Emergency hospital admissions for the elderly: insights from the Devon Predictive Model

Machine learning-based prediction model for emergency department visits using prescription information in community-dwelling non-cancer older adults

Older adults are more likely to require emergency department (ED) visits than others, which might be attributed to their medication use. Being able to predict the likelihood of an ED visit using prescription information and readily available data would be useful for primary care. This study aimed to predict the likelihood of ED visits using extensive medication variables generated according to explicit clinical criteria for elderly people and high-risk medication categories by applying machine learning (ML) methods. Patients aged ≥ 65 years were included, and ED visits were predicted with 146 variables, including demographic and comprehensive medication-related factors, using nationwide claims data. Among the eight ML models, the final model was developed using LightGBM, which showed the best performance. The final model incorporated 93 predictors, including six sociodemographic, 28 comorbidity, and 59 medication-related variables. The final model had an area under the receiver operating characteristic curve of 0.689 in the validation cohort. Approximately half of the top 20 strong predictors were medication-related variables. Here, an ED visit risk prediction model for older people was developed and validated using administrative data that can be easily applied in clinical settings to screen patients who are likely to visit an ED.

Risk of adverse events following the initiation of antihypertensives in older people with complex health needs: a self-controlled case series in the United Kingdom

BACKGROUND

We assessed the risk of adverse events-severe acute kidney injury (AKI), falls and fractures-associated with use of antihypertensives in older patients with complex health needs (CHN).

SETTING

UK primary care linked to inpatient and mortality records.

METHODS

The source population comprised patients aged >65, with ≥1 year of registration and unexposed to antihypertensives in the year before study start. We identified three cohorts of patients with CHN, namely, unplanned hospitalisations, frailty (electronic frailty index deficit count ≥3) and polypharmacy (prescription of ≥10 medicines). Patients in any of these cohorts were included in the CHN cohort. We conducted self-controlled case series for each cohort and outcome (AKI, falls, fractures). Incidence rate ratios (IRRs) were estimated by dividing event rates (i) during overall antihypertensive exposed patient-time over unexposed patient-time; and (ii) in the first 30 days after treatment initiation over unexposed patient-time.

RESULTS

Among 42,483 patients in the CHN cohort, 7,240, 5,164 and 450 individuals had falls, fractures or AKI, respectively. We observed an increased risk for AKI associated with exposure to antihypertensives across all cohorts (CHN: IRR 2.36 [95% CI: 1.68-3.31]). In the 30 days post-antihypertensive treatment initiation, a 35-50% increased risk for falls was found across all cohorts and increased fracture risk in the frailty cohort (IRR 1.38 [1.03-1.84]). No increased risk for falls/fractures was associated with continuation of antihypertensive treatment or overall use.

CONCLUSION

Treatment with antihypertensives in older patients was associated with increased risk of AKI and transiently elevated risk of falls in the 30 days after starting antihypertensive therapy.

Impact of Geriatric Admissions on Workload in the Emergency Department

Background: Due to the increase in life expectancy, both the general population and the population of patients of emergency departments (ED) are getting older. An understanding of differences, workload and resource requirements may be helpful in improving patient care. The main goal of this study was to evaluate the reasons for geriatric admissions in the ED, identify typical medical problems and assess the number of resources in order to provide more effective management. Methods: We examined 35,720 elderly patients' ED visits over the course of 3 years. The data collected included age, sex, timing and length of stay (LOS), use of various resources, endpoint (admission, discharge or death) and ICD-10 diagnoses. Results: The median age was 73 years [66-81], with more females (54.86%). There were 57.66% elderly (G1), 36.44% senile (G2) and 5.89% long-liver (G3) patients. There were more females in the older groups. The total admission rate was 37.89% (34.19% for G1, 42.21% for G2 and 47.33% for G3). The average length of the patient's stay was 150 min [81-245] (G3 180 min [108-277], G2 (162 min [92-261]) and G1 139 min [71-230]). Heart failure, atrial fibrillation and hip fracture were the most common diagnoses. Nonspecific diagnoses were common in all groups. Conclusion: The vast majority of geriatric patients required considerable resources. With increasing ages, the number of women, LOS and number of admissions increased.

