Study Title and Description
Caffeine as a risk factor for chronic daily headache: a population-based study.
Key Questions Addressed
|1||For [population], is caffeine intake above [exposure dose], compared to intakes [exposure dose] or less, associated with adverse effects on behavior*?|
Primary Publication Information
|Title||Caffeine as a risk factor for chronic daily headache: a population-based study.|
|Author||AI Scher,WF Stewart,RB Lipton,|
Secondary Publication Information
There are currently no secondary publications defined for this study.
Extraction Form: Behavior - Design Details - INCLUDED Studies
No arms have been defined in this extraction form.
|Question... Follow Up||Answer||Follow-up Answer|
|What outcome is being evaluated in this paper?||Behavior|
|What is the objective of the study (as reported by the authors)?||To investigate the possible association of dietary caffeine consumption and medicinal caffeine use with chronic daily headache (CDH).|
|Provide a general description of the methods as reported by the authors. Information should be extracted based on relevance to the SR (i.e., caffeine related methods)||Study participants were identified from adults aged 18 to 65 who had participated in a general health survey. The purpose of this initial survey was to identify individuals from the Baltimore, MD, Philadelphia, PA, and Atlanta, GA, metropolitan areas who might be eligible for clinical trials. Details on the initial survey are described elsewhere.19,20 In brief, potential cases (headache 180 times/year at baseline) and potential controls (headache 2 to 104 times/year at baseline) were identified in the general health survey. These individuals were contacted approximately 1 year after the first survey and screened for eligibility as potential cases and controls. Interviews were conducted over the telephone by multiple trained lay interviewers using computer-controlled scripts. We did not validate the interview for interobserver variation, though we did monitor the interviews for quality. In our prior validation studies, we found that interviewer did not predict physician diagnosis based on a subsequent in-person assessment. During the screening interview, individuals were asked about their current headache frequency. Those with ≥180 headache days/year at follow-up were asked how long they had had headaches at their current frequency. Potential cases (i.e., ≥180 headache days/year at follow-up) with duration of ≤5 years and controls (i.e., 2 to104 headache days/year at follow-up) were invited to complete a 30- to 40-minute follow-up interview covering exposures believed to be associated with the onset or persistence of CDH, including dietary and medicinal caffeine consumption. Details on the computer-assisted follow-up interview have been described elsewhere.19,20 The methods herein are limited to a description of caffeine exposure. Dietary caffeine consumption. Dietary caffeine consumption was estimated based on the following self-reported sources: caffeinated coffee, caffeinated tea, caffeinated soda, hot chocolate, and chocolate candy. Total caffeine consumption (mg/day) was calculated using the estimated caffeine content per serving size as shown in table 1 and based on recall of consumption from these sources in the 3-month period preceding the interview. Subjects were also asked to recall past consumption of these substances during a defined time period. For cases, this time period was the year before CDH onset (defined during the screening interview). Because cases were limited to those with CDH duration of no more than 5 years, this recall period ranged from 1 to 5 years before the interview. For controls, this time period was a randomly assigned year in the same range (e.g., 1 to 5 years before the interview). To facilitate recall, participants were asked to define a temporal anchoring event by naming two memorable things that occurred during this year. We defined high dietary caffeine consumption as the top quartile (mg/week) of caffeine from these sources. Medication use for headache. Respondents were asked how many different medications they took to treat their headaches in the 3-month period before the interview. They were then asked to name the first, second, and third most frequently used headache medications. Finally, they were asked to estimate the total number of days in the last 3 months that any type of medication was taken for headache. The same questions were asked about headache treatments in the distant past (i.e., the same distant time period previously used for dietary caffeine consumption). We defined high medicinal caffeine consumption as positive for those who named a caffeine-containing OTC medication as their first-choice headache medication. We chose this measure (rather than computing milligrams per week of medicinal caffeine) so that it would be possible to make meaningful comparisons between cases (with very frequent headaches) and controls (with a "normal" headache frequency). An exposure measure based on dose would, by definition, be higher in the cases for caffeine or any other medication. Similarly, we included only OTC caffeine containing medications as control subjects were likely to have less opportunity for exposure to prescription medication. However, results were similar when prescription caffeine-containing medications were included. Headache classification. Participants were grouped into three primary headache groups (migraine, tension type, and other) in accordance with the 1988 International Classification of Headache Disorders (ICDH-1)21 using methodology similar to previous studies. 22,23 At the first digit level, migraine and tension-type headache have similar definitions in the revised system (ICDH-2) published after these analyses were completed.3 Analysis. Preliminary comparisons between cases and controls for continuous variables (e.g., milligrams per week of dietary caffeine consumption, age, body mass index, etc.) were made using a nonparametric test (Wilcoxon rank sum test). Comparisons for categorical variables were made using the _x0003_2 test. Logistic regression was used to estimate the risk of CDH associated with high dietary or medicinal caffeine consumption (defined as either top quartile dietary caffeine or first-choice caffeine containing OTC drug) after adjusting for potential confounders of the relationship between CDH and high caffeine consumption. Potential confounders included age, gender, educational level, headache type (migraine, tension type, unclassified), race (white, nonwhite), marital status (currently married, previously married, never married), and body mass index (normal, overweight, obese). Confounders were retained in the final model if they remained significantly associated with case status after adjusting for all other confounders. Interaction terms