Article Text
PDF
Abstract
Background NHS England’s ‘Enhanced Health in Care Homes’ specification aims to make the healthcare of care home residents more proactive. Primary care networks (PCNs) are contracted to provide this, but approaches vary widely: challenges include frailty identification, multidisciplinary team (MDT) capability/capacity and how the process is structured and delivered.
Aim To determine whether a proactive healthcare model could improve healthcare outcomes for care home residents.
Design and setting Quality improvement project involving 429 residents in 40 care homes in a non-randomised crossover cohort design. The headline outcome was 2-year survival.
Method All care home residents had healthcare coordinated by the PCN’s Older Peoples’ Hub. A daily MDT managed the urgent healthcare needs of residents. Proactive healthcare, comprising information technology-assisted comprehensive geriatric assessment (i-CGA) and advanced care planning (ACP), were completed by residents, with prioritisation based on clinical needs.
Time-dependent Cox regression analysis was used with patients divided into two groups:
Control group: received routine and urgent (reactive) care only.
Intervention group: additional proactive i-CGA and ACP.
Results By 2 years, control group survival was 8.6% (n=108), compared with 48.1% in the intervention group (n=321), p<0.001. This represented a 39.6% absolute risk reduction in mortality, 70.2% relative risk reduction and the number needed to treat of 2.5, with little changes when adjusting for confounding variables.
Conclusion A PCN with an MDT-hub offering additional proactive care (with an i-CGA and ACP) in addition to routine and urgent/reactive care may improve the 2-year survival in older people compared with urgent/reactive care alone.
- GENERAL PRACTICE
- Health policy
- PRIMARY CARE
- Quality improvement
- Information technology
Data availability statement
No data are available.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
NHS England’s ‘Enhanced Care in Care Homes’ (EHCH) model is implemented in widely differing ways by different primary care networks, with outcomes being difficult to assess.
WHAT THIS STUDY ADDS
We evaluated an Older Peoples’ Multidisciplinary Hub in a primary care network in Plymouth, comparing the survival of care home residents who received proactive care in addition to routine/urgent care (intervention) versus routine/urgent care only (control). At 2 years, care home residents in the intervention group were more likely to be alive compared with control (48% vs 9%), representing a 39.6% absolute survival advantage (p<0.001).
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
This model of care is generaliseable and scalable to all primary care networks looking to develop Older Peoples’ Multidisciplinary Hubs to fulfil their EHCH contractual requirements and (although out of scope for this quality improvement report) has been extended to support proactive care for older people in their own homes.
Introduction
NHS England’s Long-Term Plan1 includes the ‘Enhanced Health in Care Homes’ (EHCH) model,2 which offers proactive care for residents in care homes, many of whom live with frailty, dementia and multimorbidity. In 2020, primary care networks (PCNs) were contracted to deliver EHCH, which specified that all care homes should receive weekly proactive multidisciplinary ‘care home rounds’ based on ‘the principles and domains of a comprehensive geriatric assessment (CGA)’.2
A CGA is ‘a multidimensional, multidisciplinary process that identifies medical, social and functional needs, and the development of an integrated/coordinated care plan to meet those needs’.3 CGA is beneficial in acute and hospital-at-home settings, with patients more likely alive and in their own homes at 6–12 months.4 5 Community-based CGA may also improve physical function and independence,4 6 reduce hospital admissions7 and increase survival.4 8 However, there is a lack of evidence on the impact of proactive CGA for care home residents.9
Meeting the EHCH contractual requirements has been challenging. Rising patient demand, complexity and workforce shortages have challenged capacity.10 11 Furthermore, there is a skills gap; primary care teams have limited experience with delivering a CGA, and specific training was not provided as part of EHCH. Lastly, there is no guidance or specification for how primary care-led CGA should be structured.12
Since Autumn 2018, Pathfields Medical Group, a single-practice PCN, developed a dedicated Older Peoples’ Hub with an MDT supporting older people who lived in care homes or were housebound and living with frailty. Initially, it offered urgent care, but since May 2019, the Hub began offering proactive care. An information technology-assisted CGA (i-CGA) tool was developed within SystmOne, enabling structured, check-listed, high-quality assessments, with minimum administrative burden. Weekly proactive care clinics assessed complex patients, offered i-CGA and provided case management where needed. Online supplemental table 1 describes i-CGA and the Hub’s proactive care elements.13
Supplemental material
Evaluation of proactive care and i-CGA commenced in 2021, and by March 2024, we published research confirming i-CGA improved the quality of advanced care planning (ACP), compared with routine NHS care and may improve unplanned admissions.14–16 In addition, we noted another unexpected finding; patients receiving i-CGA appeared to have reduced mortality.14–16 However, the study was underpowered to conclude this, due to a small sample size. Accordingly, we recruited more subjects, and in this report, we present the findings looking at the headline outcome of mortality at 2 years.
