Abstract
Background
Prescribing cascades occur when a new drug is prescribed to treat an adverse drug event caused by an existing medication, resulting in unnecessary, or potentially hazardous additional drugs. To date, there are no published studies assessing the prevalence of prescribing cascades in older hospitalised adults.
Objective
To investigate the prevalence of prescribing cascades in hospitalised older adults.
Methods
We conducted a prospective observational study of adults aged ≥ 65 years with multimorbidity and polypharmacy presenting to hospital with acute unselected medical or surgical illness. Prescribing cascades were identified using two predefined validated explicit cascade lists, i.e. ThinkCascades, and a list derived from a recently published systematic review of prescribing cascades in community-dwelling adults, referred to here as the ‘Doherty list’. Potential prescribing cascades were classified as ‘definite’, ‘probable’, ‘possible’, ‘uncertain’ or ‘indeterminate’ according to pre-specified criteria.
Results
The study included 385 consecutive patients (55.1% female, mean age 80.2 years, standard deviation 7.3 years). A total of 281 potential prescribing cascades (drug A → drug B) were identified in 152 patients (39.4%). Probable or possible prescribing cascades were identified in 48 patients (12.4%) using the Doherty list and in 44 patients (11.4%) using ThinkCascades. Patients exposed to potential prescribing cascades experienced greater levels of polypharmacy than patients not exposed to prescribing cascades (median interquartile range [IQR] of 12 [9–14] daily drugs versus 9 [IQR 7–11], p < 0.001).
Conclusions
Potential prescribing cascades were highly prevalent in older hospitalised adults. Practical tools are needed to assist prescribers in prevention, recognition and management of inappropriate prescribing cascades.
Supplementary Information
The online version contains supplementary material available at 10.1007/s40266-025-01201-9.
Key Points
| Prescribing cascades are clinically significant as they can result in prolonged burden of symptoms and accumulation of unnecessary medications which increases the risk of medication-related harm. |
| This study demonstrates that potential prescribing cascades are substantially prevalent in older adults with multimorbidity and polypharmacy. |
| Potential prescribing cascades are more common in patients with hyperpolypharmacy (prescribed ≥ 10 daily medications) and higher mean Charlson Comorbidity Index scores. |
Introduction
A prescribing cascade occurs when a new drug is prescribed to treat a side effect or an adverse drug event (ADE) caused by an existing medication and can be classified as either appropriate or inappropriate [1, 2]. Appropriate prescribing cascades occur when a new medication is intentionally prescribed to manage a well-recognized drug side effect, with the benefits of initiation outweighing the risk, such as laxatives to counteract the common and predictable problem of constipation arising from opioid treatment [3]. In contrast, inappropriate prescribing cascades occur when drug side effects are misdiagnosed as new conditions, leading to the inappropriate prescription of an unnecessary or potentially dangerous secondary medication to treat the misdiagnosed new condition or symptom, thereby increasing the risk of medication-related harm [3]. One well-known example of an inappropriate prescribing cascade is the prescription of diuretics to treat lower limb oedema arising as a side effect from prior treatment with dihydropyridine calcium channel blockers, the oedema being misdiagnosed as a sign of heart failure rather than as a side effect of the calcium channel blocker. Inappropriate prescribing of cascades may result in prolonged and excessive burden of symptoms arising from adverse drug events, avoidable additional medication burden and further inappropriate use of medications [4–6].
Numerous examples of prescribing cascades have been reported in literature through case reports, case series, case-control studies, cross-sectional studies and original research [7, 8]. The likelihood that a concurrent drug pair represents a prescribing cascade may be investigated through so-called prescription sequence symmetry analysis (PSSA) methodology [7, 9, 10]. PSSA compares the number of patients prescribed the index drug (drug A) before introduction of the marker drug (drug B) to those who were prescribed the marker drug before the index drug (so-called crude sequence ratio). PSSA may be affected by prescribing trends such that many studies also report a null-effect sequence ratio or adjusted sequence ratio, which is the crude sequence ratio divided by the null-effect sequence ratio, to account for prescribing trends. Although PSSA is an effective and validated method of establishing whether a drug pair constitutes a potential prescribing cascade, it cannot determine the actual clinical relevance of a suspected inappropriate prescribing cascade, nor cannot it establish causality [9].
A recent systematic review by Doherty et al., designed to identify published prescribing cascades in community-dwelling adults, examined 101 published studies. From their systematic analysis of literature, the authors highlighted the 25 most commonly identified prescribing cascades in community-dwelling adults, the majority of which were identified by studies that utilised PSSA methodology [7]. Separately, the iKASKADE consortium recently published a consensus list (called “ThinkCascades”) of nine clinically relevant prescribing cascades validated by the Delphi consensus method [11]. Not surprisingly, the list of the 25 most common prescribing cascades identified by Doherty et al. and the ThinkCascades list overlap to some extent. Nevertheless, they represent an advance in prescribing cascades research in that they embody explicit validated lists of cascades not previously described in literature.
