Unexpectedly long hospital stays as an indicator of risk of unsafe care: an exploratory study

A clinical prediction model based on interpretable machine learning algorithms for prolonged hospital stay in acute ischemic stroke patients: a real-world study

Objective

Acute ischemic stroke (AIS) brings an increasingly heavier economic burden nowadays. Prolonged length of stay (LOS) is a vital factor in healthcare expenditures. The aim of this study was to predict prolonged LOS in AIS patients based on an interpretable machine learning algorithm.

Methods

We enrolled AIS patients in our hospital from August 2017 to July 2019, and divided them into the "prolonged LOS" group and the "no prolonged LOS" group. Prolonged LOS was defined as hospitalization for more than 7 days. The least absolute shrinkage and selection operator (LASSO) regression was applied to reduce the dimensionality of the data. We compared the predictive capacity of extended LOS in eight different machine learning algorithms. SHapley Additive exPlanations (SHAP) values were used to interpret the outcome, and the most optimal model was assessed by discrimination, calibration, and clinical utility.

Results

Prolonged LOS developed in 149 (22.0%) of the 677 eligible patients. In eight machine learning algorithms, prolonged LOS was best predicted by the Gaussian naive Bayes (GNB) model, which had a striking area under the curve (AUC) of 0.878 ± 0.007 in the training set and 0.857 ± 0.039 in the validation set. The variables sorted by the gap values showed that the strongest predictors were pneumonia, dysphagia, thrombectomy, and stroke severity. High net benefits were observed at 0%-76% threshold probabilities, while good agreement was found between the observed and predicted probabilities.

Conclusions

The model using the GNB algorithm proved excellent for predicting prolonged LOS in AIS patients. This simple model of prolonged hospitalization could help adjust policies and better utilize resources.

Determinants of Prolonged Length of Hospital Stay in Patients with Severe Acute Ischemic Stroke

OBJECTIVE

Long hospitalizations are associated with a high comorbidity and considerable hospital cost. Admissions of severe acute ischemic stroke are prone to longer hospitalizations. We aimed to explore the issue and method for improving the length of stay.

METHODS

From the prospective Stroke Registry between January 2019 and June 2020, acute ischemic strokes with an admission National Institutes of Health Stroke Scale ≥ 15 were identified. Prolonged length-of-stay was defined as in-hospital-stay ≥ 30 days. All clinical characteristics were collected, and all do-not-resuscitate documentations were categorized if the order had been written within 7 days of onset.

RESULTS

A total of 212 patients were eligible for severe stroke. Of these, 42 (19.8%) had prolonged length-of-stay and 170 had non-prolonged length-of-stay (median 43 vs. 13 days). The prolonged group was younger, mostly men, and was more likely to be in an independent state and more likely to receive reperfusion therapy, and there was a higher frequency of late do-not-resuscitate orders if signed. Although there was a lower in-hospital mortality rate in the prolonged group (12% vs. 23%), there was a higher proportion with a severe functional state (Modified Rankin Scale = 4-5) among the survivors (97% vs. 87%).

CONCLUSIONS

Severe acute ischemic stroke patients with a prolonged length-of-stay were younger, mostly male, more likely to receive reperfusion therapy, less likely to have an early do-not-resuscitate order if signed, and more likely to have poor functional status at discharge, although there was a lower rate of in-hospital mortality.

Determination of pharmacy students' patient safety approach using the theory of planned behaviour: a mixed-method study

OBJECTIVES

To investigate pharmacy students' attitudes, subjective norms, perceived behavioural control, intentions and their behaviour towards patient safety using a theory of planned behaviour framework.

DESIGN

Mixed-methods research.

SETTING

Private university in Malaysia.

PARTICIPANTS

Pharmacy undergraduate students participated in the study. There were 18 students participated in the qualitative study and 272 students responded to the survey questionnaire.

