Circadian Biomarkers Before and After the Transition to Spring Daylight Saving Time Authors: Mark S. Rea, John D. Bullough, Mariana G. Figueiro Published: 2026 ABSTRACT ObjectivePeriodically, concerns are raised within the research community about the effects on circadian and sleep-related outcomes resulting from changes to and from daylight saving time (DST). To better understand these concerns, we collected varied circadian and sleep-related outcomes from healthy individuals immediately before and after the spring DST transition.Materials and Methods Subjects wore calibrated, personal light sensors that measured light exposure (amount, duration, time) and wrist actigraphs to quantify the synchrony between the 24-hour light-dark exposure and the 24-hour activity-rest cycles. Sleep onset, offset, and time in bed were obtained from the wrist actigraphs. Dim light melatonin onset (DLMO) was also measured before and after DST transition. Predicted minimum core body temperature (CBT min p ) times were calculated from the light exposure data. Karolinska Sleepiness Score (KSS) ratings were collected four times a day. Health-related outcomes were not examined. Results The spring DST transition had modest (statistically insignificant) effects on the weekly averages of time in bed (≤ 21 min), subjective sleepiness (KSS < 0.48), and the synchrony between the light and activity cycles (phasor magnitude ≤ 0.08 units). After the transition to DST, CBT min p was slightly delayed, whereas DLMO and mid-sleep were slightly advanced. ConclusionFor this cohort, the spring transition to DST had little if any effect on a variety of circadian and sleep-related outcomes. The transition appears to have caused a small disassociation in the relative timing of the circadian phase markers, which may or may not be of some significance for health and wellbeing. Introduction There has been a long-standing debate about the health implications related to the transitions between standard time (ST) and daylight saving time (DST). 1 2 3 4 5 Recent authors have repeated concerns that the transitions to and from DST may disrupt the circadian system, potentially disrupting sleep and inviting a variety of health-related maladies. 6 7 8 9 Martín-Olalla and Mira are not, however, convinced that the literature is clear about a direct connection between negative health effects and the transitions to and from DST. 10 They state that there is no strong, convincing link between negative health outcomes and DST transitions, but rather, that these changes in clock times are “[…] just one of the many complexities […] in modern extratropical societies.” Neumann and von Blanckenburg, who recently reviewed the literature, summarize by saying “From an overall perspective, it is evident that current research on the effects of clock changes and DST does not generally provide consistent results,” 7 a point that is reinforced by the conclusion of another recent review that evidence relating DST to human health is “limited and heterogeneous.” 11 Yet, at the outset of their overview, Neumann and von Blanckenburg say that DST transitions have “[…] clearly negative effects on health […] which are driven by the disturbed circadian rhythm and the resulting sleep problems.” 7 So, the literature suggests, perhaps inconclusively, that transitions to and from DST may compromise circadian entrainment and, consequently, sleep quality and health. Circadian entrainment can be defined as a continuous variable characterizing the strength of synchronization between the 24-hour light-dark cycle (input) and one or more 24-hour biomarker cycles (output). This definition is consistent with others of circadian entrainment based upon the temporal regularity of different biomarkers over the daily cycle, such as the dim light melatonin onset (DLMO), mid-sleep, and minimum core body temperature (CBT min ). 6 12 13 14 15 Circadian disruption occurs when the daily timing or order of these biomarkers varies, which, depending upon the magnitude of these variations, may or may not translate into measurable adverse health effects. Because the timing of the circadian system is largely governed by the 24-hour light-dark cycle, 16 a sudden change in the photic period (e.g., transcontinental flight) can alter the temporal regularity and possibly the relative timing of these biomarkers. Since people's behavior, particularly those in urban/suburban environments, is very often governed by local clock time rather than sunrise and sunset, 17 it is possible that behavioral biomarkers, like mid-sleep, might be minimally affected by the transition to DST (i.e., the local time of mid-sleep remains constant). By comparison, physiological biomarkers, like CBT min or DLMO, might be affected if the light-dark cycle on the retina suddenly changes, because this photic cycle directly affects the superchiasmatic nuclei (SCN). The objective of the present study was to contribute to the debate about the societally mandated seasonal changes to local clock time by assessing how the springtime transition to DST affected the temporal regularity of several biomarkers as well as subjective impressions of sleepiness in a sample of individuals