Developing and validating a predictive model for future emergency hospital admissions

Although many emergency hospital admissions may be unavoidable, a proportion of these admissions represent a failure of the care system. The adverse consequences of avoidable emergency hospital admissions affect patients, carers, care systems and substantially increase care costs. The aim of this study was to develop and validate a risk prediction model to estimate the individual probability of emergency admission in the next 12 months within a regional population. We deterministically linked routinely collected data from secondary care with population level data, resulting in a comprehensive research dataset of 190,466 individuals. The resulting risk prediction tool is based on a logistic regression model with five independent variables. The model indicated a discrimination of area under the receiver operating characteristic curve of 0.9384 (95% CI 0.9325-0.9443). We also experimented with different probability cut-off points for identifying high risk patients and found the model's overall prediction accuracy to be over 95% throughout. In summary, the internally validated model we developed can predict with high accuracy the individual risk of emergency admission to hospital within the next year. Its relative simplicity makes it easily implementable within a decision support tool to assist with the management of individual patients in the community.

Trends in emergency department use by older people during the COVID-19 pandemic

Aim

To examine changing trends in presentation of older people to the emergency department during the COVID-19 pandemic compared to 2018/2019.

Findings

On average 4 fewer people aged ≥70 years presented to the ED in the first 6 months of the COVID-19 pandemic (March-August 2020). There was a 20% reduction in presentations of stroke and cardiac complaints but a 25% increase in falls/injuries following easing of lockdown restrictions.

Message

It is imperative that we consider enabling strategies to ensure older people access unscheduled care in a timely manner when necessary.

Purpose

Reports suggest that many older people deferred seeking healthcare during the COVID-19 pandemic due to fear of contracting COVID-19. The aim of this study was to examine trends of emergency department (ED) use by older people during the first wave of the COVID-19 pandemic compared to previous years.

Methods

The study site is a 1000-bed university teaching hospital with annual ED new-patient attendance of > 50,000. All ED presentations of patients aged ≥ 70 years from March to August 2020, 2019 and 2018 inclusive (n = 13,989) were reviewed and compared for presenting complaint, Manchester Triage Score, and admission/discharge decision.

Results

There was a 16% reduction in presentations across the 6 months in 2020 compared to the average of 2018/2019. On average, 4 fewer people aged ≥ 70 years presented to the ED per day in 2020. Much of this was concentrated in March (33% fewer presentations) and April (31% fewer presentations), when the country was in ‘lockdown’, i.e. non-essential journeys were banned. There was a 20% reduction in patients presenting with stroke and cardiac complaints. In the 3 months following easing of restrictions, there was a 25% increase in falls and orthopaedic injuries when compared to 2018/2019.

Conclusion

This study demonstrates a significant decline in the number of older people presenting to the ED for unscheduled care, including for potentially time-dependent illnesses such as stroke or cardiac complaints. Given the possibility of further lockdowns, it is imperative that we consider enabling strategies to ensure older people access unscheduled care in a timely manner when necessary.

Gray Transformation of Internal Medicine

Aim To study the profile of patients hospitalized in an internal medicine facility of a public hospital and explore if there is a shift in the age distribution of hospitalized patients toward an older age group. Methods The study was conducted at a tertiary care hospital in Delhi where the department of medicine has six units and a particular unit is in charge of the emergency room one day a week and every sixth Sunday. A total of 716 patients hospitalized in the medicine ward through the emergency services each Wednesday and every sixth Sunday (i.e, admission days of a particular unit) during a period for six months (November 2017 to April 2018) were retrospectively identified and their name, age, sex, comorbidities, final diagnosis, duration of hospital stay, and the outcome were noted. This was compared with similar data collected from a previous study conducted in the same setting in the years 2010-2011. The data were analyzed using Stata version 13 software (StataCorp, College Station, Texas). Findings were compared using the chi-squared test or t-test, wherever applicable. Results The mean age of patients hospitalized in 2010-2011 was 43.9 years, and this had increased to 48 years in 2018-2019 suggesting that the average patient being hospitalized in general medicine wards of a public hospital was now approximately five years older. Thirty-nine percent (39%) of the individuals admitted were more than 60 years of age in 2018-2019 as compared to 28.0% in 2010-2011. There was a significant rise in the cases of cardiovascular and gastrointestinal diseases and a decline in poisoning and infectious diseases from the year 2010-2011 to 2018-2019. A significantly higher prevalence of cardiovascular diseases, diabetes, and respiratory diseases was found in the current study among older adults. Conclusion There has been a significant shift in the distribution of hospitalized patients in the internal medicine in-patient wards toward the older age group during the last decade. Also, there is a significant difference between the disease profiles of older and young patients. A comprehensive approach to geriatric medicine should be introduced and followed.

