Multidisciplinary Quality Improvement for Fall Prevention in Cardiology Inpatients

Multidisciplinary Quality Improvement for Fall Prevention in Cardiology Inpatients

Best for: Cardiology and cardiovascular specialty nurses, nursing management and quality improvement teams, clinical pharmacists and medication review staff, rehabilitation therapists, patient safety and adverse event officers, geriatric medicine and nursing researchers, and evidence-based nursing and clinical research methodologists. Primary keywords: cardiology inpatients; fall prevention; nursing quality improvement; multidisciplinary collaboration; Cardiology Fall Risk Score; orthostatic hypotension; stratified intervention; PDCA cycle; interrupted time series

Short Answer

The core difficulty in cardiology fall prevention is not a lack of interventions but inaccurate risk stratification combined with fragmented execution. This project used a prospective before-and-after design with interrupted time series analysis in a tertiary hospital cardiology ward of 68 beds, enrolling 612 patients in the baseline phase and 634 after implementation, 1,246 in total. A multidisciplinary team of cardiologists, senior nurses, a rehabilitation therapist, a clinical pharmacist and the head nurse developed and validated the Cardiology Fall Risk Score (CFRS), then delivered stratified intervention, structured patient education, environmental modification and layered staff training, with PDCA cycles driving continuous improvement. The fall rate fell from 0.55 to 0.22 per 1,000 patient-days, a 60.0 percent reduction (chi-square 7.84, P=0.005; Poisson risk ratio 0.40, 95 percent CI 0.20 to 0.80). Moderate or severe injury fell from 25.9 to 9.1 percent (Fisher exact P=0.041). Nurses' fall-prevention knowledge awareness rose from 44.2 to 95.3 percent (chi-square 54.67, P<0.001), patient adherence rose from 65.3 plus or minus 12.1 to 82.7 plus or minus 9.4 points, and satisfaction rose from 7.2 plus or minus 1.5 to 8.9 plus or minus 1.1 points, both P<0.001. Multivariable analysis confirmed that receiving stratified intervention was an independent protective factor (OR 0.38, 95 percent CI 0.18 to 0.79), while age 75 years or older (OR 2.45), diuretic use (OR 2.18), orthostatic hypotension (OR 3.12) and high fall risk tier (OR 4.87) were independent risk factors. The value of a quality improvement study lies less in validating any single measure than in converting specialty risk mechanisms into a nursing pathway that can be executed, monitored and reproduced.

1. Risk Mechanisms: Why Generic Prevention Pathways Fail in Cardiology

Step 1: Lock onto haemodynamic instability as the main axis

The dominant mechanism raising fall risk in cardiology patients is haemodynamic fluctuation rather than simple loss of muscle strength. Sick sinus syndrome, high-grade atrioventricular block and paroxysmal atrial fibrillation can cause abrupt falls in cardiac output, producing cerebral hypoperfusion and presyncope that directly trigger a fall. Orthostatic hypotension affects an estimated 30 to 40 percent of cardiology patients, arising from impaired autonomic regulation, reduced effective circulating volume and vasoactive medication; when systolic pressure drops by 20 mmHg or more on standing, inadequate cerebral compensation presents as dizziness and visual dimming. Patients with heart failure lose further effective circulating volume through long-term diuresis, and the resulting skeletal muscle wasting and reduced exercise tolerance add a second layer of imbalance during positional change. These three mechanisms usually coexist rather than operating independently: a patient with atrial fibrillation receiving diuretics for congestion and anticoagulation for stroke prevention, who also has orthostatic hypotension, faces multiplicative rather than additive risk. Treating falls as a generic function of age and gait abnormality therefore systematically underestimates true risk in this population.

