Medication overuse,healthy lifestyle behaviour and stress in chronic headache: Results from a population-based representative survey

Abstract

Aim

This cross-sectional study investigated associations between chronic headache (CH) with and without medication overuse, healthy lifestyle behaviour, and stress.

Methods

Questionnaires were sent to 129,150 adults. Those with headache ≥15 days per month for three months were classified as having CH then further described as having medication-overuse headache (MOH) or CH without medication overuse. Associations between headache and daily smoking, physical inactivity, obesity, excessive drinking, illicit drug use, and high stress were analysed by logistic regression.

Results

CH with and without medication overuse (prevalence 1.8% and 1.6%, respectively) had strong, graded associations with stress. Associations with daily smoking, physical inactivity, and obesity were significant only for MOH. Odds for MOH were highest among people who had all three factors compared to those who had none (OR 2.8 in women and 5.1 in men). High stress plus any of these three factors had synergistic effects in MOH but not clearly in those who had CH without overuse. Associations between CH subtypes and excessive drinking or illicit drug use were not statistically significant.

Conclusion

Results suggest strong links between healthy lifestyle behaviour and stress in MOH. Stress reduction and promoting healthy behaviour are highly relevant in MOH management.

Keywords

Chronic headache medication-overuse headache secondary headache disorders health behaviour life stress smoking sedentary lifestyle obesity alcohol abuse illicit drugs

Introduction

Medication-overuse headache (MOH) has been described as a biobehavioural disorder: There is a biological basis for the condition as well as behavioural factors explaining the ‘initiation and sustaining dynamics’ of overuse (1). Medication overuse could be viewed as a form of harmful health behaviour in people with frequent headache. The intriguing question is whether this co-exists with other health behaviour and lifestyle factors such as smoking, physical inactivity, obesity, excessive alcohol intake, and illicit drug use.

In population-based studies that have explored the links between healthy lifestyle behaviours and chronic headache (CH) (2 –4) few identified people with MOH as a distinct group (5 –7). In these studies, behaviours were analysed as single factors. The approach of studying combined effects of healthy lifestyle behaviours has been applied in the study of chronic noncommunicable diseases (8,9) but only in a few studies on headache. In one study, no association was found between headache and a health index based on four factors: smoking, physical activity, alcohol consumption, and body mass index (BMI). However, the authors looked at migraine and tension-type headache in general, not their chronic forms (10). Another study found that overweight, smoking, and physical inactivity were associated with recurrent headache independently and in combination but the study included only 12- to 19-year-olds and did not examine medication overuse (11).

Another unexplored question is whether stress mediates the relationship between healthy lifestyle behaviour and MOH. Stressful events or fluctuating stress levels can trigger headache (12 –14), and there appears to be an association between stress and analgesic use for headache among those with fewer coping resources (15). A longitudinal study showed how increasing stress is associated with increasing headache days even after controlling for medication overuse (16).

Population-based studies have shown that high stress is related to smoking (17), physical inactivity (18), obesity (19), and alcohol use disorder (20). Stress has been consistently associated with illicit drug use (21). Stress could be the reason for unhealthy lifestyle choices; and at the same time, attempts to make healthy lifestyle changes could be limited by external stressors or psychological distress (22).

To our knowledge, there are neither large population-based studies on stress and MOH nor studies on how stress is related to healthy lifestyle behaviour in people with MOH.

The aims of our study were: 1) to examine the prevalence of daily smoking, physical inactivity, obesity, excessive alcohol use, illicit drug use, and high stress among individuals with MOH and those with chronic headache without medication overuse (CHnoO); and 2) to examine whether there are associations between MOH/CHnoO, stress, and healthy lifestyle behaviours (independently and in combination); and whether the association between headache and these behaviours is modified by stress.

We hypothesised that medication overuse in CH is more likely among people who have other harmful health behaviours, and furthermore, that the relationship between healthy lifestyle behaviours and MOH is mediated by stress.

Methods

Study population and data gathering

This cross-sectional population-based study is part of the 2010 Danish National Health Survey (DNHS) (23). A self-administered questionnaire was used to gather data on headache frequency, chronic illness, over-the-counter (OTC) medication use, healthy lifestyle behaviour, and stress from 129,150 residents aged ≥16 years sampled randomly using the Civil Registration System (24). The stratified sampling methodology has been previously described (23,25). This representative sample made up 7.0% and 5.2% of two regional populations.

Using a unique personal identification number, data from the DNHS were linked at the individual level to other nationwide registers. Prescription medication overuse was assessed using Danish National Prescription Registry (DNPR) records from 2009 (26). Demographic (age, sex, ethnicity) and socioeconomic data (educational attainment, work status and income) were retrieved from national registers (27). Civil, partnership, and family status data were gathered from the questionnaire.

Case ascertainment

Respondents with ≥15 days of headache per month over the last three months were classified as having CH. Those with CH plus OTC medication use ≥15 days per month were classified as having MOH. This case definition approximated the International Classification of Headache Disorders third edition beta (ICHD-3 beta) diagnostic criteria for MOH due to simple analgesic overuse (28). OTC medications were not specified. The cut-off level was conservative if the overused medications were combination preparations (for which ≥10 days per month is considered as overuse).

DNPR data were retrieved for respondents with CH who purchased drugs coded under Anatomical Therapeutic Chemical groups M01A (anti-inflammatory, antirheumatic products), N02A (opioids), N02B (other analgesics, antipyretics), and N02C (antimigraine drugs). Defined daily doses (DDDs) per month were calculated, with one DDD being an adult’s average maintenance dose per day for a drug used for its main indication (29). The levels considered as cut-off for overuse were ≥30 DDDs per month for simple analgesics; and ≥20 DDDs per month for opioids, triptans, ergotamines, combination preparations, and combinations of different drugs not individually overused (25). DDDs were calculated as an average per month in order to approximate the diagnostic criteria of overuse for more than three months. Use of DDDs as a drug utilisation metric has been described by the World Health Organisation (WHO) (29). Use of prescription registry data for estimating MOH prevalence and incidence has been described by Hagen et al. (5), who used the same cut-off level of ≥20 DDDs per month for triptans and opioids. The cut-off level of ≥30 DDDs per month for simple analgesics is a conservative threshold that is equivalent to daily intake of the full dose of these medications.

Butalbital, a barbiturate used in some countries for the treatment of tension-type headache, is not available in Denmark.

Healthy lifestyle behaviour and stress

Data on healthy lifestyle behaviour were gathered using the DNHS questionnaire. Smoking status was dichotomised as daily smoker, yes/no. Physical inactivity was assessed using a validated questionnaire battery (30). Results were dichotomised as low physical activity (yes/no) based on a cut-off of <30 minutes/day doing moderate to hard physical activity at work or during leisure.

Overweight (BMI 25–29.9 kg/m²) and obesity (≥30 kg/m²) were computed from self-reported height and weight.

