Good Research Practices for Comparative Effectiveness Research: Defining, Reporting and Interpreting Nonrandomized Studies of Treatment Effects Using Secondary Data Sources: The ISPOR Good Research Practices for Retrospective Database Analysis Task Force Report—Part I

IMPORTANCE Failure of bone fracture healing occurs in 5% to 10% of all patients. Nonunion risk is associated with the severity of injury and with the surgical treatment technique, yet progression to nonunion is not fully explained by these risk factors.OBJECTIVE To test a hypothesis that fracture characteristics and patient-related risk factors assessable by the clinician at patient presentation can indicate the probability of fracture nonunion.DESIGN, SETTING, AND PARTICIPANTS An inception cohort study in a large payer database of patients with fracture in the United States was conducted using patient-level health claims for medical and drug expenses compiled for approximately 90.1 million patients in calendar year 2011. The final database collated demographic descriptors, treatment procedures as per Current Procedural Terminology codes; comorbidities as per International Classification of Diseases, Ninth Revision codes; and drug prescriptions as per National Drug Code Directory codes. Logistic regression was used to calculate odds ratios (ORs) for variables associated with nonunion. Data analysis was performed from January 1, 2011, to December 31, 2012, EXPOSURES Continuous enrollment in the database was required for 12 months after fracture to allow sufficient time to capture a nonunion diagnosis. RESULTSThe final analysis of 309 330 fractures in 18 bones included 178 952 women (57.9%); mean (SD) age was 44.48 (13.68) years. The nonunion rate was 4.9%. Elevated nonunion risk was associated with severe fracture (eg, open fracture, multiple fractures), high body mass index, smoking, and alcoholism. Women experienced more fractures, but men were more prone to nonunion. The nonunion rate also varied with fracture location: scaphoid, tibia plus fibula, and femur were most likely to be nonunion. The ORs for nonunion fractures were significantly increased for risk factors, including number of fractures (OR, 2.65; 95% CI, 2.34-2.99), use of nonsteroidal anti-inflammatory drugs plus opioids (OR, 1.84; 95% CI, 1.73-1.95), operative treatment (OR, 1.78; 95% CI, 1.69-1.86), open fracture (OR, 1.66; 95% CI, 1.55-1.77), anticoagulant use (OR, 1.58; 95% CI, 1.51-1.66), osteoarthritis with rheumatoid arthritis (OR, 1.58; 95% CI, 1.38-1.82), anticonvulsant use with benzodiazepines (OR, 1.49; 95% CI, 1.36-1.62), opioid use (OR, 1.43; 95% CI, 1.34-1.52), diabetes (OR, 1.40; 95% CI, 1.21-1.61), high-energy injury (OR, 1.38; 95% CI, 1.27-1.49), anticonvulsant use (OR, 1.37; 95% CI, 1.31-1.43), osteoporosis (OR, 1.24; 95% CI, 1.14-1.34), male gender (OR, 1.21; 95% CI, 1.16-1.25), insulin use (OR, 1.21; 95% CI, 1.10-1.31), smoking (OR, 1.20; 95% CI, 1.14-1.26), benzodiazepine use (OR, 1.20; 95% CI, 1.10-1.31), obesity (OR, 1.19; 95% CI, 1.12-1.25), antibiotic use (OR, 1.17; 95% CI, 1.13-1.21), osteoporosis medication use (OR, 1.17; 95% CI, 1.08-1.26), vitamin D deficiency (OR, 1.14; 95% CI, 1.05-1.22), diuretic use (OR, 1.13; 95% CI, 1.07-1.18), and renal insufficiency (OR, 1.11; 95% CI, 1.04-1.17) (multivariate P < .001 for all).C...

show abstract

“…5 We tested this hypothesis in a large payer database of patients with fracture in the United States. 8 …”

mentioning

confidence: 99%

Epidemiology of Fracture Nonunion in 18 Human Bones

et al. 2016

View full text Add to dashboard Cite

show abstract

“…CER is not a license to do away with RCTs but rather represents an opportunity to complement and go beyond RCTs into areas not otherwise addressed. [18]. Thomas and Peterson have encouraged the use of prospectively defined statistical analysis plans for observational research similar to those routinely provided with RCTs, including specification of the population, endpoints, objectives, hypotheses, and statistical methods (such as the handling of missing data and methods for addressing potential confounding and interaction) [19].…”

Section: Observational Studiesmentioning

confidence: 99%

Comparative Effectiveness Research in Oncology

Lyman

2013

The Oncologist

View full text Add to dashboard Cite

Although randomized controlled trials represent the gold standard for comparative effective research (CER), a number of additional methods are available when randomized controlled trials are lacking or inconclusive because of the limitations of such trials. In addition to more relevant, efficient, and generalizable trials, there is a need for additional approaches utilizing rigorous methodology while fully recognizing their inherent limitations. CER is an important construct for defining and summarizing evidence on effectiveness and safety and comparing the value of competing strategies so that patients, providers, and policymakers can be offered appropriate recommendations for optimal patient care. Nevertheless, methodological as well as political and social challenges for CER remain. CER requires constant and sophisticated methodological oversight of study design and analysis similar to that required for randomized trials to reduce the potential for bias. At the same time, if appropriately conducted, CER offers an opportunity to identify the most effective and safe approach to patient care. Despite rising and unsustainable increases in health care costs, an even greater challenge to the implementation of CER arises from the social and political environment questioning the very motives and goals of CER. Oncologists and oncology professional societies are uniquely positioned to provide informed clinical and methodological expertise to steer the appropriate application of CER toward critical discussions related to health care costs, cost-effectiveness, and the comparative value of the available options for appropriate care of patients with cancer. The Oncologist 2013;18:752-759 Implications for Practice: Escalating health care costs, largely driven by expensive new technologies and therapies, have prompted calls for more rigorous assessment of the effectiveness, safety, and overall value. At the same time, the time and resources required for large randomized controlled trials, as well as their recognized limitations, necessitate consideration of other forms of comparative effectiveness research for evaluating meaningful clinical outcomes. Through the results derived from selective and appropriate use of a range of comparative effectiveness methodological tools, clinicians, patients, payers, and policy makers can make more rationale choices in selecting the most appropriate, efficient, and cost-effective cancer care. At the same time, practicing oncologists and other stakeholders must be aware of the challenges presented by such research approaches as well as the considerable progress that has been made in addressing these limitations in order to appropriately weigh the totality evidence on critical health care issues in every day oncology practice. INTRODUCTIONComparative effectiveness research (CER) attempts to compare the benefits and harms of alternative strategies for diagnosing, treating, or preventing disease in patients. Driven primarily by new therapies and technologies, health care expenditures c...

show abstract

“…Contrary to any preconceptions, the use of databases for pharmacoepidemiology research is challenging -"It is not a simple process of get some data and do some statistics" (Harpe, 2009). Challenges of conducting studies throughout databases include concerns about study design, careful understanding of the underlying health care system in which the data were generated, data quality, limited ability to control confounding in the absence of randomization and to handle bias, and data analysis, among others (Berger et al, 2009). …”

Section: Use Of Automated Databases: Pitfalls and Methodological Chalmentioning

confidence: 99%