Ileostomy formation, although effective in reducing the morbidity related to colorectal complications, is associated with the highest readmission rates from colorectal procedures.1–4 In fact, ileostomies matched kidney transplants for the highest 30-day all-cause readmission rate at 29.1% readmitted, placing a huge burden on the health care system.5 In 2010, there were 23,392 index hospital stays with ileostomy creation in the United States. This presents an opportunity for substantial improvement.
Dehydration is the most common cause of readmission present in approximately 40% of postileostomy readmissions and has significant morbidity for patients.6 , 7 Furthermore, ileostomy-related morbidity and readmissions can impact adjuvant treatment in cancer settings because they have been shown to be associated with delayed adjuvant chemotherapy in patients with rectal cancer.8 Therefore, preventing dehydration and subsequent readmission is a goal that can improve morbidity, health care utilization, and even cancer survival.
Although there has been recent interest in understanding and preventing readmissions after ileostomy creation, many unanswered questions remain. To date, predictors of readmission for dehydration rather than all-cause readmission are limited in the literature. All-cause readmissions have been predicted by factors that are not amenable to prevention postdischarge, such as serious in-hospital complications during the initial admission, whereas other predictors such as formation of loop versus end ileostomy and length of stay are more amenable to intervention and are more likely to be associated with dehydration.7 As such, some factors that predict all-cause readmission may not be applicable for identifying patients specifically at risk for readmission with dehydration. In addition, patients who have had previous ileostomies have been excluded in previous analyses. Although it could be that patients who have had an ileostomy may feel experienced and knowledgeable with regard to stoma management and hydration, thereby being less likely to be readmitted, this has never been reported. Therefore, there are important questions that still need to be addressed in an effort to understand which patients are at high risk of readmission with dehydration to develop targeted interventions to prevent this common event.
The goal of this study was to evaluate the factors associated specifically with 30-day readmissions with dehydration, rather than all-cause readmissions. The identification of dehydration-specific readmission predictors is warranted to create ileostomy-dehydration programs with evidence-based target patients that can lead to decreased ileostomy-related morbidity. We expected dehydration to be the leading cause of readmission, and a priori, we hypothesized that patient factors available before discharge would be associated with increased risk of readmission for dehydration, and that a history of an ileostomy would decrease the risk of readmission for dehydration.
All patients aged ≥18 who underwent creation of an ileostomy from January 1, 2014, through December 31, 2016, by surgeons within the Colorectal Surgery Division of our tertiary-care center were identified. Patients were all enrolled in standardized pathways for their postoperative care, and they had inpatient and outpatient wound, ostomy and continence nursing care as well. During the study period, an enhanced recovery after surgery protocol was implemented at our institution. Patients were identified from the division inpatient billing list (n = 2739) through the use of Current Procedural Terminology codes and/or procedure names and, ultimately, verification through operative note review. Patients with a preexisting ileostomy on the day of surgery that was not recreated during the procedure according to the operative note were excluded. Otherwise, patients with a history of an ileostomy, whether in the past or currently, were included if a new ileostomy was created (ie, end ileostomy converted to loop or small-bowel resection including existing ileostomy with new ileostomy creation). Of 273 patients identified, 10 died during the index admission, and, therefore, could not be readmitted within 30 days of discharge, and are excluded. One patient was excluded because of a missing laboratory value for a final cohort of 262 patients.
Each medical record was meticulously examined in chronological order by both a surgically trained reviewer and a research assistant. Study variables were captured by using REDCap (Research Electronic Data Capture, Vanderbilt University).9 A methodical inquiry of each patient’s chart was performed, including phone encounters and scanned-in documents from encounters at other institutions, to account for patient, operative, postoperative, and, if present, readmission factors.