living an urban/suburban lifestyle. The study was conducted over a 4-week period centered on the DST transition in March 2024. Material and Methods Subjects A total of 20 healthy adult subjects (16F/4M, mean age 42 years, SD: 13 years) participated in the study, which ran from February 26, 2024, to May 3, 2024. The present paper is only concerned with the first 4 weeks, bracketing the DST transition on March 10, 2024. Subjects were required to be able to use text messaging, were free of cataracts or macular degeneration, and were not excluded from obtaining a driver's license due to visual deficiencies. Use of oral melatonin and travel outside the U.S. Eastern time zone were not permitted for 2 weeks before and throughout the study duration. Subjects signed an informed consent form approved by the Institutional Review Board at our institution. Protocol Subjects visited our lab in Albany, New York (approximately 42.65° N, 73.75° W), where they received a Speck personal light sensor (Blue Iris Labs), calibrated in terms of circadian-effective light, also called circadian stimulus (CS), 18 19 20 21 to be worn as a pendant during waking hours. During sleep, subjects were instructed to place the Speck next to their bed. Personal, calibrated light exposures were continuously measured throughout the study. Subjects also received an Argon One Pi 4 V2 network router (Argon40) to relay data to the experimenters. Subjects wore a wrist actigraph continuously, except during bathing. Due to a limited number of actigraphs from the same manufacturer, subjects 1 through 6 received a Micro Motionlogger Watch (MMW; Ambulatory Monitoring), while the others received an Actiwatch Spectrum Plus (ASP; Philips Respironics). The MMW devices contain a three-channel accelerometer, and the number (or “count”) of times acceleration exceeded an internally specified threshold in each 1-min measurement epoch is recorded. The ASP devices contain an accelerometer capable of detecting vertical acceleration between 0.5 and 2 g at frequencies between 0.35 and 7.5 Hz; the duration and magnitude of acceleration are integrated and converted to count values in each 1-min measurement epoch. Both sets of time-series actigraphy data, always rendered in terms of local clock time, were used to determine the clock time (hours) for the onset (start) and offset (end) of sleep every day. Tests carried out at our laboratory, independent of the study protocol, confirmed that sleep onset/offset times determined from the MMW and APS were highly correlated ( r = 0.98, p < 0.01) so the data from both devices were combined for subsequent analyses. Subjects also received saliva sampling tubes (Salivette, Sarstedt) that included a plain cotton swab to be used on the 2 nd (March 8, 2024) and the 4 th (March 22, 2024) Friday evenings for assessing DLMO at home. Subjects received text messages each morning of the study, including weekends, near their usual wake time asking them what time they went to bed the previous night, what time they woke up that morning, and whether they used an alarm clock to wake. They also received text messages 4 times daily—wake time, lunchtime (≈ 12 PM), dinnertime (≈ 6 PM), and bedtime—asking them to rate their sleepiness on the 9-point Karolinska Sleepiness Scale (KSS). 22 Subjects collected their own saliva samples while in their homes on Fridays, one at the end of week 2 (just prior to the DST transition) and the other at the end of week 4 (2 weeks after the DST transition). The first saliva sample was collected 2 hours before the subject's usual bedtime and every half hour for the next 4.5 hours (10 samples per night). Starting 30 minutes before the first sample was to be collected, subjects were in dim light while wearing orange glasses (NoCry Blue Light Blocking Over Glasses Safety Glasses) that absorb nearly all spectral power shorter than 530 nm. Participants were instructed not to eat during this period but were allowed to drink water within 15 minutes after each sample collection. Subjects placed each sample in their home freezer immediately after collection and subsequently brought their frozen samples to the laboratory at a convenient appointed time. Melatonin levels were assayed in our laboratory using a commercially available radioimmunoassay kit (Labor Diagnostika Nord). The assay's limit of detection specified by the manufacturer was 2 pg/mL, with intra- and inter-assay coefficients of variability of 11.4 and 12.7%, respectively. Analysis Personal light exposure time-series data from the Speck devices were used with the CS oscillator model 23 to predict the time of core body temperature minimum (CBT min p ) at the end of each week. The Speck device averages and then stores the light data over slightly different time intervals (approximately 1 minute). The stored data were converted into precisely equal time intervals throughout the entire time-series for processing by the model. The time series data were divided into 7 24-hour periods to determine CBT min p at the end of each week. For the first 2 weeks of the study, the time series data started at 6 AM on Monday, February 26, when the study formally began. The time series data for the first week was continuously run through the CS oscillator model until values of CBT min p reached an asymptote. This value was used as the starting CBT min p for the second week; the same procedure was used for all subsequent weeks. The average mid-sleep time was calculated as the midpoint between the actigraphic sleep onset and sleep offset times each night and averaged by week for each subject. The average time in bed was similarly calculated as the mean difference between the sleep onset and offset times. Light exposure and actigraphy data were used together to perform phasor analysis. 