Drug exposure misclassification in pharmacoepidemiology: Sources and relative impact

Comparison design and evaluation power in cohort and self-controlled case series designs for post-authorization vaccine safety studies

Background

Post-authorization safety studies (PASSs) of vaccines are important. PASSs enable the evaluation of association between vaccination and adverse events following immunization through common study designs. Clinical trials during vaccine development typically include a few thousand to 10,000 participants while a PASS might aim to detect a few adverse events per 100,000 vaccine recipients. While all available data may be utilized, prior consideration of power analyses are nonetheless crucial for interpretation in cases where statistically significant differences are not found.

Methods

This research primarily examined cohort study design and self-controlled case series (SCCS) design, estimating the power of a PASS under plausible conditions.

Results

Both the cohort study and SCCS designs necessitated large sample sizes or high event counts to guarantee adequate power. The SCCS design is particularly suited to evaluating rare adverse events. However, extremely rare events may not yield sufficient occurrences, thereby resulting in low power. Although the SCCS design can more efficiently control for time-invariant confounding in principle, it solely estimates relative measures. A cohort study design might be preferred if confounding can be adequately managed as it also estimates absolute measures. It may be an easy decision to use all the data at hand for either design. We found it necessary to estimate the sample size and number of events to be used in the study based on a priori information and anticipated results.

Time lapses between distribution of influenza vaccines to health authorities and their administration by General Practitioners (GPs) to older adults: a retrospective study over five influenza seasons in Italy

BACKGROUND

Delays in influenza vaccine delivery and administration can hinder vaccine coverage and protection. This study examines the differentials in distributing and administering adjuvanted trivalent (aTIV) and quadrivalent influenza vaccines (aQIV) to older adults in Italy's primary care setting and its potential impact on hospitalization risk over 5 epidemic seasons.

METHODS

Using a primary care database, individuals aged ≥ 65 years were selected. The proportion of vaccine distribution to regional authorities and subsequent administration by GPs was estimated using census data. Using quantile (median) regression, we examined the relationship between velocities of vaccine distribution and administration (doses/week) and the incidence of hospitalizations.

RESULTS

Over the 5 influenza seasons, the velocity of distribution and administration of aTIV/aQIV ranged 341-833 and 152-270 median doses/week; no trend was yielded for the difference between these velocities (p = 0.189) or vaccine coverage (p = 0.142). An association was observed for each differential dose/week between distributed and administered vaccines and all-cause hospitalizations with a 10% increase in 2017-2018, 54% in 2018-2019, and 12% in 2020-2021 season.

CONCLUSIONS

These findings highlight the importance of minimizing the time lapse between vaccine distribution and administration to mitigate the impact of influenza and address factors that contribute to vaccination barriers.

Drug safety signal detection in a regional healthcare database using the tree-based scan statistic and comparison to 3 other mining methods

AIMS

To evaluate and compare the relative performance of the tree-based scan statistic (TreeScan) with the crude cohort study, Bayesian confidence propagation neural network (BCPNN) and Gamma Poisson Shrinker (GPS) in detecting statin-related adverse events (AEs) in an electronic healthcare database.

METHODS

Data from a Chinese healthcare database from 2010 to 2016 were evaluated. We identified statin users based on prescription information in their out-/in-patient records, and AEs were defined according to the ICD-10 codes in patients' diagnosis records. TreeScan was applied to detect AE signals related to statin use and was compared with 3 other methods based on sensitivity, specificity, positive predictive value, negative predictive value, accuracy, the Youden index, area under the precision-recall curve and the area under the receiver operating characteristic curve.

RESULTS

A total of 224 187 patients were enrolled and divided into 85 758 statin users and 138 429 nonusers. TreeScan generated 29 positive signals, of which 9 were known AEs. The sensitivities of TreeScan, BCPNN and GPS were all 69.2%, which was higher than that of the crude cohort study (46%). The specificity (82.3%), positive predictive value (31.0%), negative predictive value (95.9%), accuracy (81.0%), Youden index (51.5%) and area under the receiver operating characteristic curve (75.8%) of TreeScan were the highest among the 4 methods.

CONCLUSION

TreeScan outperformed the crude cohort, BCPNN and GPS in detecting statin-related AEs in an electronic healthcare database. Therefore, it can be used as a complementary tool for other signal detection methods in drug safety surveillance.

"Take up to eight tablets per day": Incorporating free-text medication instructions into a transparent and reproducible process for preparing drug exposure data for pharmacoepidemiology

PURPOSE

Routinely collected prescription data provides drug exposure information for pharmacoepidemiology, informing start/stop dates and dosage. Prescribing information includes structured data and unstructured free-text instructions, which can include inherent variability, such as "one to two tablets up to four times a day". Preparing drug exposure data from raw prescriptions to a research ready dataset is rarely fully reported, yet assumptions have considerable implications for pharmacoepidemiology. This may have bigger consequences for "pro re nata" (PRN) drugs. Our aim was, using a worked example of opioids and fracture risk, to examine the impact of incorporating narrative prescribing instructions and subsequent drug preparation assumptions on adverse event rates.

