Sleep characteristics and self‐reported sleep quality in the oldest‐old: Results from a prospective longitudinal cohort study

Summary

Little is known about the correlation between subjective perception and objective measures of sleep quality in particular in the oldest‐old. The aim of this study was to perform longitudinal home sleep monitoring in this age group, and to correlate results with self‐reported sleep quality. This is a prospective longitudinal home sleep‐monitoring study in 12 oldest‐old persons (age 83–100 years, mean 93 years, 10 females) without serious sleep disorders over 1 month using a contactless piezoelectric bed sensor (EMFIT QS). Participants provided daily information about perceived sleep. Duration in bed: 264–639 min (M = 476 min, SD = 94 min); sleep duration: 239–561 min (M = 418 min, SD = 91 min); sleep efficiency: 83.9%–90.7% (M = 87.4%, SD = 5.0%); rapid eye movement sleep: 21.1%–29.0% (M = 24.9%, SD = 5.5%); deep sleep: 13.3%–19.6% (M = 16.8%, SD = 4.5%). All but one participant showed a weak (r = 0.2–0.39) or very weak (r = 0–0.19) positive or negative correlation between self‐rated sleep quality and the sleep score. In conclusion, longitudinal sleep monitoring in the home of elderly people by a contactless piezoelectric sensor system is feasible and well accepted. Subjective perception of sleep quality does not correlate well with objective measures in our study. Our findings may help to develop new approaches to sleep problems in the oldest‐old including home monitoring. Further studies are needed to explore the full potential of this approach.

1. INTRODUCTION

Population aging is a major problem in most developed countries around the world. Keeping elderly people healthy and independent as long as possible contributes not only to their quality of life, but is also crucial to reduce the need for professional care and with that to decrease the financial burden for the society (Bloom et al., 2015). In the past few years, sleep has become increasingly recognized as an important contributor to health and well‐being (Ramar et al., 2021). Recently, sleep disturbances have been added to the more classical risk factors as a major cardiovascular risk factor (Lloyd‐Jones et al., 2022). However, sleep research is mainly performed in persons with suspected sleep disorders either under laboratory conditions or by using portable devices at home, by population‐based surveys using questionnaires or interviews (West & Kushida, 2021), or more recently by using telemedicine technologies (Shamim‐Uzzaman et al., 2021). Results from longitudinal sleep measurements at individuals' homes are rare and are mainly derived from wearables such as smart watches (Aubert & Brauers, 2008; Fallmann & Chen, 2019; Perez‐Pozuelo et al., 2020). This is particularly true for elderly people. Studies have shown that older adults living in an assisted living home are prevalent to sleep disturbances. There are multiple factors that contribute to this condition (e.g. difficulties to fall and stay asleep, pain, and noise). Poor sleep as assessed by self‐reporting and by objective measurement (wrist actigraphy) has been found to be associated with worse quality of life due to decreased independence during activities of daily living and an increased state of depression (Martin et al., 2008). Dietary habits are another important factor with a potential impact on a person's sleep behaviour. Previous studies have shown that sleeping too short (< 7 hr) or too long (> 8 hr) is associated with an unhealthy diet, although the causality between these variables is still unknown (Štefan et al., 2018). A comparison between older adults with and without insomnia complaints in regard to measures of sleep, mood, lifestyle, health and sleep‐requirement expectations revealed greater discrepancies between current sleep patterns and sleep‐requirement expectations in poor sleepers compared with good sleepers (Morin & Gramling, 1989). The main difference separating older persons with poor sleep from those with a good sleep was the time spent awake after the sleep onset and the frequency of such events. One finding of particular interest in this study was that medical disorders, pain conditions and drug usage (other than sleep aids) did not distinguish the two groups (Morin & Gramling, 1989). In view of the fact that complaints about sleep problems are common in particular in this age group, an opportunity to monitor sleep in elderly people over several days or even longer time periods at their home in a simple and suitable way is highly desirable.

Recently, we have reported the potential of contactless and continuous sleep monitoring in detecting early signs of health problems in community‐dwelling older adults (Schütz et al., 2021).

The aim of this exploratory prospective longitudinal study was to analyse sleep characteristics by contactless and continuous home sleep monitoring, and to correlate findings with subjective perception of sleep quality in a naturalistic sample of oldest‐old individuals over a period of 1 month.

