PHYSICIANS and BREASTFEEDING: BELIEFS, KNOWLEDGE. SELF-EFFICACY and COUNSELLING PRACTICES MARIA JEAN BURGLEHAUS. B.Sc, McGill University, 1991

Transcription

1 PHYSICIANS and BREASTFEEDING: BELIEFS, KNOWLEDGE SELF-EFFICACY and COUNSELLING PRACTICES by MARIA JEAN BURGLEHAUS B.Sc, McGill University, 1991 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE MASTERS of SCIENCE in THE FACULTY OF GRADUATE STUDIES (Department of Health Care and Epidemiology) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA April 1995 Maria Jean Burglehaus, 1995

2 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission: Department The University of British Columbia Vancouver, Canada DE-6 (2/88)

3 ABSTRACT Reversing the problems of breastfeeding failure and bottlefeeding choice depends on physicians who are skilled in breastfeeding support and management. The following study aimed to assist an urban British Columbian hospital intending to develop breastfeeding education for its physicians. The study sought specifically to determine which predisposing factors to target in order to improve physicians' willingness, motivation and ability in counselling about and managing breastfeeding. The aims of the investigation were: 1) to provide baseline data on the physicians' attitudes, beliefs, knowledge and self-efficacy concerning breastfeeding counselling; and 2) to identify which measures including cognitive factors, gender, specialty, years in practice and personal or spousal breastfeeding experience might be independently predictive of physicians' self-reported counselling behaviours. A pilot-tested survey was mailed to the offices of all 325 obstetricians, pediatricians, family practitioners and general practitioners with privileges at the hospital. Response rate was 67.3 percent. The female physicians were much more likely to respond than male physicians; response rates of 86 and 57 percent respectively. Ninety percent of the respondents reported always or usually discussing breastfeeding with their patients prenatally (65% always do) and 88% reported always or usually attempting to convince mothers to breastfeed if they intend to bottlefeed (41% always do). Correlational data showed weak to moderate associations between the reported cognitive factors and counselling behaviours. The strongest of these associations suggest that physicians counsel more if they feel confident and if they believe strongly in the immune properties of breastmilk.

4 Female physicians expressed greater self-efficacy in counselling the mom about breastfeeding problems and in positioning the baby at the breast (p <.001). Self-efficacy scores were higher for both male and female physicians whose children were breastfed (females, p <.001; males, p<.01). The measures of self-confidence, knowledge and beliefs were added to a regression model containing measures of gender, specialty, years in practice and personal or spousal breastfeeding experience to determine whether additional variance in the counselling behaviour could be accounted for. Physicians attempted to convince women to breastfeed if they believed in the immune properties of breastmilk (Beta =.63, p=.08) and were confident in their own breastfeeding counselling (Beta =.21, p=.004). Likewise, encouraging women to continue breastfeeding in the face of breastfeeding problems was related to belief in the immune properties of breastmilk (Beta =1.04, p=.021) and confidence in breastfeeding counselling (Beta =.20, p=.038). Knowledge of how to position the infant at the breast was associated with whether physicians reported discussing breastfeeding with the patients (p<05). However, when asked to identify which of the two pictures of the infant at the breast is in fact correct, 12% of physicians chose the wrong picture. Physicians also felt that supplementation was indicated for twins, infants with a difficult latch (suckling technique), and in cases requiring an emergency C-section. These are not, however, medical indications for supplementation. Despite more than a decade of research suggesting the need for breastfeeding education to physicians, little if any education has been provided. Revision of medical school curricula and development of continuing education are necessary to improve physicians' interest in breastfeeding and to provide specific skills that will increase physicians' competence and self-efficacy in counselling patients about breastfeeding.

