Effects of daylight savings time transition on compliance and effectiveness of positive airway pressure therapy

Abstract

Spring transitions into daylight savings time (DST) result in increase of sleep latency and fragmentation and decrease of sleep time and efficiency. We evaluated the effects of DST on patterns of positive airway pressure (PAP) use and its effectiveness by utilizing continuous tracking of PAP therapy available with cloud-based sleep care management systems. 62 compliant adult OSA patients from Jacksonville, Florida, USA on stable continuous PAP (CPAP)/autoPAP (APAP) therapy were enrolled and PAP usage and residual apnea–hypopnea index (AHI) were collected for the “pre-DST period” (Sun–Mon–Tue, 7–9/03/2021) and “DST period” (Sun–Mon–Tue, 14–16/03/2021) from compliance reports generated or downloaded via cloud-based sleep care management systems. Demographic variables, average compliance and effectiveness of PAP during the two weeks that included both intervals of interest (from Wednesday 3rd to Wednesday 17th) were further analyzed. Statistics included repeated measures ANOVA, non-parametric Wilcoxson’s rank sum tests, independent and paired T tests, and Chi-square test. Majority of patients were Caucasian (73%); with average age of 57.5 ± 11 years, 443.1 ± 124 min of nightly PAP use, and 0.97 ± 0.06% of PAP compliance. 73% of patients were male, with no significant differences noted for sleep variables between genders. There was significant decrease in PAP usage duration between pre-DST Mondays and DST Mondays (delta_normalized_PAP_duration = -0.18, p = 0.0027). AHI demonstrated significant decrease on DST Mondays, followed by significant increase on DST Tuesdays (deltaAHI = − 0.54 and 0.47 respectively). This study demonstrated that the effects of DST on duration of PAP use and sleep disruption monitored by AHI are seen days after DST transition, even in the zones with very stable light/dark cycles like Florida, USA.

Introduction

Daylight savings time (DST) is the practice of adjusting the clock time with the presence of daylight two times in a calendar year to synchronize activity during wakefulness to the periods of the peak sun-generated light and heat. DST arguably leads to long-term effects of decreasing cost of energy consumption, increasing physical activity and traffic safety [1–10] although the actual effect on energy savings seems quite small and dependent on the location’s latitude [11–13], at times even causing increased environmental hazards [8]. DST is currently being implemented by approximately one quarter of the world’s population in over 70 countries that are often at higher latitudes and thus with a greater seasonal variation of available sunlight; DST’s impact, however, continues to be argued, as both the European Sleep Research Society and the American Academy of Sleep Medicine released positions that DST should be eliminated in favor of a national, year-round, fixed time [14–16]. Furthermore, in addition to presumed long-term effects of DST, the opponents of DST point out to increasing data showing short-term effects of DST transitions on behavior, cognition, physiology and disease [17–21] which are likely related to significant desynchronization between peripheral body pacemakers and the central pacemaker in the suprachiasmatic nucleus (SCN) governing biological functions down to a cellular level [22, 23]. Direct effects causing dysregulation of sleep structure, duration and efficiency seen during transitions into and out of DTS have also been described [16, 24–26]. Specifically, spring transition into DTS was shown to increase sleep latency, increase slow-wave sleep and sleep fragmentation, and decrease sleep time and efficiency, with the effect being more prominent in the evening types of study participants, and with the adjustment period of sleep schedule during weekend lagging by adjustment during work week for almost two weeks [25, 27–29]. DST effects are further affected by seasonal daylight exposure differences, particularly in zones with higher latitudes [30].

Although 4 h of PAP use during 24-h period commonly defines minimal acceptable levels of adherence to therapy, current evidence suggests a continuous dose–response relationship between hours of use and therapeutic response [31–36]. In our study, we have for the first time compared patterns of PAP use and PAP effectiveness preceding and during spring transition into DTS and discussed their consequences on individual and population level.

Materials and methods

Subjects and definitions

A total of 62 diurnally active patients with medical history of CPAP- or APAP-treated OSA that were scheduled for their yearly clinic follow-up at the University of Florida, Jacksonville clinic were enrolled in this study during the recruitment period, from May 2021 to September 2021. Inclusion criteria were age ≥ 18 y, compliance to PAP for ≥ 95% of nights during the 90 day period preceding the visit, and settings that have not been changed for the period of at least 12 months. Nightly compliance to PAP therapy was defined as ≥ 4 h of PAP use per night [36, 37]. Exclusion criteria were current shift work or history of shift work, presence of circadian rhythm disorder, pregnancy, severe major psychiatric disorder (e.g., schizophrenia, psychosis, and depression).

