PRISMA-7 and Risk for Short-Term Adverse Events in Older Patients Visiting the Emergency Department: Results of a Large Observational and Prospective Cohort Study

Abstract

Background

The “Program of Research on the Integration of Services for the Maintenance of Autonomy” (PRISMA-7) is the reference tool for the assessment of older patients visiting the emergency departments (EDs) in the province of Quebec (Canada). This study aimed to examine 1) whether the PRISMA-7 high-risk level for disabilities was associated with the length of stay in ED and in hospital, and hospital admission; and 2) performance criteria (i.e., sensitivity, specificity, positive predictive value [PPV], negative predictive value [NPV], likelihood ratios [LR]) of the PRISMA-7 high-risk level for the length of stay in ED and hospital, and hospital admission in older ED users.

Methods

A total of 12,983 older ED users of the Jewish General Hospital (Montreal, Quebec, Canada) were recruited in this observational and prospective cohort study. All enrolled participants had a PRISMA-7 assessment upon their arrival at ED. The length of stay in ED and hospital, and hospital admission were used as outcomes.

Results

A PRISMA-7 high-risk level was associated with an increased length of stay in ED and hospital (β ≥2.1 with P≤0.001 and Hazard ratio (HR)= ≥1.2 with P≤0.001) as well as in hospital (HR=1.27 with P≤0.001) in patients on a stretcher. All performance criteria were low (i.e., <0.78). Patients with a PRISMA-7 high-risk level were discharged significantly later from ED and hospital compared to those with low-risk level (P=0.001).

Interpretation

A PRISMA-7 high-risk level was associated with a long length of stay in ED and hospital, and hospital admission in patients on a stretcher but had poor performance criteria for these adverse events, suggesting that it cannot be used as a prognostic tool in older ED users.

Introduction

In Canada, older ED users account for up to a quarter of all ED visits (1., 2., 3.). They are the fastest increasing group of patients visiting ED (3., 4., 5., 6., 7.). The ED is not an ideal place for frail older patients to be cared for but once in ED, it is essential to assess and address their needs. ED care plan is widely designed for acute disease-based approach (5). Thus, substantial resources are devoted to diagnose and treat acute chief complaints in EDs, but multi-morbidities and disabilities are usually ignored that may expose to short-term ED adverse events (3, 7). For instance, older ED users are more prone to longer ED and hospital stays, or higher hospital admission compared to younger ED users (3., 4., 5., 6., 7.). These three short-term ED adverse events are one of the major challenges facing EDs and, thus, require appropriate preventive actions.

In Quebec, a province of Canada, the “Program of Research on the Integration of Services for the Maintenance of Autonomy” (PRISMA-7) is the Ministry of Health and Social services' reference tool for the screening of frailty in older ED users (8). PRISMA-7 stratifies older adults into two risk groups for disabilities: low versus high-risk. It has a high sensitivity for identifying frailty in older community dwellers (9, 10). Recently, a systematic review underscored that PRISMA-7 has all characteristics for being a usable clinical tool in ED because it is a brief (i.e., <5 min) and simple tool (i.e., 7 clinical questions assessing health and functionality with close-end answers) providing a multidimensional assessment (10). PRISMA-7 prognostic value for the incident short-term ED adverse events in older ED users has never been examined. We hypothesised that a PRISMA-7 high-risk level for disabilities could be associated with the occurrence of short-term ED adverse events. This study aimed to examine 1) whether the PRISMA-7 high-risk level for disabilities was associated with the length of stay in ED and in hospital, and hospital admission; and 2) performance criteria (i.e., sensitivity, specificity, positive predictive value [PPV], negative predictive value [NPV], likelihood ratios [LR]) of the PRISMA-7 high-risk level for the length of stay in ED and hospital, and hospital admission in older ED users.

