Subjects and data collection
A consecutive series of eligible bronchial asthma patients and their mothers who visited the out-patient clinic of the Pediatrics Department of the National Fukuoka Hospital between February 1 and August 31, 2001 were invited to participate in the study. Criteria for enrollment included: (i) patient age 2 to 12 years; (ii) duration of bronchial asthma one year or longer; (iii) the necessity of repeat visits to the clinic (or hospitalization) over the past year; and (iv) medications necessary on a regular basis for the past year. The mothers of the patients who met the criteria were introduced by the pediatrician in charge to one of the authors. The author explained the study purpose and methods in person, gave instructions for completing the questionnaire, and invited them to participate in the study. The mothers who agreed were then handed the baseline questionnaire and a self-addressed stamped envelope, along with a book coupon of 1000 yen as a token of gratitude. Some of the participating mothers completed the questionnaire and returned it immediately, and the others completed it at home and mailed it back. A reminder call was made when there was no return mail within two weeks. All the participating mothers had learned the basic knowledge about bronchial asthma, including the etiology, mechanisms, diagnosis, treatment, and management, typically through a one-day seminar at the hospital. One year after the baseline survey, i.e., between February 1 and August 31, 2002, a follow-up mail that included the follow-up questionnaire, a self-addressed stamped envelope, and a 1,000 yen book coupon was sent to the mothers who had returned the baseline questionnaire. Data from medical records was also used to obtain clinical information including serum immunoglobulin E (Ig E) level, measured within two years before the baseline survey. The study protocol was approved by the ethics committee of the Institute of Health Science, Kyushu University.
The baseline questionnaire included questions about the mother's age, patient's date of birth, mother's occupation, mother's education, cohabiters, age at diagnosis, factors related to deteriorations, comorbid allergic conditions, family history of allergic conditions, intensity and frequency of attacks in the past year, cohabitation with smokers, adherence, medications, mother's parenting style (see below), and mother's stress or coping behaviors (see below). Factors related to deteriorations were assessed by picking relevant items from a list that consisted of five categories with 4-7 items in each, i.e., an infection category with cold, tonsillitis, etc.; an inhalation category with dust, pollen, etc.; a physical category with exercise, fatigue, etc.; a mental category with anxiety, grief, etc.; and a meteorological category with coldness, heat, etc. The list for comorbid allergic conditions included items for atopic dermatitis, allergic rhinitis, food allergy, drug allergy, urticaria, and others. The intensity and frequency of attacks in the past year was assessed using an intensity × frequency matrix, in which the mother was guided to indicate 1-3 combinations (cells) that met the status of her child's attacks (see Additional file 1). This table was designed based on the Japanese Pediatric Guidelines for the Treatment and Management of Asthma (JPGL) 2000 , which was developed by the Japanese Society of Pediatric Allergy and Clinical Immunology (JSPACI). Adherence was assessed using a set of four questions: (i) Does your child take or inhale medicines as prescribed by his/her doctor? (ii) Do you take care in order to reduce allergens, such as cleaning rooms and bedclothes? (iii) Does your child visit the clinic regularly as recommended by the doctor? (iv) When an attack happens does your child take or inhale medicines immediately? The answers were given on a 7-point rating scale, 1 to 7, where 1 and 7 respectively correspond to "no" and "yes". The adherence score was the sum of the item scores; Cronbach's coefficient alpha was 0.72 in the present sample. Questions on medications asked about the frequency of use in the past week for each inhaled corticosteroid, oral and inhaled β2-stimmulants, oral theophylline, and oral and inhaled anti-allergic agents.
The follow-up questionnaire included questions about the intensity and frequency of attacks in the past year, adherence, and medications, that were the same as those in the baseline questionnaire.
Assessment of mother's parenting style
The mother's parenting style was assessed using the "Parent Form" of the "Ta-ken Diagnostic Test for Parent-Child Relationship" (T-PCRp) . T-PCRp is a self-administered questionnaire that was developed, based on Symonds' theory of parent-child relationship , to assess "unfavorable" parenting attitudes in terms of three axes; "rejection--acceptance", "dominance--obedience", and "inconsistency and disagreement". Additional file 2 shows the construction of this instrument, which consists of five parenting attitudes and two scales corresponding to each of the attitudes. A brief description of key characteristics of the scales is also noted in the table. Each scale consists of 10 questions, and the answers are given on a 3-point rating scale, such as "no", "sometimes" and "always" or "no", "fairly much" and "very much". To each answer, 1, 2 or 3 points were assigned, so that a higher score indicates a stronger tendency. Reliability based on the split-half method was high (r = .90) . Because all 10 scales represent unfavorable attitudes toward parenting, we hypothesized that a strong tendency toward any of the attitudes could negatively affect the status of a child's asthmatic symptoms (as indicated by the symbols "•" or "••" in the table of Additional file 2). Special emphasis, however, was placed on three scales, "passive rejection", "active rejection", and "doting", on which mothers of children with strong airway contraction reportedly score high (indicated by "••" in the table) .
