The study cohort was composed of healthy adults older than 18 years whose deidentified data were abstracted from client records for research analysis. Data were collected from individuals who had been screened for contraindications and exclusion criteria as previously reported [13]. Briefly, participants were excluded if they reported a diagnosed chronic disease, had untreated high or low blood pressure, or were pregnant or breastfeeding. All participants completed a consent that allowed data to be used for research purposes.

Data were examined retrospectively and determined to be exempt by the Norwich University institutional review board and research ethics committee and informed consent was waived (HHS IORG #0004914; institutional review board no. 00005859; date of registration: April 30, 2023).

Participants were followed for an average of 8.55 months (SD 3.13; median 6.77, IQR 6.77-9.73), with total follow-up duration ranging from 5 to 13 months for those who had at least 1 follow-up visit. During this time, they frequently participated in recreational W-BC at a commercially operated studio (Restore Hyper Wellness). The mean amount of W-BC received was 6.78 (SD 4.26) times per month with the cumulative total ranging from 13 to 157 W-BC sessions over the course of the study. On average, participants completed 1-2 sessions per week throughout the intervention. Participants were encouraged to engage in regular physical activity and received guidance on healthy dietary patterns, although these behaviors were not monitored.

Participants were escorted to a private changing room where they removed personal clothing and put on a provided hat, socks, slippers, mask, and gloves, as well as a towel or robe. Once changed, they were escorted to the cryotherapy chamber for 3 minutes of cold exposure at –110 ℃. Chambers were electrically cooled using a main cooling unit in a full body chamber (Zimno Tech). Participants were observed by an employee trained in administering the protocol and could exit the chamber at any time. At the completion of 3 minutes, the chamber door was opened and the participants exited, returned to the private changing room, and redressed in their own clothing.

Blood was drawn at baseline and then incrementally at follow-up visits. Participants were asked to fast for 8-12 hours prior to blood draw. On average, there were 2.13 (SD 2.59) follow-up visits with a mean time of 170 (SD 84) days between blood draws. The minimum time between blood draws was 17 days, and the maximum was 292 days. Blood biomarkers measured include serum high-sensitivity C-reactive protein (hsCRP), fasting glucose, hemoglobin A_1c (HbA_1c), low-density lipoprotein (LDL) and high-density lipoprotein (HDL), and triglycerides. All blood was drawn from the median cubital vein with a 20-gauge intravenous catheter needle. Drawn samples were sent to Quest Diagnostics for analysis. hsCRP was analyzed with an immunoturbidimetric assay (test code 10124); fasting glucose, serum lipids, and triglycerides were analyzed with spectrophotometry (test codes 6881 and 7600, respectively); and HbA_1c was analyzed with an enzymatic assay (test code 496). For hsCRP, the reporting range by Quest Diagnostics was 0.3-10 mg/L, with values outside the range being listed as <0.3 or >10 mg/L. There were no identified thresholds for fasting glucose, HbA_1c, or lipids.

All data analysis and statistical modeling were carried out in R (version 4.3.1) [14]. Data processing was performed using tidyverse packages including readr and readxl for file opening; dplyr, hms, lubridate, tibble, and stringr for wrangling; and ggplot2 for visualizations [15]. Two data points were missing a parsable date and those records were removed. Out of range hsCRP measurements were included as their threshold value, that is, <0.3 was included as 0.3.

Statistical significance was defined as a P<.05 threshold while anything failing to meet that threshold but with P<.10 was noted as being mild association and of possible interest. Paired t tests on all complete pairs for a given analyte were used to compare start and end measurements for all biomarkers in Table 1.

For both the primary analysis (W-BC impact on hsCRP) and the secondary analysis (W-BC impact on biomarkers of metabolism) multiple linear regression models were constructed. An initial model was fit with the analyte of interest standardized by the participant-specific mean as the response, and the predictors were age, sex, days from study start, and number of W-BC sessions in the past 30 days (1 month). After inspection of model diagnostics for violations of the linear modeling assumptions, outliers were removed based on a Cook’s distance threshold of 4/n where n is the sample size. A refined model was then fit without outliers and inspected again for model diagnostics. Finally, the reduction method F test was used to determine whether a reduced model that removed age and sex was preferred. This was done because age and sex were not found to be significant in any of the initial models. Based on the F test results, the final model was chosen (Figure 1).

Fifteen participants (female: n=9) were included in the analysis (Table 1). Ages ranged from 32 to 59 years. Average weight at baseline was 84.97 (SD 14.61) kg. Ethnicity or race and height were not recorded. Two were lost to follow-up and an intention-to-treat statistical model was used.

The number of W-BC sessions in the month prior to blood measurement was associated with a statistically significant decrease in hsCRP (P<.01). The final linear model for hsCRP was fit on 29 observations from 15 participants. Neither age nor sex was associated with changes in hsCRP and was not included in the final model (Figure 2A). The effect size was –0.1413 mg/L in hsCRP per W-BC session (95% CI –0.2289 to –0.0537).