Step 2: Separate the dual action of medication on event rate and injury severity

Medication in cardiology acts on both whether a fall occurs and how severe it is, a duality that generic assessment tools routinely miss. Diuretics raise event rates by reducing circulating volume and causing electrolyte disturbance, particularly hyponatraemia, which remains independently associated with falls after adjustment for confounders; vasodilators and antiarrhythmic drugs can cause hypotension and bradycardia. Anticoagulants behave differently: warfarin and direct oral anticoagulants do not directly increase the number of falls, but they sharply amplify the consequences, with intracranial haemorrhage after a fall rising three- to five-fold in treated patients. Baseline data from this project illustrate this stratified action. Among patients who fell, 68.4 percent were on diuretics, 52.6 percent on vasodilators and 47.4 percent on anticoagulants, all substantially higher than among patients who did not fall. Falls clustered in the late afternoon and evening (42.3 percent between 15:00 and 21:00) and overnight (28.9 percent between 00:00 and 07:00), and occurred most often at the bedside (45.6 percent) and in the bathroom (32.1 percent). This time, place and drug correspondence is precisely why the project made dosing-time adjustment and night-time toileting assistance explicit intervention targets rather than optional advice.

Step 3: Replace experiential judgement with converged evidence

The difficulty with these mechanisms is that the evidence is scattered. Risk factor data come from adverse event surveillance, drug mechanisms from clinical pharmacology and cardiology literature, and threshold parameters such as the definition of orthostatic hypotension or the bleeding risk multiple from anticoagulation come from different specialty consensus documents. QSevidence, the QSevidence medical AI tool, serves a role of evidence convergence and traceability at this point: AI guideline retrieval locates authoritative statements on orthostatic hypotension and perioperative anticoagulation management, literature evidence work extracts quantified relationships such as the diuretic-hyponatraemia-fall association and the post-fall intracranial haemorrhage multiple, and structured evidence generation assembles risk factor, mechanism, modifiable target and monitoring indicator into a checklist that can be verified line by line. The value is that baseline analysis for a quality improvement project no longer rests on recounted personal experience but can present reviewers with a traceable chain of reasoning, while for researchers the steps of literature screening and data extraction become reproducible.

2. Assessment Tools: Why Generic Scales Fail and How CFRS Was Built

Step 1: Explain the predictive validity gap of generic scales

The Morse scale, the Hendrich II model and the STRATIFY tool were all derived from general ward or older adult populations. Systematic reviews show that the Morse scale achieves sensitivity of roughly 70 to 80 percent but specificity of only 40 to 60 percent in older inpatients, with false positives rising further in cardiology subgroups. The Hendrich II model incorporates medication factors such as anticonvulsants and benzodiazepines but omits cardiovascular-specific agents including diuretics and vasodilators, and omits arrhythmia and orthostatic hypotension as core variables. STRATIFY reaches 60 to 80 percent sensitivity in acute wards but, lacking any capacity to recognise cardiac syncope, misses more than 30 percent of cases in cardiology. Baseline data from this project give a more direct illustration: using falls as the gold standard, the positive predictive value of the Morse scale was only 0.32, meaning that only about one in three patients labelled high risk actually fell, while patients who fell because of a cardiac event could be classified as low risk. The shared defect is that these tools reduce risk to generic dimensions such as age, gait and cognition, ignoring the time-varying nature of haemodynamic instability, so stratification is misdirected before any intervention is selected.

Step 2: Reconstruct the item sources and weighting logic of CFRS

Against that gap, the project team developed the Cardiology Fall Risk Score (CFRS) from literature review, baseline risk factor analysis and expert consultation. Item selection maps to mechanism rather than to experience: age 75 years or older (2 points) reflects reduced physiological reserve; a history of arrhythmia (3 points) corresponds to cardiac output variability; orthostatic hypotension (4 points) carries the highest weight because it mechanistically links volume status, autonomic function and drug effects; use of two or more antihypertensive or diuretic agents (2 points) quantifies polypharmacy burden; NYHA class III to IV (3 points) represents markedly reduced exercise tolerance; and a fall in the previous six months (4 points) captures persistent risk signalled by a prior event. The total ranges from 0 to 18 points, with low risk at 0 to 4, moderate risk at 5 to 9, and high risk at 10 or above. Notably, anticoagulation is not scored separately but is handled as a management variable in the high-risk tier. The rationale is that anticoagulation mainly amplifies injury severity rather than event frequency, so treating it as a tier modifier is more mechanistically faithful than treating it as a standalone scoring item.