Excessive drinking was defined as binge drinking (>5 units at one time at least once a week); large alcohol intake (>35 units in a week); or a score of ≥2 in the modified CAGE-C questionnaire (31).

Questions on illicit drug use were given only to those aged 16–34 years (about 20% of the sample). They were asked about use of cannabis, amphetamines, ecstasy, cocaine, lysergic acid diethylamide (LSD), heroin, euphoria-inducing mushroom, and other narcotics.

Stress was measured using the Perceived Stress Scale (PSS) with one-month recall (32). The PSS correlates with the Major Depression Inventory in the Danish general population (33). Ten questions were used to measure the degree to which situations in one’s life are deemed as stressful. Each question was scored 0–4 on a Likert scale. Higher scores indicated more stress. A total score was not computed if a respondent had >3 missing answers. If there were ≤3 missing, these items were scored as 2 (neutral on the scale).

Respondents were grouped into quintiles based on total PSS scores (range 0–40). Scores for each quintile were 0–6, 7–10, 11–13, 14–18, and 19–40. The fifth quintile with scores ≥19 was designated as having high stress.

Confounders

Sociodemographic variables were considered possible confounders in the relationships between MOH/CHnoO and healthy lifestyle behaviour. In this sample, both MOH and CHnoO are more prevalent among women, the middle-aged, non-ethnic Danes, and those with low socioeconomic position (25).

Comorbid illnesses were considered as possible confounders. Chronic pain conditions and psychiatric illness have been associated with MOH (5). These conditions are also related to healthy lifestyle behaviour as well as analgesic use. Furthermore, comorbid illnesses could be sources of stress independent of headache. Analyses were thus controlled for self-reports of comorbid osteoarthritis, rheumatoid arthritis, discus prolapse/back pain, and short- and long-term psychiatric illness (yes/no).

Physical activity is likely to be influenced by smoking and obesity; and smoking is related to alcohol drinking. Mutual adjustments were thus considered in the regression models.

Statistical analysis

Distributions of healthy lifestyle behaviours and stress quintiles were calculated. Chi-square tests were used to compare distributions across headache groups by comparing CH vs. no CH, then comparing MOH vs. CHnoO. Because of the large number of tests conducted, a cut-off p value of <0.001 was considered statistically significant.

Logistic regression analyses (adjusted for possible confounders) were performed with headache as the outcome variable and one healthy lifestyle behaviour or stress quintile as the explanatory variable. Each of these models was adjusted for sociodemographic variables as well as comorbid chronic pain and psychiatric illness. The variables used are listed in the description of Table 1, Model 1.

Table 1.

Associations between chronic headache with and without medication overuse and stress score, smoking, alcohol drinking, physical activity, body mass index and illicit drug use.

	Chronic headache without medication overuse, odds ratio (95% CI)
	Model 1	Model 2	p value^d	Model 1	Model 2	p value^d
Females
Perceived stress quintile			<0.0001			<0.0001
Second quintile vs. lowest quintile	1.1 (0.7–1.8)	1.1 (0.7–1.9)		1.2 (0.7–1.9)	1.0 (0.6–1.7)
Third quintile vs. lowest quintile	1.8 (1.1–2.9)^e	1.8 (1.1–3.0)^e		1.5 (0.9–2.5)	1.5 (0.9–2.4)
Fourth quintile vs. lowest quintile	2.0 (1.3–3.2)^e	2.2 (1.4–3.6)^e		2.2 (1.4–3.3)^e	2.0 (1.3–3.2)^e
Highest quintile vs. lowest quintile	4.0 (2.6–6.4)^e	4.1 (2.6–6.7)^e		4.4 (2.9–6.8)^e	3.9 (2.5–6.1)^e
Daily smoking (yes vs.no)	1.0 (0.7–1.4)	1.0 (0.7–1.3)	0.8792	1.5 (1.2–1.9)^e	1.4 (1.1–1.8)^e	0.0031
Excessive alcohol drinking (CAGE ≥2, yes vs. no)	0.8 (0.6–1.2)	0.8 (0.6–1.2)	0.3561	1.0 (0.7–1.3)	0.9 (0.7–1.3)	0.6743
Low physical activity (yes vs. no)	1.2 (0.9–1.5)	1.1 (0.8–1.4)	0.4512	1.6 (1.2–2.0)^e	1.4 (1.1–1.8)^e	0.0039
Body mass index^a			0.6977			0.1005
Overweight vs. <25 kg/m²	0.9 (0.7–1.2)	0.9 (0.7–1.2)		1.3 (1.0–1.6)^e	1.2 (1.0–1.6)
Obese vs. <25 kg/m²	1.2 (0.9–1.7)	1.1 (0.7–1.5)		1.4 (1.1–1.9)^e	1.3 (1.0–1.8)
Illicit drug use including cannabis last month (yes vs. no)^b	0.7 (0.2–2.5)	–		0.9 (0.2–3.9)	–
Illicit drug use excluding cannabis last month (yes vs. no)^b	n.a.	–		2.2 (0.3–18.2)	–
Lifetime use of any illicit drug (yes vs. no)^b	0.7 (0.4–1.4)	–		1.0 (0.5–2.1)	–
Global tests for models (Wald test)	p < 0.0001^c	p < 0.0001		p < 0.0001^b	p < 0.0001
Males
Perceived stress quintile			<0.0001			<0.0001
Second quintile vs. lowest quintile	3.5 (1.7–7.1)^e	4.5 (2.1–9.3)^e		1.0 (0.5–2.1)	1.0 (0.5–2.2)
Third quintile vs. lowest quintile	3.3 (1.6–6.9)^e	3.6 (1.7–7.5)^e		1.7 (0.8–3.6)	1.9 (0.9–4.0)
Fourth quintile vs. lowest quintile	6.8 (3.5–13.4)^e	8.9 (4.5–17.9)^e		4.2 (2.3–7.9)^e	4.3 (2.2–8.2)^e
Highest quintile vs. lowest quintile	12.0 (6.1–23.7)^e	15.5 (7.7–31.0)^e		8.8 (4.6–16.6)^e	10.3 (5.5–19.3)^e
Daily smoking (yes vs.no)	1.2 (0.8–1.7)	1.0 (0.7–1.5)	0.9141	1.5 (1.1–2.2)^e	1.5 (1.0–2.1)^e	0.0315
Excessive alcohol drinking (CAGE ≥2, yes vs. no)	1.0 (0.7–1.5)	0.9 (0.6–1.3)	0.5268	1.3 (0.9–1.9)	1.3 (0.9–1.9)	0.1852
Low physical activity (yes vs. no)	1.1 (0.8–1.6)	1.0 (0.7–1.5)	0.8951	1.7 (1.2–2.4)^e	1.4 (1.0–2.0)^e	0.0435
Body mass index^a			0.4574			0.0386
Overweight vs. normal	1.0 (0.7–1.3)	1.1 (0.7–1.5)		1.5 (1.0–2.2)^e	1.5 (1.0–2.3)^e
Obese vs. normal	1.4 (0.9–2.2)	1.3 (0.8–2.2)		1.8 (1.1–2.8)^e	1.8 (1.1–2.8)^e
Illicit drug use including cannabis last month (yes vs. no)^b	0.3 (0.0–2.2)	–		1.3 (0.3–4.9)	–
Illicit drug use excluding cannabis last month (yes vs. no)^b	1.1 (0.1–10.9)	–		1.0 (0.1–6.2)	–
Lifetime use of any illicit drug (yes vs. no)^b	0.8 (0.3–2.1)	–		1.9 (0.7–5.1)	–
Global tests for models (Wald test)	p < 0.0001^c	p < 0.0001		p < 0.0001^c	p < 0.0001