Patient factors captured were age, sex, Medicaid insurance at the time of surgery, patient zip code, race, marital status, smoking status, presence of Charlson Comorbidities, BMI, history of ileostomy, preoperative creatinine, and chronic use of diuretics with or without discharge on diuretics. The date of surgery was the date of ileostomy creation, whether during the index procedure or as a return to the operating room for a complication from a previous procedure. Operative factors captured were elective surgery versus nonelective, indication for surgery, operative approach, type of surgery, and surgical wound contamination classification. Elective admissions were planned before the admission; an elective patient was occasionally admitted 1 to 2 days before surgery. Nonelective surgeries included urgent and emergent cases that were not planned before admission. In the postoperative course, data capture included complications during the index admission (ileus/small-bowel obstruction; superficial, deep or organ space surgical site infection; anastomotic leak; urinary tract infection; pneumonia; sepsis; blood transfusion; venous thromboembolism; acute kidney injury; return to the operating room; myocardial infarction or stroke/transient ischemic attack), Alvimopan use (none, preoperative use only, postoperative use only, or both), and creatinine (postoperative day 1 and on day of discharge). Days from surgery until ileostomy output was defined by output >250/24 hours.7 Alvimopan timing varied according to elective versus nonelective status as well as postoperative course. High ileostomy output was defined as a recorded 24-hour output of ≥1500 mL requiring medical intervention (intravenous fluid bolus, addition of antimotility agents, and/or high output was the sole reason preventing discharge per surgical team note) at any point throughout the hospitalization, but not necessarily immediately before (24 or 48 hours) discharge. Length of stay (LOS) after ileostomy creation and discharge disposition location were also captured.
Readmission was defined as a readmission within 30 days of discharge from the index ileostomy creation admission; readmissions included both inpatient and observation stays that may be up to 2 midnights long. Days to first readmission was calculated from discharge and readmission date. The cause or causes of readmission were abstracted from the chart review, and strict criteria were used to define dehydration. Dehydration was defined using strict and objective clinical criteria; a subjective patient report of dehydration without corresponding objective data was not recorded as dehydration. Dehydration could be diagnosed by urine electrolyte abnormality at readmission (urine sodium <20 mEq/L, fractional excretion of sodium <0.5%, or urine osmolality >450), or by clinical criteria by admitting physicians (diagnosis of dehydration, high stoma output described, or poor oral intake or vomiting described) and an objective sign of dehydration at the time of readmission (blood urea nitrogen/creatinine ratio >20, creatinine ≥150% of baseline, bicarbonate <20, hyperkalemia >5.0 or hypokalemia <3.4, hypernatremia >149 or hyponatremia <133, sinus tachycardia >110, mean arterial pressure <65, or documented orthostatic hypotension).7
We opted to power for a small effect size given that this was exploratory and designed for quality improvement implementation. For regression equations using 6 or more predictors, approximately 30 participants per predictor are needed for sufficient power to detect a small effect size.7 , 10 With a sample size of 262, we were powered to use up to 8 predictors.
Bivariate analysis was performed by comparing factors by outcome using the χ2 test and t test as appropriate. Factors with a p value <0.1 were entered manually into the multivariable logistic regression for 30-day readmission due to dehydration. Analysis was performed using SAS Software (Version 9.3, SAS Institute, Cary, NC). The study was approved by the University of Rochester Medical Center Institutional Review Board.
Patient characteristics are described in Table 1. Seventy-five percent of patients were aged <65 (mean age, 54), 53% were male, and 14% had a history of ileostomy. The most common reason for ileostomy creation was colorectal cancer (32%). Twelve ileostomies (5%) were created as stand-alone procedures. A minimally invasive approach was intended in 56% of cases. Thirty-four percent of ileostomies were created as part of nonelective cases. In comparison with elective cases, nonelective cases did not significantly vary in median age (55.6 vs 56.5, p = 0.446), median number of comorbidities (1 vs 1, p = 0.976), and having Medicaid insurance (14% vs 13%, p = 0.821). High ileostomy output during the index admission was diagnosed in 26% of cases, and 18% of all cases had a creatinine >1.0 on discharge day. All patients had 30 days of follow-up postdischarge.
Among all patients who underwent ileostomy creation, the rate of readmission with dehydration was 11%, whereas the all-cause readmission rate was 30%. Characteristics of patients stratified by 30-day readmission with dehydration or not are described in Table 2. Mean days to readmission for any cause was 8.5, whereas for dehydration it was 11.6 days, although the most common time to readmission in both groups was 3 days.