Method
This quality improvement project was set up on 1 March 2019 in Pathfields Medical Group PCN in Plymouth, England, and we report outcomes until 30 September 2022.
All PCN-registered permanent residents of older people care homes and Pathfields-registered patients discharged from hospital to care home were included in the assessment under a current UK arrangement known as ‘discharge-to-assess’ (D2A). Patients were excluded if they left the PCN and registered with another surgery or returned home during follow-up.
For analysis, we treated the data as a non-randomised crossover cohort study. Two groups were defined:
Control group: They received (if needed) Hub-coordinated urgent care, otherwise routine care and, to a variable extent, may have included some additional interventions described in online supplemental table 1.
Intervention group: During the study, residents moved progressively from the control to the intervention group on completion of a proactive i-CGA cycle, comprising all activities in online supplemental table 1.
Residents were not randomised; i-CGAs were prioritised based on clinical need and structured around dedicated care home sessions. The intent was to offer all residents proactive care. In our dataset, many patients started on routine/urgent care only (control), before receiving additional proactive care (intervention).
The primary outcome of interest in this analysis was survival/mortality. Frailty diagnosis (categorised by mild, moderate and severe) was made using the Pathfields Tool,17 a case-finding tool built in primary care IT. It invited clinicians to record a frailty diagnosis on saving the record following a patient encounter annually. Clinician diagnosis was made by combining the Rockwood Clinical Frailty Scale18 and longitudinal clinical knowledge of the patient.
Statistical analysis
To control any survival bias, time-dependent statistical analysis was performed in R19 based on the time each resident spent in the control and intervention groups.
Differences in mortality between the two groups were tested using Cox proportional hazards regressions, with group (control/intervention) as a time-dependent variable. This type of regression is specifically used for analysing survival data. The time-dependent variable component allows analysis of subjects with variables that change over time, for example, changing from the control to the intervention group. The model accounts for the time patients spent in each group, providing a more robust test and reducing survival bias.
Cox proportional hazards model analysis comprised the following steps:
A simple model where the group (control/intervention) was the only predictor variable.
Multiple-predictor model controlling for additional potential confounding variables.
Kaplan-Meier curves were plotted for the data. Both frequentist and Bayesian versions of analyses were conducted. We report several statistics from frequentist regression models. The Wald statistic (z) and p-value are reported together to indicate statistical significance (p<0.05). Positive Wald values indicate a positive relationship between variables and negative values an inverse relationship. The HR quantifies how much more/less likelihood the intervention versus control groups have of dying during the study period. An HR of 2 or 0.5 would mean double or half the likelihood, respectively, of mortality within 2 years.
There are no Bayesian proportional hazards regression packages in R that allow for time-dependent variable analysis. Therefore, Bayesian proportional hazards regressions were conducted on a simplified version of the data, where the group was a three-level variable: intervention, control or both (ie, instead of patients being coded by time spent in each group). Evidence for effects was tested using Bayesian 95% credible intervals (Bayesian versions of CIs) for each coefficient, with intervals discrete from zero providing evidence (similar to p<0.05). The control group was used as the reference category, and the intervention (never in control) and intervention (control first then intervention) groups were tested against this. The coefficient from the Bayesian coefficient (BC) is similar to the frequentist Wald statistic, where positive and negative values indicate a positive or inverse relationship, respectively.
Results
At the time of evaluation, 429 eligible patients had completed follow-up (figure 1).
CONSORT flow diagram of patients included in this quality improvement project.
Characteristics in table 1 are presented according to the final group residents were in and include all participants from each group, regardless of whether they survived or died by the end of follow-up.
