Impact of frailty on hiatal hernia repair: a nationwide analysis of in-hospital clinical and healthcare utilization outcomes Free

Summary

Previous studies recommend a watch-and-wait approach to paraesophageal hernia (PEH) repair due to an increased risk for mortality. While contemporary studies suggest that elective surgery is safe and effective, many patients presenting with PEH are elderly. Therefore, we assessed the impact of frailty on in-hospital outcomes and healthcare utilization among patients receiving PEH repair. This retrospective population-based cohort study assessed patients from the National Inpatient Sample database who received PEH repair between October 2015 to December 2019. Demographic and perioperative data were gathered, and frailty was measured using the 11-item modified frailty index. The outcomes measured were in-hospital mortality, complications, discharge disposition, and healthcare utilization. Overall, 10,716 patients receiving PEH repair were identified, including 1442 frail patients. Frail patients were less often female and were more often in the lowest income quartile compared to robust patients. Frail patients were at greater odds for in-hospital mortality [odds ratio (OR) 2.83 (95% CI 1.65–4.83); P < 0.001], postoperative ICU admissions [OR 2.07 (95% CI 1.55–2.78); P < 0.001], any complications [OR 2.18 (95% CI 1.55–2.78); P < 0.001], hospital length of stay [mean difference (MD) 1.75 days (95% CI 1.30–2.210; P < 0.001], and total admission costs [MD $5631.65 (95% CI $3300.06–$7.963.24); P < 0.001] relative to their robust patients. While PEH repair in elderly patients is safe and effective, frail patients have an increased rate of in-hospital mortality, postoperative ICU admissions, complications, and total admission costs. Clinicians should consider patient frailty when identifying the most appropriate surgical candidates for PEH repair.

INTRODUCTION

Prior to the 2000s, surgical intervention for paraesophageal hiatal hernias (PEHs) was recommended regardless of symptom status to prevent acute strangulation and its associated morbidity and mortality.^1–3 However, in 2002, Stylopoulos et al. found that watchful waiting was superior to elective PEH repair for asymptomatic or minimally symptomatic patients 65 years or older, citing a 1.1% rate of emergency surgical intervention and an associated 5.4% mortality rate.⁴ This resulted in a shift toward watchful waiting and delayed elective repair of symptomatic PEHs.⁵

Symptomatic PEHs often present between 65 and 75 years of age. Despite this predilection to elderly patients, there is increasing interest in offering elective surgical intervention to older patients with symptoms in the context of improved outcomes with modern minimally invasive techniques.^6,7 While most studies have demonstrated that laparoscopic PEH repair can be performed safely and effectively in an elderly population with a significant improvement in quality of life, conflicting findings remain.^8–10 A cohort study of 2681 patients undergoing laparoscopic PEH repair found a higher rate of minor complications among older patients, but no difference in mortality and major morbidity.¹¹ In contrast, one study assessed 267 patients undergoing PEH repair and found that older patients had a higher rate of major morbidity, but no difference in mortality.¹² Thus, careful patient selection for elective PEH repair on the basis of more than age alone is likely necessary. Standardized frailty assessments may be a solution in assessing surgical candidacy for PEH repair among older individuals.

Frailty is an important prognostic factor, as it approximates physiologic reserve and has been shown to predict postoperative outcomes and the ongoing need for healthcare support (i.e. nursing home and home care) at the time of discharge.¹³ However, many previous studies do not predict PEH repair outcomes based on frailty.¹⁴ Additionally, the impact of frailty on the cost-effectiveness of PEH repair is unknown. The purpose of this study was to investigate the impact of frailty on in-hospital outcomes and healthcare utilization among patients receiving PEH repair.

