Prioritization of indicators of the quality of care provided to older adults with frailty by key stakeholders from five canadian provinces

Background To meet the needs of older adults with frailty better, it is essential to understand which aspects of care are important from their perspective. We therefore sought to assess the importance of a set of quality indicators (QI) for monitoring outcomes in this population. Methods In this mixed-method study, key stakeholders completed a survey on the importance of 36 QIs, and then explained their ratings in a semi-structured interview. Stakeholders included older adults with frailty and their caregivers, healthcare providers (HCPs), and healthcare administrators or policy/decision makers (DMs). We conducted descriptive statistical analyses of quantitative variables, and deductive thematic qualitative analyses of interview transcripts. Results The 42 participants (8 older adults, 18 HCPs, and 16 DMs) rated six QIs as more important: increasing the patients’ quality of life; increasing healthcare staff skills; decreasing patients’ symptoms; decreasing family caregiver burden; increasing patients’ satisfaction with care; and increasing family doctor continuity of care. Conclusions Key stakeholders prioritized QIs that focus on outcomes targeted to patients and caregivers, whereas the current healthcare systems generally focus on processes of care. Quality improvement initiatives should therefore take better account of aspects of care that are important for older adults with frailty, such as having a chance to express their individual goals of care, receiving quality communications from HCPs, or monitoring symptoms that they might not spontaneously describe. Our results point to the need for patient-centred care that is oriented toward quality of life for older adults with frailty. Supplementary Information The online version contains supplementary material available at 10.1186/s12877-022-02843-9.


Additional file 1.
Scoping review on the current healthcare services and models care offered to frail seniors Objective This scoping review aimed answering the following research questions: (1) what are the current healthcare services and models care offered to frail seniors? (2) What are the healthcare resource utilization and care outcomes for frail seniors? and (3) What factors influence the healthcare resource utilization and care outcomes of frail seniors? We used a recognized scoping review methodology to answer these questions, 1 2 and reviewed three types of reports about Canadian frail seniors: reports evaluating the impact of an intervention on the clinical quality of FS care (Type-I), reports describing an innovative intervention or model to improve FS care without any formal evaluation (Type-II), and reports describing the quality of care across various FS cohorts (Type-III). For each of the three report types, we identified and extracted indicators of the clinical quality of frail senior care.

Types of reports
We included reports targeting Canadian FS of any type and format: empirical studies of any method (quantitative, qualitative or mixed methods) or design (descriptive, case study, before-and-after, randomized trials), national and provincial quality-related projects, programs, strategic plans, or initiatives that targeted FS. We also included reports on healthcare resource utilization and healthcare outcomes from administrative databases of FS in Canada. If a report covered both the Canadian population, and a population from another country, then it was included provided that the information (data or discussion) on Canadian FS could be extracted. We included reports published in French and/or English.

Types of participants
We included reports on FS aged 65+ years, their family caregivers, and/or their healthcare providers. For the purpose of the review, the participants studied in the reports were defined as being 'frail', (1) if they were described as frail, and the authors justified the use of the term 'frail' (e.g. by citing a frailty index), (2) if the description of participants provided in the report allowed us to class them as being frail using the Canadian Study of Health and Aging (CSHA) Clinical Frailty Scale, 3 or if two or more domains of the Edmonton Frail Scale 4 were described as health problems of the participants -meeting one scale or the other was required for inclusion, (3) if participants were described as living in long-term care facilities or palliative care home (hospices), or (4) if the participants were clearly identified to be at the end-of-life, terminally ill, or in palliative care. If a report covered both frail and not frail people, it was included provided that the information on frail people could be extracted. Similarly, if a report covered both seniors and non-seniors, we included it if we could extract information (data or discussion) about frail seniors.

Types of clinical quality of care indicators
We only included reports that discussed at least one clinical quality indicator measure. Following the Agency for Healthcare Research and Quality taxonomy, 5 we searched for reports on the clinical quality of care of FS measured at levels of healthcare structure (macro), resources use and costs, process of healthcare services, and healthcare patient/caregiver/provider outcome. We also searched for qualitative information to identify potential reasons for variations in healthcare resource utilization and outcomes across patient cohorts, settings of care, or provinces and deepen our understanding of current healthcare services and models of care.