To account for possible time-related confounding, time was adjusted for by including days since baseline launch day. Time was found to be a significant predictor of hsCRP although with a very small effect size (P<.01; effect size=–0.0047 mg/L, 95% CI –0.0066 to –0.0028). Importantly, even with time in the model, the number of W-BC sessions was still strongly significant. In addition, W-BC sessions had a larger effect size being –0.1413 mg/L as compared with –0.0047 mg/L for time. If scaled to 1-month windows, the effect of W-BC was still larger than time. Namely, multiplying days by 30 and W-BC sessions by the monthly average of 6.78 gives monthly effects sizes of –0.9579 mg/L and –0.1409 mg/L for W-BC and time, respectively. Thus W-BC sessions have a statistically significant impact on reducing hsCRP while also having a larger effect as compared with time (Table 2).

To elucidate the durability of the observed effect, sequential time windows were tested (Figure 3A). Windows less than 3 weeks were not associated (P>.10) with hsCRP. At 3 weeks, a mild association appeared (P>.10), and at 1 month the association was significant even after adjusting for the multiple testing windows (P<.05). For windows of 2 months or longer, the association remained, though only mildly (P>.10). Because participants were tracked only for around 9 months on average, longer windows were not constructed.

The number of W-BC sessions in the prior 1 month failed to reach the threshold for statistical significance (P=.07). The final linear model was fit on 26 observations of fasting glucose from 14 participants. Neither age nor sex was associated with fasting glucose and was not included. The estimated effect size was –0.7337 mg/dL in fasting glucose per W-BC session (95% CI –1.5236 to 0.0562) meaning more W-BC was mildly associated with a small decrease in fasting glucose (Figure 2B). Time indicated a possible small increase in fasting glucose over time although failed to reach significance (P=.06; effect size=0.0114, 95% CI –0.0008 to 0.0237). HbA_1c did increase slightly after 9 months (P<.01); however, the change was quite minimal and within normal ranges (difference=0.13% and <5.7% at both time points). Only hsCRP was reduced significantly (P=.03) in before versus after measurements.

Durability results revealed that slightly longer time windows may be more appropriate for studying fasting glucose. After correction for multiple testing using the false discovery rate method, the shorter 1-month window was not statistically significant (adjusted P>.05). However, windows of 2 and 3 months (60 and 90 days) showed a significant association between W-BC and decreased fasting glucose (adjusted P<.05; Figure 3B).

For the lipid biomarkers, namely, LDL cholesterol, HDL cholesterol, and triglycerides, no association with W-BC was found (P>.10), although LDL trended lower over the time period examined (P=.07).

This study investigated long-term physiologic effects associated with inflammation and aging in healthy adults by examining hsCRP and biomarkers of metabolism as a function of W-BC. Overall, W-BC showed beneficial impacts for inflammation and fasting glucose but not serum lipids. The effect with hsCRP appears time dependent and indicated that the impact of W-BC on inflammatory markers may take up to 3 weeks to manifest, with the effect significant and durable for up to 9 months with continued use.

The results of this study agree with, and extend, prior research on W-BC related to inflammation. Studies in patients with arthritis showed that W-BC helped reduce inflammatory markers such as IL-6, tumor necrosis factor, and serum CRP [16-18]. Similar anti-inflammatory benefits from W-BC have been observed in studies of men with overweight and obesity, as well as postmenopausal women with type 2 diabetes [10-12]. While prior studies provided a groundwork for the acute anti-inflammatory effect of W-BC in populations who already suffer from chronic inflammation, this study sought to extend the generalizability of such an anti-inflammatory effect to healthy individuals without specific disease conditions.

To the best of our knowledge, there is only 1 other study that specifically examined healthy individuals and long-term durability of the anti-inflammatory effect with W-BC. Lubkowska et al [19] measured cytokine levels in healthy men for up to 6 weeks after W-BC and found a beneficial association between W-BC and serum cytokines. Their results, along with the current findings, demonstrate that W-BC can decrease inflammation in healthy adults, both in the short term (ie, 6 weeks) and the long term (ie, up to 9 months).

The current findings also add a durable clinical marker of systemic inflammation associated with future disease risk (hsCRP) to the cytokines previously studied. This is key because cytokines alone may be less accurate biomarkers than CRP for disease risk due to issues with stability, half-life, and serum variability. For example, IL-6, one of the main cytokines analyzed by Lubkowska et al [19, shows distinct diurnal variation and therefore time of day can be a confounder [20,21]. In contrast, CRP shows no diurnal variations lending it stability and reproducibility for comparison [22]. CRP may also be a better overall marker of disease risk by not only showing acute inflammation but also having persistence and stability in plasma. Research in cardiovascular disease indicates that hsCRP may be a better marker for future diagnoses of hypertension, myocardial infarction, and coronary heart disease [23,24]. Together, the current findings and prior outcomes elucidate potential mechanistic and clinical implications of short- and long-term W-BC.

