In a published letter commenting on research by Zhao and colleagues evaluating the MX3 hydration testing system in resting and post-exercise adults, researchers argue that significant methodological constraints prevent a meaningful evaluation of the portable device’s diagnostic accuracy.
The correspondence focuses on the recent investigation titled Assessment of the MX3 hydration testing system in adults at rest and following exercise in the heat, noting that while assessing portable hydration monitoring tools is necessary, key study design choices compromised the validity of the reported conclusions.
A primary concern raised in the letter involves the selection of urine specific gravity as a reference standard to verify dehydration and euhydration status for comparison against salivary osmolarity measurements made by the MX3 device. Rather than employing a recognized criterion reference such as plasma osmolality, the study relied on single spot urine samples obtained from participants upon their initial arrival.
According to the commentary, random spot urine collections are known to be vulnerable to recent fluid intake and fail to accurately reflect shifts in hydration status during active dehydration. Because the authors compared salivary metrics against a non-criterion measure without acknowledging its established shortcomings or quantifying independent measurement errors, the resulting claims regarding device validity remain severely limited.
The letter also highlights methodological issues surrounding saliva sample collection, pointing out that reported salivary osmolarity values in the study included measurements below 25 milliosmoles per kilogram. Because typical resting saliva in euhydrated individuals ranges between 40 and 70 milliosmoles per kilogram, values beneath this threshold are considered non-physiologically relevant.
These abnormally low values, the letter suggests, likely stemmed from water contamination or air bubbles introduced during sampling, such as tapping test strips against a dry tongue. Although the original study indicated test strips were applied according to manufacturer recommendations, it omitted critical procedural controls, such as requiring participants to swallow residual mouth fluids and presenting a fresh, bubble-free sample of sufficient volume.
The presence of measurements outside the normal physiological range contributed to elevated variance, including a high standard deviation recorded during post-exercise testing in the study’s second trial. In contrast, prior work by Winter and colleagues that adhered to strict collection protocols observed neither non-physiological values nor excessive variance while demonstrating repeatability with the MX3 device.
Regarding test-retest reliability, the letter criticizes the study’s comparison of MX3 measurements taken across two trials spaced forty-eight hours apart. Determining instrument reliability requires the underlying biological variable to remain stable, yet the trial lacked objective markers, such as baseline serum osmolality, urine osmolality, or standardized body mass, to confirm consistent hydration across the two-day interval.
While the original authors noted that forty-eight-hour biological changes in salivary osmolarity likely drove the observed variability, they provided no dietary or fluid intake data to support that explanation. The letter contends that this between-trial variation reflects unmanaged biological fluctuation rather than device error, noting that Winter and colleagues demonstrated an intraclass correlation coefficient of 0.75 when testing samples taken three to five minutes apart.
Finally, the commentary argues that the study’s conclusions regarding the device’s poor ability to detect hypohydration were severely underpowered. Across forty total trials, only six met the hypohydration threshold of urine specific gravity exceeding 1.020, and only five exceeded a two percent body mass loss, resulting in an inadequate sample size to evaluate classification accuracy.
Beyond questioning the specific reference standard employed by Zhao and colleagues, the commentary emphasizes the broader methodological principles governing validation of novel diagnostic technologies. To make valid scientific inferences regarding a newly introduced hydration monitoring device, investigators must not rely merely on agreement with another non-criterion measure. Comprehensive evaluation protocols should incorporate a recognized criterion standard, such as plasma osmolality, while independently quantifying measurement error inherent to each testing technique. Furthermore, statistical evaluations must utilize analytical frameworks that explicitly account for potential error across both variables simultaneously. In the absence of these essential components, assertions regarding the validity or reliability of a novel testing system remain unverified and should be framed with caution.
The critique also expands upon the physiological limitations of spot urine samples by referencing established findings within the published literature. Multiple prior investigations, including research by Cheuvront and colleagues in 2015, Hew-Butler and colleagues in 2018, and Sommerfield and colleagues in 2016, have demonstrated that random spot urine collections are exceptionally sensitive to recent fluid ingestion. Consequently, spot samples do not reflect ongoing hydration adjustments during periods of active dehydration. The commentary notes that Zhao and colleagues did not acknowledge these recognized shortcomings, despite earlier literature by McDermott and colleagues in 2017 highlighting the vulnerability of spot urine specific gravity checks. Relying on an unvalidated spot check as the reference benchmark inevitably undermines any comparison with salivary metrics.
A substantial portion of the letter scrutinizes the specific data published in Table 2 of the Zhao report, where recorded salivary osmolarity ranges display profound physiological discrepancies. The letter points out that Table 2 contains a significant number of values measuring below 25 milliosmoles per kilogram. In established physiological literature, including investigations by Munoz and colleagues in 2013 as well as Winter and colleagues in 2024, normal resting saliva in euhydrated adults consistently spans between 40 and 70 milliosmoles per kilogram. Because readings under 25 milliosmoles per kilogram fall outside documented biological norms, the commentary argues that these entries represent non-physiologically relevant artifacts rather than authentic salivary osmolarity states.
In explaining how such distorted measurements could arise, the authors of the letter examine the physical mechanisms involved in saliva sample acquisition. Non-physiologically relevant values can easily be generated when test strips encounter water contamination or air bubbles. These bubbles may either be preexisting on the participant’s tongue or introduced when attempting to collect a sample from an excessively dry oral surface. Although Zhao and colleagues mentioned that diagnostic test strips were tapped onto the tongue in accordance with manufacturer guidelines, their manuscript provides no detailed description regarding the specific physical state of the sampled saliva, leaving critical sampling variables entirely unaccounted for.
