Dataset Information

Approaches to Optimize Medication Data Analysis in Clinical Cohort Studies.

ABSTRACT:

Objectives

Methods for pharmacoepidemiologic studies of large-scale data repositories are established. Although clinical cohorts of older adults often contain critical information to advance our understanding of medication risk and benefit, the methods best suited to manage medication data in these samples are sometimes unclear and their degree of validation unknown. We sought to provide researchers, in the context of a clinical cohort study of delirium in older adults, with guidance on the methodological tools to use data from clinical cohorts to better understand medication risk factors and outcomes.

Design

Prospective cohort study.

Setting

The Successful Aging After Elective Surgery (SAGES) prospective cohort.

Participants

A total of 560 older adults (aged ≥70 years) without dementia undergoing elective major surgery.

Measurements

Using the SAGES clinical cohort, methods used to characterize medications were identified, reviewed, analyzed, and distinguished by appropriateness and degree of validation for characterizing pharmacoepidemiologic data in smaller clinical data sets.

Results

Medication coding is essential; the American Hospital Formulary System, most often used in the United States, is not preferred over others. Use of equivalent dosing scales (e.g., morphine equivalents) for a single medication class (e.g., opioids) is preferred over multiclass analgesic equivalency scales. Medication aggregation from the same class (e.g., benzodiazepines) is well established; the optimal prevalence breakout for aggregation remains unclear. Validated scale(s) to combine structurally dissimilar medications (e.g., anticholinergics) should be used with caution; a lack of consensus exists regarding the optimal scale. Directed acyclic graph(s) are an accepted method to conceptualize causative frameworks when identifying potential confounders. Modeling-based strategies should be used with evidence-based, a priori variable-selection strategies.

Conclusion

As highlighted in the SAGES cohort, the methods used to classify and analyze medication data in clinically rich cohort studies vary in the rigor by which they have been developed and validated.

SUBMITTER: Duprey MS

PROVIDER: S-EPMC7744438 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:INTRODUCTION:In outcome research, incomplete follow-up is a major, yet potentially correctable source of bias. Cross-sectional surveys may theoretically increase completeness of follow-up, but low response rates are reported typically. We investigated whether a pre-notification letter improved patient availability for follow-up phone interviews and thereby improved cross-sectional survey yield. METHODS:A consecutive series of vascular patients was randomly divided into a trial and a validation population. The trial population was then randomized 1:1 to one of two cross-sectional contact strategies: Strategy 1 consisted of direct contact attempts by up to 12 systematically timed phone calls, whereas Strategy 2 used a personalized pre-notification letter to arrange for scheduled phone call interviews. Response rates, average time and efforts needed per patient and overall survey duration were compared. Subsequently, trial findings were externally validated in the validation population. RESULTS:Of 728 consecutive patients, 370 were allocated to the trial population. Trial patients contacted by strategy 1 (n = 183) had a similar profile when compared to trial patients contacted by strategy 2 (n = 187). Follow-up periods following surgery (54.3 versus 53.6 months) and all-cause mortality rates (21.3% versus 18.7%) were comparable between the trial groups. Cross-sectional information on survival outcomes was almost complete after both contact strategies (99.5% versus 98.9%, P = 1.0). In 144/187 strategy 2 patients (77%) interviews were scheduled successfully necessitating significantly less contact attempts (median of 1.3 versus 2.3 per patient, P<0.0001). However, invested time per patient was similar between the groups (median of 10.1 versus 9.6 minutes), and survey strategy 1 completed earlier (median time to contact 4 versus 11 days, P<0.0001). Therefore, strategy 1 was validated in the validation population (n = 358): a low lost to follow-up rate below 1% (P = 1.0) was reconfirmed necessitating an average of 2.3 contact attempts per patient. CONCLUSIONS:Both contact strategies were equally successful in contacting almost all patients cross-sectionally. If systematically timed, direct phone calls were less complicated to organize and faster completed. Given the low time and effort per patient, outcome studies should invest in systematic follow-up surveys to minimize attrition bias.