Search strategy
The population of the "frail seniors" can be described in many ways, with frailty having many signs and symptoms. The clinical characteristics of frailty are varied, and there are several different scales created to diagnose, describe or delineate frailty. At first, we tried identifying this patient population in the medical literature using clinical characteristics, but it resulted in a hugely sensitive search, encompassing most of the discipline of gerontology and nearly every description of interactions between the elderly and the medical system. Hence, to find articles describing models of care or best practices for the population as a whole, we chose to combine keywords indicative of frailty (frail*, vulnerab*, "at risk", "high risk", high-risk, "low function", dependent) near to keywords indicative of older age (older, elder*, senior*) in the titles or abstracts. This resulted in a highly targeted set of articles that discussed the population more directly. In addition to finding the articles which directly address the frail senior population, three further stipulations were used to select articles of interest. Firstly, currency was important, so the results were limited to articles published in the last 5 years. Next, any articles directly addressing the Canadian population where selected -any mention of Canada or any of its provinces or territories was used for this filter.

Electronic sources of data
We searched the following academic databases (2009-December 14 th , 2014): Ageline, CINAHL, Cochrane Library, DARE, EMBASE, PsycINFO, Sociological Abstracts, PubMed, Ovid MEDLINE, and Ovid AMED. An experienced information scientist combined keywords from controlled vocabulary and free text to optimize the identification of relevant reports, and adapted this strategy for each database (Appendix 1). A second information scientist peer reviewed these search strategies. We also consulted with co-investigators, a geriatrician and a researcher specialized in the care of frail seniors to improve the specificity of the strategy.

Other sources of data
We also searched the reference lists of primary articles and reviews, and ISI Web of Science and Google Scholar (in July 2015) for studies citing the most relevant reports included.

Selection of reports
The research team initially met to discuss decisions surrounding study inclusion and exclusion. Then, a calibration exercise with a random sample of approximately 100 reports ensured a proper training of reviewers, iteratively using 10 reports each time, until they reached a minimum level of 80% agreement. Two reviewers (among MAL, MBB, MM, NC, MAC, VVT) then read titles and abstracts of reports identified through the search strategy to assess which reports met the inclusion criteria. The team conducted a calibration exercise again before selecting reports based on the full text of those retained. After calibration, they were assessed by two reviewers to ensure they met the inclusion criteria. Disagreements regarding study inclusion were resolved by discussion between the two reviewers and when disagreements occurred, a third reviewer determined final inclusion.

Data extraction and management
For extraction, we used a web-based standardized data extraction form supported by specialized software (DistillerSR, Evidence Partners). We extracted data relative to the report (e.g. authors, date of publication), to the study (e.g. type of study, scope), and to the content (e.g. participant characteristics, description of intervention, impact, covariates, whether the report covers a 'hot topic' -defined as a theme of actuality). We categorized the included intervention (i.e. health, professional, financial, or regulatory interventions) according to the taxonomy developed by the Effective Practice and Organisation of Care editorial group of the Cochrane collaboration. 6 Extraction was initially iterative with continuous data extraction and updating of the data extraction form. Again, a calibration exercise with a random sample of approximately 20 reports ensured proper training of reviewers: two persons independently extracted data from the first five included studies and met to compare their results and determine whether their approach to data extraction was consistent with the research question and purpose. They extracted 5 reports at a time, iteratively, until they reached a minimum level of 80% agreement. Following this calibration, a first person completed the extraction of the remaining reports, a second person verified it, and when there was disagreement among them, a third senior reviewer (MM) made the final decision on the extraction or referred to the project leader (AMCG) to make a decision.

Quantitative data
We prepared a descriptive numerical summary of the characteristics of the included reports (e.g. number of studies included, type of author, language, province, targeted audience). We grouped and pooled data in tables according to the predefined Types of report (I-impact of intervention; IIdescription of intervention; III-comparison of various FS cohorts), type of setting, patient characteristics, province, and study design. We estimated frequency of each variable using SAS (version 9.4). For the Type-I reports evaluating the impact of an intervention or Type-II that described an intervention, we presented the clinical quality indicator (CQI) targeted by the intervention, a short description of the intervention, the categories of intervention and the impact of the intervention (statistically significant improvement, deterioration or no significant effect). For the Type-III reports comparing various FS cohorts, we presented the CQI targeted in the report, the covariates studied and their association to the CQI.

