Patient risk profiling in acute medicine: the way forward?

Development of a risk prediction nomogram for disposition of acute toxic exposure patients to intensive care unit

Early risk stratification of acutely poisoned patients is essential to identify patients at high risk of intensive care unit (ICU) admission. We aimed to develop a prognostic model and risk-stratification nomogram based on the readily accessible clinical and laboratory predictors on admission for the probability of ICU admission in acutely poisoned patients. This retrospective cohort study included adult patients with acute toxic exposure to a drug or a chemical substance. Patients' demographic, toxicologic, clinical and laboratory data were collected. Among the 1260 eligible patients, 180 (14.3%) were admitted to the ICU. We developed a generalized prognostic model for predicting ICU admission in patients with acute poisoning. The predictors included the Glasgow coma scale, oxygen saturation, diastolic blood pressure, respiratory rate and blood bicarbonate concentration. The model displayed excellent discrimination and calibration (optimistic-adjusted area under the curve = 0.924 and optimistic-adjusted Hosmer and Lemeshow test = 0.922, respectively) when internally validated. Additionally, we developed prognostic models that determine ICU admission in patients with specific poisonings. Furthermore, we constructed risk-stratification nomograms that rank the probability of ICU admission in these patients. The developed risk-stratification nomograms help decision-making regarding ICU admission in acute poisonings. Future external validation in independent cohorts is necessary before clinical application.

Consensus on predictors of clinical deterioration in emergency departments: A Delphi process study

AIM

The study aim was to determine relevance and applicability of generic predictors of clinical deterioration in emergency departments based on consensus among clinicians.

METHODS

Thirty-three predictors of clinical deterioration identified from literature were assessed in a modified two-stage Delphi-process. Sixty-eight clinicians (physicians and nurses) participated in the first round and 48 in the second round; all treating hospitalized patients in Danish emergency departments, some with pre-hospital experience. The panel rated the predictors for relevance (relevant marker of clinical deterioration) and applicability (change in clinical presentation over time, generic in nature and possible to detect bedside). They rated their level of agreement on a 9-point Likert scale and were also invited to propose additional generic predictors between the rounds. New predictors suggested by more than one clinician were included in the second round along with non-consensus predictors from the first round. Final decisions of non-consensus predictors after second round were made by a research group and an impartial physician.

RESULTS

The Delphi-process resulted in 19 clinically relevant and applicable predictors based on vital signs and parameters (respiratory rate, saturation, dyspnoea, systolic blood pressure, pulse rate, abnormal electrocardiogram, altered mental state and temperature), biochemical tests (serum c-reactive protein, serum bicarbonate, serum lactate, serum pH, serum potassium, glucose, leucocyte counts and serum haemoglobin), objective clinical observations (skin conditions) and subjective clinical observations (pain reported as new or escalating, and relatives' concerns).

CONCLUSION

The Delphi-process led to consensus of 19 potential predictors of clinical deterioration widely accepted as relevant and applicable in emergency departments.

Comparative influence of Acute Illness Severity and comorbidity on mortality

BACKGROUND

The extent to which illness severity and comorbidity determine the outcome of an emergency medical admission is uncertain. We aim to quantitate the relative effect of these factors on mortality.

METHODS

We evaluated all emergency medical admission to our institution between 2002 and 2018. We derived an Acute Illness Severity Score (AISS) and Comorbidity Score from admission data and International Classification of Diseases codings. We employed a multivariable logistic regression model to relate both to 30-day in-hospital mortality.

RESULTS

There were 113,807 admissions in 58,126 patients. Both AISS, Odds Ratio (OR) 4.4 (95%CI 3.5, 5.5), and Comorbidity Score, OR 1.91 (95%CI 1.67, 2.18), independently predicted 30-day in-hospital mortality. The two highest AISS risk groups encompassed 46.5% of admissions with predicted mortality of 5.9% (95%CI 5.7%, 6.1%) and 14.4% (95%CI 13.9%, 14.8%) respectively. Comorbidity Score >=10 occurred in 17.9% of admissions with a predicted mortality of 13.3%. AISS and Comorbidity Score interacted to adversely influence mortality; the threshold effect for Comorbidity Score was reduced at high levels of AISS.

CONCLUSION

High AISS and Comorbidity Scores were predictive of 30-day in-hospital mortality and relatively common in emergency medical admissions. There is a strong interaction between the two scores.

Outcomes in acute medicine - Evidence from extended observations on readmissions, hospital length of stay and mortality outcomes

BACKGROUND

The Acute Medical Admission Unit (AMAU) model of care has been widely deployed, we examine changes in hospital readmission rates, length of stay (LOS) and 30-day in-hospital mortality over 16 years.

METHODS

All emergency medical admissions between 2002 and 2017 were examined. We assessed 30-day in-hospital mortality, readmission rates, and LOS using logistic regression and margins statistics modelled outcomes against predictor variables.