Project description:Background: The benefits of medication optimization are largely uncontroversial but difficult to achieve. Behavior change interventions aiming to optimize prescriber medication-related decisions, which do not forbid any option and that do not significantly change financial incentives, offer a promising way forward. These interventions are often referred to as nudges. Objective: The current systematic literature review characterizes published studies describing nudge interventions to optimize medication prescribing by the behavioral determinants they intend to influence and the techniques they apply. Methods: Four databases were searched (MEDLINE, Embase, PsychINFO, and CINAHL) to identify studies with nudge-type interventions aiming to optimize prescribing decisions. To describe the behavioral determinants that interventionists aimed to influence, data were extracted according to the Theoretical Domains Framework (TDF). To describe intervention techniques applied, data were extracted according to the Behavior Change Techniques (BCT) Taxonomy version 1 and MINDSPACE. Next, the recommended TDF-BCT mappings were used to appraise whether each intervention applied a sufficient array of techniques to influence all identified behavioral determinants. Results: The current review located 15 studies comprised of 20 interventions. Of the 20 interventions, 16 interventions (80%) were effective. The behavior change techniques most often applied involved prompts (n = 13). The MINDSPACE contextual influencer most often applied involved defaults (n = 10). According to the recommended TDF-BCT mappings, only two interventions applied a sufficient array of behavior change techniques to address the behavioral determinants the interventionists aimed to influence. Conclusion: The fact that so many interventions successfully changed prescriber behavior encourages the development of future behavior change interventions to optimize prescribing without mandates or financial incentives. The current review encourages interventionists to understand the behavioral determinants they are trying to affect, before the selection and application of techniques to change prescribing behaviors. Systematic Review Registration: [https://www.crd.york.ac.uk/prospero/], identifier [CRD42020168006].

Project description:ObjectiveIn large cohort studies comorbidities are usually self-reported by the patients. This way to collect health information only represents conditions known, memorized and openly reported by the patients. Several studies addressed the relationship between self-reported comorbidities and medical records or pharmacy data, but none of them provided a structured, documented method of evaluation. We thus developed a detailed procedure to compare self-reported comorbidities with information on comorbidities derived from medication inspection. This was applied to the data of the German COPD cohort COSYCONET.MethodsApproach I was based solely on ICD10-Codes for the diseases and the indications of medications. To overcome the limitations due to potential non-specificity of medications, Approach II was developed using more detailed information, such as ATC-Codes specific for one disease. The relationship between reported comorbidities and medication was expressed by a four-level concordance score.ResultsApproaches I and II demonstrated that the patterns of concordance scores markedly differed between comorbidities in the COSYCONET data. On average, Approach I resulted in more than 50% concordance of all reported diseases to at least one medication. The more specific Approach II showed larger differences in the matching with medications, due to large differences in the disease-specificity of drugs. The highest concordance was achieved for diabetes and three combined cardiovascular disorders, while it was substantial for dyslipidemia and hyperuricemia, and low for asthma.ConclusionBoth approaches represent feasible strategies to confirm self-reported diagnoses via medication. Approach I covers a broad spectrum of diseases and medications but is limited regarding disease-specificity. Approach II uses the information from medications specific for a single disease and therefore can reach higher concordance scores. The strategies described in a detailed and reproducible manner are generally applicable in large studies and might be useful to extract as much information as possible from the available data.

Project description:IntroductionThe effects caused by differences in data acquisition can be substantial and may impact data interpretation in multi-site/scanner studies using magnetic resonance spectroscopy (MRS). Given the increasing use of multi-site studies, a better understanding of how to account for different scanners is needed. Using data from a concussion population, we compare ComBat harmonization with different statistical methods in controlling for site, vendor, and scanner as covariates to determine how to best control for multi-site data.MethodsThe data for the current study included 545 MRS datasets to measure tNAA, tCr, tCho, Glx, and mI to study the pediatric concussion acquired across five sites, six scanners, and two different MRI vendors. For each metabolite, the site and vendor were accounted for in seven different models of general linear models (GLM) or mixed-effects models while testing for group differences between the concussion and orthopedic injury. Models 1 and 2 controlled for vendor and site. Models 3 and 4 controlled for scanner. Models 5 and 6 controlled for site applied to data harmonized by vendor using ComBat. Model 7 controlled for scanner applied to data harmonized by scanner using ComBat. All the models controlled for age and sex as covariates.ResultsModels 1 and 2, controlling for site and vendor, showed no significant group effect in any metabolites, but the vendor and site were significant factors in the GLM. Model 3, which included a scanner, showed a significant group effect for tNAA and tCho, and the scanner was a significant factor. Model 4, controlling for the scanner, did not show a group effect in the mixed model. The data harmonized by the vendor using ComBat (Models 5 and 6) had no significant group effect in both the GLM and mixed models. Lastly, the data harmonized by the scanner using ComBat (Model 7) showed no significant group effect. The individual site data suggest there were no group differences.ConclusionUsing data from a large clinical concussion population, different analysis techniques to control for site, vendor, and scanner in MRS data yielded different results. The findings support the use of ComBat harmonization for clinical MRS data, as it removes the site and vendor effects.