Development and validation of an interpretable Markov-embedded multilabel model for predicting risks of multiple postoperative complications among surgical inpatients: a multicenter prospective cohort study.

BACKGROUND: When they encounter various highly related postoperative complications, existing risk evaluation tools that focus on single or any complications are inadequate in clinical practice. This seriously hinders complication management because of the lack of a quantitative basis. An interpretable multilabel model framework that predicts multiple complications simultaneously is urgently needed. MATERIALS AND METHODS: The authors included 50 325 inpatients from a large multicenter cohort (2014-2017). The authors separated patients from one hospital for external validation and randomly split the remaining patients into training and internal validation sets. A MARKov-EmbeDded (MARKED) multilabel model was proposed, and three models were trained for comparison: binary relevance, a fully connected network (FULLNET), and a deep neural network. Performance was mainly evaluated using the area under the receiver operating characteristic curve (AUC). The authors interpreted the model using Shapley Additive Explanations. Complication-specific risk and risk source inference were provided at the individual level. RESULTS: There were 26 292, 6574, and 17 459 inpatients in the training, internal validation, and external validation sets, respectively. For the external validation set, MARKED achieved the highest average AUC (0.818, 95% CI: 0.771-0.864) across eight outcomes [compared with binary relevance, 0.799 (0.748-0.849), FULLNET, 0.806 (0.756-0.856), and deep neural network, 0.815 (0.765-0.866)]. Specifically, the AUCs of MARKED were above 0.9 for cardiac complications [0.927 (0.894-0.960)], neurological complications [0.905 (0.870-0.941)], and mortality [0.902 (0.867-0.937)]. Serum albumin, surgical specialties, emergency case, American Society of Anesthesiologists score, age, and sex were the six most important preoperative variables. The interaction between complications contributed more than the preoperative variables, and formed a hierarchical chain of risk factors, mild complications, and severe complications. CONCLUSION: The authors demonstrated the advantage of MARKED in terms of performance and interpretability. The authors expect that the identification of high-risk patients and the inference of the risk source for specific complications will be valuable for clinical decision-making.

Prediction of complications associated with general surgery using a Bayesian network.

BACKGROUND: Numerous attempts have been made to identify risk factors for surgery complications, but few studies have identified accurate methods of predicting complex outcomes involving multiple complications. METHODS: We performed a prospective cohort study of general surgical inpatients who attended 4 regionally representative hospitals in China from January to June 2015 and January to June 2016. The risk factors were identified using logistic regression. A Bayesian network model, consisting of directed arcs and nodes, was used to analyze the relationships between risk factors and complications. Probability ratios for complications for a given node state relative to the baseline probability were calculated to quantify the potential effects of risk factors on complications or of complications on other complications. RESULTS: We recruited 19,223 participants and identified 21 nodes, representing 9 risk factors and 12 complications, and 55 direct relationships between these. Respiratory failure was at the center of the network, directly affected by 5 risk factors, and directly affected 7 complications. Cardiopulmonary resuscitation and sepsis or septic shock also directly affected death. The area under the receiver operating characteristic curve for the ability of the network to predict complications was >0.7. Notably, the probability of other severe complications or death significantly increased when a severe complication occurred. Most importantly, there was a 141-fold higher risk of death when cardiopulmonary resuscitation was required. CONCLUSION: We have created a Bayesian network that displays how risk factors affect complications and their interrelationships and permits the accurate prediction of complications and the creation of appropriate preventive guidelines.

Generating real-world evidence compatible with evidence from randomized controlled trials: a novel observational study design applicable to surgical transfusion research.