2. METHODS

2.1. Study setting

The study has been performed in an assisted living home in central Switzerland, where each participant was living in an equally designed 2½ room apartment in the same building with equal environmental conditions over the same study period of 28 days. The study duration has been selected to compensate for bias due to extraordinary events.

2.2. Study participants

Fourteen consecutive single living seniors in the age group between 80 and 100 years without a known sleep disorder and being able to handle their life without major assistance have been invited to participate at the study, of whom 13 agreed with the study procedures. One senior had a water bed, which is not suitable for sleep measurement by using an Emfit QS device, and one senior declined because of a combination of serious musculoskeletal pain and depression.

All participants provided written informed consent to participate in the study, and to use their data for research and publications without restrictions. The study has been conducted based on principles in the Declaration of Helsinki, and study procedures have been approved by the regional Ethics Committee.

2.3. Ambient sensor system

The practicality and reliability of the ambient sensor system that has been used in this study for sleep and activity analysis has been previously confirmed in this setting (Saner et al., 2020; Saner et al., 2021; Schütz et al., 2019; Schütz et al., 2022). Heart rate, respiration rate and toss‐and‐turn counts have been recorded in bed with a commercially available EMFIT QS device (Emfit, Finland). This is a contact‐free piezoelectric sensor, which is placed under the subject's mattress. These sensors use thin quasi‐piezoelectric films measuring even light pressure differences as produced by the beating heart. The EMFIT QS sensor has been fixed under the participant's mattress, in proximity to the chest. The device requires no further manual intervention (Saner et al., 2020). Recent results suggest that this sensor can accurately measure heart rate and respiration rate (Aittokoski et al., 2019). Sleep staging algorithms of Emfit‐based sensors have shown a reasonable agreement in comparison with polysomnography‐based ratings (Kelly et al., 2012; Kortelainen et al., 2010). Monitoring data are transmitted to the cloud in real‐time through local WiFi and, subsequently, the Global System for Mobile Communications (GSM) network. The device extracts a variety of vital signs, including heart rate, respiration rate, heart rate variability, movements in bed, sleep duration, and sleep‐onset delay (Saner et al., 2020).

Physical activity in the apartment and toilet visits were quantified using a commercially available passive infrared (PIR) motion‐sensing system from DomoSafety S.A., Switzerland (Schütz et al., 2019).

2.4. Sleep parameters

The analysed sleep parameters include commonly measured quality‐of‐sleep metrics, such as sleep efficiency, number of awakenings, rapid eye movement (REM) sleep, deep sleep and light sleep (Ohayon et al., 2004). Additionally, body movements, further referred to as the number of toss‐and‐turn events, were analysed. Body movement can be broadly described as a binary discretization (dichotomization) of the activity signal over short‐time epochs of a few seconds. This is conceptually similar to what is being done when measuring activity counts with wearable accelerometers. All parameters are derived from proprietary algorithms from the manufacturer (Schütz et al., 2021).

2.5. Subjective sleep assessment

Participants received a questionnaire for daily subjective sleep assessment of the past night using a 10‐point visual analogic scale (from 0 = very bad to 5 = moderately good to 10 = excellent). In case of disturbed sleep, the reason for this problem had to be indicated.

2.6. Data processing

After initial parsing, all collected data were stored in medical‐grade, secure data centres according to Swiss regulations. The recorded sleep data of EMFIT QS presented in this work were extracted based on proprietary algorithms from Emfit, and aggregated on a daily basis for graphical display (mean aggregation for heart rate, respiration rate and heart rate variability; sum for larger movements in bed; Saner et al., 2020).

3. STATISTICAL ANALYSIS

3.1. Sleep score calculation

The sleep score is calculated by a mix of sleep duration, REM and deep sleep duration and interruptions, as follows:

$$\text{Sleep Score}=\begin{array}{l}(\left(t_{\text{slee}{\mathrm{p}}_{\mathrm{tot}}}+t_{\mathrm{REM}}\times 0.5+t_{\text{DEEP}}\times 1.5\right)\\ -\left(t_{\text{awake}}\times 0.5+\frac{n_{\text{wakenings}}}{15}\right))\times 8.5\end{array}$$ (1)

where $t_{\text{slee}{\mathrm{p}}_{\mathrm{tot}}}$: total duration of sleep (hr); $t_{\mathrm{REM}}$: duration of REM sleep (hr); $t_{\text{DEEP}}$: duration of DEEP sleep (hr); $t_{\text{awake}}$: duration awake (hr); $n_{\text{wakenings}}$: number of wakenings (−).