CAGE: score on questionnaire on excessive alcohol use; CI: confidence interval; n.a.: Not applicable, there was no reported illicit drug use (excluding cannabis) among young women (16–34 years old) with chronic headache but no medication overuse. Logistic regression analyses were performed using chronic headache with or without medication overuse as the outcome variable in the total sample. Analyses were weighted for stratified sampling and nonresponse.

Model 1: Regression analysis with each explanatory variable adjusted for sociodemographic variables: age as a continuous variable, Western ethnicity (yes/no), civil status, living alone (yes/no), children at home (yes/no), education, work status and income plus self-report of comorbid osteoarthritis, rheumatoid arthritis, discus prolapse/back pain, and short- and long-term psychiatric illness (yes/no).

Model 2: Model 1 with mutual adjustment for other health behaviours and perceived stress. Illicit drug use was not included in this model.

The categories underweight and normal were combined due to the small number of underweight males with valid data on other variables in the model. ^bQuestions on illicit drug use were given only to respondents 16–34 years old. ^cAll models were highly significant on global (Wald) tests, p < 0.0001. ^dP values refer to type 3 analyses of effects for including each variable in regression Model 2. ^eThe CI does not overlap the null value (odds ratio = 1.0).

Regression analyses were also performed with headache as the outcome variable and four healthy lifestyle behaviour variables (smoking, CAGE-C score, physical activity, BMI) plus stress quintile as covariates. These models were also adjusted for sociodemographic variables and comorbid illness (see description of Table 1, Model 2). Type 3 analyses of effects of including the covariates were performed and a p value of <0.05 was used as an indication that adding the variable significantly improved the model fit. Illicit drug use was not included in these models because data were available on young adults only.

Statistical interaction was assessed by testing the inclusion of each of the following terms in the full regression models (Model 2): smoking*BMI, smoking*physical activity, smoking*stress, BMI*physical activity, BMI*stress, physical activity*stress.

Distributions were computed for three healthy lifestyle behaviours independently and in combination (i.e. having 0, one, two, or three behaviours). These three were identified as significant explanatory variables (p < 0.05) for MOH in the above regression analyses. Associations between headache and these combinations were analysed by logistic regression under the hypothesis that having many harmful health behaviours would be more strongly associated with MOH.

Stress was examined as a possible effect modifier in the relationship between healthy lifestyle behaviour and headache. Odds ratios (ORs) were computed for each behaviour in combination with high or low stress levels to investigate synergistic effects.

All analyses were adjusted for stratified sampling and possible differential non-response using weights computed by Statistics Denmark. Adjustment for non-response was based on register information on sex, age, municipality of residence, educational level, income, marital status, ethnic background, number of visits to the general practitioner in 2007, hospitalisation in 2007 (yes/no), occupational status, owner/tenant status, and protection from inquiries during statistical surveys (23).

SAS software (version 9.3, SAS Institute Inc, USA) was used in all prevalence and regression analyses.

Ethical considerations

In Denmark, all residents have a 10-digit Central Person Register (CPR) number. To maintain anonymity, CPR numbers were encrypted by the national statistics agency (Statistics Denmark) and replaced with 12-digit personal identification numbers which were consistently used as identifiers in all data sources. The reference database relating numeric codes to individual identities were not accessible to the authors. The use of Danish nationwide databases for health research, particularly linkage at the individual level, has been previously described (27). Links to different registers were also implemented for DNHS 2010 (23).

The goal of the DNHS was stated in the questionnaire: to gather information about residents’ health and illness in order to guide ongoing efforts to improve public health. Invited participants were informed that additional health-related information – including prescription medication – will be retrieved from national registers. Participation was voluntary. Informed consent for use of data was given by respondents upon returning the questionnaire.

The study was approved by the Danish Data Protection Agency according to the Danish Act on Processing of Personal Data. Approval from the Danish Health Research Ethics Committee System was not required according to Danish law, as the research project was purely based on data from questionnaires and registers, and did not involve collection of human biological material. Authors M.W. and C.G. had access to data for statistical analysis.

Results

The response rate was 53.1% (68,518 people). There were 2087 respondents identified as having CH, further classified as 1110 with MOH and 977 with CHnoO. The prevalence of CH adjusted for stratified sampling and non-response was 3.3% (confidence interval (CI): 3.2–3.5%). The adjusted prevalence of MOH was 1.8% (CI: 1.7–1.9%) and for CHnoO, 1.6% (CI: 1.4–1.7%). Detailed characteristics of the study population are presented in Table 2. Prevalence of CH, MOH and CHnoO by sociodemographic categories have been previously reported (25).

Table 2.

Demographic characteristics of respondents: distribution by sex, age, civil/family status, ethnicity, education, work status and income (column percentage totals = 100).

