Causes of Death Among Patients Infected with HIV at a Tertiary Care Hospital in China: An Observational Cohort Study

LSTM-Based Prediction Model for Tuberculosis Among HIV-Infected Patients Using Structured Electronic Medical Records: A Retrospective Machine Learning Study

Background

Both HIV and TB are chronic infectious diseases requiring long-term treatment and follow-up, resulting in extensive electronic medical records. With the exponential growth of health and medical big data, effectively extracting and analyzing these data has become the research hotspot. As a fundamental aspect of artificial intelligence, machine learning has been extensively applied in medical research, encompassing diagnosis, treatment, patient monitoring, drug development, and epidemiological investigations. This significantly enhances medical information systems and facilitates the interoperability of medical data.

Methods

In our study, we analyzed longitudinal data from the electronic health records of 4540 patients, gathered from the National Clinical Research Center for Infectious Diseases in Shenzhen, China, spanning from 2017 to 2021. Initially, we employed the fine-tuned ChatGLM to structure the electronic medical records. Subsequently, we utilized a multi-layer perceptron to classify each patient and determined the presence of tuberculosis in HIV patients. Using machine learning-based natural language processing, we structured these records to build a specialized database for HIV and TB co-infection. We studied the epidemiological characteristics, focusing on incidence patterns, patient characteristics, and influencing factors, to uncover the transmission characteristics of these diseases in Shenzhen. Additionally, we used Long Short-Term Memory to create a predictive model for TB co-infection among HIV patients, based on their medical records. This model predicted the risk of TB co-infection, providing scientific evidence for clinical decision-making and enabling early detection and precise intervention.

Results

Based on the refined ChatGLM model tailored for structured electronic health records, the accuracy of symptom extraction consistently surpassed 0.95 precision. Key symptoms such as diarrhea and normal showed precision rates exceeding 0.90. High scores were also achieved in recall and F1 scores. Among 4540 HIV patients, 758 were diagnosed with concurrent tuberculosis, indicating a 16.7% co-infection rate, while syphilis co-infection affected 25.1%, underscoring the prevalence of concurrent infections among HIV patients. Utilizing electronic health records, a Multilayer Perceptron classifier was developed as a benchmark against Long Short-Term Memory to predict high-risk groups for HIV and tuberculosis co-infections. The Multilayer Perceptron classifier demonstrated predictive ability with AUROC values ranging from 0.616 to 0.682 on the test set, suggesting opportunities for further optimization and generalization despite its accuracy in identifying HIV-TB co-infections. In tuberculosis intelligent diagnosis based on laboratory results, the Long Short-Term Memory showed consistent performance across 5-fold cross-validation, with AUROC values ranging from 0.827 to 0.850, indicating reliability and consistency in tuberculosis prediction. Furthermore, by optimizing classification thresholds, the model achieved an overall accuracy of 81.18% in distinguishing HIV co-infected tuberculosis from simple HIV infection.

Conclusion

Combining the Multilayer Perceptron classifier with Long Short-Term Memory represented an advanced approach for effectively extracting electronic health records and utilizing it for disease prediction. This underscored the superior performance of deep learning techniques in managing both structured and unstructured medical data. Models leveraging laboratory time-series data demonstrated notably better performance compared to those relying solely on electronic health records for predicting tuberculosis incidence. This emphasized the benefits of deep learning in handling intricate medical data and provided valuable insights for healthcare providers exploring the use of deep learning in disease prediction and management.

Supervised machine learning algorithms to predict the duration and risk of long-term hospitalization in HIV-infected individuals: a retrospective study

Objective

The study aimed to use supervised machine learning models to predict the length and risk of prolonged hospitalization in PLWHs to help physicians timely clinical intervention and avoid waste of health resources.

Methods

Regression models were established based on RF, KNN, SVM, and XGB to predict the length of hospital stay using RMSE, MAE, MAPE, and R2, while classification models were established based on RF, KNN, SVM, NN, and XGB to predict risk of prolonged hospital stay using accuracy, PPV, NPV, specificity, sensitivity, and kappa, and visualization evaluation based on AUROC, AUPRC, calibration curves and decision curves of all models were used for internally validation.