METHODS

A convergent parallel-mixed method design, involving a quantitative survey and qualitative focus group discussions was used among pharmacy students in a private university in Malaysia. Qualitative data of transcribed verbatim texts were then subjected to a thematic content analysis framework. Multiple correlations were undertaken using the quantitative data to examine how the dependent variable (self-reported knowledge) related to the independent variables (attitudes, behavioural intentions, subjective norms, perceived behavioural control.

PRIMARY OUTCOME

Pharmacy students' attitudes, subjective norms, perceived behavioural control, behavioural intentions constructs led to their behaviour towards patient safety.

SECONDARY OUTCOME

The quantitative study revealed that there was a moderate positive correlation between students' self-reported knowledge and attitudes (r=0.48, p=0.03).

RESULTS

Pharmacy students' attitudes and perceived behavioural control constructs had positive correlations with pharmacy students' self-reported knowledge on patient safety. There was no correlation between students' self-reported knowledge and subjective norms (r=0.27, p=0.23). There was a weak positive correlation between students' self-reported knowledge and perceived behavioural control (r=0.39, p=0.04). There was no correlation between students' self-reported knowledge and behavioural intention (r=0.20, p=0.56).

CONCLUSIONS

Theory of planned behaviour constructs such as attitudes, subjective norms, perceived behavioural control and behavioural intentions of pharmacy students, defined their behaviour towards patient safety. Pharmacy students' attitudes and perceived behavioural control constructs were correlated with their self-reported knowledge on patient safety.

A Discrete Density Approach to Bayesian Quantile and Expectile Regression with Discrete Responses

For decades, regression models beyond the mean for continuous responses have attracted great attention in the literature. These models typically include quantile regression and expectile regression. But there is little research on these regression models for discrete responses, particularly from a Bayesian perspective. By forming the likelihood function based on suitable discrete probability mass functions, this paper introduces a discrete density approach for Bayesian inference of these regression models with discrete responses. Bayesian quantile regression for discrete responses is first developed, and then this method is extended to Bayesian expectile regression for discrete responses. The posterior distribution under this approach is shown not only coherent irrespective of the true distribution of the response, but also proper with regarding to improper priors for the unknown model parameters. The performance of the method is evaluated via extensive Monte Carlo simulation studies and one real data analysis.

Platelet to Lymphocyte Ratio Associated with Prolonged Hospital Length of Stay Postpeptic Ulcer Perforation Repair: An Observational Descriptive Analysis

Background

The predictive role of platelet to lymphocyte ratio (P/LR) in patients with perforated peptic ulcer (PPU) is not well-studied. We aimed to investigate the association between the P/LR ratio and the hospital length of stay (HLOS) for surgically treated PPU.

Method

This is a retrospective observational study for surgically treated adult cases of PPU at Hamad Medical Corporation during the period from January 2012 to August 2017. Patients were categorized into two groups based on their HLOS (<I week vs. >I week). The receiver operating characteristic (ROC) curve was plotted to determine the cutoff value for lymphocyte count, neutrophil to lymphocyte ratio, and P/LR ratio for predicting the prolonged hospitalization.

Results

One hundred and fifty-two patients were included in the study. The majority were young males. The mean age was 38.3 ± 12.7 years. Perforated duodenal ulcer (139 patients) exceeded perforated gastric ulcer (13 patients). The HLOS > 1 week was observed in 14.5% of cases. Older age (p = 0.01), higher preoperative WBC (p = 0.03), lower lymphocyte count (p = 0.01), and higher P/LR ratio (p = 0.005) were evident in the HLOS > 1 week group. The optimal cutoff value of P/LR was 311.2 with AUC 0.702 and negative predictive value of 93% for the prediction of prolonged hospitalization. Two patients died with a mean P/LR ratio of 640.8 ± 135.5 vs. 336.6 ± 258.9 in the survivors.

Conclusion

High preoperative P/LR value predicts prolonged HLOS in patients with repaired perforated peptic ulcer. Further larger multicenter studies are needed to support the study findings.