	No chronic headache		All with chronic headache		Chronic headache without medication overuse		Medication-overuse headache
	n	%	n	%	n	%	n	%
Female	36,834	55.4	1404	67.3	623	63.8	781	70.4
Male	29,597	44.6	683	32.7	354	36.2	329	29.6
Age group (years)
16–24	6535	9.8	222	10.6	147	15.0	75	6.8
25–34	7294	11.0	180	8.6	101	10.3	79	7.1
35–44	11,329	17.1	354	17.0	169	17.3	185	16.7
45–54	12,166	18.3	419	20.1	161	16.5	258	23.2
55–64	13,230	19.9	445	21.3	181	18.5	264	23.8
65–79	13,075	19.7	387	18.5	176	18.0	211	19.0
80+	2802	4.2	80	3.8	42	4.3	38	3.4
Civil status^a
Married or registered partnership	37,755	58.5	1,091	54.8	462	50.2	629	58.8
Separated or divorced	6048	9.4	235	11.8	99	10.7	136	12.7
Widow or widower	4398	6.8	147	7.4	67	7.3	80	7.5
Single	16,344	25.3	518	26.0	293	31.8	225	21.0
Partnership status^a
Living alone	18,380	28.6	660	33.2	333	36.1	327	30.8
Living with partner	45,883	71.4	1,325	66.8	590	63.9	735	69.2
Children 15 years or younger^a
Without child at home	43,549	70.3	1,296	69.2	586	66.8	710	71.3
With child at home	18,425	29.7	577	30.8	291	33.2	286	28.7
Ethnicity^b
Danish	61,360	92.4	1,779	85.2	838	85.8	941	84.8
Non-Danish, Western	2256	3.4	59	2.8	37	3.8	22	2.0
Non-Danish, non-Western	2815	4.2	249	11.9	102	10.4	147	13.2
Highest educational attainment^a
Primary/secondary school	16,176	25.0	760	37.8	348	36.9	412	38.5
Vocational education	29,941	46.3	930	46.2	427	45.3	503	47.0
Academy or bachelor degree	11,659	18.0	253	12.6	126	13.4	127	11.9
Master’s or PhD degree	6831	10.6	70	3.5	42	4.5	28	2.6
Work status^a
Employed	42,024	64.9	1,028	51.1	509	53.9	519	48.6
Student	4411	6.8	140	7.0	96	10.2	44	4.1
Unemployed	250	0.4	17	0.8	11	1.2	6	0.6
Receiving sickness benefits	405	0.6	52	2.6	26	2.8	26	2.4
On social welfare	535	0.8	87	4.3	28	3.0	59	5.5
Early pensioner	1996	3.1	241	12.0	73	7.7	168	15.7
Old–age pensioner	15,166	23.4	447	22.2	201	21.3	246	23.0
Income (USD)^a
0–<18,000	3214	5.1	140	7.1	65	7.3	75	7.0
18,000–<27,000	8097	12.9	386	19.7	153	17.2	233	21.7
27,000–<45,000	15,246	24.2	606	30.9	271	30.6	335	31.1
45,000–<68,000	22,102	35.1	603	30.7	284	32.0	319	29.6
68,000–<95,000	10,164	16.1	176	9.0	88	9.9	88	8.2
95,000+	4115	6.5	52	2.6	26	2.9	26	2.4

Respondents with missing data are not included in the table. ^bThis category was dichotomised as Western ethnicity (yes/no) in regression analyses.

Prevalence proportions by age, sex, and sociodemographic status have been previously reported. (See Westergaard et al. (25)).

Comparison of groups with and without CH

By definition, the proportion of people in the highest stress quintile was about 20% of the entire sample. However, as many as 52.9% of women and 52.2% of men with CH were in the highest stress quintile. The CH group was also more likely than those with no CH to report daily smoking, physical inactivity, overweight and obesity (all p < 0.0001); but not excessive drinking (Table 3).

Table 3.

Distribution of respondents according to stress score, smoking, alcohol drinking, physical activity, body mass index and illicit drug use by headache group.

	Total respondents	Respondents in this category^e	Distributions for respondents		p-value All CH vs. No CH	Distributions for respondents with		p value MOH vs. CHnoO
	without chronic headache, % (CI)	with chronic headache, % (CI)	chronic headache but no overuse, % (CI)	medication-overuse headache, % (CI)
Females
Highest quintile based on perceived stress scores	37,540	7342	21.1 (20.6–21.6)	52.9 (49.8–56.0)	<0.0001	48.0 (43.3–52.7)	56.9 (52.8–60.9)	0.0051
Daily smoking	37,250	6744	18.9 (18.4–19.4)	28.7 (25.9–31.6)	<0.0001	25.1 (20.9–29.3)	31.8 (27.9–35.6)	0.0239
Excessive alcohol drinking
CAGE-C score ≥2	37,102	4072	10.4 (10.0–10.7)	8.1 (6.6–9.7)	0.0131	8.3 (5.9–10.6)	8.0 (6.0–10.1)	0.8914
Binge drinking, over five units	37,318	2609	7.5 (7.2–7.8)	6.8 (5.4–8.3)	0.4198	8.6 (6.2–11.0)	5.4 (3.7–7.2)	0.0322
Large alcohol intake, over 35 units a week	36,670	298	0.9 (0.8–1.0)	1.1 (0.4–1.7)	0.5383	0.8 (0.0–1.6)	1.3 (0.4–2.2)	0.4788
Low physical activity, <30 minutes’ exercise per day	36,236	10,545	29.3 (28.7–29.8)	43.0 (39.8–46.1)	<0.0001	34.9 (30.3–39.5)	49.7 (45.4–54.0)	<0.0001
Body mass index	37,379
under 18.5 kg/m²		1361	4.0 (3.8–4.3)	5.0 (3.7–6.3)	0.0967^a	7.1 (4.8–9.4)	3.3 (1.9–4.6)	0.0029^a
18.5–<25 kg/m²		21,685	58.8 (58.2–59.4)	47.9 (44.8–51.1)		52.7 (48.0–57.3)	44.0 (39.8–48.2)
25–29.9 kg/m²		9636	24.9 (24.3–25.4)	27.4 (24.6–30.2)	<0.0001^b	22.4 (18.5–26.3)	31.5 (27.6–35.4)	0.0001^b
≥30 kg/m²		4697	12.3 (11.9–12.7)	19.7 (17.2–22.2)	<0.0001^c	17.9 (14.1–21.6)	21.2 (17.8–24.6)	0.2019^c
Males
Highest quintile based on perceived stress scores	29,898	4015	14.7 (14.2–15.2)	52.2 (47.7–56.7)	<0.0001	45.1 (38.8–51.3)	59.4 (53.1–65.7)	0.0016
Daily smoking	29,434	5948	21.9 (21.3–22.5)	36.6 (32.0–41.1)	<0.0001	35.5 (29.3–41.8)	37.5 (31.0–44.0)	0.6655
Excessive alcohol drinking
CAGE-C score ≥2	29,599	7334	23.3 (22.8–23.9)	24.6 (20.8–28.5)	0.5053	23.4 (18.2–28.6)	25.9 (20.2–31.6)	0.5209
Binge drinking, over five units	29,645	5121	19.0 (18.5–19.6)	19.3 (15.4–23.1)	0.9001	19.4 (14.4–24.5)	19.1 (13.2–25.0)	0.9357
Large alcohol intake, over 35 units a week	29,237	1337	4.8 (4.5–5.1)	8.3 (5.6–11.0)	0.0016	9.6 (5.5–13.6)	7.0 (3.4–10.7)	0.3585
Low physical activity, <30 minutes’ exercise per day	29,268	7181	23.3 (22.7–23.8)	41.3 (36.7–45.9)	<0.0001	35.5 (29.4–41.6)	47.3 (40.6–54.1)	0.0109
Body mass index	29,870
under 18.5 kg/m²		327	1.5 (1.3–1.6)	3.1 (1.4–4.9)	0.0056^a	2.5 (0.5–4.4)	3.8 (1.0–6.7)	0.4125^a
18.5–<25 kg/m²		13,009	46.5 (45.9–47.2)	33.2 (28.9–37.6)		36.9 (30.8–42.9)	29.6 (23.3–35.8)
25–29.9 kg/m²		12,472	39.2 (38.5–39.8)	41.4 (37.0–45.9)	<0.0001^b	39.6 (33.5–45.8)	43.3 (36.8–49.8)	0.1960^b
≥30 kg/m²		4062	12.9 (12.4–13.3)	22.2 (18.4–25.9)	<0.0001^c	21.0 (15.9–26.1)	23.3 (17.9–28.7)	0.5497^c
16–34 years old (both sexes) ^d
Illicit drug use including cannabis last month	13,851	941	7.6 (7.1–8.1)	6.2 (2.9–9.6)	0.4667	4.2 (1.3-7.0)	9.3 (2.3-16.4)	0.1222
Illicit drug use excluding cannabis last month	13,820	193	1.6 (1.4–1.9)	0.8 (0.0–1.7)	0.1787	0.3 (0.0-1.0)	1.6 (0.0-3.4)	0.1481
Lifetime use of any illicit drug	13,820	2219	17.5 (16.7–18.2)	14.4 (10.4–18.5)	0.1771	11.4 (6.9-15.8)	19.0 (11.5-26.5)	0.0730

