Lighting and Sleep Quality in the Elderly: A Systematic Review to Inform Future Research Design

Abstract

Exposure to light is an important factor in regulating sleep and sleep quality, especially for elderly people with a high risk of sleep problems. A systematic literature review was conducted to explore the current understanding of the relationship between light and sleep quality of the elderly, and to identify methodological gaps and soundness of existing studies to inform the design of future experimental studies. Specific focus is given to healthcare centres and similar settings due to their controlled environment and the high prevalence of sleep disturbances. Out of 406 publications screened from four databases—namely Google Scholar, Semantic Scholar, Lens.Org, and Scopus—380 studies remained after removing duplicates, and 19 studies published after 2002 that were relevant to the review topic were selected based on the PRISMA 2020 guidelines. The selected studies were analysed using six key aspects, which reflect typical components of experimental design such as participants’ characteristics, experiment and exposure duration, mode of light exposure, lighting and light interventions, experiment procedure, and data collection methods. The results indicated that many studies have limitations in terms of the accuracy and generalisability of findings in representing the entire elderly population due to issues with experimental design or control of the participants’ attendance. The results suggest that future studies should increase the duration of light intervention to around 21–35 days and the number of participants to around 14 and 47. The issues identified from the experimental designs of the selected studies provide valuable insights for establishing guidelines and recommendations for future studies.

1. Introduction

Poor sleep quality is common among older adults and has detrimental effects on their overall health and well-being as they age [1]. With the proportion of people over the age of 65 projected to reach 16% of the global population by 2050 [2], this demographic underscores the growing importance of addressing sleep-related issues in this age group. Given the critical role of sleep in cognitive function [3], emotion regulation [4], and physical health [5], understanding the factors that influence sleep quality in older adults is essential for their well-being. Among these factors, the relationship between light and sleep quality has become a prominent area of study, as light exposure directly affects circadian rhythms, which regulate melatonin production and the sleep–wake cycle [6,7]. However, indoor environments in healthcare and similar facilities typically offer insufficient circadian stimulation [8], and older adults often experience limited mobility that restricts their access to outdoor light [9], further compounding the issue.

The term ‘sleep quality’ is not clearly and consistently defined in the scientific literature, as its scope and definition often depend on the context or research field [10,11,12]. Relating to elderly people, it is commonly used in association with sleep disturbance [13], sleep disorders [14], sleep efficiency [15], and dementia [16]. In this paper, the term ‘sleep quality’ is used and defined to account for all possible relationships to the measurement of sleep, including total sleep time, sleep onset and offset, time in bed, number of awakenings, and wake-up time [17,18,19]. Sleep quality in older adults is influenced by a combination of physiological, neurological, and environmental factors, which can be broadly categorised into age-related changes in visual perception, circadian system responses to light, and built environment or lifestyle conditions.

The human visual system is similar in general but also unique to each individual. The ageing of the human body alters the visual system, which affects the visual perception of lighting [20]. In the elderly, age-related changes in the eye, such as lens yellowing, can cause the surrounding environment to appear yellow, which has a higher risk for causing sleep disturbances [21]. Reduced pupil size also allows less light to reach the retina, further reducing daily natural light exposure [22]. All of these changes in the visual system, in response to lighting conditions, ultimately lead to poorer sleep quality.

To understand the full impact of lighting on sleep, it is important to consider not only age-related visual changes, but also the role of specific photoreceptors in circadian regulation. While the rods and cones are primarily responsible for visual perception, the intrinsically photosensitive retinal ganglion cells (ipRGCs) in the eye play a crucial role in regulating the human circadian system [23]. IpRGCs are light-sensitive, particularly to the short wavelengths of the spectrum. They have been shown to influence the timing and quality of sleep by affecting melatonin secretion, which governs the sleep–wake cycle [24]. Short-wavelength light exposure in the morning also promotes alertness and cognitive performance [25], while during the evening, it suppresses melatonin production and delays sleep onset [26].

In addition to age-related physical changes affecting visual perception—which, in turn, lead to a decline in sleep—various environmental and lifestyle factors also affect sleep quality. For instance, poorly designed buildings that limit natural light, or mobility-related issues that can cause less access to natural daylight, can reduce sleep quality [27,28]. The decrease in natural light exposure also decreases vitamin D serum levels in the elderly, lowering their physical health and sleep quality [29]. After sunset, patients’ symptoms worsen, especially those with dementia living in healthcare centres, nursing homes, or similar institutions, potentially leading them to develop sundowning syndrome [30,31], which relates to the disturbances in circadian rhythms and sleep quality [32].

The design of light interventions, especially architectural indoor lighting technologies, can mitigate adverse sleep-related effects and promote better sleep quality in the elderly. Many experimental studies also indicate that electrical light has an association with sleep quality and circadian control of the elderly, especially those suffering from dementia [13,16,30,33,34]. Various types of electrical light exposure experiments, whether from ceiling lights [16], low-brightness light goggles [6], or high-brightness lightboxes [35], all indicate significant positive effects on the sleep quality of elderly people.