24 The synchrony between the 24 hour light-dark pattern and the 24 hour activity-rest pattern over 1-week (phasor magnitude), provides a weekly measure of behavioral circadian entrainment. The temporal offset between the 24 hour light-dark cycle and the 24 hour activity-rest cycle (phasor angle), provides a weekly estimate of how delayed or advanced the activity-rest pattern is with respect to the light-dark exposure pattern. Subjective sleepiness ratings 22 for each subject during each week were averaged across each day of the week and for each collection time (morning, lunchtime, dinnertime, bedtime). The DLMO thresholds were calculated using two methods, either fixed threshold or the two standard deviations (2SD) method, depending on the sampled melatonin concentrations. If the melatonin concentrations of the first three samples were below 2 pg/mL, we used a fixed threshold to define DLMO. The fixed threshold used was either 2 or 4 pg/mL, depending on the values of the subsequent concentrations. Otherwise, we used the 2SD method, calculating a threshold based on the average melatonin concentration of the first 3 time points, plus 2SDs of those same 3 times. Once DLMO thresholds were calculated for each participant, DLMO times were calculated by linear interpolation, using the samples on either side of the threshold. Statistical Procedures Unless otherwise indicated, one-way repeated measures analyses of variance (ANOVA) 25 based on the general linear model (to account for missing or incompletely balanced data) were used to assess differences between the 4 weeks of the study for each outcome measure. The statistical significance criterion for rejecting the null hypothesis in all analyses was p < 0.05. When a significant main effect was present, pairwise comparisons (using Tukey's correction) between weeks were performed. 25 Results Table 1 shows the mean values for each outcome measure during each week of the study. Supplementary Material Table S1 includes the weekly values for each individual participant, indicating when specific data were missing or incomplete. Table 1Weekly outcome measure mean ± SEM and sample sizesOutcome measureSample size*(n)Outcomes, mean ± SEM Main effect ( p- value) Week 1Week 2Week 3Week 4 CBT minp (hours) 20 04:19 ± 00:17 a 04:25 ± 00:16 a,b 04:40 ± 00:17 b 04:38 ± 00:20 a,b 0.04 Actigraphic sleep Sleep onset time (hours)15–17 23:10 ± 00:22 a,b 23:23 ± 00:16 a 22:50 ± 00:17 b 23:05 ± 00:19 a,b 0.03 Sleep offset time (hours)15–1707:21 ± 00:2607:14 ± 00:2407:05 ± 00:2907:11 ± 00:290.49 Mid-sleep time (hours)15–1703:16 ± 00:2203:19 ± 00:1802:58 ± 00:2003:08 ± 00:220.06 Time in bed (hours) ꝉ 15–1708:11 ± 00:2307:50 ± 00:2008:14 ± 00:2608:06 ± 00:250.40 Phasor magnitude 16–170.33 ± 0.020.25 ± 0.020.32 ± 0.030.32 ± 0.070.11 Phasor angle (hours) 16–1701:14 ± 0:1801:06 ± 0:2200:56 ± 0:1101:30 ± 0:200.57 KSS ‡ 0.77 Morning205.12 ± 0.405.20 ± 0.435.08 ± 0.434.72 ± 0.43— Lunch203.38 ± 0.273.63 ± 0.233.58 ± 0.273.67 ± 0.28— Dinner204.07 ± 0.234.17 ± 0.284.08 ± 0.274.52 ± 0.30— Bedtime206.92 ± 0.277.02 ± 0.276.85 ± 0.236.67 ± 0.28— DLMO (hours) 14–16—22:59 ± 0:20—22:58 ± 0:210.21Abbreviations: CBT min p , predicted minimum core body temperature time; DLMO, dim light melatonin onset; KSS, Karolinska sleepiness scale; SEM, standard error of the mean. Notes: *Missing data resulting from device failure or the failure of participants to follow instructions resulted in fewer than 20 values for some outcomes. ꝉ Time in bed is shown as elapsed time (in hours), whereas other actigraphic sleep measures are shown as local clock times. ‡ The significance level for the main effect of week is shown for KSS. A statistically significant effect of time of day ( p < 0.001) and a significant interaction between week and time of day ( p = 0.01) were also found for KSS but are not reported here. Significance: No statistically significant differences exist for pairwise comparisons between weekly mean values marked with the same superscripted lowercase letter. The threshold for statistical significance using Tukey's correction was p < 0.05. Weekly outcome measure mean ± SEM and sample sizes Abbreviations: CBT min p , predicted minimum core body temperature time; DLMO, dim light melatonin onset; KSS, Karolinska sleepiness scale; SEM, standard error of the mean. Notes: *Missing data resulting from device failure or the failure of participants to follow instructions resulted in fewer than 20 values for some outcomes. ꝉ Time in bed is shown as elapsed time (in hours), whereas other actigraphic sleep measures are shown as local clock times. ‡ The significance level for the main effect of week is shown for KSS. A statistically significant effect of time of day ( p < 0.001) and a significant interaction between week and time of day ( p = 0.01) were also found for KSS but are not reported here. Significance: No statistically significant differences exist for pairwise comparisons between weekly mean values marked with the same superscripted lowercase letter. The threshold for statistical significance using Tukey's correction was p < 0.05. With respect to local clock time, sleep onset and sleep offset times both tended to occur earlier (∼ 30 and ∼ 15 min, respectively) after the DST transition, so naturally the circadian phase marker, mid-sleep times, also advanced. Overall, sleep behavior, including mid-sleep, during weeks 1 and 2 appeared to be temporally offset from CBT min p by the same amount ( Fig. 1 ). For weeks 3 and 4, after the transition to DST, the temporal offset between mid-sleep and CBT min p not only increased, but the increase was the same for both weeks. Specifically, mid-sleep times advanced by about 15 min and CBT min p delayed by about 15 min. There was no noticeable change in DLMO times, which were determined for the ends of weeks 2 and 4. Mean mid-sleep, minimum core body temperature time (CBT min p ) and dim light melatonin onset (DLMO) times (mean ± standard error [SE]) in weeks 1 to 4, normalized to the mean of the pre-DST weeks (1-2). To understand the relative delay in CBT min p after DST that is illustrated in Fig. 1 , the hourly average CS exposures during weeks 2 and 3 (immediately before and after the DST transition), as well as the difference between these weeks, are shown in Fig. 2 . In week 3, relative to week 2, there was slightly less light exposure between 6 and 9 AM, slightly more between 9 AM and 12 PM, and a larger increase in light exposure during the afternoon (between 12 and 5 PM) and early evening (between 5 and 7 PM). Mean ± SE hourly circadian stimulus (CS) exposure ( A ) and activity ( B ) profiles for all subjects before (week 2) and after (week 3) the daylight saving time (DST) transition, as well as the differences between the 2 weeks. In the present study, phasor analysis was used as an operational definition of behavioral circadian entrainment, where the phasor magnitude represents the strength in the synchrony between the 24-hour activity-rest cycle and the 24-hour light-dark cycle, and the phasor angle represents the temporal offset between the activity-rest and light-dark cycles. 24 Despite the temporal separation between CBT min p and mid-sleep time after DST, neither the phasor magnitudes nor the phasor angles exhibited significant changes over the study period. As seen in both panels of Fig. 2 , changes in the activity-rest pattern after DST tended to mirror changes in the light-dark pattern. Fig. 3 shows that for the daytime hours when light was available, changes in light exposure after the transition to DST were directionally consistent with changes in physical activity. Consequently, there was a lack of change in the phasor magnitudes and angles after DST. Changes in average hourly activity for each daytime hour from week 2 to 3 (positive values indicate an increase in week 3 relative to 2), plotted as a function of the changes in average hourly CS exposure for the same daytime hours, also from week 2 to 3. The KSS ratings did not exhibit a statistically significant main effect of week. In week 3, the morning KSS ratings did have somewhat higher values (∼ 0.3–0.5 points greater) for the first 2 days immediately following the DST transition than they did for the rest of that week. In fact, the average morning KSS ratings in week 3 were slightly (albeit non-significantly) lower (i.e., participants reported being less sleepy) than in week 2 ( Table 1 ). Discussion In general, the spring transition to DST was not particularly disruptive to this population of urban/suburban adults when assessed on a weekly basis. Subjects experienced slightly greater morning sleepiness the first day or two immediately after the transition, but after 3 days, it returned to the same level as before the transition. Importantly, mid-sleep time and time in bed were not significantly different before and after the DST transition, nor were phasor magnitudes or angles. Phasor magnitudes and angles did not change with the DST transition because not only did subjects receive more personal light exposure later in the day after the DST transition ( Fig. 2A ), but they also increased their activity during those times ( Fig. 2B ). This increase in personal light exposure during the afternoon and evening following the DST transition was likely responsible for the statistically significant delay in CBT min p . Although there was a statistically significant delay in CBT min p and a nearly significant advance in mid-sleep after the DST transition, both changes were on average only about 15 to 20 minutes. A small but nonsignificant increase (∼ 10 minutes) in the average time in bed was observed in the weeks following the DST transition, mainly because of the earlier sleep onset in week 3 relative to week 2. In comparison, Kantermann et al. 26 reported an average seasonal (winter vs. summer) swing in sleep duration of about 20 minutes (shorter sleep during the summer) among 55,000 individuals. The variations in sleep duration for any given week was also about 20 minutes. 