CH: chronic headache; CHnoO: chronic headache without medication overuse; MOH: medication-overuse headache; CI: 95% confidence interval. All proportions were adjusted for stratified sampling and non-response. Proportions for daily smoking, excessive alcohol drinking, low physical activity, and highest stress quintile were analysed as dichotomies, yes vs. no. Only the proportions classified as “yes” are shown in the table. Column totals for body mass index (%) equal 100. Missing responses were excluded from the analyses. P values refer to chi-square tests comparing headache groups.

P value for test comparing proportion with body mass index (BMI) under 18.5 kg/m² (yes/no). ^bP value for test comparing proportion with BMI ≥25 kg/m² (overweight or obese, yes/no). ^cP value for test comparing proportion with BMI ≥30 kg/m² (obese, yes/no). ^dQuestions regarding illicit drug use were given only to those aged 16–34 years. ^eRefers to the number of respondents with high stress, daily smoking, excessive alcohol drinking, or low physical activity. All respondents were classified according to categories of BMI.

Of 14,231 young adults who returned the questionnaire, 97% answered the questions on illicit drug use. Young men were more likely to report illicit drug use compared to young women (p < 0.0001); however, lifetime or previous month’s illicit drug use were not significantly different among those with and without CH (all p > 0.05).

Comparison of CH groups with and without medication overuse

Among those with MOH, 21.9% (CI: 19.0–24.9%) were obese; 33.8% (CI: 30.4–37.1%) smoked daily, 48.9% (CI: 45.2–52.5%) were physically inactive, and 57.7% (54.3–61.2%) were in the highest stress quintile (Table 3, summarised by sex).

A greater proportion of people with MOH were in the highest stress quintile compared to those with CHnoO. Women with MOH more commonly reported physical inactivity (p < 0.0001) and BMI ≥25 kg/m² (p = 0.0001) compared to women with CHnoO. Among men, the proportion with low physical activity was slightly higher in the MOH group (p = 0.01), and there were no clear differences for BMI (p > 0.05).

There were no statistically significant differences between MOH and CHnoO in terms of daily smoking (p > 0.01) or excessive drinking (p > 0.01) in both sexes.

Associations between headache, healthy lifestyle behaviour and stress

For both MOH and CHnoO, there was a strong graded relationship with stress scores even after adjusting for sociodemographic factors and comorbidities (Table 1, Model 1). The greatest ORs for headache were observed among those with the highest levels of stress, particularly among men.

There were significantly increased odds for MOH in individuals who were daily smokers, physically inactive, overweight, or obese. Similar associations were not statistically significant for CHnoO. There were no significant associations with CAGE-C score or illicit drug use for both CH subtypes (Table 1, Model 1).

Multivariate analyses (Table 1, Model 2) showed that for MOH, addition of the following variables significantly improved model fit (p < 0.05): daily smoking, physical inactivity, and stress quintile. BMI was a significant explanatory variable for men but not for women. The picture for CHnoO was very different, with stress quintile being the only significant explanatory variable (p < 0.0001).

There was a significant interaction between BMI and stress as explanatory variables for MOH (p < 0.0001) as shown by the addition of an interaction term in Table 1, Model 2. All other two-way interactions were not significant (p > 0.05).

Analyses of the relationships between headache and three behaviours singly or in combination are summarised in Table 4. ORs for MOH were highest among those with all three behaviours compared to those with none (OR 2.8 in women and 5.1 in men). In contrast, ORs for CHnoO did not deviate significantly from 1.0 for the different combinations among women and nearly all combinations among men.

Table 4.

Associations between chronic headache with and without medication overuse, and daily smoking, low physical activity, and obesity (independently or in combination).

	Chronic headache without medication overuse, odds ratio (95% CI)	Medication-overuse headache, odds ratio (95% CI)
Females
Not smoking daily, physically active, BMI <30 kg/m²	1.0 (reference)	1.0 (reference)
Daily smoking	1.1 (0.7–1.6)	1.7 (1.2–2.4)^a
Low physical activity	1.1 (0.8–1.6)	1.9 (1.4–2.6)^a
Obese	1.0 (0.6–1.7)	1.9 (1.2–2.9)^a
Daily smoking + low physical activity	1.3 (0.8–2.1)	2.4 (1.7–3.6)^a
Daily smoking + obese	1.2 (0.4–3.2)	2.1 (1.1–4.3)^a
Low physical activity + obese	1.6 (1.0–2.7)	1.7 (1.0–2.7)^a
Daily smoking + low physical activity + obese	0.9 (0.3–2.8)	2.8 (1.4–5.3)^a
Males
Not smoking daily, physically active, BMI <30 kg/m²	1.0 (reference)	1.0 (reference)
Daily smoking	1.1 (0.7–1.8)	1.9 (1.1–3.2)^a
Low physical activity	1.1 (0.7–1.9)	1.9 (1.2–3.1)^a
Obese	1.8 (1.1–3.1)^a	1.3 (0.7–2.3)
Daily smoking + low physical activity	1.7 (1.0–3.1)	1.9 (1.1–3.5)^a
Daily smoking + obese	1.3 (0.4–4.2)	2.1 (0.8–5.9)
Low physical activity + obese	1.3 (0.6–2.8)	2.4 (1.3–4.5)^a
Daily smoking + low physical activity + obese	0.9 (0.2–3.6)	5.1 (2.2–12.2)^a

BMI: body mass index; CI: confidence interval.

The CI does not overlap the null value (odds ratio = 1.0). In two instances for chronic headache without medication overuse, a confidence limit <1.0 was rounded up to 1.0.

Analyses were weighted for stratified sampling and nonresponse. Respondents with missing data on any of the three healthy lifestyle behaviour variables or possible confounding variables were excluded. Odds ratios were computed from logistic regression analysis adjusted for age as a continuous variable, Western ethnicity (yes/no), civil status, living alone (yes/no), children at home (yes/no), education, work status, income, and self-report of the following chronic illnesses (yes/no): osteoarthritis, rheumatoid arthritis, discus prolapse/back pain, and short- and long-term psychiatric illness. Global test for all models, p < 0.0001.