3.2. Correlation between objective sleep score and individual perception of sleep quality

After a run‐in period of 2 days, the objective sleep scores per night were collected and calculated by the EMFIT QS system during 28 days of sleep monitoring. In addition to the objective sleep study data, participant subjective sleep scores were collected using questionnaires. The self‐perceived sleep scores were reported on a continuous interval from 0 to 10 (equivalent to 100 mm in scale length) and extracted using a stencil of vertical lines with a resolution of 2 mm. This interpretation may introduce evaluation bias, but was found to be justified as many participants did not use the categorical scoring scale (“Ganz schlecht”, “Eher schlecht”, “Mässig”, “Gut”, “Sehr gut”) as many scoring marks were distributed between the defined categories. In cases where certain nights did not show explicit scoring marks “X” but written text describing the sleep experience, such text was transferred to a scoring. Although assigning a score to text may not seem intuitive and consistent with participant interpretations, a similar source of error is present in conversations between patient and physician. The underlying data were tested for normal distribution using Q‐Q plots as a qualitative exploration method and Shapiro–Wilk as a quantitative statistics method. The latter test showed that only the variables of P03 may be normally distributed (t = 0.929, p = 0.073).

To analyse the correlation between the two variables “Self‐rated” and “Sleep score”, the monotonic relationship using the Spearman's rank correlation coefficient (r) was calculated. Spearman was used based on the non‐normal distribution of the data and the assumption that the participants have different perceptions of sleep quality. Some participants rated their sleep with a score of about 7 despite having issues in falling asleep, while others that encountered the same issue rated their sleep with a score below 3. This example demonstrates that the rating of the self‐perceived sleep quality depends on the individual impressions of what is good sleep and therefore justifies the assumption of monotonic relationships between the two variables. The resulting correlation plots show a generally weak or very weak relationship between both variables with the exception of the plot of participant P11, which showed a moderate, negative correlation. Additionally, the participants show different numbers of nights (N) included in the correlation analysis, due to missing EMFIT data or not completed questionnaires.

4. RESULTS

4.1. Study participants

During the study period of 28 days, participants (age 83–100 years, mean 93 years, 10 females) were continuously monitored by the EMFIT QS bed sensor and a pervasive ambient sensor system DomoCare® (DomoHealth SA, Switzerland). Acceptance rate was 93%. Sensor data are complete for 301/336 registration days. All study participants completed the study period. All participants have been able to perform their daily activities without help, and to move within the house and for short walks outside with or without a walker. No participant suffered from acute illness during the study period, whereas chronic diseases were prevalent as expected in this age group (Table 1). Diagnoses included stable ischaemic heart disease (West & Kushida, 2021), atrial fibrillation (Shamim‐Uzzaman et al., 2021), pacemaker (Ramar et al., 2021), arterial hypertension (Perez‐Pozuelo et al., 2020), diabetes (Lloyd‐Jones et al., 2022), mild to moderate renal failure (Ramar et al., 2021), mild depression (Ramar et al., 2021) and mild chronic obstructive pulmonary disease (COPD; Bloom et al., 2015). One participant was taking a mild antidepressant, (participant 01), one participant a tranquilliser (participant 05), and one participant a γ‐aminobutyric acid (GABA)‐receptor‐modulator because of difficulties in falling asleep (participant 02).

TABLE 1.

Baseline characteristics of study participants.