However, despite these promising findings, the wide variation in experimental protocols, light sources, intensity levels, exposure durations, and participant characteristics limits the comparability and generalizability of the results. This heterogeneity presents a significant research gap: the field currently lacks a unified, well-established framework and consistent methodological standards to guide future investigations. Without clearer experimental parameters and reporting criteria, it is difficult to draw robust conclusions or develop evidence-based lighting interventions tailored to the needs of older populations. Therefore, setting clear criteria is critical. The following research questions should be addressed to enhance understanding: What is the current state of knowledge on the effects of light on sleep quality in the elderly, and what are the gaps and weaknesses in the experimental designs of existing studies?

The primary aim of this literature review is to answer these questions, with a focus on the sleep quality among older adults living in healthcare or similar institutional settings. While the existing literature has explored the general relationship between light and sleep, there has been no systematic synthesis specifically examining experimental design parameters and their applicability to institutional settings. This review contributes to the existing body of literature by providing a structured comparison of study designs, lighting system characteristics, and measurement methods. In doing so, it aims to establish a foundation for developing more consistent and effective experimental protocols, ultimately supporting future interventions to optimise indoor lighting for the elderly.

2. Methods

2.1. Data Sources

The search for relevant existing studies was conducted within four databases, namely Google Scholar, Semantic Scholar, Lens.Org, and Scopus. The search terms used were a combination of ‘light’ or ‘lighting’ with ‘sleep’ in the title of the publication, while having ‘sleep quality’ with ‘elder’ or ‘senior’ or ‘old’ within the titles, abstract, or keywords of the publication. The search string used in Google Scholar was (“light” OR “lighting”) AND “sleep” [title], “sleep quality” AND (“elder” OR “senior” OR “old”). The main search for suitable research publications was carried out in Google Scholar on the 24th of August 2024, with 355 resulting publications. Other databases were searched using the same search string. However, they provided fewer publication results than Google Scholar—7, 23, and 21, respectively. However, the results from all four databases were combined to undergo the selection process in order to prevent certain publications only being available on certain databases.

2.2. Study Selection

The total search results from all four databases consisted of 406 potential publications; all publications within the data set underwent the selection process using the PRISMA 2020 flow chart [36]. Publications were included according to the inclusion criteria given below:

First phase—Records Screening

• Only peer-reviewed journal articles are included to focus on the latest research findings.
• Since ipRGCs were established in 2002 by David Berson and serve an essential role in controlling and changing the human circadian system [37], only studies published after 2002 are included.
• Studies solely on light medication or light therapy are excluded, as this review focuses on architectural lighting aspects instead of medical lighting.

Second phase—Reports Screening

• Only studies reporting empirical data on the effects of light on elderly people’s sleep quality were considered.
• Studies with all experimental designs are included.
• Studies without light interventions or controlled light exposure are considered, as long as the measurement of light exposure with analysis on the effects of light on sleep quality or factors relating to the sleep quality of elderly people is included.
• Only studies with statistically significant results are included to focus on proven relationships between sleep quality and light for elderly people. Although it may limit generalisability and introduce potential bias, this will increase the quality of results by focusing on interventions with demonstrated effectiveness.

The search results from the four databases were obtained using the publication searching software ‘Publish or Perish 8’ [38]; they were then imported into the publication sorting software ‘Zotero 7.0.22’ [39] for screening. The combined data set of 406 publications from four databases was selected and screened according to eight main selection criteria (Figure 1). First, the data set was sorted, where duplicates of the same publications were removed (26 cases); none of the publications were marked as ineligible by the publication searching software (0 cases) or removed for other reasons (0 cases), such as no given URL from the software. Second, a total of 380 publications underwent records screening, with 91 publications being excluded, where three selection criteria were applied to the titles and abstracts of the publications. The majority of the excluded publications in this stage were omitted due to their publication date being before the year 2002 (48 cases); some were excluded as the publication did not provide an author (8 cases), or the publications were focused on medical therapy rather than general lighting (28 cases). The remaining 289 publications were all successfully retrieved. Next, they were read in full as part of the reports screening, with 270 publications being excluded using an additional four selection criteria. The majority of the excluded publications in this stage had no direct relation to the elderly or healthcare centres (242 cases). Most were focused on experiments with animals (i.e., rats), and some publications did not include any experiments (15 cases) or significant findings from their results in their experiments (11 cases); some publications were retrospective studies, where all the data used in the analysis were obtained from public or national databases (2 cases). A total of 19 publications passed the entire selection process, and they are further analysed in this literature review.

2.3. Light Therapy Selection

Two publications associated with light therapy were still included and passed the selection process [35,40]. This is because these light-therapy-focused studies are difficult to differentiate from general lighting-focused studies, mainly when those studies also focus on elderly people and sleep quality [41,42]. Many therapy-focused studies used the word “therapy” in their titles [41,42,43,44] but were excluded since their methods and results provided no insights or suggestions into an architectural lighting setting. Some studies had an association with light therapy or used other similar words, such as “light treatment” [35] or “bright light exposure” [14,40]; these studies were included when they took the architectural light setting into consideration, or if their use of light exposure techniques could be further developed or provided suggestions for architectural lighting products or systems.

2.4. VOSviewer Key Word Analysis

Overall, the 19 publications provide a range of different experimental methods examining the possible effects of light on the sleep quality of elderly people. These were further analysed against the total 380 publications—before the criteria selection process—on their diversity and inclusion. VOSviewer 1.6.20 [45] was used to generate network maps identifying thematic trends, starting with keyword analysis.