Although it is nearly 30 years since the original concept of prescribing cascades was described by Rochon and Gurwitz, there are no studies describing the real-world prevalence of prescribing cascades in older multimorbid people exposed to polypharmacy, i.e. the population at highest risk of inappropriate prescribing cascades [12]. This deficiency in the prescribing cascades literature relates largely to the lack of validated explicit prescribing cascades lists prior to 2022. In addition, the need to carefully evaluate the probability of true cascades being observed has not been addressed through prospective prevalence studies in real-world multimorbid older patients experiencing polypharmacy. Older people undergoing hospitalization because of acute illness represent one of the most multimorbid patient cohorts who experience long-term polypharmacy [13]. Accordingly, the principal aim of this study was to investigate the prevalence of prescribing cascades in hospitalised older adults with multimorbidity and polypharmacy using the explicit prescribing cascades lists of Doherty et al. and ThinkCascades in combination.
Methods
This study was a single-centre observational prospective prevalence study conducted in a major tertiary referral medical centre in Cork, Ireland. The local Clinical Research Ethics Committee approved this research study. Participants were recruited between October 2022 and July 2024.
Eligible participants were adults aged ≥ 65 years admitted through the emergency department with acute surgical or medical illness, with preadmission multimorbidity (i.e. ≥ 3 chronic co-morbidities diagnosed prior to admission) and associated polypharmacy (i.e. ≥ 5 prescribed daily medications). Exclusion criteria were: acute psychiatric illness (not including delirium) and patients at end of life or admitted directly to the intensive care unit (ICU). Patients were also excluded if they or their legally authorised representative (if they lacked capacity) were unable or unwilling to give consent to study participation. Apart from their high prevalence of multimorbidity and polypharmacy, hospitalised older patients were selected as the focus of this study because hospital admission provides a unique opportunity to perform a comprehensive medication review, facilitating the identification of prescribing cascades [14].
To achieve a statistically valid detection of prescribing cascades, a sample size of 384 was determined. This was calculated on the basis of a margin of error of 5% at 95% confidence. This sample size was calculated using an assumed prevalence of 50%, an estimate that requires the largest sample size ensuring adequate power regardless of the true prevalence within the population. Furthermore, the sample size was calculated under the assumption of a large at-risk population (more than 5000) [15, 16].
Patients’ data were collected at two timepoints, i.e. at recruitment within 72 h of admission and at discharge. At recruitment, three trained research physicians (R.D., S.M., and L.K.) conducted patient and/or carer interviews, contacted patients’ community pharmacists and reviewed medical records to obtain the following patient information: age, sex, admitting speciality, long-term regular medications, admitting diagnosis, specific diagnoses and cumulative co-morbidity as quantified by the Charlson Co-morbidity Index score (CCI), number of reported falls in the preceding 12 months, number of admissions in the preceding 12 months, delirium status on admission based on the 4AT screening test and Clinical Frailty Scale (CFS) score. At discharge, the research physicians obtained the following information: length of hospital stay (days), discharge destination (home or long-term care facility) and updated list of regular medications.
Potential prescribing cascades were identified at the point of recruitment using the ThinkCascades list (Table 1) and the list compiled by Doherty et al. (hereafter referred to as the ‘Doherty list’) (Table 2). Patients’ long-term daily medications were screened for potential prescribing cascades using the two explicit cascades lists separately and in combination. When potential cascades were identified, the temporal relationship between drug A and drug B was determined, where ascertainable, through discussions with the community pharmacist and review of patients’ medical records. Establishing the temporal relationship involved firstly determining whether drug A was initiated before drug B and, where possible, assessing the timeframe between initiation of both drugs. If drug B was initiated after drug A, further investigation was conducted to determine whether an ADE was implicated as the potential indication for drug B. Patients were interviewed and their medical records carefully scrutinized, including correspondence letters and inpatient progress notes, to determine whether drug B was initiated to treat a known ADE caused by or relating to drug A. In this assessment, only ADEs documented on the ThinkCascades list and the Doherty list were considered as indicating potential cascades. Potential ADE’s were only included if the ADE was documented in the patients’ medical records. The likelihood of a prescribing cascade being present was categorized as ‘definite’, ‘probable’, ‘possible’, ‘uncertain,’ or ‘indeterminate’ according to pre-defined criteria (Table 3). This classification system was developed in alignment with the World Health Organisation-Uppsala Monitoring Centre (WHO-UMC) system for standardised case causality assessment [17]. Similar to the WHO-UMC’s ADR causality assessment, this system cannot provide an accurate quantitative measurement of the drug A/drug B cascade relationship likelihood, but it helps reduce assessor disagreement and improve the overall evaluation of potential prescribing cascades. The standardized process to ascertain prescribing cascade likelihood is illustrated in Fig. 1.
Table 1.
