J ULL1 PROPOSED 10 criteria to describe a distinct type of

Transcription

1 1204 Validity of a Measure of the Frequency of Headaches With Overt Neck Involvement, and Reliability of Measurement of Cervical Spine Anthropometric and Muscle Performance Factors Leigh Blizzard, MEC, Karen A. Grimmer, PhD, Terence Dwyer, MD ABSTRACT. Blizzard L, Grimmer KA, Dwyer T. Validity of a measure of the frequency of headaches with overt neck involvement, and reliability of measurement of cervical spine anthropometric and muscle performance factors. Arch Phys Med Rehabil 2000;81: Objective: To test the validity of a questionnaire to measure frequency of headaches related to the neck. A secondary goal was to test the reliability of field measurement of associated cervical spine anthropometric and muscle performance factors. Design: Intermethod and test-retest comparisons. Setting: Two municipalities in a rural area of Tasmania, Australia. Participants: One hundred subjects were selected as a representative sample of never-injured adults from a comprehensive listing of the source population, and 93 participated. Main Outcome Measures: Subjects distinguished headaches matching three criteria for overt neck involvement, reported past-month frequency by questionnaire, and recorded in a diary occurrences during the next month. Measurements of height, weight, neck column length and circumference, lateral flexion and extension range of movement, cervical short flexor muscle endurance, and cervical long flexor and extensor strength were taken at the start and end of the month. Results: Headache frequency was associated with neck stiffness and neck ache. There was moderate agreement (weighted.66) between questionnaire and diary, but better agreement (.72) for subjects who maintained their usual patterns of recreational sport. For this group, the correlation with true frequency was r u.87. The anthropometric and muscle performance factors were reliably measured (intraclass correlations.96 to 1.00,.78 to.86) despite minor improvement in muscle performance on retest. Conclusions: The questionnaire measure has construct validity. Neck-related headaches are a temporally stable presentation in never-injured subjects who maintain customary sporting activity. Measurement error was consequential, but less so for this group than for the study subjects generally. The anthropometric and muscle performance measurements were reliable, but slight improvements on retest suggest the need for multiple measurements. Key Words: Headache; Neck muscles; Anthropometry; Reproducibility of results; Rehabilitation. From the Menzies Centre for Population Health Research, University of Tasmania (Blizzard, Grimmer, Dwyer); and Centre for Physiotherapy Research, School of Physiotherapy, University of South Australia (Grimmer). This study was conducted in Huonville, Tasmania, Australia. Accepted January 10, No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit upon the authors or upon any organization with which the authors are associated. Reprint requests to Leigh Blizzard, Menzies Centre for Population Health Research, GPO Box , Hobart, Tasmania, 7001, Australia /00/ $3.00/0 doi: /apmr by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation J ULL1 PROPOSED 10 criteria to describe a distinct type of headache 2-11 believed to result from faulty biomechanics of the cervical spine. 1-5,12-14 Three of her criteria identify overt neck involvement, and symptoms associated with the neck had been previously used as descriptors of this type of headache. 5-6,15-17 In particular, stiff or aching necks and sustained positions of the neck have been related directly to headache events. By asking a representative sample of 427 never-injured adults to distinguish and recall the past-month frequency of headaches that matched Jull s three neck-specific criteria, we undertook the first population-based study of neck-related headaches. 18 Questionnaires had been successfully used to investigate the prevalence of migrainous and nonmigrainous headaches, but none had previously been developed for the evaluation of neck-related headaches. Jull s criteria have never been validated in a controlled study, and the questionnaire incorporating just three of them was untested. The primary purpose of this study was to assess the validity of the our new questionnaire measure. We examined the construct validity of the new measure by testing whether the frequency of headaches was associated with reports of stiff or aching necks and unaccustomed neck movements. Extending an earlier preliminary analysis, 24 we compared the questionnaire responses for the previous month with the results of 31-day prospective diary recording of headaches during the following month. A secondary objective was to assess the accuracy of field measurement of cervical spine anthropometric and muscle performance factors that may be involved in the cause of these headaches. In the absence of disease or trauma, it has been proposed that these headaches result from sustained, long-term abnormal physiologic loads on the neck, which compromise cervical joint function. 