A 2020 case series by Stjernbrandt and colleagues in the International Journal of Circumpolar Health examined 12 Swedish participants selected for cold-sensitive hands. Median superficial dorsal-hand laser-speckle perfusion was 52.5 perfusion units before a two-minute hand immersion in 12 °C water and 51.3 immediately afterward, with no group-level before–after difference. At the participant level, however, six values rose and six fell [2].

Slow index-finger temperature recovery, or recovery whose timing could not be confirmed before observation ended, occurred across different baseline spatial patterns and in both perfusion directions. Yet the study did not test, participant by participant, how symptom severity, detection thresholds, temperature recovery, spatial pattern, and perfusion shift were related [2]. The nearly flat median therefore does not imply uniform individual responses, but the six-up–six-down split does not establish physiological subtypes either. These findings belong to a selected 12-person immersion series; they do not describe natural episodes or the frequency of either direction in a wider population.

A 2025 BMC Women's Health experiment by Yamazaki and colleagues divided 20 healthy Japanese women aged 20–22 using a 10-item cold-sensitivity questionnaire that included symptom items. Measured hand skin temperature was similar between the questionnaire-defined groups during the nominal 31 °C precooling period and the subsequent nine-minute descent to 23 °C, yet the cold-sensitive group gave colder ratings. Temperature and sensation still changed together within each cooling condition, so the result is not that temperature was irrelevant; reported cold differed at comparable measured temperatures. The experiment ended after cooling and did not observe rewarming [5].

Across three independent experiments reported by Ezquerra Romano and colleagues in the Proceedings of the Royal Society B in 2025, each with 12 healthy volunteers, concurrent punctate touch reduced sensitivity to a brief, focal, thermal-camera-monitored fall in hand temperature. Response bias was analyzed separately, but the proposed neural account was not directly confirmed. The outcome was binary detection in a tactile context, not a rating of cold intensity or discomfort, so those sensory tasks cannot be substituted for one another [6].

In a 2023 Journal of Applied Physiology experiment by Moes and colleagues, 14 healthy young men underwent repeated whole-body cold-water immersions. Under nitrous oxide, they reported less coldness and discomfort even as core cooling progressed further, while the decline in an eight-site mean skin-temperature composite did not differ between conditions. The study was almost completely unblinded and compared pharmacologically manipulated, provoked whole-body conditions rather than naturally occurring hand or foot episodes. Detection, cold intensity, discomfort, local temperature, composite skin temperature, and core cooling must therefore remain named as separate tasks and measures; this experiment does not establish a cause or treatment effect for recurrent distal cold [7].

In 2023, Eglin and colleagues gradually cooled a footplate for a selected group with chronic non-freezing cold injury and controls. The injury group rated their feet colder at the 33 °C baseline and at 20 °C and 15 °C, and reported greater discomfort at several steps, although great-toe temperature and local laser-Doppler cutaneous vascular conductance were similar between groups [1]. Three participants in the injury group stopped before 15 °C, reducing the number available for the final comparison. This mismatch is a group result from one ordered laboratory protocol; it does not show that either the reported sensation or the local measurements were wrong, and it does not identify a cause.

In a separate test from the same study, the left foot was immersed in 15 °C water for two minutes and then allowed to recover for ten minutes in approximately 30 °C air. The injury group rated their feet colder and more uncomfortable than two control groups, and their coldest-toe temperatures were lower at minutes five and ten [1]. Great-toe temperature did not differ between groups, however, the mean-toe result did not cross the conventional statistical significance threshold, and great-toe cutaneous vascular conductance was also similar. That conductance was a site-specific quantity derived by dividing great-toe laser-Doppler flux by mean arterial pressure, so it cannot be extended to whole-foot perfusion or a general state of blood flow.

By contrast, Norrbrand and colleagues tested the right hand and right foot separately in 70 healthy Swedish military cadets in 2017. After prewarming, each site underwent 30 minutes of immersion in 8 °C water followed by 15 minutes of spontaneous rewarming; at minutes five, ten, and fifteen of recovery, both toe temperature and local thermal-sensation ratings were lower for the foot than for the hand [4]. Felt cold and skin temperature aligned at those selected times. Direct blood flow was not measured, however, and the hand and foot underwent separate provoked tests, so this agreement cannot be treated as the default relation during naturally occurring episodes.