Prescribing cascades according to the ThinkCascades list
| No. | Drug A | Suspected ADE | Drug B | Total (probable/possible/ uncertain) | Probable cascade | Possible cascade | Uncertain cascade | Remains prescribed at discharge | Indeterminate cascade | Prescribed in the Sequence drug B→ drug A |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Calcium channel blocker | Peripheral oedema | Diuretic | 39 | 0 | 17 | 22 | 25 | 3 | 13 |
| 2 | Diuretic | Urinary incontinence | Overactive bladder medication | 30 | 4 | 17 | 9 | 19 | 2 | 16 |
| 3 | Antipsychotic | extrapyramidal symptoms | Antiparkinsonian agent | 3 | 1 | 2 | 0 | 3 | 0 | 2 |
| 4 | Benzodiazepine | Cognitive impairment | Cholinesterase inhibitor or memantine | 1 | 0 | 1 | 0 | 0 | 0 | 1 |
| 5 | Benzodiazepine | Paradoxical agitation or agitation secondary to withdrawal | Antipsychotic | 11 | 0 | 2 | 9 | 7 | 1 | 2 |
| 6 | SSRI/ SNRI | Insomnia | Sleep agent (e.g. benzodiazepine, benzodiazepine receptor agonist, sedating anti-depressant, melatonin) | 16 | 1 | 7 | 8 | 10 | 8 | 31 |
| 7 | NSAID | Hypertension | Antihypertensive | 2 | 0 | 2 | 0 | 1 | 0 | 1 |
| 8 | Urinary Anticholinergics | Cognitive impairment | Cholinesterase inhibitor or memantine | 2 | 0 | 2 | 0 | 1 | 0 | 0 |
| 9 | Alpha-1 Receptor Blocker | Orthostatic hypotension, dizziness | Vestibular sedative (e.g. betahistine, antihistamines, benzodiazepines) | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Total | 104 | 6 | 50 | 48 | 66 | 14 | 66 | |||
NSAID nonsteroidal anti-inflammatory drug; SSRI selective serotonin reuptake inhibitor; SNRI serotonin-norepinephrine reuptake inhibitor
Table 2.
Prescribing cascades according to the list published by Doherty et al.
| No. | Drug A | Suspected ADE | Drug B | Total (probable/possible/ uncertain) | Probable cascade |
Possible cascade |
Uncertain cascade |
Remains prescribed at discharge | Indeterminate cascade | Prescribed in the Sequence drug B→ drug A |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | DH-CCB | Oedema | Loop diuretic | 29 | 0 | 10 | 19 | 20 | 1 | 6 |
| 2 | Amiodarone | Hypothyroidism | Thyroxine | 2 | 1 | 1 | 0 | 2 | 0 | 1 |
| 3 | Inhaled CS | Oral candidiasis | Topical antifungals | 2 | 0 | 2 | 0 | 2 | 0 | 0 |
| 4 | Neuroleptics/antipsychotics | Parkinsonian symptoms/extrapyramidal symptoms | Anti-Parkinson’s meds/PD diagnosis | 3 | 1 | 2 | 0 | 3 | 0 | 2 |
| 5 | Acetylcholinesterase inhibitors | Urinary incontinence | Drugs for urinary frequency and incontinence/drugs for treating LUTS symptoms | 3 | 0 | 2 | 1 | 2 | 0 | 4 |
| 6 | Metoclopramide | Parkinsonian symptoms | Levodopa | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| 7 | ACEi | Cough | Antitussive | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| 8 | NSAID | GI symptoms | Anti-ulcer med | 0 | 0 | 0 | 0 | 0 | 0 | 3 |
| 9 | Ranitidine | Heart failure | Furosemide | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| 10 | Rosiglitazone | ‘Failure’ | Furosemide | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| 11 | SGLT2 | Genital infections | Antifungal | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| 12 | DOAC | Depression | Antidepressant | 21 | 0 | 2 | 19 | 16 | 1 | 28 |
| 13 | High ceiling diuretics (Loop diuretics) | Hypokalaemia | Potassium | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| 14 | Statins | LUTS | Drugs for urinary frequency and incontinence | 59 | 0 | 17 | 42 | 40 | 9 | 11 |
| 15 | Statins | Skin soft tissue infection | Antibiotics—dicloxacillin/flucloxacillin | 2 | 0 | 1 | 1 | 2 | 0 | 0 |
| 16 | Statins | Depression | Antidepressant | 70 | 0 | 10 | 60 | 60 | 5 | 21 |
| 17 | Statins | Muscle cramps | Quinine | 1 | 0 | 0 | 1 | 1 | 0 | 2 |
| 18 | Brinzolamide | Heart failure | Furosemide | 0 | 0 | 0 | 0 | 0 | 1 | 0 |
| 19 | Latanoprost | Heart failure | Furosemide | 3 | 0 | 1 | 2 | 2 | 0 | 1 |
| 20 | Carbamazepine | Hypothyroidism | Levothyroxine | 1 | 0 | 1 | 0 | 1 | 0 | 1 |
| 21 | Valproate | Hypothyroidism | Levothyroxine | 1 | 0 | 1 | 0 | 1 | 0 | 0 |
| 22 | Lithium | Drug induced tremor/parkinsonism | Anti-Parkinson drug | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| 23 | Lithium | Hypothyroidism | Thyroxine | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| 24 | Benzodiazepine | Dementia | Anti-dementia drug | 1 | 0 | 1 | 0 | 0 | 0 | 1 |
| 25 | SSRI | Urinary incontinence | Drugs for urinary frequency and incontinence (or incontinence products) | 12 | 0 | 10 | 2 | 7 | 0 | 4 |
| Total | 210 | 2 | 61 | 147 | 159 | 17 | 85 | |||
ACEi angiotensin-converting enzyme inhibitors; CS corticosteroid; DH-CCB, dihydropyridine calcium channel blocker; DOAC direct oral anticoagulant; GI gastro-intestinal; LUTS lower urinary tract infections; NSAID nonsteroidal anti-inflammatory drug; PD Parkinson’s disease; SSRI selective serotonin reuptake inhibitor; SGLT2 sodium-glucose cotransporter-2
Table 3.