1,10,12,25,26 We therefore studied the test-retest reliability of measurement of factors considered to affect the performance of the neck, such as body size and head positioning, 27,28 range of neck movement, and the endurance and strength of cervical muscles. 26,29,30 In each case, we used measurement methods of the type appropriate in large-scale fieldwork. Studying the cause of headaches with overt neck involvement is important because they are a common source of pain and illness in the community. We found that 28% of subjects in a large representative sample of never-injured persons experienced these headaches at least twice per month, and another 35% had them occasionally. 18 Opportunities for successful therapeutic management are presently limited by a lack of information about the causal pathways involved in their presentation. An important use of the new questionnaire method will be to mount analytical epidemiologic studies, the results of

2 MEASUREMENT OF NECK-RELATED HEADACHES, Blizzard 1205 which will eventually provide the evidence base to justify intervention. METHODS Subjects The sample selection methods are described in detail elsewhere. 18,24 In brief, 450 adults were selected by simple probability sampling using computer-generated random numbers from a published and numbered list of all registered electors living in a defined geographic region of Tasmania, a state of Australia. All 450 persons were contacted, and only 9 refused to participate, but 14 others were subsequently excluded to restrict the sample to never-injured persons who were not on medication for migraine, not regularly using nonsteroidal antiinflammatory drugs or analgesics, not pregnant, and not breastfeeding. The 427 eligible subjects completed a questionnaire and attended a measurement session in the winter of The first 100 of those 427 subjects were asked to participate in this study, and 93 agreed. Those 38 women and 55 men included 22 persons aged 18 to 29 years, 29 aged 30 to 39 years, 25 aged 40 to 49 years, 9 aged 50 to 59 years, and 8 aged 60 to 84 years. They kept a diary record of headaches during the following month, were remeasured at the end of that month, and completed a brief follow-up questionnaire at that time. The same investigator (KAG), who was blind to the questionnaire data, made all measurements. An independent scribe recorded all measurements. Study Measurements Two interview and measurement sessions, one calendar month apart, were conducted for each subject. The same (blinded) measurer conducted each session. Headache by questionnaire (previous month). At the first measurement session, information on the frequency of headache in the previous month was captured by a self-administered questionnaire, which has been reproduced elsewhere. 18 All 427 subjects were given colloquial written interpretations of the three criteria describing overt neck involvement (table 1), and were asked to recall the frequency during the previous month with which they had experienced headaches that matched these three criteria. The response categories are shown in table 1. The questionnaire also included questions on neck stiffness and neck ache, occupation, type of employment (full-time, parttime, period since last employment), use of spectacles, type of dentures, usual frequency of participation in favorite interests and hobbies, usual frequency of recreational sport, and the type of sport involved. First measurement of anthropometry and muscle performance (start of the diary month). Standing height was measured in centimeters with the barefoot subject standing as tall as possible, head vertically aligned with a wall tape with eyes focused on a letter on a wall chart that was level with the straight-ahead gaze. Weight was measured standing in kilograms, using the same set of regularly calibrated bathroom scales for all subjects. Neck length and neck circumference were measured with a flexible nonstretch tape measure. Neck column length was measured with the subject sitting straight on a stool, with hips, knees, and ankles at right angles, and head and eyes positioned as described for standing height. The anterior neck column length was measured from the point of the chin to the level of notch of the xiphisternum, with the tape measure placed on the skin and conforming to the neck contours. The posterior column length was measured in the Table 1: Two Measures of the Monthly Frequency of Headaches With Overt Neck Involvement Method 1: First Questionnaire (Previous Month) How often did you have a headache like this in the last month? Number of subjects a. Didn t have a problem with them 38 b. Daily 6 c. Every couple of days 4 d. Once a week 11 e. About once every 2 weeks 6 f. About once a month 11 g. Every couple of months or so Method 2: Diary (Following Month) Circle the days on the enclosed calendar, starting on.../.../..., when you have a headache which matches the description provided. Number of subjects 28 (n 2) or 30 days circled 3 22 (n 1) or 23 days circled 3 11, 13, 15, 16, or 19 days circled 7 4(n 2) or 6 days circled 4 3 days circled 5 2 days circled 7 1 day circled 23 0 days circled Headache description: 1. You have pain at the back of the skull and/or in the neck when you have this headache. 