Pharmacoepidemiology and Drug Safety最新文献

Purpose: Confounding is a key concern in observational studies using healthcare databases. The high-dimensional propensity score (HDPS) algorithm is an approach for generating and prioritising proxy variables, leveraging all available information in a database to mitigate residual confounding. This study aims to implement HDPS approaches in a novel setting using primary and secondary data available from Hong Kong (HK).

Methods: Using data from HK, we implemented HDPS in a cohort study investigating the use of different antihypertensive drug classes and incident dementia risk. The top 250 HDPS covariates were included in inverse probability of treatment weighting in addition to investigator-specified variables. Diagnostics evaluated the performance of the HDPS. Sensitivity analyses included varying the number of HDPS covariates and removing potentially influential or inappropriate covariates.

Results: 434 506 new-users of antihypertensives were included. With a traditional PS approach, no evidence for an association was observed for each antihypertensive comparison. After HDPS implementation, the estimate for beta-blockers shifted from no evidence (Hazard ratio (HR): 0.93, 95% confidence interval (CI): 0.86-1.02) to moderate evidence of a reduced hazard of incident dementia compared to angiotensin-converting enzyme inhibitors (HR: 0.90, 95% CI: 0.82-0.98). A greater overall covariate balance between comparison groups was achieved after the inclusion of HDPS covariates and potential frailty markers were identified as influential.

Conclusions: We successfully implemented the HDPS in HK data, observing improved covariate balance across a wider set of potential confounders. HDPS also identified possible database-specific frailty markers which could be considered more widely when specifying adjustment variables in this setting.

Introduction: The extraordinary circumstances due to the COVID-19 pandemic and related restrictions altered the management of mental health disorders, including the use of antipsychotics. We aimed to examine the changes in antipsychotic utilization and expenditure in Turkey throughout pandemic-associated restriction periods.

Methods: Nationwide drug sales and projected prescribing data from 01.03.2018 to 31.12.2022 were obtained from IQVIA Turkey. We assessed average monthly consumption, expenditure, and quarterly prescribing levels across three periods: "before restrictions" (BfR, 01.03.2018-31.03.2020), "during restrictions" (DuR, 01.04.2020-31.03.2022), and "after restrictions" (AfR, 01.04.2022-31.12.2022). Consumption and prescribing levels were measured using "defined daily dose/1000 inhabitants/day" (DID) parameter.

Results: Antipsychotic consumption throughout periods increased from 8.4 ± 0.6 DID in BfR to 9.9 ± 1.6 DID in DuR (p < 0.001), and to 10.1 ± 0.9 DID in AfR (p < 0.001 vs. BfR). Atypical antipsychotics followed the overall trend, whereas typical antipsychotics remained stable from DuR to AfR, deviating from this pattern. Antipsychotic expenditure rose from €16.7 m ± 1.1 m in BfR to €19.0 m ± 2.7 m in DuR (p < 0.001), then shifted to €18.3 m ± 1.9 m in AfR (p > 0.05 vs. BfR and DuR). High-cost antipsychotic use increased after the pandemic onset (p < 0.001) and remained elevated in AfR (p < 0.001). Prescribing for schizophrenia declined from 2.2 ± 0.3 DID in BfR to 1.3 ± 0.2 DID in DuR (p < 0.001), then escalated to 1.8 ± 0.3 DID in AfR (p = 0.015 vs. DuR).

Conclusions: Our study revealed an upsurge in antipsychotic utilization in Turkey with the start of the pandemic. A range of factors may have contributed, notably the impact of policies facilitating the dispensing of chronic medications without prescription or a tendency towards polypharmacy.

Introduction: Venous thromboembolism (VTE) is labeled as an adverse effect of the adeno-vector-based vaccines AstraZeneca and Johnson & Johnson. We aimed to study whether there was an increase in general practitioner (GP) consultations for VTE after COVID-19 vaccination.

Methods: An exposure-anchored self-controlled cohort study was performed among COVID-19 vaccinated persons aged ≥ 12 years who were registered in the PHARMO Data Network and Nivel Primary Care Database in the Netherlands. The focal window was set at 28 days after each COVID-19 vaccination and the referent window at all time outside the focal window. Adjusted incidence rate ratios (aIRR), adjusting for SARS-CoV-2 infection, were calculated using Poisson regression.