No.	Age (years), gender	Comorbidities	Medication per day
1	93, m	stable CVD, diabetes	Rosouvastatin 20 mg, Empagliflozin 10 mg, Insulin, Trazodon 100 mg
2	87, f	stable CVD, SVES, hypertension	Bilol 10 mg, Ramipril 5 mg, Simvastatin 10 mg, Zolpidem 10 mg
3	86, f	AFib, myelofibrosis	Procoralan 5 mg, Apixaban 10 mg
4	96, f	hypertension, osteoporosis	Amlodipin 5 mg, Pregabalin 150 mg, Venlaflaxin 75 mg
5	100, f	hypertension, palpitations, depression	Oxazepam 10 mg, Meto‐Zerok 25 mg, Comilorid 2.5/25 mg
6	93, f	stable CVD, hypertension	Amlodipin 10 mg, Moxonidin 0.2 mg, Irbesartan 160 mg, Clopidogrel 75 mg
7	92, f	stable CVD, PM, hypertension, diabetes	Atorvastatin 40 mg, Candesartan 8 mg, Metoprolol 50 mg, Torasemid 10 mg, Endoxaban 30 mg
8	100, f	stable CVD, hypertension, AFib	Atorvastatin 20 mg, Torasemid 5 mg, Xarelto 20 mg, Oxycodone/Naloxane2 × 1
9	95, f	stable CVD, AFib, PM, hypertension, anaemia, coxarthrosis pain	Irbesartan 150 mg, Torem 10 mg, Coumadin, Zaldiar 37.5/325 mg
10	97, m	stable CVD, heart failure, renal failure, hypertension, diabetes	Irbesartan 300 mg, Torasemid 10 mg, Carvedilol 5 mg, Aspirin 100 mg
11	84, f	mild asthma bronchiale, hypertension, diarrhoea	Crestastatin 10 mg, Losartan 100 mg, Nebivolol 2.5 mg
12	93, f	stable CVD, AFib, hypertension	Torasemid 5 mg, Bilol 5 mg, Apixaban 20 mg

AFib, atrial fibrillation; CVD, cardiovascular disease; f, female; m, male; PM, pacemaker; SVES, supraventricular extrasystole.

4.2. Sleep characteristics

Sleep duration: The sleep duration varied from 239 to 561 min (M = 418 min, SD = 91 min).

Sleep efficiency: The ratio between the time in bed and the time asleep was 83.9%–90.7% (M = 87.4%, SD = 5.0%).

Sleep score overall: The sleep score varied considerably between 44 and 98 (M = 74.8, SD = 17.4). The mean sleep score was < 70 in 4/12 participants.

Sleep score in patients with sleep medication: Participant 01: mean 78.6; participant 02: mean 98.1; participant 05: mean 76.7.

Sleep stages: Participants were in REM sleep 21.1%–29.0% (M = 24.9%, SD = 5.5%), in deep sleep 13.3%–19.6% (M = 16.8%, SD = 4.5%), and in light sleep 54.9%–62.1% (M = 58.2%, SD = 4.3%) of total sleep time (Figure 1). The percentage of the different sleep phases was stable over 28 days (Table 2).

FIGURE 1.

Stacked bar plot illustrating the percentage distribution of the three major sleep phases during the sleep at night recorded by an EMFIT system. This sleep study was performed during 1 month with the first 2 days spent installing the EMFIT systems for each participant and 28 days as an effective measurement phase. Each bar represents an averaged percentage distribution of the three sleep phases: “Deep” (blue); “REM” (orange); and “Light” (yellow); calculated from the observed population (N = 12). The recorded month shows a similar daily percentage distribution for Deep (M = 16.8%, SD = 4.5%), REM (M = 24.9%, SD = 5.5%) and Light (M = 58.2%, SD = 4.3%) sleep phases. REM, rapid eye movement.

TABLE 2.

Average duration of sleep stages per night.

Nights	Participants	Light		REM		Deep
	N	M	%	M	%	M	%
1	System installation
2	System installation
3	4	222	54.9	120	29.7	63	15.5
4	9	232	58.7	90	22.8	73	18.5
5	9	248	57.3	121	27.9	64	14.8
6	12	229	59.7	92	23.9	63	16.3
7	12	224	58.3	86	22.3	74	19.3
8	12	255	60.4	97	22.9	70	16.6
9	12	240	59.0	94	23.0	73	18.0
10	11	243	61.9	87	22.1	63	16.0
11	12	232	58.1	104	26.1	63	15.8
12	12	226	58.3	93	24.0	68	17.7
13	12	238	58.2	93	22.8	78	19.0
14	10	230	58.4	89	22.6	75	19.0
15	11	255	57.4	108	24.2	82	18.4
16	11	269	57.8	121	25.9	76	16.4
17	12	239	59.5	101	25.1	62	15.4
18	11	227	57.4	106	26.8	62	15.8
19	11	203	56.8	93	26.1	61	17.1
20	10	262	58.8	112	25.0	72	16.2
21	9	276	58.3	126	26.5	72	15.2
22	10	237	59.0	95	23.6	70	17.4
23	11	250	57.5	113	26.0	71	16.4
24	11	254	55.7	130	28.5	72	15.8
25	11	256	58.0	117	26.5	69	15.6
26	12	241	59.3	91	22.5	73	18.1
27	11	251	59.8	96	22.8	73	17.4
28	11	257	58.4	108	24.4	76	17.2
29	11	268	57.8	128	27.7	68	14.6
30	11	236	56.7	110	26.5	70	16.8

