural populations that can be attributed to each of the 4 pathways described below. We calculated the proportion of rural–urban complication disparities (any, SSI, AL) mediated by surgical approach. In the context of this study, the 4-way effect decomposition included (1) the controlled direct effect (proportion of the rural–urban complication difference not mediated by surgical approach), (2) the pure indirect effect (proportion of the rural–urban complication difference attributable to surgical approach alone), (3) the reference interaction (interaction between rurality and surgical approach not related to mediation), and (4) the mediated interaction (proportion of the rural–urban disparity due to both mediation by surgical approach and interaction between rurality and surgical approach; Fig. 2). The pure indirect effect most closely represents a causal mediation pathway.²⁹ Age (older than 75 years vs younger than 75 years), sex, race (White, non-White), stage (0 to III, IV, unknown), acuity of presentation, CCI, and hospital type were included in effect decomposition models as covariates; CCI was categorized into 0 and 1 or more for these models.

Figure 1.

Directed acyclic graph (DAG) illustrating the interconnected relationships between the exposure (rurality), the mediator (surgical approach), the outcome (complication), and covariates included in our 4-way effect decomposition model. CCI, Charlson Comorbidity Index.

Figure 2.

Illustration of each of the calculations included in the 4-way effect decomposition (ie mediation analysis) results as described in the text.

RESULTS

Patient characteristics

We identified 33,183 patients who met the inclusion and exclusion criteria (urban = 27,245 [82%], rural = 5,938 [18%]). Rural patients were younger (p < 0.0001), more often men (p = 0.0022), White (p < 0.0001), and had lower CCIs (p < 0.0001). Site of cancer, stage, and receipt of chemotherapy were similar between the groups. Rural patients presented less emergently (36.8% vs 38.3%, p = 0.0307), but rural patients underwent more open colectomies (69.7% vs 61.1%, p < 0.0001), whereas urban patients underwent more MIS (38.9% vs 30.3%, p < 0.0001). Rural patients were less often treated at academic centers (p < 0.0001) and National Cancer Institutes (p < 0.0001; Table 1). Rates of MIS were similar across time among rural and urban populations each year of the study period (Supplemental Digital Content 2, https://links.lww.com/JACS/A521).

Table 1.

Population Demographic and Clinical Characteristics

Patient characteristic	Overall (N = 33,183)	Urban (N = 27,245; 82%)	Rural (N = 5,938; 18%)	p Value
Age, y, median (interquartile range)	77 (71–83)	77 (71–83)	76 (71–82)	<0.0001
Sex, n (%)				0.0022
Male	14,723 (44.4)	11,982 (44.0)	2,741 (46.2)
Female	18,460 (55.6)	15,263 (56.0)	3,197 (53.8)
Race and ethnicity, n (%)				<0.0001
Non-Hispanic White	26,092 (78.6)	20,880 (76.6)	5,212 (87.8)
Non-Hispanic Black	2,643 (8.0)	2,357 (8.7)	286 (4.8)
Hispanic	2,650 (8.0)	2,320 (8.5)	330 (5.5)
Asian	1,527 (4.6)	1,478 (5.4)	49 (0.8)
Other race	271 (0.8)	210 (0.8)	61 (1)
Charlson Comorbidity Index, median (range)	1 (0–13)	1 (0–13)	1 (0–11)	<0.0001
Emergent operation, n (%)	12,631 (38.1)	10,444 (38.3)	2,187 (36.8)	0.0307
Region, n (%)				<0.0001
Northeast	9,532 (28.7)	8,946 (32.8)	586 (9.9)
Midwest	2,830 (8.5)	1,918 (7.0)	912 (15.4)
South	11,930 (36.0)	8,484 (31.1)	3,446 (58.0)
West	8,891 (26.8)	7,897 (29.0)	994 (16.7)
Site, n (%)				0.1000
Right colon	19,999 (60.3)	16,481 (60.5)	3,518 (59.3)
Transverse colon	3,500 (10.6)	2,890 (10.6)	610 (10.3)
Left colon	8,801 (26.5)	7,162 (26.3)	1,639 (27.6)
Overlapping sites/unspecified	883 (2.7)	712 (2.6)	171 (2.9)
Stage, n (%)				0.3345
Stage 0	797 (2.4)	648 (2.4)	149 (2.5)
Stage I	9,047 (27.3)	7,433 (27.3)	1,614 (27.2)
Stage II	12,641 (38.1)	10,408 (38.2)	2,233 (37.6)
Stage III	6,595 (19.9)	5,432 (19.9)	1,163 (19.6)
Stage IV	3,613 (10.9)	2,936 (10.8)	677 (11.4)
Stage unknown	490 (1.5)	388 (1.4)	102 (1.7)
Procedure, n (%)				<0.0001
Open colectomy	20,786 (62.6)	16,645 (61.1)	4,141 (69.7)
Minimally invasive colectomy	12,397 (37.4)	10,600 (38.9)	1,797 (30.3)
Hospital type, n (%)				<0.0001
Academic	19,890 (59.9)	17,277 (63.4)	2,613 (44.0)
National Cancer Institute	2,679 (8.1)	2,371 (8.7)	308 (5.2)

