Readmissions and Emergency Room Visits Following Outpatient and Inpatient Herniorrhaphies.

Despite recent advances in abdominal wall reconstruction, postoperative readmissions remain a substantial source of morbidity and additional costs. For example, the reported nationwide readmission rate following ventral hernia repairs (HR) is 20% [ 1 ], which is higher than after complex oncologic procedures [ 2 ]. As a result, strategies to prevent readmissions after herniorrhaphies are exponentially drawing interest. Most published-to-date studies of readmissions following ventral and inguinal HR utilize nationwide data, such as the National Readmission Database [ 3 , 4 ], other national registries [ 5 - 7 ], American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP) [ 8 ], or institutional data [ 9 - 11 ]. While these studies are useful for understanding inpatient HR, which presumably have higher complexity, they largely omit outpatient repairs and emergency department visits (EDVs), with rare exceptions [ 12 ]. Additionally, no studies describe a time-wise breakdown of postoperative encounters. We aimed to fill the gaps in the current understanding of the type (readmission or EDV) and timing of unplanned postoperative encounters after HR. To achieve comprehensive coverage for both inpatient and ambulatory HR, as well as postoperative readmissions and EDVs, we linked several state-level administrative databases of the Healthcare Cost and Utilization Project (HCUP). We aimed to determine the incidence, timing, and primary reasons for EDVs and readmissions after HR. To our knowledge, integrating multiple HCUP databases for a comprehensive review of postoperative events represents a novel approach.

A total of 26,218 unique patients who underwent inpatient (n=3,297; 12.6%) or outpatient (n=22,921; 87.4%) abdominal wall HR (AHR) in Maryland during 2016 (n=12,749) and 2017 (n=13,469) were identified (Figure 1 ). Of those, 2,267 (8.6%) patients had a readmission or an EDV within 30 days after their hernia surgery. The timing of the first postoperative encounter by its type is listed in Table 1 . The majority (81.0%) of postoperative encounters within the first 48 hours after surgery were EDVs. The fraction of EDVs within later encounters was in the 70.5-76.4% range. Within 90 days postoperatively, 953 (3.6%) patients were readmitted, and 2,733 (10.4%) visited the ED. SASD: State Ambulatory Surgery and Services Databases; SID: State Inpatient Database; SEDD: State Emergency Department Database; POD: postoperative day * fraction of a total number of encounters within a given time frame postoperatively Individuals who had no 30-day ED or hospital encounters tended to be Caucasian males, with fewer comorbidities, privately insured, with higher income, and married. They had more groin HR than ventral HR, and no concomitant procedures. Conversely, those who had 30-day postoperative encounters were more likely to be African American patients, have Medicaid/Medicare, and have co-morbidities. They had more concomitant procedures, such as adhesiolysis, bowel resection, or cholecystectomy, and were more likely to receive blood products or ICU care. Expectedly, the incidence of the most common chronic conditions was higher in patients with higher readmissions or EDV. There was no difference in patients’ age or approach in bivariate analysis (Table 2 ). a by independent samples Mann-Whitney U test (for skewed distributions) b by Chi-squared test c column percentages (i.e., 69% of patients without 30-day encounters were males) d Pearson Chi-Square statistic e Mann-Whitney U standardized test statistic (Z-score) SD: standard deviation; IQR: interquartile range; MIS: minimally-invasive surgery (laparoscopy/robotics); LOA: lysis of adhesions; OSA: obstructive sleep apnea; TIA/CVA: transient ischemic attack/cerebrovascular accident; BMI: body mass index; EtOH: ethanol (alcohol); ICU: intensive care procedures (e.g., mechanical ventilation, placement of central lines, administration of vasopressors, tube feeds); OSA: obstructive sleep apnea; CAD: coronary artery disease; HF: heart failure; MIS: minimally invasive surgery Please note that due to negligible fraction of missing data, total values within the variable may slightly differ. Structure of 30-day postoperative encounters and associated charges Primary diagnoses of EDVs or hospitalizations of patients are listed in Figure 2 . Most frequent reasons for postoperative EDV were urinary complaints (n=101 (24.1%), n=46 (10.6%), and n=33 (4.0%) on postoperative days (PODs) 0-2, 3-7, and 8-30, respectively), followed