Results

In regression models, XGB model performed best in the internal validation (RMSE = 16.81, MAE = 10.39, MAPE = 0.98, R2 = 0.47) to predict the length of hospital stay, while in classification models, NN model presented good fitting and stable features and performed best in testing sets, with excellent accuracy (0.7623), PPV (0.7853), NPV (0.7092), sensitivity (0.8754), specificity (0.5882), and kappa (0.4672), and further visualization evaluation indicated that the largest AUROC (0.9779), AUPRC (0.773) and well-performed calibration curve and decision curve in the internal validation.

Conclusion

This study showed that XGB model was effective in predicting the length of hospital stay, while NN model was effective in predicting the risk of prolonged hospitalization in PLWH. Based on predictive models, an intelligent medical prediction system may be developed to effectively predict the length of stay and risk of HIV patients according to their medical records, which helped reduce the waste of healthcare resources.

Clinical Characteristics Associated with Poor Prognosis of Acquired Immunodeficiency Syndrome Patients Complicated with Disseminated Talaromycosis marneffei

Purpose

To analyze the clinical characteristics of AIDS with dTSM, especially in patients with poor prognosis.

Patients and Methods

One hundred and seventy AIDS patients were enrolled in this single-center retrospective study. The epidemiological characteristics, clinical manifestations, laboratory tests, imaging examination, and treatment outcome were collected. Logistic regression analysis was used to estimate the risk of mortality in AIDS patients with dTSM. The predictive value was evaluated using the receiver operating characteristic (ROC) curve.

Results

From 2015 to 2022, the incidence of AIDS with dTSM in the Wenzhou region increased yearly, mainly in young adults. The mortality rate was 16.47%. The most common clinical manifestations were lymph-node enlargement (92.35%) and fever (78.24%). Multivariate logistic regression analysis showed that procalcitonin (PCT), blood urea nitrogen (BUN), shock, and antiretroviral therapy (ART) were the risk factors for poor outcomes. The model comprised four risk factors and showed an excellent prediction performance, with an AUC of 0.987 in the training cohort (95% CI: 0.946-0.999) and 0.976 in the validation cohort (95% CI: 0.887-0.999).

Conclusion

This study suggested that PCT, BUN, shock, and ART were associated with the prognosis and outcome of AIDS with dTSM and had a specific predictive value.

Pulmonary Infections and Outcomes in AIDS Patients with Respiratory Failure: A 10-Year Retrospective Review

Background

Respiratory failure in acquired immunodeficiency syndrome (AIDS) patients was the leading cause of intensive care unit (ICU) admission in our center. We aimed to describe the pulmonary infections and outcomes for respiratory failure in AIDS patients.

Methods

A retrospective study was conducted on AIDS adult patients with respiratory failure who were admitted to the ICU in Beijing Ditan hospital, China, from January 2012 to December 2021. We investigated pulmonary infections complicated by respiratory failure in AIDS patients. The primary outcome was ICU mortality, and a comparison between survivors and nonsurvivors was performed. Multiple logistic regression analysis was used to identify predictors of ICU mortality. The Kaplan-Meier curve and Log rank test were used for survival analysis.

Results

A total of 231 AIDS patients were admitted to ICU with respiratory failure over a 10-year period with a male predominance (95.7%). Pneumocystis jirovecii pneumonia was the main etiology of pulmonary infections (80.1%). The ICU mortality was 32.9%. In multivariate analysis, ICU mortality was independently associated with invasive mechanical ventilation (IMV) [odds ratio (OR), 27.910; 95% confidence interval (CI, 8.392-92.818; p = 0.000) and the time before ICU admission (OR, 0.959; 95% CI, 0.920-0.999; p = 0.046). In the survival analysis, patients with IMV and later admission to ICU had a higher probability of mortality.

Conclusion

Pneumocystis jirovecii pneumonia was the primary etiology for respiratory failure in AIDS patients admitted to the ICU. Respiratory failure remains a severe illness with high mortality, and ICU mortality was negatively associated with IMV and later admission to ICU.

