03.03.2014 Views

How to investigate Adherence to Antiretroviral Treatment ... - INRUD

How to investigate Adherence to Antiretroviral Treatment ... - INRUD

How to investigate Adherence to Antiretroviral Treatment ... - INRUD

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

CHAPTER 4. SURVEY DESIGN<br />

Sampling Facilities<br />

Carrying out a survey of these indica<strong>to</strong>rs could be done in a single facility or in a sample of<br />

facilities. A survey <strong>to</strong> examine adherence in a large programme or system of care, such as the<br />

National AIDS Program, should include a minimum of 20 health facilities. If a system of care<br />

includes fewer facilities, then all of them should be included in a survey <strong>to</strong> measure system<br />

performance. A survey in a single facility only reflects the performance in that facility so it<br />

cannot be used <strong>to</strong> represent the performance in the country (unless it is the only facility<br />

providing treatment).<br />

Facilities are best selected randomly within specific strata defined by key characteristics such<br />

as geographic location, facility type, and facility management.<br />

The sample of facilities should be as randomly chosen as is feasible, taking in<strong>to</strong> account the<br />

logistics of travel and the days the clinics are open.<br />

The retrospective sample of patients is 100, so that it is preferable <strong>to</strong> only choose facilities<br />

that had at least 100 patients on ARVs six months ago. <strong>How</strong>ever, if one wants <strong>to</strong> look at<br />

smaller facilities, this can be done but instead of sampling, all patients’ records should be<br />

looked at.<br />

The data <strong>to</strong> calculate each indica<strong>to</strong>r should be collected at each facility. The sample sizes<br />

suggested are sufficient for a moderately reliable set of adherence measures at each facility<br />

(such as when moni<strong>to</strong>ring performance over time) and a very reliable cross-sectional or<br />

longitudinal estimates of these measures in the system as a whole. An explanation of sample<br />

sizes and their accuracy is included in Appendix 4.<br />

Sampling Retrospective Patient Records<br />

The purpose of this exercise is <strong>to</strong> give us all the adherence indica<strong>to</strong>rs except the self report of<br />

the exit interview. As such it is the single most important exercise of the survey.<br />

Retrospective data from attendance records and pharmacy records are useful because they<br />

allow computation of indica<strong>to</strong>rs related <strong>to</strong> success of short-term and long-term adherence,<br />

defaulting, and clinical outcomes. It is necessary <strong>to</strong> take a sample of 100 patients who<br />

attended the clinic during the month seven months before. To end up with information on 100<br />

patients, it is advisable <strong>to</strong> sample 120 patients from the list of those who attended during that<br />

month as some records may be unavailable.<br />

This means that if the data collection is taking place in June, you need the patients who<br />

attended in November the year before. This is because you need <strong>to</strong> follow the patient for six<br />

whole months and if the patient attended on the last day of November, then six months from<br />

then would be the last day of May. Depending on the month of data collection, the months <strong>to</strong><br />

sample patient attendance from is documented in table 6.<br />

23

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!