Observational study to determine the utility of hospital administrative data to support case finding of English patients at higher risk of severe healthcare-related harm

OBJECTIVES

To identify ways of using routine hospital data to improve the efficiency of retrospective reviews of case records for identifying avoidable severe harm DESIGN: Development and testing of thresholds and criteria for two indirect indicators of healthcare-related harm (long length of stay (LOS) and emergency readmission) to determine the yield of specified harms coded in Hospital Episode Statistics (HES).

SETTING

Acute National Health Service hospitals in England.

PARTICIPANTS

HES for acute myocardial infarction (AMI), bowel cancer surgery and hip replacement admissions from 2014 to 2015.

INTERVENTIONS

Case-mix-adjusted linear regression models were used to determine expected LOS. Different thresholds were examined to determine the association with harm. Screening criteria for readmission included time to readmission, length of readmission and diagnoses in initial admission and readmission. The association with harm was examined for each criterion.

RESULTS

The proportions of AMI cases with a harm code increased from 14% among all cases to 47% if a threshold of three times the expected LOS was used. For hip replacement the respective increase was from 10% to 51%. However as the number of patients at these higher thresholds was small, the overall proportion of harm identified is relatively small (15%, 19%, 9% and 8% among AMI, urgent bowel surgery, elective bowel surgery and hip replacement cohorts, respectively). Selection of the time to readmission had an effect on the yield of harms but this varied with condition. At least 50% of surgical patients had a harm code if readmitted within 7 days compared with 21% of patients with AMI.

CONCLUSIONS

Our approach would select a substantial number of patients for case record review. Many of these cases would contain no evidence of healthcare-related harm. In practice, Trusts may choose how many reviews it is feasible to do in advance and then select random samples of cases that satisfy the screening criteria.

Effects of surgery start time on postoperative cortisol, inflammatory cytokines, and postoperative hospital day in hip surgery: Randomized controlled trial

BACKGROUND

The aim of this study was to compare morning surgery (Group A), characterized by high cortisol levels, with afternoon surgery (Group B), characterized by low cortisol levels, with respect to cortisol, inflammatory cytokines (interleukin [IL]-6, IL-8), and postoperative hospital days (POHD) after hip surgery.

METHODS

The study was conducted in a single center, prospective, randomized (1:1) parallel group trial. Patients undergoing total hip replacement or hemiarthroplasty were randomly divided into two groups according to the surgery start time: 8 AM (Group A) or 1-2 PM (Group B). Cortisol and cytokine levels were measured at 7:30 AM on the day of surgery, before induction of anesthesia, and at 6, 12, 24, and 48 hours (h) after surgery. Visual analogue scale (VAS) and POHD were used to evaluate the clinical effect of surgery start time. VAS was measured at 6, 12, 24, and 48 h postoperatively, and POHD was measured at discharge.

RESULTS

In total, 44 patients completed the trial. The postoperative cortisol level was significantly different between the two groups. (24 h, P < .001; 48 h, P < .001). The percentage of patients whose level returned to the initial level was higher in Group B than in Group A (P < .001). Significant differences in IL-6 levels were observed between the two groups at 12, 24, and 48 h after surgery (P = .015; P = .005; P = .002), and in IL-8 levels at 12 and 24 h after surgery (P = .002, P < .001). There was no significant difference between the two groups in VAS and POHD. However, only three patients in Group A were inpatients for more than 3 weeks (P = .233).

CONCLUSIONS

Afternoon surgery allowed for more rapid recovery of cortisol to the baseline level than morning surgery, and IL-6 and IL-8 were lower at 1-2 days postoperatively. The results of this study suggest that afternoon surgery may be considered in patients with postoperative delayed wound healing or inflammation because of the difference in cortisol, IL-6 and 8 in according to surgery start time.

CLINICAL TRIAL REGISTRATION NUMBER

NCT03076827 (ClinicalTRrial.gov).