Study protocol

For each patient, compliance and effectiveness of the therapy was assessed by reports generated via cloud-based sleep care management systems, either Philips Respironics Care Orchestrator or ResMed Air View, depending on the manufacturer of the PAP device used. Specifically, PAP usage duration in minutes and residual AHI were recorded daily for “DST period”, defined as three day period from 14 to 16th of March (i.e. Sunday of DST and the following two days), and for “pre-DST period”, from 7 to 9th of March (i.e. period preceding DST by seven days). Apneas and hypopneas were as defined in the absence of SaO2 measurements by Phillips and Resmed capture programs detailed in ResMed and Phillips Product Manuals, (available at https://www.resmed.com/en-us/sleep-apnea/cpap-parts-support/cpap-product-manuals). For each patient, demographic variables of sex, race, age, and baseline compliance to PAP were collected. Descriptive summaries were frequencies and percentages for categorical data and medians and quartiles for continuous variables. Demographic variables of sex, race and age were tested using Chi-square. Female versus male patients were compared using the non-parametric Wilcoxson’s rank sum tests and independent T tests. Analysis of differences between pre-DST and DST periods was done with Wilcoxon’s signed-rank test and paired T test. All analyses were done in Excel 2016 and SAS for Windows Version 9.4.

Results

Of the total of 62 patients, 73% were male and 27% were female, with most of them (73%) being Caucasian (C) and minority (27%) African American (AA) (Table 1). No Asian or Hispanic patients were enrolled in the study. The average (± S.D.) patient’s age was 57.5 ± 11 y, with the youngest patient in the study being 33 and the oldest 84 years of age. Average PAP usage duration for all dates collected (i.e., pre-DST and DTS period, 6 days total) was 443.1 ± 124 min with average compliance of 0.97 ± 0.06%.

Table 1.

Analysis of demographic variables

Variable	Category	Male (45, 73%)	Female (17, 27%)	Overall (62,100%)	P value
Race	AA	13 (29)	4 (24)	17 (27)	0.67
	C	32 (71)	13 (76)	45 (73)
All tests done using Chi-squared

Variable	Group	N	Mean	Std Dev	Min	1st Quartile	Median	3rd Quartile	Max	P-value
Age	Male	45	56.97	± 12.29	33	49	55	66	84	0.60
	Female	17	58.88	± 6.58	42	57	59	62	73
Sleep duration	Male	45	436.52	129.0	262	361	420	471	931	0.23
	Female	17	455.21	± 97.89	246	376	441	496	713
Sleep Efficacy	Male	45	0.97	± .06	0.8	0.95	1.0	1	1	0.08
	Female	17	0.97	± .06	0.8	0.93	0.93	1	1
All tests done using the non-parametric Wilcoxson’s rank sum tests

f = female, m = male, AA = African American, C = Caucasian

There was no significant difference in race, age, sleep duration and sleep efficacy variables between male and female groups of patients (Table 1). Daily average PAP usage duration profile did not significantly differ between men and women for analyzed days (Fig. 1); therefore, further analysis was carried out with grouped data of all the enrolled patients, regardless of sex.

Fig. 1.

Difference between average PAP use duration (± SD) between sexes calculated for each date. m = males, f = females, Pre-DTS = days one week before DST, DTS = first 3 days of DST

First, we set to analyze change in PAP usage duration during the pre-DST and DST. For each patient, PAP usage durations during these six days were normalized to 0–1 range: shortest PAP usage duration (minPAPD) and longest PAP usage duration (maxPAPD) were identified, and all PAP usage duration times were assigned a value from 0 to 1 by formula nPAPD = (PAPD-minPAPD)/(maxPAPD-minPAPD). Duration of the PAP use on Monday following DST (i.e. DST day2) was significantly shorter than on pre-DST Monday (i.e. pre-DST day2; Fig. 2, p < 0.01, T = -3.129) while PAP use did not significantly change on DST day1 and 3 compared to their pre-DST counterparts.

Fig. 2.

Average normalized PAP use duration in study population. Pre-DTS = days one week before daylight savings time, DTS = first 3 days of daylight savings time. ** indicates statistical significance of p < 0.01 between pre-DST and DST Mondays

We further analyzed differences in effectiveness of PAP therapy between pre-DST and DST intervals, as represented by residual AHI. Residual AHI values during three pre-DST and three DST days were all within mild OSA range with minAHI of 0, maxAHI of 14.9, median AHI of 1.35 and mean AHI of 1.95. While average AHI did not change for DST day1 (Sunday), there was small, but statistically significant decrease in absolute AHI on DST day2 (Monday) compared to pre-DST Monday (mean change = − 0.54 AHI) followed by significant increase of AHI on DST day3 (Tuesday) compared to pre-DST Tuesday (mean change =  + 0.47 AHI, Fig. 3). When normalized, this change represented decrease of 18% of AHI from pre-DST day2 to DST day2 and 15% increase of AHI on DST day3 compared to pre-DST day3, Fig. 4).

Fig. 3.

Absolute values of residual AHI during pre-DST and DST period. ** = p < 0.01, * = p < 0.05

Fig. 4.

