Article Text
PDF
Abstract
Objective Non-malignant pleural effusions (NMPE) are common in hospitalised patients. Data on NMPE inpatients are scarce and the factors influencing the prognosis are unknown.
Design This was a retrospective cohort study.
Setting and participants We conducted a retrospective cohort of inpatients (n=86 645) admitted to the Chinese PLA General Hospital from 2018 to 2021, based on electronic medical records. The observations of 4934 subjects with effusions confirmed by chest radiological tests (CT or X-ray) without a diagnosis of malignancy were followed during admission. Logistic regression was used to analyse organ damage and other factors associated with in-hospital death. Patients were clustered according to their laboratory indicators, and the association between the clustering results and outcomes was studied.
Outcome The outcome of this study was in-hospital mortality.
Results Among 4934 patients, heart failure + pneumonia + renal dysfunction was the most common (15.12%) among 100 different diagnostic groups. 318 (6.4%) patients died during hospitalisation. Lung (OR 3.70, 95% CI 2.42 to 5.89), kidney (OR 2.88, 95% CI 2.14 to 3.90) and heart (1.80, 95% CI 1.29 to 2.55) damage were associated with in-hospital mortality. Hierarchical clustering of laboratory indicators (estimated glomerular filtration rate, white blood cell count, platelet count, haemoglobin, N-terminal pro-B-type natriuretic peptide, serum albumin) demonstrated the ability to discriminate patients at high risk of in-hospital death.
Conclusion Comorbidities and multiorgan failure are the prominent characteristics of NMPE patients, which increase the risk of in-hospital mortality, and comprehensive intervention for specific comorbidity patterns is suggested.
- Heart failure
- Respiratory infections
- Chronic renal failure
Data availability statement
Data are available upon reasonable request. Some or all datasets generated during and/or analysed during the current study are not publicly available but are available from the corresponding author upon reasonable request.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
STRENGTHS AND LIMITATIONS OF THIS STUDY
This study provides detailed data on inpatients with non-malignant pleural effusions, focusing mainly on comorbidities associated with pleural effusions.
This study has the largest sample size compared with other studies of non-malignant pleural effusions.
One limitation is that we could not determine the exact cause of the effusion, which was the same in most clinical scenarios.
Introduction
Pleural effusion (PE) is a common clinical manifestation in hospitalised patients and is associated with significant morbidity and mortality. It affects more than 3000 people per million population each year.1 In China, a multicentre cross-sectional study reported that the estimated prevalence of PE was 4684 per million, and the median cost of hospitalisation was over US$2275.55.2 In the USA, PEs affect approximately 1.5 million people per year.3
Many conditions can contribute to PE. To date, more than 50 causes have been recognised.1 PEs are commonly divided into malignant PEs and non-malignant pleural effusions (NMPE),4 the latter usually associated with cardiac, renal or hepatic dysfunction. As a common and obvious manifestation, PE has been used as a clinical predictor of disease severity, such as the acute pancreatitis severity bedside index, pneumonia severity index, and so on.5 6 Although malignant PEs are usually associated with worse prognosis, high mortality in patients with NMPE due to organ dysfunction has been noted.4 A previous study reported that patients with NMPE secondary to organ dysfunction had an extremely high 1-year mortality rate ranging from 25% to 50%.7
However, existing research on NMPE was mostly limited to patients undergoing thoracentesis, with a small sample size and lacking detailed clinical features and risk factors for prognosis. This study describes the characteristics and in-hospital mortality of NMPE patients in medical wards based on the electronic medical record (EMR) with a large sample size, aiming to help identify high-risk inpatients requiring careful monitoring or active intervention.
Methods
Study design and population
We conducted a retrospective cohort study of internal medicine inpatients admitted to PLA General Hospital, the largest tertiary care academic hospital in China. The EMR of PLA General Hospital is a unified database with data collected by each department.8 We included all patients who (1) were admitted to the internal medicine wards, excluding the oncology and haematology departments, between 1 January 2018 and 31 December 2021; (2) were diagnosed with PE by chest radiological examinations (CT or X-ray). The date of the first radiological text finding PE was used as the index date. Patients with a discharge diagnosis of malignancy such as lung cancer, breast cancer or leukaemia were excluded. All patients were included for the first admission if they had multiple admissions.
