Introduction

Many older inpatients are at high risk of adverse clinical outcomes such as longer length of stay, reduced physical function, increased dependency, admission to a care home, readmission to hospital and death [1,2,3]. It is well recognised that patient factors including physical function, illness severity, cognition, comorbidity, presenting medical diagnosis, polypharmacy, frailty and age can influence the outcomes of hospital admission [4]. However, the organisation of care can also impact on healthcare outcomes.

Best clinical practice includes the identification of older individuals at risk of poor healthcare outcomes routinely in order to optimise patient-centred service delivery [5, 6]. Routine early identification of high-risk patients may help healthcare professionals assess individuals’ needs and provide appropriate interventions [7]. However, it is unclear whether and how those people are identified in current practice. The purpose of this study was to define current practice in one hospital with regard to the use of routine clinical data to identify patients at high-risk of poor healthcare outcomes.

Main text

Methods

Interviews and focus groups were used to investigate the current practice of clinical staff in identifying older patients at risk of poor healthcare outcomes. In addition, quantitative data was abstracted from patients’ clinical records to contextualise the study patients and routine admission practice. The study was conducted in five acute medical wards (120 beds) for older people in one hospital in England including three female and two male wards.

Qualitative data obtained from staff interviews and focus groups

In-depth semi-structured interviews and focus groups were conducted by the first author (KI) with the assistance of a moderator (CO) with a range of healthcare staff on the study wards. Interview schedules were designed to obtain information on the current roles and experiences of each participant in relation to identifying high-risk patients (see Additional file 1). Written consent was obtained and the conversations were audio-recorded. Participants were recruited until no more new concepts emerged (i.e. sampling saturation) [8].

Interviews and focus groups were transcribed verbatim, password protected, and anonymised. Data was analysed thematically using Framework method [9]. A descriptive framework containing three main themes (described below) was identified from the initial analysis of the transcripts. Two researchers (KI, HCR) completed the analysis with regular discussion. A software program (NVivo 10) [10] was used to facilitate data analysis.

Quantitative data abstraction from clinical records

A sample of 60 patients’ clinical records were reviewed by two researchers (KI & CO) to characterise the older patients, contextualise the wards and describe the usual care provided on admission. Demographic data on age, sex, date of admission, domicile status, and reasons for admission were obtained. Information recorded about diagnosis, resuscitation status and routine assessment measures applied to patients on admission were collected. Exclusion criteria included patients who were at end of life or had been admitted for less than three days at the time of data collection.

Quantitative data were analysed using descriptive statistics and the statistical software package IBM SPSS statistics 22. Data were summarised using mean (standard deviation, SD), median (interquartile range, IQR) and number (percent, %) as appropriate. The association between clinical outcomes (e.g. length of stay and discharge destination) and risk assessments scores was completed using Chi square or linear regression as appropriate.

Results

Qualitative results

A total of five focus groups and three interviews were conducted with 22 staff participants across the five study wards with different sex, professional bands, and years of experience. The sample comprised: seven nurses, four dieticians, four therapists (two physiotherapists and two occupational therapists) and seven doctors: two consultants, two foundation year (FY) doctors (doctors within the first 2 years post-graduation), and three specialty registrars (trainee doctors specialising in Geriatric Medicine). Three main themes developed from the analysis reflecting participants’ views and experiences of identifying older patients at risk of poor healthcare outcomes (Table 1).

Table 1 Themes, sub-themes and illustrative data extracts
Full size table

Theme 1: “Clinical judgment”

Staff relied on their “clinical judgment” to interpret patient’s needs and to recognise those who might be at risk of poor healthcare outcomes. A number of factors influenced clinical judgment including: reasons for admission, familiarity with patients, patient’s general condition, visible frailty, and patient’s ability to manage at home. One consultant reported that severity of acute illness and high number of comorbidities as well as poor response to treatments enabled him to assess the likelihood of deterioration. Another consultant reported that familiarity with patients (see Quote 2, Table 1) assisted his judgment. Nursing staff stated that they recognised high-risk patients from their general condition. Visible frailty was reported to influence registrars’ clinical judgment (see Quote 4, Table 1). Similarly, dieticians described that they encouraged nursing staff to use their clinical judgment to recognise frail patients.

Consultants and therapists reported that patients’ inability to manage at home despite a care package was an important factor in judgment that these patients might need higher levels of care such as placement in a nursing home. Nursing staff and FY doctors stressed that the need to change discharge destination was the main predictor of prolonged length of stay (Quotes 6 and 8, Table 1).

Theme 2: “Therapy assessment”

Registrars reported that therapists’ assessment of patients played a crucial part in identifying older people at risk of functional decline or longer hospital stay (Quotes 9 and 10, Table 1). Therapy staff and registrars considered patients who lack motivation and fail to engage with therapy to be at higher risk of poor outcomes. Therapists recognised that disengagement could be sometimes due to inability to understand and follow instructions as a result of cognitive impairment such as dementia (Quote 13, Table 1).

Theme 3: “Difficulties and challenges”

Participants reported a number of challenges that could hinder their ability to identify high-risk patients including: delirium, predicting risk of falls, delayed clinical judgment, and lack of communication. One consultant reported that the presence of delirium made judgment of a patient’s true risk status more difficult (Quote 14, Table 1). Registrars and therapists reported that all patients were at risk of functional decline and deconditioning, hence the importance of therapy assessment and treatment. However, consultants believed that it was hard to predict how much patients will decline functionally during their admission and that they usually take a ‘best guess’ which can be wrong sometimes (Quote 15, Table 1).