BACKGROUND: Numerous observational studies have revealed an increased risk of death and complications with transfusion, but this observation has not been confirmed in randomized controlled trials (RCTs). The "transfusion kills patients" paradox persists in real-world observational studies despite application of analytic methods such as propensity-score matching. We propose a new design to address this long-term existing issue, which if left unresolved, will be deleterious to the healthy generation of evidence that supports optimized transfusion practice. METHODS: In the new design, we stress three aspects for reconciling observational studies and RCTs on transfusion safety: (1) re-definition of the study population according to a stable hemoglobin range (gray zone of transfusion decision; 7.5-9.5 g/dL in this study); (2) selection of comparison groups according to a trigger value (last hemoglobin measurement before transfusion; nadir during hospital stay for control); (3) dealing with patient heterogeneity according to standardized mean difference (SMD) values. We applied the new design to hospitalized older patients (aged ≥60 years) undergoing general surgery at four academic/teaching hospitals. Four datasets were analyzed: a base population before (Base Match-) and after (Base Match+) propensity-score matching to simulate previous observational studies; a study population before (Study Match-) and after (Study Match+) propensity-score matching to demonstrate effects of our design. RESULTS: Of 6141 older patients, 662 (10.78%) were transfused and showed high heterogeneity compared with those not receiving transfusion, particularly regarding preoperative hemoglobin (mean: 11.0 vs. 13.5 g/dL) and intraoperative bleeding (≥500 mL: 37.9% vs. 2.1%). Patient heterogeneity was reduced with the new design; SMD of the two variables was reduced from approximately 100% (Base Match-) to 0% (Study Match+). Transfusion was related to a higher risk of death and complications in Base Match- (odds ratio [OR], 95% confidence interval [CI]: 2.68, 1.86-3.86) and Base Match+ (2.24, 1.43-3.49), but not in Study Match- (0.77, 0.32-1.86) or Study Match+ (0.66, 0.23-1.89). CONCLUSIONS: We show how choice of study population and analysis could affect real-world study findings. Our results following the new design are in accordance with relevant RCTs, highlighting its value in accelerating the pace of transfusion evidence generation and generalization.

Protein intake and risk of inflammatory bowel disease: A meta-analysis.

BACKGROUND AND OBJECTIVES: Although the association between dietary protein intake and inflammatory bowel disease (IBD) risk has been investigated, the results are inconsistent. Therefore, we conducted a meta-analysis to reassess the relationship between dietary protein intake and IBD risk. METHODS AND STUDY DESIGN: The PubMed, Web of Knowledge, and Wanfang databases were searched for pertinent studies through January 31, 2020. Relative risks (RRs) with 95% confidence intervals (CIs) were derived using a random-effect model. Subgroup analyses according to disease type, geographic location, and sex; sensitivity analysis; and publication bias analysis were performed. RESULTS: The current report includes 8 articles consisting of 12 studies with 1069 cases and 330,676 participants. The pooled RR (95% CI) of the highest vs. the lowest categories of dietary protein intake for the IBD risk was 1.561 (0.384-6.347) in cohort studies and 1.060 (0.663-1.694) in case-control studies. Evidence of heterogeneity was found both in cohort studies (I2=86.4%, p=0.007) and in case-control studies (I2=49.0%, p=0.039). However, the association was significant among Asian populations (RR=1.675, 95% CI=1.096-2.559) but not in other populations. We did not find any relationship of dietary protein intake with the risk of either Crohn's disease or ulcerative colitis. CONCLUSIONS: Based on limited information, the highest dietary protein intakes among Asians may increase the risk of IBD, undifferentiated for ulcerative colitis or Crohn's disease. This may reflect dietary patterns for which protein is a marker rather than implicate protein itself.

Nutritional Status According to the Short-Form Mini Nutritional Assessment (MNA-SF) and Clinical Characteristics as Predictors of Length of Stay, Mortality, and Readmissions Among Older Inpatients in China: A National Study.

BACKGROUND: Studies are scarce in China that explore the association of nutritional status, measured using the Short-Form Mini Nutritional Assessment (MNA-SF) and biochemical data, on adverse clinical outcomes among older inpatients. In this study, we aimed to determine the prevalence of malnutrition in tertiary hospitals of China and the associations between malnutrition and adverse clinical outcomes. METHODS: This prospective study involved 5,516 older inpatients (mean age 72.47 ± 5.77 years) hospitalized in tertiary hospitals between October 2018 and February 2019. The tertiary hospitals refer to the hospital with more than 500 beds and can provide complex medical care services. The MNA-SF was used to assess nutritional status. Multiple logistic regression and negative binomial regression were used to analyze the relationship between nutritional parameters and risk of hospital length of stay (LoS), mortality, and rehospitalization. RESULTS: We found that 46.19% of hospitalized patients had malnutrition or malnutrition risk, according to the MNA-SF. Death occurred in 3.45% of patients. MNA-SF scores 0-7 (odds ratio [OR] 5.738, 95% confidence interval [CI] 3.473 to 9.48) were associated with a six-fold higher likelihood of death, and scores 8-11 (OR 3.283, 95% CI 2.126-5.069) with a three-fold higher likelihood of death, compared with MNA-SF scores 12-14 in the logistic regression model, after adjusting for potential confounders. A low MNA-SF score of 0-7 (regression coefficient 0.2807, 95% CI 0.0294-0.5320; P < 0.05) and a score of 8-11 (0.2574, 95% CI 0.0863-0.4285; P < 0.01) was associated with a significantly higher (28.07 and 25.74%, respectively) likelihood of increased LoS, compared with MNA-SF score 12-14. MNA-SF scores 0-7 (OR 1.393, 95% CI 1.052-1.843) and 8-11 (OR 1.356, 95% CI 1.124-1.636) were associated with a nearly 1.5-fold higher likelihood of 90-day readmission compared with MNA-SF scores 12-14 in the logistic regression model. Moreover, hemoglobin level, female sex, education level, former smoking, BMI 24-27.9 kg/m2, age 75 years and above, and current alcohol consumption were the main factors influencing clinical outcomes in this population. CONCLUSIONS: Malnutrition increases the risk of hospital LoS, mortality, and 90-day readmission. The use of nutritional assessment tools in all hospitalized patients in China is needed. The MNA-SF combined with hemoglobin level may be used to identify older inpatients with a high risk of adverse clinical outcomes. These findings may have important implications for the planning of hospital services.

