08954356/90$3.00+ 0.00 Copyright 0 1990Pergamon Press plc

J Clto EpidemiolVol. 43, No. 9, PP. 983-994, 1990 Printed in Great Britain. All rights reserved

THE RELATIONSHIP OF HEADACHE SYMPTOMS SEVERITY AND DURATION OF ATTACKS DAVID D. CELENTANO,’ WALTER F. STEWART’

WITH

and MARTHA S. LJNET~

‘Division of Behavioral Sciences and Health Education, and ZDepartment of Epidemiology, School of Hygiene and public Health, The Johns Hopkins University, 624 North Broadway, Baltimore, MD 21205 and )Analytic Studies Section, Biostatistics Branch, Epidemiology and Biostatistics Program, Division of Cancer Etiology, National Cancer Institute, Bethesda, MD 20892, U.S.A. (Received in revised form

18 September

1989; received for publication

21 February 1990)

Abstract-Efforts to develop clinically useful headache classification schemes have generally focused on linking specific symptom groupings with specific headache subtypes. An alternative conceptual approach, the “severity model” of headache, considers a continuum of headache ranging from mild to severe forms with specific headache subtypes distinguished by level of severity rather than unique constellations of symptoms. A population-based telephone interview was carried out among 10,169 subjects aged 12-29 to estimate the prevalence of serious headaches and better characterize symptoms that accompany headache attacks. In an analysis of frequency of occurrence, pain and duration of recent (within 4 weeks prior to interview) headache attacks, the data revealed that common symptoms (such as forehead pain and pain in the back of the head, neck and shoulders) were reported frequently, but headaches with these symptoms were generally characterized by low levels of pain and short duration. Although not an original study objective, the data were analyzed to determine whether distinct symptom constellations could be identified or whether symptoms overlapped between headache types. Symptoms of migraine were frequently experienced concomitant with tension-type symptoms; the resultant headaches were usually characterized as moderate in intensity. In contrast, symptoms usually associated with migraine in the absence of concomitant tension-type symptoms were infrequently experienced, but resulted in headaches causing the greatest disability. The data provide some support for the severity model of headache. Headache sification

Migraine headache

Tension-type

INTRODUCTION

headache

Symptoms

Clas-

ties [2]. Descriptions of the epidemiologic features of headache become problematic when different classifications or definitions are used.

A variety

of symptoms are commonly associated with severe or disabling headache, one of the chief complaints and leading reasons for medical care visits [l]. Despite a number of efforts to develop clinically useful classification schemes for headache, there is substantial controversy concerning the specific criteria that differentiate the various types and whether headaches characterized by different symptom constellations are, in fact, distinct clinical enti-

Several general approaches have been used to categorize headache types including clinical consensus [3,4], statistical approaches [5,6], and distinctions based on epidemiologic parameters [7-111. Each of these approaches, however, has methodological problems. Classifications derived from clinical consensus and statistical algorithms have utilized data almost exclusively from headache clinic or practice 983

984

DAVID D. C%JSITANOet al.

patients [12]. Headache types comprising such classification schemes represent both the manner in which symptoms cluster as well as factors motivating individuals to seek care [2]. On the other hand, categorizations derived from population samples have generally been based on relatively few reported headache symptoms with little attention to important aspects of headache such as intensity and duration, indicators of the severity of attack [13]. Recently, a number of investigators have emphasized the “severity model” or psychobiological model of headache rather than classifications employing numerous categories [14-l 61. The severity model, which postulates that “migraine” is at the severe end of a headache continuum ranging from mild (forehead pain only easily relieved by analgesics) to severe and disabling [17], represents both a different philosophical approach to headache as well as a possible solution to the problem of nosology. Proponents of this model assert that the variants of migraine, per se, do not constitute distinct headache entities. Featherstone [ 171 summarizes the evidence supporting a headache continuum. First, both migraine and tension-type headache (muscle contraction) symptoms independently correlate more strongly with severity than with groupings of defined symptom configurations [16, 181; e.g. Featherstone cites unilaterality and throbbing as severe and characteristic of both types of headache [19,20]. Second, pathophysiological phenomena, such as platelet aggregation and vasomotor disturbances, are the same for both migraine and tension-type headache [21-231. Third, few differences in sensitivity to pharmaceutical agents have been demonstrated in studies of treatment for the two types of headache [24-271, although ergot compounds and calcium channel blockers are more effective in migraine than tension-type headaches [21,28]. No treatment differences are noted in the two headache types using biofeedback techniques [21,23,29]. Despite this evidence, to date the headache continuum model has not been assessed using data from random population samples. In 1986-87 we carried out a population-based telephone interview survey to estimate the prevalence of serious headaches among 10,169 subjects aged 12-29 and to describe in detail headache symptoms, pain, duration, and other characteristics of each subject’s most recent headache attack. Although not among the original study objectives, in this report the data were