The main keywords before the selection process (Figure 2) consisted of ‘questionnaire’, ‘intensity’, ‘wake cycle’, ‘phase’, and ‘system’, and the main keywords after the selection process consisted of ‘light’, ‘sleep efficiency’, ‘architectural lighting design’, and ‘light exposure’. This shows that the selection process successfully eliminated publications that did not fit into the topic of this paper, focusing more on light, sleep, and lighting design. The network map counted through the selected 19 publications demonstrates the limitation of knowledge between sleep efficiency and light exposure. This depicts the research gap between light and sleep in the elderly in current studies. The keyword ‘light’ included in the 19 publications is linked to ‘blue light’, reflecting the inclined usage of blue light in experiments or analysis in existing publications. This results in more coverage on the short wavelengths, suggesting a possible missing understanding on longer wavelengths above 495 nanometres or shorter wavelengths below 400 nanometres in the visible spectrum, highlighting another key research gap where future studies need to have broader light wavelength coverage in their experiment. The keyword ‘architectural lighting design’ is the odd one out in the network map—it has no connection with other keywords, depicting the need for research to link interior lighting design to sleep efficiency and light exposure.

2.5. VOSviewer Bibliographic Analysis

To gain insights into the structure of the research field, bibliometric analysis using VOSviewer 1.6.20 [45] was conducted on 380 publications. This analysis aimed to map the broader scholarly network, including key institutions, collaborative clusters, and thematic groupings related to light and sleep in older adults.

The resulting co-authorship and co-citation networks (Figure 3) revealed a fragmented landscape with few strongly interconnected clusters, indicating limited collaboration across research groups. While certain contributors (e.g., Figueiro, M. G.) appeared frequently, institutional-level collaboration and methodological standardisation remained weak. These findings align with the gaps identified in the systematic review, highlighting the need for greater coherence and integration in future studies, particularly regarding experimental design and reporting standards. To ensure comprehensiveness, a proportion of Figueiro’s publications [6,19,33,46,47] was included in the final analysis. This inclusion strategy will strengthen the analytical framework and also connect to findings from the previous Section 2.4, where Figueiro’s research emerged as being closely aligned with the aim of this paper.

3. Results

3.1. Methodology Analysis of 19 Selected Publications

3.1.1. Publications Without Baseline Measurement

The 19 publications that passed the selection criteria deliver various experimental methods, light exposure types, and light interventions. They include the use of natural light or electrical light exposure, exposure time during the morning or at night, and high or low CCT or illuminance (Appendix A).

The baseline measurement in an experiment is essential to provide reference data for comparison, especially in minimising the effects of individual differences, further strengthening the experiment’s validity. Seven out of the nineteen publications [6,13,15,16,21,48,49] did not include an outlined baseline measurement, but three of the publications included two light interventions for comparison [6,13,16]. The three studies with two light interventions all found that one treatment group showed better sleep quality than the other. Additionally, none of the studies included a control group for comparison with participants’ standard daily lighting. Figueiro’s experiment in 2009 [6] only included participants with ‘normal’ and ‘early’ Munich Chrono Type Questionnaire results, excluding elderly participants with sleep onsets or sleep times later than 11:30 pm, as most elderly people have sleep disorders or a higher chances of sleep disorders which have a delayed sleep onset or sleep time [50]. This limits the applicability of the experiment in representing all elderly people regardless of whether they have early or late bedtimes.

Liu’s experiment [13] indicated the baseline measurement; however, they did not specify its time duration and data collection period, leaving it unclear as to whether they are consistent for a controlled accurate comparison. In Liu’s experiment [13], participants in each treatment group were collectively exposed to the designed light scheme. This group exposure could introduce potential bias due to psychological factors such as peer pressure. Despite lacking a baseline or control group, Dal’s experiment [16] did not exclude participants taking medications, nor did it account for intake consistency, which the study acknowledged as potentially affecting the results. Furthermore, Dal’s experiment [16] has six cycles within its experiment procedure, while only four out of thirteen participants completed more than two cycles, and only one participant completed the entire six cycles. The results were analysed separately and combined with other participants’ data, regardless of their completed cycle times, which could have reduced accuracy due to the different experiment durations for each participant. The analysis and findings from the single participant who completed the entire experiment provide a limited representation of the broader population of people with dementia.