RESULTS

There were 106,586 admissions in 54,928 patients over 16 years. Calculated per patient the 30-day in-hospital mortality was 8.9% (95%CI 8.6% to 9.2%) and showed a relative risk reduction (RRR) of 61.1% from 12.4% to 4.8% over the 16 years (p = .001). Calculated per admission the 30-day in-hospital mortality was 4.5% (95%CI 4.4% to 4.6%) with a RRR of 31.9% from 2002 to 2017. Over this extended period 48.7% of patients were readmitted at least once, 9.3% >5 times and 20 patients >50 times each. The median LOS was 5.9 days (IQR 2.4, 12.9) with no trend of change over time. Total readmissions increased as a time dependent function; early readmissions (<4 weeks) fluctuated without time trend at 10.5% (95%CI 9.6 to 11.3). A logistic regression model described the hospital LOS as a linear function both of comorbidity and the utilisation of inpatient procedures and services.

CONCLUSION

30-day in-hospital mortality showed a linear trend to reduce over time at unaltered LOS and readmission rates. LOS showed linear dependency on clinical complexity; interventions aimed at reducing LOS may not be appropriate beyond a certain point.

Predicting Outcomes in Emergency Medical Admissions Using a Laboratory Only Nomogram

Background

We describe a nomogram to explain an Acute Illness Severity model, derived from emergency room triage and admission laboratory data, to predict 30-day in-hospital survival following an emergency medical admission.

Methods

For emergency medical admissions (96,305 episodes in 50,612 patients) between 2002 and 2016, the relationship between 30-day in-hospital mortality and admission laboratory data was determined using logistic regression. The previously validated Acute Illness Severity model was then transposed to a Kattan-style nomogram with a Stata user-written program.

Results

The Acute Illness Severity was based on the admission Manchester triage category and biochemical laboratory score; these latter were based on the serum albumin, sodium, potassium, urea, red cell distribution width, and troponin status. The laboratory admission data was predictive with an AUROC of 0.85 (95% CI: 0.85, 0.86). The sensitivity was 94.4%, with a specificity of 62.7%. The positive predictive value was 21.2%, with a negative predictive value of 99.1%. For the Kattan-style nomogram, the regression coefficients are converted to a 100-point scale with the predictor parameters mapped to a probability axis. The nomogram would be an easy-to-use tool at the bedside and for educational purposes, illustrating the relative importance of the contribution of each predictor to the overall score.

Conclusion

A nomogram to illustrate and explain the prognostic factors underlying an Acute Illness Severity Score system is described.

Survival analysis of weekend emergency medical admissions

BACKGROUND

We previously reported weekend emergency admissions to have a higher mortality; we have now examined the time profile of deaths, by weekday or weekend admission, in all emergency medical patients admitted between 2002 and 2014.

METHODS

We divided admissions by a weekday or weekend (After 17.00 Friday-Sunday) hospital arrival. We examined survival following an admission using Cox proportional hazard models and Kaplan-Meier time to event analysis.

RESULTS

In total 82 368 admissions were recorded in 44, 628 patients. Weekend admissions had an increased mortality of 5.0% (95% CI 4.7, 5.4) compared with weekday admissions of 4.5% (95% CI 4.3, 4.7) ( P = 0.007). The univariate adjusted Odds Ratio (OR) of death for a weekend admission was significantly increased OR = 1.15 (95% CI 1.05, 1.24) ( P = 0.001). Mortality following an admission declined exponentially over time with a long tail, ∼25% of deaths occurred after day 28. Only 11.4% of deaths occurred on the weekend of the admission. Survival curves showed no mortality difference at 28 days ( P = 0.21) but a difference at 90 days ( P = 0.05). The higher mortality for a weekend admission was attributable to late deaths in the cohort with an extended stay; compared with weekday, these weekend admissions were more likely to be older and have greater co-morbidity.

CONCLUSION

Survival rates following a weekend or weekday admission were similar out to 28 days. The higher overall mortality for weekend admissions is due to divergence in survival between 28 and 90 days. Most deaths in weekend admissions occurred when the hospital was fully staffed.

Time patterns in mortality after an emergency medical admission; relationship to weekday or weekend admission

BACKGROUND

The aim of this study was to detail the time profile and frequency distribution of mortality following an emergency admission and to compare these for weekday and weekend admissions.

METHODS

We profiled in-hospital deaths following emergency medical admission between 2002 and 2014. We determined the frequency distribution, time pattern, causality and influence of day of admission on mortality out to 120days. We utilized a multivariable regression model (logistic for in-hospital mortality and truncated Poisson for count data) to adjust for major predictor variables.

RESULTS

There were 82,368 admissions in 44,628 patients with 4587 in-hospital deaths. The 30-day in-hospital mortality declined from 8.2% in 2002 to 3.7% in 2014. The mortality pattern showed an exponential decay over time; the time to death was best described by the three-parameter Weibull model. The calculated time to death for the 5th, 10th, 25th, 50th, 75th, and 90th centiles were 0.5, 1.2, 3.8, 11.1, 26.3 and 49.3days. Acute Illness Severity Score, Chronic Disabling Disease Score, Charlson Co-Morbidity Index and Sepsis status were associated with mortality. The risk of death was initially high, lower by day 3, and showed a cumulative increase over time. The mortality pattern was very similar between a weekday or weekend admission; however, the risk of death was greater at all time points between 0 and 120days for patients admitted at a weekend OR 1.08 (95% CI 1.01-1.15).

CONCLUSION

We have demonstrated the pattern of mortality following an emergency admission. The underlying pattern is similar between weekday and weekend admissions.