2. You often wake up with this type of headache. 3. This type of headache is usually associated with holding your head in the one position for a long time; and/or constant neck movement; and/or with tension. same manner from the bony occipital prominence on the skull to the level of spinous process of C7. Neck circumference was measured around the neck column at the level of the spinous process of C7 and the larynx/trachea junction. A tape measure was held firmly against the skin, conforming to the contours of the neck column while allowing the subject to swallow comfortably. Measurements of lateral flexion and extension of the cervical spine were taken with a Myrin goniometer, a using the test positions described by Klaber Moffat and colleagues. 31 For all mobility measurements, a Velcro strap was positioned around the skull with the lower edge of the strap made level with the top of the ears. To measure left lateral flexion, the goniometer was centered at zero on the strap in the center for the forehead while subjects sat with their head in the position for measuring neck column length. Subjects were then asked to move their left ear toward the point of their left shoulder, and to stop when the end of comfortable range was reached. The examiner guided the movement to ensure directional accuracy, but did not apply overpressure at the end of range. At the end of comfortable range, the degrees of lateral flexion movement were read from the goniometer. Right lateral flexion was similarly measured using instructions directing subjects to move to the right. To measure extension, the goniometer was placed on the strap directly above the superior-most tip of the helix of the right ear, and was centered to zero, while subjects sat upright looking ahead (as for measuring the length of neck column). Subjects

3 1206 MEASUREMENT OF NECK-RELATED HEADACHES, Blizzard were instructed to bend their neck back and look upward to the ceiling until they reached the end of their comfortable extension range. The degrees of extension were read from the goniometer. The endurance of the cervical short flexor muscles was measured as the time over which a common exercise position 14,32 could be held. This involved lying supine, tucking the chin in, and then lifting the head with the chin still in the tucked position. The time between assuming the test position and chin thrust (failure to maintain the chin-tucked position, which was visually identified when the distance between the chin and the neck first increased) was measured with a stopwatch. Chin thrust was the most easily detected event in the sequence of events, culminating in loss of holding the test position of the head. Strength of the long cervical flexors and cervical extensors was estimated with bilateral antigravity and manual resistance tests, as described by Janda. 33 Testing was initiated with no head support, in supine (for cervical flexors) and prone (for cervical extensors positions). Strength was described in one of five categories: inability to maintain the position against gravity (grades 1, 2), maintenance of the test position against gravity only (grade 3), maintenance of the test position against gravity with some manual resistance (grade 4), and maintenance of the test position against full manual resistance (grade 5). Headache by diary (during the diary month). The 100 subjects invited to participate in this study were asked to mark on a calendar with which they were provided those days during the next month on which they had a headache matching the three criteria. The calendars were retrieved at the second measurement session, and the number of headache days marked on each was counted and recorded. Remeasurement of anthropometry and muscle performance (end of the diary month). The second measurement session was held one calendar month after the first. To blind the investigator to the previous measurements, the second measurements were recorded on new data sheets, and the first recordings were not retrieved until all remeasurements were completed. Follow-up questionnaire (end of the diary month). A brief follow-up questionnaire was completed after the second measurements. It contained the same questions as the first questionnaire on neck stiffness ( Did you suffer from any stiff necks in the past month? ) and neck ache ( Did you suffer from any aching necks in the past month? ). Subjects who reported a stiff neck during the diary month were identified from responses to the follow-up questionnaire; those who had an aching neck were similarly identified. Furthermore, by comparing subjects