3.1.2. Publications with Light Intervention

The other four publications that do not include an outlined baseline measurement also lack specifically designed light interventions [15,21,48,49]. Within the four publications, the studies by Wei [48] and Kessel [21] focus on blue-light interaction with human eye lenses; it can be assumed that their baseline measurement is the first set of data measurements, although it is not explicitly stated in their study. Without a control group, the results of the experiment are limited in accuracy with the possibility of bias [51], as all participants in Wei’s experiment [48] were aware of the lens implantation. The resulting improvement in sleep quality may be altered by participants’ psychological expectations associated with the usage of the lens implantation. Furthermore, relying solely on the Pittsburgh Sleep Quality Index (PSQI) for data collection, without the use of additional objective measurement tools, may compromise the accuracy of the results due to the subjective nature of self-reported responses. Kessel’s experiment, on the other hand [21], had more participants than Wei’s experiment [48], with a total of 970 participants. The high number of participants reduces the possibility of biases in the results, and the combined use of questionnaires and photographic examination further enhances the reliability of the results. The other two publications [15,49] focused on the relationship between light and sleep quality without controlling the participants’ light exposure—participants were allowed to be exposed to both electrical lights and natural daylight. Since both light exposure and sleep quality are measured, a baseline measurement is not required for the impact evaluation of light and sleep quality. The field study conducted by Aarts [15] had a longer duration, spanning over one year, and thus providing a broader data range to examine over different seasons; however, the long time span limits the number of participants able to follow the full experiment. Data from Aart’s experiment [15] are also not consistent with each participant in terms of measurement dates and duration, and participant results were analysed together regardless of their experiment duration, which compromises the accuracy of the results. Lok’s experiment [49], on the other hand, had a shorter experiment duration of one week, which covered less data with the possibility of seasonal effects on participant results. This can be overlooked by the high number of participants (887), but a one-week experiment that is inclusive to natural daylight is not able to incorporate all possible weathers and seasons; this can have psychological effects [52], further impacting sleep quality [53].

3.1.3. Publications with Baseline Measurement

The twelve remaining publications all have a baseline measurement consisting of two light interventions [17,18,33,35] and one light intervention [14,19,40,46,47,54,55,56]. However, most publications indicate that there are missing or discarded results or data from participants; this could be due to the adherence of elderly participants to the entire experiment [14,19,40,46,47,54,56], or the limited time until the experiment deadline [46]. Elderly participants tend to lose interest in continuing during the process of the experiment [14,17,18,19,35,40,56]; they also tend to have health-related difficulties, such as undergoing surgery or not feeling comfortable [14,56], or withdraw due to changes from their initial eligibility where they are no longer able to meet the criteria for involvement [14,35,54]. Figueiro’s 2008 experiment [46] indicates that, due to experimental deadline, the final set of data intended to measure the time it takes for participants’ circadian rhythms to return to baseline after light exposure was collected before, rather than after, light exposure. As a result, this data set did not adhere to the experimental protocol and was subsequently discarded—though, if collected as scheduled, it might have provided valuable findings. Palmer’s experiment [17], on the other hand, had limited control of participants’ exposure to light, as was expected in the initial experiment method. In this experiment, light exposure came from a movable luminaire; the participants were expected to sit near it at a certain distance. However, it was reported that participants tended to move around or reposition the luminaire, which affected the amount of light falling into their eyes, thus reducing the accuracy of the data and results.

3.1.4. Other Limitations

The informed limitations of the experiments have also been analysed. Figueiro’s experiment [46] included people from the lighting research centre as participants; this bring the validity of the experiment into question as the participants were researchers with background knowledge of the lighting field and possibly the experiment. McCurry’s experiment [14] included an education group where the participants were educated on light exposure and sleep quality, such as their potential triggers of nighttime awakenings and developing individual sleep plans while still undergoing the light exposure experiment. The comparison of the education group to other treatment groups and control group neglects the possible psychological effects of the results, bringing uncertainty to the accuracy and validity of the results.

The 19 selected publications establish a foundation for examining light and sleep quality interactions in elderly populations within architectural contexts, though the analysis of six key parameters and variables remains essential to advance our thoroughness of understanding.

3.2. Six Key Parameter Analysis of 19 Selected Publications

The key parameters and variables examined within the selected 19 publications include the appropriate number and demographics of experiment participants, the lighting exposure time compared to the total experiment duration, the types of lighting exposure or interventions in relation to indoor luminaires, the recommended illuminance and Correlated Colour Temperature (CCT) for light exposure, the experimental procedure, including preliminary or baseline measurements and washout periods, and, lastly, suitable methods and instruments used for data measurement and collection.

3.2.1. Participant

The two main aspects considered for participants are the number of participants and the participants’ ages. Out of the 19 publications examined (Figure 4), only two had more than 200 participants in their experiment [21,49]. Of these, one was focused on eye lenses, while the other was focused on natural light. Neither publication included any light interventions nor controlled light exposure. Therefore, their use of participants is redundant and will not be taken into account if the suggested future study does include light intervention or designed light exposure. The other 17 publications all have participant numbers under 150, with a mean of 33 participants. The expected ages of participants range from 50 to 100 across 18 publications, with only Kessel’s study [21] including age groups below 50 (from 30 to 60); this is also the study with the highest number of participants in the experiment. Kessel’s study [21] will be excluded from the participant consideration parameter as both the number of participants and the age groups are outliers from the other 18 publications.

3.2.2. Experiment and Duration of Light Exposure

The experiment and exposure duration results are scattered, with no apparent consistency individually or relations in between (Figure 5). Some publications that did not have a light intervention [21] or light exposure treatment [21,48,49] are shown as empty in Figure 5. Excluding those publications, the duration of the experiment ranges from 7 days to 365 days (1 year), and the duration of exposure ranges from 0.5 h to 4032 h (168 days). Furthermore, as indicated before regarding missing or discarded results or data, publications with long experiment durations or exposure [14,15,19,54] all have issues with participants’ adherence [14,19,54], discontinuation due to lack of interest [14,19], or changes to the original protocol due to time constraints [15].