analyzed to determine whether symptoms clustered into distinct entities commonly referred to as specific types of headache (e.g. migraine, tension-type) or whether symptoms overlapped between headache types. We evaluated the interrelationship among nine headache symptoms accompanying each subject’s most recent attack within 4 weeks of interview, particularly focusing on the extent to which migraine headache symptoms were also experienced by individuals with symptoms of tension-type headache. We also analyzed the relationship of specific headache symptoms occurring in an attack with the duration and estimated pain for that attack to ascertain symptom clusters predictive of disabling headache. A central question was whether symptoms characteristic of migraine headache were associated with headaches of greater severity and duration than those symptoms usually linked with tension-type headache. To examine the type vs continuum models of headache classification, if headache symptoms defining a specific headache type (e.g. migraine headache) clustered together and were not associated with symptoms defining another headache type (e.g. tension-type headache), one could conclude that classifications based on symptom constellations were justified. However if the data demonstrate that symptoms overlap between headache types, then the data would support the severity model of headache. While there are no published criteria specifying precise cut-off points for significance of symptom overlap, our data can be examined for trends. Recent analytic approaches for assessing whether or not symptom groupings are statistically significant generally require paired data or correlational approaches. However, multidimensional scaling techniques are also particularly well suited for the present analysis. Multidimensional scaling [30] can be used to examine how symptoms cluster together while simultaneously considering each symptom’s relationships to all other symptoms. If this analysis reveals clusters of symptoms which conform to clinically described entities (such as the classification scheme proposed by the Ad Hoc Committee on Classification of Headache [3]), it is reasonable to conclude that the severity model is not supported; alternatively, if there are no distinctive clusters of symptoms and if extensive overlap of tension-type and migraine symptoms is observed, then the severity model may be a more accurate explanation for the observed data.

Headache Symptoms

A final important issue which we hoped to address as we examined the relationships of specific symptoms with reported pain and duration of the most recent headache attack, was the clinical relevance of these (competing) classification schemes. If the severity hypothesis is true, we would anticipate symptoms of migraine with aura to demonstrate the highest levels of subjective pain and prolonged duration, with other migrainous symptoms being somewhat less severe, followed by symptoms of tension-type headache, with simple forehead pain only attacks the least severe headache type. MATERIALS

AND METHODS

In 1986-1987 we conducted a populationbased telephone survey of headache among residents ages 12-29 in Washington County, Maryland. The age group selected for study was chosen based on literature reports of the peak age of onset of migraine headache symptoms [13]. Using random digit dialing [31], 10,169 Washington County residents ages 12-29 years were interviewed, representing approximately half of the county residents in that age group. The response rate was 94.2%; only 4.3% of eligible subjects refused, and we were unable to contact the remainder [32]. The high response rate suggests that the respondents were representative of the study population. The investigators constructed a standardized questionnaire to elicit information about subjects’ headache histories and to inquire in depth about each subject’s most recent headache attack that occurred in the prior 4 weeks. Using six survey instruments employed in previous studies of headache as a guide [8, 16,33-361, we inquired about lifetime and recent occurrence of an array of symptoms associated with headache. We report on 9 such symptoms, selected because some have been strongly linked with migraine (e.g. accompanying nausea and unilaterality of pain) while others have been associated primarily with tension-type attacks (e.g. pain in the back of the head, neck and shoulders). All interviews were conducted by trained female interviewers using a closed-ended questioning approach. In the 15 minute interview, information was obtained about history of headache, details about headaches occurring in the 4 weeks prior to interview, including estimates of the duration and severity of the most recent headache attack, as well as a history of specific symptoms associ-

985

ated with headache not due to a blow to the head or a cold or influenza. Among the symptoms ascertained were (with the abbreviated symptoms given in parentheses): nausea and/or vomiting accompanying a headache (nausea); numbness and tingling in one arm or leg during a headache (numbness); visual abnormalities appearing as “spots, lines and heat waves” prior to a headache (scotoma); partial loss of vision before a headache (vision loss); pain on one side of the head (unilateral); headache brought on by exposure to light (light induce); light making the headache worse (photophobia); pain in the back of the head, neck and shoulders (neck pain); and a feeling of a tight band around the head (tight band). Information obtained about additional symptoms was not used in the present analysis because these symptoms are infrequently used in making headache type diagnoses [3]. The nosology to be used in presentation of the findings is based on the recently published diagnostic and classification system of the International Headache Society [3]. Thus, the terms migraine with and without aura replace the categories of classic and common migraine and tension-type headache is used instead of the former designation of muscle-contraction headache. Aura symptoms (commonly referred to as prodromal or premonitory) include neurological phenomena of the types listed above. Tension-type headaches are accompanied by bilateral pain, pain occurring in muscles of the back of the head, neck and shoulders, and the frequently cited sign of “a feeling of a tight band” around the head. Self-reported subjective experience of the pain associated with the subject’s most recent headache was assessed by asking the respondent to rate the accompanying headache pain “on a scale from 1 to 10, with 1 being very mild pain and 10 being very severe pain, such as slamming your fingers in a car door.” Finally, for the same headache we inquired as to how long (in minutes and hours) the headache lasted. A total of 6347 respondents (2496 males and 3851 females), reported their most recent headache occurring within 4 weeks of interview (62.4% of all subjects interviewed). In the analysis, for each symptom, the respondents reporting occurrence of that symptom with their most recent headache are compared with those who did not experience that symptom with their last headache. The extent of overlap of symptoms which occurred with this headache was analyzed using two methods. First, to determine the joint

DAVIDD. CELENTANO et al.