Results: In total, 2 133 853 persons were included. The highest increase in GP consultations for VTE was observed after Johnson & Johnson vaccination (aIRR: 3.14, 95% CI: 1.50-6.57), and a slight increase after Pfizer/BioNTech dose 1 (aIRR: 1.24, 95% CI: 1.09-1.40). Risk groups were 12-60 year-olds with increased GP consultations for VTE after Johnson & Johnson (aIRR: 2.30, 95% CI: 1.44-3.69) and Pfizer/BioNTech (aIRR: 1.29, 95% CI: 1.11-1.50), and in specific groups of males aged 12-60 years. Also, females using hormone-containing contraceptives or hormone replacement therapy (HRT) showed increased GP consultations for VTE after AstraZeneca (aIRR: 2.87, 95% CI: 1.13-7.33) and Pfizer/BioNTech (aIRR: 1.48, 95% CI: 1.10-2.01).

Conclusion: Increased GP consultations for VTE were observed after both vector and mRNA vaccination, in particular among males, 12-60 year olds, and females using hormone-containing contraceptives or HRT.

Purpose: A prescribing cascade (PC) occurs when a medication (index) causes an adverse drug reaction (ADR), which is addressed by prescribing additional medication (marker). Medication initiated in the hospital may cause post-discharge ADRs and PCs, especially when multiple healthcare providers are involved. The study aimed to assess the cumulative incidence of potential PCs post-discharge and identify the healthcare providers involved in prescribing the marker medication.

Methods: A cohort study was conducted among adult patients admitted in one hospital between 2019 and 2023, who initiated medication associated with preselected PCs (n = 20). A PC was defined as the initiation of a marker medication which may be intended to treat an ADR induced by the index medication. Data from the hospital and the Nationwide Medication Record System were used to identify potential PCs post-discharge. The primary outcome was the cumulative incidence of PCs, estimated for PCs with ≥ 10 patients initiating the index medication. The secondary outcome was the percentage of cases where the marker medication was prescribed by a healthcare provider outside the hospital, for PCs with ≥ 10 patients initiating the marker medication. Descriptive statistics were used.

Results: Among 24 282 patients initiating index medication, 502 potential PCs were observed. The cumulative incidence was estimated for 17 PCs, ranging from 0% to 12.3%. Across 12 PCs with ≥ 10 patients, percentages of marker medications prescribed outside the hospital ranged from 31.8% to 92.8%.

Conclusion: The cumulative incidence of potential PCs post-discharge can be substantial with marker medication often initiated by healthcare providers outside the hospital.

Purpose: Extended-release naltrexone (XR-NTX, monthly injection) is used to treat opioid use disorder (OUD). In claims data, XR-NTX may be identified by drug or procedure codes. In the US, Medicaid is a predominant payer of OUD treatment and differences in state Medicaid policies may produce variation in XR-NTX coding. We aimed to describe documentation of XR-NTX in multi-state Medicaid data.

Methods: Using 2016-2019 National Medicaid data (TAF) from 26 states, we identified individuals with an XR-NTX specific drug or procedure code and evidence of OUD during ≥ 5 months continuous Medicaid enrollment (N = 26 169). At the individual's first observed XR-NTX treatment, we described state-level variation in the types of codes, file source, and presence of procedure codes for injection (including nonspecific codes).

Results: An XR-NTX drug code was the first record of treatment for 98% of patients; this percentage was high in all states except one. Just 25% of patients had a procedure code for injection (XR-NTX specific code or non-specific injection code) during the first treatment with marked variation across states, ranging from 7% to 87%. The percentage of patients with evidence of a second XR-NTX treatment was higher among patients with an injection code at initial treatment (61%) than among patients without an injection code (49%).

Conclusions: We found inconsistent patterns of XR-NTX codes across states indicating claim-based definitions should consider both drug and procedure codes to fully capture XR-NTX service delivery. Multiple definitions should be considered in sensitivity analyses given substantial variability in coding practices across states.

Purpose: Diabetic ketoacidosis (DKA) is an important identified risk of treatment with the sodium-glucose cotransporter 2 inhibitor (SGLT2i) dapagliflozin, particularly in patients with type 1 diabetes mellitus (T1DM). We evaluated the DKA incidence rate (IR) among dapagliflozin-treated Japanese patients with T1DM receiving concomitant insulin.

Methods: This observational, cohort, post-marketing study utilized data from the JMDC Payer database. All patients had T1DM and were prescribed insulin; the dapagliflozin group included dapagliflozin-treated patients and the control group included SGLT2i nonusers. The study outcomes were DKA events (a composite of DKA resulting in hospitalization and/or death). Using a prevalent new-user design, time-stratified sequential propensity score (PS) matching was implemented.