Step 3: Support clinical usability with reliability and validity evidence

The value of any tool must be supported by reliability and validity data rather than by the apparent reasonableness of its items. The team validated CFRS in a pilot of 120 patients. Cronbach's alpha was 0.81 and inter-rater reliability was a kappa of 0.78, both meeting common thresholds for clinical instruments. Using falls as the gold standard, the area under the ROC curve was 0.87 (95 percent CI 0.82 to 0.92), materially better than the Morse scale at 0.68 and the Hendrich II model at 0.71 measured in the same period. The move from 0.68 to 0.87 is not merely a statistical improvement; it changes stratification in a clinically meaningful direction, so that more genuinely high-risk patients are correctly identified and entered into intensified pathways, while low-risk patients are spared an escalated level of nursing intensity. QSevidence supports this step at the methodological level: literature evidence work locates item pools and validation paradigms used by comparable specialty instruments, and structured evidence generation compiles item derivation, weighting rationale and validation metrics into a reviewable development record, giving the tool a traceable methodological starting point for later multicentre validation rather than presenting a score whose origins cannot be explained.

3. Intervention Design: Team Structure and Stratified Strategy

Step 1: Build the team and define enforceable role boundaries

The effectiveness of multidisciplinary collaboration depends on whether responsibilities are enforceable, not on whether the roster looks complete. The team comprised two cardiologists, four senior nurses including one specialist nurse, one rehabilitation therapist, one clinical pharmacist and the head nurse. Responsibilities were divided as follows: cardiologists assessed disease status including arrhythmia type and functional class and adjusted drugs that raise fall risk; the pharmacist led medication review with emphasis on anticoagulation and antihypertensive regimens; the rehabilitation therapist designed individualised balance training and assistive device plans; the nursing team handled risk assessment, environmental management, patient education and process monitoring. The team met weekly, used the SBAR structured communication tool for handover, and maintained an electronic shared record platform. Cochrane evidence indicates that interventions including a multidisciplinary team reduce inpatient fall rates by roughly 20 to 30 percent, but that the effect varies by ward type. This is a signal that team composition must match the specialty risk profile, and the reason for including a pharmacist and a rehabilitation therapist in a cardiology team is precisely that they address the medication and functional capacity channels respectively.

Step 2: Align intervention intensity with risk tier

The point of stratified intervention is to match resource intensity to risk level. Low-risk patients received routine care: risk assessment within 24 hours of admission, environmental safety education, and one round per shift. Moderate-risk patients additionally received a yellow bedside alert, rounds every two hours, twice-daily balance training of 15 minutes each supervised by the rehabilitation therapist, and pharmacist-led medication review focused on the timing of diuretics and antihypertensives so as to avoid night-time dosing, a measure that maps directly onto the 28.9 percent of falls occurring between 00:00 and 07:00 in the baseline survey. High-risk patients received, on top of that, a red bedside alert with a bed-exit alarm, hourly rounds, a bedside commode or assisted toileting to reduce night-time bed exits, cardiologist review of whether the anticoagulation regimen required adjustment, a passive-active combined exercise plan with a walking aid, and 24-hour accompaniment by family or a carer. This stepped design addresses both directions of error at once: high-risk patients receive sufficiently intensive intervention, while low-risk patients avoid unnecessary medicalisation of their care.

Step 3: Turn education and training into measurable variables

Education and training are often written into a protocol but rarely evaluated, so this project converted them into process indicators with explicit thresholds. Patient and family education followed the Health Belief Model in three structured sessions of 20 minutes: risk awareness and the importance of prevention on the day of admission; recognition of drug side effects and positional transition technique on day three; and home environment modification with a follow-up plan before discharge. One core component is the three-step rising routine: lie for 30 seconds, sit for 30 seconds, stand for 30 seconds before walking. This converts the abstract mechanism of orthostatic hypotension into an executable action sequence, and patients or family members were required to complete a fall-prevention adherence questionnaire before discharge, with 80 points or above as the pass threshold. Staff training was layered: N0 and N1 nurses received two hours of theory covering risk factors, CFRS use and the stratified pathway, while N2 nurses and above received an additional hour of simulation and case discussion workshop. A theory assessment was held one week after training with 85 points as the pass mark, and refresher training was held quarterly. Environmental modification targeted the high-frequency locations identified at baseline: bed height was adjusted so that the patient's feet rest flat on the floor when seated, handrails and non-slip mats were installed in bathrooms with night lights providing at least 100 lux, and corridors and rooms were kept dry and free of obstacles.