3.2.3. Mode of Light Exposure

In total, four different types of light exposure were used within the 19 publications: 9 publications designed an indoor room light exposure procedure, 2 publications had natural daylight exposure, 4 had light box exposure, 1 had light goggles, and 3 had no light interventions (Figure 6). Room light exposure had the highest percentage of usage due to its close relation in observing the relationship between electric lighting and sleep quality; other modes of light exposure all had issues in their applicability when providing suggestions for use in an architectural setting.

3.2.4. Lighting and Light Interventions

Illuminance and CCT are the two variables mentioned the most out of the 19 publications, especially where the experiment used electrical lights for the light exposure of the participants. The 19 publications provide a diverse range of light interventions, with some having two illuminance or two CCTs, and some with changing illuminance or CCTs within a set range throughout the experiment (Figure 7). Six publications had no data due to their experiment not including any light intervention [21,48,56], only including daylight exposure [15,49], or not providing the relevant data [19].

3.2.5. Experiment Procedure

Most experiment procedures have three sections: baseline, washout, and light interventions (Figure 8). Most publications only include one baseline measurement with one light intervention experiment (9 cases). Three publications did not include a baseline measurement but had at least two different lighting conditions or light interventions [8,13,16]. Four publications had no detailed experiment procedure since they did not provide a timeline or duration of each section of their experiment. Only two publications included all sections with a baseline measurement before each of the two different light interventions, with a washout period in between [18,33], providing a larger quantity of data collection. This increased accuracy and provided a more comprehensive range of data comparison between baseline and light interventions, or between the light interventions themselves, thus increasing the reliability and validity of the results.

3.2.6. Method of Data Collection

There were 24 different types of method used for data collection across the 19 publications (Figure 9). Most methods were used only once, with 22 types of method used three or fewer times. Overall, seven data collection methods were objective, while the remaining 15 types were subjective. The two methods used the most were the Pittsburgh Sleep Quality Index (PSQI) and actigraphy. A further analysis of these two methods will be carried out in the following sections.

3.3. p-Value Analysis of the 19 Selected Publications

3.3.1. PSQI and p-Value

Nine publications included the PSQI as one of their data-measuring methods (

Table 1. The p-value of the PSQI results from each publication is indicated. The # represents the publication number and is shown in Appendix A. NS = No Significance.

Publication Number	PSQI
#4	0.037
#6	NS
#7	<0.001
#9	NS
#13	<0.001 *
#14	0.026 *
#16	0.01
#17	<0.001
#19	0.013

* If multiple p-values exist, only the highest significant p-value is taken.

), and seven out of nine publications indicated a statistically significant finding from the PSQI results. This suggests that the PSQI is highly efficient in identifying and discovering potential correlations between light and sleep quality. Two publications [19,55] indicated multiple uses of the PSQI from multiple data measurements. They supplied a broader understanding of the effect of time on the relationship between light and sleep quality; however, only the highest significant p-values were shown among all questions.

3.3.2. Actigraphy and p-Value

Actigraphy has the highest usage at 12 cases across the 19 publications (

Table 2. The p-value of the actigraphy results from each publication is indicated, and the significance of the p-value results is highlighted in the background. The # represents the publication number and is shown in Appendix A. NS = No Significance, TIB = Time in Bed, TST = Total Sleep Time, IS = Interdaily Stability, IV = Intradaily Variability, SE = Sleep Efficiency, SOL = Sleep Onset Latency, WASO = Wake After Sleep Onset.

Publication Number	TIB	TST	Light/Dark Ratio	IS	IV	Actual Sleep Time	SE	SOL	Daytime Naps	WASO
#5	<0.05	<0.05					NS			NS
#6			0.04	NS	NS		NS
#7				0.08	0.049	0.3	0.15	0.08	0.1
#8		0.429					0.113	0.592		0.186
#9
#11	0.28	0.63					0.61		0.89
#12		0.13		0.01	0.04		0.07			0.17
#13				0.28	0.92	0.86
#14		0.62			0.67		0.187	0.48		0.453
#17					0.97		0.09
#18					0.001
#19		0.001					0.001	0.005

); it measures the movement and activity of the wearer. However, only half of the publications (six cases) indicated statistically significant findings under different variables of the actigraphy results. No consistency was found in the variables of the actigraphy results between the publications; each publication provided a different coverage of the actigraphy results. Therefore, variables of the results obtained using actigraphy that were reported in certain publications but not in others are left blank. The two variables with the highest indication rate are Sleep Efficiency (nine cases) and Intradaily Variability (seven cases), with Intradaily Variability having the highest statistically significant indication rate (three cases). This brings greater insight into future light and sleep quality studies including those two variables in the possible usage of actigraphy. One of the publications [15] specified the usage of actigraphy in data collection and the significant relationship between light and sleep quality. However, it did not specify which variables derived from actigraphy were measured, especially when the actigraphy results were combined with the participants’ sleep logs; it is unclear which data were used that led to the significant findings of this experiment.

4. Discussion

This discussion integrates six key parameters identified in the results section, along with relevant inclusion criteria, experimental limitations, and methodological inconsistencies in the existing studies; this collectively informs a set of recommendations for future experimental studies and designs. The analysis emphasises the need for standardising methodology, data reporting, and participant control.