METHODS

Data source

A retrospective population-based cohort study was performed utilizing the 1 October 2015 to 31 December, 2019 data from the Healthcare Cost and Utilization Project (HCUP) National Inpatient Sample (NIS), managed by the Agency for Healthcare Research and Quality (AHRQ).¹⁵ The timeline was chosen to capture the years that NIS started utilizing the International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM) codes. The NIS is the largest public all-payer inpatient database in the US; it approximates a 20% stratified sample of community hospital discharges, and its included hospitals cover more than 97% of the population, providing a nationally representative sample of the patient population and hospital characteristics. It records information on roughly 7 million hospitalizations annually, including weighted data to help make population estimates.¹⁵ Hamilton Integrated Research Ethics Board (HiREB) ethics approval was not required for this study due to the nature of the database being de-identified and publicly available.

Population

The NIS captures 30 admission diagnoses and 15 admission procedures through the ICD-10-CM codes. Adult patients (age ≥ 18 years) who had an elective or emergency admission with a primary diagnosis code of hiatal hernia (K441, K440, and K449) were included. Included patients who underwent emergency surgery did so due to incarceration (K440) or strangulation (K441). The study group was narrowed down further by identifying the patients with a primary procedure code of hiatal hernia repair (0BQS0ZZ, 0BQT0ZZ, 0BQS3ZZ, etc.). These diagnosis and procedure codes have been used in similar studies and the comprehensive list of codes are listed in Supplemental Table 1. Patients were excluded if (1) their primary (first) procedure in hospital was not a hiatal hernia repair and (2) missing information on age, sex, type of hospital admission, in-hospital mortality, and length of stay to ensure a homogeneous cohort of patients.

Modified frailty index

The modified frailty index (mFI), as previously described by Velanovich et al. consists of 11 variables: (1) functional health (totally or partially dependent); (2) diabetes mellitus (insulin- or non-insulin dependent); (3) hypertension requiring medication; (4) peripheral vascular disease, history of revascularization, or amputation for peripheral vascular disease, rest pain, or gangrene; (5) congestive heart failure within 30-day period prior to surgery; (6) history of myocardial infarction within 6 months prior to surgery; (7) chronic obstructive pulmonary disease or pneumonia; (8) coronary artery disease, history of angina, previous coronary surgery, or prior percutaneous coronary intervention; (9) history of delirium/impaired sensorium; (10) history of transient ischemic attack; and (11) history of cerebrovascular accident.¹⁶ Each item was allocated the same weight (1 point) when calculating the index. The items were adapted to the NIS database through corresponding ICD-10-CM codes (Supplemental Table 1). The mFI was calculated as the number of items present in an individual patient divided by the number of total items. Patients were stratified into two groups based on an mFI cut-off of 0.27, which has been suggested by previous studies to differentiate between individuals of ‘frail’ and ‘robust’ status.¹⁷

Study variables

The patient characteristics included for analysis were age, sex, race category (White, Black, Hispanic, and others), class of the body mass index (BMI ≤30 kg/m², 30–40 kg/m², ≥40 kg/m²), insurance status (Medicare, Medicaid, private insurance, self-pay, and others), and income quartile. Comorbidities were assessed through the Elixhauser Comorbidity Software for ICD-10-CM to distinguish them from complications.^18,19 Hospital characteristics collected were teaching status, rural status, region (Northeast, Midwest, South, and West), and bed size (small, medium, and large). Disease characteristics such as elective or emergency admission status and diagnosis of gastric volvulus (K562) or strangulated hiatal hernia (K441) were collected. Additionally, operative characteristics such as route of surgery (open versus minimally invasive) and concurrent procedures such as fundoplication (0DV30ZZ, 0DV33ZZ, 0DV34ZZ, etc.), hiatal hernia repair with mesh (0BUT07Z, 0BUT0JZ, 0BUT0KZ, etc.), gastropexy (0DS60ZZ, 0DS64ZZ, 0DS67ZZ, etc.), and partial or total gastrectomy (0DB60Z3, 0DB60ZZ, 0DB70ZZ, etc.) defined by ICD-10-CM was collected.