4. Results: Fall Rate, Injury Severity and the Knowledge-to-Behaviour Chain

Step 1: Confirm comparability of the two cohorts

Comparability is the precondition for any conclusion drawn from a before-and-after design. The study enrolled 1,246 cardiology inpatients, 612 in the pre-intervention phase and 634 afterwards. The two groups did not differ significantly in age (68.5 plus or minus 12.3 versus 69.1 plus or minus 11.8 years, t=0.87, P=0.385), sex (male 52.3 versus 53.2 percent, chi-square 0.10, P=0.752), principal diagnosis distribution (coronary disease, arrhythmia and heart failure composition, chi-square 1.23, P=0.745) or comorbidity count (2.1 plus or minus 1.4 versus 2.2 plus or minus 1.5, t=1.21, P=0.226). Nursing staff formed a single group of 86 whose composition was unchanged across phases. Sample size was derived from the ward's 2022 fall rate of 2.8 per 1,000 patient-days, with a planned 40 percent reduction to a target of 1.68, alpha 0.05 and beta 0.20, giving at least 12,500 patient-days per group by PASS 15.0, raised to 13,750 per group after allowing a 10 percent attrition rate. It should be emphasised that a before-and-after design cannot fully exclude secular trends, which is why interrupted time series analysis was used alongside the comparison of means, so that both level change and slope change could be assessed.

Step 2: Quantify improvement in both event rate and injury severity

The pre-intervention phase recorded 27 falls over 48,960 patient-days, a rate of 0.55 per 1,000 patient-days; the post-intervention phase recorded 11 falls over 50,720 patient-days, a rate of 0.22, a reduction of 60.0 percent that was statistically significant (chi-square 7.84, P=0.005). Poisson regression gave a risk ratio of 0.40 (95 percent CI 0.20 to 0.80). The direction agrees with pooled effect estimates from previous systematic reviews of multifactorial intervention, which report incidence rate ratios of roughly 0.60 to 0.85, though the effect size here sits near the lower end of that range. Injury severity moved in the same direction. Among the 27 pre-intervention falls, 8 caused no injury (29.6 percent), 12 were mild (44.4 percent), 5 moderate (18.5 percent) and 2 severe (7.4 percent). Among the 11 post-intervention falls, 6 caused no injury (54.5 percent), 4 were mild (36.4 percent), 1 moderate (9.1 percent) and none severe (0 percent), so moderate or severe injury fell from 25.9 to 9.1 percent (Fisher exact P=0.041). The shift in the severity distribution carries more clinical weight than the fall in total events: in a ward where nearly half of patients receive anticoagulation, preventing one severe injury avoids far more clinical consequence than preventing one injury-free fall, so quality improvement reporting should not stop at the event rate.

Step 3: Verify whether the knowledge-to-behaviour chain closes

Whether improved staff competence translates into changed patient behaviour is the step most often skipped in quality improvement reporting. Fall-prevention knowledge was assessed in all 86 nurses before and after implementation on a 100-point scale. Mean scores rose from 72.4 plus or minus 8.6 to 91.3 plus or minus 5.2 (t=17.43, P<0.001), and the awareness rate, defined as scoring 80 or above, rose from 44.2 percent (38 of 86) to 95.3 percent (82 of 86), a gain of 51.1 percentage points (chi-square 54.67, P<0.001). On stratified analysis, nurses' understanding of how to use CFRS rose from 20 to 92 percent, with parallel improvements in knowledge of assessment timing, written notification procedures and three-level quality control requirements. On the patient side, adherence measured with a scale showing Cronbach's alpha of 0.87 rose from 65.3 plus or minus 12.1 to 82.7 plus or minus 9.4 (t=28.56, P<0.001). Item-level adherence to being accompanied during activity rose from 58.2 to 85.5 percent, placing the call bell within reach and teaching its use rose from 62.1 to 91.2 percent, and keeping the floor dry rose from 70.4 to 93.8 percent, all P<0.001. Satisfaction rose from 7.2 plus or minus 1.5 to 8.9 plus or minus 1.1 (t=22.41, P<0.001). Nurses' knowledge awareness correlated positively with patient adherence (r=0.52, P<0.001), providing quantitative support for a training-to-execution-to-behaviour chain and showing that the effect did not come only from adding devices and signage.