Predictors of death among TB/HIV co-infected patients on tuberculosis treatment in Sichuan, China: A retrospective cohort study

Mycobacterium tuberculosis is the most common opportunistic infection among patients with human immunodeficiency virus (HIV) infection, and it is also the leading cause of death, causing approximately one-third of acquired immune deficiency syndrome deaths worldwide. China is on the World Health Organization's global list of 30 high-tuberculosis (TB) burden countries. The objective of this study was to evaluate the mortality rate, survival probabilities, and factors associated with death among patients with TB/HIV co-infection undergoing TB treatment in Sichuan, China. A retrospective cohort study was conducted using the Chinese National TB Surveillance System data of TB/HIV co-infected patients enrolled in TB treatment from January 2020 to December 2020. We calculated the mortality rate and survival probabilities using the Kaplan-Meier estimator, and a Cox proportional hazard model was conducted to identify independent risk factors for TB/HIV co-infection mortality. Hazard ratios and their respective 95% confidence intervals were also reported in this study. Of 828 TB/HIV co-infected patients, 44 (5.31%) died during TB treatment, and the crude mortality rate was 7.76 per 1000 person-months. More than half of the deaths (n = 23) occurred in the first 3 months of TB treatment. Overall survival probabilities were 97.20%, 95.16%, and 91.75% at 3rd, 6th, and 12th month respectively. The independent risk factors for mortality among TB/HIV co-infected patients were having extra-pulmonary TB and pulmonary TB co-infection, history of antiretroviral therapy interruption, and baseline cluster of differentiation 4 T-lymphocyte counts <200 cells/μL at the time of HIV diagnosis. Antiretroviral therapy is important for the survival of TB/HIV co-infected patients, and it is recommended to help prolong life by restoring immune function and preventing extra-pulmonary TB.

Spectrum and mortality of opportunistic infections among HIV/AIDS patients in southwestern China

We describe the opportunistic infections (OIs) of HIV/AIDS to understand the spectrum, mortality, and frequency of multiple coinfected OIs among HIV/AIDS patients in southern China, where OIs are severe. We carried out a retrospective cohort study of hospitalized HIV-infected individuals at the Fourth People's Hospital of Nanning, Guangxi, China, from Jan. 2011 to May. 2019. The chi-square test was used to analyze cross-infection; the Kaplan‒Meier analysis was used to compare mortality. A total of 12,612 HIV-infected patients were admitted to this cohort study. Among them, 8982 (71.2%) developed one or more OIs. The overall in-hospital mortality rate was 9.0%. Among the patients, 35.6% coinfected one OI, and 64.4% coinfected more than two OIs simultaneously. Almost half of the patients (60.6%) had CD4 + T-cell counts < 200 cells/μL. Pneumonia (39.8%), tuberculosis (35.3%), and candidiasis (28.8%) were the most common OIs. Coinfected cryptococcal meningitis and dermatitis are the most common combined OIs. The rate of anaemia (17.0%) was highest among those common HIV-associated complications. Multiple OIs are commonly found in hospitalized HIV/AIDS patients in southwestern China, which highlights the need for improved diagnosis and treatment.

Machine learning-based in-hospital mortality prediction of HIV/AIDS patients with Talaromyces marneffei infection in Guangxi, China

Objective

Talaromycosis is a serious regional disease endemic in Southeast Asia. In China, Talaromyces marneffei (T. marneffei) infections is mainly concentrated in the southern region, especially in Guangxi, and cause considerable in-hospital mortality in HIV-infected individuals. Currently, the factors that influence in-hospital death of HIV/AIDS patients with T. marneffei infection are not completely clear. Existing machine learning techniques can be used to develop a predictive model to identify relevant prognostic factors to predict death and appears to be essential to reducing in-hospital mortality.

Methods

We prospectively enrolled HIV/AIDS patients with talaromycosis in the Fourth People’s Hospital of Nanning, Guangxi, from January 2012 to June 2019. Clinical features were selected and used to train four different machine learning models (logistic regression, XGBoost, KNN, and SVM) to predict the treatment outcome of hospitalized patients, and 30% internal validation was used to evaluate the performance of models. Machine learning model performance was assessed according to a range of learning metrics, including area under the receiver operating characteristic curve (AUC). The SHapley Additive exPlanations (SHAP) tool was used to explain the model.

Results

A total of 1927 HIV/AIDS patients with T. marneffei infection were included. The average in-hospital mortality rate was 13.3% (256/1927) from 2012 to 2019. The most common complications/coinfections were pneumonia (68.9%), followed by oral candida (47.5%), and tuberculosis (40.6%). Deceased patients showed higher CD4/CD8 ratios, aspartate aminotransferase (AST) levels, creatinine levels, urea levels, uric acid (UA) levels, lactate dehydrogenase (LDH) levels, total bilirubin levels, creatine kinase levels, white blood-cell counts (WBC) counts, neutrophil counts, procaicltonin levels and C-reactive protein (CRP) levels and lower CD3⁺ T-cell count, CD8⁺ T-cell count, and lymphocyte counts, platelet (PLT), high-density lipoprotein cholesterol (HDL), hemoglobin (Hb) levels than those of surviving patients. The predictive XGBoost model exhibited 0.71 sensitivity, 0.99 specificity, and 0.97 AUC in the training dataset, and our outcome prediction model provided robust discrimination in the testing dataset, showing an AUC of 0.90 with 0.69 sensitivity and 0.96 specificity. The other three models were ruled out due to poor performance. Septic shock and respiratory failure were the most important predictive features, followed by uric acid, urea, platelets, and the AST/ALT ratios.