4.1. Participant

A summary (Appendix A) is provided on all six parameters (Section 3.2.1, Section 3.2.2, Section 3.2.3, Section 3.2.4, Section 3.2.5 and Section 3.2.6) indicated in the “Selected Publication Analysis” section of this literature review. From the 19 publications, the average number of participants included in the experiments, excluding the two outlier publications [21,49], is 34.7 (1 dp) participants, and the average age of the participants is 71.6 (1 dp) years old. Therefore, the initial suggested number of participants should be around 35 for future studies (Figure 10) to ensure sufficient comparison of results between participants and to find relationships between light and sleep quality. Moreover, as mentioned previously regarding missing and disregarding results from the adherence of participants, or changes in participants, initial information, the initial suggested number of participants will be further increased to 50, which sits within the suggested range from the power analysis. This also allows for a possible 30% unavoidable reduction in results as was seen in most publications [14,17,18,19,35,40,47,54,57]. Based on the quartile calculations (Q1 = 14, Q2 = 25, Q3 = 47), the suggested number of participants for future studies could range from 14 to 47.

The age of participants mostly fell within the range of 60–85 (Figure 10). As the focus of this study is the effect of light on the sleep quality of elderly people, where the traditional age consideration for elderly is 65 years and older [57], people older than 85 should also not be excluded as long as they are able and willing to participant in the experiment. Therefore, the initially suggested range of 60 to 85 years old will be changed to 65 years and older.

4.2. Mode of Light Exposure, Experiment, and Duration of Light Exposure

The duration of the experiment is associated with all other parameters and variables, including the choice of inclusion of a possible washout session or control group, the time needed for each light intervention, and the set aim of the study. For this literature review, the duration of the experiment should account for studying the effects of light on sleep quality in people aged 65 and older in an architectural setting, using exposure to room/electrical lighting and/or natural daylight.

Out of the 11 publications that included either room light or natural light as the mode of exposure (

Table 3. Edit of Appendix A, where only publications with room light or natural light mode of exposure are included, showing only parameters of the number of participants in the control group, experiment duration, number of light interventions, duration of light exposure, baseline measurements, washout time, and mode of exposure. The # represents the publication number and is shown in Appendix A.

Publication Number	Number of Participants in Control Group	Duration (Days)	Baseline Measurements (B)	Number of Lighting Intervention (LI)	Duration of Light Exposure	Washout Time (W)	Mode of Exposure
#2	11	56	n/a	2	1 h	n/a	Room Lighting
#3	n/a	126	n/a	2	21 days	n/a	Room Lighting
#6	n/a	90	14 days	1	28 days	7 days	Room Lighting
#7	n/a	98	7 days	2	28 days	28 days	Room Lighting
#8	n/a	36	7 days	2	2–3 h	n/a	Room Lighting
#9	n/a	<365	n/a	n/a	5 days	n/a	Natural Light
#12	n/a	210	70 days	1	98 days	n/a	Room Lighting
#13	n/a	175	7 days	1	168 days	n/a	Room Lighting
#16	n/a	63	7 days	1	28 days	n/a	Room Lighting
#18	n/a	7	n/a	n/a	n/a	n/a	Natural Light
#19	n/a	41	9 days and 7 days	2	9 days	7 days	Room Lighting

), 2 publications did not have any room/electric light interventions [15,49]. Although they involved natural light interventions and human experiments, they are excluded from further analysis as the aim of this review focuses more on electrical light interventions. Other publications that used room light intervention (Nine cases) did not account for the natural daylight exposure of participants in the morning. Most publications indicated the measuring of light received by the participants from the usage of actigraphy [17,18,19,33,46,49,54], but did not identify the amount of light received from natural daylight or indoor electrical light interventions. This limits the effectiveness of the electrical light interventions on the participants’ sleep quality and further highlights the importance of including baseline measurements, especially in excluding natural daylight from electrical illuminance.

Seasons also have a psychological effect on people’s moods [52]. An experiment duration that spans more than three months or crosses multiple seasons may limit the accuracy of results when compared to baseline measurements taken in a different season [15,19,54]. However, the inclusion of more than one light intervention with the results compared between the light interventions will eliminate such issues, as long as the environmental conditions are consistent with the light interventions running in parallel [16].

Among the five publications that used room light as the mode of exposure with a baseline measurement and experiment duration spanning within a single season [17,18,33,46,47], Palmer’s experiment [17] negatively stands out, as the duration of light exposure was measured in hours rather than days. It had less control of the overall light received by the participants as the light intervention only happened before the participants’ bedtimes. The participants were required to sit next to a dim light luminaire for two to three hours, regardless of their activity during the light intervention; unlike the other four publications, the light intervention ran for the entire day from the participant’s wake time to sleep time. Therefore, Palmer’s experiment is excluded, while the remaining four publications have a median light exposure duration of 28 days, baseline measurement of 7 days, and a washout time of seven days.

The suggested experiment duration for this review should include a baseline measurement around seven days—in order to differentiate the illuminance received by the participants from natural daylight and light intervention—along with a washout time of around seven days if more than one light intervention is included—around 21–35 days for each light intervention—and a total experiment duration that does not exceed more than three months or run across more than one season.