986

occurrence of pairs of symptoms with the most recent headache, odds ratios were calculated. A relative odds of 2.0 means that given the presence of one symptom, the occurrence of the second symptom is twice as likely to be reported than among individuals who did not experience the first symptom. Because the reports of symptoms are all binary (present/absent), the statistical requirements for parametric analyses (e.g. multinormal distribution) precluded strategies based on correlations or commonly used clustering algorithms. Thus, the second approach utilized a non-parametric multidimensional scaling procedure to simultaneously consider the joint occurrence of all 9 symptoms. We used the matrix of odds ratios to determine a Euclidean distance model solution. The fit of mutual distances between all symptoms was obtained, in which small distance denotes a high degree of association. Next, the dimensionality of the space and the projection of the points on axes of the space was determined. An iterative solution was used to represent a proximity clustering solution in n dimensional space [30]. The output is displayed separately for males and females as graphical representations of the relationship between the headache symptoms. Both two and three-dimensional models were estimated; only the three-dimensional solutions are presented, as they better fit the data. Proponents of the headache continuum hypothesis argue that headaches vary in the pain and disability they cause differentially by symptoms rather than by clinical categorizations of headache types. To address this issue, we first analyze the estimated pain and duration of the subject’s most recent headache within the past 4 weeks separately by sex for single symptoms and then by pairs of headache symptoms. To simultaneously consider effects of all 9 symptoms on the level of disability associated with the most recent headache, ordinary least squares regression analyses for pain (on a scale ranging from 1 to 10) and duration (in hours) of the most recent headache are presented by the occurrence of the 9 headache symptoms, entered as independent variables separately for males and females, following adjustment for respondent age.

males and females who experienced that symptom with their most recent headache and the proportions who had ever had each symptom (indicating lifetime prevalence) are displayed in Fig. 1. Simple forehead pain was the most frequently cited symptom, accompanying half of the subjects’ most recent headaches (data not shown). Unilateral pain and pain in the back of the head, neck and upper shoulders (the latter generally associated with tension-type headache) were also common symptoms. Typical migraine symptoms occurred significantly less frequently. The only symptom with a notable sex difference was nausea and/or vomiting accompanying headache (described by 7% of men and 14% of women). There was no substantial difference by sex for neurological symptoms, including: numbness and tingling of one arm or leg during

70,

(a)

60

50 $ Q4o 5 r 30 !J! 20 10 0

Headache

symptoms

(males)

Fig. I(a). Percentage distributions of headache symptoms with most recent headache and lifetime prevalence.

RESULTS

Frequency

of recent headache symptoms

Recent headache (within the past 4 weeks) was reported by 57.1% of males and 76.5% of females. For each symptom the proportions of

Headache

symptoms

( females )

Fig. l(b). Percentage distributions of headache symptoms with most recent headache and lifetime prevalence.

Headache Symptoms

981

Table 1. Odds ratios of joint occurrence of symptoms associated with most recent headache in past 4 weeks Washington County resident males (n = 2509) (below diagonal) and females (n = 3873) (above diagonal) ages 12-29 years (1), (2) (3) (4) (5) (6) (7) (8) (9)

Nausea Numbness Scotoma Visual loss Unilateral Light induce Photophobia Neck pain Tight band

(1) 4.00 4.13 5.82 1.84 1.33 2.16 1.70 2.91

(2) 3.50 3.88 8.77 1.69 2.05 3.73 3.39 3.75

(3) 3.50 6.53 9.95 2.35 2.16 4.13 1.51 3.30

the most recent headache (2.6 and 2.8%, males and females, respectively), seeing “spots, lines and heat waves” prior to the headache (6.5 and 8.8%, respectively), and partial loss of vision before the attack (2.1% for each sex). Other symptoms, such as photophobia were also distributed equally among men and women. When joint occurrence of symptoms was assessed using relative odds ratios, a substantial level of concomitant occurrence of most headache symptoms was found (Table 1). Among males (entries below the diagonal), the neurological signs (numbness and tingling in one arm or limb, prodromal visual disturbances, and partial visual loss before headache) are strongly (and significantly) related to each other and to nausea and/or vomiting accompanying headache. For example, the relative odds for the joint occurrence of nausea and numbness (OR = 4.0) indicates that men who had nausea with their most recent headache were four times as likely as those without nausea to report numbness and/or tingling in one limb during the same headache attack. Each of the symptoms occurred significantly more frequently among those with another symptom (i.e. all values exceed 1.O with only three entries not achieving statistical significance). Some symptoms, however, appear to aggregate somewhat less commonly (indicated by an OR of less than 2.0). These include unilaterality, pain in the back of the head, neck and shoulders, and bright lights inducing a headache. However, for both males and females, overlap in symptoms is quite common, including substantial overlap of the migraine and tension-type headache symptoms. The results of the multidimensional scaling analysis for the co-occurrence of the 9 symptoms are displayed in Figs 2 and 3 for males and females, respectively. For men (Fig. 2) a generalized spatial clustering of migraine symptoms with tension-type symptoms is found. The symptoms of nausea and/or vomiting, headache

(4) 2.96 9.52 6.30 5.63 3.27 4.92 2.46 4.24

(5) 2.29 2.91 2.19 2.38 1.72 1.97 1.38 1.48

(6) 1.26 2.03 2.26 1.96 1.34 12.54 1.19 1.98

(7) 2.81 2.47 2.88 3.20 1.76 10.61 1.53 2.26

(8) 2.31 5.24 1.96 1.72

1.49 1.24 1.24 1.55

(9) 2.89 3.31 2.19 2.80 1.50 1.42 2.31 1.77 -

induced by bright lights, and photophobia occur in close proximity on the first two dimensions, X and Y. Both of the tension-type headache symptoms (pain in the back of the head, neck and shoulders and feeling of a tight band around the head) cluster together, but are also spatially linked with the cluster consisting of nausea and/or vomiting, lights induce the photophobia on the 2 dimension. Two symptoms are notable for their lack of association with other symptoms, namely, numbness and/or tingling in one arm or leg, and partial loss of vision preceding headache. The lack of separate clustering for the migraine and tension-type symptoms provides empiric confirmation for the severity or continuum model for headache and does not support the classification based on joint symptom occurrence, such as the older [4] and more recent clinical consensus categorization schemes [3]. Figure 3 displays the results of the scaling procedure for females, where a distinct clustering of the neurological symptoms characterizing headache can be seen (numbness and tingling, prodromal visual disturbances and partial loss of sight) being located close together on all three axes. Interestingly, a feeling of a tight band around the head is similarly located on two of the same dimensions. The only remaining symptom which might be characterized as clustering with the three neurologic symptoms is nausea and/or vomiting, being located in the same quadrant (i.e. sharing two dimensions) as the neurological signs and of the same magnitude on the third axis. These results also demonstrate the relative independence of the two photophobia symptoms and the lack of cluster group membership of unilaterality. Also notable is the uniqueness of pain in the back of the head, neck and shoulders, which is distinct from the other tension-type headache symptoms (tight band). The spatial clustering pattern observed for females, while different to that

988

DAVID

D.