In 2020, Maley and colleagues compared a thermistor taped to the right index-finger pad with an infrared spot about 1 cm farther distally in 52 young men. After five minutes of prewarming at 35 °C, the bagged hand was immersed for 30 minutes in approximately 8 °C stirred water, and both setups followed ten minutes of rewarming. Averaged across all recorded times, the infrared setup read 1.80 °C warmer, but the mean gap narrowed from 2.55 °C at minute one to 1.15 °C at minute ten [8]. The reported limits of agreement extended toward zero and into reversed differences. Under the authors’ prespecified criteria, the two setups were not interchangeable measurements separated by a fixed offset in this protocol.

This comparison cannot determine which device was closer to the true local temperature. The camera had been calibrated recently, but there was no independent same-location standard; the thermistor sampled skin covered by tape, whereas the infrared device sampled a single uncovered pixel at another site. Havenith and Lloyd’s counterpoint that year left calibration, spatial gradients during distal-to-proximal rewarming, region selection, tape and contact microclimate, insulation, and response time as competing explanations [8,9]. Because it added no new participants or replication data, the defensible conclusion is a time-varying setup difference with unresolved error allocation—not that either device universally reads high or low.

Even with the same device, anatomy and sampling schedule changed the course that appeared. In Norrbrand and colleagues’ separate hand and foot tests, fingers began warmer than toes, the hand sites cooled faster initially, toes were colder late in the 30-minute immersion, and fingers remained much warmer throughout the 15-minute recovery [4]. Temperature-defined vasodilation waves were also more common in fingers, but they were not direct perfusion measurements. Thermocouples sampled every second, whereas infrared recovery imaging began at minute one after removal and possible towel drying, with the delay to the first image unquantified; initial slope, late-cooling temperature, and post-removal recovery therefore cannot substitute for one another, and hand values cannot be transferred to the foot.

In Stjernbrandt and colleagues’ 12-person cold-sensitive-hand series, recovery was defined as index-finger pulp temperature returning to within 2.0 °C of its starting value; 30-second acquisitions were made every three minutes, and observation stopped at 29 minutes. Four observations had not crossed the threshold by then, leaving their recovery times unconfirmed rather than establishing either 29-minute recovery or failure to recover [2]. Weller and colleagues’ 2026 meta-analysis of temperature-defined cold-induced vasodilation also found substantial differences in immersed part, digit, water and ambient temperature, prewarming, and the time window used for mean temperature, with very high heterogeneity; toes were excluded, and post-immersion recovery, sensation, and direct-flow endpoints were not pooled [10]. Temperature and recovery values can therefore be compared only when their device, exact site, sampling schedule, time origin, and endpoint are stated together.

KDCA frostbite information updated on 6 May 2026 ties its account to cold exposure and depth of tissue involvement, describing early cold, pain, burning, pallor, or altered sensation and deeper courses that may include blistering and tissue damage [11]. When a compatible exposure is accompanied by local tissue findings, the immediate question is acute injury rather than whether felt cold, skin-surface temperature, local perfusion, and rewarming agree. Its hospital and conditional field-rewarming instructions concern suspected cold injury and delayed access to care; they are not a self-test or routine warming plan for recurrent cold hands or feet.

The NHS chilblains page, reviewed on 22 April 2026, concerns small itchy patches that appear after cold exposure, often on fingers or toes, with possible burning and red or purple colour change [12]. Once such a lesion appears, the relevant question becomes a post-cold skin change rather than the measurement of isolated cold sensation. A low skin-temperature reading or slow rewarming does not establish chilblains, and this UK care pathway is not a universal urgency rule.