Qualitative data
Open-text fields of the extraction form were analyzed qualitatively using an inductive thematic analysis approach. We then articulated the findings, first through the three report types that correspond to the environmental scan aims, and then through sub-themes that emerged during extraction and coding. Lastly, the research team met to consider the meaning of the findings as they relate to the aims, and discuss implications for future research, practice, and policy.

Search results
Using the academic databases we found 453 reports describing current healthcare services and outcomes to be considered for inclusion to the review (Figure 1.1). The grey literature search identified 580 further references, and 17 additional references were identified from the other sources. After removal of duplicates, 1004 references were retained for consideration for inclusion in the review.

Study selection
We included 93 studies: 22 studies of the impact of interventions to improve the quality of healthcare for FS, 35 describing an intervention (without a formal evaluation of their impact), and 42 comparing indicators of the quality of health care between various patient cohorts. Three of the studies comparing patient cohorts were also comprised among the intervention studies. Reasons for excluding reports comprised: not being written in French and/or English (N= 79), not covering the Canadian health care system (N = 599), not targeting seniors (N=604), not targeting frail seniors (n=850), and/or not being one of the three types of reports included (N= 117).

Description of the reports
Most of the included reports were published in 2012 (N=27), followed by 2013 (N=18) and 2014 (N=14). They were more frequently authored by researchers (N= 86) and/or clinicians (N= 66), and the authors were often based in universities (N= 82). Most reports were written in English (N= 85), whereas a small number were in French (N= 7); only one was written in both English and French. The conclusions of the included reports more often applied to the provincial level (N= 37), but some covered national (N= 11), regional, and/or local levels (N= 45). Most of the included reports presented the results of an empirical study, which was defined here as new data collection or secondary analysis (review) (N= 86) (Table 1.1). Among the reports of an empirical study, most presented the results of quantitative descriptive studies without a control group (N=49/93), 7 presented randomized trials, 15 non-randomized studies with control group, 22 qualitative studies and 1 presented a literature review.
Ontario and Quebec were the provinces most often the focus of the work in the included reports (N=37 reports), followed by British Columbia, Nova Scotia, and Alberta (N=11 each). More reports covered settings of care outside of hospitals (N=70): most commonly in long-term care (N=26) and home care settings (N=21). Several reports also covered hospital settings (N=27), more often Geriatric care units and Emergency departments (N= 5). Most report targeted frail seniors (N=75/93), several targeted healthcare providers of diverse professions (N=25), and a few targeted caregivers and decision makers.
Frail seniors were most often recognized as being 'frail' through a classification by our research team using the description of participants provided in the report, by checking if they met the CSHA clinical frailty scale 3 or Edmonton Frail scale 7 (Table 1.2). This strategy allowed identifying 83% of the included reports. Only for 26/75 (35%) reports did the authors of the report cite a frailty index to justify that they classified participants as being 'frail'. Among the reports in which the participants were recognized as 'frail', based on a classification by the research team, several frailty domains were recognized by the team.
Among the studies that targeted FS, some reported the mean age of the FS included (N= 45/93 reports; mean = 81 years old, range = 72-87), others their median age (N=2/93 reports; median = 85; range = 85 -86). Some reported the proportion of women (N =47/93 reports; mean = 67% women; range 0-100%). We also extracted if the report included mostly women (N= 40/93 reports), mostly men (N = 7/93), or if this information was not available. None of the included studies reported the proportion of FS belonging to a specific population group (e.g. Asian, Black, Filipino, Latin American, Aboriginal, or Caucasian). Only two reports described some of their frail senior participants as immigrants; the majority did not provide any information on the citizenship of the frail senior participants. Among the reports that targeted FS participants, most targeted FS living in a community (N=24/93 report), but several also targeted FS living alone (N=23), or living with a partner or caregiver (N=21). This information was lacking from many of the included reports (N=39) (Table 1.9). Most of the included reports did not specify if the targeted seniors were living in rural or urban settings (N=61/93 reports) -10 targeted FS seniors living in rural settings, and 3 FS living in urban settings.
In the reports that targeted caregivers of FS, we did not find any information on the proportion actually living with FS.
The included reports covered several targeted 'hot topics' (Table 1.3).