5 TABLE OF CONTENTS Abstract» Table of Contents List of Tables List of Figures Acknowledgement iv vi viii ix INTRODUCTION 1 Benefits of breastfeeding 1 Infant feeding recommendations 2 Physicians do not meet breastfeeding counselling recommendations 3 Chapter 1: Literature Review... 6 Theoretical background 6 Research objectives 16 Chapter 2: Methodology, Variables and Data Analysis 18 Methodology 18 Instrument development and pilot-testing 18 Mailing of the questionnaire 19 Dependent and Independent Variables 20 Methods of analysis 22 Reliability analysis 23 Descriptive statistics and univariate and correlational analyses 24 Regression analyses 25 Chapter 3: Results 27 Feedback from the pilot test 27 Results of the mailing-out and data collection processes 27 Response rates and characteristics of the sample 28 Reliability analyses results 34

6 Univariate and correlational analyses results 36 Counselling behaviours: descriptive statistics and differences by gender, specialty, years in practice, ethnicity and personal or spousal breastfeeding experience 36 Physicians' beliefs, self-efficacy and knowledge: descriptive statistics and differences by gender, specialty, years in practice, ethnicity and personal or spousal breastfeeding experience 41 Counselling behaviours by self-efficacy, beliefs and knowledge 43 Predisposing variables: gender, specialty, years in practice, ethnicity and personal or spousal breastfeeding experience differences 45 Results of the regression analyses 49 Chapter 4: Discussion 54 Counselling behaviours 55 Methodological limitations 61 Implications for future research 64 Implications for policy and education 66 Bibliography 71 APPENDIX 1: Final Survey 77 APPENDIX 2: Selected Regression Analyses Results 86

7 List of Tables Table 1. Summary of breastfeeding studies: settings and response rates 13 Table 2. Components to be included in the thesis 16 Table 3. Flow of the mailing-out and data collection processes 28 Table 4. Response rate by physician specialty 29 Table 5. Response rate by gender 31 Table 6. Response rate by years in practice 32 Table 7. Response rate by ethnic origin 32 Table 8. Gender and years in practice by physician specialty of the respondents and study population 33 Table 9. Personal or spousal breastfeeding experience 34 Table 10. Counselling behaviours 35 Table 11. Occurrence of counselling by counselling behaviours categories 38 Table 12. Counselling behaviours by gender and ethnicity 39 Table 13. Dichotomized counselling behaviours by gender 40 Table 14. Counselling behaviours by specialty 40 Table 15. Belief levels concerning breastfeeding 42 Table 16. Self-efficacy and knowledge variables 43 Table 17. Counselling behaviours by self-efficacy, beliefs and knowledge 43 Table 18. Beliefs by specialty 45 Table 19. Beliefs by gender and by spousal or personal breastfeeding experience Table 20. Mean self-efficacy scores by gender 47 Table 21. Knowledge by specialty, gender, years in practice and personal or spousal breastfeeding experience 47 Table 22. Preferred methods to receive information on breastfeeding 48 Table 23. Reports of indications for supplementation 49

8 Table 24. Final regression models 51 Table 25. Findings from this study and previous studies: comparison and contrast Table 26. Counselling behaviours: a comparison to similar studies 55

9 List of Figures Figure 1. Short-term and mid-term impacts of the breastfeeding education program. 11 Figure 2. Hypothesized regression model and potential components 26

10 ACKNOWLEDGEMENT I would like to acknowledge the thesis supervisor, Dr. Lawrence W. Green, whose teachings and guidance have been integral to the thesis. My sincere thanks to Dr. Samuel Sheps and Dr. Lorie A. Smith for their crucial input and dedication. Funding from the Department of Family Practice of B.C. Women's Hospital and Health Centre Society and by the B.C. Medical Services Foundation as well as the use of computer software at the Institute of Health Promotion Research, UBC, were much appreciated. I would also like to thank Margaret Cargo affectionately for her advice, encouragement and motivational enthusiasm.

11 1 INTRODUCTION Benefits of breastfeeding Recent reviews of the overall reduction in risk of death with breastfeeding suggest that one-third to one-half of current infant deaths in North America are because of a failure to breastfeed fully (i.e., to give breastmilk exclusively for the first 4 to 6 months of age, then breastmilk plus solid food until 12 months) (Damus et al., 1988; Madeley et al. 1986). One study estimated that between 1 and 12 months of age, the U.S. infant that is not fully breastfed runs three times the risk of dying compared to the fully breastfed child (Adebonojo, 1972). Problems that a baby can experience when fed with formula or other non-human milks include more respiratory and gastrointestinal illnesses, allergies and ear infections, reduced psychological development and neurodevelopment and increased risk of sudden infant death (Casey & Hambidge, 1983; Lucas et al., 1990 & 1992; Damus et al., 1988). Epidemiologists currently suggest that breastfeeding appears to provide substantial protection for the mother too: protection against breast cancer (Layde et al., 1989; Byers et al., 1985), ovarian cancer (Whittemore, 1993) and osteoporosis (Aloia, 1985). Partially breastfeeding infants in which breastfeeding is augmented with an artificial baby milk is not recommended, unless medically indicated; 3 to 5% of mothers are medically unable to breastfeed. All levels of health care have endorsed breastfeeding as the preferred method of infant feeding. The Canadian Paediatric Society (1991) supports the WHO - UNICEF recommendation that infants should be breastfed exclusively for the first four to six months of life. The first milk or colostrum is particularly nutritious. While continued breastfeeding is best, any number of weeks of breastfeeding is beneficial. Solids are introduced at five to six