Association between Frailty and 90-Day Outcomes amongst the Chinese Population: A Hospital-Based Multicentre Cohort Study.

INTRODUCTION: Frailty has gained increasing attention as it is by far the most prevalent geriatric condition amongst older patients which heavily impacts chronic health status. However, the relationship between frailty and adverse health outcomes in China is far from clear. This study explored the relation between frailty and a panel of adverse health outcomes. METHODS: We performed a multicentre cohort study of older inpatients at 6 large hospitals in China, with two-stage cluster sampling, from October 2018 to April 2019. Frailty was measured according to the FRAIL scale and categorized into robust, pre-frail, and frail. A multivariable logistic regression model and multilevel multivariable negative binomial regression model were used to analyse the relationship between frailty and adverse outcomes. Outcomes were length of hospitalization, as well as falls, readmission, and mortality at 30 and 90 days after enrolment. All regression models were adjusted for age, sex, BMI, surgery, and hospital ward. RESULTS: We included 9,996 inpatients (median age 72 years and 57.8% male). The overall mortality at 30 and 90 days was 1.23 and 1.88%, respectively. At 30 days, frailty was an independent predictor of falls (odds ratio [OR] 3.19; 95% CI 1.59-6.38), readmission (OR 1.45; 95% CI 1.25-1.67), and mortality (OR 3.54; 95% confidence interval [CI] 2.10-5.96), adjusted for age, sex, BMI, surgery, and hospital ward clustering effect. At 90 days, frailty had a strong predictive effect on falls (OR 2.10; 95% CI 1.09-4.01), readmission (OR 1.38; 95% CI 1.21-1.57), and mortality (OR 6.50; 95% CI 4.00-7.97), adjusted for age, sex, BMI, surgery, and hospital ward clustering effect. There seemed to be a dose-response association between frailty categories and fall or mortality, except for readmission. CONCLUSIONS: Frailty is closely related to falls, readmission, and mortality at 30 or 90 days. Early identification and intervention for frailty amongst older inpatients should be conducted to prevent adverse outcomes.

Prevalence of cognitive impairment in Chinese older inpatients and its relationship with 1-year adverse health outcomes: a multi-center cohort study.

BACKGROUND: Previous studies on the relationship between cognitive impairment and adverse outcomes among geriatric inpatients are not representative of older inpatients in China because of insufficient sample sizes or single-center study designs. The purpose of our study was to examine the prevalence of cognitive impairment and the relationship between cognitive impairment and 1-year adverse health outcomes in older inpatients. METHODS: This study was a large-scale multi-center cohort study conducted from October 2018 to February 2020. Six tertiary hospitals across China were selected using a two-stage cluster sampling method, and eligible older inpatients were selected for the baseline survey and follow-up. The Mini Cognitive Scale and the FRAIL scale were used to screen for cognitive impairment and frailty, respectively. The EuroQol-5 Dimension-5 Level questionnaire was used to assess health-related quality of life (HRQoL). We used a generalized estimating model to evaluate the relationship between cognitive impairment and adverse outcomes. RESULTS: The study included 5008 men (58.02%) and 3623 women (41.98%), and 70.64% were aged 65-75 years, and 26.27% were aged 75-85 years. Cognitive impairment was observed in 1756 patients (20.35%). There were significant differences between participants with cognitive impairment and those with normal cognitive function for age, gender, surgery status, frailty, depression, handgrip strength and so on. After adjusting for multiple covariates, compared with patients with normal cognitive function, the odds ratio for 1-year mortality was 1.216 (95% confidence interval [CI]: 1.076-1.375) and for 1-year incidence of frailty was 1.195 (95% CI: 1.037-1.376) in patients with cognitive impairment. Similarly, the regression coefficient of 1-year HRQoL was - 0.013 (95% CI: - 0.024-- 0.002). In the stratified analysis, risk of adverse outcome within 1 year was higher in older patients with cognitive impairment aged over 75 years than those aged 65-74 years. CONCLUSIONS: We revealed that cognitive impairment was highly correlated with occurrence of 1-year adverse health outcomes (death, frailty, and decreased HRQoL) in older inpatients, which provides a basis for formulating effective intervention measures. TRIAL REGISTRATION: Chinese Clinical Trial Registry, ChiCTR1800017682 , registered 09 August 2018.