Results: Between March 2019 and March 2021, 5886 insulin-treated patients with T1DM were eligible, and 327/5886 patients newly received dapagliflozin. After up to 1:3 PS-matching, 327 dapagliflozin-treated patients and 980 matched controls were analyzed. The IRs (95% confidence interval) of DKA in the PS-matched cohort were 1.54 per 100 person-years (/100 PY; 0.42-3.95) and 1.14/100 PY (0.55-2.10) in the dapagliflozin and control groups, respectively, with a hazard ratio of 1.28 (0.40-4.09). The IRs in the unmatched cohort were 1.54/100 PY (0.42-3.95) and 0.97/100 PY (0.72-1.27) in the dapagliflozin and control groups, respectively.

Conclusion: DKA incidence in dapagliflozin-treated T1DM patients was slightly higher than in SGLT2i nonusers. As the confidence intervals overlapped, these results did not suggest any meaningful differences or unexpected higher risk of DKA, consistent with previous reports in other countries.

Background: Given inconsistent findings from previous epidemiologic studies on the association between antidepressant exposure and colorectal cancer (CRC), our study provides a rigorous investigation to clarify the temporality of this association, including early-age onset (EAO) and average-age onset (AAO) CRC.

Methods: We conducted a population-based case-control study using administrative health databases from British Columbia, Canada. We included CRC cases and controls, matched (1:10) on age, sex, and index date (i.e., CRC diagnosis date/matched date). Antidepressant exposures were ascertained by duration (i.e., varying windows from 15 to 1 year before CRC diagnosis), drug classes (tricyclic antidepressants (TCAs), selective serotonin reuptake inhibiters (SSRIs), other), cumulative dose and treatment intensity. We used multivariable conditional logistic regression models and interpreted odds ratios as relative risks.

Results: Among 10,171 CRC cases (688 EAO-CRC; 9483 AAO-CRC) and 90 928 controls, antidepressants exposure in the 15-year window was associated with a lower risk of CRC overall (adjusted relative risk [aRR] 0.84; 95% CI 0.80, 0.89), EAO-CRC (aRR 0.54; 95% CI 0.44, 0.66), and AAO-CRC (aRR 0.87; 95% CI 0.83, 0.92). Across narrowing exposure windows, associations persisted up to 7 years before CRC diagnosis, then weakened. Inverse associations were also observed for TCAs (aRR 0.83; 95% CI 0.77, 0.89) and SSRIs (aRR 0.86; 95% CI 0.81, 0.91) and CRC. Cumulative dose and treatment intensity showed no associations.

Conclusions: Across all age groups, antidepressant exposure in the earlier exposure windows (15-7 years) was associated with a lower CRC risk, with the strongest effect at the 15-year window.

Purpose: This study extends a version of the high-dimensional propensity score (HDPS) recently modified for the UK electronic health record setting, by enriching primary care data with hospital data. The performance of this modified approach is assessed via the estimation of a well-established association, the reduced risk of upper gastrointestinal bleeding (UGIB) in cyclo-oxygenase-2 inhibitor (COX-2i) users versus non-steroidal anti-inflammatory drug (NSAID) users.

Methods: We conducted an active-comparator, new-user cohort study using UK primary care data from the Clinical Practice Research Datalink GOLD database, with linkages to hospitalisation and mortality records. We included individuals with osteoarthritis initiating NSAIDs or COX-2is between 2000-2004. We used Cox proportional hazards models to estimate the hazard ratio (HR) for UGIB, adjusting for confounders using investigator-specified and HDPS-derived propensity scores. Sensitivity analyses were conducted varying the number of HDPS covariates included and the covariate assessment period.

Results: We identified 74 443 and 25 742 new users of NSAID and COX-2i users, respectively. The unadjusted HR for UGIB comparing COX-2i and NSAID users was 1.28 (95% CI: 0.95-1.72). Of the included HDPS covariates, 26% originated from the hospitalisation dimension, a source not considered in previous applications indicating the considerable information contained in these data on proxies of potential confounders. The modified-HDPS obtained similar results to the other studies, shifting the HR closer to the expected association (HR 0.86; 95% CI: 0.58-1.26).

Conclusion: We demonstrate the ability of the modified-HDPS to obtain similar results to comparable pharmacoepidemiological studies and randomised trials, highlighting the potential benefit of these approaches in UK EHRs more widely and the value of adding hospital data to enrich the pool of covariates available for the HDPS algorithm.