12 2 months. The infant is breastfeeding less often and the mother's body adapts so that each breastfeed becomes more potent in terms of immunological properties (Minchin, 1985). Infant feeding practices fall short of recommendations Currently, in Canadian hospitals, 80% of new mothers initiate breastfeeding. This figure does seem to be an improvement over the 1982 rate of 70%; it is definitely much better than initiation rates of 38% in 1963 (Health and Welfare Canada, 1990). However, less than one in four newborns leave hospital having been fed only breastmilk and 71% of infants receive supplements in hospital. This high rate of supplementation leads one to question the usefulness of an initiation rate figure of 80 per cent. Among the supplemented infants there is a high rate of premature termination of breastfeeding. Availability of formula gives staff an excuse not to teach mothers how to overcome difficulties (Frank et al., 1987; Bergevin & Kramer, 1983). It has been estimated that approximately 50% of those who stopped breastfeeding in the first six weeks might have continued to breastfeed if they had received better help or teaching in hospital (Health and Welfare Canada, 1990). The continuation rates for breastfeeding have not improved in the past 20 years and only 30% of infants are breastfed 1 to six months (Avard & Hanvey, 1989). Supplementing breastmilk with formula at home has been negatively associated with duration of breastfeeding (Frank et al., 1987; Bergevin & Kramer, 1983). Sixty per cent of infants have started solids before 4 months (Davis, 1991). 1 1t is not clear if the authors mean breastfed exclusively to six months or partially breastfed to six months

13 3 The main reason for breastfeeding termination is the woman's perception that she does not have enough milk (Health and Welfare Canada, 1990). This factor accounts for a 20% decrease in breastfeeding rates between four to six weeks. The infant's growth spurt occurs at this time and the infant will be suckling more often to stimulate the mother's body to produce a greater milk supply. Without proper support and information, the mother perceives this increased suckling as a sign that she has insufficient milk (Cerutti, 1981). Some worried mothers will give supplements of formula (a practice that was modeled in the hospital) starting what often becomes a vicious cycle. Supplementation impedes the suckling necessary to stimulate an increased milk supply and the resulting inadequate milk supply perpetuates the need for supplementation. All too often the result is premature termination of breastfeeding. Physicians' breastfeeding counselling practices do not meet recommendations Studies in the 1970s and 1980s suggest that although verbal support for breastfeeding may be strong among pediatricians, strong advocacy and actual support of breastfeeding are much less common. More recently, a sample of 59 pediatricians reported very favorable attitudes toward breastfeeding promotion by pediatricians. Nevertheless, 51 percent reported they routinely recommend breastfeeding while 48 percent reported not making any recommendation but respecting the mother's choice (Michelman et al., 1990). Lawrence (1982) found that 33 percent of physicians from several specialties never initiate the topic of breastfeeding with the mother. Reames (1985) found that physicians discouraged mothers from breastfeeding for reasons that are not considered to be usual contraindications, e.g., Cesarean section, premature birth, maternal diabetes, or because the mother works. Some of the practices recommended by the physicians, such as offering supplemental bottles to breastfed infants, oral contraceptives for breastfeeding mothers, and delayed time of breastfeeding initiation have been shown to affect adversely the success of breastfeeding.

