Considerable progress continues to be made with regards to the value and use of disease associated polygenic scores (PGS). PGS aim to capture a person’s genetic liability to a condition, disease, or a trait, combining information across many risk variants and incorporating their effect sizes. They are already available for clinicians and consumers to order in Australasia. However, debate is ongoing over the readiness of this information for integration into clinical practice and population health. This position statement provides the viewpoint of the Human Genetics Society of Australasia (HGSA) regarding the clinical application of disease-associated PGS in both individual patients and population health. The statement details how PGS are calculated, highlights their breadth of possible application, and examines their current challenges and limitations. We consider fundamental lessons from Mendelian genetics and their continuing relevance to PGS, while also acknowledging the distinct elements of PGS. Use of PGS in practice should be evidence based, and the evidence for the associated benefit, while rapidly emerging, remains limited. Given that clinicians and consumers can already order PGS, their current limitations and key issues warrant consideration. PGS can be developed for most complex conditions and traits and can be used across multiple clinical settings and for population health. The HGSA’s view is that further evaluation, including regulatory, implementation and health system evaluation are required before PGS can be routinely implemented in the Australasian healthcare system.

Mortality risk is known to be associated with many physiological or biochemical risk factors, and polygenic risk scores (PRSs) may offer an additional or alternative approach to risk stratification. We have compared the predictive value of common biochemical tests, PRSs and information on parental survival in a cohort of twins and their families. Common biochemical test results were available for up to 13,365 apparently healthy men and women, aged 17−93 years (mean 49.0, standard deviation [SD] 13.7) at blood collection. PRSs for longevity were available for 14,169 study participants and reported parental survival for 25,784 participants. A search for information on date and cause of death was conducted through the Australian National Death Index, with median follow-up of 11.3 years. Cox regression was used to evaluate associations with mortality from all causes, cancers, cardiovascular diseases and other causes. Linear relationships with all-cause mortality were strongest for C-reactive protein, gamma-glutamyl transferase, glucose and alkaline phosphatase, with hazard ratios (HRs) of 1.16 (95% CI [1.07, 1.24]), 1.15 (95% CI 1.04–1.21), 1.13 (95% CI [1.08, 1.19]) and 1.11 (95% CI [1.05, 1.88]) per SD difference, respectively. Significant nonlinear effects were found for urea, uric acid and butyrylcholinesterase. Lipid risk factors were not statistically significant for mortality in our cohort. Family history and PRS showed weaker but significant associations with survival, with HR in the range 1.05 to 1.09 per SD difference. In conclusion, biochemical tests currently predict long-term mortality more strongly than genetic scores based on genotyping or on reported parental survival.

Introduction: Very low high-sensitivity troponin-T (hs-cTnT) concentrations on presentation can rule out acute myocardial infarction (AMI), but the ability to identify patients at low risk of 30-day major adverse cardiac events (MACE) is less clear. This study examines the sensitivity of low concentrations of hs-cTnT on presentation to rule out 30-day MACE. Methods: This prospective cohort study enrolled emergency department chest pain patients with non-ischemic ECGs who underwent AMI rule-out with an hs-cTnT assay. The primary outcome was 30-day MACE; secondary outcomes were individual MACE components. Because guidelines recommend using a single hs-cTnT strategy only for patients with more than 3-hours since symptom onset, a subgroup analysis was performed for this population. Outcomes were adjudicated based on review of medical records and telephone follow-up. Results: Of 1,167 patients enrolled, 125 (10.7%) experienced 30-day MACE and 97 (8.3%) suffered AMI on the index visit. More than one-third (35.6%) had presenting hs-cTnT concentrations below the limit of detection (5ng/L), which was 94.4% (95%CI 88.8-97.7%) sensitive for 30-day MACE and 99.0% (95%CI 94.5-100%) sensitive for index AMI. Of 292 (25.0%) patients with hs-cTnT < 5ng/L and at least 3-hours since symptom onset, only 3 experienced 30-day MACE (sensitivity 97.6%, 95%CI 93.2-100%) and none suffered AMI within 30-days (sensitivity 100%, 95%CI 96.3-100%). Conclusion: Among patients with non-ischemic ECGs and >3-hours since symptom onset, low hs-cTnT concentrations on presentation confer a very low risk of 30-day MACE. In the absence of a high risk clinical presentation, further risk stratification is likely to be low yield.

