Frailty in elderly patients with acute colonic diverticulitis is associated with worse in-hospital outcomes: a nationwide analysis

Introduction

Acute colonic diverticulitis (ACD) refers to the inflammation of colonic diverticula, which can occur in approximately 10-25% of individuals with diverticulosis [1]. This condition can lead to various systemic and local complications, including sepsis, pericolonic abscesses formation, fistulas, bowel obstructions, bleeding, and perforations [2]. The risk of developing acute diverticulitis increases with age, ranging from 10% in individuals under 50 years old to 33% in those between 60 and 69 years old [3].

Like ACD, frailty is also believed to be highly prevalent in the elderly [4]. It is defined as a vulnerability to adverse health outcomes secondary to decreased resistance or reserve to stressors, resulting from a decline in the performance of multiple integrated physiological systems, and it is closely related to aging [5,6]. Various methods have been proposed to assess frailty based on objective performance (phenotypic frailty), such as the Fried Frailty Phenotype, or based on comorbidities, disabilities or social factors (deficit accumulation or index frailty) [5,7]. Various screening tools have been developed based on these methods; however, to date, there is no gold standard method of screening for frailty. This poses a major challenge to the development of successful interventions [8,9]. The presence of frailty can potentially affect inpatient outcomes related to conditions that preferentially affect the elderly, and the identification of frailty in these patients can provide an opportunity for early intervention in the population at risk.

Therefore, we investigated the effects of frailty on in-hospital outcomes of ACD. For this purpose, we used the Hospital Frailty Risk Score, which is based on the International Classification of Diseases, Tenth Revision (ICD-10) codes and was developed using the Hospital Episode Statistics inpatient database [10]. Our study was performed using the National Inpatient Sample (NIS), the largest national inpatient database in the United States of America (USA).

Materials and methods

Statistical analysis

The analyses were performed using STATA version 17.0 (StataCorp, College Station, Texas, USA). ACD patients were stratified into frail (Hospital Frailty Risk Score ≥5) and non-frail (Hospital Frailty Risk Score ≤4) cohorts. To mitigate the inherent selection bias in this retrospective study, we decided to perform a propensity-matched analysis. A propensity score was calculated for each hospitalization, based on biodemographic and hospital characteristics, as well as a list of comorbidities detailed in Tables 1 and 2. Frail patients were then matched to non-frail patients using a 1:1 matching method within 0.05 standard deviation of the calculated propensity score, and the covariate balance was analyzed using a covariance plot before and after matching. A receiver operating characteristic (ROC) plot was generated for evaluation of the performance of age combined with frailty, versus age only, in predicting in-hospital mortality (Fig. 1). Matched cohorts were analyzed for primary and secondary outcomes using two-tailed non-parametric tests, including Pearson’s chi-squared test for categorical variables and the Wilcoxon rank-sum (Mann-Whitney) test for continuous variables. The categorical variables were reported as frequency (N) and percentage (%), while continuous variables were reported as median and interquartile range (IQR). The inpatient cost was adjusted for inflation up to January 2023 using the consumer price index, in order to calculate the inflation-adjusted median inpatient cost [12]. A P-value of 0.05 or less was set as the threshold for statistical significance, and all P-values were 2-sided. The study was exempt from institutional review board approval or patient consent, as the NIS databases contain de-identified patient information and are available publicly. The study findings are reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [13].

Table 1 Biodemographic and hospital characteristics of acute colonic diverticulitis-related hospitalizations in the United States of America in 2016-2020 stratified by frailty status

Table 2 Elixhauser comorbidities among acute colonic diverticulitis-related hospitalizations in the United States in 2016-2020 stratified by frailty status

Figure 1 ROC plots for age and frailty combined (red) vs. age only (black) as primary predictors of in-hospital mortality ROC, receiver operating characteristic; AUC, area under the curve

Results

A total of 516,725 patients were identified and included in our study after application of the inclusion and exclusion criteria; of these, 241,095 (46.7%) were categorized as non-frail and 275,630 (53.3%) as frail patients, as detailed above. Frail patients were older, and presented a higher burden of comorbidities as indicated by their higher Elixhauser Comorbidity Index. The majority of patients in both cohorts were white females, hospitalized in large teaching hospitals in the southern USA (Table 1).

Our matching process resulted in no significant differences in biodemographics, hospital characteristics or major comorbidities among the matched cohorts, as indicated by a P-value >0.05 (Table 1 and Table 2). The difference in median hospital frailty risk score between the 2 cohorts remained statistically significant after matching (7.2 vs. 2.3, P<0.001), as the matching process was performed after the initial patient stratification. The ROC plot showed a better prediction of in-hospital mortality using age and frailty combined compared to age alone (Fig. 1). Matched frail patients showed significantly greater in-hospital mortality (1.1% vs. 0.2%, P<0.001), median LOS (4 days vs. 3 days, P<0.001) and inflation-adjusted median inpatient cost (10442 vs. 9028 USD, P<0.001) compared to their non-frail counterparts. Frailty in ACD was also significantly associated with higher rates of complications, including abscess formation, intestinal perforation, gastrointestinal fistula formation, sepsis without shock, sepsis with shock, acute kidney injury, hypovolemic or hemorrhagic shock, need for blood transfusion, cardiac arrest, and need for intensive care (all P-values <0.001). However, frail patients experienced less frequent gastrointestinal bleeding compared to non-frail patients.