Prescribing cascade likelihood criteria
| Cascade likelihood | Assessment criteria | |
|---|---|---|
| 1 | Definite |
Drug B follows Drug A Drug B initiated to treat ADE caused by drug A Initiation of drug B cannot be explained by other disease/indication Withdrawal challenge of drug A results in improvement/cessation of the side-effect |
| 2 | Probable |
Drug B follows drug A Drug B initiated to treat ADE caused by drug A Initiation of drug B unlikely explained by another disease or symptom |
| 3 | Possible |
Drug B follows drug A Drug B possibly initiated to treat ADE caused by drug A (documentation of ADE in medical records after drug A prescribed and before drug B prescribed) Initiation of drug B could be explained by other disease/indication |
| 4 | Uncertain |
Drug B follows drug A Unclear if drug B initiated to treat ADE caused by drug A (no documentation of ADE in medical chart) Initiation of drug B could be explained by other disease/indication |
| 5 | Indeterminate |
Drug A and drug B co-prescribed Cannot determine temporal relationship between drug A and drug B Data cannot be supplemented or verified |
ADE adverse drug event
Fig. 1.
Standardized process to ascertain prescribing cascade likelihood
Prevalence, the primary outcome of interest, was determined as the proportion of patients identified with a potential prescribing cascade, categorised as definite, probable, possible or uncertain. Patients were then classified into two groups—those with a potential prescribing cascade and those without—to evaluate statistical difference in demographic and clinical characteristics. The selected covariates were chosen on the basis of review of the relevant literature of related studies on inappropriate prescribing in older multimorbid adults [18–20].
At discharge, patients’ medications were reviewed to determine whether any potential prescribing cascades identified at recruitment were still included in their regular prescriptions.
Statistical Analysis
Statistical analysis was conducted using SPSS® version 28.0 software. Descriptive statistics included the mean and standard deviation (SD) for normally distributed variables and the median and inter-quartile range (IQR) for non-parametric variables. Deceased patients were excluded from the analysis of medication prescriptions at discharge. Differences in the distribution of categorical variables were compared using Pearson’s chi-squared test and continuous variables using the independent t-test. The Mann–Whitney U test was used to determine independence of two nonparametric variables. A probability value of < 0.05 was considered statistically significant. The probability values were selected to show the differences between the two groups – those with potential prescribing cascades and those without potential prescribing cascades – allowing for testing of the null hypothesis, which assumes no difference between the groups. To ensure strong reliability in the classification of potential prescribing cascades and ADE’s among the three researchers, inter-rater reliability was calculated using the intra-class correlation coefficient, a two-way mixed model, assessing for absolute agreement.
Results
In total, 385 patients were enrolled in this study, of whom 212 patients (55.1%) were female with a mean (SD) age of 80.2 (7.3) years. Baseline patient characteristics are presented in Table 4. The median (IQR) number of comorbid chronic conditions in patients with a potential prescribing cascade was 8 (6–10) versus 7 (5–9) in patients without a potential prescribing cascade (p = 0.003). The most common diagnoses were hypertension (72.7%), dyslipidaemia (50.6%) and atrial fibrillation (34.2%). A total of 3950 daily medications were prescribed to the 385 patients, i.e. an average of 10.3 daily drugs per patient. Statistically significant p values are highlighted in bold in Table 4.
Table 4.