Introduction: Two published studies reported natriuretic peptides can aid in risk-stratification of Emergency Department (ED) syncope. We sought to assess the role of N-Terminal pro Brain Natriuretic Peptide (NT pro-BNP) to identify syncope patients at risk for serious adverse events (SAE) within 30 days of the ED visit, and its value above that of the Canadian Syncope Risk Score (CSRS). Methods: We conducted a multicenter prospective cohort study at 6 large Canadian EDs from Nov 2011 to Feb 2015. We enrolled adults who presented within 24-hours of syncope and excluded those with persistent altered mentation, obvious seizure, and intoxication. We collected patient characteristics, nine CSRS predictors (includes troponin), ED management and NT pro-BNP levels. Adjudicated serious adverse events (SAE) included death, cardiac SAE (arrhythmias, myocardial infarction, serious structural heart disease) and non-cardiac SAE (pulmonary embolism, severe hemorrhage and procedural interventions within 30-days). We used two tailed t-test and logistic regression analysis. Results: Of the 1359 patients (mean age 57.2 years, 54.7% females, 13.3% hospitalized) enrolled, 148 patients (10.9%; 0.7% deaths, 7.9% cardiac SAE including 6.1% arrhythmia) suffered SAE within 30-days. The mean NT pro-BNP values, when compared to the patients with no SAE (499.8ng/L) was significantly higher among the 56 patients who suffered SAE after ED disposition (3147ng/L, p=0.001), and among the 35 patients with cardiac SAE after ED disposition (2016.2ng/L, p=0.02). While there was a trend to higher levels among patients who suffered arrhythmia after the ED visit, it was not statistically significant (1776.4ng/L, p=0.07). In a model with CSRS predictors, the adjusted odds ratio for NT pro-BNP was 8.0 (95%CI 1.8, 35.9) and troponin was 3.8 (95%CI 1.7, 8.8). The addition of NT pro-BNP did not significantly improve the classification performance (p=0.76) with areas under the curves for CSRS was 0.91 (95%CI 0.88, 0.95) and CSRS with NT pro-BNP was 0.92 (95%CI 0.88, 0.95). Conclusion: In this multicenter study, mean NT pro-BNP levels were significantly higher among ED syncope patients who suffered SAE including cardiac SAE after ED disposition. Though NT pro-BNP was a significant independent predictor of SAE after ED disposition, it did not improve accuracy in ED syncope risk-stratification when compared to CSRS. Hence, we do not recommend NT pro-BNP measurement for ED syncope management.

Introduction: Lower gastrointestinal bleeding (LGIB) can result in serious adverse events, including recurrent bleeding, need for intervention and death. Endoscopy is important in the management of LGIB, however gastroenterologists have limited resources to safe endoscopy. Risk stratification of LGIB patients can aid physicians in disposition decisions. Objective: to develop a clinical decision tool to accurately identify LGIB patients presenting to the emergency department (ED) who are at risk for 30-day serious adverse events. Methods: We conducted a health records review and included 372 adult ED patients who presented with an acute LGIB. The outcome was a 30-day composite outcome consisting of all-cause death, recurrent LGIB, need for intervention to control the bleed and ICU admission. A second researcher confirmed data-collection of 10% of the data and we calculated a -value for inter-rater reliability. We analyzed the data using stepwise backwards selection and SELECTION=SCORE option and calculated the diagnostic accuracy of the final model. Results: Age 75 years, hemoglobin 100 g/L, INR 2.0, a bloody stool in the ED and a past medical history of colorectal polyps were significant predictors in the multivariable regression analysis. The AUC was 0.83 (95% CI 0.77-0.89), sensitivity 0.96 (0.90-1.00), specificity 0.53 (0.48-0.59), and negative likelihood ratio 0.08 (0.02-0.30) for a cut-off score of 1. Conclusion: This model showed good ability to identify LGIB patients at low risk for adverse events as evidenced by the high AUC, sensitivity and negative likelihood ratio. Future, large prospective studies should be done to confirm the data, after which it should be validated and implemented.