Impact of interventions to improve FS care (Type-I)
Among the interventions that were formally evaluated in an experimental study (Type-I), there were health interventions (i.e. interventions targeting the patient's health) (N=14), organizational interventions, (N=6), professional interventions (N=3), and one regulatory intervention (Table 1.4-A). Among those evaluating a health intervention, 11 reported a preventive intervention, five reported treatment interventions and one reported a screening intervention. Among the preventive interventions, exercise was the intervention most studied (N=4) and it generally improved the targeted clinical quality indicators, namely physical functioning, self-efficacy, cognitive performances, physical capacities (gait, balance, grip strength), and quality of life. [8][9][10][11] All the studied organizational interventions showed some benefits, 12-17 either on patient-level outcomes (e.g. increased empowerment or satisfaction with care, reduced unmet needs), healthcare provider outcomes (improved skills), resource utilization (e.g. increased number of home visits, reduced length of hospital stay, reduced ER use), or structure (institutional practices), although one also had negative impacts 13 on patient-level outcomes (unmet needs), caregiver (burden), resource use (ER and hospital use).

Innovative interventions described and not formally evaluated (Type-II)
The reports describing an innovative intervention or model to improve FS care without formally evaluating their impact (Type-II) ( Table 1.

Analysis by clinical quality indicators (CQI)
The CQI most often compared across various frail senior cohort were: decreased mortality (N= 9), decrease in the rate of visits to the emergency department (N= 5), decrease in hospital use (N= 5), decrease in drug use (N= 4), and decrease in activities of daily living (N= 4) ( Table 1.

6).
Mortality was reported to decrease with increases in continuity of care (comprehensive of care, informational), 18 in males, 19 and in persons presenting with chronic obstructive lung disease. 20 On the other hand, increases in mortality were reported with frailty, 21-23 advanced disease (cancer), 23 comorbidities, 20 setting of care (LTC and ambulatory), history of falls, 20 use of psychotropic drugs 24 and symptoms and diseases (Parkinson, pressure ulcers, stroke). 20 Rates of visits to the emergency department decreased with increasing age 25 and informational continuity of care. 18 However, rates of emergency visits increased in men, 26 in patients with cognitive impairment suspicion 27 and in those using hospital and LTC institutions. 28 No significant association was reported between the rate of visits to the ER and 11 covariates (age, sex, frailty, comorbidities, marital status, education, symptoms and diseases, advanced disease, extensive cancer treatment received continuity of care and neighborhood income).
Hospital use was significantly associated with 15 CQI. It decreased with local anesthesia (vs. general) 29 or in greater community size (ref: <10 000). 30 In contrast, a report showed that hospital use increased with increasing age, frailty, low strength social network, increasing number of medications, history of falls, health status (mild, moderate to high, ref: stable), and comorbidities. 30 Other reports also showed increases in hospital use in people who used long-term care, 31 with length of antibiotic prescriptions, 32 and with increasing age. 33 Men used less drugs than women. 33 However, covariates such as increasing age, 33 length of antibiotic prescription 32 and using long-term care settings 31 were associated with increases in drug use. No significant association was observed between drug use and neighborhood income, having a pharmacy chain affiliation (ref: no affiliation), LTC care or hospital use.
In a single report, the patients' activities of daily living (ADL) were demonstrated to increase with the patient's familiarity with home settings. 34 In contrast, other reports showed that ADL decreased or were unaffected by frailty 35-37 and by patient's familiarity with home settings. 34

