Local to
national
Understanding the flow of case information is necessary in order to evaluate the surveillance system effectively. In the Caribbean, the flow of case information occurs:
-
from the healthcare facilities (clinics, hospitals and government/ private physician offices) to parish/county/district/health region level, as appropriate for the country
-
from district/county/parish to the national level
-
from the national level to the Caribbean regional level (CAREC).
Measuring the
true frequency
of HIV infection
/AIDS incidence
Completeness of reporting measures the proportion of all true cases that are reported to the surveillance system. This definition of completeness should not be confused with measuring the completeness of information that is collected on a case report form. Ensuring that a case report form is filled out correctly and completely is an important component of a high quality surveillance system and will be discussed as part of evaluating the validity of case data.
Methods to measure
completeness
Completeness of reporting should be evaluated for a specified time period, such as one year. You can measure completeness of reporting in one of two ways:
-
expanding surveillance activities to find (and report) any missed cases
-
estimating the proportion of all cases that were reported in a specified time period, using a capture-recapture methodology.
Expanding case
finding
Expanding case finding has the benefit of both measuring the completeness of reporting and of identifying missed cases that can be reported and added into the surveillance system.
Expanding case finding, continued
Expanding case finding can be done by:
Some examples of increasing active surveillance activities include:
-
visiting healthcare facilities from which passive reports have been received and reviewing the clinic records for missed cases
-
reviewing laboratory records at a facility to identify laboratory tests done that are likely to be indicative of HIV/AIDS, such as CD4 tests, HIV antibody or viral load tests or tests for Pneumocystis carinii.
Identifying a secondary database to match against previously reported cases can be challenging in certain settings. It requires that:
-
reported cases be maintained in an electronic database that lists each case separately and includes a name or unique identifier
-
the secondary database includes persons with HIV infection and lists each individual separately by the same identifier as the HIV/AIDS case registry
-
the secondary database has sufficient information to identify a person with HIV infection.
Examples of secondary databases that have been used for identifying missed cases and measuring the completeness of reporting include:
-
vital registries of deaths in which the cause of death is listed (and would include HIV and /or AIDS)
-
a registry of patients receiving HIV-disease-related medications.
To calculate the completeness of reporting, divide the number of reported cases by the number of reported cases plus the missed cases (the total of known cases). Usually completeness is presented as a percent.
Number of previously reported cases
Number of previously reported cases + newly identified (that is, missed) cases
Either of these methods is likely to result in the identification of cases that were not reported. Once these cases have been identified they should be reported.
Capture-recapture
methods
Capture-recapture is a method that estimates the completeness of reporting but does not identify missed cases to report. This system can be used in areas in which reports from multiple sources that concern a single individual with HIV infection are received at the surveillance unit. For example, consider reports that pertain to one person received from:
-
laboratory reporting
-
case reporting from a clinic
-
case reporting from a hospital.
The capture-recapture methodology assumes that if all of the reports are collected, this represents the true universe of cases. To help conceptualise how this method works, consider the following example, in which HIV case reporting is done using a unique code and that case reports come from several sources.
A database is developed, into which all reports concerning all HIV-infected persons are recorded. The first column represents five distinct individuals reported with HIV. The next two columns are facilities from which reports are received. One might be a clinic, another might be a laboratory, etc. A ‘1’ means that a report on this person was received from this site. A ‘0’ means that a report was not received from this facility.
Table 5.2. Example of case reports from two facilities.
-
Case number
|
Facility A
|
Facility B
|
X239
|
1
|
0
|
H750
|
0
|
0
|
S000
|
1
|
1
|
W298
|
0
|
1
|
T298
|
0
|
1
|
Use this information to develop a 2X2 table, as follows:
Reported from:
Facility A
-
Facility B
|
|
Yes
|
No
|
Yes
|
a
|
b
|
No
|
c
|
x
|
In this table, ‘a’ represents the cases reported from both sources, ‘b’ are the cases reported only from Facility B, ‘c’ are the cases reported only from Facility A, and ‘x’ represents missed cases. Then, ‘x’ can be estimated by the following formula: (bc)/a. Once ‘x’ is determined, the
Capture-recapture methods, continued
total number of estimated cases and completeness of reporting can be calculated.
Here are the results, using the example of case reports from Facilities A and B.
Reported from:
Facility A
-
Facility B
|
|
Yes
|
No
|
Yes
|
2
|
2
|
No
|
1
|
x
|
(2X1)/2=1
Thus, X =1. Our total number of cases is the sum of all boxes. This equals 6. The completeness of reporting is the number of counted cases divided by the total estimated cases. In our example, this is 5/6 or 0. 83. Completeness is usually presented as a percent. So in this example, completeness is 83.3%.
Capture-recapture can also be used in a single setting. In this situation, two time periods are used. For example, instead of Facility A we would collect reports from a single clinic for six months (January through June). These reports would be recorded. Reports from the same clinic will be collected for the period July through December. In this situation your record log would look like this.
-
Case number
|
Reports from Jan-June
|
Reports from July-Dec
|
X239
|
1
|
0
|
H750
|
0
|
0
|
S000
|
1
|
1
|
W298
|
0
|
1
|
T298
|
0
|
1
|
The analysis would be conducted as described for reports from two different clinics.
Although these examples provide the conceptual aspects to capture-recapture, there are a number of assumptions that must be made. One of these is that the reports must all be independent from one another. We know that in this situation, case reports are not independent. Therefore, for an accurate estimate of the completeness of reporting, adjustments must be made. A number of statistical tests can be used to determine the
Capture-recapture methods, continued
interaction between reporting sites and to develop appropriate adjustments to the analysis. References regarding use of capture-recapture methods for estimating completeness of reporting are listed at the end of this unit.
Share with your friends: |