Of the readmissions, 37% were readmitted with a diagnosis of dehydration, and dehydration was the sole reason for readmission in 26% of all those readmitted (Table 3). Among all readmissions, the median readmission LOS was 6 days (interquartile range, 3–11), and the most common LOS was 3 days. Among those readmitted for dehydration, the median readmission LOS was 6 days (interquartile range, 4–10), and the most common LOS was 2 days. Of patients readmitted, 22% had an intraperitoneal infection, 9% had an extraperitoneal infection, and 10% had an ileus/small-bowel obstruction. Among patients readmitted, 24% required interventional radiology drainage. Nine patients were readmitted with dehydration and another reason; most commonly, 5 patients had an intra-abdominal infection underlying their fluid/electrolyte imbalance. One patient’s dehydration was driven by abdominal pain in the setting of pulmonary embolism with new extensive inferior vena cava thrombus in the setting of a preexisting inferior vena cava filter. One patient had rib osteomyelitis and chest wall abscess as the reason for the index admission during which the patient happened to develop sigmoid volvulus; at readmission, the primary reason was persistent chest wall infection. One patient was discharged home with intravenous fluids through a central line, and readmission was primarily due to a central line infection and secondary dehydration. Of these patients readmitted with dehydration and an additional reason for readmission, dehydration was the ultimate cause for 1 patient who had a subsequent demand-mediated myocardial infarction; this patient’s index admission was due to obstructing colon cancer that was metastatic on pathology.
In bivariate analysis, age, sex, history of an ileostomy, a diagnosis of high ostomy output during the index admission, diuretics at home, and a discharge creatinine >1.0 were associated with readmission with dehydration. A change in postoperative creatinine >0.5 mg/dL in comparison with baseline was not associated with dehydration. Although indication for surgery was not significantly related with readmission for dehydration, some variation is seen across different surgery types. For instance, 21% of IPAA, 12% of partial colectomies, and 25% of small-bowel resections/lysis of adhesions were readmitted with dehydration. However, 0% of total proctocolectomy with end ileostomy, 5% of low anterior resections, and 6% of total colectomies were readmitted with dehydration. Patients with a history of an ileostomy were more likely to have specific procedures: 29% had a total proctocolectomy with end ileostomy, 25% IPAA, and 25% small-bowel resections/lysis of adhesions.
In multivariable logistic regression, 30-day readmission with dehydration was independently associated with older age (65+), male sex, history of an ileostomy, high ileostomy output during index admission, and a discharge creatinine >1.0. If the analysis is performed with outcome as readmission only for dehydration (yes/no), similar results are obtained leading to the same inferences (Table 4).
After careful evaluation of 262 patients who underwent ileostomy creation, we found an all-cause readmission rate of 30% within 30 days of discharge. Among those readmitted, dehydration was the most common cause at 37% and was the sole reason for readmission in 26%. These findings are in concordance with data from the past decade demonstrating an overall postileostomy all-cause readmission rate of 28% to 35% and dehydration to be the most common reason for readmission present in 41% to 44% of patients who are readmitted.7 , 11 , 12
One of the most interesting findings was that a history of an ileostomy was not associated with a decreased risk of readmission with dehydration, which we had hypothesized, but was actually associated with an increased risk. Numerous studies have looked at readmissions after ileostomy formation and have excluded patients with a history of an ileostomy without established justification.7 , 11–13 Although the reason for this may be that patients with history of an ileostomy are expected to be more knowledgeable and capable of managing stomas and hydration status, no direct evidence to support this exists. Meanwhile, other studies have not excluded such patients, but have also not evaluated this risk factor.8 , 14 , 15 Yet, a new ileostomy is likely to be more proximal, which could lend to increased dehydration.14 This is important to consider given that ileostomy dehydration programs and clinical practice guidelines are targeted toward new ileostomates only.12 , 16
Another compelling finding was that high ileostomy output at any time during the index admission was associated with a 3-fold increased risk of readmission for dehydration. This is meaningful because previous studies have looked at stoma output at 24 or 48 hours before discharge but have not found any significance.7 , 14 Similarly, if we run our logistic regression model using continuous ileostomy output within 24 hours of discharge, this variable is not significant in our model. Yet, when high ileostomy output does occur, it is usually addressed and decreased before discharge, thus, leading to what is expected to be a safe discharge. Hence, we believe that using the diagnosis of a high ileostomy output requiring an intervention during the index admission, which was present in 55% of readmissions with dehydration, better captures the ileostomy-related burden of each patient, rather than the immediate predischarge volume that is expected to be safe. This could be used as a target for aggressive follow-up after discharge.