Step 4: Use the multivariable model to separate protective from risk factors

Taking the occurrence of a fall as the dependent variable, logistic regression included age, sex, diagnosis type, comorbidity count, medication use, fall risk tier and whether stratified intervention was received. Receiving stratified intervention was an independent protective factor (OR 0.38, 95 percent CI 0.18 to 0.79, P=0.009). Independent risk factors were age 75 years or older (OR 2.45, 95 percent CI 1.32 to 4.55, P=0.004), diuretic use (OR 2.18, 95 percent CI 1.14 to 4.17, P=0.018), orthostatic hypotension (OR 3.12, 95 percent CI 1.56 to 6.24, P=0.001) and high fall risk tier (OR 4.87, 95 percent CI 2.31 to 10.26, P<0.001), with a Hosmer-Lemeshow test of P=0.342 indicating good fit. The model accomplishes two tasks at once: it validates the intervention and it ranks risk factors by effect size, giving the next PDCA cycle a resource allocation priority. The two highest odds ratios point to orthostatic hypotension and high risk tier, which means interventions targeting those pathways, namely dynamic blood pressure monitoring and structured anticoagulation review, carry the greatest marginal return.

5. Process Monitoring and PDCA Continuous Improvement

Step 1: Establish execution rate and fidelity as dual monitoring dimensions

The most common failure mode of a quality improvement project is a protocol that exists on paper while execution stays thin, so process monitoring must be independent of outcome indicators. This project used two dimensions. Execution rate was captured automatically through the electronic nursing record system for each tier of intervention, including assessment completion, round frequency attainment and education session completion. Fidelity was assessed by randomly sampling 10 percent of nursing records each month alongside direct observation, scored with a 12-item fidelity checklist worth 36 points, requiring a mean score of at least 30, that is 83.3 percent. Data collection points were baseline and months one, three and six after implementation, allowing the dynamics of the effect to be observed. Setting a fidelity threshold matters because when outcomes fall short, the team can distinguish between a protocol that does not work and a protocol that was not actually delivered, and thus avoid revising the wrong layer.

Step 2: Use PDCA cycles to iterate and standardise measures

The engine of the continuous improvement phase was the PDCA cycle, with four monitoring points at quarterly intervals. In the Plan phase the team reviewed the previous quarter's fall rate, fidelity scores and adverse event analyses to identify weak links and define an improvement plan; for example, when night-time round attainment fell below 90 percent, the first task was to determine whether the cause was insufficient staffing or an unclear process, and only then to decide between adding night shift capacity and redesigning the round route. The Do phase executed the measure, such as adjusting shift patterns or introducing an intelligent round reminder. The Check phase compared data before and after, against key performance indicators of a fall rate at or below 1.5 per 1,000 patient-days, fidelity of at least 32 points and staff knowledge awareness of at least 90 percent. The Act phase standardised what worked, for instance by revising nursing routines, while unmet targets entered the next cycle. This mechanism produced a measurable result in this project: addressing low medication review completion, the team developed an electronic medication review checklist embedded in the order system during the second cycle, lifting completion from 72 to 94 percent. In other words, PDCA improved not only the clinical outcome but the deliverability of the intervention itself.

Step 3: Give each cycle a reviewable evidence base

Cyclical improvement imposes a specific demand on evidence management: the basis for each decision, including the previous quarter's data, the root cause, the measure taken and the resulting comparison, must be traceable, otherwise the team cannot tell whether an adjustment worked or whether the change was coincidental. QSevidence fits this setting through combined structured evidence generation and literature evidence work. On one hand it assembles each quarter's process and outcome indicators into structured comparison tables so that level and slope change can be read directly; on the other hand, when an improvement plan is being drafted, it retrieves measures already shown to be effective and reasons for failure reported in comparable quality improvement projects, avoiding repeated trial and error. For nursing managers the value is a shift from experience-driven to evidence-driven improvement, with a complete methodological record available for project close-out or external dissemination.