The first 2 days of the month were not measured (System Installation). The number of participants (N) with successfully measured sleep stages varied from 4 to 12 participants per night. The average sleep duration (M) was reported in minutes. The percentages (%) of average sleep for the single sleep stages (Light, REM, Deep) were rounded to one decimal place, resulting in an error of ± 0.1%.

REM, rapid eye movement.

Toss and turns: There was a wide variety in regard to movements in bed defined as toss and turns, ranking from a mean of 16 per night to 106 per night (mean 33 per night).

Bed exits: Participants reported between 0 and 4 bed exits per night, mean 1.7 exits.

Exclusion of sleep disorders: By adding EMFIT QS data in regard to heart rate, respiration and activity data from PIR motion sensors in the apartment, we could exclude sleep disorders such as sleep apnea syndrome and cases of serious insomnia with daytime sleepiness and/or excessive daytime sleeping.

4.3. Subjective perception of sleep quality

Information about subjective perception of sleep quality was available for 329/336 nights. Participants reported that the procedures related to sleep measurements have been perceived as easy and non‐disturbing after the 4‐weeks measurement period. One participant (P05) did not report the first 7 days. Participant P07 did report 22 days with written text instead of the scoring scale. Three participants rated their sleep quality generally as impaired: one because of chronic itching (participant 07); one because of musculoskeletal pain (participant 09); and one because of having to get‐up up to 4 times per night due to nycturia (participant 06).

Correlations between subjective indications of sleep quality on a 10‐point visual analogue scale with a total sleep score on a 100‐point scale are shown below (Figure 2).

FIGURE 2.

Correlation analysis of the Self‐rated sleep quality compared with the Sleep score measured using the EMFIT QS systems. Each plot shows the relationship between the Self‐rated score (x‐axis) extracted from the questionnaires and the EMFIT sleep score (y‐axis) for every night (N) recorded and the resulting Spearman correlation coefficient (r). Additionally, the graphs show regression curves (coloured lines) with confidence bands (shaded area along the regression curve).

4.4. Correlation between sleep score and subjective perception

Figure 2 shows the individual results for the correlation between sleep score and self‐rated sleep score. All participants showed a weak (r = 0.2–0.39) or very weak (r = 0–0.19) positive or negative correlation, except for one participant (P11) who showed a moderate (r = 0.40–0.59) negative correlation (r = −0.55).

4.5. Population correlation between subjective perception and sleep parameters

The population Spearman correlation coefficient (ρ) of Table 3 shows very weak negative correlations (ρ = 0.0–0.19) between the self‐rated scores and the measured sleep parameters by the EMFIT QS, except for the toss and turn parameter, which shows a moderate (ρ = 0.40–0.59) negative correlation (ρ = −0.44).

TABLE 3.

Population correlation (ρ) of self‐rated score and sleep parameters.

Parameters	Self‐rated
	Ρ
Sleep score	−0.11
Sleep efficiency	−0.14
Duration in light	−0.17
Duration in REM	−0.11
Duration in deep	−0.04
Toss and turn	−0.44
Bed exits	−0.15

N = 295.

REM, rapid eye movement.

4.6. Failure rate of EMFIT QS systems

The EMFIT QS systems successfully recorded 301 nights from a total of 336 measured nights, which results in a mean device failure rate of 10.4%. The number of device failures per participant (n) varied from 0 to 12 failed measurements. The highest failure rate of 42.9% was observed for participant P08 (Table 4).

TABLE 4.