CELENTANO

et

al.

6

Fig. 2. Multidimensional scaling of symptoms experienced with most recent headache among Washington County males aged 12-29 years. 1, nausea & vomiting; 2, numbness in limb; 3, visual scotoma; 4, partial visual loss; 5, unilateral pain; 6, light induce; 7, photophobia; 8, muscle pain; 9, tight band.

found for males, also suggests that certain migraine symptoms appear to cluster on two of the same dimensions with one of the tensiontype symptoms.

Pain associated

with recent headache

symptoms

Table 2 displays the pain estimates according to single and paired headache symptoms. Women report on average pain level of 4.9, vs

I

6

2.0

f 70

6

1

9

0.5

f

4

2

1

-1.0

-

2.00

---_

---_ -2.5

.. X

-3

Fig. 3. Multidimensional scaling of symptoms experienced with most recent headache among Washington County females aged 12-29 years. 1, nausea & vomiting; 2, numbness in limb; 3, visual scotoma; 4, partial visual loss; 5, unilateral pain; 6, light induce; 7, photophobia; 8, muscle pain; 9, tight band.

HeadacheSymptoms

989

Table2. Estimatesof painassociatedwithsymptoms(single and joint) associated with most recentheadachein past 4 weeks; Washington County residents, by gender (2509 males; 3873 females). Subjective estimate of pain M F

(1) 5.9 6.5

(2) 6.2 7.0

(2) Numbness

M F

7.1 7.8

-

(3) Visual spots

M F

6.6 7.2

6.6 7.7

-

(4) Scotoma

M F

7.1 7.9

7.7 8.4

6.4 6.9

-

(5) Unilateral

M F

6.8 6.9

7.0 7.2

5.9 6.7

6.6 6.9

-

(6) Light induce

M F

6.2 6.8

7.4 7.7

5.8 6.6

5.9 6.6

6.1 5.8

-

(7) Photophobia

M F

6.6 7.1

7.3 7.8

5.8 6.8

6.2 7.4

6.3 6.2

6.6 5.6

-

(8) Neck pain

M F

6.5 6.9

6.6 7.2

6.4 7.0

6.8 6.5

5.3 6.2

5.6 6.0

6.2 6.2

-

(9) Tight band

M F

6.2 7.3

6.8 7.6

6.1 7.1

6.7 7.2

8.9 6.4

6.6 6.2

6.6 6.6

5.8 5.8

(1) Nausea

yi 6.2

an average pain level of 4.5 for men. For both males and females, the most severe headaches were those accompanied by numbness and tingling during the attack. Other symptoms associated with particularly severe headaches were nausea and/or vomiting and partial loss of vision preceding the headache. By contrast, lower levels of pain were linked with pain in the back of the head, neck, or shoulders and a feeling of a tight band around the head (symptoms more commonly associated with tension-type headaches). In general, symptoms accompanying headaches characterized by the highest pain levels (e.g. those with neurological manifestations) generally demonstrated linear increasing trends by age for both males and females (data not shown). Age-adjustment was therefore employed in the regression analyses. The reported levels of pain varied according to the occurrence of pairs of symptoms (Table 2). Among males, the reported pain of the last headache is always higher when pairs of symptoms are experienced than when only single symptoms were reported. For example, when individuals report both nausea and/or vomiting during a headache (zero-order pain level of 5.9) and numbness and/or tingling of one arm or leg (with a value of 6.2), the reported pain increases to a level of 7.1, a substantially higher value. The highest pain levels are seen for nausea and/or vomiting accompanying a headache with concomitant numbness and/or tingling in limb or preceded by partial loss of vision, as well as

(4) 6.1 6.2

y; 5:l

y; 512

(7) 5.5 5.7

$ 5.5

(9) 5.6 5.9

-

for headaches characterized by numbness and/or tingling and partial visual loss, unilaterally, or photophobia (all with pain levels over 7.0). The highest pain levels were described by males who experienced unilateral headaches accompanied by a feeling of a tight band about the head. It is noteworthy that pain levels were substantially elevated for headaches characterized by pain in the back of the head, neck and shoulders and/or the feeling of a tight band around the head, when these tension-type symptoms occurred jointly with those symptoms traditionally attributed to migraine headaches (prodromal visual disturbances, nausea and vomiting, and unilaterality). Among women, almost identical results were observed. Substantially elevated pain levels were found for tension-type symptoms accompanying nausea and vomiting, as well as for numbness and tingling. Again, pain levels were at least one unit higher than those for the individual migraine or tension-type symptoms, when there was joint occurrence of migraine with tension-type symptoms. Duration of recent headache symptoms

Table 3 presents for males and females the average reported duration of headaches associated with the headache symptoms singly and jointly. For both males and females, the length of headache varied according to age (data not shown) and symptoms, from a headache duration of several minutes to those lasting several

DAVID D. CELENTANO et al.