Communication of the routine risk assessments between staff was a challenge. These assessments were not communicated to doctors unless there was a need for medical advice. Registrars reported that nursing notes were unfriendly documents with much repetition. Importantly, FY doctors and consultants reported needing several days to make a clinical judgment of patients’ needs which could potentially delay interventions (Quotes 19 and 20, Table 1). FY doctors and registrars thought the adoption of a simple clinical tool to highlight those high-risk patients earlier in the admission could be helpful.

Quantitative results

A random sample of 60 patients’ (35 female) clinical records across the five study wards were reviewed. Patients’ demographic variables are detailed in Table 2. This group of patients (mean age 86.7 years) had complex needs and multiple risk factors: 42 (70%) had more than five active comorbidities at time of admission and 51 (85%) were taking more than five medications; 50 (83%) had both high comorbidities and polypharmacy. Forty-six (77%) of patients had dementia and 35 (58%) had do not attempt resuscitation (DNAR) forms completed. Twenty-four (40%) of patients were discharged to a new destination; more than half of these went to nursing homes and six died.

Table 2 Characteristics of participants
Full size table

A number of routine risk assessments and relevant care plans were completed by nursing staff on admission (see Table 3). These included assessment of nutrition (completed for 85%), falls risk (95%), mobility (85%), and pressure ulcer risk (88%). Eight (16%) patients were found to be at high-risk of malnutrition but in fact 11 (18%) were referred to the dietetic team and 19 (32%) patients were prescribed oral nutritional supplements. Forty-eight (80%) of patients were seen by the therapy team early on admission. Only two patients were recorded to be frail. No specific tool was used for the assessment of frailty.

Table 3 Routine risk assessments performed on admission to hospital wards
Full size table

No significant association was found between risk of malnutrition, falls and pressure ulcers and discharge destination. Similarly risk of malnutrition, pressure ulcers and handling and mobility risk assessments were not associated with length of stay. However, there was a significant association between risk of falls and length of stay in this small sample.

Discussion

This study described how older inpatients at higher-risk of poor healthcare outcomes such as longer length of stay in hospital, reduced physical function, and discharge to care home were identified in routine hospital practice. Staff relied on their “clinical judgment” and “therapy assessment” to recognise high-risk patients. The multi-disciplinary setting (medical, therapy, and nursing staff) appeared to facilitate clinical judgement. However, a number of challenges in making this clinical judgment were reported leading to possible delays. Few risk assessment measures completed routinely by nursing staff on admission to the wards. Yet, the purpose of these assessments was to identify the risk of a single adverse outcome and they were not used collectively to highlight patients at risk of poor healthcare outcomes. Patients’ clinical records lacked documentation of staff clinical judgment.

Clinical judgment can be defined as “an interpretation or conclusion about a patient’s needs, concerns, or health problems, and/or the decision to take action (or not), use or modify standard approaches, or improvise new ones as deemed appropriate by the patient’s response” [11]. It is complex and requires a flexible ability to recognize prominent aspects of an undefined clinical situation, interpret their meaning, and respond appropriately. It relates to the experience of individual clinicians. Senior clinicians often possess experience and knowledge to inform their decisions. Junior clinicians might be less confident in their clinical judgment and tend to follow protocols to eliminate variability in patient care [12]. Similarly in our study, there was a discrepancy in practice between consultants and junior doctors which in turn could have impacted on their clinical judgement.

There are concerns about the accuracy of clinical judgment on the functional status of older patients in hospitals, and to predict length of stay and survival from critical illness and mortality [13, 14]. Some staff interviewed in this study questioned the accuracy of their clinical judgment sometimes suggesting that a simple measure could be highly relevant in daily practice. This measure should be easy to use in clinical settings, quick, cheap and reliable. A number of assessment instruments have been developed to predict risk of adverse outcomes among hospitalised older inpatients [15,16,17,18]. However, they have not been proven to be accurate and reliable [19].

Gait speed has been recommended as the most suitable assessment tool to be implemented in standard clinical evaluation of community dwelling older people [20]. However in hospital where many patients are acutely ill and unable to walk, grip strength measurement could be more suitable [21]. Low grip strength is associated with poor current and future health including increased falls [22], morbidity [23], and death [24]. Further research is needed to evaluate the feasibility of its implementation in routine clinical practice [25].

Limitations

This study was conducted in one hospital in England and the current practice identified in this study may not be necessarily generalised to other hospitals and departments. Therefore, it is important to replicate this work to compare variation in clinical practices. Second, the study reported only what was documented in clinical records and in staff interviews. Discussions between clinical staff during the ward rounds and multi-disciplinary team meetings were not captured. However, we believe that the interviews and focus groups with multiple healthcare professionals addressed this concern.

Conclusions

Staff relied on their clinical judgment to identify high risk patients. Making such a judgment was often difficult and could occur several days after admission potentially delaying specific interventions. A number of risk assessments were carried out routinely but used mainly to identify single healthcare needs rather than holistically to identify individuals at risk of poor healthcare outcomes to the multi-disciplinary team. We believe that early identification of those patients using a valid measure alongside staff clinical judgment could be highly relevant. Future research will assess the feasibility and acceptability of using grip strength measurement as a tool to identify older patients at risk of poor healthcare outcomes.