Self-reported Sleep Duration, Sleep Quality, and Breast Cancer Risk in a Population-based Case-Control Study

Jennifer Girschik; Jane Heyworth; Lin Fritschi

Am J Epidemiol. 2013;177(4):316-327.

Abstract and Introduction

Abstract

Breast cancer is one of the most commonly diagnosed invasive cancers. Established risk factors account for only a small proportion of cases. Previous studies have found reductions in sleep duration and quality in the general population over time. There is evidence to suggest a link between poor sleep and an increased risk of breast cancer. In this study, we investigated the relationship between breast cancer and sleep duration and quality in Western Australian women. Data were obtained from a population-based case-control study conducted from 2009 to 2011. Participants completed a self-administered questionnaire that included questions on sleep. Odds ratios and 95% confidence intervals were calculated using unconditional logistic regression. Sensitivity analysis for potential selection and misclassification bias was also conducted. We found no association between self-reported sleep duration on workdays and risk of breast cancer (for <6 hours, odds ratio (OR) = 1.05 (95% CI: 0.82, 1.33); for 6–7 hours, OR = 0.96 (95% CI: 0.80, 1.16); and for >8 hours, OR = 1.10 (95% CI: 0.87, 1.39), compared with the reference category of 7–8 hours' sleep). In addition, we found no association between sleep duration on nonworkdays, subjective sleep quality, or combined duration and quality and risk of breast cancer. This study does not provide evidence to support an association between self-reported sleep duration or quality and the risk of breast cancer.

Introduction

Breast cancer is the most commonly diagnosed invasive cancer and the leading cause of cancer death among women, both in Australia and globally.^[1,2] While some risk factors, such as primary genetic mutations, reproductive history, weight gain, and alcohol consumption, are well established, they do not account for a significant proportion of breast cancer cases.^[2]

Both international and Australian studies have reported changes in sleep over time in developed countries. Some studies have suggested a reduction in habitual sleep duration of 13–36 minutes over the last 20–30 years,^[3–7] although reductions in sleep duration have not been consistently found.^[8,9] Studies of sleep quality have reported increases in the prevalence of subjective poor-quality sleep ranging from 5% to almost 40%.^[3,4,10]

While the immediate effects of poor sleep, such as tiredness, loss of concentration, and injuries, are well recognized,^[11] the chronic effects of poor sleep have not been extensively studied. However, there is some evidence to suggest a link between poor sleep and a range of long-term health effects, including increased risk of cancer.^[12–18]

Two plausible biological models have been proposed that would explain how poor sleep can directly influence the development of cancer: impaired immune function and metabolic pathways that lead to obesity.^[2,12,15,19] An additional indirect mechanism has also been proposed which suggests an increased risk of cancer due to altered melatonin release.^[20,21] Melatonin release is regulated by the light/dark cycle rather than by sleep per se, and previous studies have used sleep as a proxy for "exposure to darkness".^[22] All 3 proposed pathways have some evidence to support them, but the true process by which sleep might influence cancer risk is unknown.

Five studies have investigated the relationship between sleep duration and breast cancer and have shown mixed results.^[22–26] Kakizaki et al.^[23] reported a statistically significant trend of decreasing risk with increasing sleep duration, while McElroy et al.^[24] reported a statistically significant trend of increasing risk with increasing duration. The 3 remaining studies also found significant trends, but only after study populations were restricted or polytomous analysis was conducted.^[22,25,26] Only 1 of those studies also examined subjective sleep quality;^[22] no association with breast cancer was found.

Our aim in this study was to investigate the relationship between breast cancer and 3 domains of sleep (sleep duration on workdays, sleep duration on nonworkdays, and subjective sleep quality) in a population of Western Australian women.