The role of stress as an effect modifier in the relationship between healthy lifestyle behaviours and headache was explored by analysing associations according to stress levels (Table 5). The odds for both MOH and CHnoO were significantly higher in the context of high stress. However, in the MOH group, associations were strongest for those with a harmful health behaviour (daily smoking, physical inactivity or obesity) plus high stress.

Table 5.

Odds ratios for chronic headache with and without medication overuse according to stress levels combined with daily smoking, low physical activity, or obesity.

Health behaviour	Chronic headache without medication overuse, odds ratio (95% CI)^a		Medication-overuse headache, odds ratio (95% CI)^a
Low stress^b	High stress^c	Low stress^b	High stress^c
Females
Daily smoking
No	1.0 (reference)	2.8 (2.1–3.7)^f	1.0 (reference)	2.7 (2.1–3.6)^f
Yes	1.2 (0.8–1.7)	2.0 (1.2–3.3)^f	1.5 (1.1–2.1)^f	3.8 (2.7–5.4)^f
Low physical activity^d
No	1.0 (reference)	2.5 (1.8–3.5)^f	1.0 (reference)	2.8 (2.0–3.9)^f
Yes	1.1 (0.8–1.6)	2.6 (1.8–3.9)^f	1.6 (1.2–2.2)^f	3.5 (2.5–5.0)^f
Obese^e
No	1.0 (reference)	2.6 (1.9–3.5)^f	1.0 (reference)	2.7 (2.0–3.5)^f
Yes	1.3 (0.8–2.0)	2.7 (1.6–4.4)^f	1.4 (1.0–2.0)	2.9 (1.9–4.5)^f
Males
Daily smoking
No	1.0 (reference)	3.4 (2.2–5.3)^f	1.0 (reference)	4.1 (2.6–6.7)^f
Yes	1.4 (0.9–2.2)	3.0 (1.7–5.3)^f	1.7 (1.0–2.7)^f	5.6 (3.2–9.6)^f
Low physical activity^d
No	1.0 (reference)	3.0 (1.9–4.9)^f	1.0 (reference)	3.6 (2.9–7.1)^f
Yes	1.1 (0.6–1.7)	2.9 (1.7–5.0)^f	1.5 (0.9–2.4)	5.0 (2.9–8.5)^f
Obese^e
No	1.0 (reference)	2.8 (1.8–4.4)^f	1.0 (reference)	3.5 (2.2–5.6)^f
Yes	1.1 (0.7–1.9)	4.3 (2.3–7.9)^f	1.3 (0.8–2.4)	4.9 (2.8–8.5)^f

CHnoO: chronic headache without medication overuse; MOH: medication-overuse headache; CI: 95% confidence interval.

Analyses were weighted for stratified sampling and nonresponse. Respondents with missing data on the health behaviour variables or confounding variables were excluded. Global test for all the models, p < 0.0001.

Odds ratios were computed for each health risk behaviour using those with no health risk behaviour and low stress as the reference group (OR 1.0). Logistic regression analyses were adjusted for sociodemographic variables and self-report of the following chronic illnesses (yes/no): osteoarthritis, rheumatoid arthritis, discus prolapse/back pain, and short- and long-term psychiatric illness. ^bLow stress refers to being in the first to fourth stress quintiles based on Perceived Stress Scale scores. ^cHigh stress refers to being in the highest stress quintile based on Perceived Stress Scale scores. ^dPhysical activity was adjusted for daily smoking and obesity in addition to sociodemographic factors and comorbid illness. ^eObesity was adjusted for physical activity in addition to sociodemographic factors and comorbid illness. ^fThe confidence interval does not overlap the null value (odds ratio = 1.0).

The odds for MOH were particularly strong among daily smokers with high stress vs. non-smokers with low stress (four-fold higher in women and six-fold higher in men). There was a similar trend for the physically inactive with high stress vs. the physically active with low stress (four-fold higher odds for MOH in women and five-fold higher in men). ORs for MOH were three-fold higher in women and five-fold higher in men who were obese with high stress compared to the non-obese with low stress.

In the CHnoO group, an additive or synergistic effect of obesity plus stress was seen in men, but not clearly for women, and not for combinations of high stress plus daily smoking or physical inactivity.

Discussion

People with MOH had different healthy lifestyle behaviour and stress profiles compared to people with CHnoO. Daily smoking, physical inactivity, and obesity were associated with MOH independently and in combination; and these associations were particularly strong in the context of high stress. In contrast, CHnoO appeared to be strongly associated with stress but not with healthy lifestyle behaviours.

In this cross-sectional study, it is not possible to determine the direction of causality: whether headache is the result or cause of medication overuse. Neither is it possible to determine which occurred first: headache, harmful health behaviour or stress. Nonetheless, healthy lifestyle behaviour and stress are modifiable; and targeting these factors may be key to preventing MOH.

Associations with stress, smoking, low physical activity and obesity

The relationship between CH and high stress is consistent with the findings of a longitudinal study that described increase in headache days for every 10-point increase in stress measured on a visual analogue scale (16). In the present study, which used a validated stress questionnaire in a large population, associations appeared to follow a ‘dose-response’ pattern, supporting previous findings that higher ‘exposure’ to stress increases the odds for headache (14) and analgesic use for headache (15). Associations with stress remained strong after adjusting for sociodemographic variables and comorbid illness, suggesting that the relationship is probably independent of these factors.

The association with smoking has been observed in other studies on frequent headache (2,4,34) and MOH (5,6) although this is not a consistent finding (7). It is probable that the association between smoking and CH is mediated by stress, especially in the subset with MOH.

Studies on the relationship between MOH and overweight/obesity showed conflicting results (5,6). The association may depend on the predominant type of primary headache in the population, i.e. stronger with chronic migraine (2,3,35). The association might also depend on the presence of other harmful health behaviours or underlying psychological distress (19) which may be precipitating headaches.

The observed link between MOH and physical inactivity are in line with previous observations (5). People with MOH might adopt sedentary lifestyles to avoid migraine attacks. On the other hand, physical inactivity could be a risk factor for non-migraine headache (36) and intake of more medication.

No associations with excessive drinking or illicit drug use

A longitudinal study found no elevated risk for MOH among those with a CAGE score of ≥1 (5). We found no significant association between MOH and a CAGE score of ≥2, a highly sensitive cut-off value (0.94) for detecting problematic alcohol use (31).

The relationship between excessive alcohol intake and headache is complex and studies show conflicting results (10). People with a predisposition to frequent headache tend to avoid alcohol because it is a headache trigger (2,4,5,6,14,34). On the other hand, chronic overusers of alcohol may suffer from alcohol-induced ‘hangover’ headache next day (34). People with chronic tension-type headache may be more likely to drink excessively than those with chronic migraine (2).

