Legacy Drug-Prescribing Patterns in Primary Care

The overall purpose of the study was to examine the prevalence of legacy pharmacy prescriptions for adult patients in a primary care setting. The authors explored prescription timelines for bisphosphonates, antidepressants and proton pump inhibitors (PPIs) and hypothesized that many of these medications have extended prescriptions (termed "legacy prescriptions) outside of their intended therapeutic duration. The authors claimed that the extended use of these drugs contribute to polypharmacy and prescribing cascades.

This study uses a retrospective cohort design to analyze identified data on 50,813 patient demographics and prescriptions collected over 6 years from the McMaster University Sentinel and Information Collaboration network in Hamilton, Ontario, which encompassed a wide range of socioeconomic statuses. The group discussed that this was a benefit of doing this study in Canada, due to the same insurance provider. Patients were included if they had taken antidepressants, bisphosphonates, and PPIs, which was selected by Anatomic Therapeutic Council (ATC) coding. Use of antidepressants for 15 months, bisphosphonates for 5.5 years, or PPIs for 15 months minimum were designated legacy prescriptions.

We found that the inclusion of bisphosphonates in this particular study to be inappropriate in comparison to the other two medications, given the researchers defined legacy prescription for bisphosphonates to be greater than 5.5 years (compared to PPIs and antidepressants greater than 1.5-year window) and data-collection for the study was just a 6-year window. The difference between legacy prescription qualifications and the timeframe for data collection was too narrow in the opinion of the group to appropriately assess legacy prescriptions involving bisphosphonates despite the researchers' reported conservative definitions for legacy status.

The group also discussed that the grouping all of the antidepressants (with the exception of MAO-Is) is problematic since there are multiple types of antidepressants with different indications and side effects. Also, the fact that MAO-Is are excluded was interesting. In the next study the antidepressant classes should be stratified, and/or MAO-Is should be included for completeness of assessment. Each class of antidepressants used should also be categorized by the diagnoses associated with each antidepressant prescription. SSRI use for premenstrual syndrome should also be taken into account or event excluded from future studies as prescriptions in this case are for 1 to 2 weeks per month while symptoms persist. In addition, PPIs could be prescribed as PRN, but have a long prescribing date, and it is unknown if those prescriptions should be considered legacy.

Since there were few methods of evaluating legacy prescriptions, the authors developed their own method of summing the difference as well as the stop-start duration. Summing the difference involved calculating the length of each prescription, and stop-start duration consisted of calculating the difference between the first-ever start date and last-ever stop date for each drug. Using both resulted in the most accurate selection of legacy prescriptions, and prescriptions that only fit one or the other method were excluded. The group found that this was an acceptable method of determining legacy prescribing.

While the rate of legacy prescribing was about 40% for antidepressants and PPIs, the group found it interesting that bisphosphonates had 14% legacy and 73% non-legacy prescriptions, possibly due to the relatively short course of data collected vs threshold for legacy prescriptions. The group also found it interesting that the highest percentage (17%) of dual legacy prescribing consisted of both antidepressants and PPIs, likely due to the effects of antidepressants on the GI system. Patients, that had received legacy prescriptions at some point, were also much more likely to still be on the legacy prescription at the end of data collection, and the group discussed that this was most likely due to physicians unwilling to change medications when patients are stable.

The group also felt that the study could have improved with more information about prescriptions, such as: age group (especially for antidepressants and bisphosphonates); gender (increased risk of osteoporosis and depression in women); and dosage to determine the necessity of the legacy prescription. However, this may not have been possible with a large data set. The group also considered the possibility that prescriptions are not updated on patient records or low patient compliance with regular visits, which could cause a higher legacy prescribing percentage. However, there is not enough information given by the authors to make this conclusion.

Regardless of outlying and confounding factors, the exogenous amount of legacy prescriptions cannot be denied. We believe that this study will raise awareness and cause providers to review medication prescriptions and update medication lists within the hospital EMR. This study, although limited to only 3 drug groups, is pertinent when considering other chronically prescribed medications in which it is important to be mindful of adverse side effects associated with long-term use. Additionally, patients and providers should be educated and stay up-to-date on prescription recommendations, treatment timelines, and associated side effects at various time-points of treatment.

Future studies can be conducted as clinical trials in different outpatient settings, with the primary outcome being the difference in legacy prescribing following the implementation of different policies, mainly in clinics with continuity of care which will have better compliance. Possible methods to decrease legacy prescribing include provider education, follow-up appointment notifications linked with long- term prescriptions in the computer system, or making long-term prescriptions with multiple refills more difficult to obtain. Physician education could entail reinforcing when to stop prescriptions or a more complete medication history. In addition, certain prescriptions may be needed to be taken over long periods of time, for example, patients with relapsing depression in a short time period may need to be on SSRIs for 15 months or more. Finally, when prescribing medications with higher percentages of legacy prescriptions, patients could be given educational materials about the long-term use of medications and when it is most often appropriate to discontinue the medications pending provider approval. These are possible approaches to decreasing legacy prescribing, but it is unknown if they will address the problem and will need to be further studied.

References

Cheung KS, Chan EW, Wong AYS, et al. Long-term proton pump inhibitors and risk of gastric cancer development after treatment for Helicobacter pylori: a population-based study. Gut. 2018;67(1):28-35. doi: 10.1136/gutjnl-2017-314605

Hengartner MP, Angst J, Rossler W. Antidepressant use prospectively relates to a poorer long-term outcome of depression: results from a prospective community cohort study over 30 years. Psychother Psychosom. 2018;87(3):181-183. doi: 10.1159/000488802

Mangin D, Lawson J, Cuppage J, et al. Legacy drug-prescribing patterns in primary care. Ann Fam Med. 2018; 16(6):515-520. doi: 10.1370/afm.2315

Watts NB. Long-term risks of bisphosphonate therapy. Arq Bras Endocrinol Metab. 2014;58(5):523-529. doi: 10.1590/0004-2730000003308

Competing interests: None declared

Congratulations on this thoughtful piece - it is interesting, informative, and important. I have a few thoughts to share:

The term "legacy prescribing" is great, although I wonder if it is not pointed enough to get the idea across compared to a more judgmental term such as "prescriptions of excessive duration" or something similar. On a related note, there is some emerging work around deprescribing that suggests that patients are put off by this term ("deprescribing"), and as the field moves forward it will be useful to ensure that the relevant actors understand and buy in to the concepts revealed through language.

The validation work on methods was a highly useful addition, although I have one question: how were refills accounted for in the prescription data? For example, if I prescribed a 30-day supply of a PPI with 11 refills, and a month later told the patient to stop taking the medication, is there any way of accounting for this or would this patient be counted as taking the medication for the full year? And, are there fields for indicating whether a medication is discontinued? There are some companies that are working on ensuring that discontinuation orders are communicated to community pharmacies in the U.S. (for prescribers who don't work in closed health systems), and major gaps remain, but I don't know if there are better systems in Canada.

The results are very interesting, although sadly not surprising. And, as the authors noted there was likely undercounting of bisphosphonate legacy prescriptions given the duration of available data. For next steps, it would be particularly interesting to dig deeper into the details of who has legacy prescribing, how many of these prescriptions are actually inappropriate, and what factors drive failure to discontinue these medications.

Thanks for sharing this important work.

Competing interests: None declared