METHODS

R-packages for extracting free-text medication prescription instructions in a structured form (doseminer) and an algorithm for transparently processing drug exposure information (drugprepr) were developed. Clinical Practice Research Datalink GOLD was used to define a cohort of adult new opioid users without prior cancer. A retrospective cohort study was performed using data between January 1, 2017 and July 31, 2018. We tested the impact of varying drug preparation assumptions by estimating the risk of opioids on fracture risk using Cox proportional hazards models.

RESULTS

During the study window, 60 394 patients were identified with 190 754 opioid prescriptions. Free-text prescribing instruction variability, where there was flexibility in the number of tablets to be administered, was present in 42% prescriptions. Variations in the decisions made during preparing raw data for analysis led to marked differences impacting the event number (n = 303-415) and person years of drug exposure (5619-9832). The distribution of hazard ratios as a function of the decisions ranged from 2.71 (95% CI: 2.31, 3.18) to 3.24 (2.76, 3.82).

CONCLUSIONS

Assumptions made during the drug preparation process, especially for those with variability in prescription instructions, can impact results of subsequent risk estimates. The developed R packages can improve transparency related to drug preparation assumptions, in line with best practice advocated by international pharmacoepidemiology guidelines.

Association of anticholinergic drug use with postoperative mortality among patients with hip fracture. A nationwide cohort study

PURPOSE

Anticholinergic (AC) drugs are associated with various determinantal outcomes. Data regarding the effect of AC drugs on mortality among geriatric hip fracture patients are limited and inconsistent.

METHODS

Using Danish health registries, we identified 31,443 patients aged ≥65 years undergoing hip fracture surgery. AC burden was assessed 90 days before surgery by the Anticholinergic Cognitive Burden (ACB) score and number of AC drugs. Logistic and Cox regression producing odds ratios (OR) and hazard ratios (HR) for 30- and 365- day mortality, adjusting for age, sex, and comorbidities were computed.

RESULTS

AC drugs were redeemed by 42% of patients. The 30-day mortality increased from 7% for patients with ACB score of 0 to 16% for patients with ACB score of ≥5, corresponding to an adjusted OR 2.5 (CI: 2.0-3.1). The equivalent adjusted HR for 365-mortality was 1.9 (CI: 1.6-2.1). Using count of AC drugs as exposure we found a stepwise increase in ORs and HRs with increased number of AC drugs; Compared to non-users, adjusted ORs for 30-days mortality were 1.6 (CI: 1.4-1.7), 1.9 (CI: 1.7-2.1), and 2.3 (CI: 1.9-2.7) for users of 1, 2 and 3+ AC drugs. HRs for 365-day mortality were 1.4 (CI: 1.3-1.5), 1.6 (CI: 1.5-1.7) and 1.8 (CI: 1.7-2.0).

CONCLUSION

Use of AC drugs was associated with increased 30-day and 365-day mortality among older adults with hip fracture. Simply counting the number of AC drugs may be a clinically relevant and easy AC risk assessment tool. Continued effort to reduce AC drug-use is relevant.

Primary nonadherence to drugs prescribed by general practitioners: A Dutch database study

AIM

Primary nonadherence (PNA) is defined as not filling the first prescription for a drug treatment. PNA can lead not only to poor patient outcomes but also to exposure misclassification in written prescription databases. This study aims to estimate PNA in primary care in the Netherlands and to investigate associated factors.

METHODS

Patients from the Nivel Primary Care Database (Nivel-PCD) who received a new prescription (>1 year not prescribed) from a general practitioner in 2012 were linked to pharmacy dispensing information of consenting pharmacies based on sex, year of birth, four-digit postal code and at least 50% matching Anatomical Therapeutic Classification codes. PNA was defined as not having a prescription dispensed within 30 days from the prescribing date. PNA was assessed overall and per drug class. The associations between PNA and several patient- and prescription-related characteristics were assessed using mixed-effects logistic regression models.

RESULTS

After matching 86 361 of 396 251 subjects (21.8%) in the Nivel-PCD records to the pharmacy records, this study included 65 877 subjects who received 181 939 new drug prescriptions. Overall, PNA was 11.5%. PNA was lowest for thyroid hormones (5.5%) and highest for proton pump inhibitors (12.8%). Several factors were associated with PNA, such as having comorbidities (odds ratio [OR] 1.46, 95% confidence interval [CI] 1.37-1.56 for >3 active diagnoses, compared to no active diagnoses) or reimbursement status (OR 2.78, 95% CI 2.65-2.92 for not reimbursed drugs compared to fully reimbursed drugs).

CONCLUSIONS

A total of 11.5% of newly prescribed drugs were not dispensed. This can lead to overestimation of the actual drug exposure status when using written prescription databases.