Acute health care services use among nursing home residents in Germany: a comparative analysis of out-of-hours medical care, emergency department visits and acute hospital admissions

BACKGROUND

Nursing home (NH) residents often utilise acute health care services. However, comparative data on those are lacking.

AIMS

Investigating German NH residents' use of out-of-hours medical care (OOHC), visits to emergency departments (EDs) and acute hospital admissions (AHAs).

METHODS

Using claims data of 1665 residents for 2014-2015, we conducted a retrospective cohort study, examining the incidence rates involving the different services. Multivariate Poisson regression analyses were performed to calculate relative risks (RRs). Differences in the utilisations over the days of the week and of the reasons for contacts were assessed.

RESULTS

In total, 3576 contacts occurred (mean age 80.5 years, women 66.3%), resulting in an overall incidence rate of 2.7 per person-year (95% confidence interval 2.6-2.8). Strongest predictors were polypharmacy (RR 1.79; 95% CI 1.50-2.12), followed by male sex and higher care dependency. Among the three services AHAs showed the highest rates. Injuries were the most common reasons for visiting EDs, whereas for OOHC use and AHAs, coded diagnoses covered a broader spectrum. Utilisation of the services on weekdays varied, particularly for OOHC.

DISCUSSION

Polypharmacy, a higher care dependency and male sex seem to play a role in predicting acute health care services. Considering the distribution of the diagnoses of all three types, certain patterns concerning the symptoms' acuity become apparent.

CONCLUSIONS

Our findings revealed high acute health care services use among NH residents in Germany and differences among the three available services. This information can be used to design studies for investigating the appropriateness of these contacts.

["Emergency anesthesia" in geriatric patients]

BACKGROUND

The share of elderly patients undergoing emergency surgery is constantly increasing. Their postoperative outcome remains poor, even if surgery itself is survived in the short or medium term.

OBJECTIVES

Important aspects of anesthesiologic care for older emergency patients based upon recent literature and guideline recommendations are presented.

METHODS

Selective review of the literature, considering national and international guidelines, meta-analysis and Cochrane reviews.

CONCLUSION

Anesthesiologic care can significantly influence the perioperative outcome of elderly emergency surgery patients. In this context, emergency anesthesiology exceeds mere anesthesia itself and applies to the overall perioperative management.

New AI laboratory for the NHS

The public must be engaged in AI innovations to ensure real benefits for health

Development and validation of an admission prediction tool for emergency departments in the Netherlands

Objective

Early prediction of admission has the potential to reduce length of stay in the ED. The aim of this study is to create a computerised tool to predict admission probability.

Methods

The prediction rule was derived from data on all patients who visited the ED of the Rijnstate Hospital over two random weeks. Performing a multivariate logistic regression analysis factors associated with hospitalisation were explored. Using these data, a model was developed to predict admission probability. Prospective validation was performed at Rijnstate Hospital and in two regional hospitals with different baseline admission rates. The model was converted into a computerised tool that reported the admission probability for any patient at the time of triage.

Results

Data from 1261 visits were included in the derivation of the rule. Four contributing factors for admission that could be determined at triage were identified: age, triage category, arrival mode and main symptom. Prospective validation showed that this model reliably predicts hospital admission in two community hospitals (area under the curve (AUC) 0.87, 95% CI 0.85 to 0.89) and in an academic hospital (AUC 0.76, 95% CI 0.72 to 0.80). In the community hospitals, using a cut-off of 80% for admission probability resulted in the highest number of true positives (actual admissions) with the greatest specificity (positive predictive value (PPV): 89.6, 95% CI 84.5 to 93.6; negative predictive value (NPV): 70.3, 95% CI 67.6 to 72.9). For the academic hospital, with a higher admission rate, a 90% probability was a better cut-off (PPV: 83.0, 95% CI 73.8 to 90.0; NPV: 59.3, 95% CI 54.2 to 64.2).