1 2 3 4

Next Section

Abstract and Introduction
Materials and Methods
Results
Discussion
References

Table 1. Distribution of Demographic and Risk Factors Among Women With and Without Breast Cancer, Breast Cancer Environment and Employment Study, 2009–2011
	Controls (n = 1,695)		Cases (n = 1,133)
	No.	%	No.	%
Age group, years
<45	179	10.6	157	13.9
45–49	201	11.9	171	15.1
50–54	229	13.5	149	13.2
55–59	288	17.0	175	15.5
60–64	303	17.9	180	15.9
65–69	250	14.8	135	11.9
≥70	245	14.5	166	14.7
Country of birth
Australia/New Zealand	1,139	67.2	725	64.0
United Kingdom/Ireland	362	21.4	244	21.5
Europe	75	4.4	64	5.7
Asia	62	3.7	56	4.9
Other	57	3.4	44	3.9
No. of children
0	183	10.8	148	13.1
1	123	7.3	109	9.6
2	663	39.1	437	38.6
3	460	27.1	284	25.1
4	184	10.9	98	8.7
≥5	82	4.8	57	5.0
Age at first childbirth, years^a
<20	164	10.9	106	10.8
20–24	605	40.0	404	41.0
25–29	498	32.9	274	27.8
≥30	245	16.2	201	20.4
Breastfed children^a
No	145	9.6	116	11.8
Yes	1,367	90.4	869	88.2
Family history of breast cancer
No family history	1,210	71.4	688	60.7
Unknown family history	4	0.2	4	0.4
Some family history	357	21.1	283	25.0
Increased-risk family history	124	7.32	158	14.0
Menopausal status
Postmenopausal	1,279	75.5	782	69.0
Premenopausal	416	24.5	351	31.0
Use of hormone replacement therapy
None	999	58.9	711	62.8
Estrogen only	254	15.0	118	10.4
Progesterone only	72	4.3	27	2.4
Estrogen + progesterone	62	3.7	68	6.0
Mixture of treatments	150	8.9	110	9.7
Nonhormonal treatment	32	1.9	33	2.9
Unknown treatment	126	7.4	66	5.8
Highest level of education completed
Year 9, 10, or 11	592	34.9	401	35.4
Year 12	390	23.0	233	20.6
Trade/apprenticeship	414	24.4	247	21.8
University degree	299	17.6	252	22.2
Usual no. of alcoholic drinks per week
0	271	16.0	185	16.3
<1	211	12.5	129	11.4
1–3	395	23.3	265	23.4
4–6	227	13.4	142	12.5
7–9	138	8.1	114	10.1
≥10	453	26.7	298	26.3
Weight status at age 30 years, compared with others of the same age and height
Weighed a lot less	80	4.7	82	7.2
Weighed a little less	458	27.0	312	27.5
Weighed the same	793	46.8	528	46.6
Weighed a little more	296	17.5	169	14.9
Weighed a lot more	68	4.0	42	3.7

^aLimited to participants who had 1 or more children.

Table 2. Age-adjusted and Multivariate-adjusted Odds Ratios for the Association Between Sleep and Breast Cancer Risk, Breast Cancer Environment and Employment Study, 2009–2011
	Controls		Cases		OR	95% CI	Adjusted OR^a	95% CI
	No.	%	No.	%	OR	95% CI	Adjusted OR^a	95% CI
Duration of sleep on workdays, hours (n = 2,828)
<6	238	14.0	163	14.4	1.04	0.83, 1.32	1.05	0.82, 1.33
6–7	544	32.1	341	30.1	0.95	0.79, 1.14	0.96	0.80, 1.16
7–8	681	40.2	461	40.7	1.00		1.00
>8	232	13.7	168	14.8	1.05	0.83, 1.33	1.10	0.87, 1.39
Duration of sleep on nonworkdays, hours (n = 2,828)
<6	166	9.8	118	10.4	1.07	0.82, 1.39	1.06	0.80, 1.39
6–7	369	21.8	235	20.7	0.96	0.78, 1.17	0.96	0.78, 1.18
7–8	670	39.5	461	40.7	1.00		1.00
>8	490	28.9	319	28.2	0.92	0.76, 1.11	0.94	0.78, 1.13
Sleep quality (n = 2,827)^b
Very good	420	24.8	281	24.8	1.00		1.00
Fairly good	852	50.3	580	51.2	1.03	0.85, 1.24	1.06	0.88, 1.28
Fairly bad	331	19.6	208	18.3	0.97	0.77, 1.22	0.98	0.77, 1.24
Very bad	90	5.3	65	5.7	1.14	0.80, 1.63	1.11	0.77, 1.61
Duration on workdays (hours)/quality (n = 2,826)^c
<6/good	96	5.7	69	6.1	1.10	0.79, 1.55	1.12	0.80, 1.58
6–7/good	378	22.3	244	21.5	0.99	0.80, 1.21	1.02	0.82, 1.26
7–8/good	591	34.9	394	34.8	1.00		1.00
>8/good	207	12.2	154	13.6	1.10	0.86, 1.40	1.14	0.89, 1.47
<6/bad	142	8.4	94	8.3	1.03	0.77, 1.38	1.03	0.67, 1.39
6–7/bad	164	9.7	97	8.6	0.93	0.70, 1.24	0.90	0.67, 1.21
7–8/bad	90	5.3	67	5.9	1.13	0.80, 1.59	1.15	0.81, 1.63
>8/bad	25	1.5	14	1.2	0.82	0.42, 1.60	0.93	0.47, 1.83