between all exposures and gender were also evaluated as preliminary comparisons revealed heterogeneity by gender for some exposures. For consistency, the same set of confounding factors selected in the main model was used in all subanalyses below. Subanalyses. Because caffeine consumption varies by age and gender17 and might also be related to headache type, we tested for interaction of caffeine with age group (_x0002_40, _x0001_40 years), gender, and primary headache type (migraine, tension type, unclassified). The p values were reported for the interaction term (age _x0004_ caffeine, gender _x0004_ caffeine, headache type _x0004_ caffeine) calculated from models including all participants. We also report odds ratios (ORs) calculated from stratified models (e.g., comparing female cases with female controls, male cases with male controls, younger cases with younger controls, etc.) as an estimate of the individual association of caffeine and CDH for each of these subgroups. Heterogeneity of risk associated with caffeine consumption was also considered for some case subgroups. In particular, we evaluated whether risk varied by whether headaches were continuous or noncontinuous, whether or not cases had consulted physicians for their headaches (because of physician advice regarding caffeine), and whether CDH was of longer or shorter duration. For the heterogeneity analysis, CDH cases were subdivided as follows: 1) last year physician consultation status for headache (0 _x0001_ control, 1 _x0001_ consulting case, 2 _x0001_ nonconsulting case), 2) duration of CDH (0 _x0001_ control, 1 _x0001_ case _x0002_2 years, 2 _x0001_ case _x0001_2 years), and 3) CDH subtype (0 _x0001_ control, 1 _x0001_ continuous case, 2 _x0001_ noncontinuous case). We estimated ORs for each of these case groups using multinomial (polytomous) regression, which is equivalent to logistic regression for nominal outcomes with more than two choices.24 All analyses were performed using Stata (StataCorp., 2003, Stata Statistical Software, release 8.0, College Station, TX).|
|How many outcome-specific endpoints are evaluated?||1|
|What is the (or one of the) endpoint(s) evaluated? (Each endpoint listed separately)||headache|
|List additional health endpoints (separately).|
|List additional health endpoints (separately)|
|Notes||chronic daily headache|
|What is the study design?||Case-Control|
|Randomized or Non-Randomized?|
|What were the diagnostics or methods used to measure the outcome?||Subjective|
|Optional: Name of Method or short description||headache questionnaire|
|Caffeine (general)||Caffeine (general)|
|What was the reference, comparison, or control group(s)? (e.g. high vs low consumption, number of cups, etc.)||top quartile of consumption vs lower quartile|
|What were the listed confounders or modifying factors as stated by the authors? (e.g. multi-variable components of models. Copy from methods)||Potential confounders included age, gender, educational level, headache type (migraine, tension type, unclassified), race (white, nonwhite), marital status (currently married, previously married, never married), and body mass index (normal, overweight, obese). Confounders were retained in the final model if they remained significantly associated with case status after adjusting for all other confounders. Interaction terms between all exposures and gender were also evaluated as preliminary comparisons revealed heterogeneity by gender for some exposures.|
|Provide a general description of results (as reported by the authors).||Caffeine consumption in the pre-CDH time period was 232 mg/day (median 147 mg/day; IQR _x0001_ 47 to 311) and also consisted primarily of coffee consumption (68%). The cut-point for the top quartile of current dietary caffeine consumption was 287 mg/day and for past caffeine consumption 311 mg/day, equivalent to roughly three 8-oz cups of caffeinated coffee. Either top quartile or caffeine-containing medication. Overall, cases were more likely to be high current (41 vs 32%; OR =_x0001_ 1.49, p = 0.026) or past (40 vs 29%; OR = 1.58, p =_x0001_ 0.013) caffeine consumers than controls (see table E-1 on the Neurology Web site at to www.neurology.org). After adjusting for potential confounders, high current caffeine exposure lost significance (OR _x0001_ 1.36, p _x0001_ 0.12), whereas past high caffeine exposure remained significant (OR _x0001_= 1.50, p _x0001_ 0.05). Analyses were repeated including prescription caffeine-containing medication. ORs were unchanged for current caffeine exposure and slightly increased for past caffeine exposure (OR =_x0001_ 1.60, p =_x0001_ 0.022).|
|Did the authors perform a dose-response analysis (or trend/related analysis)?||No|
|What were the authors's observations re: trend analysis?|
|What were the author's conclusions?||Current high caffeine consumption did not differ between CDH cases and controls. The pattern we observed is consistent with the hypothesis that caffeine consumption is a risk factor for CDH as caffeine consumption was increased in the period prior to CDH onset. It is not consistent with the hypothesis that caffeine consumption increases as a consequence of increasing headache frequency. Were that the case, we would have expected caffeine consumption to be elevated both before and after CDH onset. This association might indicate a causal relationship between caffeine consumption and CDH onset or might be secondary to unmeasured confounding factors. For example, sleep disturbances and depression are related to caffeine consumption and are possible predisposing factors for CDH development for some individuals. Although the relationships reported herein remained similar after adjusting for current depression and current sleep problems, we do not have data on the pre-CDH status of these factors.|
|What were the sources of funding?||Supported by GlaxoSmithKline, the Migraine Trust, and the American Headache Society.|
|What conflicts of interest were reported?||N/A (see sources of funding)|
|Does the exposure (dose) need to be standardized to the SR?|
|Provide calculations/conversions for the exposure based on the decision tree in the guide (for all endpoints/exposure levels of interest).|
|List all the endpoint(s) followed by the dose (mg) which will be used in comparison to Nawrot. Characterize value as LOAEL/NOAEL, etc. if possible.||headache LOAEL = 311 mg/day|
|Notes regarding selection/listing of endpoints and exposures/doses to be compared to Nawrot.||headache refers to risk of chronic daily headache|
|What is the importance of the study with respect to the adverseness of the outcome?||Critcal|
No baseline characteristics have been defined for this extraction form.
Results & Comparisons
No Results found.
|Arm or Total||Title||Description||Comments|
No quality dimensions were specified.
No quality rating data was found.