Baseline characteristics and univariate analysis of association
| Variables | No cascade group (n = 233 (60.6%)) | Cascade group (n = 152 (39.4%)) | P-value |
|---|---|---|---|
| Female (%) | 131 (55.9) | 80 (52.6) | 0.438 |
| Age, median (years, IQR) | 80.5 (74.75–86) | 80 (75–85) | 0.95 |
| Living arrangements pre-admission | |||
| Home, number of patients (%) | 218 (93.6) | 139 (91.4) | 0.435 |
| Long-term care facility, number of patients (%) | 15 (6.4) | 13 (8.6) | |
| Charlson Comorbidity Index, mean (SD) | 5.57 (1.68) | 6.13 (2.06) | 0.006 |
| Number of daily medications, median (IQR) | 9 (7–11) | 12 (9–14) | < 0.001 |
| Hyperpolypharmacy, number of patients (%) | 87 (37.3) | 107 (70.3) | < 0.001 |
| Clinical frailty Scale score, median (IQR) | 5 (4–6) | 5 (4–6) | 0.476 |
| Comorbidities, median (IQR) | 7 (5–9) | 8 (6–10) | 0.003 |
| Fall in last 12 months, number of patients (%) | 123 (52.7) | 79 (52.0) | 0.875 |
| Admitted in last 12 months, number of patients (%) | 128 (55.0) | 88 (57.9) | 0.567 |
| 4AT score, median (IQR) | 1 (0–2) | 0 (0–2) | 0.886 |
| Admitted under geriatric medicine, number of patients (%) | 42 (18) | 18 (11.8) | 0.102 |
| Admitted under | |||
| Medical speciality, number of patients (%) | 218 (93.6) | 141 (92.8) | 0.760 |
| Surgical speciality, number of patients (%) | 15 (6.4) | 11 (7.2) | |
| Primary diagnosis (ICD code) | |||
| Fall (W19) (%) | 50 (21.5) | 29 (19.1) | 0.572 |
| Acute lower respiratory infection (J22) (%) | 22 (9.4) | 25 (16.4) | 0.040 |
| Urinary tract infection (N39.0) (%) | 17 (7.3) | 8 (5.3) | 0.429 |
| Syncope (R55) (%) | 14 (6) | 4 (2.6) | 0.125 |
| Decompensated congestive heart failure (I50.0) (%) | 13 (5.6) | 5 (3.3) | 0.298 |
| Length of hospital stay, days (IQR) | 9 (4–17) | 11 (5–21) | 0.276 |
| Survival outcome (death), number of patients (%) | 21 (9) | 10 (7) | 0.391 |
| Discharge destination | |||
| Home, number of patients (%) | 183 (78.5) | 113 (74.3) | 0.93 |
| Long-term care facility, number of patients (%) | 50 (21.5) | 39 (25.7) |
4AT Arousal, Attention, Abbreviated Mental Test—4 and Acute change
Using both the ThinkCascades list and the Doherty list in combination, a total of 281 individual potential prescribing cascades were identified in 152 patients, i.e. 39.4% prevalence, with 63 patients (16.3%) classified as having possible cascades and 7 patients (1.8%) classified as having probable cascades (Table 5). Owing to overlap between the Doherty list and the ThinkCascades list, the total number of unique patients and prescribing cascades in the combined analysis is lower than the sum of the individual lists. Each patient and prescribing cascade was counted only once. Cascades categorized as ‘indeterminate’ were excluded from overall prevalence calculations. The overall prevalence of potential prescribing cascades, including those classified as probable, possible and uncertain, is highlighted in bold in Table 5.
Table 5.
Prevalence of potential prescribing cascades
| Potential prescribing cascades (includes probable, possible and uncertain) | Probable prescribing cascades | Possible prescribing cascades | Uncertain prescribing cascades | Remains prescribed at discharge | Indeterminate (sequence drug A↔ drug B unknown | Prescribed in the sequence drug B→ drug A | ||
|---|---|---|---|---|---|---|---|---|
| Combined lists | Individual patients no. (%) | 152 (39.4%) | 7 (1.8) | 63 (16.3) | 101 (26.2) | 110 (28.6%) | 25 (6.5) | 80 (20.7) |
| No. individual prescribing cascades | 281 | 7 | 98 | 176 | 215 | 30 | 142 | |
| Doherty list | Individual patients no. (%) | 135 (35.1%) | 2 (0.5) | 46 (11.9) | 95 (24.7) | 107 (27.8) | 15 (3.9) | 64 (16.6) |
| No. individual prescribing cascades | 210 | 2 | 61 | 147 | 159 | 17 | 85 | |
| Think cascades list | Individual patients no. (%) | 75 (19.4%) | 6 (1.5) | 39 (10.1) | 39 (10.1) | 46 (11.9) | 11 (2.9) | 46 (11.9) |
| No. individual prescribing cascades | 104 | 6 | 50 | 48 | 66 | 14 | 66 |
The ratio of drug pairs prescribed drug A followed by drug B to those prescribed drug B followed by drug A was 1.57 (104/66) using the ThinkCascade list and 2.47 (210/85) using the Doherty list.
The four most common probable or possible prescribing cascades according to the Doherty list were:
statin → drugs for urinary frequency and incontinence (n = 17)
dihydropyridine calcium channel blocker → loop diuretic (n =10)
statin → antidepressant (n = 10)
selective serotonin reuptake inhibitor → drugs for urinary frequency and incontinence (or incontinence products) (n = 10).
By comparison, the four most common probable or possible prescribing cascades according to the ThinkCascades list were:
diuretic → overactive bladder medication (n = 21)
calcium channel blocker → diuretic (n = 17)
selective serotonin reuptake inhibitor/serotonin–norepinephrine reuptake inhibitor → sleep agent (e.g. benzodiazepine, benzodiazepine receptor agonist, sedating anti-depressant and melatonin) (n = 8)
antipsychotic→ antiparkinsonian agent (n = 3).
Potential prescribing cascades were more common in patients with hyperpolypharmacy (i.e. taking ≥ 10 daily long-term medications) compared with patients taking ≤ 9 daily medications (70.3% versus 37.3%, p < 0.001). In addition, patients exposed to potential cascades had higher mean Charlson Comorbidity Index (CCI) scores compared with patients not exposed to potential cascades (mean 6.13 [SD 2.06] versus 5.57 [SD 1.68]), p = 0.006). When comparing patients with a prescribing cascade categorised as ‘probable’ or possible’ in combination to ‘uncertain’ no significant differences were observed except for length of hospital stay (median: 7 days [IQR 3–13] versus 6 days [IQR 6.25–22], p = 0.018) (Appendix 1).