4.3. Lighting and Light Interventions

Moving to lighting and light interventions, out of the 19 publications, three included a control group, seven included two light interventions, and 8 included one light intervention. Except for one publication [13], the other studies that included two light interventions without a control group directly compared the two interventions with each other. On the other hand, publications that only have one light intervention with no control group would have a baseline measurement for light exposure comparison. Therefore, all publications with light interventions are valid and reliable. In contrast, publications that relied solely on observation [15,21,48,49] are not considered for their experimental methods due to the absence of light interventions. However, their data collection methods and modes of exposure are still included.

Across the 19 publications (Figure 11), most illuminance used in experiments falls within the range of 30–500 lux; CCT, on the other hand, falls within the range of 2700–6500 Kelvin. Shishegar’s experiment [18] included a comparison between a dynamic CCT and a constant stable CCT, examining their effect on the sleep quality of elderly people; this can be implemented in the consideration of illuminance, where dynamic illuminance is compared to constant stable illuminance. Most publications using dynamic illuminance also utilised dynamic CCT [18,54]. Since most publications focus on comparing illuminance, the CCT is suggested to be the same between different light interventions.

4.4. Method of Data Collection

The methods of data collection should include both objective and subjective measurements to cover a broader range of data analysis; this will increase the reliability and accuracy of the results [58], especially in addressing sleep quality and sleep disorders [59]. Within the objective methods of data collection (Figure 7), actigraphy has the highest rate of usage, which will be one of the suggested methods for future studies. Regarding the suggestion of including and measuring natural daylight, devices such as portable spectroradiometers (similar to daysimeters) should be included to accurately measure the light received by the participants on top of the actigraphy. It was indicated in some publications that participants accidentally blocked the actigraphy on their wrist, thus affecting the accuracy when measuring light exposure [15,49]. On the other hand, regarding the subjective data collection methods, PSQI has the highest rate of usage. Some publications also used the PSQI alongside the CSDD (Cornell Scale for Depression in Dementia), CMAI (Cohen–Mansfield Agitation Inventory) and MDS-ADL (Minimum Data Set-Activities of Daily Living) [16,30,44]. The four types of questionnaires have all been suggested to be included in future studies, along with the use of a sleep log or sleep diary. Logs or diaries maintained by the participants or caretakers can be used as a complement to the results of the actigraphy in accurately identifying the sleep schedule of the participants [15,17,35].

4.5. Cohen’s d on PSQI of 19 Selected Publications

An effect size analysis was conducted using the Campbell Collaboration’s Cohen’s d calculator [60] on the 19 selected publications, focusing on the p-values associated with the PSQI outcomes. Of these 19 publications, only 9 reported p-values for the PSQI results. The PSQI was chosen because it was used in a similar way across all nine studies, with consistent variables. Other methods, like actigraphy, include many outcome measures that are inconsistently reported across the studies. Additionally, these methods involve varying independent variables, which complicates the calculation and comparison of effect sizes in a reliable and standardised manner.

Overall, the calculated effect sizes for the nine studies indicate that when the p-values were statistically significant, the effect sizes were high, ranging from 0.6824 to 21.25. Conversely, when the p-values were not statistically significant, the corresponding effect sizes were low, ranging from 0.1545 to 0.46.

Shishegar’s study [18] compared the effects of constant CCT lighting (2700 Kelvin) and dynamic CCT lighting, ranging from 2700 Kelvin to 6500 Kelvin, on the sleep quality of older adults, with both interventions incorporating dynamic illuminance from 100 lux to 500 lux. For the constant 2700 Kelvin lighting condition, the p-value was not statistically significant (p = 0.727), and the effect size was low (d = 0.1545). In contrast, the dynamic CCT lighting condition yielded a statistically significant p-value (p = 0.018) and a high effect size (d = 1.126). These findings suggest that a dynamic CCT light intervention has a better effect size and provides more significant results when using PSQI than a constant 2700 Kelvin light intervention.

When comparing Shishegar’s results [18] with those of Figueiro’s experiment [33], both of which investigated constant CCT lighting at 2700 Kelvin, the principal difference lay in the illuminance pattern, with Figueiro’s experiment employing constant illuminance of 150 lux and Shishegar’s experiment employing dynamic illuminance ranging from 100 lux to 500 lux. Regardless of whether the light intervention has constant or dynamic illuminance, the effect sizes remain low, and the p-values are not significant for the PSQI. For example, in Figueiro’s experiment, effect sizes remained low at 0.46, and the p-values for PSQI outcomes were not statistically significant at 0.13. However, Figueiro’s experiment also included a set of light interventions with constant CCT at 6000 Kelvin with constant illuminance at 350 lux, this light intervention resulted in a high effect size at 21.25 and p-values < 0.001, suggesting possibilities that brighter and cooler light may provide higher effect size and better PSQI results on testing sleep quality of the elderly.

4.6. Participant Inclusion Criteria

Key factors or measurable parameters from the 19 publications that fell outside the six main parameters include the initial criteria for participant inclusion, allowance for medication, control of participant movement, and the measurement of light interventions.