990

Table 3. Comparison of average duration (in hours) associated with symptoms (single and joint) with most recenl : headache in past 4 weeks; Washington County residents, by gender (2509 males; 3873 females). Reported duration of most recent attack

(1)

(2)

M F

8.9 16.2

11.2 16.7

(2) Numbness

M F

14.1 18.2

-

(3) Scotoma

M F

10.8 19.6

17.8 17.2

-

(4) Visual loss

M F

10.3 24.2

24.7 18.4

16.0 13.2

-

(5) Unilateral

M F

9.1 18.2

9.5 19.1

13.4 16.3

9.6 16.8

-

(6) Light induce

M F

10.2 21.0

9.4 22.0

12.2 17.2

5.3 14.4

7.4 14.3

(7) Photophobia

M F

11.4 20.7

19.4 19.2

14.5 17.7

12.3 19.3

8.3 16.0

8.7 11.3

-

(8) Neck pain

M F

10.0 20.5

13.8 19.2

17.1 19.1

15.5 17.4

6.8 15.3

7.8 15.1

9.6 16.7

-

(9) Tight band

M F

10.2 18.0

15.8 21.2

12.1 17.8

11.7 16.9

9.5 13.9

7.9 15.3

11.3 15.0

10.5 16.3

(1) Nausea

(3) 11.3 14.0

days. The duration of headaches with both migrainous and tension-type symptoms generally increased with increasing age of the subjects, with the exception of unilaterally. For many symptoms, the duration of the most recent headache was nearly twice as long in the oldest respondents compared with adolescents. The average duration in hours of the most recent headache according to the joint occurrence of symptoms (Table 3) shows that for men, the symptoms that lead to higher estimates of pain, namely nausea and/or vomiting, numbness or tingling in one arm or leg, and partial visual loss with headache, were also found to be associated with longer headaches. Photophobia was also found to be associated with other symptoms in producing headaches of extended duration, as was pain in the back of the head, neck and shoulders, a symptom associated with tension-type headache. For females, a similar result was found, although the headaches were all of substantially longer duration than those reported by males. For women, a feeling of a tight band about the head (tension-type headache) accompanying other symptoms consistently increased headache duration to a greater extent than was found for men; headaches with pain in the back of the head, neck and shoulders and those with a sensation of a tight band around the head lasted anywhere from 13.9 to 21.2 hours. In no case did overlap of tension-type with migrainous symptoms lead to decreased duration of headache attack.

(4) 10.0 13.7

(5) 7.3 11.7

(7) (76)3 8.1 10:4 12.3

(8) 7.2 12.2

(9) 9.0 11.2

-

-

Regression analyses for pain and duration of recent headache

In Table 4 we consider the relation between the reported pain and duration of each symptom after adjusting for the co-occurrence of all other symptoms (and age). Among males, the factors contributing most strongly to elevated headache pain include all of the symptoms inquired about with the exception of light induced headache and partial visual loss. However, the overall explained variance was quite low for this analysis (R* = 0.03, p < O.Ol), and the results should be interpreted with caution. Both migraine and tension-type headache symptoms are found to exert significant and nearly equivalent effects on reported pain, even after adjusting for all other symptoms considered. For females, despite higher reported levels of headache pain, the same symptoms were also significantly associated with pain and the regression coefficients were also nearly identical. The overall explained variance in this model was much higher than that found for men (R* = 0.09, p < O.OOl), although the minimal dispersion in reported pain placed limits on the explained variation in these models. However, different results were found for the two sexes in the regression analyses of headache duration (of the most recent headache). Among men, nausea and vomiting, visual prodromal disturbances (symptoms classified as migrainous), and pain behind the head, neck or

Headache Symptoms

991

Table 4. Summary of regression analyses of symptoms associated with most recent headache pain (age-adjusted), by sex, Washington County, Maryland, 19861987* Females

Males b

Symptoms

Nausea and/or vomiting 1.06 Numbness and/or tingling of one arm or leg 0.98 Visual “spots, lines, heat waves” 0.55 Partial loss of vision 0.51 Unilateral pain 0.42 Light induced -0.02 Photophobia 0.64 Pain behind head. neck or shoulders 0.37 0.65 Tight band around head

SE(b) 0.16 0.26 0.16 0.25 0.09 0.13 0.12 0.10 0.11

r

6.82 3.80 3.38 1.94 4.50 -0.18 5.52 3.70 5.80

P

0.001 0.001 0.001 0.052 0.001 0.856 0.001 0.001 0.001

b

1.30 1.15 0.62 0.36 0.50 -0.07 0.64 0.35 0.63

SE(b) 0.09 0.20 0.11 0.22 0.07 0.10 0.09 0.08 0.09

I

13.76 5.82 5.36 1.53 7.03 -0.90 7.53 4.63 6.94

P

0.001 0.001 0.001 0.103 0.001 0.366 0.001 0.001 0.004

*n = 2508 males and 3872 females.

shoulders (tension-type headache) were independently related to longer duration of attack. None of the remaining 6 symptoms attained statistical significance; the overall equation was statistically significant (R2 = 0.12, p < 0.001). Among females (R2 = 0.19, p < O.OOl), 6 of the 9 symptoms were significantly linked with increased duration, including both migraine and tension-type symptoms. However, the tensiontype symptom of a feeling of a tight band around the head, previously found to be important when pairs of symptoms were considered, failed to attain statistical significance in the regression analysis. Overall, the explained variation was higher for duration than for pain, reflecting in part the increased dispersion of reported lengths of the most recent headache attacks. DISCUSSION