Methods

Population and design

Older patients who visited the ED of the Jewish General Hospital (Montreal, Quebec, Canada) between September 2017 and July 2018 were recruited in this observational and prospective cohort study. The inclusion criteria were age ≥ 75, unplanned ED visit, PRISMA-7 completed and agree to participate in the study. The exclusion criterion was concomitant participation in an experimental study. During the 10-month period of recruitment, 14,288 patients aged ≥ 75 visited the ED of the Jewish General Hospital. A total of 12,983 (90.9 %) met the selection criteria. Among these patients, 10,971 (84.5 %) were on a stretcher and 2,012 (15.5 %) were ambulatory patients.

Baseline assessment

The triage nurse at ED performed upon the arrival a standardized clinical assessment and collected information on age, sex, place of living before the ED visit including home, residence, hospital when patient was transferred from another hospital for an administrative reason (i.e., patient regularly followed at the Jewish General Hospital) and the hospital administrative health area (i.e., if the patient was living the administrative area of the Jewish General Hospital). The reasons for ED visit were collected and categorized into 5 sub-types: Organ failure defined as an acute organ decompensation; Mobility disorders defined as gait and/or balance impairment and/or fall with or without fall-related injuries; Neuropsychiatric disorders defined as delirium, dementia, and/or behavioral disorders; Cancer defined as a group of diseases involving abnormal cell growth with the potential to invade or spread to other parts of the body; And social issues defined as the absence of acute disease symptoms combined with an acute increase in the use of formal and/or informal home and social services, leading to an inability to cope at home. Furthermore, the Canadian ED Triage and acuity scale was performed to assess medico-surgical urgency levels ranged from level 1 defined as resuscitation (i.e., conditions that are threats to life or limb or pose an imminent risk of deterioration and require aggressive interventions) to level 5 defined as non-urgent (i.e., conditions that may be acute but non-urgent, as well as conditions that may be part of a chronic problem with or without evidence of deterioration) (11).

PRISMA-7 assessment

The triage nurse was in charge to collect the information for the filling of PRISMA-7 questions which are: 1) Are you more than 85 years old?; 2) Are you male?; 3) In general, do you have any health problems that require you to limit your activities?; 4) Do you need someone to help you on a regular basis?; 5) In general, do you have any health problems that require you to stay at home?; 6) In case of need, can you count on someone close to you?; and 7) Do you regularly use a stick, walker, or wheelchair to move around? The scoring method used by the PRISMA-7 is to assign 1 point to the answer yes and no points to the answer no. It stratifies inpatients into two disability risk levels: low (score 0 to 2) and high (score 3 to 7) risks (8).

Follow-up and outcomes

Dates and hours for ED visit, hospital admission and discharge were extracted from the administrative database of the patients' digital file. The length of stay in ED stays, expressed in hours, was defined as the delay between the hour and day of the ED visit and the hour and day of the ED discharge (either to hospital or place of living). The length of stay in hospital, expressed in day, was defined as the delay between the day of the ED visit and the day of the hospital discharge. A long length of stay was defined as being in the highest tertile of length of stay in ED (i.e., > 21 hours) and hospital (i.e., > 12 days) for patients on stretcher. For patients not on a stretcher the highest tertile of length of stay in ED was >7 hours)

Ethics

This study was conducted in accordance with the ethical standards set forth in the Helsinki Declaration (1983). The study was classified as a clinical quality-improvement program for older ER users' care plan by the Ethics Committee and the West-Central Montreal Health Review Office for quality programs of the Jewish General Hospital (Montreal, Quebec, Canada). Verbal inform consent were obtained for all participants following a systematic and standardized process used in the ED ward where the study has been performed. Participants or their legal representatives when appropriate, were informed that their medical information may be used for research purpose. If they disagreed, they informed the physician taking care of them and a note was recorded in their chart. The Ethic Committee of the Jewish General Hospital approved this process.