Male sex was also an independent predictor of readmission with dehydration. Although previous ileostomy-specific studies have not shown an association between male sex and overall or dehydration-specific readmissions, multiple studies have shown that, of those readmitted, a larger proportion are men at up to 59%.7 , 14 Similarly, a study of 42,348 patients who underwent colectomy for cancer found that male sex, as well as stoma formation, were independent risk factors for 30-day readmissions.17 Although this may seem surprising, the fact that men tend to be have worse health outcomes than women resonates in the literature, and worse outcomes in men are thought to reflect factors, including behaviors associated with male norms of risk taking and adventure, health behavior paradigms related to masculinity, and the fact that men are less likely to visit a doctors when they are ill, and when they see a doctor, men are less likely to report symptoms of illness.18
Results from other articles vary widely (Table 5), likely because of small sample sizes, inconsistent variable definitions and capture, and various follow-up times. Given that patients are generally optimized before discharge, identifying those who are more likely to fail in the outpatient setting is of utmost importance to implement targeted interventions. Our data suggest that, even with an enhanced recovery program, patients with ileostomies remain at high risk for readmission. Active prevention of dehydration in patients identified predischarge could change the natural history of ileostomy dehydration in the postdischarge period. Because of this issue, hospitals have begun implementing ileostomy care pathways to reduce readmissions due to dehydration. A recent single-center prospective study showed that by instituting a pathway consisting of preoperative teaching, standardized education materials, in-hospital engagement, observed management, and postdischarge tracking of intake and output, 30-day readmission rates in patients with ileostomies dropped from 35.4% to 21.4% overall, and from 15.5% to 0% for patients readmitted with dehydration.12 However, such markedly improved results have not been replicated. Another institution showed a decrease of 58% in their 30-day overall readmission rate for patients undergoing ileostomy formation after implementing a pathway consisting of regular home visits by a visiting nurse agency.15 Both of these studies did not risk-adjust for patient and hospitalization characteristics. It is unclear what the best intervention is, because existing program evaluations have been small, not risk-adjusted, and not replicated. Nevertheless, we believe the outpatient component of such programs is where the most improvement remains to be created, and we have identified 5 risk factors (history of ileostomy, older age, male sex, high ileostomy output during index admission, and a discharge creatinine >1.0) to identify patients prior to discharge that may benefit from targeted programs. Perhaps engaging in telemedicine for close outpatient follow-up and intervention may be part of the solution.19
This study’s strengths are founded on detailed data collection of contemporary patients with a priori definitions that can readily be replicated. This is the first study to include comprehensive patient-level information, including distance to hospital, marital status, and a history of an ileostomy, all of which can impact stoma and hydration management. Of importance is the fact that the most common readmission LOS was 2 days; as such, a brief search of in-patient readmissions would not have captured the true burden of ileostomy complications and morbidity, because many of these patients would be categorized as observation status and may not be captured in basic readmission searches. However, observation stays are still a burden to the health care system and, even more importantly, to the patient experience.
Our study is not without limitations. Our retrospective data rely on the accuracy of the chart and its comprehensiveness; however, electronic medical records had been fully implemented in outpatient and inpatient settings for 2 years before the start of the study period, and completeness was only a problem for 1 patient. Readmissions to outside institutions could have been missed if records of such encounters or related phone conversations were not documented; nonetheless, phone-call documentation and scanned-in documents were queried for readmission information. Bias in classifying variables was limited by collection in chronological order; as such, study personnel did not know the outcome while capturing index admission variables. Although our study addresses a ubiquitous issue, our specific effect estimates are limited in generalizability because of the wide variation in care practices across hospitals and surgeons. Last, the data were collected primarily for quality improvement implementation, but nevertheless yielded granular information.
Notwithstanding these limitations, our study does identify a history of ileostomy as a novel risk factor for readmission with dehydration. Such patients have often been excluded from readmission programs and evaluations, but our data suggest that they would benefit. The other risk factors identified, older age, male sex, high ileostomy output during index admission, and a discharge creatinine >1.0, are readily available, objective measures that can be used to identify patients at risk for readmission with dehydration and future quality improvement initiatives.
Readmissions are common after ileostomy creation and are most often due to dehydration. Although dehydration efforts have focused on patients with new ileostomies, our data show that patients with a history of an ileostomy are actually at risk for readmission for dehydration. Further studies aimed at the reduction of readmission with dehydration after ileostomy are warranted and should include patients with a history of an ileostomy.