Fasting glucose was also beneficially impacted by W-BC in this study, though the exact relationship is less clear. At 1 month, the association between W-BC and fasting glucose trended downward (P<.10). Durability analysis then revealed significant associations between W-BC and decreased fasting glucose for 2- and 3-month windows (adjusted P<.05). After 3 months, no linear relationship was found. Thus, the effect of W-BC on fasting glucose may be more time-constrained such that some benefits are seen for intermediate time windows but may not be durable chronic adaptations. This time-dependent interpretation is further supported in light of the lack of change seen in fasting glucose from the start to the end of the 9-month study. Interestingly, a very small increase in HbA_1c was noted at the final follow-up period (P<.01). However, the change was quite minimal (difference=0.13%) and within normal ranges at both time points (5.12 [0.27%] and 5.25 [0.29%], respectively) and not considered clinically significant.

From a clinical perspective, the current results extend inflammatory effects of W-BC to known markers of metabolic dysregulation. Specifically, decreases in blood glucose are known to modulate inflammation [25]. Thus, the beneficial impact on inflammation seen from W-BC may be achieved at least in part due to reductions in blood glucose. Indeed, the study by Wiecek et al [12] noted decreases in both blood glucose and CRP. While the previous study specifically examined women with type 2 diabetes and therefore elevated glycemia, the participants in this study were healthy, nondiabetic individuals. Thus, these results further extend the beneficial impact of W-BC on fasting glucose to those without an existing hyperglycemic diagnosis. Further targeted research would be needed to establish a definitive connection between the glucose modulating effects of W-BC and its anti-inflammatory benefits.

No significant effect from W-BC was found for the lipid markers of metabolism in either the comparison of study start to end or the linear modeling for specific time windows. This result slightly contradicts prior research finding a benefit of W-BC for lipids. In particular, Lubkowska et al [26] found that W-BC combined with exercise as a program to treat overweight resulted in improved lipid metabolism including decreased LDL and triglycerides and increased HDL. However, this study included exercise in a protocol targeted at treating overweight, thus applicability for W-BC specifically is less clear. A more recent study investigating lipid metabolism in healthy adults demonstrated that daily W-BC was associated with decreased LDL and triglycerides and increased HDL [27]. However, this study was analyzing acute effects immediately after daily application of W-BC while the present study, which found no significant effect on lipid metabolism, is focused on longer-term impacts. Thus, it is plausible that W-BC could impact lipid metabolism in the short term while not having a lasting durable benefit. It is also possible that a true effect exists, but this study was simply underpowered to discern it.

One major limitation of this study is sample size. Both the number of participants and the number of follow-up visits were small. To correct for any time-based confounding, time has been included in the models; however, other possible sources of confounding could exist. In addition, there was no control group. In general, control groups for W-BC are difficult since a participant cannot be blinded to whether they receive cryotherapy. In addition, this study was conducted in a free-living real-world environment rather than a controlled clinical setting. Such a design lends both benefits and drawbacks. The main drawback of this free-living environment is that external factors could lead to possible confounding. For example, weight loss in individuals with obesity has been linked to a decrease in inflammatory markers including CRP [28,29]. As weight loss information was not available, time was included as a proxy for any weight loss or other baseline improvements in health that participants may have experienced. However, time alone likely does not capture all confounding effects and thus the impact of external uncontrolled factors cannot be ignored.

In spite of the limitations, a free-living study also has unique benefits. In particular, such a design enables immediate applicability of results to real-world environments. This study did not involve a clinical protocol necessitating strict compliance, yet the results still demonstrated a significant benefit from recreationally participating in cryotherapy. Thus, the takeaways from these results may be applied more generally to the public in that individuals can simply incorporate cryotherapy into their daily lives.

Future directions of study could include analysis of dosage for optimal effect. For example, if the benefit of increased W-BC sessions plateaus at some point, then individuals can limit unnecessary time, expense, and exposure by participating in only an optimal number of W-BC sessions. Dosage could also be dependent on the durability of the effect. Given the rapidly rising popularity of cryotherapy, high-quality randomized controlled trials are needed to investigate these more specific outcomes.

This study explores the impact of W-BC on healthy participants outside of any specific exercise protocol or disease state. In doing so, W-BC has been demonstrated to be potentially beneficial to the general population as an anti-inflammatory treatment. Furthermore, the anti-inflammatory effects show lasting durability for impact on future chronic inflammation. Finally, this study demonstrated that W-BC may have a beneficial, though time-constrained, impact on glucose.