14 4 In a study of pediatricians, very few provided educational support for breastfeeding mothers such as prenatal classes, special counselling or postnatal telephone calls (Michelman etal, 1990). Physicians continue to publish articles, sharing their discoveries that because of their own knowledge gaps they had been providing misinformation to breastfeeding mothers (Newman, 1991). Such advice that leads to decreased milk supply and to breastfeeding failure includes encouraging (or not informing mothers of the consequences of) scheduled feeding, soothers, estrogen-containing contraceptive pills (progestin-only or mini-pills are compatible), limited suckling to prevent nipple damage, or a bottle a day to let the mother rest.' Bagwell (1993), found that physicians felt that fathers feel close to their infants if they can bottlefeed them and that breastfeeding is not a good method of weight reduction nor is it compatible with the use birth control pills. Moxley and Kennedy (1994) present data from the Parent-Baby Information Line in Ottawa, a telephone service to answer parents' question. Over 25% of the calls concern breastfeeding and breastfeeding support groups where mothers can get assistance in breastfeeding problems. A study of Arizona's 61 hospitals providing obstetrical services reported practices that interfere with breastfeeding and promote bottlefeeding; provision of pacifiers and supplemental water or glucose, issuance of formula packs at discharge, and a first feed of sterile water. A positive significant relationship was identified for policies advocating breastfeeding and the prevalence of breastfeeding encouragement from professional staff (Strembel etal., 1991). Greer & Apple (1991), and Newman (1991) warn that a neutral stand about the benefits of breastfeeding by physicians is in fact damaging. These researchers particularly

15 5 discourage the practice of giving formula to mothers. They recommend that physicians increase their breastfeeding management skills, and their knowledge of breastfeeding support groups and of how to advise women on breastfeeding at work. A campaign supporting exclusive breastfeeding among women having chosen breastfeeding was successful in increasing the number of women breastfeeding at one month post-partum (66% of 506 intervention group vs 52% of 151 controls). The intervention was aimed at providing mothers with information, providing support after delivery, raising environmental awareness and educating health professionals. Educating health professionals was particularly efficacious in reducing the physical and medical problems associated with breastfeeding (Macquart-Moulin et al., 1990). A meta-analysis of nine controlled clinical trials found that nursing support with telephone follow-up increased the duration of breastfeeding up to four weeks (Bernard-Bonnin et al., 1989).

16 6 CHAPTER 1: LITERATURE REVIEW Theoretical background B.C. Women's Hospital intends to provide a breastfeeding educational program to the physicians providing care for its maternity patients. The current study aimed to assist the hospital by surveying its physicians' attitudes, beliefs, knowledge concerning breastfeeding and self-efficacy concerning breastfeeding counselling. The co-sponsored World Health Organization - UNICEF Baby-Friendly Hospital Initiative (BFHI) has been an impetus to the educational program. The Initiative aims to promote a social and organizational environment that will enable and reinforce health professionals' supportive behaviours concerning breastfeeding. For example, they recommend early initiation of breastfeeding postpartum, keeping the baby in mother's hospital room 24 hours a day and avoidance of soothers and unnecessary supplementation. The maternity care facilities are to have a breastfeeding policy that is regularly communicated to all staff. The Initiative also states that maternity care facilities are to provide their staff with training in breastfeeding support.

17 7 Continuing education to physicians should be preceded by an analysis of motivators for and obstacles to the desired change in behaviour. This can be referred to as an educational diagnosis (Green & Kreuter, 1991). It is inefficient and sometimes ineffective to train someone in skills to enable a behaviour when that person lacks prior motivation. We have seen evidence that many physicians lack confidence in the efficacy or importance of some preventive maneuvers; of those who accept the value of the procedures, many doubt their own competence or the ability of the patient to make changes. Unless these beliefs are dealt with first, there is little point in training physicians in preventive or health promotion skills (Green & Kreuter, 1991, p ). Based on the literature, an educational diagnosis would seem relevant in planning breastfeeding educational programs directed toward physicians. An educational diagnosis is a process in which information is obtained concerning three categories of behavioural influence: predisposing, enabling, and reinforcing factors. These categories are convenient in that they group the more specific behavioural influences such as knowledge, attitudes and beliefs, skills, incentives, and rewards under broader rubrics according to the measures that might be used to change behaviour. Green et al. (1988) offer this classification as a useful conceptual framework for analyzing physician behaviour and planning interventions to change physician behaviour. They specify that the three domains are not mutually independent. Predisposing factors refer to antecedents to behaviour that provide the rationale or motivation for a behaviour. These factors are within the realm of psychology and "represent the cognitive and affective dimensions of knowing, feeling, believing, valuing, and having selfefficacy or confidence" (Green & Kreuter, 1991, p. 154). Enabling factors are the antecedents to behaviour that allow a motivation to be realized. They include skills, resources or barriers. Reinforcing factors are factors subsequent to a behaviour that provide the