The results demonstrate significant overlap of occurrence of symptoms traditionally used to classify headaches into specific types. The results from multidimensional scaling clearly show that symptoms of migraine and tensiontype headache most commonly occur in combination during the subjects’ most recent headache attack, rather than showing distinct patterns of symptoms conforming to separate entities. This general pattern was found for both males and females, although the clustering was somewhat different between the sexes. These data, then, demonstrate support for the severity model of headache. In the present investigation, the symptoms generally associated with more severe headache, for both men and women, are those traditionally characterized as defining migraine headache [37-40]. However, in both the bivariate and regression analyses, elevated pain and duration

were also found to be linked with symptoms of tension-type headache. For all regression models evaluated, nausea and/or vomiting, prodroma1 scotoma, and the tension-type symptom of pain in the back of the head, neck and shoulders were strongly related to greater severity of headache pain and longer duration of the attack for males and females. The only other symptoms with nearly consistent elevations in reported disability were numbness and/or tingling in one limb, unilateral pain, and photophobia (each statistically significant in 3 of the 4 regression models), all symptoms of migraine headache. It is important to note that headache symptoms believed to characterize tension-type headache were also commonly experienced during the same headache attack as were symptoms generally associated with migraine headache. Significant determinants of headache pain intensity and duration included tension-type headache symptoms as well as migraine symptoms in both the bivariate and in the regression results. In the only other population based survey of headache which specifically addressed the prevalence and overlap of musculoskeletal, vascular and autonomic symptoms in headache [16], the investigators concluded that “increasingly severe and frequent headache attacks are likely accompanied by the increasing involvement of both migraine and tension-type headache symptoms, and not by a qualitative change in the types of symptoms experienced.” Our data do demonstrate some differences in symptom relationships with frequency, severity and duration of attacks according to the standard headache classifications. Migrainous symptoms were generally found to occur in headaches of longer duration and those characterized by higher levels of pain (especially among women), although there was some overlap observed with tension-type symptoms.

992

DAVID D. CELENTANO et

al.

Table 5. Summary of regression analyses of symptoms associated with most recent headache duration (age-adjusted), by sex, Washington County,-. Maryland, 1986-1987’ _ Males Svmntoms Nausea and/or vomiting Numbness and/or tingling of one arm or leg Visual “spots, lines, heat waves” Partial loss of vision Unilateral pain Light induced Photophobia Pain behind head, neck or shoulders Tight band around head

Females

b

SE Cbl

t

2.44 2.66 4.11 0.93 1.04 0.08 1.05 1.34 1.00

0.96 1.60 1.01 1.64 0.57 0.78 0.72 0.61 0.70

2.53 1.66 4.05 0.56 1.82 0.10 1.45 2.19 1.43

LJ

0.011 0.096 0.001 0.572 0.068 0.919 0.146 0.029 0.154

b

5.85 4.46 2.20 0.50 2.93 -0.44 3.03 2.97 1.17

SE(h)

0.77 1.61 0.94 1.83 0.58 0.79 0.70 0.63 0.74

t

n

7.57 2.76 2.33 0.27 5.02 0.56 4.33 4.75 1.58

0.001

8:E 0.784 0.001 0.574 0.001 0.001 0.114

*n = 2508 males and 3872 females.

Symptoms commonly cited as diagnostic of migraine headache (nausea and/or vomiting, unilaterality, visual disturbances and photophobia) were significantly associated with prolonged duration and elevated reports of pain. However, with the exception of unilateral pain, these symptoms occur relatively rarely in our population. Similar findings were also described by Featherstone [ 171in a series of 50 consecutive headache clinic practice patients and by Kaganov et al. [16] based on data derived from a mailed questionnaire sent to a random sample of Calgary, Alberta households. The linear relationship found in the latter study [16] between symptom frequency, disability, and pain was also replicated in the findings from our study. Similarly, Waters [40] found a high overlap among migraine and other headache symptoms in an epidemiologic study using mailed questionnaires carried out in Wales. Finally, the results of Diehr et al. [6,8] also demonstrate significant overlap of symptoms among patients classified as having tension and migraine headache. Proponents of the continuum model of headache would argue that results showing overlapping occurrence of migraine and tension-type symptoms support their position. A contrary view, however, would hold that headache attacks characterized by symptoms of both migraine and tension-type headaches merely represent “mixed headache” or two types of headaches occurring simultaneously. This latter position suggests that while it is clear that tension-type headaches generally are less painful, of shorter duration and produce less disability than migraine headaches, this is insufficient evidence to warrant the conclusion that all headaches can be summarily reduced to a single dimension (i.e. a sole continuum). Our data on pain demonstrate that headaches char-

acterized by both unilaterality and feeling of a tight band produce the greatest experience of pain among males; this can be interpreted as evidence of the simultaneous experience of two types of headache rather than a “midpoint” between tension-type and migraine headache. Until a biologically plausible interpretation of headache symptoms emerges, conclusions regarding the validity of the headache severity model will have to be based, in part, on further epidemiologic evidence [45,46]. This analysis of the relationship between symptoms, pain, and duration associated with headache was carried out using telephone interview data obtained from one of the largest randomly selected population-based surveys of subjects believed to be at the peak age for new onset of severe headache. The focus of the analysis was on the detailed characterization of headaches described within 4 weeks prior to interview, which were reported by one-half of males and nearly three-quarters of females. Since the analysis was confined to headaches occurring within the previous 4 weeks, subjects with more frequent headaches were over-represented in the sample. As such, the findings are not strictly representative of all headache attacks among persons 12-29 years of age in the total population, but should be generalizable to those occurring within the defined time interval in the age group studied. All of the information obtained in this study was collected using a standardized interviewing approach using non-clinical interviewers. It is possible that non-clinicians might have been less attentive to responses which did not exactly fit response categories. Clearly, the questionnaire did not include a complete inventory of symptoms. Among the limitations of the standardized instrument, which was devised for a telephone interview, is the absence of an