Exposure and outcome
The exposure of the study is the diagnosis of PE confirmed by CT or X-ray images, which were indicated by two radiologists independently. The outcome was the in-hospital mortality reviewed from EMR.
Covariates
Baseline characteristics were obtained from the EMR, including demographics, departments, diagnosis, procedures, treatment information and laboratory results. The laboratory results closest to the index date (date of CT or X-ray examination) were recorded.
Potential pathogenic diagnoses thought to be associated with PE were recorded, including heart failure, pneumonia, renal dysfunction, nephrotic syndrome, connective tissue disease, cirrhosis, tuberculosis, pancreatitis and pulmonary embolism. The above diagnoses were defined according to their corresponding International Statistical Classification of Disease and Related Health Problems, 10th Revision (ICD-10) codes (online supplemental e-Table 1). To improve the sensitivity of the diagnoses, the supplementary criteria for the diagnosis of heart failure referred to the European Society of Cardiology 2021 guidelines for the diagnosis and management of acute and chronic heart failure,9 including left ventricular systolic dysfunction (left ventricular ejection fraction <50%) on echocardiography during hospitalisation or reduced diastolic function on echocardiography; structural changes in the atria and ventricles; moderate-to-severe aortic stenosis or aortic regurgitation, moderate-to-severe mitral regurgitation or mitral stenosis; abnormal ECG, with N-terminal pro-B-type natriuretic peptide (NT-proBNP) >150 pg/mL. The extended definition of renal dysfunction referred to the highest eGFR during hospitalisation of less than 60 mL/min per 1.73 m2, calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD EPI) equation formula.10 The diagnosis of pneumonia referred to the ICD-10 codes on admission (online supplemental e-Table 1) or implied pneumonia on chest CT or X-ray. The diagnosis of anaemia was defined as haemoglobin <120 g/L in males and <110 g/L in females. Effusions were classified as either transudative or exudative according to Light’s criteria11 and as either bilateral or unilateral based on image reports.
Supplemental material
Patient and public involvement
Patients or the public were not involved in the design, or conduct, or reporting, or dissemination plans of our research.
Statistical analysis
The association of organ damage with an increased risk of in-hospital death was assessed by logistic regression. Multiple imputations were used to overcome missing data. We focused on three types of organ damage: ‘heart’ (diagnosed with heart failure), ‘kidney’ (diagnosed with renal dysfunction or nephrotic syndrome) and ‘lung’ (diagnosed with pneumonia) and estimated the significance of their interactions using likelihood ratio tests. Initially, logistic regression models were fitted without adjustment. Model 1 was adjusted for age and sex (female vs male). Model 2 was additionally adjusted for mechanical ventilation, thoracentesis and intensive care unit (ICU) admission. ORs with 95% CIs and p values were reported for comparison. Covariates for all models were selected on the basis of clinical plausibility without using probabilistic selection criteria. To evaluate the impact of the imbalanced outcome, the analysis was repeated with Firth’s logistic regression.12 Patients were clustered according to their laboratory indicators (eGFR, WBC count, platelet count, haemoglobin, NT-proBNP, serum albumin) and the association between clustering results and outcomes was studied. All analyses were performed using R V.4.2.0. Two-sided nominal p values <0.05 were considered statistically significant.
Results
Baseline characteristics
A total of 86 645 inpatients were screened between 1 January 2018 and 31 December 2021. Of these, 63 395 (73.1%) underwent chest radiography (CT/X-ray), of whom 6956 had PEs. After excluding 2019 patients with a discharge diagnosis of malignancy and 3 with an unclear diagnosis, 4934 patients were included in the analysis, representing 7.78% of internal medicine inpatients who underwent chest radiological tests. The flow chart is shown in online supplemental e-Figure 1.
Median age of NMPE patients was 64 years old (IQR 49–78 years old). More than one-half were men (63.4%), 3155 (63.9%) had bilateral PEs, 2848 (57.7%) had hypertension, 3002 (60.8%) had anaemia, 7.9% underwent thoracentesis, 16.5% were admitted to the medical ICU, 3117 (63.2%) were treated with loop diuretics and 3302 (66.9%) were treated with antibiotics. The median length of hospital stay was 18.9 days (table 1).