Outcomes

The outcomes measured were in-hospital mortality, complications, discharge disposition, and healthcare utilization (cost, length of stay, and prolonged length of stay). Prolonged length of stay was defined as a dichotomous outcome for patients who had a total length of stay greater than 75% percentile of the entire cohort. The in-hospital perioperative complications were defined using secondary ICD-10-CM diagnosis and procedure codes included in the AHRQ Patient Safety Indicators or that explicitly identify postoperative outcomes. Furthermore, system-specific postoperative complications were collected by grouping postoperative complications into respiratory, cardiovascular, gastrointestinal, genitourinary, wound, infections, and shock using previously utilized methods. Specific outcomes relevant to gastrointestinal surgery such as ileus (K560 and K567), nausea and vomiting (R110 and R112), and urinary retention (R330, R338, and R339) were identified through previously established ICD-10-CM codes.^17,18 Discharge disposition was categorized into home, short-term hospital, skilled-nursing facility, home healthcare, and others. Due to the nature of the NIS database not having patient identifiers or linkage with other administrative databases, only in-hospital outcomes could be captured and out-of-hospital outcomes such as readmissions could not be captured.

Statistical analysis

All statistical analyses were performed using STATA (StataCorp version 17, College Station, TX). Discharge-level weight from HCUP was used to calculate survey-weighted national estimates. Patient characteristics are presented as frequencies (%) for categorical variables and means (standard deviations) for continuous variables. Statistical analyses for categorical and continuous variables were performed using the chi-square test (for means) or Mann–Whitney test (for medians), and two sample t-test, respectively. All statistical tests were two-sided with the threshold for significance set at P < 0.05. A backward stepwise regression model including patient characteristics (age, sex, race, BMI class, insurance status, income quartile, and Elixhauser comorbidities score), hospital characteristics (rurality and teaching status), admission characteristics, and procedure characteristics (open versus minimally invasive, concurrent fundoplication, gastropexy, or gastrectomy) was used to assess outcomes. The clinical importance of the covariate, expert opinion, and the Wald test with a P-value less than 0.25 as a threshold was required for entry into the multivariate analysis to account for all possible confounding. The covariate of hospital region, bed size, and mesh use for hiatal hernia repair was eliminated from the backward stepwise regression model and was deemed to be clinically less meaningful for outcomes of interest. Subgroup analyses of outcomes were conducted by surgical approach and type of admission.

RESULTS

The NIS population included 10,716 patients receiving PEH repair, identifying 1442 patients as frail with a mean age of 71.3 years (Table 1). With a sample representative of 20% of hospitals in the US, discharge-level weighted analysis produced a national estimate of 53,580 individuals receiving PEH repair, with 7210 frail and 46,370 robust patients. Baseline characteristics of our cohort are demonstrated in Table 1. Robust patients were more likely to be female relative to frail patients (73.9 vs. 69.9%). Frail patients were more likely to be older, black, in the lowest income quartile, and had a higher Elixhauser comorbidity score (Table 1). The rate of emergency admission was higher among frail patients compared to robust patients (33.9 vs. 22.1%). Equally, frail patients more often received open surgery versus robust patients (22.3 vs. 16.8%). Frail patients more often received concurrent gastropexy relative to robust patients (11.1 vs. 7.9%).

In-hospital mortality and morbidity

Table 2 demonstrates the impact of frailty on clinical outcomes. Frail patients experienced higher incidence of in-hospital mortality compared to robust patients (2.9% vs. 0.5%). Higher rates of postoperative intensive care unit (ICU) admission were also seen among frail individuals versus those that were robust (8.2 vs. 2.4%). After adjustment for clinically significant variables, frail patients were at almost three-fold higher odds of experiencing in-hospital mortality (odds ratio [OR] 2.83, 95% CI 1.65–4.83; P < 0.001) and were at over two-fold higher odds for postoperative ICU admission (OR 2.07, 95% CI 1.55–2.78; P < 0.001) relative to robust patients.

