How Do We Prevent Readmissions After Major Surgery?

Introduction

Readmissions are publicly reported and associated with financial penalties for hospitals. A landmark study using Medicare claims data by Jencks and colleagues [1] in 2009 highlights the sobering prevalence of readmissions nationwide with one-fifth of hospitalized Medicare patients being readmitted within 30 days of discharge. Among patients receiving surgical care, 15.6% were readmitted with 2% experiencing death within 30 days compared with 21.1% readmission and 4.1% mortality among medical patients. Further, 67.1% and 51.5% of medical and surgical patients, respectively, were rehospitalized or died within the first year after the index discharge with an estimated excess cost of $17 billion annually. Tsai and colleagues [2] in 2013 found that nearly 1 in 7 surgical patients are readmitted within 30 days of a major operation using similar Medicare data.

The Affordable Care Act [3] implemented the Hospital Readmission Reduction Program in 2012 to incentivize hospitals with higher than expected 30-day readmission rates to adopt processes of care to reduce costly unplanned readmissions through reimbursement penalties. The Centers for Medicare and Medicaid Services (CMS) defines planned readmissions as rehospitalization for bone marrow or organ transplantation, chemotherapy/radiation therapy, rehabilitation, obstetric delivery, or a CMS-identified potentially planned procedure with a principle diagnosis code not related to an acute condition or complication of care [4]. Unplanned readmissions related to the care of acute myocardial infarction, heart failure, and pneumonia were targeted in fiscal year 2012 with expansion to chronic obstructive pulmonary disease and total hip or knee arthroplasty in 2015. Similar implications for the care of patients undergoing coronary artery bypass grafting are already underway for fiscal year 2017 with other surgical procedures likely to follow.

Given the prevalence of surgical readmissions and these financial implications, investigators strive to describe and identify characteristics and predictors of readmission in efforts to develop feasible and cost-effective strategies that reduce readmissions [5]. Such efforts assume that surgical readmissions are predictable and thereby preventable with risk factor identification and intervention. A brief review of known predictors and reasons for surgical readmission are included along with the effectiveness of published readmission reduction interventions to better determine if readmissions are in fact preventable.

Predictors of readmission

Surgical readmissions are tracked through a variety of ways including the University Healthcare Consortium and Medicare along with current quality improvement initiatives in the Veterans Administration Surgical Quality Improvement Project (VASQIP) and the National Surgical Quality Improvement Project (NSQIP). Definitions of 30-day readmission range from the time of the index surgery (VASQIP, NSQIP) or discharge (CMS, University Healthcare Consortium, Medicare), resulting in different rates depending on the reported time frame. With the tremendous quality and financial implications posed by surgical readmissions, a campaign to describe and identify risk factors of readmissions emerged [6], [7], [8]. According to Lucas and colleagues [9], events influencing patient outcomes and readmission after surgery are vast and can generally be summarized into biologic, health care, and social factors. We modified their categories to include distinctions between patient and surgical factors (Table 1).

Table 1.

A modified conceptual model of factors affecting readmission

Biologic factorsa	Social factorsa	Surgery factors	Health care factors
Disease process Demographics Comorbidities	Marital and caregiver status Socioeconomic status Substance abuse Social support Health beliefs Health care self-efficacy Health care social norms	Preoperative planning Appropriate indication Acuity of operation Magnitude of operation Prevalence of known complications Surgeon expertise Anesthesia Blood loss and use Infection control	Quality Critical care Postoperative care Nursing Allied health Social work Discharge planning Home health services Outpatient follow-up Care coordination Complications

^aPatient-level factors contribute the majority toward predicting readmissions.

Modified from Lucas DJ, Haider A, Haut E, et al. Assessing readmission after general, vascular, and thoracic surgery using ACS-NSQIP. Ann Surg 2013;258(3):430–9.

Readmission rates are highly variable among different surgical procedures. Patient-level factors composed of demographics, comorbidities, and disease process/indication for surgery are highly associated with readmission across diverse surgical specialties and procedures. Although readmission rates vary by procedure mix, the indication for surgery may further explain readmission variance within similar specialties and procedures. Readmission rates varied by indication for lower extremity arterial bypass with higher readmission risk among critical limb ischemia with rest pain or tissue loss compared with claudication or other indications [10], [11]. Further studies are needed to understand indication among other surgical procedures; however, providers should consider procedure indication in risk factor analyses and may describe a more high-risk subset of patients for targeted readmission interventions.