Statistics

The participants' baseline characteristics were summarized using means, standard deviations (SD) or frequencies as appropriate. Between group-comparisons were performed using unpaired t-test and Chi-square, as appropriate. P-values ≤0.001 were considered statistically significant because of multiple comparisons (Table 1). Multiple linear and Cox regressions examined the association of a PRISMA-7 high-risk level (used as independent variable) with the length of stay in ED and hospital, and in hospital admission (used as dependent variable with separated model for each variable) adjusted on participant's baseline characteristics. The performance criteria examined were sensitivity, specificity, PPV, NPV and LR. The elapsed time to discharge from ED and hospital by survival Kaplan-Meier curves and log-rank test was examined. P-values ≤0.05 were considered statistically significant. All statistics were performed using SPSS (version 24.0; SPSS, Inc., Chicago, IL).

Table 1.

Comparisons of baseline characteristics of participants visiting emergency department on stretcher or not (i.e., ambulatory patients) with a PRISMA-7 assessment (n=12,983)

	Patients on stretcher						P-value*
	Yes		P-value*		No	P-value*	Patients on stretcher versus patients not on stretcher
	Prisma-7 score (/7)			Prisma-7 score (/7)
	<3	≥3		<3	≥3		Prisma-7	Prisma-7
	(n= 5,789)	(n= 5,182)		(n= 1,375)	(n= 637)		score (/7) <3	score (/7) ≥3
Age (years), mean±SD	83.4 ± 6.0	85.7 ± 5.9	≤0.001	81.2 ± 5.0	83.6 ± 5.3	≤0.001	≤0.001	≤0.001
Female, n (%)	3745 (64.7)	2475 (47.8)	≤0.001	927 (67.4)	211 (33.1)	≤0.001	0.056	≤0.001
Localization before ED visit, n (%)
Hospital Health areas†	3124 (54.0)	2882 (55.6)	0.083	801 (58.3)	369 (57.9)	0.890	0.004	0.267
Place of living
Home	4152 (71.7)	3757 (72.5)	0.364	1254 (91.2)	581 (91.2)	0.995	≤0.001	≤0.001
Residence	817 (14.1)	801 (15.5)	0.047	25 (1.8)	18 (2.8)	0.146	≤0.001	≤0.001
Hospital‡	820 (14.2)	624 (12.0)	0.001	96 (7.0)	38 (6.0)	0.395	≤0.001	≤0.001
Reason for ED visit, n (%)
Organ failure¶	2914 (50.3)	2594 (50.1)	0.770	621 (45.2)	254 (39.9)	0.026	0.001	≤0.001
Mobility disorders§	946 (16.3)	869 (16.8)	0.547	355 (25.8)	180 (28.3)	0.249	≤0.001	≤0.001
Neuropsychiatric disorders#	868 (15.0)	727 (14.0)	0.152	218 (15.9)	111 (17.4)	0.376	0.424	0.021
Social issue‖	921 (15.9)	844 (16.3)	0.591	160 (11.6)	80 (12.6)	0.553	≤0.001	0.015
Cancer	140 (2.4)	148 (2.9)	0.152	21 (1.5)	12 (1.9)	0.558	0.045	0.157
Canadian ED Triage and acuity scale, n (%)
Level 1 — Resuscitation	80 (1.4)	43 (0.8)	0.006	0	0	-	0.001	0.021
Level 2 — Emergent	1549 (26.8)	1158 (22.3)	≤0.001	125 (9.1)	41 (6.4)	0.044	≤0.001	≤0.001
Level 3 — Urgent	3042 (52.5)	2905 (56.1)	≤0.001	507 (36.9)	257 (40.3)	0.135	≤0.001	≤0.001
Level 4 — Less urgent	1006 (17.4)	994 (19.2)	0.015	524 (38.1)	240 (37.7)	0.852	≤0.001	≤0.001
Level 5 — Non-urgent	112 (1.9)	82 (1.6)	0.162	219 (15.9)	99 (15.5)	0.825	≤0.001	≤0.001
PRISMA-7 score (/7), mean±SD	1.5±0.6	4.7±1.4	≤0.001	1.5±0.6	3.7±1.1	≤0.001	0.929	≤0.001
Length of stay in ED (hours), mean±SD	15.8±14.6	18.6±15.6	≤0.001	3.4±2.2	3.5±2.1	0.353	≤0.001	≤0.001
Hospital admission, n (%)	1686 (29.1)	1965 (37.9)	≤0.001	0	0	-	≤0.001	≤0.001
Length of stay in hospital (days), mean±SD	12.4±17.4	14.6±18.5	≤0.001	2.3±1.5	7.0±8.4	0.284	≤0.001	0.558

