Retrospective Database Analysis

Question	Answer
Randomized Controlled Trials (RCTs)	1) a research mtd that provides efficacy data. 2) Provide d strongest evidence that the drug is efficacious. 3) Efficacy indicates whether the drug "work" in a controlled study.
Randomization	used to decrease baseline differences between patient groups
Observational Studies	(1) follow what happens to patients who receive medical care in a “real-world” environment. (2) measure the “effectiveness” of treatments. (3) Effectiveness indicates whether the treatment “does work” in everyday medical practice
Prospective observational studies	record treatments and outcomes as they occur.
Retrospective observational studies	analyze treatments and outcomes that have already occurred.
Advantages of RCTs	(1) the randomization of patients improves the chances that patient characteristics are similar between study groups at baseline - this in turn, makes the efficacy differences more credible. (2)
Disadvantages of RCTs	(1) The time frame of RCTs may be too short to determine long-range outcomes, especially for chronic diseases such as diabetes and heart disease. (2) the sample size may be too small to detect some differences in outcomes such as rates of side effects. (3) RCTs may not be representative of the average patient population because of the study’s inclusion and exclusion criteria. (4) Patients enrolled in an RCT may behave differently or receive different levels of services compared with usual medical care.
Advantages of Observational Studies	(1) can overcome some of the restrictions of RCTs (2) especially if theyare retrospective, are less expensive and less time-consuming than RCTs because researchers do not have to enroll patients into the study. (3) Retrospective databases allow researchers to analyze many years' worth of patient information from a large number of records in a short period at a relatively low cost. (4) Observational data can reflect a more realistic picture of treatment patterns for a wider population of patients and factor in patients' true adherence to treatment regimens.
Disadvantages of Observational Studies (1)	(1) the information may be incomplete, inaccurate, or biased. (2) Most retrospective databases are collected and used primarily for payment (reimbursement) of treatment and services rather than to assess patient outcomes. (3) Patient information may be incomplete if a patient switches insurance companies during the study period or becomes ineligible to receive services during part of the study period.
Disadvantages of Observational Studies (2)	1. Data may also be incomplete if the patient pays out of pocket for some of his/her health care expenses, if the service is not covered by the patient's insurance plan, or if the patient has more than one form of insurance (e.g., both Medicaid and Medicare, insurance from an employer and the spouse's employer).
Disadvantages of Observational Studies (3)	(1) Data provided on diagnoses may be inaccurate. (2) - ICD-9 codes = Miscoding of diagnoses may occur, intentionally or unintentionally which may lead to undercooking or overcoming of diseases (ICD-9 codes determine reimbursement). (3) selection bias (4) channeling bias
Advantages of Prospective randomized, controlled trial	(1) More scientifically rigorous (2) Needed for FDA approval (3) Less chance of baseline differences (4) Can collect clinical and PRO [Patient reported outcome] outcome data
Disadvantages of Prospective randomized, controlled trial	(1) Only select patients (2) Short-term follow-up (3) High cost of recruitment and follow-up (4) Protocol-driven costs unrealistic in standard practice (5) Only highly adherent patients (6) Small sample size
Advantages of Retrospective database	(1) Includes broader range of patients (2) Lower cost to conduct study (3) Larger sample size (4) Longer time period (5) Easier to recalculate with different criteria (sensitivity analyses)
Disadvantages of Retrospective database	(1) Selection bias (2) May be only one population (3) May be missing data (4) Data may be miscoded (5) Clinical or PRO outcome data may be lacking
Examples of types of retrospective databases in PharmacoEconomics	(1) Electronic Medical Records (2) National Health Surveys (3) Insurance and Claims Records
Sensitivity analysis	For example, inclusion or exclusion criteria (e.g., age range, diagnoses, adherence to medication) for patient data used in the baseline analyses can be expanded or restricted to determine if results are robust (i.e., insensitive) for these different criteria ranges.
selection biases	For example, patients with a more severe form of a disease or who have been treatment resistant to standard therapy may receive (be “selected” to receive) the newer medication.