Baseline characteristics for all residents in the quality improvement project, grouped accordingly by the end of follow-up: control or information technology-assisted comprehensive geriatric assessment (i-CGA) (all those who underwent i-CGA). Also presented separately according to whether they had an i-CGA and were never in the control group and those who started in the control group before receiving an i-CGA and thus later moved to the i-CGA group
Mortality overall
The Kaplan–Mieer plot (figure 2) indicates at 2 years, 48.1% and 8.6% survival in the intervention and control groups, respectively (39.6% absolute risk reduction (ARR); 70.2% relative risk reduction (RRR); and number needed to treat (NNT)=2.5); HR=0.30, z=−9.08, p<0.001, BC (intervention only)=−1.14 (95% CI: −1.66,–0.76), BC (control then intervention)=−1.24 (95% CI: −1.49,–0.98).
Kaplan–Meier survival (days) for the information technology-assisted comprehensive geriatric assessment and control groups. Mortality at 1 and 2 years is indicated by dotted vertical lines.
Mortality after controlling for other variables
Table 1 shows all-cause mortality during the pre-vaccination period of the pandemic (in our locality, this was from 17 March to 31 December 2020). This demonstrates the intervention group had a higher all-cause mortality than the control group during this timeframe. To explore the mortality differences further, we ran single-predictor models for any health conditions that had a higher prevalence in the control group (ie, might increase mortality in this group) versus either the intervention (never in control) or intervention (control first then intervention). From these, female sex was significantly associated with lower mortality at 2 years (HR=0.66, z=−3.14, p=0.002), while the presence of heart failure (HR=1.37, z=2.00, p=0.046) and being in a nursing home (HR=2.39, z=4.16, p<0.001) or dual nursing/residential home (HR=1.88, z=3.43, p<0.001) significantly increased mortality at 2 years. Moderate frailty (HR=0.68, z=−2.67, p=0.008) and mild frailty (HR=0.59, z=1.90, p=0.058) were associated with lower mortality than severe frailty, although this was only significant for moderate frailty, possibly due to the lower n in the mild frailty group (see table 1). Other variables (diabetes, chronic obstructive pulmonary disease, cancer) were not significant (p>0.05). Therefore, we ran a multiple predictor model for group, which controlled for frailty level, male sex, heart failure and institution type.
The i-CGA group continued to be associated with reduced mortality after controlling for these other variables (see online supplemental table 2). More severe frailty and being in a nursing home or a dual nursing and residential home (compared with residential only) continued to be significantly associated with higher mortality. There was mixed evidence of heart failure being associated with higher mortality, after controlling for these other variables, and there was no longer a significant effect of sex (see online supplemental table 2). These additional variables mean there are 72 groups that can be compared for ARR—too many to report in full. For example, in female patients with mild frailty, without heart failure in residential-only settings, there was 64.8% and 21.3% survival in the intervention and control groups, respectively, at 2 years (43.5% ARR; NNT of 2.3). By comparison, for male patients with severe frailty and heart failure in nursing-only settings, there was 10.7% and 0.04% survival in the intervention and control groups, respectively (10.7% RRR; NNT of 9.4).
Supplemental material
However, if additional variables are held at their average value (mean sex and heart failure across all patients and mode of institution and frailty level), there was a 44.3% and 5.48% survival in intervention and control groups, respectively (ARR 38.8%; RRR 71.9%; NNT 2.6; HR=0.28, z=−8.92, p<0.001, BC (intervention only)=−1.06 (95% CI: −1.55,–0.67), BC (control then intervention)=−1.30 (95% CI: −1.61,–1.03).
Discussion
Summary: Survival benefit seen in the intervention group
After adjusting for confounding variables, care home residents receiving urgent and proactive care with i-CGA through an Older Peoples’ MDT Hub experienced significant survival benefits compared with residents who received urgent care alone.
We offer several putative mechanisms for improved survival. First, the activities taking place in the intervention group (eg, optimisation of long-term conditions, medication and personalisation of care (eg, relaxing hypertension targets in people prone to falls)) could improve overall health and reduce predisposing risk of conditions associated with increased mortality (eg, delirium and falls).20 21
Second, i-CGA improves the efficiency and effectiveness of proactive care. It raises warnings when high-risk medication is about to be prescribed to older people with frailty, preventing potentially harmful prescribing. It also allows clinicians to rapidly sift through the entire care home population, targeting patients on high-risk medications for priority review. Data in online supplemental table 1 show our prescribing rates for high-risk drugs are consistently lower than those of the published literature (antimuscarinics 1.1% vs 4.9%; opiates 8.9% vs 22.4%; tricyclics 1.7% vs 3.9%; and anti-psychotics 13.3% vs 21%).22 23 Furthermore, it offers better quality measures such as improved ACP in the intervention arm (see baseline characteristics).14–16
The third reason is a heavy focus on continuity, which has been shown to improve survival.24 This was achieved in two ways: first, having a dedicated team for older people improves relational continuity; and second, i-CGA enables informational continuity - all patients automatically receive care and support plans, which are also shared with other healthcare organisations in the locality. This could improve care if the patient becomes more unwell and urgent/emergency care is needed.