26 Consequently, systematic changes related to the DST transition may not be large enough to detect in the present 4-week study. Moreover, DLMO did not change at all with respect to local time. Overall, the transition to DST in the spring seemed to have little effect on sleepiness, time in bed, or circadian phase. So, these results are consistent with the intended effects of DST on society, making afternoon/evening light more available to people. Notwithstanding, it is interesting to note that for the 2 weeks prior to DST the relative timing of the 3 circadian phase markers (CBT min p , mid-sleep, and DLMO) remained constant. After the transition, however, these markers seemed to disassociate slightly, whereby mid-sleep advanced by about 15 to 20 minutes, CBT min p delayed by about 15 minutes, and DLMO was unchanged with respect to local time. This slight disassociation might indicate that there was, in fact, circadian disruption following DST. Whether this relatively small dissociation has societal health implications is unknown, but the present data suggest that such occurrences among circadian phase markers could be an important topic for future epidemiological research. In the interest of future research into the impact of DST on circadian entrainment and sleep quality, it is important to emphasize that the present study was limited to assessing the light-dark exposure pattern changes after DST for an urban/suburban cohort. Their behavior was largely, but not completely, governed by local clock time. Not only their light exposure profiles, but their activity-rest patterns also changed. In particular, the weekly temporal light profiles showed that subjects had greater light exposure (and were somewhat more active) in the afternoon and evening after DST. To accurately model the impact of DST on biomarkers like CBT min p from other populations, it is always necessary to measure personal, calibrated light exposure patterns before and after the transition. Xu et al. 27 and Rea et al. 28 utilized predictions of light-induced changes to CBT min p to interpret the effect of springtime DST transition on circadian phase. Xu et al. 27 adapted the model published by Jewett et al., 29 and Rea et al. 28 utilized the CS oscillator model. 23 Both the Jewett et al. and CS oscillator models are refinements of the one initially proposed by Kronauer et al. 29 30 In both analytical studies, estimated readaptation times to DST were on the order of approximately 10 days. For modeling changes in CBT min p due to the transition, the authors of both studies assumed the light-dark patterns would be the same (in local clock time) before and after the transition because urban/suburban individuals spend most (90%) of their time indoors. This being the case, the authors of both studies simply assumed that 1 hour could be added to the CBT min p local time at the end of ST to the new local time at the beginning of DST. If, for example, CBT min p was 4:30 AM local time at the end of ST (1:59 AM), CBT min p would simply be changed to 5:30 AM local time from the start of the DST time-series data (3 AM). Since we had time-series data before and after the transition to DST, the present study could directly test whether this assumption was appropriate. We employed two techniques to handle the transition to DST time for predicting CBT min p to ensure the length of the daily time-series data was always equal to 24 hours following the transition to DST (rather than 23 hours for the first day). For the first technique, the CS values were all equal to zero at both 1:59 AM ST and 1 min later at 3 AM DST. These CS values were inserted for 1 hour prior to 3 AM DST, making the length of the time-series data for the 1 st day equal to 24 hours. For the 2 nd technique, the time-series data for calculating CBT min p started at 3 AM (rather than 6 AM), essentially ignoring the “missing” hour from 2 to 2:59 AM. Both techniques yielded CBT min p times that were within 1 to 2 minutes of each other for every subject, so the procedures were deemed equivalent; the latter one was used for reporting. It is important to note that we found that by simply adding 1 hour to the time series data following the transition to DST, 27 28 changes in CBT min p values would have been twice the magnitude of those obtained from the two techniques examined here, suggesting a much more prolonged period of circadian re-entrainment and a period much more disparate with changes in mid-sleep. Disparities between published analytical results and the findings from the present study call into question the validity of the assumptions underlying those analytical studies, 27 28 at least under many conditions representative of the healthy suburban/urban adults in our present investigation. Overall, these results suggest that the subjects in the present study were able to compensate for the change in local clock time that accompanies DST with little difficulty. Support for