Our finding of no association between MOH and illicit drug use (among young adults) should be viewed with care since data were not gathered from middle-aged adults who had the highest prevalence of MOH. Furthermore, the majority of MOH cases were due to OTC analgesic overuse whereas drug-dependent behaviour is more likely among people who overuse prescription analgesics with addictive properties.

Evidence for a central dysfunction?

In examining the relationships between smoking, obesity and MOH in a population-based study, Straube et al. (6) considered the possibility of a frontal dysfunction in people with MOH. In clinic-based research, there is growing evidence for structural and functional brain abnormalities in relation to headache and medication overuse: orbitofrontal hypometabolism (37), reversible dysfunction in the ventromedial prefrontal cortex (38), changes in grey matter volumes in several brain areas including the midbrain and orbitofrontal cortex (39,40), and alterations in resting functional connectivity (40).

It is not clear which of these structural and functional changes in the brain are the result or cause of MOH. Pain processing and pain modulation alterations occur in chronic pain states. However, changes related to pain catastrophising, dependence-related behaviour, and a ‘drive’ toward increasing pain medication consumption could be specific to MOH (40). We could speculate that these processes manifest as maladaptive behavioural responses among people with MOH, observed in this study as higher perceived stress and unhealthy lifestyle choices.

Strengths and limitations

The advantages of using data from the DNHS 2010 were: largest sample size used for an MOH study, representative sampling, and use of validated questionnaires for stress, physical activity, and alcohol drinking. Respondents were unlikely to have been selected according to headache symptoms because the DNHS inquired about many chronic illnesses. Data from the DNHS could be linked at the individual level to the DNPR and registers with reliable sociodemographic data. Data on prescription medication purchases were not affected by problems of recall.

There were some limitations, foremost being that in this cross-sectional study, it is not possible to infer the direction of causality among the variables examined. Detailed headache histories and clinical examinations were not obtained so primary headache diagnoses could not be made. Reports on healthy lifestyle behaviours could be subject to social desirability bias. Nearly all young adult participants (97%) answered the questions on illicit drug use but there was no subsequent validation of responses.

Chronic pain conditions listed in the DNHS did not include gastrointestinal complaints, which have also been associated with MOH (5). Psychiatric illness was not characterised by validated questionnaires. It was beyond the scope of the study to check self-reports of illness against patient registers.

Sources of stress were not explored and could be specific to this population. An earlier large-scale population-based study in Denmark found that low education, heavy smoking, lack of physical exercise, small social network, and poor working conditions were associated with high perceived stress. There were also sex differences: living in a neighbourhood with low average education was associated with stress in women; while living in a neighbourhood with high crime rates and low ethnic diversity was associated with stress in men (41).

Stress from traumatic life events (13,42) and economic/work-related problems (4,42) have been found to be closely associated with headache. In this study population, stress from separation or divorce might explain why MOH is significantly more common in this civil status category compared to others; although it should be pointed out that there are no significant differences in prevalence when comparing current partnership status (living alone or with partner); and whether a person with MOH has children living at home (25). It has been suggested that day-to-day variations in stress – rather than an ‘average’ level of high stress – may be more predictive of headache onset (12). This was not explored in this cross-sectional study.

Given the 53.1% participation, non-response bias was minimised by adjusting for the effects of demographic, socioeconomic and health status variables on non-response rates (23). Differences in health service access according to socioeconomic position are minimal in the Danish context because health care is universal and largely free.

Clinical implications

Although this cross-sectional study cannot determine whether harmful health behaviours and stress are antecedents or consequences of headache, it is highly probable that these are modifiable risk factors for headache chronification and medication overuse. Reducing stress could help prevent both CHnoO and MOH. The benefit might be greater if stress reduction is coupled with changing health behaviour among people with MOH.

In the primary care setting where health behaviour is routinely discussed, advice against medication overuse for headache may be enough to prevent and treat MOH (43). However, attempts to make healthy lifestyle changes, including reduction of medication overuse, could be limited by existing external stressors or psychological distress which must also be addressed (22).

In the tertiary care setting, close contact with health care providers during detoxification could provide the structure for healthy lifestyle changes in general, not just in terms of rational medication use. The relationship between high stress and the tendency to engage in harmful health behaviours should be examined closely in patients with MOH.

Conclusion

Healthy lifestyle behaviour and stress should be assessed in all patients presenting with CH. There is a need to understand relationships between behaviour and the development of MOH, with high stress being a possible effect modifier in this relationship. The majority of patients with MOH may require assistance with healthy lifestyle changes and stress reduction. These should be considered in the management protocol for patients with MOH.

Clinical implications

Healthy lifestyle behaviour and stress should be assessed in all patients presenting with chronic headache.

People with chronic headache who smoke daily, are physically inactive or obese may be more likely to overuse headache medications, particularly in the context of high stress.

Stress reduction and health behaviour change are highly relevant in the management of medication-overuse headache.

Footnotes

Conflicts of interest

MLW has received grants from Pfizer, Lundbeck Fund and Allergan to present results of related studies at international conferences. RHJ has given lectures for Pfizer, Berlin-Chemie, Allergan, Merck, and Autonomic Technologies; is a member of the advisory boards of Autonomic Technologies, Medotech, and Neurocore; and is co-director of Lifting The Burden and EHMTIC. CG and EHH have nothing to declare.

Funding

The Region Zealand Health Profile 2010 and The Capital Region Health Profile 2010 were funded by The Region Zealand and The Capital Region of Denmark. MLW received research grants from IMK Almene Fond, and the patient organization Migræne og Hovedpineforeningen in connection with this study. Funding sources were not involved in writing this article nor in the decision to submit this work for publication. This study did not receive corporate sponsorship.

Acknowledgements

The authors thank the Research Centre for Prevention and Health, especially data manager Carsten Agger, and the team responsible for the Danish National Health Survey; Inger Helt Poulsen for facilitating use of data from Region Zealand; Statistics Denmark for management of the raw data; and everyone who responded to the surveys.

References

Saper

Hamel

Lake

AE III

. Medication overuse headache (MOH) is a biobehavioural disorder. Cephalalgia 2005; 25: 545–546.

Schramm

Obermann

Katsarava

. Epidemiological profiles of patients with chronic migraine and chronic tension-type headache. J Headache Pain 2013; 14: 40–40.

Bigal

Rapoport

. Obesity and chronic daily headache. Curr Pain Headache Rep 2012; 16: 101–109.

Molarius

Tegelberg

Ohrvik

. Socio-economic factors, lifestyle, and headache disorders – A population-based study in Sweden. Headache 2008; 48: 1426–1437.

Hagen

Linde

Steiner

. Risk factors for medication-overuse headache: An 11-year follow-up study. The Nord-Trøndelag Health Studies. Pain 2012; 153: 56–61.

Straube

Pfaffenrath

Ladwig

. Prevalence of chronic migraine and medication overuse headache in Germany – the German DMKG headache study. Cephalalgia 2010; 30: 207–213.