Average normalized AHI during pre-DST and DST period. ** = p < 0.01, * = p < 0.05

Discussion

Growing research regarding benefits vs. harms of DSL has given evidence of short-term worsening and destabilization of a number of physiological and disease processes spanning frequency and severity of cardiovascular events and stroke, mortality of respiratory disease, success of in vitro fertilization (IVF), severity and frequency of mental disorder symptoms, and frequency of injuries and ER visits, just to name a few [18–20, 22, 38–41]. It has been argued that the effects noted are likely related to significant desynchronization between peripheral pacemakers and the central pacemaker governing biological functions down to a cellular level [22, 23, 25, 28, 29]. Although daylight duration and seasonality result in significant differences in circadian physiology, the effects of DST transitions at different latitudes have been less well studied. One could hypothesize more prominent effect of DST transition occurring on higher latitudes. This, in fact, has been focus of our ongoing research in Vermont, USA where both DST and seasonal daylight changes are likely to influence PAP therapy.

While DST transitions directly affect duration of overnight PAP therapy, the effects of DST on effectiveness of PAP were not demonstrated before. Cloud based PAP therapy monitoring allowed us to directly examine DST effects on patients that otherwise demonstrate high compliance and effectiveness of PAP. Enrolling such patients in this study minimized possibility that observations reflect behavioral/disease variation, although, of course, this possibility must be mentioned as a limitation of this study, since patients were not directly queried about their circumstances during pre-DST and DST periods.

Our study demonstrated the effect on duration of PAP therapy on the DST + 1 day, as well as the change of the effectiveness of PAP and OSA severity during days following DST as demonstrated by residual nightly AHI. In our sample, the most prominent change occurred on the DST + 2 night, thus likely affecting daytime functioning in the middle of the week. Thus, for example, one could expect driving safety to be lowest during Wednesdays after DST as it follows the most disrupted night, as measured by AHI.

Obvious limitations here are small patient sample and relatively short follow-up time of three days post DST. Although females seemed to use PAP longer and recovered faster during DST period, both absolute and normalized data of PAP use did not statistically differ between genders, thus possibly reflecting relatively small sample size in our study. Ideally, large multicenter patient group followed for longer period of 1 or 2 weeks (i.e., until normalization of sleep parameters is noted) would clearly demonstrate the extent and duration of specific changes observed. Recent national study of Zhao et al. on mortality related to DST did not show association between spring DST and multiple causes of mortality for following eight weeks [41]. Although our result seemingly contradicts with this finding, it in fact reflects a different point, that there is the worsening of PAP use duration and sleep disruption in the patients diagnosed and treated for OSA, peaking on different days following DST, as opposed to population mortality rates analyzed predominantly on the basis of racial and age groups. Zhao et al. also suggested significant decrease in mortality risk during 8 weeks following DST transition in autumn that sets back the clock by 1 h (thus presumably increasing total sleep time) [25]. Although not a part of the study presented here, we are planning to include analysis of both spring and autumn DST transitions on different latitudes in patients with OSA diagnosis in future.

Our result showing a decrease of AHI on DST + 1 night might be related to shortening of sleep and preserving the first part of the overnight sleep with relative paucity of REM sleep, which is typically associated with highest frequency of obstructive respiratory events. Similarly, the following increase of the mean AHI on DST + 2 is likely related to REM sleep rebound that follows decrease of sleep duration on DST Monday. Relatively small absolute change probably relates to required good compliance and SBD control of OSA during study enrollment to eliminate day-to-day behavioral variability. However, when normalized, the change represented 30% increase in AHI from DST Mondays to Tuesdays, which, depending on the baseline AHI level, could significantly affect severity of residual breathing disturbance (Fig. 4). Daily PAP use enabled us to indirectly monitor daily differences in sleep quality in PAP users; however, it is very likely that similar worsening of AHI occurs in undiagnosed OSA patients, which represent up to 80% of all OSA cases.

Another point that should be discussed is that our data relied on PAP devices identifying frequency and the type of respiratory events, thus raising a question of PAP sensitivity and specificity of monitoring when compared to standard PSG. Previous analysis demonstrated good correlation between PAP- and formal PSG-detected events, although PAP devices demonstrated less effectiveness in determining specific type of respiratory event [36, 42, 43].

There is growing evidence that rather than defining optimal PAP therapy discreetly (as the requirement of 4 h of sleep per night and residual AHI < 5), effectiveness of PAP should be viewed as component of continuum of relative change in sleep duration and respiratory event control, where every sleep length and AHI change is likely to be associated with specific sleep risks [31–34, 42]. This study offers evidence that DST results in changes of quality in PAP therapy that are deserving of our awareness (particularly due to its delayed effect) and attempts to eliminate or minimize these changes.

Declarations

Conflict of interst

Authors Andreja Packard, Jamie S Amos, and Emir Festic declare that they have no conflict of interest.

Ethical approval

All procedures performed in this study are in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards and informed consent was obtained from all individual participants. Study was approved by the Institutional Review Board of the University of Florida, IRB201800149.