Complications at 30, 60, and 90 days postoperatively (univariate)

Rural patients experienced a higher rate of overall complications at 30 days (24.8% vs 22.4%, p < 0.0001), 60 days (28.2% vs 25.9%, p = 0.0003), and 90 days (30.2% vs 28.0%, p = 0.0005). At 30 days, rural patients experienced higher rates of pneumonia (4.8% vs 3.8%, p = 0.0001), MI (1.7% vs 1.1%, p = 0.0003), SSI (4.7% vs 3.4%, p < 0.0001), AL (8.4% vs 7.0%, p = 0.0002), and other surgical complications (3.1% vs 2.6%, p = 0.0273). Rates of respiratory failure, deep venous thrombosis/pulmonary embolus, renal failure, and hemorrhage were similar at 30, 60, and 90 days (Fig. 3A), with the exception of higher rates of hemorrhage noted at 60 days (p = 0.0397) and 90 days (2.0% vs 1.6%, p = 0.0279) in rural patients (Fig. 3B).

Figure 3.

Postoperative morbidity and complications at 30 and 90 d postoperatively. (A) Postoperative morbidity and complication at 30 d. (B) Postoperative morbidity and complication at 90 d. Complications with p value of <0.05 on chi-square analyses highlighted with boxes above bars containing p values. DVT/PE, deep vein thrombosis/pulmonary embolism; SSI, surgical site infection.

Rates of postoperative reinterventions

Within 30 days postoperatively or 30 days from discharge, 4% of patients underwent a reoperation (reexploration in the operating room, abdominal washout, drain placement, diverting ostomy, hemorrhage control). There were similar rates of inpatient reinterventions, but higher rates of postdischarge reoperations in rural patients (0.6% vs 0.3%, p = 0.0007; Supplemental Digital Content 3, https://links.lww.com/JACS/A521).

Factors associated with complications (multivariable, 30 days)

Any complication

Men (odds ratio [OR] 1.25, 95% CI 1.19 to 1.32), patients aged 75 years and older (OR 1.23, 95% CI 1.17 to 1.30), those with late-stage disease (OR 1.24, 95% CI 1.15 to 1.35), emergent presentation (OR 1.11, 95% CI 1.05 to 1.18), higher CCI (OR 1.16, 95% CI 1.14 to 1.17), left colon resection (OR 1.08, 95% CI 1.02 to 1.15), or rural residence (OR 1.10, 95% CI 1.02 to 1.18) had higher odds of having any postoperative complication. Individuals from the Midwest (OR 1.24, 95% CI 1.12 to 1.38), South (OR 1.14, 95% CI 1.06 to 1.22), and West (OR 1.19, 95% CI 1.11 to 1.29) had higher odds of complications compared with those from the Northeast as well. Race and hospital type were not significant (Fig 4A).

Figure 4.

Multivariable logistic regression results for any complication, surgical site infection (SSI), and anastomotic leak. Individuals with unknown stage were excluded from multivariable models. (A) Sociodemographic and clinical factors associated with odds of experiencing any complication. (B) Sociodemographic and clinical factors associated with odds of experiencing a SSI. (C) Sociodemographic and clinical factors associated with odds of experiencing an anastomotic leak. Error bars represent 95% CI. CCI, Charlson Comorbidity Index (modeled continuously); LCL, lower control limit; MIS, minimally invasive surgery; OR, odds ratio; UCL, upper control limit.