Evaluating the impact of organisational digital maturity on clinical outcomes in secondary care in England

All healthcare systems are increasingly reliant on health information technology to support the delivery of high-quality, efficient and safe care. Data on its effectiveness are however limited. We therefore sought to examine the impact of organisational digital maturity on clinical outcomes in secondary care within the English National Health Service. We conducted a retrospective analysis of routinely collected administrative data for 13,105,996 admissions across 136 hospitals in England from 2015 to 2016. Data from the 2016 NHS Clinical Digital Maturity Index were used to characterise organisational digital maturity. A multivariable regression model including 12 institutional covariates was utilised to examine the relationship between one measure of organisational digital maturity and five key clinical outcome measures. There was no significant relationship between organisational digital maturity and risk-adjusted 30-day mortality, 28-day readmission rates or complications of care. In multivariable analysis risk-adjusted long length of stay and harm-free care were significantly related to aspects of organisational digital maturity; digitally mature hospitals may not only deliver more harm-free care episodes but also may have a significantly increased risk of patients experiencing a long length of stay. Organisational digital maturity is to some extent related to selected clinical outcomes in secondary care in England. Digital maturity is, however, also strongly linked to other institutional factors that likely play a greater role in influencing clinical outcomes. There is a need to better understand how health IT impacts care delivery and supports other drivers of hospital quality.

Can differences in length of stay between Dutch university hospitals and other hospitals be explained by patient characteristics? A cross-sectional study

OBJECTIVES

The indicator unexpectedly long length of stay (UL-LOS) is used to gain insight into quality and safety of care in hospitals. The calculation of UL-LOS takes patients' age, main diagnosis and main procedure into account. University hospitals have relatively more patients with a UL-LOS than other hospitals. Our main research question is whether the high number of patients with a UL-LOS in university hospitals is caused by differences in additional patient characteristics between university hospitals and other hospitals.

DESIGN

We performed a cross-sectional study and used administrative data from 1 510 627 clinical admissions in 87 Dutch hospitals. Patients who died in hospital, stayed in hospital for 100 days or longer or whose country of residence was not the Netherlands were excluded from the UL-LOS indicator. We identified which patient groups were treated only in university hospitals or only in other hospitals and which were treated in both hospital types. For these last patient groups, we added supplementary patient characteristics to the current model to determine the effect on the UL-LOS model.

RESULTS

Patient groups treated in both hospital types differed in terms of detailed primary diagnosis, socioeconomic status, source of admission, type of admission and amount of Charlson comorbidities. Nevertheless, when adding these characteristics to the current model, university hospitals still have a significantly higher mean UL-LOS score compared with other hospitals (p<0.001).

CONCLUSIONS

The difference in UL-LOS scores between both hospital types remains after adding patient characteristics in which both hospital types differ. We conclude that the high UL-LOS scores in university hospitals are not caused by the investigated additional patient characteristics that differ between university and other hospitals. Patients might stay relatively longer in university hospitals due to differences in work processes because of their education and research tasks or financing differences of both hospital types.

Investigating Adverse Event Free Admissions in Medicare Inpatients as a Patient Safety Indicator

OBJECTIVE

To investigate adverse event free admissions as a potential, patient-centered indicator aligned directly with the goal of patient safety-freedom from harm.

BACKGROUND

Preventable adverse event rates in healthcare could be further reduced. These are generally measured separately, one adverse event at a time. However, this does not reveal whether different patients are affected or the same patients are experiencing multiple events.

METHODS

We examined Medicare inpatient hospital administrative datasets for 2009 to 2011, processed using standard criteria. Events were (i) death within 30 days, (ii) unplanned readmissions within 30 days, (iii) long length of stay, (iv) healthcare acquired infections, and (v) established patient safety indicators not present on admission. We defined adverse event free admissions as those without record of any of these events. National rates were calculated by diagnosis group. Risk-adjusted hospital-specific rates of adverse event free admissions were calculated using colorectal procedures as an example.