by pain control issues (n=75 (18.1%), n=108 (24.9%), n=118 (14.4%), respectively), and delayed return of bowel function (n=44 (10.5%), n=43 (9.9%), n=28 (3.4%), respectively). Average charges for such EDVs were $958.6±1,316.0, $1,204.0±944.8, and $1,227.9±1,031.3, respectively. The fraction of these three most common conditions accounted for more than 50% of all EDVs within two days after surgery. This fraction progressively declined with later presentation times. Hospitalizations had a more diverse structure of admission diagnoses. Nonetheless, the top five admitting diagnostic groups (within 30-days postoperative) constituted 51.5% of all admissions: surgical site infection (n=118; 19.8%), delayed return of bowel function (n=73; 12.2%), respiratory complications (n=33; 5.5%), sepsis (n=32; 5.4%), surgical site hematoma or seroma (n=26; 4.4%), and cardiac complications (n=25; 4.2%). Charges for SSIs, GI complaints, and respiratory complications were $17,252±22,546, $15,404±23,556, and $12,362±13,38, respectively. A separate cost analysis was not performed due to the scope of this work. Diagnostic and therapeutic interventions provided to patients in the ED are summarized in Table 3 . “Bowel obstruction/ileus” group encompasses diagnoses of bowel obstruction (88/39.5%), constipation (84/37.7%), nausea and vomiting (35/15.7%), and ileus (12/5.4%). Urinary issues include urinary retention (106/46.5%) and UTI (60/26.3%). UTI: urinary tract infection; VTE: venous thromboembolism; SSI: surgical site infection a of all ED visits within 30 days postoperative Prediction of 30-day readmission or ED utilization The most impactful predictors of 30-day encounters were Medicaid insurance, followed by inpatient surgery, Medicare, African American race, cardiac comorbidities, obesity, psychiatric disorders, and the Charlson comorbidity score. Table 4 details logistic regression models predicting 30-day postoperative encounters. Both models returned equivalent coefficients. We choose the value of the shrinking parameter λ = 0.00826 (log λ ≈ -4.8). This value is located within 1 standard error of the optimal λ value returned by LASSO (see Figure 3 ). As a result, it leads to the simplest (kept 16 out of the initial 33 predictors) but still accurate model. Backward stepwise regression kept 19 out of 33 predictors. The rate of missing data in the models was 6.4%. Metrics of models, obtained by 10-fold cross-validation on a 20% sample, are listed in Table 4 . Both models were statistically significant (p<0.001) and correctly classified 91.3% cases. The models provided very high specificity and negative predictive value, but low sensitivity (Table  5 ). a Higher value denotes more significant impact CI: confidence interval; CVA: cerebrovascular incident; GI: gastrointestinal; ICU: intensive care unit; LASSO: least absolute shrinkage and selection operator regression; MIS: minimally invasive surgery The value of 0.00826 was selected to provide the simplest model (16 out of 33 predictors) with acceptable prediction error (within 1 standard error from optimal λ value). ROC: receiver operating characteristic; PPV: positive predictive value; NPV: negative predictive value

As a vastly prevalent pathology, hernias not only present a surgical challenge but also reveal healthcare disparities. Despite a relative lack of clinical granularity, merged state-level administrative databases cover up to 95% of all HR and all unplanned healthcare encounters. This, to our knowledge, is a novel approach, which provides a real-world picture and accounts for the timeline of these encounters. Understanding of the timeline, diagnostic structure, and predictors of these events will help to develop targeted interventions. For instance, EDVs constituted 75% of postoperative encounters, peaking within the first two days after surgery. Half of these early EDVs were related to GI issues, urinary disturbances, and pain. The latter two conditions were almost always successfully treated in the ED without requiring admission. Thirty-day readmission rate was 2.3%, which is lower than previously reported. This is likely due to a larger denominator, since we have a more inclusive and comprehensive approach, more representative of standard healthcare settings where primary herniorrhaphies are performed. Readmissions peaked by the end of the first week, primarily due to GI issues and surgical site (SSI) or systemic infections, which also corresponds to previously reported data. Previously reported 30-day ED utilization rates were around 5-6.5%, which aligns with our results (6.4%) [ 12 , 16 ]. Selected previous reports on 30-day readmissions and ED utilization after herniorrhaphy are summarized in Table 6 . a Fraction of all readmissions;  b Geisinger Medical Center healthcare system (central PA) ACS-NSQIP: American College of Surgeons National Surgical Quality Improvement Program; AHR: abdominal hernia repair (includes inguinal/femoral, ventral midline & lateral, incisional); AKI: acute kidney injury;  AMA: against medical advice;  ASA: American Society of Anesthesiology class; CCI: Charlson Comorbidity Index; CCY: cholecystectomy; COPD: chronic obstructive pulmonary disease; CS: component separation; DHD: Danish Hernia Database; DNPR: Danish National Patients Registry; ED: emergency department; GU: genitourinary (urinary tract infection, urinary retention, etc); HR: hernia repair; IHR: inguinal (and femoral) hernia repair; LOS: length of stay; N/A: not available (not reported) data; NRD: National Readmission Database; SOB: shortness of breath; SPARCS: Statewide Planning and Research Cooperative System (New York); SSI: surgical site infection; SSO: surgical site occurrence (seroma, hematoma, wound healing issues); UH: umbilical hernia; VHR: ventral hernia repair. SSIs, found to be a leading cause of readmissions, underscore the critical imperative of evidence-based infection control strategies and routine surveillance for early detection and intervention. SSI and other wound-related complications cause half of readmissions within one year after incisional [ 16 ] and inguinal HR. Similarly, our study revealed that SSIs were the leading cause (nearly 20%) of all 30-day readmissions. Vigilant monitoring for signs and symptoms of SSI, particularly during the first week post-surgery, allows for prompt initiation of outpatient therapy, including wound care and antibiotics when indicated, to avoid preventable readmissions. We found that delayed return of bowel function (including constipation, diet intolerance, bowel obstruction, and ileus) was the most common reason for readmission within the first week after herniorrhaphy. Certain ileus-reducing strategies, such as peripherally acting µ-receptor antagonists (alvimopan, methylnaltrexone), demonstrated efficacy in patients undergoing abdominal surgery and using opioid patient-controlled analgesia [ 23 ]. Multimodal analgesics, regional anesthesia, and judicious management of volume status can be useful as well. Inadequate pain control emerged as a primary reason for postoperative EDVs, emphasizing the potential of Enhanced Recovery After Surgery (ERAS) protocols and shared decision-making to optimize pain management. The growing use of ERAS protocols is a hopeful strategy to achieve more reliable pain control postoperatively [ 24 ]. Urinary retention was the second most common reason for EDV within the first week of herniorrhaphy. It was almost always successfully treated in the ED and did not require admission. The use of modern anesthetic reversal agents (sugammadex) and the avoidance of general anesthesia when possible [ 25 ] are two mitigating strategies to minimize ED visits for postoperative urinary retention. Predictors of readmissions and EDVs The most impactful independent predictors of 30-day encounters were Medicaid insurance, followed by inpatient surgery, Medicare, African-American origin, cardiac comorbidities, obesity, psychiatric disorders, and the Charlson comorbidity score. This highlights the significant impact of socio-economic factors on the postoperative course, comparable in magnitude to that of medical comorbidities. Medicaid patients frequently have higher rates of comorbidities and social determinants that contribute to poor health outcomes. This population is likely to face barriers to accessing elective surgical consultation and proper follow-up care. Disparities based on race, income, and insurance status were observed in readmissions and EDVs, highlighting the need for targeted interventions to address these inequalities. Notably, our study revealed disparities in race, income, and insurance status in both LASSO and backward regression models, adding to a growing body of literature on socioeconomic predictors in hernia outcomes [ 1 , 26 , 27 ]. The diverse racial and ethnic composition, socioeconomic status, and urban/rural location of the Maryland population allowed us to demonstrate the effect of disparities on readmissions and EDVs. Furthermore, Medicaid and African American heritage were the first and fourth most significant predictors of 30-day encounters, independent of each other. These two populations more frequently have acute presentations