This study had two main objectives: First, the MCTQ, a recently developed questionnaire for the quantitative assessment of chronotype, was validated (3.2. & 4.1.) and its structure was optimized towards a maximum amount of information with a minimum number of ques- tions (3.3. & 4.2.). Second, potential biological and social influences on the sleep-wake cycle were scrutenized in order to 'clean' chronotype from environmental bias (3.4. & 4.3.).
4.4.1. Improvement of the MCTQ
Questions revealed to be redundent by PCA can be removed from the questionnaire with- out losing much information. Only SOW/Fand SEW/F, are necessary to determine chronotype. A discrimination between BT and the time of switching off the light in order to fall asleep on one hand and SE and time of leaving bed on the other hand makes the MCTQ unambigeous, thus these questions are added. However, the new questions do not give more information but make clear that BT, switching off lights, and SO are different questions, as well as SE and getting out of bed.
The MCTQ also asks for subjective assessment of chronotype on a scale from 0 (extreme early type) to 6 (extreme late type). Unlike the distribution of MSFSc, the distribution of sub- jective chronotype does not resemble a normal distribution. Most people assess themselves as moderate late types (28%), probably due to early work times which most people regard as too early (Roenneberg et al, 2004; Appendix1, Fig.3). Subjective chronotype rating thus does not reflect actual sleeping behaviour. In contrast, MSFis a reliable estimate of actual sleep times assessed by sleep logs which in turn are reported to reliably represent DLMO (Martin & Eastman, 2002). Therefore, subjective chronotype rating will be not included in the improved version of the MCTQ.
DIPW/Fcould be a reference point for a decreased state of alertness during the day, con- trolled by the circadian and/or the homeostatic oscillator. As discussed before, DIPW/Fis highly influenced by social schedules, e.g. meal times. It is doubtful if DIPW/Freflects an in- dependent structure of chronotype or if it is rather an artefact of social constraints. There- fore, DIPW/Fwill not be considered in the improved version of the MCTQ.
For the calculation of MSFSc, the general ratio of work days and free days of 5:2 was used. Asking for actual number of work days per week allows an individual weighting of work days and free days. The improved MCTQ asks for the number of work days per week.
After removement of redundant questions and the addition of three more questions, the MCTQ comprises less than one page (Appendix 3) and can be filled out in 3 minutes. A reli-
Discussion
tal or at the general practitioner to improve medication or for employers to assess optimal work times for employees. Also, participants in the online chronotype survey could focus on the essential questions. Short questionnaires are always apprecitated by people and could improve the reliability of results.
4.4.2. Properties of chronotype
Following the two-oscillator model (see chapter 1.5.5.2.), the change of sleep and wakeful- ness is regulated by a circadian oscillator and a homeostatic or hourglass oscillator. In this model, the circadian oscillator mainly determines the position of sleep phase within the 24 hour light-dark cycle while maintainance of sleep is likely to be an interplay of both oscillators (Dijk & von Schantz, 2005; Dijk et al, 2000). Phase of entrainment and α/ρ (sleep duration) can be determined with the MCTQ. Phase could describe the behavioural output of the cir- cadian oscillator, α/ρ the output of the homeostatic or both oscillators. On the population level, both properties are statistically independent of each other.
The term chronotype referred to phase of entrainment and not to sleep duration in the pilot study of Roenneberg et al(2003a). However, both contribute to the variation of the daily change of sleep and wakefulness which involves many brain areas (Pace-Schott & Hobson, 2002). The genetic basis of chronotype (phase of entrainment) has been described in hu- mans and in non-human mammals (1.4.5.) and is assigned to the SCN (1.4.3.). Also, there is evidence that sleep duration (α/ρ) is genetically controlled (Franken et al, 2001). Non- clock genes and clock genes could be involved in the manifestation of sleep duration (Tafti & Franken, 2002; Naylor et al, 2000). The two oscillator model from Pittendrigh and Daan (1976b; 1.4.6.2.) might explain the adaptability to changes in day length (Daan et al, 2001). Two distinct oscillators could also explain the achievement of sleep duration (α/ρ). Different phase angles of oscillators, respectively to dawn and dusk, could cause both sleep phase and sleep duration (Fig. 4.1.).
12 18 24 6 12 12 18 24 6 12 12 18 24 6 12 12 18 24 6 12 12 18 24 6 12 12 18 24 6 12
Fig. 4.1.: A model of two os- cillators defining both sleep phase and sleep duration. Every oscillator has its own properties (e.g. tau, respon- sivness to entraining sig- nals,…) locking differently to dawn (morning oscillator, blue) and dusk (evening oscillator, red). Black bars represent sleep.
Discussion
Although an influence on sleep homeostasis has been shown in mice only for the Clock
gene (Naylor et al, 2000) and not for components of the hypothesized morning and evening oscillator (Per1/2 and Cry1/2, Daan et al, 2001), the general framework of sleep duration could be assigned to or at least influenced by the circadian clock.
4.4.3. Future directions for a proper chronotyping
This study showed the potential influence of biological and social factors. However, only associations and not causalities were obtained. Results should, therefore, be used to formu- late hypotheses and design specific experiments for single factors.
When considering the tendency of overestimation of extreme chronotypes, the MCTQ is a highly reliable instrument for the assessment of chronotype. In contrast to the Horne- Žstberg MEQ, the quantitative measure of chronotype can be corrected for biological and social factors. The improved version of the MCTQ can be used to determine chronotype un- der normal life conditions in a first round. Depending on the purpose of further investiga- tions, different chronotypes can be easily selected. Also, a quick an reliable assessment of chronotype is possible whereever it is needed (e.g. physicians, employers,…). Based on re- sults of this study and further experiments on biological and social factors, one should be able to break down chronotype into genetic causes by quantifying as many non-genetic influ- ences as possible. This can help to reduce false positive results and to improve the attempts at identification of new human clock genes.
Summary