Assessment of the mother's stress and coping behaviors
The mother's stress and coping behaviors were assessed using the Stress Inventory (SI). SI is a self-administered questionnaire that was developed, based on Grossarth-Maticek's theory of disease-prone stress/personality , to assess response styles to stressors principally related to an interpersonal relationship or to chronic stress posed by the response style . The detailed developmental process and psychometric properties [22, 23], brief explanations , an English translation of items , and some empirical results suggesting its external validity [24, 25] are described elsewhere. Additional file 3 presents the 12 SI scales and brief descriptions. Each scale typically consists of four questions, and the answers receive a 1 to 6 rating, where 1 and 6 respectively correspond to "yes" and "no" or to "almost always" and "rarely". Cronbach alphas  and test-rest reliability coefficients ranged from 0.60 to 0.90 and from 0.66 to 0.82, respectively . SI was developed principally to assess stress-coping styles or chronic stress that were thought to be risk factors for cancer and cardiovascular disease, and its measured constructs are based on the subject's individual experiences. However, because the stresses that this instrument focuses on are mainly those posed by the pattern of the interpersonal relationship, we thought that some of the patterns might affect those around her through her language and behaviors, especially her children, who are close others. Specifically (as indicated by "••" in the table of Additional file 3), we hypothesized that the following maternal conditions would lead to lowered functioning of care and/or language and behavior that can be stressors to a child, and in turn to negative effects on the child's asthma: (i) a mother's chronic irritation and anger as represented by the SI scales "object dependence of anger" and "annoying barrier", (ii) mother's egocentric and egoistic behaviors as represented by "egoism", and (iii) mother's ambivalent and unstable behaviors, such as anger and depression in interpersonal connections, as represented by "object dependence of ambivalence". We also hypothesized that the following conditions ("•" in the table) would also be predictive of a child's poorer disease status, though more weakly than the above three, because these were thought to be related to lowered functioning of care, but not to stressors to the child: (iv) decreased sense of control over stressful situations ("low sense of control"), (v) chronic sense of hopelessness and depression ("object dependence of loss"), (vi) chronic feelings of unfulfilled needs for acceptance by others ("unfulfilled needs for acceptance"), (vii) altruistic tendency accompanied by stress ("altruism"), and (viii) an extreme tendency to react rationally to stressful interpersonal situations ("rationalizing conflicts/frustrations").
The designated cells in the intensity × frequency matrix in the baseline and follow-up questionnaires were related to the disease severity according to the JPGL 2000 classification, as demonstrated in Additional file 4. When more than one severity was indicated, the highest was adopted. Patients with a condition better than Mild disease were classified as in Remission. Thus, the severity was determined for one of the four categories, Remission, Mild, Moderate, and Severe. A composite variable for anti-asthmatic agents was created according to the JPGL 2000 scoring system, where the medication score was calculated as the sum of points that are given to agents, e.g., 10 points per 5 mg of oral predonisolone, 10 points per 500 μg of inhaled betamethazone, 2 points per 1 oral dose of long-acting theophylline, 1 point per 1 inhaled dose of cromoglycate, etc. An ordinal variable was created for the number of comorbid allergic conditions, where the values were the counts of designated items taken from among atopic dermatitis, allergic rhinitis, and others. An ordinal variable was also created for the number of adult cohabiters, where the values were 1 = mother only, 2 = mother plus 1, and 3 = mother plus 2+.
Statistical analysis was performed in five steps as follows. First, a series of univariate regression analyses was used to identity baseline non-psychosocial factors that were significantly (P <.05) associated with the disease severity at 1y follow-up. Thus, the outcome variable was severity at 1y follow-up (Remission = 1, Mild = 2, Moderate = 3, Severe = 4), and the explanatory variables considered were gender, age at baseline (in years), age at diagnosis (in years), duration (in years), severity at baseline (1, 2, 3, 4), the medication score (continuous), number of comorbid allergic conditions (0, 1, 2, 3), parental history of bronchial asthma (present/absent), number of factors reported to cause deterioration, serum Ig E (logarithm of measurement value), mother's age (in years), mother's education (in years), mother's occupation (housewife/working mother), number of adult cohabiters (1, 2, 3), and living with smokers (present/absent). The above identified variables were then put into a single multiple regression model, in order to identify non-psychosocial covariates that were to be considered for the subsequent analyses of psychosocial factors. Second, a series of univariate regression analyses was performed for psychosocial factors, i.e., mother's parenting styles and mother's stress/coping behaviors. Thus, for each model the outcome variable was the severity at 1y follow-up and the explanatory variable was one of the scale scores of T-PCRp and SI. A series of multiple regression models was then used to adjust for the above identified non-psychosocial covariates. Third, in order to examine a possible confounding effect of adherence on the association between the mother's psychosocial factors and the child's disease severity, both adherence at baseline and adherence at 1y follow-up were added to the series of multiple regression models, including a psychosocial variable and the non-psychosocial factors. Fourth, possible differences in the effects of the mothers' parenting style and stress over their children's asthma according to the child's age were examined by repeating the above analysis, stratified by age categories of < 7 years and 7 and over years (before or after school age). Finally, associations found to be significant in the above stratified analysis were further examined in terms of odds ratio (OR). ORs of moderate/severe disease vs. remission/mild disease at follow-up for high and intermediate categories relative to low category, as classified by tertiles of scale scores, were estimated using multiple logistic regression adjusted for non-psychosocial covariates. All P-values were two-sided, and considered significant at P < .05. All analyses were done using the SAS version 9.3 (SAS Institute Inc., Cary, NC).