requiring emergent surgeries and lower rates of optimal outcomes, leading to almost doubled cumulative costs compared to holders of private insurance [ 26 ]. Implementing targeted programs to improve access to elective surgical consultations, enhancing follow-up care, and addressing social determinants of health are critical steps toward achieving equitable surgical outcomes for all patients. Utilization of EDs The high utilization of ED services postoperatively prompts consideration of alternative management strategies for complications, potentially reducing unnecessary EDVs. One in 16 patients in our population visited the ED within 30 days after their HR. The overwhelming majority of EDVs in our population included certain diagnostic or therapeutic interventions (Table  3 ), suggesting that these visits may not always be classified as "unnecessary". EDVs for urinary disorders or acute pain almost never resulted in readmission, indicating that these two causes can be effectively addressed in the ED setting. Brown et al. noted that despite a decrease in readmissions after common general-surgery procedures in recent years, the rate of EDVs has changed minimally [ 12 ]. They postulated that this increase could be a consequence of the Hospital Readmission Reduction Program, implemented in 2009. Some patients who historically would be readmitted now undergo workup and management under ED observation. On the other hand, all our patients who visited the ED (by design of SEDD) were discharged home, suggesting that these patients could potentially have been managed in outpatient settings such as a physician's office, urgent care, or by visiting nurses. Limitations The main limitation of this study, utilizing administrative databases, is the inability to account for numerous potential clinical predictors of unplanned 30-day encounters. Clinical and demographic heterogeneity also makes it difficult to generate a single universally accurate forecasting model. Accordingly, our model has high specificity but low sensitivity, and does not serve as a screening tool. Rather, based on the diverse Maryland population of all abdominal hernia patients, it captured common risk factors for readmissions and EDVs regardless of hernia type and patient characteristics. However, we realize that state-specific patient mix prevents nationwide generalizations. Our dataset was limited to 2016-2017, and it is possible that there are new unforeseen factors such as limited access to healthcare during the COVID-19 pandemic, improved ERAS protocols, or wider use of telemedicine. However, we believe that the core factors driving unplanned healthcare encounters following HR have remained consistent over time. The nature of state-level databases prevents tracking patients' encounters in neighboring states. State-level HCUP databases also prevent tracking individuals across years, so if the patient underwent an HR in December and got readmitted in January the following year, this encounter could not be captured.

In this large, state-level analysis linking inpatient, outpatient, and emergency department databases, nearly one in 12 patients experienced an unplanned healthcare encounter within 30 days of hernia repair. EDVs markedly outnumbered readmissions (6.4% vs. 2.3%), most occurring within the first 48 hours and driven primarily by urinary or gastrointestinal complaints and inadequate pain control. Readmissions were largely attributable to SSIs and GI complications. Socioeconomic and insurance-related disparities emerged as the most significant independent predictors of unplanned encounters, underscoring that healthcare access and equity are as critical to postoperative outcomes as medical and surgical factors. Although the retrospective design and reliance on administrative databases limit clinical granularity and procedure-specific detail, this approach offers a standardized, real-world perspective that equalizes for variations in hernia morphology and surgical technique and reveals risk factors common across a wide spectrum of herniorrhaphies. Our findings highlight the importance of targeted perioperative strategies-including patient education, multimodal analgesia, bowel regimen optimization, infection surveillance, and improved follow-up access-to reduce avoidable postoperative readmissions and EDVs. Future prospective studies integrating greater clinical detail, patient-reported outcomes, and socioeconomic variables are warranted to validate and refine these risk models, ultimately guiding interventions that improve recovery and equity in abdominal wall reconstruction.