Finally, our earlier evaluation15–17 was underpowered due to the small sample size but showed a reduction in hospitalisation in the intervention arm. This is important as hospitalisation is also associated with delirium, deconditioning and functional decline so a reduction in admissions may also improve survival. Further work is underway evaluating hospitalisation in this larger cohort.
Taken together, these are feasible mechanisms that allow the intervention population to become more robust and less likely to experience acute insults, thus reducing the likelihood of deterioration and death.
Strengths and limitations
This service evaluation used exploratory retrospective analysis of routinely collected data; thus, our methodology is open to bias, most obviously selection bias (eg, de-prioritising residents with terminal diagnoses for proactive i-CGA). However, the user-friendly ACP documentation process during i-CGA meant that staff reported frequently choosing to use the tool in end-of-life situations. This is evidenced in table 1, where the i-CGA group had a higher number of patients with advance care plans. This was particularly important given the higher all-cause mortality in the intervention group (18% vs 13% in the control group) during the pre-vaccination period of the COVID-19 pandemic, a time when excess deaths were reported from care homes.25
We controlled for potential survival bias and differences between the two groups using Cox proportional hazards regressions, with group (control or intervention) as a time-dependent variable. This type of regression is specifically used for analysing survival data and accounts for the time patients spent in each group (reducing survival bias). Additionally, we controlled for the effects of potentially confounding variables (in this case, frailty severity, sex, heart failure and type of institution).
Comparison with existing literature
There is currently limited comparable data published on the impact of proactive CGA for care home residents, a recognised gap in evidence-based practice.9
Previous findings looking at the effect on mortality of complex community-based interventions, or CGA specifically, have been mixed. A Cochrane review demonstrated survival benefits following hospital-based CGA,4 and a 20% reduction in mortality following proactive community-based CGA was seen in older adults in Germany.8 However, most individual CGA-based studies have not shown clear survival benefits. One key issue when looking at mortality is the selection of patients and the duration of follow-up. Many CGA-intervention studies target more frail individuals, where high mortality rates may provide little time for the treatment effect to be realised. The converse applies in less frail individuals where low mortality means longer follow-up required to detect differences.
Implications for research and practice
If the processes outlined in the TIDier checklist (online supplemental table 1) are followed in their entirety, this model of care is generalisable and scalable to all primary care networks looking to develop Older Peoples’ Multidisciplinary Hubs to fulfil their EHCH contractual requirements and (although out of scope for this quality improvement report) has been extended to support proactive care for older people in their own homes.
To build on this evidence, the authors will conduct further analysis of hospitalisation during this and future periods using this cohort. We also propose a multi-site study using these interventions for residents in care homes, conducted outside of the pandemic, using a wider range of important outcome measures including patient-reported outcome measures (eg, quality of life, depression), healthcare outcomes (eg, falls/fractures, delirium episodes), healthcare utilisation metrics, a health economic analysis, qualitative interviews (with patients, families, health and social care staff) and incorporation of patient and public involvement.
Data availability statement
No data are available.
Ethics statements
Patient consent for publication
Ethics approval
Not applicable.
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors All authors were involved in one or more of the planning, conduct of the service and reporting of the work described in the article. The lead author is responsible for the overall content and accepts full responsibility for the work. DA, as the lead author, attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.
Funding Adam Gordon is an NIHR Senior Investigator and is part funded by the NIHR Applied Research Collaboration East Midlands (ARC-EM). Stuart Spicer was funded and supported by the National Institute for Health Research Applied Research Collaboration South West Peninsula (NIHR PenARC). Suzy Hope was supported by the National Institute for Health and Care Research Exeter Biomedical Research Centre and National Institute for Health and Care Research Exeter Clinical Research Facility. The views expressed are those of the authors and not necessarily those of the NIHR or the Department of Health and Social Care.
Competing interests The IT-assisted Comprehensive Geriatric Assessment (i-CGA) is owned by Target Health Solutions (THS, a company that enhances primary care IT). DA and JB are directors in THS.
Patient and public involvement Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.