RESULTS

There were 23,991,193 admissions after exclusions. Approximately, 64% went through the acute inpatient Medicare system without record of anything untoward. Multiple events were recorded in 22·7% admissions; 15% of these experienced more than 2 adverse events. Risk-adjusted hospital-specific rates of adverse event free admissions for colorectal procedures showed 131 out of 3786 hospitals below the 99·8% lower control limit of the national upper quartile.

CONCLUSIONS

Overall, only 60% of admissions were recorded as adverse event free. Multiple adverse events were common. Even if events are under recorded, this measure could provide an easily understandable and useful baseline for clinicians and managers.

Quantifying the effect of complications on patient flow, costs and surgical throughputs

BACKGROUND

Postoperative adverse events are known to increase length of stay and cost. However, research on how adverse events affect patient flow and operational performance has been relatively limited to date. Moreover, there is paucity of studies on the use of simulation in understanding the effect of complications on care processes and resources. In hospitals with scarcity of resources, postoperative complications can exert a substantial influence on hospital throughputs.

METHODS

This paper describes an evaluation method for assessing the effect of complications on patient flow within a cardiac surgical department. The method is illustrated by a case study where actual patient-level data are incorporated into a discrete event simulation (DES) model. The DES model uses patient data obtained from a large hospital in Oman to quantify the effect of complications on patient flow, costs and surgical throughputs. We evaluated the incremental increase in resources due to treatment of complications using Poisson regression. Several types of complications were examined such as cardiac complications, pulmonary complications, infection complications and neurological complications.

RESULTS

48 % of the patients in our dataset experienced one or more complications. The most common types of complications were ventricular arrhythmia (16 %) followed by new atrial arrhythmia (15.5 %) and prolonged ventilation longer than 24 h (12.5 %). The total number of additional days associated with infections was the highest, while cardiac complications have resulted in the lowest number of incremental days of hospital stay. Complications had a significant effect on perioperative operational performance such as surgery cancellations and waiting time. The effect was profound when complications occurred in the Cardiac Intensive Care (CICU) where a limited capacity was observed.

CONCLUSIONS

The study provides evidence supporting the need to incorporate adverse events data in resource planning to improve hospital performance.

Hospital length of stay after lung transplantation: Independent predictors and association with early and late survival

BACKGROUND

Duration of index hospitalization after lung transplantation (LTx) is an important variable that has not received much attention. We sought to determine independent predictors of prolonged hospital length of stay (LOS) and its association with early and late outcomes.

METHODS

The United Network of Organ Sharing database was queried for adult patients undergoing LTx between 2006 and 2014 (N = 14,320). Patients with dual organ or previous transplantation and patients who died during the first 25 days after LTx were excluded (n = 12,647, mean age 55.2 years ± 13.1). Primary outcome was prolonged LOS (>25 days) (3,251/12,647; 25.7%). Donor, recipient, and procedure-related variables were analyzed as potential predictors of prolonged LOS. Association of prolonged LOS with 1-year and 5-year survival was evaluated using Cox proportional hazards analysis.

RESULTS

Independent predictors of prolonged LOS included serum albumin, lung allocation score, functional status, and need of extracorporeal membrane oxygenation or ventilator support at the time of transplant; donor age >40 years; gender mismatch (female donor to male recipient); donor body mass index; African American ethnicity; ischemic time >6 hours; and double LTx. Prolonged LOS was independently associated with increased mortality at 1 year (hazard ratio, 3.96; 95% confidence interval, 3.48-4.50; p < 0.001) and 5 years (hazard ratio, 2.00; 95% confidence interval, 1.79-2.25; p < 0.001).

CONCLUSIONS

A significant proportion of patients have a prolonged LOS after LTx, and several recipient, donor, and procedure-related variables are independent predictors of this outcome. Patients with prolonged LOS after LTx have significantly increased risk of death at 1 year and 5 years.