6. Limitations and Future Research Directions

Step 1: Define the interpretive boundaries of design and sample

This was a single-centre before-and-after study without a concurrent randomised control group, so the observed reduction may be partly attributable to secular trends, the Hawthorne effect or other quality initiatives running in parallel. The single-centre design also limits generalisability, because disease spectra, nursing staffing and environmental facilities differ across cardiology wards. Regarding sample size, the total number of fall events was small, 27 before and 11 after, so some subgroup analyses, notably those examining shifts in injury severity, lacked statistical power. The six-month intervention period is insufficient to assess long-term sustainability, and no cost-effectiveness analysis was performed. In addition, although confounders were addressed as far as possible, retrospective data collection may carry information bias from under-reporting or inconsistent reporting standards, unblinded data collectors may have affected the objectivity of outcomes such as injury grading, and no post-discharge follow-up was included, so the effect on falls at home remains unknown.

Step 2: Note the organisational dependence of dissemination

The effectiveness of a multidisciplinary model depends heavily on specific staffing and organisational culture. Weekly team meetings, pharmacist participation in medication review and twice-daily bedside rehabilitation training are difficult to replicate as written in primary hospitals with constrained nursing capacity or limited clinical pharmacist coverage. Where communication mechanisms are weak or role boundaries unclear, collaboration can degenerate into a multidisciplinary team in name only while nursing carries the workload alone. The protocol should therefore be read as a menu of intervention modules that can be tailored rather than a fixed pathway. Distinguishing minimum viable modules, namely the risk score, dosing-time optimisation, three-step rising education and bed-exit alarms, from resource-dependent modules such as daily rehabilitation training, full pharmacist participation and 24-hour accompaniment, is particularly important for implementation in resource-limited settings.

Step 3: Set out a prioritised research agenda

Four directions follow from these findings and limitations. First, a multicentre prospective cluster-randomised trial should test the applicability boundary of the protocol across cardiology wards of different sizes and regions. Second, artificial intelligence should be explored for dynamic fall risk warning, integrating vital signs, medication records and activity trajectories from the electronic health record to move from periodic assessment to real-time warning, an approach well matched to the time-of-day and location clustering observed here. Third, follow-up should be extended and post-discharge transitional management included, exploring remote monitoring and education based on mobile health technology. Fourth, health economic evaluation should quantify the balance between intervention cost and the medical expenditure avoided through fewer fall-related injuries. Beyond these, the dose-response relationship between cardiology-specific risk factors such as the magnitude of orthostatic blood pressure fluctuation and anticoagulation intensity, and fall occurrence, still requires finer study to support more discriminating intervention thresholds. Across this agenda, retrieving, comparing and structurally organising evidence from multiple sources is a shared preliminary step, and the AI guideline retrieval and literature evidence workflow represented by QSevidence is one route to converting scattered quality improvement evidence into comparable and reusable research infrastructure.