Basic characteristics of internal medicine inpatients with non-malignant pleural effusions
The incidence of NMPE was highest in nephrology (19.1%), followed by respiratory (11.4%) and rheumatology (7.8%) (online supplemental e-Table 2).
Among the 4934 patients with NMPE, there were 100 different combinations of diagnoses. We used the Venn diagram in online supplemental e-Figure 2 to show the 36 sets of diagnoses that had more than 10 cases. The most common diagnosis combination was renal dysfunction + heart failure + pneumonia (15.12%), followed by pneumonia (14.47%), and then heart failure + pneumonia (11.76%). The most common diagnosis of NMPE in subjects over 60 years of age was renal dysfunction + heart failure + pneumonia (18.19%), followed by heart failure + pneumonia (16.94%), and then pneumonia (16.62%). In subjects younger than 60 years old, the most common diagnosis was pneumonia (11.63%), followed by renal dysfunction + heart failure + pneumonia (11.06%), followed by renal dysfunction + heart failure (9.98%) (online supplemental e-Table 3). The demographic characteristics (age and sex) and comorbidities (diabetes, hypertension and anaemia) in patients with different combinations of diagnoses are shown in online supplemental e-Table 4 and 5. Patients with cirrhosis, tuberculosis + pneumonia and renal dysfunction + connective tissue disease + pneumonia were more likely to have unilateral effusions (figure 1). Of 367 patients with pleural fluid examination results, a total of 183 (49.9%) patients had exudative effusions (online supplemental e-Table 6).
Proportion of bilateral effusions in patients with NMPE according to different diagnoses. Diagnostic combinations with fewer than 20 cases were not shown. CI, cirrhosis; CTD, connective tissue disease; HF, heart failure; NS, nephrotic syndrome; PA, pancreatitis; PE, pulmonary embolism; PN, pneumonia; RD, renal dysfunction.
In-hospital mortality of NMPE patients
318 patients died during hospitalisation, giving a mortality rate of 6.4%. In-hospital mortality tended to increase with age, and the mortality rate in patients under 60 years of age was 1.1% (figure 2). Online supplemental e-Figure 3 shows in-hospital mortality according to different diagnostic combinations.
In-hospital mortality rates of medical inpatients with NMPE in different age ranges.
Multiorgan injury and in-hospital death
Of the 4934 NMPE patients, 58.9% had pneumonia, 53.7% had heart failure, 42.4% had renal dysfunction and 15.5% had nephrotic syndrome, making a total of 4390 patients with damage to one of the above 3 organs, representing 89.0% of subjects. The percentage of concomitant damage to one of the other two organs was 86.3%, 79.7% and 68.4% in patients with heart, kidney and lung damage, respectively. 1000 patients had concomitant damage to the three organs, accounting for 37.7% of heart, 40.8% of kidney and 34.4% of lung damage (online supplemental e-Figure 4).
Online supplemental e-Figure 5 and figure 3 show the results of univariate and multivariate logistic models investigating the association between multiorgan damage and in-hospital death. Age, mechanical ventilation and ICU admission were independently associated with the risk of in-hospital death. Comparing crude, model 1 and model 2, the ORs for heart and lung damage decreased after including additional covariates in the models, while the OR for kidney damage increased. In model 2, the three types of organ damage were all associated with a significant increase (p<0.001) in the likelihood of in-hospital death (lung: OR 3.70, 95% CI 2.42 to 5.89; kidney: OR 2.88, 95% CI 2.14 to 3.90; heart: OR 1.80, 95% CI 1.29 to 2.55). To investigate whether combinations of organ damage should be considered, a model with additional interaction terms was built. The likelihood ratio test between this interaction model and model 2 gives a p value of 0.53, indicating that the added interaction terms may not improve the model. Firth logistic regression model was almost identical to the original model (online supplemental e-Table 7), so the problem of imbalanced data might be negligible in our analysis.
The results of multivariate logistic regressions modelling in-hospital death. Crude was the unadjusted model, model 1 was adjusted by age and gender (female compared with male), and model 2 was additionally adjusted for mechanical ventilation, thoracentesis drainage and admission to ICU.