Conclusion

The XGBoost machine learning model is a good predictor in the hospitalization outcome of HIV/AIDS patients with T. marneffei infection. The model may have potential application in mortality prediction and high-risk factor identification in the talaromycosis population.

Talaromyces marneffei can cause a fatal deeply disseminated fungal infection- talaromycosis. It is widely distributed in Southeast Asia and spreading globally, the disease is insidious and responsible for significant deaths. Clinicians need easy-to-use tools to make decisions on which patients are at a higher risk of dying after infecting T. marneffei. In this study, conducted in Southern China, we have evolved XGBoost machine learning model. 15 clinical indicators and laboratory measures were used to estimate a patient’s risk of dying in the hospital due to the T. marneffei infection. The study showed that the machine learning model has good predictive ability when tested in an internal testing population of patients. We expect that the model could help clinicians assess a patient’s risk of death in just the time of admission to help decide on early treatment timing of high-risk patients who are likely to die.

The burden of serious non-AIDS-defining events among admitted cART-naive AIDS patients in China: An observational cohort study

The objective of this study was to elucidate the burden, risk factors, and prognosis of serious non-AIDS-defining events among admitted cART-naive AIDS patients in China. The evaluation of the burden, risk factors and prognosis of serious NADEs was carried out among 1309 cART-naive AIDS patients (median age: 38.2 years, range: 18–78 years) admitted in Beijing Ditan Hospital between January 2009 and December 2018. Among 1309 patients, 143 patients (10.9%) had at least one serious NADEs, including 49 (3.8%) with cerebrovascular diseases, 37 (2.8%) with non-AIDS-defining cancers, 28 (2.1%) with chronic kidney diseases, 26 (2.0%) with cardiovascular diseases, and 18 (1.4%) with liver cirrhosis. Serious NADEs distributed in different age and CD4 levels, especially with age ≥50 years and CD4 ≤350 cells/ul. Other traditional risk factors, including cigarette smoking (OR = 1.9, 95%CI = 1.3–2.8, p = 0.002), hypertension (OR = 2.5, 95%CI = 1.7–3.7, p<0.001), chronic HCV infection (OR = 2.8, 95%CI = 1.4–5.6, p = 0.004), and hypercholesterolemia (OR = 4.1, 95% CI = 1.2–14.1, p = 0.026), were also associated with serious NADEs. Seventeen cases (1.3%) with serious NADEs died among hospitalized cART-naive AIDS patients, and severe pneumonia (HR = 5.5, 95%CI = 1.9–15.9, p<0.001) and AIDS-defining cancers (HR = 3.8, 95%CI = 1.1–13.2, p = 0.038) were identified as risk factors associated with an increased hazard of mortality among these patients with serious NADEs. Serious NADEs also occurred in cART-naive AIDS patients in China with low prevalence. Our results reminded physicians that early screening of serious NADEs, timely intervention of their risk factors, management of severe AIDS-defining events, multi-disciplinary cooperation, and early initiation of cART were essential to reduce the burden of serious NADEs.

Development and external-validation of a nomogram for predicting the survival of hospitalised HIV/AIDS patients based on a large study cohort in western China

The aim of this study was to develop and externally validate a simple-to-use nomogram for predicting the survival of hospitalised human immunodeficiency virus/acquired immunodeficiency syndrome (HIV/AIDS) patients (hospitalised person living with HIV/AIDS (PLWHAs)). Hospitalised PLWHAs (n = 3724) between January 2012 and December 2014 were enrolled in the training cohort. HIV-infected inpatients (n = 1987) admitted in 2015 were included as the external-validation cohort. The least absolute shrinkage and selection operator method was used to perform data dimension reduction and select the optimal predictors. The nomogram incorporated 11 independent predictors, including occupation, antiretroviral therapy, pneumonia, tuberculosis, Talaromyces marneffei, hypertension, septicemia, anaemia, respiratory failure, hypoproteinemia and electrolyte disturbances. The Likelihood χ² statistic of the model was 516.30 (P = 0.000). Integrated Brier Score was 0.076 and Brier scores of the nomogram at the 10-day and 20-day time points were 0.046 and 0.071, respectively. The area under the curves for receiver operating characteristic were 0.819 and 0.828, and precision-recall curves were 0.242 and 0.378 at two time points. Calibration plots and decision curve analysis in the two sets showed good performance and a high net benefit of nomogram. In conclusion, the nomogram developed in the current study has relatively high calibration and is clinically useful. It provides a convenient and useful tool for timely clinical decision-making and the risk management of hospitalised PLWHAs.