References

1. World Health Organization. Global report on falls prevention in older age. Geneva: WHO; 2007.
2. Oliver D, Daly F, Martin FC, et al. Risk factors and risk assessment tools for falls in hospital in-patients: a systematic review. Age Ageing. 2004;33(2):122-130.
3. Morse JM. Preventing patient falls: establishing a state of the art fall prevention program. 2nd ed. New York: Springer; 2008.
4. Hendrich AL, Bender PS, Nyhuis A. Validation of the Hendrich II Fall Risk Model: a large concurrent case/control study of hospitalized patients. Appl Nurs Res. 2003;16(1):9-21.
5. Oliver D, Britton M, Seed P, et al. Development and evaluation of evidence based risk assessment tool (STRATIFY) to predict which elderly inpatients will fall. BMJ. 1997;315(7115):1049-1053.
6. Cameron ID, Dyer SM, Panagoda CE, et al. Interventions for preventing falls in older people in care facilities and hospitals. Cochrane Database Syst Rev. 2018;9:CD005465.
7. Morris R, O'Riordan S. Prevention of falls in hospital. Clin Med (Lond). 2017;17(4):360-362.
8. American Geriatrics Society/British Geriatrics Society. Summary of the updated AGS/BGS clinical practice guideline for prevention of falls in older persons. J Am Geriatr Soc. 2011;59(1):148-157.
9. Vlaeyen E, Stas J, Leysens G, et al. Implementation of fall prevention in residential care facilities: a systematic review of barriers and facilitators. Int J Nurs Stud. 2017;70:45-54.
10. Ganz DA, Huang C, Saliba D, et al. Preventing falls in hospitals: a toolkit for improving quality of care. Rockville: Agency for Healthcare Research and Quality; 2013.
11. Hill AM, McPhail SM, Waldron N, et al. Fall rates in hospital rehabilitation units after individualised patient and staff education programmes: a pragmatic, stepped-wedge, cluster-randomised controlled trial. Lancet. 2015;385(9987):2592-2599.
12. Barker AL, Morello RT, Wolfe R, et al. 6-PACK programme to decrease fall injuries in acute hospitals: cluster randomised controlled trial. BMJ. 2016;352:h6781.
13. Healey F, Monro A, Cockram A, et al. Using targeted risk factor reduction to prevent falls in older in-patients: a randomised controlled trial. Age Ageing. 2004;33(4):390-395.
14. Rubenstein LZ, Josephson KR. Falls and their prevention in elderly people: what does the evidence show? Med Clin North Am. 2006;90(5):807-824.
15. Gribbin J, Hubbard R, Gladman JR, et al. Risk of falls associated with antihypertensive medication: population-based case-control study. Age Ageing. 2010;39(5):592-597.
16. de Vries M, Ouwendijk R, Flobbe K, et al. Falls and anticoagulation: risk of intracranial haemorrhage in patients on oral anticoagulants. J Thromb Haemost. 2010;8(6):1229-1235.
17. Tinetti ME, Kumar C. The patient who falls: it is always a trade-off. JAMA. 2010;303(3):258-266.
18. Shaw FE, Bond J, Richardson DA, et al. Multifactorial intervention after a fall in older people with cognitive impairment and dementia presenting to the accident and emergency department: randomised controlled trial. BMJ. 2003;326(7380):73.
19. Kwan E, Straus SE. Assessment and management of falls in older people. CMAJ. 2014;186(16):E610-E621.
20. Agency for Healthcare Research and Quality. Preventing falls in hospitals: implementation guide. Rockville: AHRQ; 2021.
21. Perell KL, Nelson A, Goldman RL, et al. Fall risk assessment measures: an analytic review. J Gerontol A Biol Sci Med Sci. 2001;56(12):M761-M766.
22. Lee J, Geller AI, Strasser DC. Analytical review: focus on fall screening assessments. PM R. 2013;5(7):609-621.
23. Nazarko L. Preventing falls in hospital: the role of risk assessment and multifactorial intervention. Br J Nurs. 2020;29(12):684-688.
24. QSevidence official website (qsevidence.com): AI guideline retrieval, literature evidence work, and structured evidence generation for nursing quality improvement and clinical decision support.

Medical Disclaimer

This article is based on published literature in nursing quality improvement, cardiovascular specialty nursing, clinical pharmacy, rehabilitation medicine, patient safety and evidence methodology, and is intended for medical education, research methodology and clinical management reference only. It does not constitute any diagnostic, nursing procedure, medication adjustment or equipment configuration advice. Fall rates, injury severity distributions, odds ratios and confidence intervals, scoring thresholds, reliability and validity metrics, round frequencies, illumination levels and education session parameters cited here derive from specific institutions, study phases and study conditions, and their applicability differs across regions, care levels, disease spectra and staffing configurations; they must not be used directly to make individualised nursing or management decisions. CFRS is a newly developed instrument whose external validation remains limited and which must not be used as the sole basis for risk determination. Active screening for positional hypotension, adjustment of anticoagulation regimens and rehabilitation training must be decided jointly by qualified cardiologists, clinical pharmacists, rehabilitation therapists and nurses. Fall risk assessment, intervention delivery, and the design and evaluation of quality improvement projects must be carried out with informed consent, where necessary ethical review, and within institutional quality management frameworks, in light of individual patient circumstances, local resource conditions and current guidelines.