Device failure rate per participant using the EMFIT QS system.

Participants	EMFIT
	n	%‐FR
P01	2	7.1
P02	4	14.3
P03	2	7.1
P04	0	0
P05	2	7.1
P06	1	3.6
P07	4	14.3
P08	12	42.9
P09	1	3.6
P10	1	3.6
P11	2	7.1
P12	4	14.3

N = 28.

The failure rates (%‐FR) of the EMFIT QS devices were 0%–42.9% (M = 10.4%, SD = 11.3%).

5. DISCUSSION

This is to our best knowledge the first longitudinal sleep study with objective measurements of sleep parameters in the age group of oldest‐old seniors without specific sleep disorders. The major findings are: (1) that continuous sleep monitoring using a contactless piezoelectric sensor placed under the mattress (EMFIT QS) is feasible and well accepted in the oldest‐old over time periods of up to 1 month with an acceptance rate of 93%; (2) that sleep patterns in regard to sleep duration, sleep efficiency and sleep phases including deep, REM and light sleep are showing similar patterns as in middle ages (Ohayon et al., 2004); and (3) that the subjective perception of sleep quality does not correlate well with the sleep score as an indicator of the objective sleep quality in most individuals in this age group. This has important consequences for healthcare professionals for their approach to evaluate sleep quality in this age group, and to prescribe sleep medication or tranquillisers based on individual complaints.

Sleep problems are common in older adults. While some problems are thought to be a result of normal aging processes, many are indicative of underlying medical or psychological disorders or comorbidities (Bloom et al., 2009; Li et al., 2018). In this context, it has been shown that community‐dwelling older women with higher sleep‐onset delay (the time it takes to fall asleep) and lower sleep efficiency (fraction of time spent asleep while in bed) were significantly more likely to be placed in a nursing home (Spira et al., 2012). It has also been shown that sleep‐related problems are an important factor driving institutionalizations among older adults, and that they are often poorly recognized by caregivers (Marasinghe, 2014; Spira et al., 2012).

In an earlier study, we found promising indications that contactless sleep monitoring may have a great potential for early detection of health problems in community‐dwelling older adults (Schütz et al., 2021). The results of the present study confirm that contactless monitoring of sleep with a piezoelectric sensor (EMFIT QS) placed under the mattress is feasible and well accepted by the oldest‐old over time periods of up to 1 month. This is of particular importance because elderly people generally prefer contactless sensors to wearables (Pais et al., 2020).

Changes in sleep patterns with aging include advanced sleep timing, shortened nocturnal sleep duration, increased frequency of daytime naps, increased number of nocturnal awakenings, time spent awake during the night, and decreased slow‐wave sleep (Li et al., 2018). Whereas most of these changes seem to occur between young and middle adulthood, sleep parameters remain largely unchanged among healthy older adults (Li et al., 2018). We did not find any results published in regard to sleep phases in a similar population of oldest‐old seniors. However, the mean sleep duration was 418 ± 91 min, which corresponds to what is expected in younger and middle‐aged persons. A similar finding applies to the sleep efficiency of 87.4 ± 5%.

The fact that the subjective perception of sleep quality did not correlate well with the measured sleep patterns in most of our participants at a very high age is an important finding. The sleep score as used in our study is a metric based on differently weighted parameters such as sleep duration, awake duration and the sleep stage durations (Equation (1)). Therefore, this sleep score conceptually is an indicator of sleep quantity and to a lesser extent sleep quality. There are no proper validation studies associated with this aggregation of sleep metrics. Thus, there are no validated cut‐off scores. Nevertheless, the sleep score as used for this study has been common practice in this field for almost 15 years.

In our population, neither sleep efficiency nor duration in light, REM or deep sleep correlated well with the self‐rated sleep score. However, both are measuring different constructs, and one is not inherently better than the other. Furthermore, it is unclear how the sleep score relates to health and function. Our study population is too heterogeneous in regard to the health status, and the number of participants is too low to draw firm conclusions in this regard. In previous studies, poor sleep has been found to be associated with declining functional status and quality of life and with greater depression (Martin et al., 2008), with unhealthy nutrition (Štefan et al., 2018) and in particular with poor health (Ohayon et al., 2004). Older persons that reported poor sleep have shown both a longer time spent awake after the sleep onset and a higher frequency of such events (Morin & Gramling, 1989). Furthermore, a comparison of these two groups showed more frequent use of sleeping aids for persons with insomnia complaints. In the same study, poor sleepers showed greater discrepancies between their current sleep patterns and sleep‐requirement expectations than did good sleepers. A rather surprising finding in this study was that medical disorders, pain conditions and drug usage (other than sleep aids) did not distinguish the two groups (Morin & Gramling, 1989).