SD: Standard deviation; ED: Emergency department; PRISMA-7: “Program of Research on the Integration of Services for the Maintenance of Autonomy”; *: Comparison based on unpaired t-test or Chi-square, as appropriate; †: Patient living in the referent hospital administrative health area; ‡: Patients transferred from other hospital; ¶: Defined as an acute organ decompensation; §: Defined as gait and/or balance impairment and/or fall with or without fall-related injuries; #: Defined as delirium, dementia, behavioral disorders; ‖:Defined as the absence of symptoms of acute disease combined with an acute increase of the use of formal and/or informal home and social services leading to an inability to cope at home; P-value significant (i.e., ≤0.001 because of multiple comparisons) in bold

Results

Table 1 shows the comparisons of participant's baseline characteristics visiting ED on stretcher or not (i.e., ambulatory patients) with and without a PRISMA-7 score. Regardless of the stretcher condition, patients with a PRISMA-7 high-risk level were older (P≤0.001) with a lower proportion of females (P≤0.001) and their mean PRISMA score was higher compared to those with a low-risk PRISMA-7 level. The prevalence of level 2 of the Canadian ED triage and acuity scale was lower in patients on stretcher with a PRISMA-7 high-risk level compared to those with a PRISMA-7 low-risk level (P≤0.001), whereas the prevalence of level 3 was higher (P≤0.001). Patients on stretcher with a PRISMA-7 high-risk level had a higher length of stay in ED and hospital and hospital admission compared to those with a PRISMA-7 low-risk level (P≤0.001). Comparisons between patients on stretcher and ambulatory revealed that, regardless the PRISMA-7 score, patients on stretcher were older (P≤0.001), lived more frequently in residence (P≤0.001) and not at home (P≤0.001), came more frequently from hospital (P≤0.001) and had a higher prevalence of mobility disorders (P≤0.001). Furthermore, those with a PRISMA-7 low-risk level and on stretcher had a higher prevalence of cancer (P≤0.001), were more frequently classified levels 2 and 3 (P≤0.001) and less classified levels 4 and 5 (P≤0.001) and had a longer stay in ED and hospital as well as hospitalization (P≤0.001) compared to ambulatory patients. Patients with a PRISMA-7 high-risk level and on stretcher had a higher prevalence of organ failure (P≤0.001), were more frequently classified levels 2 and 3 (P≤0.001), less classified levels 4 and 5 (P≤0.001) and had a longer stay in ED (P≤0.001) compared to ambulatory patients.

Table 2.

Linear and Cox regression models showing the association of the length of stay in emergency department and hospital as well the hospital admission rate (dependent variable, separated model for each variable) with a PRISMA-7 high-risk level (i.e., score ≥3/7) (independent variable) adjusted on participants' baseline characteristics in participants visiting emergency department (n=12,983)

	Patients on stretcher
	Yes						No
	Linear regression*			Cox regression*†			Linear regression*			Cox regression*†
	β	[95%CI]	P-Value	HR	[95%CI]	P-Value	β	[95CI]	P-Value	HR	[95%CI]	P-Value
Length of stay in ED (hours) ‡	2.15	[1.58;2.73]	≤0.001	1.25	[1.16;1.37]	≤0.001	0.10	[−0.12;0.31]	0.375	1.23	[0.70;2.17]	0.479
Hospital admission‡		−		1.27	[1.19;1.36]	≤0.001		−			−
Length of stay in hospital (days) ‡	2.26	[1.04;3.49]	≤0.001	1.25	[1.10;1.41]	≤0.00l		−			−