SSI

Odds of experiencing an SSI were higher among patients with late stage (OR 1.28, 95% CI 1.08 to 1.51), emergent presentation (OR 1.22, 95% CI 1.08 to 1.38), rural residence (OR 1.31, 95% CI 1.13 to 1.52), higher CCI (OR 1.06, 95% CI 1.03 to 1.09), and left colon resections (OR 1.29, 95% CI 1.13 t o1.47), whereas patients aged 75 years and older had lower odds of SSI (OR 0.81, 95% CI 0.72 to 0.91); sex, race, region, and hospital type were not significant (Fig 4B).

Anastomotic leak

Men (OR 1.18, 95% CI 1.09 to 1.29), those with late-stage cancer (OR 1.26, 95% CI 1.12 to 1.42), emergent presentation (OR 1.25, 95% CI 1.15 to 1.37), and rural residence (OR 1.19, 95% CI 1.07 to 1.33) had higher odds of leak. Individuals undergoing resection of the transverse colon (OR 1.18, 95% CI 1.03 to 1.35), left colon (OR 1.21, 95% CI 1.10–1.33), overlapping and unspecified sites (OR 1.31, 95% CI 1.02 to 1.67), and from the West (OR 1.16, 95% CI 1.03 to 1.30) also had higher odds of AL. Age, race, CCI, and hospital type were not significant risk factors for AL (Fig 4C).

Factors associated with minimally invasive surgery

Patients who underwent MIS had lower odds of experiencing any complication (OR 0.67, 95% CI 0.63 to 0.71), SSI (OR 0.58, 95% CI 0.51 to 0.67), and AL (OR 0.57, 95% CI 0.52 to 0.63; Fig. 4). Patients at academic centers were more likely to undergo MIS (OR 1.22, 95% CI 1.16 to 1.28), whereas older (OR 0.89, 95% CI 0.85 to 0.93), rural (OR 0.67, 95% CI 0.63 to 0.72), and those with late-stage disease (OR 0.44, 95% CI 0.41 to 0.48), emergent presentation (OR 0.36, 95% CI 0.34 to 0.38), higher CCI (OR 0.98, 95% CI 0.96 to 0.99), transverse (OR 0.59, 95% CI 0.54 to 0.64), left (OR 0.68, 95% CI 0.64 to 0.72), or overlapping and unspecified sided cancer (OR 0.68, 95% CI 0.58 to 0.79), and those from the Midwest (OR 0.78, 95% CI 0.71 to 0.85) and South (OR 0.90, 95% CI 0.84 to 0.95) were less likely to undergo MIS (Fig. 5).

Figure 5.

Odds of undergoing minimally invasive surgery (MIS); error bars represent 95% CI. Individuals with unknown stage excluded from multivariable models. LCL, lower control limit; OR, odds ratio; UCL, upper control limit.

Mediation analysis

Compared with urban patients, rural patients had a 15.8% higher risk of experiencing any complication, a 40.9% higher risk of SSI, and a 23.5% higher risk of AL. Surgical approach mediated 28.9% of the rural–urban complication disparity, 17.8% of the SSI disparity, and 27.7% of the leak disparity. The pure indirect effect, which is the proportion of the rural–urban complication disparity mediated by the surgical approach alone and most closely represents a causal pathway, was 25.9% for any complication, 13.3% for SSI, and 23.7% for AL (Table 2).

Table 2.

Mediation Analysis (Effect Decomposition) of the Proportion of the Higher Risk of Complication Among Rural Populations as Mediated by Minimally Invasive Surgery

Complication	% Excess risk*	% Mediated†	% Controlled direct effect‡	% Mediated interaction§	% Pure indirect effect¶
Complication (any)	15.8	28.9	83.1	2.9	25.9
Surgical site infection	40.9	17.8	100.56	4.5	13.3
Anastomotic leak	23.5	27.7	88.7	4.0	23.7

Models included age, sex, race, stage, acuity, Charlson Comorbidity Index (categorized as 0 vs ≥1), and hospital type.