Conclusion

Admission probability for ED patients can be calculated using a prediction tool. Further research must show whether using this tool can improve patient flow in the ED.

Evaluating the predictive strength of the LACE index in identifying patients at high risk of hospital readmission following an inpatient episode: a retrospective cohort study

OBJECTIVE

To assess how well the LACE index and its constituent elements predict 30-day hospital readmission, and to determine whether other combinations of clinical or sociodemographic variables may enhance prognostic capability.

DESIGN

Retrospective cohort study with split sample design for model validation.

SETTING

One large hospital Trust in the West Midlands.

PARTICIPANTS

All alive-discharge adult inpatient episodes between 1 January 2013 and 31 December 2014.

DATA SOURCES

Anonymised data for each inpatient episode were obtained from the hospital information system. These included age at index admission, gender, ethnicity, admission/discharge date, length of stay, treatment specialty, admission type and source, discharge destination, comorbidities, number of accident and emergency (A&E) visits in the 6 months before the index admission and whether a patient was readmitted within 30 days of index discharge.

OUTCOME MEASURES

Clinical and patient characteristics of readmission versus non-readmission episodes, proportion of readmission episodes at each LACE score, regression modelling of variables associated with readmission to assess the effectiveness of LACE and other variable combinations to predict 30-day readmission.

RESULTS

The training cohort included data on 91 922 patient episodes. Increasing LACE score and each of its individual components were independent predictors of readmission (area under the receiver operating characteristic curve (AUC) 0.773; 95% CI 0.768 to 0.779 for LACE; AUC 0.806; 95% CI 0.801 to 0.812 for the four LACE components). A LACE score of 11 was most effective at distinguishing between higher and lower risk patients. However, only 25% of readmission episodes occurred in the higher scoring group. A model combining A&E visits and hospital episodes per patient in the previous year was more effective at predicting readmission (AUC 0.815; 95% CI 0.810 to 0.819).

CONCLUSIONS

Although LACE shows good discriminatory power in statistical terms, it may have little added value over and above clinical judgement in predicting a patient's risk of hospital readmission.

What is the relationship between age and deprivation in influencing emergency hospital admissions? A model using data from a defined, comprehensive, all-age cohort in East Devon, UK

OBJECTIVES

To clarify the relationship between social deprivation and age as two factors associated with emergency admissions to hospital.

DESIGN

Emergency admissions for 12 months were analysed for patients in the NHS NEW Devon CCG. Social deprivation was measured by the Index of Multiple Deprivation (IMD). Logistic regression models estimated the separate and combined effects of social deprivation and age on the risk of emergency admissions for people aged under and over 65.

SETTING

East Devon, UK-area of the NEW Devon CCG.

POPULATION

765 861 patients in the CCG database.

MAIN OUTCOME MEASURE

Emergency admission to any English hospital.

RESULTS

Age (p<0.001) and social deprivation (p<0.001) were significantly associated with emergency admission to hospital, but there was a significant interaction between age and social deprivation (p<0.001). From the third quintile of age upwards, age progressively overtakes deprivation and age has a dominant effect on emergency admissions over the age of 65. The effect of age was J-shaped in all deprivation groups, increasing exponentially after age 40. For patients under 65, age and social deprivation had similar risks for emergency admissions, the differences in risk between the top and bottom quintiles of IMD and age being ∼1.5 and 0.9 percentage points. In patients over 65, age had a much greater effect on the risk of admissions than social deprivation, the differences in risk between the top and bottom quintiles of IMD and age being ∼2.8 and 18.7 percentage points.

CONCLUSIONS

Risk curves for all social groups have similar shapes, implying a common biological pattern for ageing in any social group. Over age 65, the biological effects of ageing outweigh the social effects of deprivation. Our model enables CCGs to anticipate and plan for emergency admissions to hospital. These findings provide a new logic for allocating resources to different populations.

How can we define and analyse drug exposure more precisely to improve the prediction of hospitalizations in longitudinal (claims) data?

BACKGROUND

Risk prediction models can be powerful tools to support clinical decision-making, to help targeting interventions, and, thus, to improve clinical and economic outcomes, provided that model performance is good and sensitivity and specificity are well balanced. Drug utilization as a potential risk factor for unplanned hospitalizations has recently emerged as a meaningful predictor variable in such models. Drug treatment is a rather unstable (i.e. time-dependent) phenomenon and most drug-induced events are concentration-dependent and therefore individual drug exposure will likely modulate the risk. This especially applies to longitudinal monitoring of appropriate drug treatment within claims data as another promising application for prediction models.