Hospital discharge prescriptions were reviewed to determine whether potential prescribing cascades identified at recruitment remained in patients’ regular prescriptions, i.e. whether both drug A and drug B were still prescribed. Using both cascades lists in combination, 76.5% (n = 215) of the identified potential prescribing cascades remained at discharge. In isolation, 75.7% (n = 159) of the Doherty list cascades and 63.4% (n = 66) of the ThinkCascades cascades remained at discharge.
Inter-rater reliability for prescribing cascade detection and ADE classification was also evaluated to ensure acceptable reliability among the three researchers. In total, 39 cases (10%) were selected at random for evaluation by the researchers. The intra-class correlation coefficient (ICC), a two-way mixed model, assessing for absolute agreement was calculated. For prescribing cascade classification, the ICC for the consistency of ratings was 0.935 (95% CI [0.884, 0.964]) and for consistency of ratings of ADE classification, ICC was 0.896 (95% CI [0.813, 0.943]), which indicates excellent inter-rater reliability for detection and classification of potential cascades and ADE’s.
Discussion
This is the first prospective prevalence study of potential prescribing cascade prevalence in hospitalised multimorbid older adults exposed to polypharmacy. It demonstrates that prescribing cascades are substantially prevalent in this population to a clinically significant degree, particularly among those who experience hyperpolypharmacy. The analysis of drug pairs where drug A was prescribed followed by drug B compared with the reverse order suggests that the suspected prescribing cascades identified in this study are likely to be clinically valid. Although the more frequent prescribing in the sequence drug A → drug B does not definitively confirm their status as prescribing cascades, it supports the hypothesis that the potential prescribing cascades identified in this study are true prescribing cascades. This is important, as true prescribing cascades can have serious adverse consequences for patient safety and healthcare outcomes [5].
As ageing demographic trends intensify as a global phenomenon, the prevalence of multimorbidity is increasing in tandem, resulting in ever increasing levels of polypharmacy and hyperpolypharmacy [21–23]. As older people are prescribed more medications, their risk of experiencing inappropriate prescribing (IP) in general also rises [13, 24]. The present study aligns with this observed association, confirming that potential prescribing cascades as another manifestation of IP are associated with a higher number of prescribed medications [25].
While the management of patients with complex multimorbidity and associated polypharmacy has traditionally been the responsibility of general practitioners and geriatricians, the responsibility for medication management will increasingly extend to all physicians dealing with multimorbid older people in their own areas of clinical practice. In the present study, 84% (n = 326) of patients were under the care of physicians who were not geriatricians, i.e. specialists without specific training in age-related multimorbidity and associated complex polypharmacy. Without the necessary training and experience in geriatric medicine, these physicians may struggle to recognise prescribing cascades, as was evident in the present study where very little curtailment of potential cascades was evident between admission and discharge, possibly related to lack of recognition of potential cascades. This challenge is compounded by the fact that ADEs in older adults with multimorbidity are more challenging to recognise and can often present with non-specific symptoms, such as new-onset confusion, gastrointestinal upset and dizziness [4, 26, 27].
A recent systematic review by Adrien et al. assessed recommendations for addressing prescribing cascades [10]. Fewer than half of the studies included in the review provided specific guidance on reversing inappropriate prescribing cascades. Among those that did, recommendations for dose reductions were often made without confirmation of a dose-dependent association, with most studies recommending the discontinuation of the index drug but not specifying an alternative medication. The persistence of potentially problematic drug combinations at hospital discharge identified in this study highlights the challenge physicians face in recognising and addressing prescribing cascades. It also points to the need for expert medication review at admission and particularly at discharge of older patients with multimorbidity and associated polypharmacy.
The identification and management of prescribing cascades are further complicated by the limitations of existing assessment tools. The ThinkCascades list, developed through three rounds of a modified Delphi validation process and involving international experts in geriatric pharmacotherapy, produced a list of only nine common and clinically important prescribing cascades [11]. This list very likely underrepresents the true number and prevalence of clinically relevant prescribing cascades. Although the Doherty list includes the 25 most commonly published appropriate and inappropriate prescribing cascades [7], some of the prescribing cascades listed raise questions about some of the cascades’ frequency and clinical relevance. Examples include lower urinary tract symptoms arising from statins leading to drugs for urinary frequency/incontinence and depression arising from direct oral anticoagulants (DOACs) leading to antidepressant drug prescription. These examples highlight the limitations of relying heavily on statistical methods such as PSSA to define cascades without clinical data, expert endorsement and clinical plausibility to validate proposed prescribing cascades. Nevertheless, a recent follow-up study evaluating the strength of the evidence supporting the prescribing cascades identified by Doherty et al. found that 84% of these prescribing cascades were supported by moderate evidence, i.e. single high-quality study or strong evidence, i.e. multiple high-quality studies [28].