18 8 continuing reward or incentive for the behaviour and contribute to its persistence or repetition. Reinforcing factors include rewards, feedback and colleague support (Green & Kreuter, 1991). The influential factors that determine breastfeeding support among health professionals are the targets of the Baby-Friendly Hospital Initiative and can be classified as predisposing, enabling and reinforcing factors. Further research, such as an educational diagnosis, on these factors may more clearly define specific strategies for implementing the Initiative. The objective of the thesis is to determine which predisposing factors should be targeted in an educational effort directed to physicians. Predisposing factors include knowledge, attitudes, beliefs, values and perceived needs and abilities that might motivate physicians' counselling and supportive practices. Personality factors can also predispose behaviour but are excluded from consideration as they cannot be readily changed by educational or other health promotion interventions. Furthermore, demographic factors including age, gender and family size may predispose behaviour but are not included in the list because they cannot be directly influenced by educational or other interventions (Green & Kreuter, 1991). However, research on physician behaviour could consider physician specialty, gender and other demographics as such subgroups may demonstrate a need for different interventions. The following discussion of predisposing factors will make reference to an area of physician practice which has received much attention in the literature: physicians' support (attitudes, values and beliefs) for preventive practices including health promotion counselling. Apparently physicians favour preventive practices in principle, but hesitate to carry out these activities (Green et al., 1988). Many of the physician factors that have been researched in this

19 9 area will be potential points of consideration in a study of physicians' breastfeeding counselling behaviours. "Physicians' positive attitudes towards health promotion are consistent with patient attitudes toward physicians as being the primary source of health information and effective behaviour change" (Green, Cargo and Ottoson, 1994). Paradoxically, physicians appear reluctant to counsel about risk factors and behaviour modification because they do not think patients want or would follow their advice (Valente, 1986; Cummings et al, 1987; Reed, 1991). Orleans et al. (1985) reported that family physicians considered this pessimism the greatest barrier to their preventive counselling. Not only can the physicians' beliefs about their patients' abilities cause the physician to avoid preventive counselling practices but so can their own lack of confidence in this area. The physicians' training experiences do not traditionally emphasize building competencies in preventive practices. "Up until the latter part of the 1970s, most undergraduate curricula were content driven in the belief that application of Flexnerian principles (that is, a thorough understanding of the basic sciences) would provide the undergraduate with the knowledge and skills to handle most clinical problems" (Piterman, 1991). Studies suggest that recent medical graduates are more confident than other physicians in their counselling effectiveness (Green etal., 1988).

20 10 Green and Kreuter (1991) have described the predisposing, enabling and reinforcing determinants of physicians' behaviour. With respect to predisposing factors: Attitude is a rather constant feeling that is directed toward an object (be it a person, an action, a situation or an idea). Inherent in the structure of an attitude is evaluation, a good-bad dimension. An attitudinal problem among physicians apparent doubt concerning the importance of some behavioural risk factors. relates to their Values include more basic orientations, such as the role of the physician, patient autonomy, and issues of privacy of patient behavior or lifestyle outside the immediate medical realm. These are very important but have not been well studied. Beliefs include the more immediate and changeable viewpoints of the physician on matters such as patients' willingness to change their lifestyles or their ability to change their health practices. Self-efficacy is a perception of one's own capacity for success in organizing and implementing a pattern of behaviour that is new; based largely on experience with similar actions or circumstances encountered or observed in the past (Green and Kreuter, 1991, p ). There is considerable evidence of the interrelatedness of the predisposing factors. Attitudes, for example, appear in analysis as determinants, components, and consequences of beliefs, values, and behaviour (Green & Kreuter, 1991). Regarding self-efficacy, it can be said that the low level of confidence among physicians in preventive counselling can only be improved with adequate skill training. "Specifically, self-efficacy can be enhanced by breaking the complexities of the target behaviour into components that are relatively easy to manage" (Bandura, 1977). Developing awareness of specific situations in which efficacy may be low and rehearsing the desired behaviour in these situations appears to enhance self-efficacy (Gilchrist & Schinke, 1983; Kaplan, Atkins, & Reinsch, 1984).

