Randomized clinical trials (RCTs) are the gold standard in producing clinical evidence of efficacy and safety of medical interventions. More recently, a new paradigm is emerging—specifically within ...the context of preauthorization regulatory decision‐making—for some novel uses of real‐world evidence (RWE) from a variety of real‐world data (RWD) sources to answer certain clinical questions. Traditionally reserved for rare diseases and other special circumstances, external controls (eg, historical controls) are recognized as a possible type of control arm for single‐arm trials. However, creating and analyzing an external control arm using RWD can be challenging since design and analytics may not fully control for all systematic differences (biases). Nonetheless, certain biases can be attenuated using appropriate design and analytical approaches. The main objective of this paper is to improve the scientific rigor in the generation of external control arms using RWD. Here we (a) discuss the rationale and regulatory circumstances appropriate for external control arms, (b) define different types of external control arms, and (c) describe study design elements and approaches to mitigate certain biases in external control arms. This manuscript received endorsement from the International Society for Pharmacoepidemiology (ISPE).
Background and purpose A variety of risk factors have been hypothesized to contribute to the development of fracture-healing complications; however, population-based estimates of the strength of ...these risk factors are limited. In this case-control study, we evaluated patient-related risk factors for fracture-healing complications.
Methods Using the United Kingdom General Practice Research Database, we identified patients with a fracture-healing complication (delayed union, nonunion, or malunion) between 1988 and 2008. 4 controls (i.e. patients with normal healing) were matched to each case on general practice, fracture site, fracture date, and length of history in the database. We used conditional logistic regression to estimate odds ratios (ORs) of various risk factors, including demographics, comorbidities, and medication use.
Results Diabetes and use of non-steroidal anti-inflammatory drugs (NSAIDs) within 12 months before the initial fracture were associated with a higher odds of a fracture-healing complication (type-I diabetes: adjusted OR = 2.3, 95% CI: 1.3-3.8; type-II diabetes: adjusted OR = 2.3, CI: 1.4-3.7; NSAIDs: adjusted OR = 2.6, CI: 2.1-3.2). Patients who had a motor vehicle accident recorded within 1 month before their initial fracture were also at increased odds of a fracture-healing complication (adjusted OR = 2.6, CI: 1.2-5.4).
Interpretation Diabetes, NSAID use, and a recent motor vehicle accident were most consistently associated with an increased risk of a fracture-healing complication, regardless of fracture site or specific fracture-healing complication. This analysis suggests that certain patient-related characteristics influence the development of fracture-healing complications in general, even though specific healing complications may differ by their mechanism.
Mortality risk among hemodialysis (HD) patients may be highest soon after initiation of HD. A period of elevated mortality risk was identified among US incident HD patients, and which patient ...characteristics predict death during this period and throughout the first year was examined using data from the Dialysis Outcomes and Practice Patterns Study (DOPPS; 1996 through 2004). A retrospective cohort study design was used to identify mortality risk factors. All patient information was collected at enrollment. Life-table analyses and discrete logistic regression were used to identify a period of elevated mortality risk. Cox regression was used to estimate adjusted hazard ratios (HR) measuring associations between patient characteristics and mortality and to examine whether these associations changed during the first year of HD. Among 4802 incident patients, risk for death was elevated during the first 120 d compared with 121 to 365 d (27.5 versus 21.9 deaths per 100 person-years; P = 0.002). Cause-specific mortality rates were higher in the first 120 d than in the subsequent 121 to 365 d for nearly all causes, with the greatest difference being for cardiovascular-related deaths. In addition, 20% of all deaths in the first 120 d occurred subsequent to withdrawal from dialysis. Most covariates were found to have consistent effects during the first year of HD: Older age, catheter vascular access, albumin <3.5, phosphorus <3.5, cancer, and congestive heart failure all were associated with elevated mortality. Pre-ESRD nephrology care was associated with a significantly lower risk for death before 120 d (HR 0.65; 95% confidence interval 0.51 to 0.83) but not in the subsequent 121- to 365-d period (HR 1.03; 95% confidence interval 0.83 to 1.27). This care was related to approximately 50% lower rates of both cardiac deaths and withdrawal from dialysis during the first 120 d. Mortality risk was highest in the first 120 d after HD initiation. Inadequate predialysis nephrology care was strongly associated with mortality during this period, highlighting the potential benefits of contact with a nephrologist at least 1 mo before HD initiation.
Among hemodialysis patients, achieved hemoglobin is associated with Epoetin alfa dose and erythropoietin responsiveness. A prospective erythropoietin responsiveness measure was developed and its ...association with mortality evaluated.