Data source This was a retrospective cohort analysis evaluating HR (inpatient and outpatient), postoperative encounters (EDV vs readmission), and time to these encounters using the State Inpatient Database (SID), State Emergency Department Database (SEDD), and State Ambulatory Surgery and Services Databases (SASD) of Maryland, United States, 2016-2017, provided by the Agency for Healthcare Research and Quality, United States [ 13 ]. Institutional review board approval was waived as not required for the analysis of de-identified administrative data according to our institution’s policy. At the time of statistical analysis, our institution did not have access to more updated versions of these databases. State-level databases cover up to 97% of all discharges within a state [ 14 ]. These databases contain revisit variables, including a de-identified unique patient number, which allows linking of encounters of different types (hospitalization, EDV, and outpatient surgery, excluding outpatient offices and urgent care centers), and a timing variable, enclosing time gaps between encounters. As such, one can reconstruct a timeline of most healthcare encounters of a single individual within one state within a given calendar year. Study population We selected all encounters with procedural codes for abdominal wall hernia repairs (International Classification of Diseases, 10th Revision, Procedure Coding System (ICD-10-PCS) for inpatient (SID) procedures and Current Procedural Terminology (CPT®) for outpatient (SASD) procedures (see Appendices). We included only patients with primary diagnoses of abdominal wall hernias in order to isolate initial hernia patients from those who underwent herniorrhaphy in combination with a different intervention, such as primary (non-mesh) repair of umbilical hernia during laparoscopic cholecystectomy. Encounters of patients who underwent HR were supplemented by their hospitalizations (SID) and EDVs (SEDD) by linking all encounters among all state-level databases as described above. We identified and removed all duplicate records. The final study database encompassed encounters with herniorrhaphies (referred to as “index encounters”) and all subsequent patients’ encounters (hospitalizations and EDVs) that occurred during the same calendar year. Statistical considerations All analyses were performed in IBM SPSS Statistics for Windows, version 22 (Released 2013; IBM Corp., Armonk, New York, United States) and R version 4.2.2 (2022; R Foundation for Statistical Computing, Vienna, Austria, https://www.R-project.org /). We computed time intervals between the index (operative) and first subsequent encounters using the SPSS LAG function. Timing of the first postoperative encounter was classified as follows: ≤2 days, 3-7 days, 8-30 days, 31-90 days, and >90 days. We determined the structure of primary diagnoses of the first postoperative encounters by timing and encounter type (EDV versus readmission). Additionally, we further analyzed the workup and therapy provided by EDs to the patients of the three most common diagnostic groups. Patients’ demographics and comorbidities, hernia type, surgical approach, and facility characteristics were compared between patients who had any 30-day postoperative encounter and those who did not. We utilized conventional statistical tools such as the Chi-squared and Mann-Whitney U tests. We built two logistic regression models predicting 30-day postoperative readmission or EDVs. All potential predictors were critically assessed according to fundamental knowledge and clinical sense. Variables for the first model were selected by LASSO (least absolute shrinkage and selection operator) regression. This approach facilitates the selection of the most significant predictors and thus achieves a good balance of accuracy and simplicity (library “glmnet” within R) [ 15 ]. Variables for the second model were determined by backstep elimination based on likelihood ratios. For validation, we randomly split the input data into 80% training and 20% validation subsets and applied to the models 10-fold cross-validation (library “caret”).