All publications indicated clear initial criteria for the inclusion of participants in their experiment. Depending on the research aim, the criteria ranged from the exclusion of people with a major illness [6,33], physical disorder [16], dementia [18], a neurological disorder [16,56], or sleep disorder [56], to the inclusion of people with dementia [13,14,19,33,35,47,55], sleep disorders [21,33,46], or insomnia [21,40]. Relating to the aim of this review, the inclusion of participants with dementia is suggested, as the risk of dementia is related to ageing [61]. In Shishegar’s experiment [18] the Montreal Cognitive Assessment (MoCA) is indicated and used to assess the levels of dementia of the participants; participants with moderate to major dementia are excluded. Furthermore, in Figueiro’s experiment [6], the sleep types of the participants were assessed using the Munich Chrono Type Questionnaire to determine and exclude cases where melatonin would not synthesise with the light intervention time due to the participant’s sleep schedule. This approach is recommended for future studies where the light intervention period extends into the night.

4.7. Allowance of Medication

Medication is another factor indicated in some publications [6,15,18]. In Figueiro’s experiment [6], participants taking medications were excluded from the experiment, while in Dal’s [16] and Aarts’s [15] experiments, they were included. Medicine is essential for many elderly people, especially those with health-related issues [62], and it is not ethical under experimental circumstances to prevent participants from taking medications. However, it is suggested in future studies to exclude participants with irregular medication consumption, but include participants with consistent medication consumption as their results are still valid and reliable.

4.8. Participant Control and Measuring of Light Interventions

As mentioned before regarding Palmer’s experiment [17], where the light exposure of participants during the experiment was limited in terms of illuminance consistency and accuracy, the movement of participants and a moveable light source can change illuminance exposure for the participants. Although it is not ethical to limit the movement and activity of the participants, it is suggested that the light be measured at the equivalent position of the participant’s cornea throughout the experiment to reduce errors in illuminance accuracy. Furthermore, the indication and measurements of the light intervention used in all 19 experiments are identified through illuminance and are reported in varying forms: solely in terms of horizontal illuminance [13], vertical illuminance [16], or illuminance at the corneal plane of the participant’s eye [6,14,18,33,35,46,47,54,62]. This again depicts the lack of consistency in the existing studies and reinforces the importance of a clear framework for future experiments in this field. One publication [17] did not specify the illuminance measurement perspective at all, potentially leading to ambiguity regarding the experimental lighting conditions and complicating readers’ interpretation of the experimental environment. Among the 19 publications, only 5 provided a spectral power distribution (SPD) curve of the light used [16,17,18,47,55]. The SPD of the light intervention should be included in future studies to allow readers to better understand the experiment [63]. This further highlights the need to provide Melanopic Equivalent Daylight Illuminance (MEDI) values of the light with SPD, as it is a critical methodological gap in current studies. This will allow readers to also understand the non-visual effects of the light used; its importance was additionally mentioned in the CIE recommendations [63].

5. Conclusions

Ageing is accompanied by many changes in circadian rhythms, from the secretion of melatonin levels to the yellowing of eye lenses, thus reducing the intake of light. These factors contribute to decreased sleep quality, including changes in sleep patterns, total sleep time, sleep efficiency, and difficulty falling asleep. In this regard, visual perception of light affects the sleep quality of elderly people in many ways, especially those living in healthcare centres or similar facilities. This review of existing publications demonstrates the positive effects of altering and controlling the sleep quality of elderly people through light exposure from manipulated light interventions, ranging from dim indoor electrical lights and portable luminaires to dynamic lights and high-illuminance lightboxes. Most studies are limited by their experimental methods or participant selection, making it difficult to provide accurate results when examining the relationship between lighting and sleep quality, or in representing the broader elderly population. Despite the review having a limitation on the number of publications available in this field, it still offers suggestions and results that may be used for future experimental studies.

Our analysis and review of the publications mentioned provides insights and considerations leading to suggestions for experiment design in future studies (Appendix B). The suggested initial number of participants is around 14 to 47, with the suggested age of participants being equal to or more than 65 years old; inclusion criteria should not exclude participants with dementia or those who take medication, as well as participants with certain sleep types. The suggested experiment procedure should include possible baseline measurements around 7 days, more than one light intervention running for around 21 to 35 days, washout sessions between light interventions running for around 4 to 10 days, and a total experiment duration not exceeding three months, or spanning different seasons. The suggested types of light interventions are room lighting exposure with dynamic and/or constant illuminance ranging between 30 and 500 lux, and dynamic CCT ranging between 2700 and 6500 Kelvin. The suggested data collection methods include objective data measuring using actigraphy and a daysimeter, with subjective measuring using PSQI, CSDD, CMAI, and MDS-ADL, along with a sleep log or sleep diary.

Five key elements for data reporting in future experimental studies in this field are proposed: (1) detailed reporting of the SPD of the light used in the intervention, (2) the MEDI, (3) the geometry of light measurement—specifying whether measurements were taken at the horizontal, vertical, or corneal plane of the participant’s eye, (4) disclosure of any medications taken by participants, and (5) a clear statement of the inclusion criteria for participant selection.

These recommendations, drawn from existing studies on the relationship between light and sleep quality in the elderly, offer a potential framework for future experiments. Implementing them would help address the current lack of consistency and comparability across studies in this field.