A Clinical scoring model to predict mortality in HIV/TB co-infected patients at end stage of AIDS in China: An observational cohort study

We construct and validate a non-invasive clinical scoring model to predict mortality in HIV/TB patients at end stage of AIDS in China. There were 1,007 HIV/TB patients admitted to Beijing Ditan Hospital from August 2009 to January 2018 included in this study, who were randomly assigned to form derivation cohort and validation cohort. A clinical scoring model was developed based on predictors associated with mortality identified with Cox proportional hazard models. The discrimination and accuracy of model were further validated using the area under the ROC curves. The derivation and validation cohort consisted of 807 and 200 patients in 8:2 ratio, respectively. In derivation cohort, anemia (HGB < 90g/L), tuberculous meningitis, severe pneumonia, hypoalbuminemia, unexplained infections or space-occupying lesions, and malignancies remained independent risk factors of mortality in HIV/TB co-infected patients, and included in this clinical scoring model. The model indicated good discrimination, including AUC = 0.858 (95% CI: 0.782-0.943) in the derivation cohort, and AUC = 0.867 (95% CI: 0.832-0.902) in validation cohort, respectively. The predicted scores were categorized into two groups to predict the mortality: low-risk (0-2 points with mortality with 3.6-9.1%) and high-risk (4-16 points with mortality with 26.42-74.62%), in which 54.55% and 74.62% of patients with score of 5 to 11 and 12-16 were died among high-risk group. Kaplan-Meier curve indicated a significant difference in the cumulative mortality in the two groups by log-rank test (p < 0.001). A clinical scoring model to assess the prognosis in HIV/TB patients at end stage of AIDS was constructed based on simple laboratory and clinical features available at admission, which may be an easy-to-use tool for physicians to evaluate the prognosis and treatment outcome in HIV/TB co-infected patients. The model was also applicable for predicting the death of end-stage HIV/TB patients within a 12 months period after discharge.

A Model to Predict In-Hospital Mortality in HIV/AIDS Patients with Pneumocystis Pneumonia in China: The Clinical Practice in Real World

We aimed to develop and validate a predictive model to evaluate in-hospital mortality risk in HIV/AIDS patients with PCP in China. 1001 HIV/AIDS patients with PCP admitted in the Beijing Ditan hospital from August 2009 to January 2018 were included in this study. Multivariate Cox proportional hazard model was used to identify independent risk factors of death, and a predictive model was devised based on risk factors. The overall in-hospital mortality was 17.3%. The patients were randomly assigned into derivation cohort (801cases) and validation cohort (200 cases) in 8:2 ratio, respectively, in which in derivation cohort we found that 7 predictors, including LDH >350U/L, HR>130 times/min, room air PaO₂ <70mmHg, later admission to ICU, Anemia (HGB≤90g/L), CD4<50cells/ul, and development of a pneumothorax, were associated with poor prognosis in HIV/AIDS patients with PCP and were included in the predictive model. The model had excellent discrimination with AUC of 0.904 and 0.921 in derivation and validation cohort, respectively. The predicted scores were divided into two groups to assess the in-hospital mortality risk: low-risk group (0-11 points with mortality with 2.15-12.77%) and high-risk group (12-21 points with mortality with 38.78%-81.63%). The cumulative mortality rate also indicated significant difference between two groups with Kaplan-Meier curve (p<0.001). A predictive model to evaluate mortality in HIV/AIDS patients with PCP was constructed based on routine laboratory and clinical parameters, which may be a simple tool for physicians to assess the prognosis in HIV/AIDS patients with PCP in China.