As expected, patients with more comorbidities are at increased risk for readmission [12]. Investigators have developed complex readmission risk scores using single institution data in the Rothman Index incorporating granular patient factors including vital signs, nursing assessments, Braden scores, cardiac rhythms, and laboratory data in addition to comorbidities. Patients with elevated indices were associated with readmission [13], [14]. Patients with mental health diagnoses are more likely to become readmitted than those without (schizophrenia, 16% vs 11%; depression, 12.5% vs 10.8%) [15]. Frail patients and those with dependent functional status are also more likely to experience readmission after surgery [15], [16], [17]. Prior use of health care with an inpatient admission or emergency department visit within 6 months of an index operation is associated with surgical readmission [15]. Whether or not complications occur, high reported pain scores greater than 8 were highly associated with readmission among Veterans Affairs (VA) patients undergoing major surgery and warrants further investigation [15]. Preliminary work is already under way to understand how patient pain trajectories and medications change over the postoperative course and how they relate to outcomes, including readmission.

Patient demographics and social environment impacting readmission highlight disparities in surgery. Race is a known independent predictor associated with poor outcomes, including readmission. Black patients experience higher mortality, complication, and readmission rates compared with patients of different races [18], [19]. Patients at minority-serving hospitals are more likely to be readmitted after major surgery than patients at non–minority-serving hospitals [20]. Studies have shown Asian, Black, and Hispanic patients experience higher readmission rates compared with whites after surgery despite adjusting for patient, procedure, and hospital factors [15], [18], [20], [21], [22]. Low socioeconomic status, disadvantaged social determinants of health, limited education, poor social support, and poor self-efficacy or lack of self-management skills contribute to readmission [12], [23], [24]. Disparities in care are likely attributed to patient, provider, and systemic factors and remain potential targets for quality improvement [25]. Disparities research efforts are now a national funding priority of the American College of Surgeons and the National Institutes of Health.

It is well-known that postoperative complications place patients at increased risk for prolonged duration of stay, morbidity, readmission, and mortality. Similar to various studies using single institution datasets, Kassin and colleagues [26] studied NSQIP outcomes among 1442 general surgery patients from 2009 to 2012. They showed that having any postoperative complication, independent of patient- or procedure-specific factors, increased risk of readmission by a factor of 4. Gani and colleagues [19] used administrative claims data from 2009 to 2013 of 22,559 patients undergoing major surgery and found patient comorbidity, race, having a postoperative complication, and extended duration of stay predicted readmission. A systematic review of 35 studies involving vascular, general, bariatric, and colorectal operations found reliable predictors of readmission involving postoperative complications, medication-related issues, comorbidity, and postoperative duration of stay [27]. Understanding the timing of and reasons for postoperative complications will provide further insight and targets for action. Among VA patients undergoing 59,273 colorectal, arthroplasty, vascular, and gynecologic procedures from 2005 to 2009, more than one-quarter of assessed VASQIP complications occurred after discharge and ultimately resulted in readmission [28].

Multicomorbid patients with or without predischarge complications often require additional resources during their recovery necessitating discharge to a location other than their home or original admission source; however, these patients are at risk for readmission [17]. Greenblatt and others in 2010 showed that vascular patients were 2.5 times more likely to experience a readmission when discharged to a skilled nursing facility [29]. In addition, Morris and colleagues [15] also found that patients admitted from skilled nursing facilities or transferred from another hospital preoperatively were more likely to be readmitted in addition to those who were discharged to locations other than home. Rather than poor clinical care delivered at ancillary care facilities, disposition to locations other than the original admission source may reflect a sicker population with a reduced capacity to regain homeostasis after surgery.

Care provided and readmissions: Hospital, provider, or patient factors?

Hong and colleagues [30] investigated surgery readmissions after major gastrointestinal cancer operations among 110,857 patients in California from 355 hospitals in 2004 to 2011, 44 of which were vulnerable hospitals (safety net or high Medicaid payer systems). They concluded that vulnerable hospital systems and teaching status were significantly associated with 30-day readmission; however, the number of beds and procedure volume were not significantly associated with increased readmission. Vulnerable hospitals by definition care for minority or more comorbid populations that likely result in poor outcomes [20], [30]. Goodney and colleagues [31] show that hospital volume was not significantly associated with readmission among 2.5 million cardiovascular and cancer surgeries within Medicare.