METHODS AND RESULTS

To guide future research towards this direction, we firstly reviewed current risk prediction models for unplanned hospitalizations that explicitly included information on drug utilization and were surprised to find that these models rarely attempted to consider dose and frequent modulators of drug clearance such as interactions with co-medication or co-morbidities. As another example, they often presumed class effects where in fact, differences between active moieties were well established. In addition, the study designs and statistical risk analysis disregarded the fact that medication and risk modulators and, thus, adverse events can vary over time. In a simulation study, we therefore evaluated the potential benefit of time-dependent Cox models over standard binary regression approaches with a fixed follow-up period.

CONCLUSIONS

Longitudinal drug information could be utilized much more efficiently both by precisely estimating individual drug exposure and by applying more refined statistical methodology to account for time-dependent drug utilization patterns.

Improving continuity: THE clinical challenge

Continuity of care is a core feature of general practice; it creates multiple benefits for patients, doctors and society. Continuity increases trust, patient satisfaction, disclosure of information, take-up of preventive care, adherence to advice, reduction in socio-economic disadvantage, and reduces deaths. However, the level of continuity is reducing in general practice. About 15 consultations are needed with a patient for a GP to acquire enough ‘accumulated knowledge’ to develop a sense of continuing responsibility. This fosters GP sensitivity and mutual understanding, which enable GPs to provide ‘higher-level’ quality of care. The RCGP curriculum states two high-level aims: that GPs need to ‘enhance continuity of care’ and ‘build long-term relationships with patients’. This article analyses these aims by setting them in the context of international research on continuity of care.

Fifty years of longitudinal continuity in general practice: a retrospective observational study

BACKGROUND

Continuity of care has been defined as relational continuity between patient and doctor and longitudinal continuity describing the duration of the relationship. Measurement of longitudinal continuity alone is associated with outcomes including patient satisfaction, medical costs, hospital admissions and mortality.

METHODS

In one UK general practice, records were searched for patients with continuous registration for 50 or more years. Characteristics of these patients were analysed for age, gender, social deprivation, partner registration and length of registration. Trends in numbers and proportions of this group over the previous 14 years were determined. A comparison group of patients, aged 50 or more, and registered in the same practice within the last 2-4 years, was identified.

RESULTS

Patients registered for 50 years or more with a median registration of 56.2 years numbered 190 out of a population of 8420 (2.3%). These patients increased in number by 35.3% (1.7-2.3%) over 14 consecutive years. There were no differences between groups for GP consultation rate, number of repeat medications and hospital use, despite the significantly higher prevalence of multi-morbidity, depression and diabetes in patients with high continuity.

CONCLUSIONS

This is the first report of 50-year continuity in general practice. Numbers of such patients and proportions are increasing. Longitudinal continuity is easily measured in general practice and associated with important clinical outcomes.

The Oldest Old in the Emergency Department: Impact of Renal Function

OBJECTIVES

The ageing population implicates an increasing numbers of older adults attending Emergency Departments (ED). We assessed the effect of estimated glomerular filtration rate as a predictor of clinical outcomes in oldest-old patients ≥ 85 years attending the ED in an university teaching hospital.

DESIGN

Within three years, 81831 patient contacts were made in our ED. 7799 (9.5%) were older than 85 years, in whom we analyzed the impact of renal function on various outcome parameters. Furthermore, this patient group was compared to the patients < 85 years.

RESULTS

Within the group of patients ≥ 85 years, not older age, but as denominator decreased glomerular filtration rate led to significant longer hospital stays. In addition, impaired kidney function was associated with lower heart rates, lower blood pressure, lower oxygenation, a higher rate of established ambulant care setting, as well as higher mortality. Compared to younger patients, the oldest-old significantly differed with regard to medical attribution (e.g. internal medicine, surgery), sex distribution, length of hospital stay, Manchester triage score, Glasgow Coma Scale, visual analogue pain scale, heart rate, blood pressure, oxygen saturation as well as fall prophylaxis, outpatient care, and presence of relatives.