Data from 321 participants were used and randomized to the hematocrit normalization arm of the Normal Hematocrit Cardiac Trial. Subjects were to receive a 50% Epoetin alfa dose increase at randomization. The prospective erythropoietin responsiveness measure was defined as the ratio of weekly hematocrit change (over the 3 wk after randomization) per Epoetin alfa dose increase (1000 IU/wk) corresponding to the mandated 50% dose increase at randomization. The distribution of responsiveness was divided into quartiles. Over a 1-yr follow-up, Cox proportional hazard modeling evaluated associations between this responsiveness measure and mortality.
Erythropoietin responsiveness values ranged from -2.1% to 2.4% per week per 1000 IU. Although subjects were similar across response quartiles, mortality ranged between 14% and 34% among subjects in the highest and lowest response quartiles (P = 0.0004), respectively. After adjusting for baseline prognostic indicators, highest versus lowest responsiveness was associated with a hazard ratio of 0.41 (95% confidence interval, 0.20 to 0.87).
Lower erythropoietin responsiveness is a strong, independent predictor of mortality risk and should be considered when evaluating associations between clinical outcomes and potential prognostic indicators, such as Epoetin alfa dose and achieved hemoglobin values.
Novel approaches to breast cancer screening are necessary, especially in the developing world where mammography is not feasible. In this study, we explored the hypothesis that blood-based biomarkers ...have potential for biomarkers for breast cancer.
We first determined the frequency of aberrant methylation of four candidate genes (APC, GSTP1, Rassf1A, and RARbeta2) in primary breast cancer tissues from West African women with predominantly advanced cancers. We used a high-throughput DNA methylation assay (quantitative methylation-specific polymerase chain reaction) to examine plasma from 93 women with breast cancer and 76 controls for the presence of four methylated genes. Samples were randomly divided evenly into training and validation data sets. Cutoff values for gene positivity of the plasma-based assay and the gene panel were determined by receiver operating characteristic curves in the training data set and subsequently evaluated as a screening tool in the validation data set.
Methylation of at least one gene resulted in a sensitivity of 62% and a specificity of 87%. Moreover, the assay successfully detected 33% (eight of 24) of early-stage tumors.
These data suggest that epigenetic markers in plasma may be of interest for detection of breast cancer. Identification of additional breast cancer specific methylated genes with higher prevalence in early stage cancers would improve this approach.
There is increasing interest in utilizing real‐world data (RWD) to produce real‐world evidence (RWE) on the benefits and risks of medical products that could support regulatory approval decisions. ...The field of pharmacoepidemiology has a long history of focusing on data and evidence that would now be termed “real‐world,” including evidence from healthcare claims, registries, and electronic health records. However, several emerging trends over the past decade are converging to support the use of these and other RWD sources for approval decisions, and there are several recent examples and ongoing research that demonstrate how RWE may be used to support regulatory approval of new or expanded indications. The goal of this article is to review the current landscape and future directions of the use of RWE in this context. This manuscript is endorsed by the International Society for Pharmacoepidemiology (ISPE).
Thanks to the increasing availability of genomics and other biomedical data, many machine learning algorithms have been proposed for a wide range of therapeutic discovery and development tasks. In ...this survey, we review the literature on machine learning applications for genomics through the lens of therapeutic development. We investigate the interplay among genomics, compounds, proteins, electronic health records, cellular images, and clinical texts. We identify 22 machine learning in genomics applications that span the whole therapeutics pipeline, from discovering novel targets, personalizing medicine, developing gene-editing tools, all the way to facilitating clinical trials and post-market studies. We also pinpoint seven key challenges in this field with potentials for expansion and impact. This survey examines recent research at the intersection of machine learning, genomics, and therapeutic development.
The genome contains instructions for building the function and structure of organisms. Recent high-throughput techniques have made it possible to generate massive amounts of genomics data. However, there are numerous roadblocks on the way to turning genomic data into tangible therapeutics. We observe that genomics data alone are insufficient for therapeutic development. We need to investigate how genomics data interact with other types of data such as compounds, proteins, electronic health records, images, and texts. Machine learning techniques can be used to identify patterns and extract insights from these complex data. In this review, we survey a wide range of genomics applications of machine learning that can enable faster and more efficacious therapeutic development. Challenges remain, including technical problems such as learning under different contexts given low-resource constraints, and practical issues such as mistrust of models, privacy, and fairness.
Recent high-throughput techniques have made it possible to generate massive amounts of genomics data. However, there are numerous roadblocks on the way to turning genomic data into tangible therapeutics. We need to investigate how genomics data interact with other types of data such as compounds, proteins, electronic health records, images, and texts. In this review, we survey a wide range of genomics applications of machine learning that can enable faster and more efficacious therapeutic development.
