Statistical Process Improvement
Georgetown University
 

 Matched Case Control Study


 

Assigned Reading

  1. Introduction (use instructor's last name as password) Read►

Presentations

Assignments

Question 1: These data show a list of 9 patients.  Two of the patients have received an experimental treatment.  Identify which of the patients in the following table could serve as controls for the two cases.  Select the possible controls randomly using the lowest random number.  Match on age.  Show a Kaplan Meier survival curve.  For each month, show the percent of cases or controls that have not fallen since the start of the study. Solution►

Patient ID

Received Intervention

Age

Months to Fall

Random Number

Observation Period

Follow-up Period

1

Yes

65

 

3

0.24

2

No

60

 

2

0.85

3

Yes

84

2

 

0.64

4

No

82

4

 

0.7

5

No

78

 

0.87

6

No

80

3

 

0.72

7

No

79

 

 

0.86

8

No

64

 

 

0.16

9

No

70

 

2

0.17

Question 2: Establish the percent of days the following patient is in therapeutic range.  On days when blood pressure was not measured, assume that blood pressure is the same as the nearest data point less than 7 days away.  For all other days assume that the blood pressure is unknown.  Solution►

Days Since Enrollment

BP

Therapeutic Range

In 
range

Lower

Upper

0

125

120

140

Yes

20

150

120

140

No

60

130

120

140

Yes

70

110

120

140

No

90

130

120

140

Yes

120

135

120

140

Yes

Question 3: Hypertension is a pervasive medical condition in the U.S. that can often be managed through diet. A study was performed to examine the impact of diet in controlling hypertension by reducing sodium intake.  A control group was given a diet rich in fruits and vegetables, while a randomly selected study group was given a diet that also included low-fat dairy products.  Both groups initially participated in a 3-week period in which they consumed a ‘typical’ American diet.  During this period the baseline data for sodium intake were obtained (see Table below).  This is a type of statistical problem that you might face in a course in health systems research. Answer►


  • Participants 7, 8 and 12 were randomly selected to participate in the study group. Create a data table from which to perform the matching analysis. The table should include the Participant Number, Study or Control Group Designation, Daily Sodium Level, and Random Number Assignment.
  • Matching on Daily Sodium Intake, which three participants would be the best controls for Participant 7?
  • Which three participants would be the best controls for Participant 12?
  • For cases and controls, verify that the matching was successful.

Question 4: Venous thromboembolism (VTE) is a medical condition in which blood clots form within a vein that can block the vein or break off and become a life-threatening pulmonary embolism.  One treatment for VTE is the use of anticoagulants such as warfarin. Anticoagulants are effective, but their level in the patient’s blood must be actively monitored.  If it is too high, the patient is at risk for internal bleeding; too low and anticoagulant won’t have a therapeutic effect.  Dosing for anticoagulants is monitored using INR, a measure that represents the ease with which blood clots. The target INR for effective treatment of VTE is between 2.0 and 3.0. Clinical protocols suggest that patients with a transient risk for VTE be put on warfarin for a three-month period.  During this time the INR is monitored weekly.  Table below represents the experience of a typical patient. This is a type of statistical problem that you might face in a pharmacy course. Answer►

  • Plot the observation data. Calculate the days on which the patient crossed control limits.
  • Calculate the number of days the patient was in therapeutic range.
  • Calculate the unknown days, and the percentage of days in range assuming a 3-day therapeutic horizon.

More

  • A review of matched case control studies PubMed► 
  • Measurement errors in matched case control studies Pubmed►
  • Treatment of missing data in matched case control studies Pubmed►
  • Kaplan Meier survival analysis using Excel Video►

 

 
Prepared by Farrokh Alemi, Ph.D. Most recent revision 06/01/2017.  This page is part of the course on Statistical Process Improvement in EHR.