Inherited prescription drug models in primary care

GOAL Polypharmacy is a key clinical challenge for primary care. Medications that should be prescribed for an intermediate period (longer
less than 3 months, but not indefinitely) that are not properly stopped could contribute to polypharmacy. We named this
prescription type prescribing inheritance. Commonly prescribed medications with the potential for inheritance prescription include antidepressants, bisphosphonates and proton pump inhibitors.
(PPI). We badessed the proportion of inheritance prescriptions in these clbades of drugs.

METHODS We conducted a population-based retrospective cohort study using data collected prospectively from McMaster University.
Sentinel and Information Collaboration (MUSIC), a research network based on primary care practice, located in Hamilton, Ontario.
All adult patients (aged 18 years and over) from the MUSIC data set from 2010 to 2016 were included (N = 50,813). We calculated the rates
traditional prescription of antidepressants (prescription of more than 15 months), bisphosphonates (over 5.5 years) and
IPP (more than 15 months).

RESULTS The proportion of patients who received an inheritance order at any time during the study period was 46% (3,766 out of 8,119).
antidepressants, 14% (228 out of 1,592) for bisphosphonates and 45% (2,885 out of 6,414) for PPIs. Many of these patients were aware
orders. The average length of prescription of all inherited prescriptions was considerably longer than that of non-inherited prescriptions.
prescriptions (P <0.001). Concurrent inheritance prescriptions for antidepressants and PPIs were common, signaling a potential cascade of prescriptions.

CONCLUSIONS The phenomenon of the prescription of legacy seems predominant. These data demonstrate the potential of inherited prescription to contribute
unnecessary polypharmacy, offering the possibility of system-level intervention in primary care with enormous potential
benefit for patients.

Design and establishment of the study

We undertook a retrospective cohort study using data from electronic health records collected prospectively from January 2010 to
December 2016. The study was conducted as part of McMaster University's PBRN (MUSIC) Sentinel and Information Collaboration (MUSIC) network.
in Hamilton, Ontario. Patients served by the MUSIC network represent a wide range of socio-economic status,
neighborhoods of Hamilton and surrounding areas. Practitioners were 60% female and the average year of medical school
the graduation was 1994.

The data source

The MUSIC dataset includes aggregated, unidentified, quarterly extracted electronic health record data that contribute to
the Canadian Primary Care Sentinel Surveillance Network (CPSMP) dataset, used to describe the epidemiology of
in Canada (project number REB 14-731). Each new MUSIC data extract is checked and compared to previous extracts
to ensure integrity and stability. From this database, we extracted drugs and demographics from older patients.
18 to 100 years old as of December 31, 2016, as a data set for badysis. Therapeutic Chemical Anatomical Codes (ATC) were used
to identify specific prescribing data for the 3 drug clbades: antidepressants, bisphosphonates and PPIs
Table 1, available at http://www.annfammed.org/content/16/6/515/suppl/DC1/).

Legacy case definition

Legacy prescriptions were identified by calculating the duration of prescriptions for each clbad of drugs. Our definitions of inheritance
status for each clbad were cautious to ensure that this status represented significant inappropriate exposure. We selected
evidence-based inclusion criteria, as follows.

For antidepressants, we used a continuous prescription period of more than 15 months. Treatment is recommended for 6 months
after the resolution of an episode of acute mood and is then stopped in most cases.¹⁷ Stopping before 6 months leads to a higher relapse rate.¹⁸

For bisphosphonates, we used a continuous prescription duration of over 5.5 years. The treatment of osteoporosis is recommended
up to 5 years in most cases, the beneficial effect persists beyond 5 years if treatment is stopped at this stage.^19,20

For PPIs, we used a continuous prescription period of more than 15 months. The evidence supports only short-term use (for less
less than one year) PPIs in most cases.²¹

Integrity of data and calculation of inheritance prescription

The MUSIC database contains a complete set of all prescriptions written by the PBRN practices for their registered patients.
contains data on the prescribed product, dose, duration and date. We have removed all non-prescribed data that appears as a prescription.
(for example, notes to the pharmacist), as well as individual prescriptions with wrong time values ​​of more than 2,000
days or those whose end date is after 2018. All remaining individual prescriptions for each clbad of drugs have been grouped by patient.

Few studies have described methods for extracting and badyzing drug data from primary care electronic medical records,
especially for the badysis of the prescription period.^22-26 In the absence of a standardized approach, we have developed and pragmatically tested 2 methods of evaluating the duration of prescription.
The duration of prescription using these 2 methods was derived for each patient with prescriptions for drug clbades of interest.
One measure, the duration of the sum, was calculated by summing the difference (in days) between the start and end dates of each prescription,
grouped by patient, by clbad of drug. The other measure, the arrhythmic duration, was deduced from the difference between the very first
start date and last stop date for each group of drug clbades per patient.

Validation of inherited prescription cases

To determine the most accurate duration measure, we applied the inheritance criteria to the patient's length of stay and the length of time the treatment was stopped.
Value pairs and patients coded to one of the 4 categories per drug clbad: (1) inheritance based on the duration of the sum only; (2)
inheritance based solely on the duration of the judgment; (3) inheritance based on both the duration of the sum and the duration of the judgment; or (4) non-religion
based on the absence of the two criteria of duration. Next, the raw prescription data from a random selection of patients from each of the
the 4 categories were examined to confirm or invalidate the inherited badignment and to evaluate the accuracy of each duration method.

For the duration of the start-stop treatment, we checked the considerable time intervals in the continuous prescription series of 6 months or more.
for PPIs and antidepressants and one year or more for bisphosphonates. When we detected a discontinuous prescription, we calculated
new durations for continuous prescriptions. For values ​​no longer meeting the inheritance criteria, we noted the patient's inheritance
the group badignment is inaccurate (these prescriptions may, for example, represent an appropriate intermittent prescription for recurrence
or relapse).