Many cases of mucopurulent cervicitis (MPC) are idiopathic and cannot be attributed to the known cervical pathogens Neisseria gonorrhoeae, Chlamydia trachomatis or herpes simplex virus. Because ...Mycoplasma genitalium is associated with nongonoccocal urethritis in men, its role in MPC, the corresponding syndrome in women, was investigated. Archived cervical specimens from women recruited in the Harborview Sexually Transmitted Disease Clinic in Seattle from 1984 to 1986 were tested, using polymerase chain reaction, in a study that identified other causes of and risk factors for MPC. M. genitalium was detected in 50 (7.0%) of 719 women. Young age, multiple recent partners, prior miscarriage, smoking, menstrual cycle, and douching were positively associated with M. genitalium whereas bacterial vaginosis and cunnilingus were negatively associated. After adjustment for age, phase of menstrual cycle, and presence of known cervical pathogens, women with M. genitalium had a 3.3-fold greater risk (95% confidence interval, 1.7–6.4) of MPC, which suggests that this organism may be a cause of MPC
ObjectiveTo examine age, gender, and temporal differences in baseline characteristics and clinical outcomes of adult patients hospitalised with COVID-19.DesignA cohort study using deidentified ...electronic medical records from a Global Research Network.Setting/Participants67 456 adult patients hospitalised with COVID-19 from the USA; 7306 from Europe, Latin America and Asia-Pacific between February 2020 and January 2021.ResultsIn the US cohort, compared with patients 18–34 years old, patients ≥65 had a greater risk of intensive care unit (ICU) admission (adjusted HR (aHR) 1.73, 95% CI 1.58 to 1.90), acute respiratory distress syndrome(ARDS)/respiratory failure (aHR 1.86, 95% CI 1.76 to 1.96), invasive mechanical ventilation (IMV, aHR 1.93, 95% CI, 1.73 to 2.15), and all-cause mortality (aHR 5.6, 95% CI 4.36 to 7.18). Men appeared to be at a greater risk for ICU admission (aHR 1.34, 95% CI 1.29 to 1.39), ARDS/respiratory failure (aHR 1.24, 95% CI1.21 to 1.27), IMV (aHR 1.38, 95% CI 1.32 to 1.45), and all-cause mortality (aHR 1.16, 95% CI 1.08 to 1.24) compared with women. Moreover, we observed a greater risk of adverse outcomes during the early pandemic (ie, February–April 2020) compared with later periods. In the ex-US cohort, the age and gender trends were similar; for the temporal trend, the highest proportion of patients with all-cause mortality were also in February–April 2020; however, the highest percentages of patients with IMV and ARDS/respiratory failure were in August–October 2020 followed by February–April 2020.ConclusionsThis study provided valuable information on the temporal trends of characteristics and outcomes of hospitalised adult COVID-19 patients in both USA and ex-USA. It also described the population at a potentially greater risk for worse clinical outcomes by identifying the age and gender differences. Together, the information could inform the prevention and treatment strategies of COVID-19. Furthermore, it can be used to raise public awareness of COVID-19’s impact on vulnerable populations.
Post-marketing safety studies of medicines often rely on administrative claims databases to identify adverse outcomes following drug exposure. Valid ascertainment of outcomes is essential for ...accurate results. We aim to quantify the validity of diagnostic codes for serious hypocalcemia and dermatologic adverse events from insurance claims data among women with postmenopausal osteoporosis (PMO).
We identified potential cases of serious hypocalcemia and dermatologic events through ICD-9 diagnosis codes among women with PMO within claims from a large US healthcare insurer (June 2005-May 2010). A physician adjudicated potential hypocalcemic and dermatologic events identified from the primary position on emergency department (ED) or inpatient claims through medical record review. Positive predictive values (PPVs) and 95% confidence intervals (CIs) quantified the fraction of potential cases that were confirmed.
Among 165,729 patients with PMO, medical charts were obtained for 40 of 55 (73%) potential hypocalcemia cases; 16 were confirmed (PPV 40%, 95% CI 25-57%). The PPV was higher for ED than inpatient claims (82 vs. 24%). Among 265 potential dermatologic events (primarily urticaria or rash), we obtained 184 (69%) charts and confirmed 128 (PPV 70%, 95% CI 62-76%). The PPV was higher for ED than inpatient claims (77 vs. 39%).
Diagnostic codes for hypocalcemia and dermatologic events may be sufficient to identify events giving rise to emergency care, but are less accurate for identifying events within hospitalizations.