^*Percent excess risk: percent higher risk of complication associated with rural residence.

^†Percent mediated: percent of the relationship between rurality and complication risk mediated by MIS.

^‡Percent controlled direct effect: percent of the rural–urban complication difference not mediated by MIS.

^§Percent mediated interaction: percent interaction between rurality and MIS as compared with urban patients undergoing MIS.

^¶Percent pure indirect effect: percent of the rural–urban complication difference attributable to MIS alone.

MIS, minimally invasive surgery.

Survival

Rural individuals had lower survival at 30 days (94.2% vs 94.9%, p = 0.0313), 90 days (90.3% vs 91.2%, p = 0.0195), and 1 year (80.3% vs 81.8%, p = 0.0063) postoperatively compared with urban patients (Supplemental Digital Content 4, https://links.lww.com/JACS/A521).

DISCUSSION

To our knowledge, this is the first study to examine surgical approach as a mediator of the complex relationship between rurality and postoperative outcomes after CC resection. Despite presenting less emergently, rural patients had a lower likelihood of undergoing MIS, which mediated a significant proportion of the complication risk associated with rural residence. Additionally, rural patients underwent double the rate of reoperations after discharge home compared with urban patients, which has serious implications for the rural healthcare ecosystem, given that rural patients tend to live further from care.^30,31 Finally, rural patients also had lower survival 1 year after surgery, despite presenting at similar stages as urban patients.

Our results establish the surgical approach as a potential target for interventions aimed at reducing rural–urban disparities in postoperative outcomes after CC resection, as the surgical approach mediated up to 29% of the complication disparities in this study. In concert with earlier literature, we found MIS to be associated with substantially reduced odds of SSI and AL; however, we also found that rural patients were less likely to undergo MIS compared with their urban counterparts.^9,17,32–35 Given the rural patients’ disparate access to MIS despite its known advantages, we analyzed the proportion of the complication risk experienced by rural individuals that could be attributed to their lower likelihood of undergoing MIS. Rural individuals had a 40.9% higher risk of SSI and a 23.5% higher risk of AL, and 13.3% and 23.7% of those risks, respectively, can be attributed to rural patients’ lower likelihood of MIS. These findings suggest that expanding access to MIS among rural populations may help reduce postoperative complication disparities. This could involve training rural surgeons in MIS modalities as well as expanding MIS infrastructure and robotic platforms at rural hospitals, which may lead to more geographically equitable care.

Patients undergoing minimally invasive colectomy are more likely to have a fellowship-trained colorectal surgeon (CRS) compared with patients undergoing open colectomy, suggesting a potential benefit to employing fellowship-trained CRS at rural hospitals.^35,36 Additionally, patients who are high-risk for complications may benefit from referrals to centers comfortable with minimally invasive resections. Although earlier work suggests a survival advantage for rural patients traveling further distances to high-volume centers, rural patients may face transportation challenges, and traveling for care is not always an option.^10,37 Patients tend to seek care at the hospitals closest to their homes, which often means rural patients receive cancer care at low-volume centers.^38–40 Rather than placing the burden on rural individuals to travel to urban centers for care, the disparities in postoperative outcomes we observed could be addressed by expanding access to MIS for CC resections at rural hospitals.

This study provides new insights into complication-related reoperations in both the inpatient and outpatient settings. Our study’s rural patients experienced more ALs and SSIs (including deep and organ-space) compared with urban patients, and they underwent double the rate of postdischarge reoperations, including abdominal washouts and diversions. This finding is particularly concerning, given that rural patients tend to live further from hospitals and may therefore re-present later in the course of a complication.^30,31 Discharge disposition has been associated with patient outcomes, including readmission and mortality, and discharging patients to the appropriate setting can improve these outcomes.^31,41,42 Rural individuals are less likely to be discharged to an inpatient rehabilitation facility compared with urban individuals, and rural patients undergoing colectomy at critical access hospitals are less likely to be discharged with short-term care services after suegery.^17,42 Within the context of these earlier works, our results suggest that expanding home healthcare services and structured discharge planning in rural areas may reduce the disparity observed in postdischarge reoperations.