responses on the follow-up questionnaire with those from the first questionnaire, we identified subjects who reported having had a stiff or aching neck in the diary month but not in the previous month, and those who had a stiff or aching neck in the previous month but not in the diary month. There was another question on frequency of recreational sport ( How regularly did you participate in recreational sport in the past month? ), which had identical response categories to the question on usual recreational sport in the first questionnaire. By comparing subjects reports of participation on the second questionnaire with their responses to the question on usual participation in the first questionnaire, we identified those who reportedly had engaged in recreational sport more frequently or less frequently than usual during the diary month. The remaining subjects who had marked the same response category to the question on both questionnaires were regarded as having engaged in recreational sport with their usual frequency. Data Analysis To test agreement between the questionnaire and diary measurements of headache frequency, the diary numbers were categorized as follows: 28 to 30 daily, 22 to 23 daily, 11 to 19 every couple of days, 4 to 6 once a week, 3 once a week, 2 about every 2 weeks, 1 occasionally (combining the categories of about once a month or every couple of months or so ), 18 and 0 none. Alternative categorizations with 22 to 23 every couple of days and 3 about every 2 weeks produced only minor differences in results. For some analyses, the categories corresponding to two or more headaches per month were collapsed into a single category (frequent headaches), because we have shown that the six categories can be reduced to three (none, occasional headache, frequent headache) without loss of information. 18 The weighted statistic, 34 which penalizes more heavily the larger departures from perfect agreement, was calculated as a validity coefficient on the assumption of independent measures. The weights were calculated from the k 6 row (i) and column (j) numbers as (i j)2 w ij 1 (k 1). 2 To make comparisons on a numerical scale of measurement, the questionnaire categories were scored as follows: daily 28, every couple of days 14, once a week 4, about every 2 weeks 2, about once a month 1, every couple of months or so 0.5, and didn t have a problem with them 0. The product-moment correlation coefficient calculated from the ranks of the data is reported as a validity coefficient. For analyses of headache frequency as an outcome with three ordered categories (none, occasional, frequent), we used proportional odds logistic regression as described previously. 18 This allowed us to compare the odds of neck stiffness during the diary month in the three headache frequency groups. For analyses of outcomes with two levels, binary logistic regression or chi-square tests were used. To examine the test-retest measurements of anthropometric and muscle performance factors, the intraclass correlation coefficient 34 was calculated as a reliability coefficient for numerical measurements. The variance components were recovered from a one-way analysis of variance (ANOVA) model with the subjects treated as a random effect. For ordered categorical measures, the weighted statistic was used to assess reliability. Within-person differences between pairs of measurements were calculated and compared for groups of subjects. Statistical testing was applied to the ranks of the data because, in most cases, the means were influenced by extreme values. The within-person differences were regressed on the averages of the two measurements to determine if the differences increased with the size of the measurements. RESULTS The 93 subjects in this study were slightly younger (median age, 39.4yr) than the remaining 334 subjects (median age, 42.4yr) in the main study, but all age groups were represented. The two groups were alike in sex, occupation, hobbies, and frequency of headache during the past month. The result of categorizing the diary measurements of headache frequency is shown in table 1. The percentage of subjects identically categorized by questionnaire for the previous month and by diary report for the next was 62% (58 of 93), but 86% (80 of 93) of subjects were in the same or an adjacent category