The prevalence of any postoperative complication was greater among frail patients versus robust patients (29.8 vs. 12.3%). After adjustment, postoperative complications remained greater overall among frail patients compared to their robust counterparts (OR 2.18, 95% CI 1.87–2.54; P < 0.001). Frailer patients were at 26-fold greater odds for experiencing cardiovascular complications such as MI (OR 26.93, 95% CI 7.79–93.13; P < 0.001) compared to robust patients. Frail patients were also at higher odds of experiencing pneumonia, postoperative ileus, acute kidney injury, urinary retention, urinary tract infections, and infections overall relative to robust patients upon adjusted analysis (Table 2).

Frail patients more frequently were discharged to nursing (20.4 vs. 6.2%) and home healthcare facilities (16.5 vs. 8.6%) compared to robust patients. Regression analysis showed that patients identified as frail were at reduced odds of returning home after discharge (OR 0.43, 0.37–0.51; P < 0.001). Frail patients were at a greater likelihood of being discharged to nursing facilities (OR 2.03, 95% CI 1.66–2.47; P < 0.001) and home healthcare facilities (OR 1.41, 1.16–1.69; P < 0.01) compared to robust patients.

Healthcare utilization

Table 3 captures the impact of frailty status on healthcare utilization among patients receiving PEH repair. The median length of stay was 4 (interquartile range [IQR]: 2–8) days for frail patients and 2 (IQR: 1–5) days for robust patients, with an adjusted mean difference (MD) of 1.75 days (95% CI 1.30–2.21 days; P < 0.001) (Table 2). Moreover, frail patients had 84% higher odds of experiencing prolonged length of stay than robust patients (OR 1.84, 95% CI 1.58–2.16; P < 0.001). Median total admissions costs were $18,582 (IQR: $12,320–$29,694) for frail patients compared to $14,561 (IQR: $10,216–$22,343) for robust patients, with an adjusted mean difference of $5631.65 (95% CI $3300.06–$7.963.24; P < 0.001).

Subgroup analysis

Adjusted subgroup analysis by minimally invasive (Supplemental Table 2) PEH repair demonstrated that frail patients receiving minimally invasive PEH repair were at two-to-three-fold increased odds for in-hospital mortality (OR 2.91, 95% CI 1.45–5.83; P = 0.003) and any complications (OR 2.41, 95% CI 2.02–2.88; P < 0.001), as well as an increased length of stay by 1.66 days (MD 1.66, 95% CI 1.26–2.06; P < 0.001) and increased total admission costs (MD $4292.51, 95% CI $2645.76–5939.00, P < 0.001) (Supplemental Table 2). A similar subgroup was created for open procedures, in which frail patients continued to display greater odds of in-hospital mortality (OR 2.96, 95% CI 1.19–7.35; P = 0.019) and complications (OR 1.68, 95% CI 1.22–2.30; P = 0.001), as well as increased length of stay (MD 2.10, 95% CI 0.40–3.80; P = 0.016) and admission costs (MD $10,068.16, 95% CI $794.42–$19,341.89; P = 0.033) compared to robust patients (Supplemental Table 3).

Patients were also stratified by elective (Supplemental Table 4) and emergency (Supplemental Table 5) admissions. When receiving elective PEH repair, the prevalence of in-hospital mortality was 1.7% for frail patients and 0.3% for robust patients, with frail patients experiencing an almost three-fold increase in odds for in-hospital mortality compared to robust patients (OR 2.97, 95% CI 1.33–6.61; P = 0.008). Frail patients also experienced a two-fold increase in odds for any complications (OR 2.07, 95% CI 1.69–2.54; P < 0.001) (Supplemental Table 4). When stratified by emergency PEH repair, frail patients remained at increased odds for in-hospital mortality compared to robust patients (OR 2.77, 95% CI 1.34–5.76; P = 0.006) with rates of 5.3 and 1.1%, respectively, and retained greater odds for any complications (OR 2.26, 95% CI 1.77–2.88; P < 0.001) (Supplemental Table 5). In both subgroups, frail patients experienced greater length of stay and increased total admission costs.

