1. Introduction
Community-acquired pneumonia (CAP) is one of the leading causes of hospitalization and is responsible for approximately 2.5 million deaths worldwide every year [
1,
2]. In Europe, CAP also leads to high hospitalization rates, causing a significant financial burden for the healthcare system [
3,
4]. The financial impacts of CAP due to prolonged hospitalizations or increased hospitalization rates have been documented in previous studies [
5,
6,
7]. Current guidelines emphasize the importance of discharging patients as soon as they achieve clinical stability and have access to a safe environment where continuity of care can be ensured [
8]. The recommendations particularly underline the importance of increasing outpatient treatment to decrease the cost of hospitalizations and the risk of hospital-acquired complications [
8]. However, the length of hospital stay (LOHS) for patients with CAP continues to be variable and for that reason, the development of accurate models to predict the LOHS using patients’ baseline profiles from an early stage is needed. Obtaining accurate predictive models upon admission has multiple advantages. First of all, they allow us to identify the profiles of patients at risk of prolonged hospitalization, and whenever possible, to promptly act on modifiable factors. Moreover, discharge strategies can be improved. The implementation of a precise prediction model would additionally permit the evaluation of hospital performance, thereby fostering advancements in hospital management.
The LOHS in patients with CAP can be influenced by a variety of factors, including sociodemographic, health-related and hospital care-related characteristics [
9,
10,
11,
12,
13,
14,
15,
16,
17,
18,
19,
20,
21]. A number of previous studies investigating factors that influence the LOHS in CAP identified patient-related variables such as advanced age and specific comorbidities, in addition to disease severity, as predictors of a prolonged LOHS [
9,
10,
11,
12,
13]. Other studies direct their research focus to laboratory values [
14,
15,
16], while others concentrate on therapies [
17,
18,
19] or other interventions during hospitalization [
20,
21]. Due the wide variety of influencing factors, there is no uniform method for predicting the LOHS in CAP patients; moreover, as mentioned above, several studies included factors that are not available at the time of admission, hindering the chance of predicting the LOHS in the first days of hospitalization.
The primary aim of this study was to identify which factors may affect the length of stay of patients admitted for CAP. The identification of patient characteristics influencing the LOHS may help decision makers properly plan hospital management. Particularly, we retrospectively explored whether the primary outcome, the LOHS for CAP, was associated with commonly available sociodemographic and health-related variables that are measurable at the time of admission to the hospital.
Despite advances in therapy, the mortality rate associated with this disease is still high (6–10%). While a shorter LOHS may decrease hospital costs, it may also negatively impact the quality of care [
22]. Moreover, research has indicated that rehospitalization and mortality rates are high among patients with CAP who survive the initial admission. This is primarily attributed to factors related to the aging population, like the presence of multiple medical conditions and other health fragilities [
23]. Most elderly CAP patients require special attention from health care professionals after discharge to reduce rehospitalization and mortality rates [
24]. For this reason, this study analyzed factors associated with rehospitalization within 6 months and all-cause mortality (30-day and one-year mortalities) as secondary outcomes.
4. Discussion
This retrospective observational cohort study of patients with CAP showed that the LOHS is influenced by demographic factors such as an older age and female gender and by disease-specific factors like the qSOFA score and atypical pneumonia. Other factors, such as other types of comorbidities, vital signs (other than included in the qSOFA) and laboratory values at admission, were not associated with a longer LOHS.
Interestingly, our results show that women had worse outcomes compared to men. Gender differences have been observed in the clinical course, and outcomes of people with CAP and, historically, men have been found to have worse outcomes, particularly in terms of short- and long-term mortality [
27,
28]. Although little evidence in terms of the LOHS is available, our results are consistent with an international multicenter study by the Community Acquired Pneumonia Organization which followed patients for 10 years. In this study, Arnold and colleagues found that women had significantly longer LOHSs and also worse outcomes in terms of time until clinical stability and mortality within 28 days [
29]. Gender differences clearly warrant further confirmation and validation because causal inference cannot be drawn. However, if confirmed in the future, the current concept that female patients have a lower risk than males with CAP may need to be revised and the current scoring system adjusted (for example, the subtraction of 10 points for females in the Pneumonia Severity Index (PSI)).
The quick Sequential (Sepsis-related) Organ Failure Assessment (qSOFA) score is another severity assessment tool and validated prognostic model devised by Seymour et al. [
30,
31] Originally it was developed to predict sepsis using three main clinical criteria, namely altered mental status, low systolic blood pressure and high respiratory rate. In line with other studies, our results also confirm the prognostic validity of the qSOFA score in predicting the length of hospital stay [
30,
32,
33,
34]. The role of qSOFA in the LOHS was confirmed recently by Koch et al. [
35]; however, the impacts of the single items comprising the score were unclear. For this reason, in our study, we also analyzed the items of the qSOFA score separately, and we found that altered mental status (GCS < 15) and blood pressure (Systolic BP ≤ 100) were significantly predictive for the LOHS (for more details, see
Table A1 in
Appendix A). The main advantage of implementing the qSOFA score is that it does not require laboratory tests and allows for rapid and repetitive assessments. In addition to the task force’s recommendation to use the qSOFA tool to further investigate potential organ dysfunction or to initiate or escalate appropriate therapy, our results suggest that the qSOFA score can be integrated into predictive models as a risk predictor for an extended LOHS.