Introduction: The Canadian Syncope Risk Score (CSRS) was developed to identify patients at risk for serious adverse events (SAE) within 30 days of an Emergency Department (ED) visit for syncope. We sought to validate the score in a new cohort of ED patients. Methods: We conducted a multicenter prospective cohort study at 8 large academic tertiary-care EDs across Canada from March 2014 to Dec 2016. We enrolled adults (age 16 years) who presented within 24 hours of syncope, after excluding those with persistent altered mentation, witnessed seizure, intoxication, and major trauma requiring hospitalization. Treating ED physicians collected the nine CSRS predictors at the index visit. Adjudicated SAE included death, arrhythmias and non-arrhythmic SAE (myocardial infarction, serious structural heart disease, pulmonary embolism, severe hemorrhage and procedural interventions within 30-days). We assessed area under the Receiver Operating Characteristic (ROC) curve, score calibration, and the classification performance for the various risk categories. Results: Of the 2547 patients enrolled, 146 (5.7%) were lost to follow-up and 111 (4.3%) had serious condition during the index ED visit and were excluded. Among the 2290 analyzed, 79 patients (3.4%; 0.4% death, 1.4% arrhythmia) suffered 30-day serious outcomes after ED disposition. The accuracy of the CSRS remained high with area under the ROC curve at 0.87 (95%CI 0.82-0.92), similar to the derivation phase (0.87; 95%CI 0.84-0.89). The score showed excellent calibration at the prespecified risk strata. For the very-low risk category (0.3% SAE of which 0.2% were arrhythmia and no deaths) the sensitivity was 97.5% and negative predictive value was 99.7% (95%CI 98.7-99.9). For the very high-risk category (61.5% SAE of which 26.9% were arrhythmia and 11.5% death) the specificity was 99.4% and positive predictive value was 61.5% (95% CI 43.0-77.2). Conclusion: In this multicenter validation study, the CSRS accurately risk stratified ED patients with syncope for short-term serious outcomes after ED disposition. The score should aid in minimizing investigation and observation of very-low risk patients, and prioritization of inpatient vs outpatient investigations or following of the rest. The CSRS is ready for implementation studies examining ED management decisions, patient safety and health care resource utilization.

Introduction: London Health Sciences Centre (LHSC) includes two academic, urban hospitals in London, Canada. The hospital-standardized mortality ratio (HSMR) is consistently higher than provincial and national averages. Unpublished data reveals that sepsis contributes the largest number of statistically unexpected deaths to LHSC’s HSMR calculation. Factors contributing to in-hospital sepsis mortality are hypothesized to include demography, emergency department (ED) flow or sepsis treatment. Methods: Retrospective chart review of patients aged >=18 years, presenting to an LHSC ED between 01 Nov 2014 and 31 Oct 2015, with >=2 SIRS criteria and/or ED suspicion of infection and/or ED or hospital discharge sepsis diagnosis (ICD-10 diagnostic codes A4xx and R65). Data were abstracted from electronic health records. Regional, provincial and national data was retrieved from CIHI and Statistics Canada. Results: Median age and sex in London and across Canada are similar (48.2 years vs 48.9 years; 48% male vs 49% male). Baseline prevalences of diabetes, hypertension, COPD and mood disorders were similar in the Local Health Integration Network and Ontario (6% vs 7%, 19% vs 19%, 3% vs 4%, and 10% vs 8%). Median “Physician Initial Assessment,” (PIA) times for sepsis patients at LHSC were faster than median Canadian PIA times for CTAS I and II patients (CTAS I: 7 min vs 11 min, CTAS II: 34 min vs 54 min), and slower for CTAS III-V patients (CTAS III: 98 min vs 79 min, CTAS IV: 99 min vs 66 min, CTAS V: 132 min vs 53 min). Median ED length of stay for admitted, high acuity (CTAS I-III) patients was 6 h at LHSC versus 10 h across Canada.Median [IQR] time to intravenous fluid resuscitation was 60.5 min [29.8-101.2] for septic shock patients and 77.0 min [36.0-127.0] for expired patients. Median [IQR] time to antibiotics was 130 min [73.0-229.0] for sepsis patients, 106 min [60.0-189.0] for severe sepsis patients, and 82 min [42.2-142] for septic shock patients. Conclusion: Excess sepsis-related mortality at LHSC is not convincingly related to patient demographics or ED flow. Gains may be made by improving time to antibiotics and IV fluids.