Headache Symptoms

open-ended component for a description of the headache as well as detailed probing and repeating of questions as is used in the clinical setting. However, questionnaires developed for large-scale surveys such as the present investigation require a compromise between efficiency and inclusiveness. It is possible that additional important symptoms associated with headache in the general population were not included or that the absence of detailed probing may have resulted in underascertainment of information about the symptoms that were included as well as the number of recent attacks. The literature describing various definitions used for migraine, tension-type headache and other headache types appears to lack consensus about which key symptoms characterize migraine, tension-type and other kinds of headaches [41]. Nevertheless, the symptoms included in the analysis were thought to represent those symptoms most frequently cited in the literature as well as those used in 6 primary sources from which our instrument was derived [8, 16,33-361. Other limitations characterizing our study are those common to survey research: decrease in the reliability of recall with an increasing interval between the event under study (most recent headache attack) and the interview [42]; the tendency by some respondents to agree with an entire series of symptoms included in lists (acquiescent response set) [43]; and selective recall of events [44]. To attempt to overcome these limitations, we restricted study subjects in this analysis to those reporting a headache within 4 weeks of interview and focused on the occurrence of specific symptoms associated with each subject’s most recent headache. Thus, while the reliability of symptom reporting is a concern, the selection of a large random sample in a defined population with restriction of the length of recall and the focus on a specific headache (rather than a “usual” or “typical” headache) reduces the likelihood of bias related to self-reports. Finally, caution must be exercised in interpreting the findings as some of these may be the result of chance due to the large number of associations examined. Our investigation was not designed to address the association of headache symptoms with the “clinical” diagnoses of migraine, tension-type headache, and other headache types. The data presented here represent a cross-section of headaches experienced in a population. We specifically address each respondent’s most CE4319-J

993

recent headache attack with its accompanying symptoms. Clearly, this approach does not reflect either the process or possible conclusions to be drawn from a clinical diagnosis given to patients presenting with a history of symptoms. While the design that we used provides a test of the association of migraine and tension symptoms as well as their relationship to level of pain and duration accompanying the respondents’ most recent headache attack, it does not address the associations between these reported experiences and clinical diagnoses of these subjects. It is essential that the survey methodology used here be linked in future studies with clinical diagnoses of disease in order to determine the overall validity of the survey approach and to more clearly specify the utility of the severity model as a clinically relevant concept. In summary, data from our epidemiologic study lend additional support to the severity or continuum model of headache. Although three population-based studies [16,40 and this investigation] and two clinic-based studies [6, 171 provide some data to support the severity model of headache, clinical experts will probably continue to adhere to the symptom-based classification approach [e.g. 31. However, the increasing recognition by clinician specialists that patients with migraine headache may also have other types of headaches and may experience symptoms of tension-type headache during a migraine attack, creates confusion in the quest for a comprehensive classification system. Future epidemiologic studies should use instruments that include more detailed inquiries about a completely exhaustive list of symptoms, and that also incorporate aspects of the clinical probing approach. In addition, validation of results of epidemiologic studies with expert clinical evaluation will be essential if epidemiologic study results are to impact upon clinical practice, particularly diagnosis. Such future epidemiologic studies will hopefully provide the data required to conclusively distinguish headache subtypes and will enable a more rigorous test of the severity or continuum model of headache. Acknowledgements-This investigation was supported by PHS Grant No. NS-19381. awarded bv the National Institute on Neurological and. Communicative Disorders and Stroke, Department of Health and Human Services. The authors acknowledge the assistance of Dewey Ziegler, M.D. and Martha Sprecher, M.S.N. in the design of the study and collection of these data.

DAVID D. CELENTANOet al.

994

REFERENCES 1.

2.

3.

4. 5.

6.

7. 8. 9. 10.

11. 12.

13. 14.

15.

16.

17. 18. 19. 20. 21. 22.