Analysis by covariate
In the reports describing the quality of care across various FS cohorts (Type-III), we identified 45 covariates associated with improvement, deterioration or without effect on CQIs. The covariates more frequently studied were frailty (N= 20), age (N= 16), sex (N= 11), comorbidities (N= 9) and setting of care (hospital and LTC) (N= 7 and N= 7; respectively). The covariates more frequently associated with improvement in CQI were frailty, age, sex, and setting of care (hospital use and LTC use).
Increasing level of frailty was associated with improvement in 5 CQI, deterioration in 14 CQI and no effect on 12 CQI. Higher frailty scores were associated with improvement in several CQI, notably decreased polypharmacy (patients needing constant assistance with eating and those who do not use toilet), increased quality-of-care processes in geriatric assessment (comprehensiveness, informational continuity, completion of advance health care directives), increased daily oral hygiene, increase in discharge process (ability to be discharged home for previously non-institutionalized patients), and increased decisions of the patient to decline scheduled treatment or procedure. Frailty was also reported to be associated with deteriorations in several CQI, notably, increased polypharmacy in patients sad or depressed, increased anxiety, increased depression, increased post-operative morbidity, increased mortality, increased risk of fracture and falls, increased cancer treatment toxicity, decreased physical function, decreased activities of daily living (ADL) and instrumental activities of daily living (IADL), increased hospital use, increased emergency department use and increased institutionalisation). Some reported a lack of association between frailty levels and mortality, cancer treatment toxicity, rate or number of fractures, risk of fall, activities of daily living (ADL), depression, polypharmacy, daily oral hygiene, number of medical visits, emergency department use, hospital use and institutionalization.
Increasing age was associated with improvement in four CQI (decrease in polypharmacy, decrease in the use of inpatient hospital services, decrease in emergency department use and increase in the quality-of-care process in geriatric assessment). Increasing age was also reported to be associated with deteriorations in four CQI (increase in institutionalisation, increase in hospital and drug use and decrease in the use of community based-care). No association was reported between age and 11 CQI (risk of fall, mortality, cancer treatment toxicity, depression, hospital admission or death within 30 days of treatment initiation, survival, activities of daily living (ADL), number of medical visits, emergency department use, hospital and drug use).
Sex (being a woman) was associated with improvement in four CQI (decrease in polypharmacy, decrease in mortality, decrease in drug use and decrease in quality-of-care in geriatric assessment). In contrast, being a woman resulted in higher risk of institutionalization in LTC and use of emergency department while no significant association was observed between sex and seven CQI (mortality, cancer treatment toxicity, depression, survival, number of medical visits, emergency department use and hospital use).
The number of comorbidities was associated with a decrease in mortality. It was also associated with deteriorations in four CQI (increase in polypharmacy, increase in depression, increase in hospital admission or death within 30 days of treatment initiation, and decrease in survival). Some of the included reports demonstrated a lack of association between age and six CQI (hospital admission or death within 30 days of treatment initiation, mortality, cancer treatment toxicity, emergency department use, hospital use and number of medical visits).
Setting of care (hospital or long-term care facilities) was associated with improvement in 5 CQI, deterioration in 7 CQI and was not associated with 5 CQI. Acute care hospital use was associated with increased survival, but it was also associated with deteriorations in four CQI (increased polypharmacy, increased institutionalization and increased hospital and emergency department use). No significant association was reported between hospital use and mortality. Hospital use was reported to be associated with increase, or decrease in quality of care processes in geriatric assessment, or to not be associated with it across various reports. Increases in the quality-of-care process in geriatric assessment were associated with hospital use defined as hospital length of stay, whereas decreased in the qualityof-care process in geriatric assessment were associated with hospital use defined as a previous admission to the Geriatric Assessment Unit (GAU), staying in a GAU for 2 weeks and discharge to community, home or independent-living facility. Long-term care (LTC) use was associated with improvement in a few CQI (decrease in polypharmacy and hospital admission or death within 30 days of treatment initiation), and in deterioration in three CQI (increase in drug use, increase in emergency department use and increase in mortality).

Clinical quality of care indicators
In this section, we describe the results regarding the six clinical quality of care indicators (CQI) that were prioritized by key stakeholders during the Activity 3 (interviews) of the present project: (1) increasing patient quality of life, (2) increasing providers' competencies or skills, (3) reducing symptoms, (4) reducing caregiver burden, (5) increasing patient satisfaction with care, and (6) increasing continuity of care by the family physician.