4 MEASUREMENT OF NECK-RELATED HEADACHES, Blizzard 1207 on the second occasion. The weighted statistic was.66 (table 2). There was no evidence of a difference in the number of headaches reported under each method (table 2). On average, about 3 days of headache were reported each time, and the difference between them of about one third of a headache was attributed to chance ( p.44). To assess construct validity, we tested whether more subjects who experienced headaches during the diary month reported having had a stiff neck or aching neck that month than did subjects who had not experienced a headache. This proved to be the case ( p.01): 8% (3 of 38) of the no-headache-by-diary group reported having a stiff neck during the diary month, compared with 39% (11 of 28) of the occasional-headache group and 67% (18 of 27) of the frequent-headache group. The pattern was similar for aching necks. We further investigated whether those who suffered a stiff neck or an aching neck in the diary month but not in the previous month would report more headaches in the diary month than in the previous month, and conversely. Nine subjects reported a stiff neck in the diary month but not in the previous month, and 8 of these 9 subjects reported either more (4 of 9) or the same number (4 of 9) of headaches by diary for the diary month than by questionnaire for the previous month. Twelve subjects reported a stiff neck in the previous month but not in the diary month, and all 12 reported either fewer (5 of 12) or the same number (7 of 12) of headaches by diary. The pattern was similar for month-on-month reports of aching necks. The differential reports of stiff necks were associated with having departed during the diary month from usual frequencies of recreational sport. All but 1 of the 9 subjects with a stiff neck in the diary month but not in the previous month had played more sports than usual (3 of 9) or the same amount (5 of 9), and 10 of the 12 subjects with a stiff neck in the previous month but not the diary month had participated less in sports than usual (5 of 12) or the same amount (5 of 12). The pattern was similar for aching necks. We therefore assessed agreement between the two headache measures for groups of subjects classified by departure from customary recreational sport. Categorical agreement was higher for subjects who maintained their usual pattern of recreational sport during the diary month (.72) than for those who had any different pattern (.60). The correlation coefficients of ranked association were r.76 and r.49, respectively. To understand what this meant in terms of numbers of headaches, mean differences between the two numerical measurements were compared for subjects classified this way. The results are depicted in figure 1. The 24 subjects (14 women, 10 men) who engaged in sports less frequently than usual during the diary month recorded fewer headaches during that month than they had reported by questionnaire for the previous month. The 48 subjects (18 women, 30 men) who maintained their customary activity recorded about the same numbers of headaches, but the 21 subjects (6 women, 15 men) who engaged in Table 2: Comparison of Questionnaire and Diary Reports of Monthly Frequency of Headaches With Overt Neck Involvement Units of Measurement Questionnaire Diary Difference Reliability* Numerical (days of headache) 3.2 (6.3) 3.5 (7.0) 0.3 (15.8) r.65 Categorical (ordered levels 1 6) 2.3 (1.5) 2.2 (1.5) 0.1 (1.2).66 * r rank correlation coefficient. weighted kappa statistic. Fig 1. Age- and sex-adjusted mean differences between questionnaire and diary reports of monthly frequency of headaches with overt neck involvement, classified by frequency of recreational sport during the diary month. sports more frequently than usual reported more headaches. The age- and sex-adjusted mean differences were.98,.03, and.85 headaches, respectively, and the trend was significant ( p.03). Slightly higher numbers of headache were reported in the diary by older subjects, women, manual workers, those who engaged in outdoor physical hobbies, and those who usually only infrequently engaged in recreational sport (irrespective of usual levels), but the differences were not significant. Moreover, there was no difference in headache reporting between groups of subjects classified by tertiles of the anthropometric and muscle performance measurements. The results of the reliability study of anthropometric and muscle performance measurements are summarized in table 3. There were small but significant numerical increases in the measurements of weight and neck circumference, and significant improvements in lateral flexion, extension of the cervical spine, and cervical short flexor endurance. The intraclass correlation coefficients were uniformly very high, and there was no evidence that the differences between the test and retest measurements increased with the size of the measurement. For the ordered categorical variables, the statistics showed excellent test-retest agreement, but there was a significant improvement in cervical flexor strength measured on the second occasion. No association was found between headache frequency and any of the test-retest differences. DISCUSSION To study the occurrence of headaches possibly associated with the cervical spine, we narrowed the definition to those with neck-specific symptoms. In the absence of a generally accepted