DISCUSSION

Available evidence surrounding PEH repair in the elderly often does not capture the influence of frailty or use standardized measurements to define frailty. This study assessed the impact of frailty on in-hospital clinical outcomes and healthcare utilization among patients receiving PEH repair using a cohort of patients from the 2015 to 2019 NIS. Frail patients were at greater odds of in-hospital mortality, postoperative ICU admission, and postoperative complications compared to robust patients. Healthcare utilization was higher among frailer patients, with a longer length of stay of 1.75 days and a mean difference in total admissions cost of $5600 USD compared to non-frail patients. Subgroup analysis showed that frail patients maintained worse outcomes compared to robust patients, with augmented differences among patients receiving open versus minimally invasive PEH repair and among those receiving emergency versus elective PEH repair. We demonstrate that frailty is an important predictor of in-hospital mortality, complications, and healthcare utilization.

While some studies support that both elective and urgent PEH repair in elderly patients is safe and effective, other studies have demonstrated increased mortality and morbidity in this population.^{5,7,9,20–23} With these discordant results, Chimunkangara et al. showed that the mFI can be applied to patients receiving PEH repair, with correlation between increasing score, postoperative complications, and discharge destination.¹⁴ We applied the mFI among a nationally representative sample in the US and showed that frail patients experienced more postoperative complications, a greater length of stay, and a higher likelihood of being discharged to a destination other than home. Frail patients also experienced almost three-fold greater odds for in-hospital mortality after adjusting for important confounders, including patient age. This is supported by Mack et al., who found that frail patients have 2.5-fold higher odds for in-hospital mortality among 10,456 patients with PEH repair using NIS data from 2016 to 2018.²⁴ Elsewhere, Damani et al. identified 12,422 patients undergoing elective minimally invasive PEH repair, but frailty did not predict mortality in their cohort.²⁵ Notably, they used a shorter 5-item frailty index, which has not been previously validated.²⁶ The validated 11-item frailty score may more accurately capture frail patients that should not receive PEH repair. Furthermore, the original Markov Monte Carlo model recommended watchful waiting among older patients, citing that surgical intervention becomes a suitable option for elective PEH repair once mortality rates reach ≤0.5%.⁴ The present study demonstrated that robust patients have an in-hospital mortality rate of 0.3% with decreased total admission costs, suggesting that elective PEH repair in robust older patients is a viable option, and emphasizing the importance of considering frailty in assessing surgical candidates for elective PEH repair.

The increased rate of mortality, morbidity, and length of stay following PEH repair resulted in greater total admission costs among frail versus robust patients ($18,583 vs. $14,561) and was sustained when stratified by emergent ($26,716 vs. $13,674) versus elective surgery ($15,678 vs. $13,574). Though studies assessing cost-effectiveness of PEH repair are limited in this population, Morrow et al. showed that elective PEH repair was associated with increased costs, with an average cost of $11,771 per patient compared to $2207 per patient with watchful waiting.²⁷ However, up to 42.1% of patients selected for elective PEH repair are deferred for observation or inpatient care due to concerns for varying morbidity, with two-fold greater costs of admission relative to patients that successfully received elective repair.²⁸ In this context, our study underscores the potential influence of frailty assessments for selecting appropriate patients for elective PEH repair, as frailty is an important predictor of physiologic vulnerability and favourable postoperative outcomes.²⁶ Moreover, watchful waiting predictably results in an increase in the number of patients presenting with emergent PEHs.⁵ Polomsky and colleagues found that patients presenting with emergent PEH have elated healthcare costs relative to patients with elective PEH repair.²⁹ This further emphasizes the importance of our findings, as the use of frailty assessments may increase the number of patients selected to safely receive elective PEH repair and decrease the amount of patients that eventually present emergently using a watch-and-wait approach, leading to improved postoperative outcomes and reduced healthcare costs.

