Pharmacoepidemiology 's strength and limitation
Strengths and limitations (1 page)
Strength: It required Risk flags for 27 different chronic disease of each patient when extracting data to form the CCW database [1]. Rheumatoid arthritis and other comorbidities of interest including stroke, heart failure, hypertension, diabetes and depression were all recorded in this database [1]. The type of claims to qualify for rheumatoid arthritis in the CCW database included at least 2 inpatient records, skilled nursing facility (SNF) claims data, Home Health Agency (HHA) claims data, Hospital Outpatient (HOP) data or Carrier claims with DX codes (714.0~721.91) during the 2 year period while any combination of claims at least one day apart was implied to ensure quality of data [1]. By using such an informative database, we could explore patient’s disease information and study rare drug effects using data of good quality. As using the claims data, the usage of the drug in the real clinical practice could be captured and thus a more accurate result of the risk would be generated. Secondly, propensity score is a powerful way to collapse many variables and potential confounders into one proxy that the subjects’ characteristics would be more evenly distributed among the two study groups after matching. This method effectively eliminated confounding by indication and selection bias considering the situation of selectively prescription based on patient’s health condition. In addition, studies found that inclusion of predictors only predict exposure in small studies could lead to large standard errors [2]. A lot of variables involved in this study were predictors of both exposure and outcome, such as the C-Reactive Protein (CRP) that can effectively indicate the amount of inflammation. The imprecise result in previous clinical trials would be improved in this study.
Limitations: Naproxen and ibuprofen were available both by prescription and over the counter; however, the drug usage over the counter can’t be captured through the
Strength: It required Risk flags for 27 different chronic disease of each patient when extracting data to form the CCW database [1]. Rheumatoid arthritis and other comorbidities of interest including stroke, heart failure, hypertension, diabetes and depression were all recorded in this database [1]. The type of claims to qualify for rheumatoid arthritis in the CCW database included at least 2 inpatient records, skilled nursing facility (SNF) claims data, Home Health Agency (HHA) claims data, Hospital Outpatient (HOP) data or Carrier claims with DX codes (714.0~721.91) during the 2 year period while any combination of claims at least one day apart was implied to ensure quality of data [1]. By using such an informative database, we could explore patient’s disease information and study rare drug effects using data of good quality. As using the claims data, the usage of the drug in the real clinical practice could be captured and thus a more accurate result of the risk would be generated. Secondly, propensity score is a powerful way to collapse many variables and potential confounders into one proxy that the subjects’ characteristics would be more evenly distributed among the two study groups after matching. This method effectively eliminated confounding by indication and selection bias considering the situation of selectively prescription based on patient’s health condition. In addition, studies found that inclusion of predictors only predict exposure in small studies could lead to large standard errors [2]. A lot of variables involved in this study were predictors of both exposure and outcome, such as the C-Reactive Protein (CRP) that can effectively indicate the amount of inflammation. The imprecise result in previous clinical trials would be improved in this study.
Limitations: Naproxen and ibuprofen were available both by prescription and over the counter; however, the drug usage over the counter can’t be captured through the