CONCLUSION

In conclusion, in this large collective of oldest-old patients, impaired kidney function seems to be a more important determinant in adverse outcome and thus increased health care costs than age per se. Adapted strategies in EDs to adjust diagnostic and treatment strategies for this population are thus warranted.

Selection of the method to appraise and compare health systems using risk stratification: the ASSEHS approach

To face the challenge of active and healthy ageing, European Health Systems and services should move towards proactive, anticipatory and integrated care. The comparison of methods to combine results across studies and to determine an overall effect was undertaken by the EU project ASSEHS (Activation of Stratification Strategies and Results of the interventions on frail patients of Healthcare Services, EU project (No. 2013 12 04). The questions raised in ASSEHS are broad and involve a complex body of literature. Thus, systematic reviews are not appropriate. The most appropriate method appears to be scoping studies. In this paper, an updated method of scoping studies has been used to determine the questions needed to appraise the health systems and services for frailty in the ageing population. Three objectives were set (i) to detect a relevant number of risk stratification tools for frailty and identify the best-in-class, (ii) to understand the feasibility of introducing stratification tools and identify the difficulties of the process and (iii) to find evidence on the impact of risk stratification in Health Services. This novel approach may provide greater clarity about scoping study methodology and help enhance the methodological rigor with which authors undertake and report scoping studies.

Acute care of older patients in the emergency department: strategies to improve patient outcomes

Older patients in the emergency department (ED) are a vulnerable population who are at a higher risk of functional decline and hospital reattendance subsequent to an ED visit, and have a high mortality rate in the months following an ED attendance. The delivery of acute care in a busy environment to this population presents its own unique challenge. The purpose of this review is to detail the common geriatric syndromes encountered in the ED as well as the appropriate strategies and instruments, which can be utilized to support the clinical decision matrix and improve outcomes.

Multimorbidity in risk stratification tools to predict negative outcomes in adult population

INTRODUCTION

Risk stratification tools were developed to assess risk of negative health outcomes. These tools assess a variety of variables and clinical factors and they can be used to identify targets of potential interventions and to develop care plans. The role of multimorbidity in these tools has never been assessed.

OBJECTIVES

To summarize validated risk stratification tools for predicting negative outcomes, with a specific focus on multimorbidity.

METHODS

MEDLINE, Cochrane Central Register of Controlled Trials and PubMed database were interrogated for studies concerning risk prediction models in medical populations. Review was conducted to identify prediction models tested with patients in both derivation and validation cohorts. A qualitative synthesis was performed focusing particularly on how multimorbidity is assessed by each algorithm and how much this weighs in the ability of discrimination.

RESULTS

Of 3674 citations reviewed, 36 articles met criteria. Of these, 29 had as outcome hospital admission/readmission. The most common multimorbidity measure employed in the models was the Charlson Comorbidity Index (12 articles). C-statistics ranged between 0.5 and 0.85 in predicting hospital admission/ readmission. The highest c-statistics was 0.83 in models with disability as outcome. For healthcare cost, models which used ACG-PM case mix explained better the variability of total costs.

CONCLUSIONS

This review suggests that predictive risk models which employ multimorbidity as predictor variable are more accurate; CHF, cerebro-vascular disease, COPD and diabetes were strong predictors in some of the reviewed models. However, the variability in the risk factors used in these models does not allow making assumptions.

Characteristics and outcomes of older persons attending the emergency department: a retrospective cohort study

BACKGROUND

The analysis of routinely collected hospital data informs the design of specialist services for at-risk older people.

AIM

Describe the outcomes of a cohort of older emergency department (ED) attendees and identify predictors of these outcomes.

DESIGN

retrospective cohort study.

METHODS

All patients aged 65 years or older attending an urban university hospital ED in January 2012 were included (N = 550). Outcomes were retrospectively followed for 12 months. Statistical analyses were based on multivariate binary logistic regression models and classification trees.