National Center for Health Statistics, Collins JG. Prevalence of selected chronic conditions, United States, 1983-85. Advance Data From VitaI and Health Statistics, 1988, No. 155, Publ. No. (PHS) 88-1250. Hyattsville, Md: Public Health Service; 1988. Linet MS, Stewart WF. The epidemiology of migraine headache. In: Blau JN, Ed. Migrabte: Classical, Tberapeutic, Conceptual and Research Aspects. London: Chapman and Hall; 1987: 451-477. Ad Hoc Committee of the International Headache Society. Proposed classification and diagnostic criteria for headache disorders, cranial neuralgias and facial pain. Cephalalgla 1988; I((Supp1. 7): 9-96. Ad Hoc Committee on Classification of Headache. Classification of headache. JAMA 1962; 179: 127-128. Ziegler DK, Hassanein R, Hassanein K. Headache syndromes suggested by factor analysis of symptom variables in a headache prone population. J Chron DIs 1972; 25: 353-363. Diehr P, Wood RE, Barr V et al. Acute headaches: presenting symptoms and diagnostic rules to identify patients with tension and migraine headache. J Chron Dis 1981; 34: 147-158. Waters WE, O’Connor PJ. Epidemiology of headache and migraine in women. J Nemo1 Neurosurg Psychiat 1981; 34: 148-153. Diehr P, Diehr G, Koepsell J et al. Cluster analysis to determine headache types. J Cbron Dis 1982; 35: 623-633. Waters WE. The Pontypridd headache survey. Headache 1974; 14: 81-90. Brewis M, Poskanzer DC, Rolland C et al. Neurological disease in an English city. Acta Neurol Stand 1966; 24 (Suppl.): l-89. Markush RE, Karp HR, Heyman A et al. Epidemiologic study of migraine symptoms in young women. Neurology (NY) 1975; 25: 430-435. Clarke GJR, Waters WE. Headache and migraine in a London general practice. In: Waters WE, Ed. The Epidemiology of Migralae. Bracknell, Berkshire: Boehringer Ingelheim; 1974. Linet MS, Stewart WF. Migraine headache: epidemiologic perspectives. Epidemiol Rev 1984; 6: 107-139. Bakal DA, Kaganov JA. Symptom characteristics of chronic and non-chronic headache sufferers. Headache 1979; 19: 285-289. Thompson JK, Haber JD, Figueroa JL, Adams HE. A replication and generalization of the “Psychobiological” model of headache. Headacbe 1980; 20: 199-203. Kaganov JA, Bakal DA, Dunn BE. The differential contribution of muscle contraction and migraine symptoms to problem headache in the general population. Headache 1981; 21: 157-163. Featherstone HJ. Migraine and muscle contraction headaches: a continuum. Headache 1985; 25: 194-198. Newland CA, Illis LS, Robinson PK, Batchelor BG, Waters WE. A survey of headache in an English city. Res Clln Stud Headache 1978; 5: l-20. Thompson JK, Haber JD, Figueroa JL, Adams HE. A replication and generalization of the “psychobiological” model of headache. Headache 1980; 20: 199-203. Tfelt-Hansen P, Lous I, Olesen J. Prevalence and significance of muscle tenderness during common migraine attacks. Headache 1981; 21: 49-54. Cohen MJ. Psychophysiological studies of headache: is there similarity between migraine and muscle contraction headaches? Headache 1978; 18: 189-196. Muck-Seler D, Deanovic Z, Dupelj M. Platelet serotonin (5-HT) and 5-HT releasing factor in plasma of migrainous subjects. Headache 1979; 19: 14-17.

23. 24. 25. 26. 27. 28.

29.

30. 31. 32.

33.

34.

35.

36.

37.

38. 39.

40. 41. 42. 43. 44. 45. 46.

Bakal DA, Kaganov JA. Muscle contraction and migraine headache: psychophysiologic comparison. Headache 1977; 17: 208-215. Martin MJ, Rome HP. Muscle-contraction headache: therapeutic aspects. Res CIIn Stud Headache 1967; 1: 205-217. Graham JR. Migraine headache: diagnosis and management. Headache 1979; 19: 133-141. Mathew NT. Prophylaxis of migraine and mixed headache: a randomized controlled study. Headache 1981; 21: 105-109. Carasso RL, Yehuda S, Streifler M. Clomipramine and amitriptyline in the treatment of severe pain. Int J Neurosci 1979; 9: 191-194. Meyer JS, Hardenberg J. Clinical effectiveness of calcium entry blockers in prophylactic treatment of migraine and cluster headaches. Headache 1983; 23: 266-277. Werder DS, Sargent JD, Coyne L. MMPI profiles of headache patients using self-regulation to control headache activity. Headache 1981; 21: 164-169. Torgerson W. Theory and Methods of SeaBng. London: Wiley; 1958. Waksberg J. Sampling methods for random digit dialing. J Am Stat Assnc 1978; 73: 40-46. Linet MS, Stewart WF, Celentano DD, Ziegler D, Sprecher M. An epidemiologic study of headache among adolescents and young adults. JAMA 1989; 261: 2211-2216. Celentano DD, McQueen DV. The National Survey of Headache Medical Questionnaire. Prepared under contract to the National Institute of Neurologic and Communicative Disorders and Stroke, Order No. 263MD-930118. Baltimore, Md: The Johns Hopkins School of Hygiene and Public Health; 1980. Ziegler DK, Hassanein RS, Couch JR. Headache syndromes suggested by statistical analysis of headache symptoms. Cepbalalgla 1982; 2: 125-134. Thompson JK, Figueroa JL. Dichotomous vs interval rating-of headache symptomatology: an investigation in the reliability of headache assessment. Headache 1980; 20: 261-265. Bickerstaff ER. Migraine variants and complications. In: Blau JN. Ed. Mlmalne: CIlnIcal. Therapeutic, Conceptual and ‘Resear&Aspects. London: Chapman and Hall; 1987: 55-75. Packard RC. Life stress, personality factors, and reactions to headache. In: Dalessio DJ, Ed. WolB’s Headache and Other Head Paln. New York: Oxford; 1987: 370-387. Ziegler DK. Headache syndromes: problems of definition. Psychosomatics 1979; 20: 443-447. Ziegler DK. The treatment of migraine. In: Dalessio DJ, Ed. Wolff’s Headache and Other Head Pam. New York: Oxford; 1987: 87-111. Waters WE. Community studies of the prevalence of mieraine. Headache 1970; 9: 178-186. Br;yn GW. Epidemiology of migraine; a personal view. Headache 1983; 23: 127-133. Sudman S, Bradbum NM. Effects of time and memory factors on response in surveys. J Am Stat Asaoc 1973; 68: 805-815. Bradburn NM. Response effect. In: Rossi PH, Wright JD, Anderson AB, Eds. Handbook of Survey Research. New York: Academic Press; 1983: 289-328. Paganini-Hill A, Ross RK. Reliability of recall of drug usage and other health-related information. Am J Epidemiol 1982; 16: 114-122. Ziegler DK, Kodanaz A, Hassanein RS. Do certain headache syndromes occur in “pain-prone” patients? Headache l-985; 25: 90-94. Ziegler DK. Headache syndromes: problems of definition. Psycbosomatics 1979; 20: 443-447.