Another point worth discussing is the fact that atypical pneumonia was predictive for an extended LOHS. In community-acquired pneumonia, examples of typical pathogens are streptococcus pneumoniae and haemophilus influenzae, and atypical pathogens are mycoplasma pneumoniae, chlamydia pneumoniae and staphylococcus aureus [
36]. Atypical pneumonia often expresses more unspecific symptoms such as headache, low fever, dyspnea, dry cough and only slightly elevated inflammatory biomarkers; moreover, the clinical presentation can range from mild symptoms to severe illness with respiratory failure or sepsis [
37]. Approximately 7% to 20% of cases of community-acquired pneumonia are believed to be caused by atypical bacterial microorganisms which cannot be detected via Gram staining and pose challenges in terms of culturing [
38]. Moreover, the presence of atypical pathogens in some patients with community-acquired pneumonia (CAP) poses a challenge in the selection of empirical antibiotic treatment. These pathogens are inherently resistant to beta-lactam drugs, which are commonly used as an initial antibiotic treatment [
39]. This dilemma arises from the fact that adding antibiotic coverage specifically for atypical pathogens might carry the risk of adverse effects and promotes the development of antimicrobial resistance [
40]. On the other hand, withholding such coverage may potentially worsen the prognosis if an atypical pathogen is indeed the causative agent of the pneumonia [
41,
42]. Therefore, in our study, we also considered the presence of atypical pathogens as a potential predictor when examining the length of stay in patients with CAP. We recognized that the use or omission of antibiotic coverage for atypical pathogens could influence the clinical course and outcomes, including the LOHS. Hence, the observed association between atypical pneumonia and an extended length of stay in our study could potentially be attributed to the challenges involved in treatment. Specifically, the addition of antibiotic treatment coverage to address atypical pathogens might inadvertently lead to adverse effects, thereby prolonging the hospitalization period. Alternatively, the diagnostic tests employed to identify atypical pathogens may require additional time, contributing to a longer length of stay.
Our secondary outcomes included rehospitalization in the KSBL within 6 months. We detected that in our study population, rehospitalization within 6 months was significantly associated with factors such as diabetes, qSOFA score and rehabilitation after discharge. The percentage of patients who were rehospitalized within 6 months was 31.6%, which is similar to the ranges of two non-recent studies in which the assessed cumulative readmission rates were 22 and 35.6% [
43,
44]. In terms of readmission rate, in fact, it is not common to assess long-term outcomes, as stated by Prescott in a systematic review; the majority of published studies in the literature concentrate their focus on the 30-day readmission, and the percentage varies from a minimum of 16.8 to a maximum of 20.1% [
45]. The most recent study published in 2021 by Averin et al., which assessed late readmission following hospitalization for pneumonia among American adults, analyzed one-year readmission, and the proportion reached 42.3% of the study population [
46]. As previously mentioned the qSOFA score is a validated prognostic tool for sepsis; a recent study investigated the prognostic performance of the qSOFA score for in-hospital mortality and ICU admission [
47], but its accuracy in predicting long-term outcome in terms of rehospitalization within 6 months has not been established.
Interestingly, diabetes was the only chronic health condition associated with rehospitalization within 6 months. Previous studies found a relationship between diabetes and the incidence of CAP [
48] or hospitalization rate [
49] or demonstrated that patients with diabetes have worse discharge outcomes compared to patients without diabetes [
50]. A recently published systematic review and meta-analysis by Fang et al. found that diabetes mellitus was significantly associated with the hospital readmission rate among pneumonia patients (pooled OR = 1.18; 95% CI: 1.08–1.28) [
24], which is confirmed by our results. So, despite advances in treatment, diabetes is still associated with a higher risk of adverse outcomes, and healthcare providers should take this finding into account. Although CAP patients who also suffer from diabetes are at an elevated risk for adverse events and a complicated clinical course, as explained above, further studies are required in order to clarify the underlying mechanisms and the impact of a disrupted glucose metabolism on the development and clinical outcome of CAP in light of rehospitalization rates.
