Table 1. Proposed primary outcomes for evaluation of ART initiation in general adult populations.
| Outcome | Parameter | Definition |
|---|---|---|
| ART initiation within 28 days | Equation | ART initiation equals the number of treatment-eligible patients initiating ART within 28 days of first HIV-related clinic visit divided by the total number of treatment-eligible patients. |
| Timing | The interval allowed for patients to initiate ART after clinic presentation varies in previous reports, with recent studies using intervals ranging from 14 days [20] to 90 days [19]. In all the studies presented at the MATI consultation, more than half of patients in the standard of care arm initiated treatment within one month (28 days) of initial clinic presentation, confirming that one month is sufficient time for all procedures to be completed. Within 28 days of first treatment-eligible, HIV-related clinic visit was thus recommended as the standard interval for this outcome. We also note that the starting point of the interval is the first interaction with the healthcare system at which the patient is eligible for ART. Under the new WHO guidelines, this will coincide with a positive HIV test, as diagnosis and eligibility will be simultaneous. Until the treat-all approach is adopted, it is important to specify the starting point. | |
| Denominator | All patients found to be eligible for ART during the study enrollment period (study cohort). Although nearly all patients will be eligible once the treat-all approach is adopted, for the time being most countries continue to apply a CD4 count threshold for eligibility. The denominator for this outcome should include all patients who are eligible, whether or not their eligibility has been conveyed to them. Thus, a patient who has a CD4 count under the threshold but does not return to obtain the CD4 count result should be included in the denominator. The enrollment period should be specified with starting and ending dates. | |
| Numerator | Patients in the study cohort who are dispensed their first supply of ARV medications within the allowed interval (28 days). Where prescribing and dispensing are separate steps, dispensing to the patient is the preferred indicator. | |
| Six-month retention in care* | Equation | Early retention in care equals the number of patients retained in care six months after the expected date of treatment initiation divided by the number of treatment-eligible patients expected to have initiated ART. |
| Timing | Since the reason for post-initiation follow-up in studies like those proposed here is to ensure that the manner of ART initiation does not harm post-initiation outcomes, short-term retention in care is a reasonable and readily measurable outcome. Six months from the expected date of treatment initiation is recommended as the standard interval for this outcome. | |
| Denominator | All patients found to be eligible for ART during the study enrollment period (study cohort). Because we are evaluating treatment initiation, not retention in care, a patient who is eligible for treatment but never initiates should be included in the denominator. For this outcome, there is no ART initiation date for patients who are lost before initiation. For these patients, the outcome could be assessed one month (28 days) plus the specified interval (e.g., six months) after the first HIV-related clinic visit. This allows the patient the same 28 days to start ART as suggested for the ART initiation outcome, plus the same duration of potential follow-up as the patients who did start ART. The enrollment period should be specified with starting and ending dates. | |
| Numerator | Patients in the study cohort who fulfill the selected definition of retained in care, which is typically not more than a specified number of days late for the next scheduled visit. Definitions of “retained” will vary by national guidelines, data availability, and researcher norms; not more than 90 days late for the next scheduled medication pickup is a commonly used definition. This outcome can be used even when viral load tests are not done, as it depends only on clinic visit data. | |
| Early viral suppression | Equation | Early viral suppression equals the number of patients virally suppressed at first routine viral load test divided by the number of treatment-eligible patients expected to be virally suppressed. |
| Timing | The timing of this outcome will depend on when routine viral load tests are done under national guidelines in countries that adopt viral load monitoring. WHO recommends the first routine viral load test at six months after treatment initiation [26], but many countries wait until 12 months. For countries with the first routine viral load test at 12 months, viral suppression by 13 months after the expected date of treatment initiation could be used as the outcome, allowing patients a one-month window after the scheduled date while also capturing viral loads that may have been suppressed before the 12-month point. For countries with the first routine viral load test at six months, viral suppression by seven months after the expected date of treatment initiation could be used as the outcome. | |
| Denominator | All patients found to be eligible for ART during the study enrollment period (study cohort). Because we are evaluating treatment initiation, not retention in care, a patient who is eligible for treatment but never initiates should be included in the denominator and considered not to have reached the outcome. For this outcome, there is no ART initiation date for patients who are lost before initiation. For these patients, the outcome could be assessed one month (28 days) plus the specified interval (e.g., 12 months) after the first HIV-related clinic visit. This allows the patient the same 28 days to start ART as suggested for the ART initiation outcome, plus the same duration of potential follow-up as the patients who did start ART. The enrollment period should be specified with starting and ending dates. | |
| Numerator | Patients in the study cohort who have a recorded viral load below the definition threshold for suppression that is used in the country. There is variation in this threshold, and using different countries’ standards may reduce the comparability of studies. For operations research, however, it is appropriate to use the accepted local threshold (and this is likely what routinely collected data will permit, as well). Patients who do not have viral load test results recorded should not be included in the numerator for this outcome. |
*For all the outcomes in Table 1, but particularly for retention in care, patients who transfer to other clinics before reaching the outcome endpoint pose an analytic challenge, as their outcomes are usually unknown. For patients who transfer formally, it may be possible to obtain the outcome from the new clinic, or the transfer may occur close enough to an endpoint to infer it from available data. Patients who self-transfer without informing the original clinic (so called “silent transfers”) are typically counted as not retained, a practice that may overestimate actual loss to care but is difficult to correct without active tracing of defaulters or a universal patient identifier.