Literature synthesizing the evidence in this area are lacking. The most recent clinical practice guideline comes from the Society of American Gastrointestinal and Endoscopic Surgeons guidelines in 2013.³⁰ The authors recommend that patients with symptomatic PEH should receive surgery, particularly when experiencing signs and symptoms suggestive of obstruction.³⁰ These recommendations reflect conventional practice patterns, with an emphasis on operating on symptomatic patients emergently. While updated guidelines are anticipated, physicians and patients may take interest in the present study. As elderly patients carry an increasing burden of illness, and patients with PEH tend to be older individuals, the use of standardized frailty assessments are feasible and can help carefully select the most suitable patients for safe and effective elective PEH repair. Insurers may take interest in these findings, considering that patients in our cohort receiving emergency repair had higher total admission costs and that frail contributed higher total admission costs for patients receiving both elective and emergency PEH repair. Our findings stem from a nationally representative sample of over 10,000 patients in the US, suggesting that frailty assessments ensure that carefully selected older patients can receive elective PEH repair safely while avoiding the negative health consequences associated with acute PEH repair. Future evidence synthesis work must consider both the clinical and economic implications of frailty assessments in selecting optimal surgical candidates for elective PEH repair. Additional primary literature must assess long-term outcomes with outpatient follow-up using frailty assessments for patients receiving PEH repair.

The results of this study must be interpreted in the context of specific limitations. Patients included in this study were undergoing PEH repair and consequently may represent a healthier population than the general public. This may impact the generalizability of our findings. Furthermore, the use of national registry data may introduce the risk of selection bias; however, regression analysis was employed to adjust for clinically significant confounding variables. Nonetheless, these data are at risk of residual confounding due to unmeasured, impactful baseline covariates such as prior abdominal surgical history, prior PEH repair, use of certain medications preoperatively (e.g. steroids and biologics), and smoking status. Moreover, due to the current ICD-10 CM procedure codes, the type of diaphragmatic hernia such as paraesophageal, sliding, or mixed could not be identified. As such, the prevalence of specific hernia types and their effect on our results is unknown. Additionally, the NIS provides inpatient admission data but does not report outpatient data such as mortality or longer-term postoperative complications. Without outpatient follow-up data, healthcare utilization and associated costs may be underestimated, and findings cannot be extrapolated to long-term outcomes. Furthermore, due to database limitations, granular detail for PEH repair in octogenarians is unavailable.²² While the large number of included patients is certainly a strength of the present study, a small number of certain outcome events, such as myocardial infarction, resulted in wide 95% CIs upon analyzing the data and thus reduce our confidence in some of the effect estimates. Nevertheless, the data presented in this study stem from a national register and include both clinical and economic data, providing generalizable and pragmatic information for patients, physicians, and policymakers.

CONCLUSION

PEH repair in elderly patients is safe, but frail patients have an increased rate of in-hospital mortality, postoperative ICU admissions, and in-hospital overall morbidity. Taken together, clinicians should consider patient frailty when identifying the most appropriate surgical candidates for PEH repair. This can offer valuable insight and help guide preoperative patient counselling. Frailty assessments among patients receiving PEH repair also offer economic implications, as they predict increased healthcare utilization including length of stay and total admission costs. Large dataset validation and refinement using the 5-item mFI may facilitate the application of frailty assessments in practice, while long-term follow-up data are needed to confirm the durability of frailty in predicting outpatient mortality, morbidity, and healthcare utilization.

DATA AVAILABILITY

The data used in the current study is available upon request.

Financial support: This article received no funding for its work.

Conflicts of interest: Nothing to disclose.

References