RESULTS

Of N = 550, 40.5% spent ≤6 h in the ED, but the proportion was 22.4% among those older than 81 years and not presenting with musculoskeletal problems/fractures. N = 349 (63.5%) were admitted from the ED. A significant multivariate predictor of in-hospital mortality was Charlson comorbidity index [CCI; odds ratio = 1.19, 95% confidence interval: 1.07, 1.34, P = 0.002]. Among patients who were discharged from ED without admission or after their first in-patient admission (N = 499), 232 (46.5%) re-attended ED within 1 year, with CCI being the best predictor of re-attendance (CCI ≤ 4: 25.8%, CCI > 5: 60.4%). Among N = 499, 34 (6.8%) had died after 1 year of initial ED presentation. The subgroup (N = 114) with the highest mortality (17.5%) was composed by those aged >77 years and brought in by ambulance on initial presentation.

CONCLUSIONS

Advanced age and comorbidity are important drivers of outcomes among older ED attendees. There is a need to embed specialist geriatric services within frontline services to make them more gerontologically attuned. Our results predate the opening of an acute medical unit with specialist geriatric input.

Risk prediction models to predict emergency hospital admission in community-dwelling adults: a systematic review

Supplemental Digital Content is available in the text.

Background:

Risk prediction models have been developed to identify those at increased risk for emergency admissions, which could facilitate targeted interventions in primary care to prevent these events.

Objective:

Systematic review of validated risk prediction models for predicting emergency hospital admissions in community-dwelling adults.

Methods:

A systematic literature review and narrative analysis was conducted. Inclusion criteria were as follows; Population: community-dwelling adults (aged 18 years and above); Risk: risk prediction models, not contingent on an index hospital admission, with a derivation and ≥1 validation cohort; Primary outcome: emergency hospital admission (defined as unplanned overnight stay in hospital); Study design: retrospective or prospective cohort studies.

Results:

Of 18,983 records reviewed, 27 unique risk prediction models met the inclusion criteria. Eleven were developed in the United States, 11 in the United Kingdom, 3 in Italy, 1 in Spain, and 1 in Canada. Nine models were derived using self-report data, and the remainder (n=18) used routine administrative or clinical record data. Total study sample sizes ranged from 96 to 4.7 million participants. Predictor variables most frequently included in models were: (1) named medical diagnoses (n=23); (2) age (n=23); (3) prior emergency admission (n=22); and (4) sex (n=18). Eleven models included nonmedical factors, such as functional status and social supports. Regarding predictive accuracy, models developed using administrative or clinical record data tended to perform better than those developed using self-report data (c statistics 0.63–0.83 vs. 0.61–0.74, respectively). Six models reported c statistics of >0.8, indicating good performance. All 6 included variables for prior health care utilization, multimorbidity or polypharmacy, and named medical diagnoses or prescribed medications. Three predicted admissions regarded as being ambulatory care sensitive.

Conclusions:

This study suggests that risk models developed using administrative or clinical record data tend to perform better. In applying a risk prediction model to a new population, careful consideration needs to be given to the purpose of its use and local factors.

Choosing a model to predict hospital admission: an observational study of new variants of predictive models for case finding

OBJECTIVES

To test the performance of new variants of models to identify people at risk of an emergency hospital admission. We compared (1) the impact of using alternative data sources (hospital inpatient, A&E, outpatient and general practitioner (GP) electronic medical records) (2) the effects of local calibration on the performance of the models and (3) the choice of population denominators.

DESIGN

Multivariate logistic regressions using person-level data adding each data set sequentially to test value of additional variables and denominators.

SETTING

5 Primary Care Trusts within England.

PARTICIPANTS

1 836 099 people aged 18-95 registered with GPs on 31 July 2009.

MAIN OUTCOME MEASURES

Models to predict hospital admission and readmission were compared in terms of the positive predictive value and sensitivity for various risk strata and with the receiver operating curve C statistic.

RESULTS

The addition of each data set showed moderate improvement in the number of patients identified with little or no loss of positive predictive value. However, even with inclusion of GP electronic medical record information, the algorithms identified only a small number of patients with no emergency hospital admissions in the previous 2 years. The model pooled across all sites performed almost as well as the models calibrated to local data from just one site. Using population denominators from GP registers led to better case finding.

CONCLUSIONS

These models provide a basis for wider application in the National Health Service. Each of the models examined produces reasonably robust performance and offers some predictive value. The addition of more complex data adds some value, but we were unable to conclude that pooled models performed less well than those in individual sites. Choices about model should be linked to the intervention design. Characteristics of patients identified by the algorithms provide useful information in the design/costing of intervention strategies to improve care coordination/outcomes for these patients.