Among bariatric surgeries, hospital teaching status and provider procedural volume are associated with readmissions. A low-volume bariatric surgeon (<25 cases/year) had a 15 times higher risk of readmission than medium-volume bariatric surgeons (26–150 cases/year) [32]. Further, low hospital volume, specifically facilities that perform fewer than 300 bariatric cases per year, was significantly associated with increased 30-day [33] and 1-year [34] readmission. High-volume facilities having improved outcomes with lower readmission rates may reflect structural processes of care and available resources. This led to a CMS policy for bariatric surgery to be performed at only centers of excellence. Dimick and colleagues [35] showed that there were no significant differences in rates of complications before and after CMS policy implementation, meaning non–centers of excellence facilities performed bariatric surgery with similar outcomes. Although readmissions were not included in analysis, hospital and surgeon volume likely did not explain outcomes and point to patient case mix.

To understand whether surgical readmissions represent global hospital effects or a sum of team and patient-level effects, Hollis and colleagues [36] investigated 100,086 VA patients undergoing general, orthopedic, and vascular surgery at 84 facilities from 2007 to 2014. Hospitals varied by the number of specialties above or below a national median readmission rate, yet this variation was not attributed to hospital volume, region, operative complexity, or procedure mix. Upon adjustment for patient factors, little variation in readmission could be explained by facility or specialty-specific factors. Patient and procedure case mix factors contributed most as opposed to broader team or facility-level effects. In 2013, Hyder and colleagues [37] showed that patient factors accounted for 95% of the variation in readmission after hepatobiliary surgery at 300 hospitals with hospital and provider factors showing little impact on readmission. Gonzalez and others studied 479,047 Medicare vascular surgery patients undergoing lower extremity revascularization and found that readmissions were attributed to hospital patient mix rather than hospital quality [11]. Merkow and colleagues [17] investigated a more heterogeneous patient population and case mix from a national sample of 498,875 operations performed in 2012 at 346 NSQIP hospitals. Patient-level factors alone explained most of the risk for readmissions compared with models using only inpatient complications, discharge destination, or hospital characteristics. The predictive model was not substantially improved when such additional information was integrated with patient-level factors. Gani and colleagues [19] found that 82.8% of the variability in 30-day readmissions is attributed to patient-level factors, 14.5% attributed to surgical specialty, and 2.8% to individual surgeons among 22,559 patients from 2009 to 2013 among various specialties. Further, Morris and colleagues looked at 237,441 VASQIP procedures among orthopedic, noncardiac general, and vascular surgery from 2007 to 2014. Again, preoperative patient-level factors contributed the most to the predictive model for readmission with modest predictive improvement using operative and postoperative information before discharge [15]. Overall, readmissions are challenging to predict despite using granular perioperative patient-centered data known to providers at the time of discharge, however, most of the variation in readmission prediction models is explained from patient-level factors.

Readmission reasons

A foundation for such work began with single institutions from a vast array of procedures of varying sample sizes describing site-specific observed readmission reasons [6], [7], [8]. Common reasons for readmissions range from wound complications including infection and seroma, gastrointestinal disturbances (nausea, emesis, dehydration, obstruction or ileus), nutrition, postoperative pain, exacerbations of known medical conditions, or procedure-related factors (ie, fistulae or duct leak after pancreatic resection). Additionally, substance abuse, mental health issues, and issues related to social support including homelessness are reasons for readmission after general surgery [38].

It was not until large, national cohorts were analyzed that we understood more generalizable reasons for readmission that support single institutional findings and may prove as worthy targets for intervention. Merkow and colleagues [17] highlight reasons for readmission nationally among bariatric, colectomy, ventral hernia repair, hysterectomy, total hip or knee arthroplasty, and lower extremity vascular bypass. Surgical site infection (19.5%) and gastrointestinal derangements with ileus at obstruction (10.3%) were among the most common reasons for 30-day readmission followed by complications with less than 5% event rates (bleeding, pulmonary, venous thromboembolism, dehydration/poor nutrition, sepsis, stroke, or pain). Readmission rates were uniform in distribution over the postoperative period without a particular peak postdischarge day or readmission reason. Morris and colleagues [15] support surgical site infection and wound complications as the most frequent reasons for surgical readmission among general, orthopedic, and vascular procedures. The majority of wound complications occurred within the first week after discharge and increased during the second week among vascular patients. Interesting, orthopedic wound complications remained low and steady throughout the 30 days after discharge. Identifying perioperative predischarge risk factors and generating prediction applications for readmission are important; however, understanding reasons for readmission provide tremendous value in our campaign to formulate readmission reduction strategies.

