Evaluating the use of prescription sequence symmetry analysis as a pharmacovigilance tool: A scoping review
Introduction
The (prescription) symmetry sequence analysis (PSSA) method is a pharmacovigilance tool employed to rapidly identify adverse drug event adverse drug event signals and potential prescribing cascades within large administrative health databases.1 Prescribing cascades occur when an adverse drug event is misinterpreted as a new medical condition, prompting practitioners to prescribe a new medication to treat effects caused by the first medication.2 This often suboptimal prescribing practice increases a patient's risk of adverse drug events, either from the potentiation of adverse drug events caused by the first medication or from additional adverse drug events from the new medication.2 Polypharmacy, which may be a result of prescribing cascades,3 has also been associated with an increased risk for drug-drug interactions and can negatively impact patients' adherence to their current medications.4, 5, 6, 7, 8 Furthermore, prescribing cascades, especially when considered problematic, have the potential to increase healthcare utilization and costs due to the increased medication burden among patients.3
The PSSA method exists within the family of self-controlled methods and employs a case-only design including only patients who are users of both drugs of interest (i.e., index drug and marker [e.g., outcome] drug).9,10 Conceptually, PSSA is applied to investigate a scenario where drug B (marker drug) is prescribed to treat an adverse drug event possibly caused by drug A (index drug).9 For example, the PSSA method was first used to investigate whether beta-blockers (index drug) have depression-provoking effects and lead to an excess risk of being prescribed antidepressants (marker drug).1 In assessing the association between the index drug and marker drug, the PSSA uses a population of new users of both index drug and marker drug within a given timeframe and compares the number of subjects who used the index drug before the marker drug to the number of users who used the marker drug before the index drug.1 Notably, individuals who initiate both the index drug and marker drug on the same day are not included in the analysis as it is impossible to distinguish a prescribing order for such individuals.9,10 Therefore, risk is estimated by calculating the ratio (crude sequence ratio [cSR]; Fig. 1) of subjects who initiate index drug before marker drug to subjects who initiate marker drug before index drug, while accounting for secular trends of prescribing of these medications over time (adjusted sequence ratio [aSR]; Appendix A).
If there is no association between index drug and the initiation of the marker drug, symmetry is expected as there should be equal chances of initiating the marker drug before or after the index drug. However, if the index drug does indeed increase the risk of an adverse drug event leading to the initiation of a marker drug, asymmetry will occur with an accompanying aSR and lower limit of confidence interval (CI) greater than 1. Because PSSA includes only those individuals who are users of both the index drug and marker drug, time-invariant patient characteristics like age, sex, and other demographic and environmental factors are inherently controlled.1 In other words, factors that are stable over time within the study time window cannot predict the prescribing order of index drug and marker drug for a given individual, since every individual included in the study is required to be started on both index drug and marker drug.1
The objective of this scoping review is to explore published PSSA studies and associated index drug – marker drug dyad signals to expand the current list of potential prescribing cascades.11,12 Additionally, we hope to describe the utility and methodological advances of PSSA to be used to investigate future potential prescribing cascades in pharmacovigilance studies.
Section snippets
Methods
This scoping review adhered to the PRISMA Extension for Scoping Reviews (PRISMA-ScR) checklist.13
Included Prescription Sequence Symmetry Analysis Studies
The flowchart of study inclusion is illustrated in Fig. 2. A total of 678 studies were identified during the study period across all databases and through reference screening. After removing duplicates and screening for eligibility, 53 studies remained that met the inclusion criteria.1,15−55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66 Among those, 47 studies reported original PSSA results (e.g., aSRs for signals evaluated/detected),1,15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30
Discussion
In this review, we identified a number of PSSA studies which aimed to detect or evaluate prescribing cascades or aimed to investigate the utility of this methodology as a pharmacovigilance tool. Other methods have been used to identify and evaluate prescribing cascades, including cohort studies, cross-sectional studies, and case reports.67, 68, 69, 70, 71, 72, 73, 74 However, the PSSA design has a number of key characteristics that contribute to its utility in detecting and evaluating
Conclusion
The PSSA methodology has been used as a pharmacovigilance tool to screen for potential prescribing cascades using prescription data. Since its development, over 300 prescribing cascade signals have been detected and/or evaluated. Nevertheless, key considerations should be taken into account when designing a study using the PSSA design with particular attention on sensitivity analyses to evaluate time-varying biases. Furthermore, high-quality reporting of PSSA results is important to provide
CRediT author statement
Earl J. Morris: Conceptualization, Investigation, Formal Analysis, Writing – Original Draft, Review & Editing, Visualization; Josef Hollmann: Conceptualization, Investigation, Writing – Review & Editing; Ann-Kathrin Hofer: Investigation, Writing – Review & Editing; Hemita Bhagwandass: Investigation, Writing – Review & Editing; Razanne Oueini: Investigation, Writing – Review & Editing; Lauren E. Adkins: Data Curation; Jesper Hallas: Conceptualization, Writing – Review & Editing; Scott M. Vouri:
Funding
The research presented in this manuscript did not receive funding from any public, private, or not-for-profit agency.
Declaration of competing interest
The authors have declared no conflicts of interest for this article.
References (86)
- et al.
Polypharmacy, adverse drug reactions, and geriatric syndromes
Clin Geriatr Med
(2012) - et al.
Polypharmacy in the elderly: focus on drug interactions and adherence in hypertension
Clin Geriatr Med
(2009) - et al.
Isotretinoin and antidepressant pharmacotherapy: a prescription sequence symmetry analysis
J Am Acad Dermatol
(2003) - et al.
Supervised signal detection for adverse drug reactions in medication dispensing data
Comput Methods Progr Biomed
(2018) - et al.
Lithium use in older adults is associated with increased prescribing of Parkinson medications
The Ameican Journal of Geriatric Psychiatry
(2016) Evidence of depression provoked by cardiovascular medication: a prescription sequence symmetry analysis
Epidemiology
(1996)- et al.
Optimising drug treatment for elderly people: the prescribing cascade
BMJ
(1997) - et al.
Assessing the scope and appropriateness of prescribing cascades
J Am Geriatr Soc
(2019) - et al.
Adverse drug events in the outpatient setting: an 11‐year national analysis
Pharmacoepidemiol Drug Saf
(2010) - et al.
Medication non-adherence among elderly patients newly discharged and receiving polypharmacy
Drugs Aging
(2014)