It is worth mentioning that discharge into rehabilitation was found to be significantly associated with rehospitalization. Patients who were sent to rehabilitation after discharge had a higher chance of being readmitted to the hospital within 6 months compared to those who did not attend rehabilitation. This finding contradicts the initial hypothesis that rehabilitation would reduce the risk of rehospitalization. Possible explanations may include the complexity and severity of the underlying conditions requiring rehabilitation, the intensity or duration of the rehabilitation program, or other unmeasured factors that could influence the outcome. In order to further investigate the underlying reasons for the positive relationship between rehabilitation and rehospitalization, we conducted a post hoc analysis comparing the characteristics of patients who were rehabilitated after hospitalization with those who were not rehabilitated. As displayed in
Table A2 in
Appendix A, significant differences were detected. The age of patients who received rehabilitation was significantly higher compared to those who did not (medians of 82.73 and 77.35, respectively;
p-value = 0.004). Similarly, patients who underwent rehabilitation had a significantly longer LOHS (medians of 11 days and 6 days, respectively 6.00;
p-value = 0.001). Other factors, such as chronic cardiovascular disease, COPD, respiratory insufficiency, parapneumonic effusion and cardiovascular complications, also showed significant differences between the two groups. The detected significant differences between the two groups in terms of age, comorbidity burden and hospital complications might explain the positive association between rehabilitation and rehospitalization. Hence, it is necessary to carefully interpret the association between rehabilitation and rehospitalization, considering the confounding effects of these patient characteristics. Moreover, a previous study showed promising results, especially in the short-term, specifically focusing on the 30-day hospital readmission rate [
51]. The majority of the studies investigating the positive effects of rehabilitation mainly focused on different outcomes [
52,
53,
54,
55]. It is important to note that our study differs from these previous investigations as we examined rehospitalization rates within a longer time frame of six months. This extended duration allowed us to capture readmissions that may have occurred beyond the initial 30-day period and provides a more comprehensive understanding of the factors influencing rehospitalization. Further exploration is needed to better understand this unexpected association.
In terms of mortality, we observed that the in-hospital mortality rate was very low: only one patient died during the initial hospitalization, as displayed in
Table 1. This can be explained by the fact that all the patients who were transferred for palliative care or directly sent to another hospital were excluded from this study. On the contrary, we noticed that almost one-quarter of the overall mortality within one year happened within thirty days after discharge (22.9%). This trend was also confirmed by Wadhera in a study using population-based data from almost 16300 patients which was conducted in Germany. The research revealed a significant increase in mortality over time, with a 4.7% increase between in-hospital mortality (17.2%) and 30-day mortality (21.9%) [
56]. Similarly, a study conducted in the United States with a 10-year cohort of about 3 million CAP patients reported a high 30-day post-discharge mortality of 8.2% [
57]. Both multivariable logistic models for 30-day and 1-year mortality revealed that age and a cancer diagnosis were associated with a higher risk of mortality. The findings from our study reinforce prior observations that all-cause mortality during the year subsequent to hospital admission for pneumonia is linked to increasing age and a worsening comorbidity profile [
46,
58,
59]. A recent study concluded that while long-term mortality following CAP was primarily associated with comorbidities, there is potential for early post-discharge complications (within 30 days) to be attributed to CAP-related issues that may benefit from targeted interventions [
60]. However, our results did not find different predictors between the two mortality outcomes. Finally, it is important to note that the LOHS was not significantly associated with mortality nor rehospitalization, implying that a shorter LOHS did not show an increased risk of re-admission or post-discharge mortality.
Strengths, Limitations and Further Research
The novelty of our study lies in its comprehensive encompassing of three important quality indicators as research outcomes (the LOHS, rehospitalization and mortality). The prediction models included various factors such as demographic variables, health-related variables and laboratory values available at the time of admission. A further strength of our research was the possibility to investigate long-term outcomes such as mortality within one year, as these data were available for all patients. However, there are certain limitations to consider. As a retrospective study, the quality of the data depended on accurate documentation in the patient files, which may have resulted in incomplete information. It is especially important to note that the presentation of the severity index data, such as the Pneumonia Severity Index (PSI), was hindered by the absence of available data, thereby limiting the depth of the analysis regarding the severity stratification of CAP cases in this study. Furthermore, information on rehospitalization within six months was limited to a specific hospital due to privacy policies, potentially missing readmissions to other healthcare facilities. However, according to a previous study, in Switzerland, most unplanned readmissions occur within the same hospital [
61]. The conclusions of this study are limited to the definition of CAP according to the IDSA criteria [
8]. The generalizability of other definitions of CAP will have to be assessed. Overall, our study provides a foundation for future research and contributes valuable insights into other aspects of CAP, particularly focusing on the possible predictors of the LOHS, mortality and rehospitalization that are available at the time of admission. The identification of predictors available at the time of admission might help to promptly identify patients who are at a higher risk of adverse outcomes and allow healthcare providers to prioritize their care, allocate appropriate resources and develop personalized management strategies tailored to patients’ specific needs. Further studies are needed to investigate the underlying causes contributing to the association between atypical pneumonia and the LOHS. As mentioned before, predictive models could include data regarding antibiotic coverage and time until the diagnosis of atypical pneumonia. By conducting additional research, a more comprehensive understanding can be obtained, and targeted interventions to optimize patient care and reduce the burden associated with prolonged hospital stays can be developed.