ED: Emergency department; PRISMA-7: “Program of Research on the Integration of Services for the Maintenance of Autonomy”; HR: Hazard ratio; B: Coefficient of regression beta; *: Model adjusted on baseline characteristics of participants (i.e., age, sex, localization before ED visit, reason for ED visit and Canadian emergency department Triage and acuity scale level); †: To be in the highest tertile with a length of stay (for patients on stretcher >21 hours in Emergency department and >12 days in hospital; for patients not on stretcher >7 hours in Emergency department); ‡: Separated model for each dependent variable; P-value significant (i.e., ≤0.05) in bold

Regressions showed that a PRISMA-7 high-risk level was associated with increased length of stay in ED and hospital (Coefficient of regression β ≥2.15 with P−0.001 and Hazard ratio (HR)= 1.25 with P≤0.001) in patients on stretcher. No significant association was found with length of stay in ED in ambulatory patients (Coefficient of regression β =0.10 with P=0.375). A PRISMA-7 high-risk level was also associated with Hospital admission (HR=1.27 with P≤0.001) in patients on stretcher. All performance criteria were low (i.e., <0.78), regardless of the outcomes (i.e., length of stay in ED and hospital, and hospital admission) (Table 3). Finally, Kaplan-Meier distributions showed that the participants with a PRISMA-7 high-risk level were significantly discharged later from ED and hospital compared to those with a normal score (P≤0.001; Figure 1).

Table 3.

Performance criteria of a PRISMA-7 high-risk level (i.e., score ≥3/7) for long length of stay in emergency department and in hospital, and hospital admission rate

Criteria	Length of stay		Hospital admission
	Emergency department*	Hospital†
Sensitivity	0.36	0.13	0.38
Specificity	0.08	0.02	0.08
Positive predictive value	0.32	0.36	0.34
Negative predictive value	0.68	0.71	0.77
Likelihood ratio of positive test	−5.32	−0.21	−5.67
Likelihood ratio of negative test	−0.22	−0.14	−0.20

PRISMA-7: the seven questions “Program of Research on the Integration of Services for the Maintenance of Autonomy”; *: Length of stay used as a discontinuous variable defined as the highest tertile of lengths of stay (i.e., >21 hours); †: Length of stay used as a discontinuous variable defined as the highest tertile of lengths of stay (i.e., >12 days)

Figure 1.

Kaplan-Meier estimates of the probability of long emergency department (a) and in hospital (b) stays from the PRISMA-7 score in participants visiting emergency department on stretcher or ambulatory (n=12,983)

Discussion

The findings showed that a PRISMA-7 high-risk level was associated with incident short-term ED adverse events including a long length of ED and hospital stay and hospital admission, in older ED users on stretcher. However, PRISMA-7 performance criteria for risk for these ED adverse events were low, regardless of the criterion examined.