5 1208 MEASUREMENT OF NECK-RELATED HEADACHES, Blizzard Table 3: Comparison of Test and Retest Measurements of Anthropometric and Muscle Performance Factors Study Factor First Measure Second Measure Difference Reliability* Numerical measurements Height (cm) (9.2) (9.2) 0.0 (0.1) r 1.00 Weight (kg) 72.3 (12.2) 72.8 (12.2) 0.5 (2.3) r.99 Length of front of neck (cm) 16.5 (1.9) 16.5 (1.8) 0.0 (0.3) r.99 Length of back of neck (cm) 12.8 (1.7) 12.7 (1.7) 0.0 (0.5) r.98 Neck circumference (cm) 37.6 (3.7) 37.7 (3.7) 0.1 (0.2) r 1.00 Lateral flexion (deg) 39.0 (6.1) 39.5 (5.4) 0.5 (2.3) r.96 Extension of cervical spine (deg) 53.8 (11.5) 54.7 (10.3) 1.0 (4.4) r.96 Short flexor endurance (sec) 16.6 (4.3) 17.2 (3.8) 0.6 (1.4) r.96 Categorical measurements Extensor strength (levels 2 5) 4.7 (0.5) 4.6 (0.5) 0.1 (0.3).78 Flexor strength (levels 2 5) 4.2 (0.8) 4.3 (0.8) 0.1 (0.4).86 * r intraclass correlation coefficient. weighted kappa statistic. p.05. p.001. diagnostic method suitable for use in fieldwork, we asked never-injured subjects to recall headaches that matched three criteria for overt neck involvement. The associations found between headaches and stiff necks, and between headaches and aching necks, suggest that the measurements corresponded to theoretical constructs regarding these headaches. There is a possibility of recall bias because the reports of stiff necks and aching necks on the follow-up questionnaire occurred after an awareness-raising month of record keeping, but further evidence of the construct validity of the new instrument is available. In the analytical study, 18 the headaches defined this way were associated with particular characteristics of the cervical spine, being significantly more frequent in subjects with a long anterior neck length relative to posterior neck length and weakest cervical long flexor strength. In this study, the effect of unaccustomed recreational activity increased the reporting of neck symptoms in the follow-up questionnaire, and this is a plausible outcome particularly for subjects with poor cervical muscle performance. There may have been an increase in the frequency of other musculoskeletal symptoms in the trunk and shoulders, but this was not studied. If neck-related headaches have a recurring component, we should be able to measure their monthly frequency of occurrence at different points in time using different methods, and obtain similar results. The comparable mean headache frequencies, calculated from the questionnaire measurements for 1 month and from the diaries for the next, provided some evidence that headaches defined this way are a temporally stable phenomenon. The moderate category-by-category agreement between the two measurements suggests that the questionnaire captures information about their true frequency of occurrence. Some measurement error is unavoidable, but if analytical studies are to provide valid inferences about causation, error should be random and preferably minor. In particular, there should be no systematic difference in the measurements at different levels of factors possible involved in causation. However, we found that part of the month-on-month variation in measurements was attributable to unaccustomed recreational sporting activities. There was much better agreement for the 52% of subjects who maintained their customary sporting activity during the diary month than there was for the 48% who deviated from their usual pattern. Subjects who engaged in recreational sports less frequently than usual during the diary month recorded one less headache that month than they recalled for the previous month, on average, and those who engaged in more recreational sports than usual recorded one additional headache. It is plausible that month-on-month differences in headache frequency would be associated with deviations from customary sporting activity. The effect of unaccustomed recreational activity increased the reporting of neck symptoms and, in the analytical study, 18 the frequency of these headache was highest in men who usually played sports only on an occasional basis. No other source of nonrandom error was found. There was no difference between the questionnaire and diary measurements for groups of subjects differentiated by age, sex, anthropometry or muscle performance, occupational status, or hobbies. For subjects who maintained customary sporting activity, therefore, measurement error that was not common to each method was limited to random error. The interpretation of headache is a possible source of common error. Subjects were required under each method to interpret the same set of instructions when deciding which of their headaches were of the neck-related type. The three neck-specific criteria were readily adapted for use and appeared to be comprehensible to our subjects, but we have no objective test of this, and subjects may have under- or overreported on both of our methods. Other sources of error in the questionnaire method, from restrictions imposed by categorization of responses and by poor memory, are not shared by the diary method that records events at the time they happen, and so are less dependent on memory. We expect that the diary method was more accurate than the questionnaire method, but