Introduction: Early Emergency Department (ED) identification of septic patients at risk of subsequent deterioration is necessary in order to optimize disposition. High-risk patients admitted directly from the ED to the ICU have better outcomes than those admitted to the floor first. Initial ED serum lactate level has been associated with 28-day mortality in admitted patients, but there is little evidence on its use in predicting short-term deterioration. Furthermore, it is unclear whether the addition of respiratory rate (RR) to lactate would create a stronger predictive model of deterioration than either alone. Methods: Prospective cohort study of ED patients (age ≥18) screened and treated for sepsis (defined as physician suspicion of infection and 2 or more of the SIRS criteria). Lactate and vital signs were obtained within 2 hours of ED arrival. Main outcome was deterioration (defined as any of the following: death; ICU admission >24 hours; Intubation; Vasoactive medications for >1 hour; or Non-invasive positive pressure ventilation for >1 hour) within 72 hours. Patients meeting an endpoint within 1 hour of arrival were excluded. Discharged patients were contacted at 72 hours to ensure that they had not met the endpoint or presented to another institution. Results: 985 patients presenting to either of two urban high-volume EDs were enrolled, of whom 84 (8.5%) met the primary outcome. Initial serum lactate ≥ 4.0 had a specificity of 97.4% (95% CI, 94.1-100%), but a sensitivity of 27.4% (95% CI, 17-8-36.9%) for predicting deterioration. Of patients with a lactate ≥ 4.0, 4 (8.7%) were discharged home, and did not reach an endpoint at 72 hours. Lactate <2.0 had a sensitivity of 95.5% (95% CI: 93.4-97.1%) and specificity of 84.5% (95% CI: 80.4-88.6%) for ruling out 72-hour deterioration. Of patients with a lactate <2.0, 224 (56.1%) were discharged home. Combining lactate with RR (AUC: 0.72, 95% CI: 0.66-0.79) did not yield better predictive capability than lactate alone (AUC: 0.70, 95% CI: 0.64-0.76). Conclusion: Initial ED lactate is predictive of deterioration within 72 hours in patients with sepsis. The combination of lactate and RR was not more predictive of deterioration than lactate alone. This suggests that serum lactate has a role in predicting deterioration in patients with sepsis, and has utility in determining disposition.

Introduction: Concern for occult serious conditions leads to variations in ED syncope management [hospitalization, duration of ED/inpatient monitoring including Syncope Observation Units (SOU) for prolonged monitoring]. We sought to develop evidence-based recommendations for duration of ED/post-ED ECG monitoring using the Canadian Syncope Risk Score (CSRS) by assessing the time to serious adverse event (SAE) occurrence. Methods: We enrolled adults with syncope at 6 EDs and collected demographics, time of syncope and ED arrival, CSRS predictors and time of SAE. We stratified patients as per the CSRS (low, medium and high risk as ≤0, 1-3 and ≥4 respectively). 30-day adjudicated SAEs included death, myocardial infarction, arrhythmia, structural heart disease, pulmonary embolism or serious hemorrhage. We categorized arrhythmias, interventions for arrhythmias and death from unknown cause as arrhythmic SAE and the rest as non-arrhythmic SAE. We performed Kaplan-Meier analysis using time of ED registration for primary and time of syncope for secondary analyses. Results: 5,372 patients (mean age 54.3 years, 54% females, and 13.7% hospitalized) were enrolled with 538 (10%) patients suffering SAE (0.3% died due to an unknown cause and 0.5% suffered ventricular arrhythmia). 64.8% of SAEs occurred within 6 hours of ED arrival. The probability for any SAE or arrhythmia was highest within 2-hours of ED arrival for low-risk patients (0.65% and 0.31%; dropped to 0.54% and 0.06% after 2-hours) and within 6-hours for the medium and high-risk patients (any SAE 6.9% and 17.4%; arrhythmia 6.5% and 18.9% respectively) which also dropped after 6-hours (any SAE 0.99% and 2.92%; arrhythmia 0.78% and 3.07% respectively). For any CSRS threshold, the risk of arrhythmia was highest within the first 15-days (for CSRS ≥2 patients 15.6% vs. 0.006%). ED monitoring for 2-hours (low-risk) and 6-hours (medium and high-risk) and using a CSRS ≥2 cut-off for outpatient 15-day ECG monitoring will lead to 52% increase in arrhythmia detection. The majority (82.2%) arrived to the ED within 2-hours (median time 1.1 hours) and secondary analysis yielded similar results. Conclusion: Our study found 2 and 6 hours of ED monitoring for low-risk and medium/high-risk CSRS patients respectively, with 15-day outpatient ECG monitoring for CSRS ≥2 patients will improve arrhythmia detection without the need for hospitalization or observation units.