this inference is evidenced by similar KSS ratings, little effect on time in bed, and similar phasor magnitudes and angles before and after the transition. The data presented here underscore the value of including a diverse set of outcome measures, in particular ambulatory light measurements, sleep behavior, and subjective ratings, when assessing the impacts of DST-related transitions on sleep and circadian regulation or, indeed, any attempts to understand the societal impacts of changes to the 24-hour light-dark and 24-hour activity-rest cycles. We cannot claim, however, that our findings will be true for individuals with different lifestyles or for those who live in more extreme latitudes. Nor can we exclude the potential for secular trends over the 4-week study duration, such as the systematic increase in day length, or changes in the observed weekly average outdoor temperatures, which did not exhibit a slope significantly ( p > 0.05) different from zero, to contribute in some way to our findings. Perhaps most importantly, we cannot rule out the potential for health-related impacts from the DST transition. It is important to reiterate that CBT min was not measured directly in this study. However, for at least three reasons, it seems unlikely that errors in the CS oscillator model predictions could account for the modest delay in CBT min p relative to local clock time after the DST transition. First, the accuracy of the model's predictions was supported by the close correspondence between changes in CBT min p before and after a light intervention with changes in DLMO. 23 31 32 33 That validation showed that the CS oscillator model was more accurate in predicting light-induced phase changes to DLMO than was possible from other models, generally resulting in errors slightly greater than half an hour. 23 Here, however, we saw a statistically significant delay in CBT min p of ~ 15 16 17 18 19 20 minutes, so the model may not be able to make predictions to this level of accuracy. Second, mean CBT min p values in the present study were positively correlated ( r = +0.82, p < 0.01) with the mid-sleep times measured in this study. This is another well-established circadian phase marker. 34 Third, the predicted CBT min p values in weeks 1 and 2 changed little relative to each other and to the mid-sleep times during these weeks, so it is difficult to conceive of a reason why or how the model would fail only for weeks 3 and 4. The increased afternoon and evening light (and activity) after the transition to DST ( Fig. 2 ) is consistent with the CBT min p predictions from the CS oscillator model and with our phasor analysis results. One factor that might impact circadian phase-related responses to changes in light-dark patterns and in the timing of social cues brought on by DST transitions is age. For example, Duffy et al. 35 reported that older individuals exhibited smaller phase shifts than younger individuals in response to the same light-based stimuli. One might therefore expect younger individuals to more readily adapt to new light-based regimens than older individuals. Despite an age range from 26 to 60 years in the present study (with a mean of 42 years), we found no statistically significant correlations ( p > 0.05) relating age and either CBT min p or midsleep times. The lack of correlation suggests that age may not be a critically important factor in understanding if and how our results can be generalized to the population at large. Additionally, the timing of food intake can affect CBT min even while melatonin rhythms are unaffected, 36 and food intake was not monitored in the present study. Nor did we ask subjects for their overall impressions about their sleep and wellbeing before or after the transition to DST, aside from their KSS ratings throughout the day. Such data might have provided further insight into our findings. Impacts of DST on health are perhaps the most relevant and important outcome measures, but also among the most difficult to obtain. Strictly speaking, this is not a limitation of this or any other observational study, because valid health-related assessments would require insightful and converging epidemiological studies using health-outcome measurements from large populations over many years. Nevertheless, we could have asked subjects' opinions on how they felt physically and mentally before and after the DST transition. Finally, we note that the measured melatonin concentrations for most participants were low, in some cases not rising above 1 to 2 pg/mL until late in the evening, complicating the DLMO time estimates and possibly reducing their reliability. Further, DLMO was only measured in weeks 2 and 4 and not in weeks 1 and 3. These factors might explain why DLMO times were closer to bedtime than typically reported, and why the DLMO times differed little between weeks 2 and 4. Conclusion Together, a variety of sleep and circadian outcome measures led to the inference that the spring transition from ST to DST had little effect on a sample of urban/suburban participants living their normal lives. However, the transition might have affected other outcomes not formally included in the present study.