we cannot quantify this. We also expect that if there is any correlation at all between the errors made by questionnaire and error made by diary, it will be a positive one. Some subjects who underreported by questionnaire because they misunderstood the headache description, and did not think that all of their headaches were of the type described, may have also underreported by diary because they were under the same misapprehension, and conversely. Assuming that the diary was at least no less accurate than the questionnaire, and that the errors made on each are positively correlated if correlated at all, the observed correlation (r) between the questionnaire and diary measures yields an upper limit (r u r) for the association between the questionnaire measure and true headache frequency. 35 For subjects who maintained customary sporting activity, the upper limit is r u.76).87. The error is consequential, therefore, even for those with temporally stable headache patterns. The correlation was nevertheless higher than those generally found in other studies with some comparability, including comparisons of food frequency questionnaires with weighed dietary records, 36 and physical activity questionnaires with activity recordings. 37

6 MEASUREMENT OF NECK-RELATED HEADACHES, Blizzard 1209 Those study factors have given rise to valid inferences about the causation of cardiovascular disease, adult-onset diabetes, and certain types of cancers, 36,38,39 despite measurement errors larger than in our measure of the frequency of neck-related headaches. Error in the measurement of the anthropometric and muscle performance factors also could cause their estimated associations with headache frequency to be weaker than they really are, or even produce spurious associations. We now have some evidence that neither was the case in our study. 18 The reliability coefficients in the test-retest study of measurement of anthropometric and muscle performance factors were uniformly very high. We conclude that these cervical spine factors were measured with excellent reliability using the types of measurement devices appropriate for large-scale fieldwork. Important for etiologic studies, there was no evidence that the measurement differences on retest differed according to frequency of headache. There were improvements in performance on the second test, however. Perhaps being aware of what was in store for them, the subjects were less reluctant to exert themselves on the second occasion. Researchers must be aware of this possibility, and consider making provision for multiple measurements in their study designs. The differences were minor, however, and the high reliability coefficients suggest that this source of error would not substantially diminish the observable measures of effect in an analytical study of headaches of this type. The strengths of the study were the representative population sample and the use of a single, blinded measurer for all measurements. The subjects were a random selection from a comprehensive listing of a defined source population, and 93% of those approached agreed to participate. The single measurer was unaware of questionnaire information on the subjects, an independent scribe recorded the measurements, and the results of the first measurements were not again sighted until after the second measurements took place. A limitation of this study was the lack of an error-free standard with which to test the criterion validity of the headache measure. This is the case in all headache reporting. Even in a symptomatic population, there are no error-free methods of diagnosing or measuring the frequency of particular presentations of headache. Our measure was based on three of Jull s criteria 1 for headaches to be of a cervical nature, but her subjects were self-selected patients who presented for treatment of headache and relatively few in number, there was no comparison group, and no definitive validation of the assessment method. The three criteria we selected were the only ones in the headache literature at the time that described overt involvement of the neck in headache. Their use as a diagnostic tool, in the absence of other forms of cervical investigation, still requires critical evaluation to confirm their validity and sensitivity in identifying headaches associated in some way with the performance of the cervical spine. CONCLUSION The questionnaire measure has construct validity. Neckrelated headaches are a temporally stable presentation in never-injured subjects who maintain customary sporting activity. Measurement error was consequential, but less so for this group than for subjects in general. The anthropometric and muscle performance measurements were reliable, but slight improvements on retest suggest the need for multiple measurements. References 1. Jull GA. Headaches associated with the cervical spine a clinical review. In: Grieve G, editor. Modern manual therapy. London: Churchill Livingstone; p Merskey H, Bogduk