-Increasing patient quality of life
Overall, seven studies focused on patient quality of life. Four of them evaluated the impact of an intervention on quality of life (Table 1.5). Three interventions significantly increased patient quality of life (a 12-week small-group physical exercise program, 8 a multifactorial, interdisciplinary team approach to falls prevention, 38 and a mobility intervention in long-term care facilities 39 ). A single intervention had no significant impact on quality of life (assessment, collaborative care planning, health promotion, and referral to community health and social support services by homecare nurses). 40 Three additional studies described an intervention designed to improve quality of life, without formally evaluating its impact: one targeted increasing independence after illness, 41 one the improvement of oral health, 42 and one disease prevention. 43 None of the included studies described quality of life across various frail senior cohorts.

2-Increasing providers' competencies, skills, or knowledge
Overall, five reports presented interventions aiming to improve providers' competencies, skills, or knowledge. A single one, the VIDOS study, evaluated an intervention's impact on providers' competencies or skills (Table 1.12). It consisted of an interdisciplinary, multifaceted knowledge translation intervention, and it was associated with a significant improvement in providers' competency and skills. Four studies described interventions targeting provider competency and skills without reporting efficacy measures. Two proposed implementing an interprofessional model of care. The first involved a collaborative team that interacted during team-based case meetings, and aimed at enhancing interprofessional skills of geriatricians and primary care providers. 44 The second described an innovative interprofessional model offering secondary outpatients services for frail seniors (e.g. early diagnosis of dementia, assessment of complex medical conditions, and development of an interprofessional health improvement plan). 45 Team members attended training sessions and workshops organized by a regional clinical nurse. Two other interventions involved nurses working in hospitals or long-term care facilities. The first was a reflective practice intervention (RPI) including eight thematic workshops each 75 minutes long, delivered at three-week intervals, combined with reading assignments and individual exercises. 46 The RPI focused on three themes central to elder hospitalization: medication, mobilization, and discharge planning. The second was a comprehensive inservice education program for nurses and residential care-aide staff in provision of daily mouth care for elders in various long-term care facilities. 42 We found no studies describing this indicator across various frail senior cohorts.

3-Reducing symptoms
We identified 3 reports targeting a reduction of patients' symptoms. Two interventions did not have a significant effect on symptoms (a multifactorial, interdisciplinary team approach to falls prevention 38 and a nordic walking program 47 ). A third report described an intervention consisting in a hierarchical assessment of balance and mobility, without studying its impact. 48 This intervention consisted of a bedside assessment of balance and mobility to help track acute changes in the health status of older people admitted to hospital. None of the included studies reported symptoms across various frail senior cohorts.

4-Reducing caregiver burden
Caregiver's burden was measured in a single study, the PRISMA study, 13 which evaluated the efficacy of a coordination-type integrated service delivery model. The intervention was significantly associated with a deterioration of caregiver's burden. None of the included studies reported caregiver burden across various frail senior cohorts.

5-Increasing patient satisfaction with care
Four studies focused on patient satisfaction with care. Two of these evaluated the impact of an intervention. The PRISMA study showed an improvement in patient satisfaction with care after implementation of a coordination-type integrated service delivery model. 13 The other intervention evaluated the implementation of alternate housing models. 17 The descriptive studies consisted of an intervention regarding a dignity therapy for terminally ill patients 49 and the implementation of integrated models of care for frail seniors in Waterloo. 50 None of the included studies reported patient satisfaction with care in various frail senior cohorts.

6-Increasing continuity of care
None of the included reports discussed continuity of care as clinical quality indicators. It was only studied as a potential co-variate influencing other CQI.

Criterion met by participants described in report to be characterized as 'frail' Frequency (%)
The description of participants provided in the report allows to classify them as frail using the Canadian Study of Health and Aging clinical frailty scale 51 or if two or more domains of the Edmonton Frail scale 4 were described as health problems of the targeted participant 62 (83%) Participant are described as being 'frail' by the authors of the report, and the authors justify the use of the term 'frail' by citing an index.

(35%)
Participant are living in long term care facilities or palliative care home (hospices) 19 (25%) Participant are clearly identified to be at the end-of-life, terminally ill, in palliative care 6 (8%)

A-Reports on the impact of an interventions (n=22) (type-I reports)
Intervention category N Health intervention (n=14)