KDCA diabetic-foot information updated on 22 April 2026 links diabetes plus a foot wound, colour change, sensory loss, or pain to prompt clinical assessment and notes that an injury may be painless [13]. Takahara’s 2025 review likewise treats chronic lower-limb rest pain, nonhealing tissue loss, and gangrene as a separate context of threatened lower-limb tissue [14]. Together, these sources show how diabetes with foot findings or a changing lower-limb course alters the frame of the question. They do not extend a foot- or lower-limb-specific pathway to hands or isolated cold sensation, and pressure ratios such as ABI or TBI do not measure felt cold or microvascular rewarming.

In a selected thyroid-clinic study published in 2024, every response level to a combined cold-hands-and-feet question appeared in both TSH-defined groups [15]. The finding leaves systemic context as a separate question when disease history or examination findings point that way, but the adjective cold alone is neither a diagnosis nor a safety pathway. More broadly, a specific exposure, new skin or tissue finding, altered course, examination finding, or relevant disease context changes what must be asked. These are bounded illustrations from accessible, sufficiently current sources—not an exhaustive warning list or reassurance rule—and an unlisted or inaccessible context cannot be assumed safe [3].

Across 13 prespecified PubMed searches, 238 records were screened from 2026-08-16T16:06:11Z to 20:19:50Z, and the relevant passages in sources used for the synthesis were checked by 2026-08-17T03:40:34Z [3]. Within that route and date range, we did not identify a qualifying hand dataset that synchronized participant-level felt cold, site-matched skin-surface temperature, an explicitly named local flow or perfusion measure, and the complete rewarming course during naturally occurring recurrent episodes. This does not erase studies that measured only some of those observations.

The same bounded search separately identified no qualifying foot dataset that synchronized the four observations during naturally occurring recurrent episodes [3]. A useful hand dataset need not include feet, just as a useful foot dataset need not include hands. The two site-specific results must therefore remain separate rather than being combined or transferred from one anatomical site to the other.

This result does not prove that qualifying data are absent from unreviewed search results, other databases, inaccessible material, or later publications [3]. Provoked experiments, disease-specific cohorts, recalled questionnaires, historical expressions, and public-language material answer questions with different populations, sites, instruments, and clocks; they cannot be assembled into the missing natural episode [3]. Nor does this search estimate how often the observations agree or diverge, their cause or prevalence, whether recurrent distal cold is benign, or whether a vascular or sensory explanation applies [3].

Within the bounded material examined, no qualifying hand record synchronized participant-level felt cold, site-matched skin-surface temperature, an explicitly named local flow or perfusion measure, and the complete rewarming course during naturally occurring recurrent episodes; the same bounded result arose separately for feet [3]. The next observation would therefore be a prospective, repeated, within-participant study of adults who report recurrent cold sensation localized to one or both hands, one or both feet, or both site families. Hands and feet would retain separate site procedures and analyses.

During each episode, momentary wording and timing of felt cold would remain distinct from tactile reports and recalled symptoms. These reports would be aligned by actual timestamps and acquisition windows with site-matched skin-surface temperature and one explicitly named local flow or perfusion compartment whose device, site, calibration, and acquisition conditions are specified [3]. Serial samples would cover the rewarming course without relabelling the first observation as a pre-episode baseline; the record would preserve noticed onset or the first feasible observation and its delay, measure-specific nadirs and recovery origins, incomplete recovery, missing synchronization, and censoring. The design does not nominate a universal device, site, interval, recovery origin, endpoint, or threshold [3].

The repeated observations would preserve each participant’s trajectory rather than replace it with a group average. Even when the same participant contributes both hand and foot episodes, results would not be pooled by default or transferred from one site family to the other. Optional provoked episodes would also remain separately labelled and analysed from naturally occurring episodes.

This observation is not yet a validated instrument. Feasibility and synchronization, technical repeatability and device–site agreement, evaluation of Korean and English sensory wording, prespecified construct-validity testing, and external replication would each have to succeed before it could be called one [3]. Until then, the question is not diagnosis or utility, but when the four records align, when they diverge, and where synchronization or confirmation of recovery fails from noticed onset to the observation limit at a defined hand or foot site in one participant. If that record cannot be captured completely and repeatably, the observation method requires redesign rather than a stronger conclusion.