Abbreviations: CI, confidence interval; OR, odds ratio.
^aAdjusted for age, number of children, age at first birth, breastfeeding, menopausal status, use of hormone replacement therapy, duration of use of hormone replacement therapy, alcohol consumption, comparative weight at age 30 years, ever use of melatonin, and physical activity.
^bOne person was missing data for sleep quality.
^cTwo people were missing data for creating a combined sleep duration-and-quality variable.

Table 3. Adjusted Odds Ratios for the Associations Between Sleep Duration and Subjective Sleep Quality and the Risks of Pre- and Postmenopausal Breast Cancer, Breast Cancer Environment and Employment Study, 2009–2011
	Controls, %	Premenopausal Cases			Postmenopausal Cases			P Value^b
	Controls, %	%	AOR^a	95% CI	%	AOR^a	95% CI	P Value^b
Duration of sleep on workdays, hours (n = 2,828)								0.73
<6	14.0	11.4	1.00	0.62, 1.60	15.7	1.07	0.81, 1.40
6–7	32.1	28.2	1.11	0.79, 1.57	31.0	0.91	0.74, 1.13
7–8	40.2	43.0	1.00		39.6	1.00
>8	13.7	17.4	1.01	0.68, 1.51	13.7	1.17	0.88, 1.55
Duration of sleep on nonworkdays, hours (n = 2,828)								0.92
<6	9.8	7.7	1.02	0.58, 1.81	11.6	1.09	0.80, 1.48
6–7	21.8	15.1	0.90	0.59, 1.38	23.3	0.99	0.78, 1.25
7–8	39.5	42.2	1.00		40.0	1.00
>8	28.9	35.0	0.82	0.60, 1.13	25.1	1.00	0.80, 1.26
Sleep quality (n = 2,827)^c								0.80
Very good	24.8	27.1	1.00		23.8	1.00
Fairly good	50.3	51.3	1.04	0.75, 1.44	51.1	1.09	0.87, 1.36
Fairly bad	19.6	18.2	1.18	0.77, 1.82	18.4	0.91	0.69, 1.21
Very bad	5.3	3.4	1.04	0.46, 2.35	6.8	1.12	0.75, 1.68

Abbreviations: AOR, adjusted odds ratio; CI, confidence interval.
^aAdjusted for age, number of children, age at first birth, breastfeeding, use of hormone replacement therapy, duration of use of hormone replacement therapy, alcohol consumption, comparative weight at age 30 years, ever use of melatonin, and physical activity.
^b P value for the difference between pre- and postmenopausal cancer risk.
^cOne person was missing data for sleep quality.

Table 4. Adjusted Odds Ratiosa for the Associations Between Sleep Duration and Subjective Sleep Quality and the Risks of Estrogen Receptor–Positive and Estrogen Receptor–Negative Breast Cancer, Breast Cancer Environment and Employment Study, 2009–2011
	Controls, %	Estrogen Receptor– Positive Cases^a			Estrogen Receptor– Negative Cases^a			P Value^b
	Controls, %	%	AOR^c	95% CI	%	AOR^c	95% CI	P Value^b
Duration of sleep on workdays, hours (n = 2,659)
<6	14.0	14.0	1.08	0.82, 1.41	13.8	0.99	0.59, 1.68	0.86
6–7	32.1	32.2	1.08	0.88, 1.32	28.3	0.94	0.62, 1.41
7–8	40.2	38.8	1.00		40.3	1.00
>8	13.7	15.0	1.17	0.90, 1.52	17.6	1.26	0.78, 2.04
Duration of sleep on nonworkdays, hours (n = 2,659)
<6	9.8	9.9	1.00	0.73, 1.36	12.0	1.30	0.74, 2.29	0.73
6–7	21.8	22.2	1.01	0.81, 1.27	17.6	0.90	0.56, 1.46
7–8	39.5	41.0	1.00		37.7	1.00
>8	28.9	26.8	0.88	0.71, 1.09	32.7	1.11	0.74, 1.66
Sleep quality (n = 2,658)
Very good	24.8	24.0	1.00		28.3	1.00		0.52
Fairly good	50.3	50.9	1.10	0.89, 1.37	48.8	0.88	0.59, 1.31
Fairly bad	19.6	19.7	1.10	0.84, 1.43	15.1	0.73	0.43, 1.25
Very bad	5.3	5.5	1.11	0.73, 1.68	8.2	1.49	0.75, 2.97

