The triangular test is for binary rather than survival endpoints
Neal Alexander, London School of Hygiene and Tropical Medicine
5 October 2015
This is a useful paper which we have used and cited in our work on leishmaniasis (Omollo et al 2011). Clinical trials of visceral leishmaniasis therapy require a long follow-up, 6 to 12 months, to reach their outcome of definitive cure or treatment failure. It might seem efficient to define the endpoint as soon as any rescue treatment is given, rather than waiting for a fixed timepoint.
However the theory behind the triangular test is based on normal approximations to binomial outcomes (Whitehead 1983), not time-to-event ones as in survival analysis. Three of the five people in Ranque et al.’s Table 1 relapsed, and two were censored. All five appear in Table 2, however, so the censored ones were apparently included as non-relapse from their time of entry. Assigning someone a positive outcome when they have had little time at risk would seem to under-estimate relapse.
In leishmaniasis trials, analysing a survival-type outcome would suffer from the opposite problem, because a cure cannot be registered till the end of the at-risk period, while a failure can be registered at any time after the end of treatment. So, for example, failures may cause the trial to be stopped, even if a large proportion of patients were progressing well but not counting as cures — in fact not entering into the analysis at all — because they had yet to reach their definitive assessment.
We conclude that the triangular test is suited to a binary endpoint at a fixed time point, but may stop too early or too late — depending on how analysed — if negative outcomes are registered continually over time.
Omollo, R., N. Alexander, T. Edwards, E. Khalil, B. Musa, A. Abdalla, M. Wasunna, N. Njoroge, D. Kinoti, G. Kirigi, T. P. C. Dorlo, S. Ellis, M. Balasegaram, A. Musa and on behalf of the Leishmaniasis East Africa Platform (LEAP) Study Group (2011). Safety and efficacy of miltefosine alone and in combination with sodium stibogluconate and liposomal amphotericin B for the treatment of primary visceral leishmaniasis in East Africa: study protocol for a randomized controlled trial. Trials 12: 166.
Whitehead, J. (1983). The Design and Analysis of Sequential Clinical Trials. Chichester, Ellis Horwood.
The triangular test is for binary rather than survival endpoints
5 October 2015
This is a useful paper which we have used and cited in our work on leishmaniasis (Omollo et al 2011). Clinical trials of visceral leishmaniasis therapy require a long follow-up, 6 to 12 months, to reach their outcome of definitive cure or treatment failure. It might seem efficient to define the endpoint as soon as any rescue treatment is given, rather than waiting for a fixed timepoint.
However the theory behind the triangular test is based on normal approximations to binomial outcomes (Whitehead 1983), not time-to-event ones as in survival analysis. Three of the five people in Ranque et al.’s Table 1 relapsed, and two were censored. All five appear in Table 2, however, so the censored ones were apparently included as non-relapse from their time of entry. Assigning someone a positive outcome when they have had little time at risk would seem to under-estimate relapse.
In leishmaniasis trials, analysing a survival-type outcome would suffer from the opposite problem, because a cure cannot be registered till the end of the at-risk period, while a failure can be registered at any time after the end of treatment. So, for example, failures may cause the trial to be stopped, even if a large proportion of patients were progressing well but not counting as cures — in fact not entering into the analysis at all — because they had yet to reach their definitive assessment.
We conclude that the triangular test is suited to a binary endpoint at a fixed time point, but may stop too early or too late — depending on how analysed — if negative outcomes are registered continually over time.
Neal Alexander, Fabiana Alves, Sakib Burza, Tansy Edwards,
Omollo, R., N. Alexander, T. Edwards, E. Khalil, B. Musa, A. Abdalla, M. Wasunna, N. Njoroge, D. Kinoti, G. Kirigi, T. P. C. Dorlo, S. Ellis, M. Balasegaram, A. Musa and on behalf of the Leishmaniasis East Africa Platform (LEAP) Study Group (2011). Safety and efficacy of miltefosine alone and in combination with sodium stibogluconate and liposomal amphotericin B for the treatment of primary visceral leishmaniasis in East Africa: study protocol for a randomized controlled trial. Trials 12: 166.
Whitehead, J. (1983). The Design and Analysis of Sequential Clinical Trials. Chichester, Ellis Horwood.
Competing interests
No competing interests.