Unsupervised clustering results
We used the hierarchical clustering algorithm to cluster patients according to their laboratory indicators. The cluster dendrogram and visualisation of the first two principal components are shown in online supplemental e-Figure 6 and 7. Surprisingly, the resulted groups seemed to be strongly associated with in-hospital death, as shown in figure 4. Differences in outcomes between groups (p<0.001) suggest that it is feasible to predict the risk of in-hospital death from laboratory results. The group with the lowest in-hospital mortality had a significantly lower albumin level than the others, and 79.7% were diagnosed with nephrotic syndrome.
Plot of clustering results regarding patient outcome. It can be seen that the third group had much lower rate of in-hospital death.
Discussion
Our study showed that among medical inpatients who underwent chest radiology (73.1% of all medical inpatients), the incidence of NMPE was 7.78%. Heart failure + pneumonia + renal failure was the most common (15.12%) among 100 different diagnostic groups. While 318 (6.4%) patients died during hospitalisation, in-hospital mortality in patients younger than 60 years was only 1.1%. After full adjustment, patients with heart, kidney and lung damage were 1.80, 2.88 and 3.70 times more likely to die compared with patients without the above organ damage, respectively. The hierarchical clustering result demonstrated the ability to distinguish patients at high risk of in-hospital death using laboratory indicators.
We included 4934 patients with NMPE confirmed by radiological tests who were hospitalised from January 2018 to December 2021, which is the largest sample size in studies related to NMPE. It’s worth noting that 73.1% of inpatients in our study underwent chest radiography, which is higher than previous studies (43.6%–46.4%),13 14 so our subjects are more representative. In addition, most studies used only chest radiographs to identify PEs, but CT scans are more sensitive and specific for diagnosing effusions and superior to other examinations in evaluating lung lesions to find aetiologies. In our study, almost 70% of patients underwent chest CT, which provided confidence in the identification of PE and the relevant diagnosis. Finally, we also collect basic laboratory test results to provide more detailed information about the clinical characteristics of this population.
Thoracentesis should be performed in almost all patients with pleural fluid that is new or of uncertain aetiology. Although thoracentesis is a well-tolerated procedure, complications can lead to increased morbidity, mortality and healthcare costs. In our study, 8% of patients with NMPE underwent thoracentesis, and the frequency was similar to a previous study that found 10% of patients in general medical inpatients with PE, which included patients with malignant PE.13 In another prospective study in which thoracentesis was performed in 39.6% of patients, respiratory patients accounted for up to 61.4%.14 Thus, the frequency of thoracentesis in our study was representative of internal medicine inpatients, reflecting the fact that in most cases we couldn’t determine the aetiologies from the nature of the effusions, which are more dependent on the underlying diseases.
In our study, the nature of the effusions was equally likely to be exudative or transudative according to Light’s criteria. The proportion of exudates was high in pneumonia (85.1%). However, in patients with other diagnostic combinations with concomitant pneumonia, the proportion of exudates was much lower. We know that Light’s criteria have been the gold standard for differentiating exudates for almost half a century, but they were tested in strictly selected patients and restricted to a single cause. Studies on the nature of effusions contributed by both exudative and transudative causes were lacking. One retrospective study reported that in patients with chronic heart failure and concomitant pneumonia, 8 of 30 effusions remained transudate according to Light’s criteria.15 In this study, pneumonia was diagnosed based on the presence of systemic inflammation syndrome and respiratory symptoms, and patients with other complicated conditions were excluded. In our study, pneumonia was determined based on the patient’s admission diagnosis and image findings, which may explain the lower proportion of exudates in our study.