Additional parameters with potential impact on self‐rated sleep score are bed exits and toss and turns in bed, which can also be measured with the Emfit QS bed sensor. Whereas bed exits showed a very weak negative correlation (ρ = −0.15), the number of toss and turns in bed overnight parameter showed a moderate (ρ = 0.40–0.59) negative correlation (ρ = −0.44). This finding is consistent with earlier findings from a similar patient cohort (Schütz et al., 2022), where higher overall in‐bed activity (possibly restless and worse quality sleep) seemed to be associated with higher fall‐risk.

Sleep problems are a major complaint from elderly people during consultations in the doctor's office. Doctors have to rely on these complaints and make decisions based on this subjective perception of sleep quality. This has important consequences for interventions such as prescription of sleep medication or tranquillisers. As sleep medication appears to have a large impact on falls at night in older age groups (Kobayashi et al., 2021), objective assessment of sleep patterns and sleep quality could have a major impact on fall reduction and fall‐related injuries such as hip fracture. This would reduce nursing home entries and healthcare costs.

The strength of the study is based on the inclusion of 12/14 consecutive participants from this highest old age group, and the fact that study conditions have been close to a laboratory setting in regard to living conditions and environment. Another strength is that only a few data have been missing in this cohort.

5.1. Limitations

The major limitation is the small number of study participants, which is an inherent problem in view of the highly selected study population. Therefore, the findings should primarily be seen as a description of a promising concept for future sleep research in the oldest‐old population, which merits to be further explored. The algorithm for the sleep score, which is used with EMFIT QS devices, is based on experience and assumptions but not on hard facts. Furthermore, we cannot exclude that there were not any naps during the day. The ambient sensors (PIR sensors) distributed in the apartment only provide binary presence/absence information.

6. CONCLUSION

Continuous sleep monitoring using a contactless piezoelectric sensor placed under the mattress (EMFIT QS) is feasible and well accepted in oldest‐old over time periods of up to 1 month. Sleep patterns in the oldest‐old are similar to those in younger individuals if sleep disorders are absent. However, subjective perception of sleep quality seems to not correlate well with objective measures. Our findings may help to develop new approaches to sleep problems in the oldest‐old, including appropriate use of sleep medications by using objective measures of sleep quality. Further studies are needed to explore the full potential of this approach.

AUTHOR CONTRIBUTIONS

Hugo Saner: Conceptualization; methodology; data curation; investigation; validation; formal analysis; supervision; visualization; writing – review and editing. Kevin Möri: Conceptualization; methodology; validation; formal analysis; visualization; data curation; writing – review and editing. Narayan Schütz: Conceptualization; methodology; formal analysis; writing – review and editing; validation. Philipp Buluschek: Conceptualization; software; investigation; writing – review and editing; validation; data curation. Tobias Nef: Investigation; validation; supervision; writing – review and editing.

FUNDING INFORMATION

This research did not receive any specific grant from funding agencies in the public, commercial or not‐for‐profit sectors.

CONFLICT OF INTEREST STATEMENT

Dr Philipp Buluschek and Dr Narayan Schuetz haven been employed by Domo Health SA, which is the main distributor of the Emfit QS system in Switzerland. The other authors declare no potential conflicts of interest.

Saner, H. , Möri, K. , Schütz, N. , Buluschek, P. , & Nef, T. (2025). Sleep characteristics and self‐reported sleep quality in the oldest‐old: Results from a prospective longitudinal cohort study. Journal of Sleep Research, 34(2), e14348. 10.1111/jsr.14348

DATA AVAILABILITY STATEMENT

Study protocol, statistical analysis plan and individual participant data that underlie the results reported in this article will be available after de‐identification for researchers who provide a methodologically sound proposal, beginning 3 months and ending 36 months following article publication.