Wiendels

van Haestregt

Knuistingh Neven

. Chronic frequent headache in the general population: Prevalence and associated factors. Cephalalgia 2006; 26: 1434–1442.

Loef

Walach

. The combined effects of healthy lifestyle behaviors on all-cause mortality: A systematic review and meta-analysis. Prev Med 2012; 55: 163–170.

Martin-Diener

Meyer

Braun

. The combined effect on survival of four main behavioural risk factors for non-communicable diseases. Prev Med 2014; 65: 148–152.

10.

Winter

Hoffman

Meisinger

. Association between lifestyle factors and headache. J Headache Pain 2011; 12: 147–155.

11.

Robberstad

Dyb

Hagen

. An unfavorable lifestyle and recurrent headaches among adolescents: The HUNT Study. Neurology 2010; 75: 712–717.

12.

Lipton

Buse

Hall

. Reduction in perceived stress as a migraine trigger: Testing the “let-down headache” hypothesis. Neurology 2014; 82: 1395–1401.

13.

Stensland

SØ

Dyb

Thoresen

. Potentially traumatic interpersonal events, psychological distress and recurrent headache in a population-based cohort of adolescents: The HUNT study. BMJ Open 2013; 3: e002997–e002997.

14.

Yokoyama

Funazu

. Associations between headache and stress, alcohol drinking, exercise, sleep, and comorbid health conditions in a Japanese population. J Headache Pain 2009; 10: 177–185.

15.

Koushede

Holstein

Andersen

. Stress and medicine use for headache: Does sense of coherence modify the association? Eur J Pub Health 2010; 21: 656–661.

16.

Schramm SH, Moebus S, Lehmann N, et al. The association between stress and headache: A longitudinal population-based study. Cephalalgia. 2015; 35: 853–863.

17.

Dube

Caraballo

Dhingra

. The relationship between smoking status and serious psychological distress: Findings from the 2007 Behavioral Risk Factor Surveillance System. Int J Public Health 2009; 54(Suppl 1): 68–74.

18.

Okoro

Stoodt

Rohrer

. Physical activity patterns among U.S. adults with and without serious psychological distress. Public Health Rep 2014; 129: 30–38.

19.

Zhao

Ford

. Serious psychological distress and its associations with body mass index: Findings from the 2007 Behavioral Risk Factor Surveillance System. Int J Public Health 2009; 54(Suppl 1): S30–S36.

20.

Boden

Fergusson

Horwood

. Associations between exposure to stressful life events and alcohol use disorder in a longitudinal birth cohort studied to age 30. Drug Alcohol Depend 2014; 142: 154–160.

21.

Moitra

Anderson

Stein

. Perceived stress and substance use in methadone-maintained smokers. Drug Alcohol Depend 2013; 133: 785–788.

22.

McKenzie

Harris

. Understanding the relationship between stress, distress and healthy lifestyle behaviour: A qualitative study of patients and general practitioners. BMC Fam Pract 2013; 14: 166–166.

23.

Christensen

Ekholm

Glümer

. The Danish National Health Survey 2010. Study design and respondent characteristics. Scand J Public Health 2012; 40: 391–397.

24.

Pedersen

. The Danish Civil Registration System. Scand J Public Health 2011; 39 (Suppl 7): 22–25.

25.

Westergaard

Glümer

Hansen

. Prevalence of chronic headache with and without medication overuse: Associations with socioeconomic position and physical and mental health status. Pain 2014; 155: 2005–2013.

26.

Kildemoes

Sørensen

Hallas

. The Danish National Prescription Registry. Scand J Public Health 2011; 39(Suppl 7): 38–41.

27.

Thygesen

Daasnes

Thaulow

. Introduction to Danish (nationwide) registers on health and social issues: Structure, access, legislation, and archiving. Scand J Public Health 2011; 39 (Suppl 7): 12–16.

28.

Headache Classification Committee of the International Headache Society. The International Classification of Headache Disorders, 3rd edition (beta version). Cephalalgia 2013; 33: 629–808.

29.

WHO Collaborating Centre for Drug Statistics Methodology. Use of ATC/DDD, www.whocc.no/use_of_atc_ddd (2011, accessed 11 February 2015).

30.

Aadahl

Jørgensen

. Validation of a new self-report instrument for measuring physical activity. Med Sci Sports Exerc 2003; 35: 1196–1202.

31.

Zierau

Hardt

Henriksen

. Validation of a self-administered modified CAGE test (CAGE-C) in a somatic hospital ward: Comparison with biochemical markers. Scand J Clin Lab Invest 2005; 65: 615–622.

32.

Cohen

Kamarck

Memelstein

. A global measure of perceived stress. J Health Soc Behav 1983; 24: 385–396.

33.

Olesen

Mortensen

Bech

. Prevalence of major depression and stress indicators in the Danish general population. Acta Psychiatr Scand 2004; 109: 96–103.

34.

Aamodt

Stovner

Hagen

. Headache prevalence related to smoking and alcohol use. The Head-HUNT study. Eur J Neurol 2006; 13: 1233–1238.

35.

Chai

Scher

Moghekar

. Obesity and headache: Part I – A systematic review of the epidemiology of obesity and headache. Headache 2014; 54: 219–234.

36.

Varkey

Hagen

Zwart

. Physical activity and headache: Results from the Nord-Trøndelag Health Study (HUNT). Cephalalgia 2008; 28: 1292–1297.

37.

Fumal

Laureys

Di Clemente

. Orbitofrontal cortex involvement in chronic analgesic-overuse headache evolving from episodic migraine. Brain 2006; 129: 543–550.

38.

Ferraro

Grazzi

Muffatti

. In medication-overuse headache, fMRI shows long-lasting dysfunction in midbrain areas. Headache 2012; 52: 1520–1534.

39.

Riederer

Gantenbein

Marti

. Decrease of gray matter volume in the midbrain is associated with treatment response in medication-overuse headache: Possible influence of orbitofrontal cortex. J Neurosci 2013; 33: 15343–15349.

40.

Chanraud

Di Scala

Dilharreguy

. Brain functional connectivity and morphology changes in medication-overuse headache: Clue for dependence-related processes? Cephalalgia 2014; 34: 605–615.

41.

Nielsen

Curtis

Kristensen

. What characterizes persons with high levels of perceived stress in Denmark? A national representative study. Scand J Public Health 2008; 36: 369–379.

42.

Rueda-Sánchez

Díaz-Martínez

. Prevalence and associated factors for episodic and chronic daily headache in the Colombian population. Cephalalgia 2008; 28: 216–225.

43.

Kristoffersen ES, Straand J, Vetvik KG, et al. Brief intervention for medication-overuse headache in primary care. The BIMOH study: A double-blind pragmatic cluster randomized parallel controlled trial. J Neurol Neurosurg Psychiatry. Epub ahead of print 11 August 2014. DOI:10.1136/jnnp-2014-308548.