channeling bias	may occur if medication A is thought to cause an unwanted side effect such as weight gain more often than medication B. For patients who are already overweight or prone to gain weight, prescribers might opt for medication B in the hope of decreasing the chance of this side effect.
Electronic Medical Records [EMR]	(1) EMR systems provide detailed information about patient encounters and are a rich source of data for pharmacoeconomic research (2) provides more complete information tan other database mtds. (3) Costly to implement (4) Time consuming to retrieve the information needed for a study.
National Health Surveys [NHS]	provide high-quality information on health care utilization and health care costs, and they are generally representative of the US population.
The Medical Expenditure Panel Survey (MEPS) [1]	is a national health survey cosponsored by NCHS and the Agency for Healthcare Research and Quality (AHRQ) that produces estimates of health care use and health care expenditures for noninstitutionalized civilians living in the United States
The Medical Expenditure Panel Survey (MEPS) [2]	The MEPS survey includes questions from the SF-12 and EQ-5D to evaluate a patient's health status
Insurance and Claims Records [ICR]	(1) the data files usually contain specific information on a large number of patients.
Advantages of Insurance and Claims Records	(1) A major advantage of using claims data is that a large amount of patient information is available in an easily retrievable format, and the costs of setting up the data collection system have already been financed by the insurance company that needs the data to pay patient claims.
Disadvantages of Insurance and Claims Records	A major disadvantage of using insurance claims data is that they are often missing clinical information (e.g., laboratory results, patient's health-related quality of life changes) that may be needed to answer the research question.
5 issues to consider when Critiquing Retrospective Database Studies	(1) Clear Explanation of the Database (2) Clear Explanation of Patient-Selection Criteria (3) Eligibility of Patients (4) Clinical or Outcome Measures (5) Sensitivity Analyses
what to look for in "Clear Explanation of the Database"	(1) prescription data, medical service data, hospitalization data, laboratory values that are available in the database (2) databases they used for their study and for what period of time the data were collected. (3) the population that is covered by the database(s). (4) Any relevant formulary or payment restrictions specific to the insurance plan should also be discussed.
what to look for in "Clear Explanation of Patient-Selection Criteria"	(1) researchers include the criteria (i.e., rules) used to select patients for the study. (2) If specific diagnoses or the use of specific medications were used as a proxy to infer that patients had a specific condition or disease, these diagnoses or medications should be listed. (3) Exclusion criteria for patients should also be stated.
what to look for in " Eligibility of Patients"	(1) Some patients may not be “eligible” to receive insurance benefits during the entire period of study. (2) Patients might switch to another insurance plan or become ineligible for their current plan (e.g., they no longer work full time for their employer or no longer meet the criterion for “need” according to the Medicaid definition). (3) It is important to describe what steps are taken to address this issue. Many researchers list ineligibility at any time during the study period as an exclusion criterion.
what to look for in "Clinical or Outcome Measures" (1)	(1) medical claims databases are primarily used for reimbursement purposes, so they may not contain patient outcome variables such as blood pressure measurements or patient-reported outcomes (PROs)
what to look for in "Clinical or Outcome Measures" (2)	(1) Another issue that arises is that of “double-counting” outcomes (2) When the ICER is calculated and an additional hospitalization or ED visit occurs it is marked as an unwanted outcome, which appears in the denominator of the ICER and increases the costs in the numerator
what to look for in "Sensitivity Analyses"	(1) Sensitivity analyses of key design choices are relatively easy to execute and should be included. (2) additional analysis compares results using disease-specific costs and outcomes (e.g., those with a specific ICD-9) with those using total costs and outcomes the patients have incurred for the period of interest

Next up

Retrospective Database Analysis

Description

Resource summary

Similar

	Created by Sam Adeyiga about 4 years ago