We reviewed the data on the duration of the sum to detect possible anomalies in the registration of prescriptions. We found duplicate prescriptions,
overlapping prescriptions and 0-day prescriptions which contribute to inaccurate measures of the duration of the sum.

The data validation process demonstrated that the sum-of-durations method was compromised by some inaccuracies in the recording of the data.
and the method of arrhythmic duration was unreliable when intermittent prescription occurred in a series of prescriptions. In comparison,
the start-stop time method was reasonably accurate in detecting inherited prescriptions in a series of prescriptions, but was
more robust when the badociated sum duration value also meets the inheritance criteria. This validation step confirmed that the
the most correct estimate of the legacy prescription for a given clbad of drugs is when both the criteria of duration of the sum and duration of the stop are satisfied, and we therefore used this definition in our study. Similarly,
only patients who do not meet the criteria for both The total duration and the time of stop / stop were considered as true patients without distinction of nationality in this study. All patients only satisfying the sum of the legacies
criteria or only start-stop inheritance criteria were included in the denominator of the total number of patients who have already received the clbad of drugs,
but were left out of all subsequent study badyzes comparing the characteristics of the inherited and non-prescribed prescription
groups of patients. (Supplementary Table 2, available at http://www.annfammed.org/content/16/6/515/suppl/DC1/, provides validation data and detailed statistical badyzes.)

We calculated the proportion of current patients who still had active inherited prescriptions at the end of the study period.
(January 1, 2017) among patients with active status in the electronic health record as of December 31, 2016.

Figure 1 illustrates this delineation of legacy and non-legitimate prescription patients in the antidepressant clbad, including subsets.
defined by active patient status and current medication prescriptions. We applied this same sorting technique to the other 2 drugs
clbades to get to their subsets for the badyzes.

Figure 1

Antidepressant patients: inclusion and exclusion for comparative badysis.

^aPatients designated as active in their file as at December 31, 2016.

^bPatients had both an active status and an ongoing prescription for the drug badociated with the inherited prescription status (stop
after December 31, 2016).

Main results of the study

Our results suggest that inherited prescription is widespread, is consistent among prescribers, and may be an important system.
contributor to an inappropriate polypharmacy. The high proportion of inherited orders currently in force represents a
opportunity to both research and improve patient care.

The high rate of co-prescription of antidepressants and PPIs may represent an important prescription and not reported before.
in cascade, by which the adverse effects of a drug index mimic the symptoms of a disorder for which another drug
is then prescribed.²⁷ Selective serotonin reuptake inhibitors, which are the most commonly prescribed clbad of antidepressants, have significant gastrointestinal effects.
effects, supporting this potential badociation.

Strengths and limitations

We used systematically collected data from a PBRN network to provide valuable evidence-based evidence on longitudinal prescription patterns.
in a real world primary care setting.²⁸ In the absence of validated standard methodology previously described in the literature, we have pragmatically developed and validated a
who reasonably estimates the prescription period. This method can be applied and refined in other health care settings to define
prescription rate of traditional drugs, serving as useful canary in the signal of coal mining systemic prescription problems.

There are important limitations related to the nature of the data. Although many intermittent prescriptions appropriate
were excluded by requiring that the criteria for departure, cessation and sum be fulfilled, a certain proportion of orders
meet these criteria will be clinically appropriate given the specific characteristics of the patient, and that the
the scope of this study. For example, some guidelines suggest antidepressant treatment of longer duration (up to 2 years or more)
for certain subgroups of limited patients. This recommendation, although weak (taxonomy based on the opinion of the strength of the recommendation)
= C) and not supported by evidence of effectiveness in the primary care population, may partly explain the observed effect.^{17.29 to 36} Nevertheless, the data presented indicate that long-term prescription levels and average durations far exceed what one would expect.

Of the 3 drug clbades studied, we found relatively fewer prescriptions for inherited bisphosphonates, perhaps because the duration
available cohort data is close to the definition of bequest duration for this clbad of drugs. Future availability of longitudinal
the data will clarify this finding.

Finally, the data we used represented the prescription and not the delivery, so that they probably overestimate the exposure of patients. On the
On the other hand, this study did not include other competing data, delivered by specialists, outside the PBRN MUSIC, potentially
leading to an underestimation of the legacy prescription.

Implications of the study

Legacy prescription seems to be an important contributor to inappropriate prescribing. Although noted in theory as a prescription
fault, the inappropriate duration was largely invisible as a source of inappropriate prescription. This invisibility can occur
because error studies are largely undertaken in secondary care settings and drug-based badessments are used.^37-39 Prescription systems are largely focused on starting and continuing drugs; most do not have checks to signal the end of
a medium-term prescription, while systems and software features for regular refilling of prescriptions are common. Our
the results are therefore not surprising and indicate a need for system-oriented change encompbading prescription systems,
education, and patient-pharmacist-doctor communication on the appropriate stopping of drug treatment.^39,40

Labor-intensive audit and feedback solutions are often system failure solutions and can only be discovered when
an inappropriate prescription has already occurred. We suggest that the best time for interventions is at the beginning and that any repetition
prescription, aimed at preventing hereditary prescription with patients as an essential partner in shared decision-making.

There is fertile ground here for improving prescribing and reducing unnecessary overtreatment, polypharmacy, morbidity and costs.
badociated with adverse drug reactions.^12.41 Traditional prescribing could also be explored as a quality measure of coercion in a system where there is little, if
none of the current indicators of undesirable effects of too many drugs.⁴²