Introduction: Acute heart failure (AHF) is a common, serious condition that frequently results in morbidity and death and is a leading cause for hospital admissions. There is little evidence to guide ED physician disposition decisions for AHF patients. We sought to create a risk-stratification tool for use by ED physicians to determine which AHF patients are at high risk for poor outcomes. Methods: We conducted a prospective cohort study in 9 tertiary hospital EDs and enrolled adult patients presenting with shortness of breath due to AHF. Patients were assessed for standardized clinical and laboratory variables and then followed to determine short-term serious outcome (SSO), defined as death, intubation, myocardial infarction, or relapse requiring admission within 14 days. We identified predictors of SSO by stepwise logistic regression and then rounded beta coefficients to create a risk scale. Results: We enrolled 1,733 patients with mean age 77.1 years, male 54.5%, and initially admitted 50.1%. SSOs occurred in 202 (11.7%) cases (14.0% in those admitted and 9.3% in those discharged from the ED). We created the CHFRS consisting of:1. Initial Assessment a) History of valvular heart disease b) On anti-arrhythmic c) Arrival heart rate ≥ 110d) Treated with non-invasive ventilation2. Investigations a) Urea >12 mmol/L or Cr>150 µmol/L b) Serum CO2>35 mmol/L or pCO2 >60 mmHg (VBG or ABG) c) Troponin >5x Upper Reference Level 3. Fails reassessment after ED treatment:(i) Resting vital signs abnormal, (SaO2 <90% on room air or usual O2, or HR >110, or RR >28); OR(ii) Unable to complete 3-minute walk test. The risk of SSO varied from 5.0% for a score of 0, to 77.4% for a score of 9. Discrimination between SSO and no SSO cases was good with an area under the ROC curve of 0.70 (95% CI 0.66-0.74). There was good calibration between the observed and expected probability of SSO and internal validation showed the risk scores to be very accurate across 1,000 replications using the bootstrap method. Conclusion: We have created the CHFRS tool which consists of 8 simple variables and which estimates the short-term risk of SSOs in AHF patients. CHFRS should help improve and standardize admission practices, diminishing both unnecessary admissions for low-risk patients and unsafe discharge decisions for high-risk patients. This will ultimately lead to better safety for patients and more efficient use of hospital resources.

Introduction: Suspicion of arrhythmias among syncope patients is the leading cause of emergency department (ED) referrals and hospitalization. However, the risk factors for short-term arrhythmias are not well defined. We sought to develop a risk prediction tool to identify syncope patients at risk for 30-day arrhythmia or death after ED disposition. Methods: This prospective cohort study involved 6 academic EDs that enrolled adult syncope patients. We collected standardized variables at index presentation from history, clinical examination, investigations including ECG, and patients’ disposition. Adjudicated outcomes included death (due to arrhythmia or unknown cause), arrhythmia or procedural intervention to treat arrhythmias within 30-days after ED disposition. Multivariable logistic regression was used to derive the model; bootstrap sampling for internal validation and to estimate shrinkage and optimism. Results: 5,010 adult syncope patients (mean age 53.4 years, 54.8% females, and 9.5% hospitalized) were enrolled with 106 (3.6%) patients suffering arrhythmia or death within 30-days after ED disposition. Of 39 candidate predictors examined, eight were included in the final model: vasovagal predisposition, heart disease, any ED systolic blood pressure <90 or >180 mmHg, troponin (>99%ile), QRS duration >130msec, QTc interval >480msec and ED diagnosis of cardiac, or vasovagal syncope [Optimism corrected c-statistic: 0.91 (95%CI 0.87-0.93); Hosmer-Lemeshow p=0.08]. The Canadian Syncope Arrhythmia Risk Score had a risk ranging from 0.2% for a score of -2 to 74.5% for a score of 8. Sensitivity for threshold score ≤-1 was 100% (95% CI 96.5-100) and specificity for a score of ≥4 was 97.0% (95% CI 96.5-97.5). Conclusion: The Canadian Syncope Arrhythmia Risk Score can improve acute management of ED patients with syncope by better identification of those at higher-risk for short-term arrhythmia or death. Once validated, the tool can be used to guide disposition decision and can also aid in selection of patients for out-of-hospital cardiac monitoring if discharged home.