The letter stresses that manufacturer instructions require specific procedural steps that were neither verified nor documented by the researchers. While participants were instructed to refrain from eating or drinking prior to testing, the published study does not confirm whether essential saliva collection protocols were carried out. As detailed by Winter and colleagues in 2024, participants must actively swallow all accumulated oral fluids and present a fresh saliva specimen for evaluation. Furthermore, technicians must ensure that an adequate volume of saliva is present on the tongue and that the sample remains free of air bubbles. Failing to adhere to these straightforward manufacturer guidelines compromised data collection.
These sampling omissions had direct statistical ramifications on the study’s reported dispersion. The incorporation of measurements falling well outside the normal physiological range introduced considerable instability into the salivary osmolarity data, directly causing the elevated standard deviation observed during post-exercise testing in Trial 2. In stark contrast, when Winter and colleagues rigorously implemented appropriate sample collection procedures in their 2024 study, they observed neither non-physiological readings nor inflated data dispersion. That investigation documented robust reliability and repeatability for salivary osmolarity measurements using the MX3 device, demonstrating that the wide scatter in the Zhao data was a product of methodological lapses rather than hardware inaccuracy.
The commentary also levels sharp criticism at how test-retest reliability was operationalized across sessions separated by 48 hours. Citing foundational measurement principles articulated by Hopkins in 2000, the authors note that evaluating device reliability requires the underlying biological variable to remain stable across measurement intervals. Only under conditions of physiological stability can observed variations be genuinely attributed to instrument error rather than real biological fluctuations. Although Zhao and colleagues stated that participant dietary and fluid intake were logged and matched across the two-day period, the researchers presented no objective verification data, such as pre-trial serum osmolality, urine osmolality, or standardized body mass assessments, to confirm that subjects began each trial in identical hydration states.
This methodological deficiency introduced a confounding factor where between-trial variations reflected an unseparated mixture of real physiological changes and measurement error. Zhao and colleagues conceded that biological changes in salivary osmolarity over 48 hours were expected and likely contributed to the large variability observed. However, the commentary emphasizes that this concession was supported neither by empirical data quantifying dietary or fluid intake nor by independent measurements confirming hydration stability or change. Consequently, the letter argues that the recorded variability represents normal day-to-day biological fluctuation occurring under loosely managed conditions, rather than reflecting the intrinsic reliability of the MX3 hydration testing system.
To demonstrate how test-retest reliability should be properly evaluated without biological confounding, the letter cites the protocol used by Winter and colleagues in a cohort of 75 adults. In that study, researchers tested two independently gathered saliva samples collected only 3 to 5 minutes apart, thereby ensuring that participant hydration status remained biologically stable. That controlled design established moderate-to-good reliability for the MX3 testing system, yielding an intraclass correlation coefficient of 0.75. Additionally, the analysis established a standard error of measurement of 9.2 milliosmoles and demonstrated a minimal detectable change of 21.3 milliosmoles at the 90 percent confidence level, providing precise statistical parameters that were missing in Zhao’s protocol.
The commentary further identifies fundamental statistical weaknesses in the hypohydration analysis conducted by Zhao and colleagues. The original investigators concluded that the MX3 system displayed poor capability in detecting hypohydration, but this conclusion was derived from an extremely restricted data pool. Out of 40 completed trials, only six met the diagnostic threshold for hypohydration via urine specific gravity exceeding 1.020, and only five trials exceeded the threshold of two percent body mass loss. Although the study authors acknowledged low trial counts as a limitation, the letter stresses that such low numbers severely impaired statistical power, precluding any meaningful evaluation of the system’s true classification accuracy.
Subject of Research: Methodological evaluation of the MX3 salivary hydration testing system
Article Title: Methodological limitations in the assessment of the MX3 hydration testing system in relation to findings by Zhao et al.
Article References: Chana, G., Parrish, K., Luther, M., Jardine, W., Erlichster, M., & Kilpatrick, T. (2026). Methodological limitations in the assessment of the MX3 hydration testing system in relation to findings by Zhao et al.. Physiological Reports, 14(22), Article e71216. https://doi.org/10.14814/phy2.71216
Image Credits: AI Generated
DOI: 10.14814/phy2.71216
Keywords: MX3 hydration testing system, salivary osmolarity, hydration assessment, urine specific gravity, plasma osmolality, dehydration, euhydration, hypohydration, test-retest reliability, biological variability, saliva sampling, body mass loss, measurement error, intraclass correlation coefficient, standard error of measurement, minimal detectable change, exercise in the heat
Cite Scienmag News
Drew Townsend. (November 20, 2026). Letter Identifies Methodological Limitations in Study Evaluating MX3 Hydration Testing System. Scienmag. https://scienmag.com/letter-identifies-methodological-limitations-in-study-evaluating-mx3-hydration-testing-system/
Drew Townsend. "Letter Identifies Methodological Limitations in Study Evaluating MX3 Hydration Testing System." Scienmag, 20 November 2026, https://scienmag.com/letter-identifies-methodological-limitations-in-study-evaluating-mx3-hydration-testing-system/. Accessed 20 November 2026.
Drew Townsend. "Letter Identifies Methodological Limitations in Study Evaluating MX3 Hydration Testing System." Scienmag. November 20, 2026. https://scienmag.com/letter-identifies-methodological-limitations-in-study-evaluating-mx3-hydration-testing-system/