The authors would like to thank Margaret M. Odhner, NP, for her dedication in providing evidence-based care and education to inpatient ileostomates. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Center for Research Resources of the National Institutes of Health.
1. Tyler JA, Fox JP, Dharmarajan S, et alAcute health care resource utilization for ileostomy
patients is higher than expected. Dis Colon Rectum. 2014;57:1412–1420.
2. Maroney S, Chavez de Paz C, Duldulao M, et alComplications of diverting ileostomy
after low anterior resection for rectal carcinoma. Am Surg. 2016;82:1033–1037.
3. Hawkins AT, Dharmarajan S, Wells KK, Krishnamurty DM, Mutch MG, Glasgow SCDoes Diverting loop ileostomy
improve outcomes following open ileo-colic anastomoses? A nationwide analysis. J Gastrointest Surg. 2016;20:1738–1743.
4. Hensley BJ, Cooney RN, Hellenthal NJ, et alUpstate New York Surgical Quality Initiative Collaborative. Readmissions after colectomy: the upstate New York Surgical Quality Initiative Experience. Dis Colon Rectum. 2016;59:419–425.
5. Weiss AJ, Elixhauser A, Steiner CReadmissions to U.S. Hospitals by Procedure, 2010: Statistical Brief #154. In: Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. 2006. Rockville, MD: Agency for Healthcare Research and Quality (US); http://www.ncbi.nlm.nih.gov/books/NBK154387/
. Accessed September 18, 2017.
6. Messaris E, Sehgal R, Deiling S, et alDehydration is the most common indication for readmission
after diverting ileostomy
creation. Dis Colon Rectum. 2012;55:175–180.
7. Fish DR, Mancuso CA, Garcia-Aguilar JE, et alReadmission after ileostomy
creation: retrospective review of a common and significant event. Ann Surg. 2017;265:379–387.
8. Phatak UR, Kao LS, You YN, et alImpact of ileostomy
-related complications on the multidisciplinary treatment of rectal cancer. Ann Surg Oncol. 2014;21:507–512.
9. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JGResearch electronic data capture (REDCap)–a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42:377–381.
10. Wilson VanVoorhis C, Morgan BUnderstanding power and rule of thumb for determining sample size statistical power effect size. Tutor Quant Methods Psychol. 2007;3:43–50.
11. Hayden DM, Pinzon MC, Francescatti AB, et alHospital readmission
for fluid and electrolyte abnormalities following ileostomy
construction: preventable or unpredictable? J Gastrointest Surg. 2013;17:298–303.
12. Nagle D, Pare T, Keenan E, Marcet K, Tizio S, Poylin VIleostomy pathway virtually eliminates readmissions for dehydration
in new ostomates. Dis Colon Rectum. 2012;55:1266–1272.
13. Hardiman KM, Reames CD, McLeod MC, Regenbogen SEPatient autonomy-centered self-care checklist reduces hospital readmissions after ileostomy
creation. Surgery. 2016;160:1302–1308.
14. Paquette IM, Solan P, Rafferty JF, Ferguson MA, Davis BRReadmission for dehydration
or renal failure after ileostomy
creation. Dis Colon Rectum. 2013;56:974–979.
15. Shaffer VO, Owi T, Kumarusamy MA, et alDecreasing hospital readmission
patients: results of novel pilot program. J Am Coll Surg. 2017;224:425–430.
16. Hendren S, Hammond K, Glasgow SC, et alClinical practice guidelines for ostomy surgery. Dis Colon Rectum. 2015;58:375–387.
17. Greenblatt DY, Weber SM, O’Connor ES, LoConte NK, Liou JI, Smith MAReadmission after colectomy for cancer predicts one-year mortality. Ann Surg. 2010;251:659–669.
18. Baker P, Dworkin SL, Tong S, Banks I, Shand T, Yamey GThe men’s health gap: men must be included in the global health equity agenda. Bull World Health Organ. 2014;92:618–620.
19. Iqbal A, Raza A, Huang E, Goldstein L, Hughes SJ, Tan SACost effectiveness of a novel attempt to reduce readmission
creation. JSLS. 2017;21: e2016.00082.