Abbreviations: AOR, adjusted odds ratio; CI, confidence interval.
^aA total of 169 women with breast cancer did not have estrogen receptor status recorded on their Western Australian Cancer Registry records.
^b P value for the difference between estrogen receptor-positive and estrogen receptor-negative cancer risk.
^cAdjusted for age, number of children, age at first birth, breastfeeding, menopausal status, use of hormone replacement therapy, duration of use of hormone replacement therapy, alcohol consumption, comparative weight at age 30 years, ever use of melatonin, and physical activity.

Table 5. Adjusted Odds Ratios for the Association Between Duration of Sleep on Workdays and Breast Cancer Risk According to 3 Characteristics of Circadian Rhythm,a Breast Cancer Environment and Employment Study, 2009–2011
Circadian Type and Duration of Sleep, hours	Controls		Cases		Adjusted Odds Ratio^b	95% Confidence Interval
Circadian Type and Duration of Sleep, hours	No.	%	No.	%	Adjusted Odds Ratio^b	95% Confidence Interval
Phase (Morningness/Eveningness) (n = 2,811) ^c
Neither type
<6	111	12.7	79	14.1	1.26	0.89, 1.79
6–7	283	32.3	176	31.4	1.08	0.83, 1.40
7–8	373	42.5	233	41.6	1.00
>8	110	12.5	72	12.9	1.15	0.80, 1.63
Evening type
<6	57	16.6	35	15.0	0.66	0.38, 1.15
6–7	123	35.9	71	30.3	0.66	0.43, 1.02
7–8	125	36.4	98	41.9	1.00
>8	38	11.1	30	12.8	0.87	0.48, 1.59
Morning type
<6	70	15.0	47	14.2	0.99	0.63, 1.57
6–7	135	29.0	92	27.8	0.98	0.68, 1.42
7–8	180	38.6	127	38.4	1.00
>8	81	17.4	65	19.6	1.16	0.76, 1.78
Amplitude (Languidness/Vigorousness) (n = 2,808) ^c
Neither type
<6	121	14.9	76	13.1	0.83	0.58, 1.18
6–7	299	36.7	184	31.8	0.82	0.63, 1.06
7–8	301	37.0	240	41.5	1.00
>8	93	11.4	79	13.6	1.19	0.83, 1.70
Vigorous type
<6	55	17.2	47	21.6	1.78	1.02, 3.09
6–7	108	33.9	72	33.0	1.21	0.76, 1.93
7–8	120	37.6	67	30.7	1.00
>8	36	11.3	32	14.7	1.70	0.90, 3.21
Languid type
<6	58	10.6	38	11.6	1.12	0.69, 1.83
6–7	133	24.2	82	25.0	1.02	0.71, 1.46
7–8	256	46.6	151	46.0	1.00
>8	103	18.7	57	17.4	0.90	0.60, 1.35
Stability (Flexibility/Rigidity) (n = 2,807) ^c
Neither type
<6	109	12.1	75	13.0	1.04	0.73, 1.49
6–7	292	32.5	155	26.9	0.83	0.64, 1.08
7–8	374	41.7	257	44.5	1.00
>8	123	13.7	90	15.6	1.07	0.77, 1.49
Rigid type
<6	31	8.6	28	10.6	1.48	0.79, 2.77
6–7	87	24.0	80	30.3	1.31	0.86, 1.99
7–8	166	45.9	112	42.4	1.00
>8	78	21.5	44	16.7	0.74	0.46, 1.20
Flexible type
<6	94	22.2	58	20.6	0.87	0.55, 1.38
6–7	162	38.2	104	36.9	0.93	0.63, 1.36
7–8	137	32.3	90	31.9	1.00
>8	31	7.3	30	10.6	1.44	0.78, 2.65

^aPhase (morningness/eveningness), amplitude (languidness or vigorousness), and stability (the flexibility or rigidity of sleeping habits).
^bAdjusted for age, number of children, age at first birth, breastfeeding, menopausal status, use of hormone replacement therapy, duration of use of hormone replacement therapy, alcohol consumption, comparative weight at age 30 years, ever use of melatonin, physical activity, country of birth, highest level of education, family history of breast cancer, previous noncancerous breast conditions, use of oral contraceptives in the last 5 years, and ever smoking tobacco.
^cSeventeen participants were missing data on circadian phase, 20 were missing data on amplitude, and 21 were missing data on stability.