To the best of our knowledge, this is the first time that prognostic value of the PRISMA-7 high-risk level is assessed to predict short-term ED adverse events. The association between the occurrence of these adverse events and PRISMA-7 high-risk level was expected because PRISMA-7 is a recognized tool to screen frailty (7., 8., 9., 10.). Frailty may be defined as a medico-social condition that exposes to disabilities and incident adverse events due to a low level of physiological reserves (3, 12). The accumulation of severe chronic morbidities and physiological aging reduces physiological reserves contributing to frailty (12). Frailty has previously been associated with adverse events during hospitalization such as increased risk for ED visits and long length of hospital stay (3, 4). No association was found with ambulatory older ED users because this category of ED users is in better health condition compared to those on stretcher as shown by our results. The reason for ED visit was both less acute and less urgent compared to patients on stretcher, as indicated by the lower prevalence of organ failure and the lower ED triage score. Identifying older ED users at risk for short-term adverse during an ED visit is important because it may help to guide clinical practice and service improvement in ED care (5). EDs in North America are under duress because of overcrowding, delays and diversions, which increase to epidemic proportions (7). The aging of Canadian's population amplifies the magnitude of this situation because of two closely related issues. First, older ED users are the fastest increasing group of ED users. Second, they have complex and specific needs that are difficult to address (5, 7). Because EDs are configured for disease-oriented and episodic model rather than chronic multi-morbidities and disabilities, the short-term adverse outcomes are one of the main challenges that EDs face (5). Assessing the specific needs of the growing number of older ED users is a priority to reduce the occurrence of the short-term ED adverse outcomes. Our results highlight that PRISMA-7 could be a clinical tool to achieve this goal. Little is known about how to implement such tools in practice. A recent systematic review concluded that risk stratification tools should be quick, simple, easy to use for a good compliance (i.e., ≥75%) of ED staff and that PRISMA-7 had all these characteristics (13). Our results confirmed this statement because 90.9% of older users visiting ED had an assessment with PRISMA-7.

The second main finding of our study is that performance criteria of the PRISMA-7 high-risk level is low, regardless of the short-term ED adverse event considered. This result suggests that PRISMA-7 cannot be used as a prognosis tool for long length of ED and hospital stay as well as hospital admission during an ED visit. The low number of items may explain in part this result. Indeed, we previously demonstrated that the number of items of a prognostic tool may influence its predictive value (14).

The systematic assessment using PRISMA-7 of ED users ≥ 75, the huge number of them visiting the ED of Jewish General Hospital and the appropriate design to achieve the research objectives are the strengths of our study. However, some limitations need to be underscored. First, the recruitment has been performed in one centre only, that limits the external validation of our results. Second, although adjustment for clinical characteristics influencing the studied short-term adverse events was performed, there are still potential residual confounders which may influence the length of stay.

In conclusion, even if PRISMA-7 is associated with incident short-term ED adverse events, its performance to predict the risk for these events are low underscoring that it cannot be used has a prognostic tool for this purpose. Further research is required to compare PRISMA-7 with other existing prognostic tools.

Authors contribution

OB: Principal investigator, study conception and design, obtaining funding and drafting first version of the manuscript and approval of the final manuscript; MA: Co-investigator, study conception and design, drafting first version of the manuscript and approval of the final manuscript; CLP: design study, analyzing data, revising the manuscript critically for important intellectual content and approval of the final manuscript; KG, EM and CV: recruitment of participants, study management, or revising the manuscript critically for important intellectual content and approval of the final manuscript.

Consent for publication

All authors gave their approval of the final manuscript and for publication.

Competing interest

The authors OB, KG, CV, EM, MA and CPL report no disclosures relevant to the manuscript. The funding sources have no involvement in study design; in the collection, analysis and interpretation of data; in the writing of the manuscript; and in the decision to submit the article for publication.

Availability of data and materials

Data and materials are available on request. Request should be sent to the corresponding author: Olivier Beauchet, MD, PhD; Department of Medicine, Division of Geriatric Medicine, Sir Mortimer B. Davis — Jewish General Hospital, McGill University, 3755 chemin de la Côte-Sainte-Catherine, Montréal, QC H3T 1E2, Canada; e]olivier.beauchet@mcgill.ca. All requests needs a cover letter explaining the objective, justification and the referent ethic committee.

Trial registration

SimplePara>NCT04038983.

Funding

The study was financially supported by private donations (Mrs. Abbigail Shaw and Mr. Davis Schiller), the RUISSS McGill Centre of Excellence on Longevity and the Foundation of the Jewish General Hospital (Montreal, Quebec, Canada). The funding sources have no involvement in study design; in the collection, analysis and interpretation of data; in the writing of the manuscript; and in the decision to submit the article for publication.