N. Classification of chronic pain: descriptions of chronic pain syndromes and definition of pain terms. Prepared for the Taskforce on Taxonomy of the International Association for the Study of Pain. Seattle: IASP Pr; p Bogduk N. Cervical causes of headache and dizziness. In: Grieve G, editor. Modern manual therapy. London: Churchill Livingstone; p Bogduk N, Marsland A. On the concept of third occipital headache. J Neurol Neurosurg Psychiatry 1986;49: Bogduk N, Marsland A. The cervical zygapophyseal joints as a source of neck pain. Spine 1988;13: Dwyer A, Aprill C, Bogduk N. Cervical zygapophyseal joint pain. Patterns 1: a study in normal volunteers. Spine 1990;15: Aprill C, Dwyer A, Bogduk N. Cervical zygapophyseal joint pain. Patterns 11: a clinical evaluation. Spine 1990;15: Kerr FWL, Olafson RA. Trigeminal and cervical volleys. Arch Neurol 1961;5: Kimmel DL. Innervation of spinal dura mater and dura mater of the posterior cranial fossa. Neurology 1961;11: Schonstrom N, Twomey L, Taylor J. The lateral atlanto-axial joints and their synovial folds: an in-vitro study of soft tissue injuries and fractures. J Trauma 1993;35: White AA, Panjabi MM. Clinical biomechanics of the spine. Philadelphia: Lippincott; Jull GA. Headaches of cervical origin. In: Grant RE, editor. Clinics in physical therapy. Vol 17: physical therapy for the cervical and thoracic spine. New York: Churchill Livingstone; p Bogduk N, Corrigan B, Kelly P, Schneider G, Farr R. Cervical headache. Med J Aust 1985;143:202, Trott PH. Manipulative therapy techniques in the management of some cervical syndromes. In: Grant RE, editor. Clinics in physical therapy. Vol 17: physical therapy for the cervical and thoracic spine. New York: Churchill Livingstone; p Sjaastad O, Saunte C, Hovdal H, Breivik H, Gronbaek E. Cervicogenic headache: an hypothesis. Cephalalgia 1983;3: Pfaffenrath V, Dandekar R, Pollmann W. Cervicogenic headache the clinical picture, radiological findings and hypotheses on its pathophysiology. Headache 1987;27: Centonze V, Polito BM, Di Bari M, Albano O. [Symptomatic headaches in internal medicine: the classification, physiopathological and diagnostic aspects. [Italian]] Ann Ital Med Intern 1992;7 Suppl 3:23S-32S. 18. Grimmer KA, Blizzard L, Dwyer T. Frequency of headaches associated with the cervical spine and relationships with anthropometric, muscle performance, and recreational factors. Arch Phys Med Rehabil 1999;80: 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: Waters WE. The Pontypridd headache survey. Headache 1974;14: Bastos SB, de Almeida-Fihlo N, Santana VS. [Prevalence of headache as a symptom in an urban area of Salvador, Bahia] [Portuguese]. Arq Neuropsiquiatr 1993;51: Honkasalo ML, Kaprio J, Heikkila K, Sillanpaa M, Koskenvuo M. A population-based survey of headache and migraine in 22,809 adults. Headache 1993;33: Jensen R, Rasmussen BK, Pedersen B, Olesons J. Muscle tenderness and pressure pain thresholds in headache: a population study. Pain 1993;52: Grimmer KA. Relationship between occupation and episodes of headache that match cervical origin pain patterns. J Occup Environ Med 1993;35: Basmajian JV. Muscles alive. Baltimore: Williams & Wilkins, 1979.

7 1210 MEASUREMENT OF NECK-RELATED HEADACHES, Blizzard 26. Sahrmann SR. Adult posturing. In: Kraus SL, editor. TMJ disorders: management of the craniomandibular complex. New York: Churchill Livingstone; p Ayub E, Glasheen-Wray M, Kraus S. Head posture: a case study of the effects on rest position of the mandible. J Orthop Sports Phys Ther 1984;5: Rocabado M. Biomechanical relationship of the cranial, cervical and hyoid regions. J Craniomandibular Pract 1983;1: Richardson C. Rehabilitation of muscular function of the cervical region. In: Dalton M, editor. Proceedings of Headache and Face Pain Symposium, Manipulative Physiotherapists Association of Australia. Melbourne: Australian Physiotherapy Association; p Janda V. On the concept of postural muscles and posture in man. Aust J Physiother 1983;29: Klaber Moffett JA, Hughes I, Griffiths P. Measurement of cervical spine movements using a simple inclinometer. Physiotherapy 1989;75: Grimmer K. Measuring the endurance capacity of the cervical short flexor muscle group. Aust J Physiother 1994;40: Janda V. Functional muscle testing. Cornwall: Butterworths; Fleiss JL. Statistical methods for rates and proportions. 2nd ed. New York: John Wiley; Walker AM, Blettner M. Comparing imperfect measures of exposure. Am J Epidemiol 1985;121: Willett WC. Nutritional epidemiology. New York: Oxford Univ Pr; Kriska AM, Caspersen CJ. Physical activity questionnaires used in the general population. Med Sci Sports Exec 1997;29 Suppl: S English DR, Armstrong BK, Kricker A, Pleming C. Sunlight and cancer. Cancer Causes Control 1997;8: US Department of Health and Human Services. Physical activity and health: a report of the Surgeon General. Atlanta (GA): U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion; Supplier a. Obrehab Co, A Lic Co, S Solna, Sweden.