Are unplanned readmissions preventable?

According to a 2007 Medicare Payment Advisory Commission report [39] to Congress on readmissions, readmissions could be avoided if the standard of care was implemented during the index hospitalization. Nearly 3 of every 4 all-cause readmissions, mostly among medical patients, were cited as potentially preventable using computer algorithms classifying readmissions as potentially preventable if reasons related to complications or processes of care present before discharge from an index admission. Interestingly, surgical readmissions, whether owing to medical or surgical reasons, were deemed more challenging to characterize.

Preventable readmissions have been defined as “if a reasonable improvement in the health care process could have potentially avoided the need for rehospitalization,” [40] or “being related to a complication of previous medical care or a gap in care transitions” [38]. Such definitions imply that if “we” prevent complications, readmissions could be reduced. Although surgical patients experience complications at a fairly consistent rate, effectively predicting which patients will experience a complication that leads to a readmission remains challenging [41]. Strategies to reduce complications postoperatively are challenging; the Surgical Care Improvement Project targeted surgical site infections, a highly prevalent postoperative complication, with limited success. In general, the Surgical Care Improvement Project has not resulted in measurable decreases in postoperative complications [42], [43]. However, other surgical site infection reduction strategies including surgical site infection reduction bundles, have shown promise to reduce wound infection after discharge [44], [45]. Further, the concept of preventing complications assumes readmissions relate to complications occurring during the index hospitalization and reflect poor quality of care. However, Merkow and colleagues [17] showed that nearly all complications across various specialties are related to new postdischarge events rather than failed management or continuation of predischarge events. Only 2.3% of postdischarge complications resulting in a readmission were related to or exacerbations of known complications at the time of discharge. Dawes and colleagues [40] defined 21% of readmissions as preventable owing to acute related (65.5%), acute unrelated (14.5%), and chronic disease management (14.5%) reasons. Of preventable readmissions, 49.1% could have potentially been prevented through better follow-up and transitions of care upon discharge with 41.8% potentially treated in an outpatient setting.

INTERVENTIONS TO REDUCE READMISSION

Transitions of care

Strategies should focus on potentially modifiable patient-level factors preoperatively, enriching postdischarge transitions of care, and common reasons for readmission by procedure performed [15]. Transition of care protocols have demonstrated effectiveness in reducing medical readmissions and have been adapted to surgery [8], [15], [17], [27]. A systematic review by Jones and colleagues [46] highlights that the evidence for transition of care models after surgery are in its infancy, but identify key strategies shown to reduce readmissions. Among 10 studies reviewed ranging from randomized controlled trials (n = 3) to observation studies (n = 5 retrospective and n = 2 prospective), the most commonly evaluated components involve follow-up phone calls, patient education, coordinated discharge planning, follow-up with surgeon or primary care provider, and face-to-face home visits with advanced practitioners. Strategies most associated with significant readmission reductions involve coordinated discharge planning, patient education, and patient follow-up.

The most successful coordinated discharge planning interventions used checklists and a dedicated clinical individual, usually a nurse or physician assistant, who followed patients throughout the perioperative course with postdischarge care coordination. The quality of education and discharge instructions starts preoperatively to establish patient expectations and preferences. [17] Individualized education programs implemented early and continued after discharge are effective. Education material was tailored and enriched to individual patient needs through survey data assessing knowledge or concerns regarding various domains of care, including pain and wound management, gastrointestinal disturbances, or emotional problems. Other programs focus on patient self-management to improve comprehension and discover gaps in education that could be addressed, rather than discovering such gaps “after it is too late” and the patient is discharged, resulting in readmission.

Outpatient follow-up involved multiple domains with clinics, home visits, and phone calls. Home visits provided opportunities for advanced practitioners and nurses to coordinate care such as wound care or medication adjustments. Primary care provider follow-up, in addition to surgeon assessments, showed the greatest benefit among high-risk patients undergoing major surgery. Close primary care provider involvement also increased patient satisfaction [47]. Although patient phone calls were the most common element among the included studies, there is little evidence to suggest that patient phone calls are effective as an independent intervention. Studies should focus on the cost effectiveness and feasibility of transitions of care interventions.