Existing literature on postoperative complications after CC resection among rural and urban patients is conflicting, with some studies finding worse outcomes among rural patients and others finding no difference when controlling for covariates.^9,10,17 Our rural population was overall younger, had CCIs skewing lower, and presented less emergently as compared with their urban counterparts; these factors were generally associated with lower odds of complications in our multivariable models, but rural individuals still had more postoperative complications and worse survival. Some literature attributes rural patients’ lower overall survival at 5 and 10 years after CC resection to surgery at low-volume centers.¹⁰ Other proposed reasons for rural–urban survival disparities include more emergent presentations and later stages at presentation among rural patients, but the rural patients in our SEER–Medicare cohort presented less emergently and at similar stages compared with their urban counterparts.^10,16 Therefore, the lower survival rate we observed among rural individuals may be due to other complex, intersecting barriers to accessing healthcare that are difficult to quantify and are not accounted for within our covariates.^43,44

Future implications

Beyond expanding MIS at rural hospitals, rural surgical CC care could be supported in several ways. The rural–urban CRS workforce disparity has been widening over time, with fewer CRSs available per capita in rural communities.⁶ Visiting clinics, in which specialty surgeons regularly travel to underserved or rural areas for clinics, as well as telehealth visits with CRSs could be expanded to improve access to specialty colorectal care.⁶ Additionally, multidisciplinary tumor boards, which are associated with higher rates of appropriate diagnostic workup, increased clinical trial enrollment, and improved survival, can be held in the virtual setting^45–47; including rural communities in virtual multidisciplinary tumor boards could increase access to specialized CC care in these underserved regions. Visiting clinics and virtual tumor boards have been used at rural hospitals to manage oropharyngeal and hepatocellular cancers, and implementation may be feasible for CC as well.^45,48 Finally, as multimodal prehabilitation programs are associated with reduced postoperative complications, rural patients undergoing elective CC resection may benefit from prehabilitation strategies including nutritional counseling, supervised exercise, psychological support, and smoking cessation.⁴⁹ Literature suggests that virtual prehabilitation programs are feasible and acceptable to patients, including those in rural and underserved areas.^50,51 Visiting clinics, virtual multidisciplinary tumor boards, and multimodal prehabilitation (virtual or in-person) represent interventions that, in addition to expanding MIS in rural regions, could reduce the rural–urban outcome disparities we observed in this study.

Limitations

Because we used the SEER–Medicare linked dataset, our population did not include patients aged 65 years and younger and only included individuals with Medicare. Additionally, ICD-9/−10 colectomy codes do not detail whether an ostomy or anastomosis was performed at the index operation, nor whether anastomoses were performed intracorporeally vs extracorporeally, but because the standard of care for a CC resection includes an anastomosis with or without a diverting ostomy, we included all individuals in our complications analyses. Our data also do not capture information on travel distance, factors related to patient choice of hospital for surgery, surgeon specialty, or chemotherapy usage. Finally, although mediation analyses included age, sex, race, stage, CCI, acuity of presentation, and hospital type, many other factors not captured in these data may mediate rural patients’ worse outcomes. For example, hospitals with greater MIS capabilities may also be higher-volume and have resources such as enhanced recovery after surgery protocols and SSI-reduction bundles, which are associated with lower rates of complications.^52–54

CONCLUSIONS

In this study of patients who underwent CC resection, we found that surgical approach accounted for approximately 30% of the rural–urban disparity in postoperative complications. MIS was associated with lower odds of complications, including SSI and AL, but rural patients were less likely to undergo MIS. Additionally, despite having fewer comorbidities, presenting less emergently, and presenting at similar stages to urban patients, rural patients had worse postoperative outcomes and 1-year survival. Delivering high-quality surgical care to rural populations is a complex issue, but rural individuals undergoing colectomy for cancer could benefit greatly from the expansion of minimally invasive colectomy services at rural hospitals.

Supplementary Material

Supplemental digital content is available for this article.

Abbreviations and Acronyms

AL

anastomotic leak

CC

colon cancer

CCI

Charlson Comorbidity Index

CRS

colorectal surgeon

MIS

minimally invasive surgery

NCH

National Claims History

SEER

Surveillance, Epidemiology, and End Results Program

SSI

surgical site infection