There were also noteworthy differences between the two cohorts in the need for colectomy and colostomy. The frail ACD cohort exhibited a higher prevalence of open colectomy and colostomy (both P-values <0.001), whereas the rate of laparoscopic colectomy was lower in frail patients compared to their matched counterparts (P<0.001) (Table 3).

Table 3 Primary and secondary outcomes of acute colonic diverticulitis-related hospitalizations in the United States in 2016-2020 stratified by frailty status

Discussion

Diverticular disorder is a common condition, and its prevalence increases with advancing age. The impact of age is particularly evident, with rates ranging from 10% among individuals under 40 years of age to a staggering 66% among those aged 80 and above [14]. Approximately 10-25% of individuals with diverticular disease will eventually experience ACD [1]. Given the growth of the aging population, it is anticipated that the incidence of ACD will continue to rise [15]. Numerous published studies have investigated the relationship between age and outcomes related to ACD [15-18]. In our research, we investigated the impact of frailty on elderly patients hospitalized with ACD; to the best of our knowledge, no previous studies have investigated this.

Frailty is closely linked to the aging process and is characterized by an increased vulnerability to stress. This vulnerability, in turn, elevates the risks of adverse outcomes such as sickness, falls, hospitalization, delirium and disability, as well as mortality. Therefore, it is a significant concern in terms of susceptibility to poor outcomes [5,6,19]. Frailty does not have a single etiology; rather, it is a product of an imbalance in the intricate interactions between the body’s different physiological systems, contributing to compromised homeostasis [5]. The observation that a subset of frail patients lacks significant comorbidities suggests that these imbalances can result either from age-related decline, such as age-related anorexia or loss of muscle mass, or as a consequence of comorbidities [5]. In our effort to contribute to the ongoing discussion, we present a comprehensive 5-year analysis based on the NIS database, shedding light on clinical outcomes among both frail and non-frail patients with ACD. We used the Hospital Frailty Risk Score for our analysis, as, like the NIS database, it uses ICD-10 codes and aligns closely with our methodology.

Based on the Hospital Frailty Risk Score, the prevalence of frailty in our study population was found to be over 50%. The prevalence of frailty varies depending on the assessment method used. While previous studies that included patients older than 65-70 years reported a prevalence between 4% and 16.3%, our patient population exhibited a significantly higher prevalence [20-22]. The patients in our study were older than the patients included in the aforementioned studies, as we specified a minimum age of 75 years as an inclusion criterion. However, the high frailty prevalence of over 50% in our study patients might still suggest that elderly patients with ACD are at particularly higher risk, and should receive more tailored care to prevent adverse in-hospital outcomes.

The other notable findings of our study are a higher in-hospital mortality, a longer LOS, and a higher inpatient cost associated with frailty status. Additionally, frail patients experienced a higher risk of local and systemic complications of ACD requiring admission to critical care units, and a greater need for inpatient laparoscopic and open colectomy or colostomy procedures. This indicates a greater severity of ACD in frail patients compared to their propensity score-matched non-frail counterparts. When we compared age combined with frailty to age alone, age combined with frailty was better at predicting in-hospital mortality, as indicated by the area under the ROC curve (Fig. 1). Numerous published articles report age as an independent factor for elevated mortality and complications in ACD patients [15-18]. As far as we are aware, no published article currently available investigates frailty as an independent factor within this context.

This study, based on the NIS database, possesses both strengths and limitations. An important strength lies in its large sample size, which adds to the statistical power and generalizability of the study findings. Furthermore, the NIS database provides a comprehensive dataset, including patient demographics, diagnoses and procedures, allowing analysis of various variables in depth. A limitation is that the retrospective nature of the study may have introduced selection bias and limited the ability to establish causal relationships. However, the propensity-matched technique was used to eliminate these biases and unmeasured confounders. Additionally, the reliance on administrative codes for diagnoses and procedures within the NIS database could potentially lead to inaccuracies and misclassifications [23]. NIS also lacks detailed clinical information regarding vital signs, laboratory results and longitudinal follow up [24]. Despite these limitations, the study offers valuable insights into the subject matter, which can be further explored and substantiated through additional research.

In summary, this study found that frailty contributes independently to adverse inpatient outcomes related to ACD in the elderly, including higher mortality rates, prolonged hospital stays, greater healthcare costs, and higher complication rates. While age remains an important factor, frailty emerges as a distinct and substantial determinant of patient outcomes. These findings underscore the importance of factoring in frailty when clinically managing elderly ACD patients, and offer valuable insights for future research and interventions aimed at improving the care and outcomes of this vulnerable population.