NMPEs are likely to indicate a poor prognosis the same as malignant PEs, which were previously described in patients under some specific conditions, such as community-acquired pneumonia,16 acute pulmonary oedema,17 maintenance dialysis,18 acute pancreatitis19 and cirrhosis.20 These studies covered almost all common causes of PEs, attracting the attention of researchers and clinicians. In a retrospective study of 744 consecutive inpatients of internal medicine who underwent chest X-ray, the 30-day mortality rate of NMPE due to organ failure was even higher than that of malignant PE.13 In a retrospective study of 1936 inpatients who underwent chest X-rays, the 30-day mortality rate of chronic heart failure and parapneumonic effusion was 19.2% and 28.9%, respectively.21 The 1-year and 5-year prognosis of NMPE was 81%–100% and 70%–83% in the retrospective study by Ercan et al of 251 patients with incidental PE diagnosed during echocardiographic evaluation.22 In a prospective study which included 308 consecutive patients undergoing thoracentesis, concomitant benign causes accounted for the largest proportion,23 having the highest 30-day and 1-year mortality rates among NMPE (29% and 55%, respectively), followed by patients with heart failure or pneumonia alone. As a retrospective study, it is difficult to determine the real cause of PE, but it is feasible to describe by diagnoses associated with the cause of effusions. Our results showed that heart failure + pneumonia + renal dysfunction was the most common diagnostic combination, which remained the second most common diagnosis in patients under 60 years old. Among the existing studies, only DeBiasi, Jonathan T. Puchalski, and Markatis E studied concomitant benign causes when discussing aetiologies, which was the main cause of NMPE in the first two studies.14 23 24 Despite the differences in the research centres and the methods used, which make it difficult to compare the studies, all the existing studies, with the exception of that of Ercan et al, came to the similar conclusion that patients with PE, including NMPE, have a high mortality rate.7 13 21–24 The mean age of the participants in Ercan et al was the lowest in the above studies, which may be the reason for the higher survival rate. The in-hospital mortality rate of patients younger than 60 years old was only 1.1% in our study, implying that NMPE is more valuable as a prognostic marker in the elderly. Multivariate logistic regression also showed that age was an independent risk factor for in-hospital mortality, which was reported previously.14 Each of the damage in kidney, heart or lung was associated with a significant increase in the likelihood of in-hospital death and there were 20.3% of patients with concomitant lung, heart or kidney damage, which means that these patients were much more likely to die during hospitalisation compared with patients without damage to the three organs. A study of the impact of multimorbidity on death and loss of social functioning in elderly patients reported that multimorbidity affected 70.4% of people and was the cause of death for 69.3% of patients, reducing life expectancy by 7.5 years and becoming the most prevalent and burdensome clinical and social problem.25 Since the prevalence of multiorgan damage in NMPE patients is high and associated with high in-hospital mortality, the interaction between diseases and medications must be carefully considered in the treatment of those patients. Further research is required to develop stratified therapy for typical multimorbidity patterns in order to improve outcomes in these potentially high-mortality patients.
We had some limits in our study. First, as a retrospective study, it was challenging to determine the precise cause of the patient’s PE; as an alternative, we described the various combinations of diagnoses thought to be associated with the cause of effusions based on previous studies.7 Since patients with NMPE are often comorbid patients with multiple organ damage, we categorised the patients according to the diagnosis of major organ damage: heart, kidney and lung. There may be a certain amount of the misclassification. What’s more, when there are multiple benign causes, it is almost impossible to determine the contribution of each possible cause to the patient’s effusions in actual clinical work. Therefore, we used large statistics to describe and summarise, in the hope of assisting clinicians to better understand this common internal medicine problem. Second, since this was a single-centre study with potentially biased patient characteristics, the conclusion needs to be supported by a larger sample from other centres. Third, the outcome of this study was death during admission, and there was no follow-up for the long-term survival of these patients. Lastly, we didn’t discuss much about treatment of NMPE because of the high heterogeneity in aetiologies and disease severity of NMPE patients.
Conclusion
Comorbidities and multiorgan failure are the prominent characteristics of NMPE patients, which increase the risk of in-hospital mortality, and comprehensive intervention for specific comorbidity patterns is suggested.
Data availability statement
Data are available upon reasonable request. Some or all datasets generated during and/or analysed during the current study are not publicly available but are available from the corresponding author upon reasonable request.
Ethics statements
Patient consent for publication
Ethics approval
The study has been approved by the Ethics Committee of PLA General Hospital and exempted from informed consent (No. S2022-730-01). The study was exempted from informed consent for it was a retrospective observational study using EMR and no personal information was involved.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
DW and YN contributed equally.
Contributors XC and PL, First Medical Center of Chinese PLA General Hospital, are the guarantors. Conceptualisation: PL and XC. Methodology and formal analysis: PL, YN and DW. Investigation: YM. Interpretation of data: LZ, LT and YM. Drafting manuscript: DW, YN and QZ. Supervision: XS, GC, PL and XC. All authors have read and agreed to the published version of the manuscript.
Funding This work was supported by the National Key Technology R&D Program (No. 2022YFC3602005, No. 2021YFC3002203) and National Natural Science Foundation of China (No. 32141005, No. 82274327).
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.