Introduction: Patients with upper gastrointestinal bleeding (UGIB) are at risk for serious adverse events (SAE) after emergency department (ED) discharge. Endoscopy can aid in risk stratification but is not easily available. Therefore, stratifying using pre-endoscopic risk scores can aid ED physicians in disposition decisions. The aim of this study was to conduct a systematic review to assess the predictive value of pre-endoscopic risk scores for risk-stratification of ED UGIB patients. Methods: We searched 4 databases from inception to March 2015 with search terms related to “UGIB” and “ED”. Inclusion criteria were: 1) adult UGIB patients presenting to the ED; 2) risk scores without endoscopic predictors developed and validated in variceal and non-variceal UGIB patients. We excluded case reports, reviews, abstracts, animal studies and commentaries. In 2 phases (screening and full-review), 2 reviewers independently screened articles for inclusion. SAE included 30-day death, recurrent bleeding and need for intervention. Two reviewers independently extracted patient level data and the consensus data was used for analysis. We report kappa for the article selection, and pooled sensitivity, specificity, positive and negative predictive value, positive and negative likelihood ratios and accuracy with 95% CI for the risk scores. Results: We identified 3,173 articles, of which 3,065 were excluded in phase I (kappa 0.88, 95% CI 0.83-0.93). In phase II, we included 16 of the 108 remaining articles (kappa 0.84, 95% CI 0.70-0.97); 3 studied Glasgow Blatchford Score (GBS), 1 clinical Rockall score (cRockall) and 2 AIMS65; 6 compared GBS and cRockall, 3 compared GBS, a modification of the GBS and cRockall and 1 compared the GBS and AIMS65. Overall, the accuracy of the GBS, cRockall and AIMS65 was 0.47 (95% CI 0.46-0.47), 0.47 (95% CI 0.46-0.49) and 0.62 (95% CI 0.61-0.62), respectively. The accuracy for the GBS with a cut-off score of 2 was 0.73 (95% CI 0.71-0.74). Conclusion: None of the risk scores identified by our systematic review were robust and hence, cannot be recommended for use in clinical practice. However, the GBS with a cut-off score of 2 was superior over other risk scores. Future prospective studies are needed to develop robust new scores for use in ED patients with UGIB.

Rising unplanned hospital admissions are a problem in ageing populations worldwide. These admissions are associated with poor outcomes for older people, contribute to rising health care costs and impede the provision of planned care. Policy and practice in recent years has focused on identification of those at risk of unplanned admission and early intervention via a range of admission avoidance services. Despite this, unplanned admissions in older people continue to rise, and managing demand for unplanned care remains a priority. Questions remain about the risk factors for unplanned admission and the best approaches to identifying and intervening with those at risk. This review explores recent evidence on admission rates, risk factors for unplanned admission in older people, identification of those at highest risk and interventions to avert unplanned admission.

Objectives: It was hypothesised that patients admitted to forensic mental health facilities are stratified along the pathway through care according to levels of need. Level of risk and psychopathology should vary with different levels of security.

Method: Seventy-five men in a forensic hospital were interviewed by three trained clinicians using the HCR-20 (Historical Clinical Risk Assessment) – clinical and risk items, The Health of the Nation Scales – Secure (HoNOS-SECURE), PANSS (Positive and Negative Syndrome Scale), GAF (Global Assessment of Functioning) and the CANFOR (Camberwell Assessment of need Forensic Version).

Results: The mean scores on a variety of clinical measures were higher in admission/high security areas and progressively lower in rehabilitation and pre-discharge areas. As patients moved through the pathways of care, they improved in a number of areas including psychiatric morbidity, risk, function, unmet needs. The following results stratified significantly; the HCR-20 summated clinical and risk (F = 9.2, df = 5, p < 0.001), the HoNOS secure (F = 18.2, df = 5, p < 0.001), PANSS (positive, general and total), GAF, staff and user unmet needs on the CANFOR.

Conclusions: The data indicate that the theoretical organisation of the units of the hospital into high, medium and low security units to form a coherent pathway through care is reflected in practice. This is a transparent route out of secure care in which restrictions are proportionate to risk and supports proportionate to need. It is unclear whether alternative models, consisting of a series of generic unstratified units for admission and discharge, all at the same level of therapeutic security, allow for the provision of treatment programmes and relational interventions appropriate to the patient's stage of recovery and rehabilitation.