To clarify the safety profile of paracetamol for home-care patients with COVID-19: a real-world cohort study, with nested case-control analysis, in primary care

BACKGROUND AND OBJECTIVE

This study aimed to compare the prescribing patterns of paracetamol in COVID-19 with those for similar respiratory conditions and investigated the association between paracetamol use and COVID-19-related hospitalization/death.

METHODS

Using a primary care data source, we conducted a cohort study to calculate the incidence rate of paracetamol use in COVID-19 and for similar respiratory conditions in 2020 and 2019 (i.e. pre-pandemic phase), respectively. In the study cohort, we nested a case-control analyses to investigate the association between paracetamol use and COVID-19-related hospitalizations/deaths.

RESULTS

Overall, 1554 (33.4 per 1000) and 2566 patients (78.3 per 1000) were newly prescribed with paracetamol to treat COVID-19 or other respiratory conditions, respectively. Those aged 35-44 showed the highest prevalence rate (44.7 or 99.0 per 1000), while the oldest category reported the lowest value (17.8 or 39.8 per 1000). There was no association for early (OR = 1.15; 95% CI: 0.92-1.43) or mid-term (OR = 1.29; 95% CI: 0.61-2.73) users of paracetamol vs. non-users. Instead, the late users of paracetamol showed a statistically significant increased risk of hospitalization/death (OR = 1.75; 95% CI: 1.4-2.2).

CONCLUSIONS

Our findings provide reassuring evidence on the use and safety profile of paracetamol to treat early symptoms of COVID-19 as in other respiratory infections.

Adjuvanted versus non-adjuvanted standard-dose influenza vaccines in preventing all-cause hospitalizations in the elderly: a cohort study with nested case-control analyses over 18 influenza seasons

BACKGROUND

The higher effectiveness of adjuvanted trivalent influenza vaccine (aTIV) versus non-adjuvanted (na) formulations in preventing all-cause hospitalization has been demonstrated for a single influenza season and in institutionalized elderly only. This study evaluated the relative vaccine effectiveness for aTIV vs. non-adjuvanted trivalent (naTIV) and/or quadrivalent (naQIV) influenza vaccines in preventing all-cause hospitalizations across 18 influenza seasons in primary care.

RESEARCH DESIGN AND METHODS

Using Health Search Database, a nested case-control analysis was conducted in a cohort of older adults being vaccinated with aTIV or naTIV/naQIV. Conditional logistic regression was adopted to estimate the odds ratio (OR) of all-cause hospitalizations occurred during the epidemic period.

RESULTS

Of 58,252 patients vaccinated with aTIV and naTIV/naQIV for the first time, 2,504 cases of all-cause hospitalization (3.46 per 1,000 person-weeks) during the 18 influenza seasons were identified. Compared with naTIV/naQIV, aTIV was associated with a 12% reduced the odds of all-cause hospitalizations (OR 0.88; 95% CI: 0.80-0.98).

CONCLUSIONS

In an 18-season cohort of older adults, aTIV reduced the risk of all-cause hospitalizations when compared with naTIV/naQIV. Our findings confirm additional benefits for adjuvanted influenza vaccines in older adults.

A read-world study on SGLT2 inhibitors and diabetic kidney disease progression

Background

Randomized controlled trials have demonstrated the benefits of sodium–glucose cotransporter 2 inhibitors (SGLT2is) in people with type 2 diabetes mellitus (T2DM) and chronic kidney disease (CKD). However, real-world data on CKD progression and the development of end-stage kidney disease (ESKD) remains scarce. Our aim was to study renal outcomes of people with diabetic kidney disease (DKD) using SGLT2is in a highly prevalent DKD population.

Methods

Between 2016 and 2019 we recruited T2DM patients in the renal and diabetic clinics in a regional hospital in Singapore. Patients prescribed SGLT2is were compared with those on standard anti-diabetic and renoprotective treatment. The outcome measures were CKD progression [a ≥25% decrease from baseline and worsening of estimated glomerular filtration rate (eGFR) categories according to the Kidney Disease: Improving Global Outcomes guidelines] and ESKD (eGFR <15 mL/min/1.73 m²).

Results

We analysed a total of 4446 subjects; 1598 were on SGLT2is. There was a significant reduction in CKD progression {hazard ratio [HR] 0.60 [95% confidence interval (CI) 0.49–0.74]} with SGLT2is. The HR for eGFR ≥45 mL/min/1.73 m² and 15–44 mL/min/1.73 m² was 0.60 (95% CI 0.47–0.76) and 0.43 (95% CI 0.23–0.66), respectively. There was also a reduction in risk for developing ESKD for the entire cohort [HR 0.33 (95% CI 0.17–0.65)] and eGFR 15–44 mL/min/1.73 m² [HR 0.24 (95% CI 0.09–0.66)]. Compared with canagliflozin and dapagliflozin, empagliflozin showed a sustained risk reduction of renal outcomes across CKD stages 1–4.

Conclusions

This real-world study demonstrates the benefits of SGLT2is on CKD progression and ESKD. The effect is more pronounced in moderate to advanced CKD patients.