Most prescribing cascades in the present study were categorized as ‘possible’ or ‘uncertain’. Accurate identification of the prevalence of probable or definite prescribing cascades using the criteria applied in this study is likely to be challenging during medication review. Categorization of any clinical event as ‘probable’ or ‘certain’ in relation to multimorbid older patients requires rigorous assessment and therefore, as we have demonstrated, reliance on patient recall and physician or pharmacist documentation in patients’ medical records may not be sufficient. A prospective study design involving contemporaneous assessment of patients’ symptoms, as well as prescriber interviews, at the onset of possible prescribing cascades would be required for greater clarity. The feasibility of conducting such a labour-intensive study, likely with major recruitment challenges and resource implications, was beyond the scope of the present study.
A more comprehensive validated explicit list of clinically relevant prescribing cascades is likely required to enhance identification during routine medication review. In a recent commentary, O’Mahony and Rochon pointed out that the number of inappropriate prescribing cascades is potentially enormous but that more clinically relevant and common cascades can be identified more readily using a systems-based structured explicit cascades list approach that can be applied readily in a manner that is more suitable for routine medication review [29]. Such a structured, systems-based list of prescribing cascades could also be valuable as an educational tool in this important area of geriatric pharmacotherapy. Heightened awareness of potential cascades is likely to facilitate their prevention.
Study Limitations
This study has some limitations. First, the absence of a national electronic healthcare system in most public hospitals in Ireland, coupled with frequent transitions of care and missing medication initiation dates, made reconstruction of the chronology of potential prescribing challenging and at times wholly uncertain. This limited our ability to establish clear timelines between drug A and drug B initiation in some cases. However, there is no consensus in literature on the appropriate timeframe for defining prescribing cascades, which can range from a short number of weeks to several years. The timeframe can also vary according to the onset of ADE symptoms, which may vary between medications, as well as the recognition of ADEs which may depend on the timing of patients’ presentation to healthcare settings. For example, it may take many months to some years before antipsychotic medication leads to drug-induced parkinsonism leading to antiparkinsonian medication. In contrast, gastrointestinal upsets induced by acetylcholinesterase inhibitor therapy may lead to inappropriate proton pump inhibitor or antidiarrheal prescription within days. Second, as mentioned, patients’ medical records in the Irish public hospital system are still predominantly paper-based, with no widespread integrated electronic platform between community healthcare settings and hospital settings. This resulted in incomplete access to clinical data, particularly that relating to general practitioner assessments. Third, this study was conducted in a single hospital, which may limit the generalizability of the finding to other hospitals or healthcare settings. Finally, as this was an observational rather than an interventional study, we were unable to apply the definitive test for prescribing cascades, i.e. observation for resolution of ADE symptoms by discontinuing drug A.
Conclusions
This study, the first of its kind to report the prevalence of prescribing cascades in multimorbid older people exposed to polypharmacy in the acute hospitalization setting, indicates that inappropriate prescribing cascades are both prevalent and poorly recognised in hospitalised older adults in southern Ireland. Explicit prescribing cascades lists can facilitate prescribers and medication reviewers with identification and intervention towards curtailment of this under-reported element of inappropriate prescribing in multimorbid older people. Further research is needed to help the development of practical tools to support prescribers in the prevention, identification and management of inappropriate prescribing cascades.
Supplementary Information
Below is the link to the electronic supplementary material.
Declarations
Funding
Open Access funding provided by the IReL Consortium. No funds, grants, or other support was received.
Conflicts of Interest
Denis O’Mahony is an Editorial Board member for Drugs & Aging. Denis O’Mahony was not involved in the selection of peer reviewers for the manuscript nor any of the subsequent editorial decisions. The authors; Ruth Daunt, Siobhán McGettigan, Lorna Kelly and Denis Curtin have no competing interests to declare that are relevant to the content of this article.
Availability of Data and Material
The datasets generated and/or analysed in this study are available from the corresponding author on reasonable request.
Ethical Approval
This study was approved by the local Clinical Research Ethics Committee, Cork, Ireland.
Consent to Participate
Written informed consent was obtained prior to enrolment from all participants or their legally authorised representative in circumstances where they were deemed to lack sufficient decision-making capacity to provide valid informed consent.
Consent for Publication
Not applicable.
Code Availability
Not applicable.
Author Contributions
All authors contributed to the study conception and design. Data collection was performed by Ruth Daunt, Siobhán McGettigan and Lorna Kelly. Material preparation and analysis were performed by Ruth Daunt. The first draft of the manuscript was written by Ruth Daunt and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.
