Tobias Broger and colleagues1 reported an evaluation of a urine point-of-care test for tuberculosis targeting lipoarabinomannan. The authors tested frozen specimens retrieved from a biorepository and acknowledged that a limitation of their study was the potential difference in analytical performance between frozen and fresh urine samples.We declare no competing interests.We investigated the effect of freezing on lipoarabinomannan concentrations of unprocessed patient urine. We measured the concentration of lipoarabinomannan in urine samples (n=15) before and after freezing to −70°C in polyethylene terephthalate tubes using an in-house ELISA. Samples were frozen for a minimum of 1 day before being retested. The mean lipoarabinomannan concentration decreased from 138 pg/mL (SD 99) to 76 pg/mL (SD 77), while the average loss was 50%. These results indicate that fresh samples contain more detectable lipoarabinomannan, which might improve the sensitivity of the described test. The use of fresh samples might be most crucial in patients expected to have low lipoarabinomannan concentrations, such as those with a CD4 count higher than 200 cells per μL.Depending on the handling of urine before and after freezing, differences in protein content and pH can be dramatic—urine alkalises over time,2 and protein content can drop after a freeze–thaw.3 These factors could affect performance of lateral flow tests by eliminating some of the variability between urine samples, leading the assay developer to produce a less robust test, unable to handle the spectrum of fresh urine. This could yield a high rate of false positivity.On the basis of our results and the crucial changes in urine with freezing, we conclude that the difference between fresh and frozen urine samples is a non-trivial matter.