Table 6. Adjusted Odds Ratios for the Association Between Quality of Sleep and Breast Cancer Risk According to 3 Characteristics of Circadian Rhythm,a Breast Cancer Environment and Employment Study, 2009–2011
Circadian Type and Quality of Sleep	Controls		Cases		Adjusted Odds Ratio^b	95% Confidence Interval
Circadian Type and Quality of Sleep	No.	%	No.	%	Adjusted Odds Ratio^b	95% Confidence Interval
Phase (Morningness/Eveningness) (n = 2,810) ^c
Neither type
Very good	196	22.4	125	22.3	1.00
Fairly good	470	53.7	305	54.5	1.12	0.85, 1.49
Fairly bad	163	18.6	97	17.3	1.01	0.71, 1.45
Very bad	46	5.3	33	5.9	1.37	0.81, 2.32
Evening type
Very good	73	21.3	54	23.0	1.00
Fairly good	161	46.9	100	42.5	0.89	0.55, 1.42
Fairly bad	83	24.2	63	26.8	1.13	0.66, 1.92
Very bad	26	7.6	18	7.7	0.87	0.40, 1.88
Morning type
Very good	148	31.8	102	30.8	1.00
Fairly good	218	46.8	168	50.8	1.12	0.80, 1.59
Fairly bad	82	17.6	47	14.2	0.83	0.52, 1.32
Very bad	18	3.9	14	4.2	0.98	0.44, 2.16
Amplitude (Languidness/Vigorousness) (n = 2,807) ^c
Neither type
Very good	185	22.8	149	25.7	1.00
Fairly good	418	51.5	306	52.8	0.96	0.73, 1.26
Fairly bad	160	19.7	95	16.4	0.75	0.53, 1.07
Very bad	49	6.03	30	5.2	0.78	0.46, 1.33
Vigorous type
Very good	90	28.2	56	25.7	1.00
Fairly good	151	47.3	107	49.1	1.07	0.67, 1.70
Fairly bad	60	18.8	36	16.5	1.00	0.55, 1.82
Very bad	18	5.6	19	8.7	1.74	0.73, 4.18
Languid type
Very good	143	26.0	75	22.9	1.00
Fairly good	278	50.6	162	49.4	1.22	0.85, 1.75
Fairly bad	107	19.5	75	22.9	1.39	0.90, 2.16
Very bad	22	4.0	16	4.9	1.46	0.67, 3.17
Stability (Flexibility/Rigidity) (n = 2,806) ^c
Neither type
Very good	214	23.9	140	24.3	1.00
Fairly good	461	51.5	314	54.4	1.08	0.82, 1.41
Fairly bad	185	20.7	98	17.0	0.84	0.60, 1.19
Very bad	36	4.0	25	4.3	1.14	0.64, 2.04
Rigid type
Very good	95	26.2	68	25.8	1.00
Fairly good	188	51.9	133	50.4	1.08	0.72, 1.63
Fairly bad	60	16.6	53	20.1	1.39	0.82, 2.33
Very bad	19	5.3	10	3.8	1.04	0.41, 2.62
Flexible type
Very good	110	25.9	69	24.4	1.00
Fairly good	195	46.0	129	45.8	1.07	0.72, 1.60
Fairly bad	85	20.1	57	20.1	1.13	0.69, 1.83
Very bad	34	8.0	28	9.9	1.37	0.72, 2.59

^aPhase (morningness/eveningness), amplitude (languidness or vigorousness), and stability (the flexibility or rigidity of sleeping habits).
^bAdjusted for age, number of children, age at first birth, breastfeeding, menopausal status, use of hormone replacement therapy, duration of use of hormone replacement therapy, alcohol consumption, comparative weight at age 30 years, ever use of melatonin, and physical activity.
^cSeventeen participants were missing data on circadian phase, 20 were missing data on amplitude, and 21 were missing data on stability.