References
- 1.McCarthy LM, Visentin JD, Rochon PA. Assessing the scope and appropriateness of prescribing cascades. J Am Geriatr Soc. 2019;67:1023–6. [DOI] [PubMed] [Google Scholar]
- 2.Rochon PA, Gurwitz JH. The prescribing cascade revisited. Lancet. 2017;389:1778–80. [DOI] [PubMed] [Google Scholar]
- 3.Sternberg SA, Gurwitz JH, Rochon PA, et al. Prescribing cascades. In: Cherubini A, Mangoni AA, O’Mahony D, et al., editors. Optimizing pharmacotherapy in older patients: an interdisciplinary approach. Cham: Springer International Publishing; 2023. p. 59–68. [Google Scholar]
- 4.Rochon PA, Gurwitz JH. Optimising drug treatment for elderly people: the prescribing cascade. BMJ. 1997;315:1096. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Piggott KL, Mehta N, Wong CL, et al. Using a clinical process map to identify prescribing cascades in your patient. BMJ. 2020;368: m261. [DOI] [PubMed] [Google Scholar]
- 6.Gill SS, Mamdani M, Naglie G, et al. A prescribing cascade involving cholinesterase inhibitors and anticholinergic drugs. Arch Intern Med. 2005;165:808–13. [DOI] [PubMed] [Google Scholar]
- 7.Doherty AS, Shahid F, Moriarty F, et al. Prescribing cascades in community-dwelling adults: a systematic review. Pharmacol Res Perspect. 2022;10: e01008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Chen Z, Liu Z, Zeng L, et al. Research on prescribing cascades: a scoping review. Front Pharmacol. 2023. 10.3389/fphar.2023.1147921. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Morris EJ, Hollmann J, Hofer A-K, et al. Evaluating the use of prescription sequence symmetry analysis as a pharmacovigilance tool: a scoping review. Res Soc Adm Pharm. 2022;18:3079–93. [DOI] [PubMed] [Google Scholar]
- 10.Adrien O, Mohammad AK, Hugtenburg JG, et al. Prescribing cascades with recommendations to prevent or reverse them: a systematic review. Drugs Aging. 2023;40:1085–100. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.McCarthy LM, Savage R, Dalton K, et al. ThinkCascades: a tool for identifying clinically important prescribing cascades affecting older people. Drugs Aging. 2022;39:829–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Rochon PA, Gurwitz JH. Drug therapy. Lancet. 1995;346:32–6. [DOI] [PubMed] [Google Scholar]
- 13.Davies LE, Spiers G, Kingston A, et al. Adverse outcomes of polypharmacy in older people: systematic review of reviews. J Am Med Dir Assoc. 2020;21:181–7. [DOI] [PubMed] [Google Scholar]
- 14.Daunt R, Curtin D, O’Mahony D. Polypharmacy stewardship: a novel approach to tackle a major public health crisis. Lancet Healthy Longev. 2023;4:e228–35. [DOI] [PubMed] [Google Scholar]
- 15.Pourhoseingholi MA, Vahedi M, Rahimzadeh M. Sample size calculation in medical studies. Gastroenterol Hepatol Bed Bench. 2013;6:14. [PMC free article] [PubMed] [Google Scholar]
- 16.Naing L, Nordin RB, Abdul Rahman H, et al. Sample size calculation for prevalence studies using Scalex and ScalaR calculators. BMC Med Res Methodol. 2022;22:209. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.World Health Organization. The use of the WHO-UMC system for standardised case causality assessment; 2013. https://www.who.int/publications/m/item/WHO-causality-assessment. Accessed 07 February 2025.
- 18.Khezrian M, McNeil CJ, Murray AD, et al. An overview of prevalence, determinants and health outcomes of polypharmacy. Ther Adv Drug Saf. 2020;11:2042098620933741. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Alhawassi TM, Alatawi W, Alwhaibi M. Prevalence of potentially inappropriate medications use among older adults and risk factors using the 2015 American Geriatrics Society Beers criteria. BMC Geriatr. 2019;19:154. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Petrovic M, O’Mahony D, Cherubini A. Inappropriate prescribing: hazards and solutions. Age Ageing. 2022;51:afab269. [DOI] [PubMed] [Google Scholar]
- 21.World Health Organisation. Multimorbidity; 2016. http://apps.who.int/iris/handle/10665/252275. Accessed 7 October 2024.
- 22.The Lancet. Making more of multimorbidity: an emerging priority. Lancet. 2018;391:1637. [DOI] [PubMed] [Google Scholar]
- 23.Hanlon P, Nicholl BI, Jani BD, et al. Frailty and pre-frailty in middle-aged and older adults and its association with multimorbidity and mortality: a prospective analysis of 493 737 UK Biobank participants. Lancet Public Health. 2018;3:e323–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Steinman MA, Hanlon JT. Managing medications in clinically complex elders: “there’s got to be a happy medium.” JAMA. 2010;304:1592–601. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Brath H, Mehta N, Savage RD, et al. What is known about preventing, detecting, and reversing prescribing cascades: a scoping review. J Am Geriatr Soc. 2018;66:2079–85. [DOI] [PubMed] [Google Scholar]
- 26.Lavan AH, Gallagher P. Predicting risk of adverse drug reactions in older adults. Ther Adv Drug Saf. 2016;7:11–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Petrovic M, van der Cammen T, Onder G. Adverse drug reactions in older people. Drugs Aging. 2012;29:453–62. [DOI] [PubMed] [Google Scholar]
- 28.Shahid F, Doherty A, Wallace E, et al. Prescribing cascades in ambulatory care: a structured synthesis of evidence. Pharmacother J Hum Pharmacol Drug Ther. 2024;44:87–96. [DOI] [PubMed] [Google Scholar]
- 29.O’Mahony D, Rochon PA. Prescribing cascades: we see only what we look for, we look for only what we know. Age Ageing. 2022;51:afac138. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.