Associations Between Nutritional Status, Sociodemographic Characteristics, and Health-Related Variables and Health-Related Quality of Life Among Chinese Elderly Patients: A Multicenter Prospective Study.

Background: Studies that explore the nutritional status, sociodemographic factors, mental health variables, and physical health variables that affect the health-related quality of life (HRQoL) of elderly patients are scarce in China. Objective: This study aimed to examine the association between health-related quality of life (HRQoL) and nutritional status, sociodemographic characteristics, and health-related variables among Chinese elderly patients. Materials and Methods: Participants were recruited from six tertiary-level hospitals in six provinces or municipalities/cities throughout China from October 2018 to February 2019: a total of 9,996 participants aged 65 years and older were enrolled. The nutritional status and HRQoL were measured using the Mini Nutritional Assessment-Short Form (MNA-SF) and the EuroQoL Five-Dimension Visual Analog Scale, respectively. BMI was taken using standard measurement protocols. Sociodemographic characteristics included age, sex, education, marital status, ethnicity, smoking, alcohol drinking, and current residence. Mental and physical health variables such as frailty and depression were assessed using validated tested instruments. Multiple linear regression analysis was used to analyze whether the nutritional status, sociodemographic characteristics, and health-related variables were associated with HRQoL. Results: According to the MNA-SF scores at the 30- and 90-day follow-up, 9.7% and 9.1% of participants were malnourished, respectively. Higher MNA-SF scores were related to higher HRQoL scores in older patients (regression coefficient; 95% confidence interval) both at the 30-day (0.660; 0.499-0.821) and 90-day (0.622; 0.434-0.809) follow-up. However, there were no significant associations between the body mass index values and HRQoL. Sociodemographic characteristics (such as age, smoking, and current residence), physical health variables (frailty, urinary function, defecation function, sleeping condition, and falling accidents in the past 12 months), and mental health variables (depression) were the main factors influencing HRQoL in this group. Conclusion: There are several factors associated with HRQoL among the population derived from this investigation of a representative sample of the Chinese hospitalized elderly population in tertiary hospitals. These findings could have major importance for the planning of "active aging" policies and programs. Trial Registration: Chinese Clinical Trial Registry, ChiCTR1800017682, registered August 9, 2018.

Potential associated factors of functional disability in Chinese older inpatients: a multicenter cross-sectional study.

BACKGROUND: There is still controversy about the relationship between aging and changes in functional ability. This study aims to describe the level of basic activities of daily living (ADL) and higher-level instrumental activities of daily living (IADL) in different age groups and explore the factors associated with functional disability in Chinese older inpatients. METHODS: This cross-sectional study surveyed 9996 older inpatients aged 65 years and older from six tertiary hospitals in China from October 2018 to February 2019. The levels of ADL and IADL were measured by scores of the Barthel index and Instrumental Activities of Daily Living Scale. A mixed-effect generalized linear model was used to examine the association between functional disability and covariates. RESULTS: The average ADL score was 89.51 ± 19.29 and the mean IADL score 6.76 ± 2.01 for all participants. There was a trend of decreasing scores along with aging, and significant differences between age groups were also observed (P < 0.001). The most affected ADL and IADL was stair climbing and shopping, respectively. Sociodemographic characteristics (such as age), physical health variables (frailty, emaciation, hearing dysfunction, urinary dysfunction, defecation dysfunction, falling accidents in the past 12 months), and mental health variables (cognitive dysfunction, depression) were associated with functional disability. Patients from the emergency department or transferred from other hospitals and former alcohol drinkers are at risk of ADL disability. Former smoking is a risk factor for IADL disability, whereas current drinking, higher-level education, and residing in a building without elevators were likely to maintain a better IADL performance. CONCLUSIONS: Functional ability declines with aging, older inpatients are low dependency upon ADL and IADL. There are several associated factors among the participants derived from this investigation of a large-scale, multicenter, nationally representative Chinese older inpatient population. These findings potentially have major importance for the planning of hospital services, discharge planning, and post-discharge care. TRIAL REGISTRATION: Chinese Clinical Trial Registry, ChiCTR1800017682 , registered August 9, 2018.