Interestingly, racial disparities in surgical readmissions are less apparent within the Veterans Health Administration and likely reflect uniform health care and resource allocation compared with minority-serving hospitals. The Veterans Affairs’ embedded primary care systems with universal outpatient care, resource use, national electronic patient record system, and coordinated discharge processes through transitions of care may improve readmission rates [42]. Aligning with patient preferences and meeting unmet needs transitioning out of the hospital may reduce racial disparities in surgery.

Enhanced recovery pathways

Enhanced recovery, or fast-track surgery, comprises standardized approaches to deliver multimodal perioperative surgical care designed to reduce physiologic stress induced by surgery to promote early recovery. Although such pathways are associated with improved clinical outcomes by incorporating comprehensive care and strategies involved with many transitions of care pathways, few studies have demonstrated reduced readmission among enhanced recovery programs. Such programs were not designed to reduce readmission, but to deliver standardized care to improve patient satisfaction and outcomes. Recovery pathways have comparable, not reduced, readmission and complication rates with traditional care models despite earlier discharge. As standardized care pathways are improved over time, it is reasonable to hypothesize that readmissions will likely decrease in the future as quality care improves.

Prehabilitation

Given the vast contribution of preoperative patient-level factors toward the risk of readmission, preoperative risk assessment and intervention may reduce readmissions through patient optimization. Prehabilitation is a “process enhancing one’s functional and mental capacity to buffer against the potential deleterious effects of a significant stressor” [48]. Prehabilitation aims to optimize patient comorbidities in physical, nutritional, and psychosocial domains through multidisciplinary support and education [49]. Although surgical outcome results are mixed owing to heterogeneous patient populations and study design, improved functional capacity in the preoperative period in conjunction with a multidisciplinary approach is favorable. The American College of Surgeon’s recently adopted the “Strong for Surgery” prehabilitation strategy in 2015 for future wide spread implementation to address nutritional optimization (ie, caloric, protein, immunonutrition), smoking cessation, glycemic control, and polypharmacy management preoperatively [50].

Prehabilitation literature and its implementation to improve surgical outcomes is an emerging field. Evidence that prehabilitation reduces readmission has yet to be seen; there are few studies reporting readmission as an outcome of interest [51]. However, prehabilitation reduces overall complication rates, improves functional capacity, and improves time to recovery after major abdominal surgery and may be a promising future strategy in readmission reduction. Further investigation is warranted, because a paucity of prehabilitation studies have focused on quality of care measures such as surgical readmissions.

Readmissions as a quality metric

The Hospital Readmission Reduction Program strategy assumes that rehospitalizations reflect suboptimal care that, when improved to the standard of care, would mitigate readmission rates, again highlighting the challenge of classifying unplanned readmissions as predictable or avoidable. Readmissions may not always reflect poorly delivered predischarge care, because readmissions for 1 procedure may not reflect quality delivered for another procedure. Readmissions may reflect quality care to facilitate proactive intervention for brewing complications rather than monitoring potentially deadly complications in the outpatient setting [17], [42]. Disparities in care may continue to grow under the current readmission policy as potentially unidentified or unmodifiable patient factors lead to readmissions and penalize hospitals with sicker, more disadvantaged patients [20], [42].

Summary

Readmissions are challenging to predict and, therefore, difficult to prevent, calling into question as to whether this metric is an adequate indicator of surgical quality. Until a new, nationally reported marker for surgical quality is proposed, the current readmission definition requires revision. As reasons for readmission continue active characterization, quality tracking improvement programs, including NSQIP and VASQIP, do not account for reasons for readmission [28]. Almost one-half of readmissions are not captured with a complication currently assessed by national quality improvement programs such as VASQIP. Further, definitions of readmission should capture 30-day all-cause readmission and be standardized to ensure universal interpretation and reporting [7]. As discussed, it is time to look outside the hospital to address surgical readmissions with a shift of focus to patient-level factors and transitions of care interventions.

Key points.

• Surgical readmissions remain challenging to predict despite granular data available to providers at the time of discharge.

• Postoperative complications resulting in readmission are new events rather than sequela of predischarge events.

• Multifactorial patient-level factors contribute most to readmission; it is time to look outside the hospital.

• Transition of care models focusing on patient education, coordinated discharge planning, and monitored follow-up are effective strategies to reduce readmission.