Efficacy and tolerability of chemotherapy in Chinese patients with AIDS-related Burkitt lymphoma and diffuse large B-cell lymphoma: An observational study

We evaluated the efficacy and tolerability of chemotherapy in HIV-infected patients with diffuse large B-cell lymphoma (DLBCL) receiving CHOP ± R (n = 17) or Burkitt lymphoma (BL) receiving CODOX-M/IVAC ± R (n = 15). The study was conducted in Beijing Ditan Hospital from January 2009 to August 2015. The following grade 4 adverse effects were observed in BL and DLBCL patients, respectively: neutropenia (80% versus 47.1%), anaemia (46.7% versus 5.9%), thrombocytopenia (53.3% versus 11.8%), bacterial pneumonia (33.3% versus 5.9%), and sepsis (20% versus 5.9%) (p < 0.05). In the BL group, 10 (66.7%) patients died from treatment-related or tumour-related causes, 5 (33.3%) achieved complete response, 1 achieved partial response (6.7%), and 7 developed progressive disease. The 1-year overall survival and progression-free survival rates were 33.3%. Of the DLBCL patients, 3 (17.6%) died from treatment-related causes, 14 (82.4%) achieved complete response, and 3 had progressive disease. The 1-year overall survival and progression-free survival rates were 82.4%. The strongest risk factor for death was relapse between chemotherapy cycles (adjusted hazard ratio = 47.3; 95%CI, 4.2-528.6, p = 0.002). Initiating antiretroviral therapy before chemotherapy failed to improve overall survival. DLBCL patients demonstrated good responses and survival outcomes, while BL patients could not tolerate chemotherapy due to more severe toxicity, and showed poor responses and survival outcomes.

Clinical complications of antiretroviral therapy in HIV/TB patients in referral hospital, China

Aim: The aim was to highlight the clinical complications and to evaluate risk factors of mortality in Chinese HIV/TB patients. Methods: The etiology of clinical deterioration of Chinese HIV/TB patients were evaluated in 180 HIV-infected patients admitted in the Beijing Ditan Hospital between 1 January 2012 and 30 April 2014. Results & conclusion: AIDS-defining illnesses (20.0%) were the most common complication, followed by TB-associated immune reconstitution inflammatory syndrome (16.6%), drug-induced liver injury (11.1%), drug rash (11.1%), non-AIDS-defining illness (5.6%), as well as highly active antiretroviral therapy resistance (3.3%). The risk factors for mortality were tuberculous meningitis associated immune reconstitution inflammatory syndrome (OR: 152.614; CI: 18.324–1263.615; p < 0.001) and non-AIDS-defining illnesses (OR: 114.133; CI: 12.939–1006.752; p < 0.001), which will help remind physicians of the risk of clinical deterioration in HIV/TB patients after antiretroviral therapy in China.

Trends and characteristics of all-cause mortality among HIV-infected inpatients during the HAART era (2006-2015) in Shanghai, China

Globally, the overall mortality rate among HIV-infected patients has significantly declined during the HAART era. Deaths among HIV-infected inpatients need to be characterized in order to formulate intervention strategies to further improve medical care for this population and their prognosis. In the current study, deaths among HIV-infected inpatients from 2006 to 2015 at a medical center for HIV infection and AIDS patient care in Shanghai, China were retrospectively analyzed. Trends in mortality rates and the proportion of deaths caused by AIDS or non-AIDS-related illnesses were evaluated. A bivariate analysis was performed to identify the demographic and clinical factors associated with AIDS or non-AIDS-related deaths among HIV-infected inpatients. Among 6,473 HIV-infected patients who were discharged from 2006 to 2015, 326 deaths (5.04%) were identified. The yearly mortality rate declined significantly over time (χ² = 34.41, p < 0.001). Results revealed that most deaths were attributed to AIDS-related illnesses (76.9 %, 233/303), and the proportion of causes of death did not change significantly over time (χ² = 13.847, p = 0.127). Bivariate analysis identified characteristic factors associated with AIDS-related mortality. Compared to patients who died of non-AIDS illnesses, patients who died of AIDS-related illnesses had a CD4+ T cell count lower than 50 cells/μL (OR 4.587, 2.377-8.850) and fewer liver (OR 0.391, 0.177-0.866) or renal comorbidities (OR 0.188, 0.067-0.523) on admission. Results indicated that the overall in-hospital mortality rate among HIV-infected patients has declined over the past decade. However, AIDS-related illnesses were still the major causes of deaths among HIV-infected inpatients, suggesting that further efforts are needed to improve AIDS care in China.