21 11 Based on the PRECEDE model (Green & Kreuter, 1991), Figure 1 allows conceptualization of the interactive physician factors that influence their breastfeeding counselling and training of childbearing women. FIGURE 1: Short-term and Intermediate Impacts of the Breastfeeding Education Program ^1 Focus of the thesis research from pretest data l Intervention Short-term impacts Predisposing Demographics Attitudes/Beliefs Knowledge Self-efficacy Interest in BE Intermediate impacts Physician practices in support of breastfeeding Breastfeeding Education to Physicians Enabling Skills Access to BE Reinforcing Support from colleagues (BE: Breastfeeding Education to Physicians) If the premise is that education would help shape attitudes, beliefs, self-efficacy and a knowledge base conducive to optimal counselling practices by physicians, then the hypothesis is that a relationship exists between these predisposing factors and counselling behavior. An association between lower than desirable levels of counselling and lower levels in one or more of the predisposing factors (attitudes, beliefs, self-efficacy, and knowledge) would tend to support this hypothesized relationship.

22 12 Explaining the incongruency between current and recommended practices Physicians have attributed the incongruency between current and recommended practices to a lack of formal training (Michelman et al, 1990; Lowe, 1990; Reames, 1985) and a need for further education (Lowe, 1990; Reames, 1985). Lactation education to health professionals has been critiqued in the literature (Hefti, 1992; Livingstone, 1992; Newman, 1991; Tanaka et al., 1990). During obstetric rotation the students and house officers become familiar with procedures known to inhibit lactation such as administering drugs during labour and giving complementary bottle feeds (Rajan & Oakley, 1990). Pediatric assignments include learning how to design human milk substitutes (Naylor, 1990). Published articles and textbooks contain conflicting advice causing confusion among health professionals (Livingstone, 1992). Table 1 summarizes the settings and response rates of several breastfeeding studies concerning physicians. Lowe (1990) investigated levels of knowledge among 161 midwives, 83 general practitioners, 50 nurses, four obstetricians and two pediatricians. A knowledge score was the outcome variable and independent variables included health professionals' personal experiences and years in practice. The scores ranged from 7 to 35 out of 35. A decrease in knowledge was observed with advanced age and more years since training (Lowe, 1990). The survey demonstrated a higher knowledge score (69%) among the 132 women health professionals who had a positive experience of breastfeeding than among the 79 male health professionals (score of 65%). These women also scored higher than did the 65 women who had never breastfed and the 24 women with negative breastfeeding experiences. Lowe (1990) did not statistically analyze these knowledge scores.

23 Table 1. Summary of breastfeeding studies: settings and response rates Study No. of respondents Response Population Rate (%) Lawrence 381 Ped, 306 O/G 64 Ped, 51 O/G U.S. National listing FP 50 FP Reames 88 Ped, 69 O/G 88 Ped, 69 O/G Randomly selected FP, 46 GP 71 FP, 46 GP from 94 U.S. hospitals providing maternity care Michelman et al Peds 86 Large U.S. urban area, telephone survey Lowe 161 Midwives, 83 GP 73 GP Gippsland, Australia 1990 Goldstein & Freed, FP residents 69 Residents in 11 of 14 programs, U.S. Bagwell et al., Dietitians, 75 Dietitians Health professionals 158 Nurses 56 Nurses seeing low income 90 Physicians 39 Physicians women, U.S.

24 14 Bagwell et al. (1993), examined attitudes and knowledge concerning breastfeeding. Interestingly, the knowledge questions address how strongly the respondent agrees or disagrees that breastmilk is completely nutritious and protective against infection. The current study designates these questions as belief questions. Similarly, Bagwell et al. (1993) make use of attitude questions concerning the appropriateness of breastfeeding promotion to mothers facing challenges to breastfeeding such as young age, plans to return to work or mental challenges. They asked physicians whether they would recommend breastfeeding to these mothers. The responses are therefore attitudinal and not self-reports of behaviour, in contrast to the studies by Michelman et. al. (1990) and Goldstein & Freed (1993). Ninety physicians participated in the study by Bagwell et al. (1993): 67 were men, 17 were women and six did not indicate gender. There was a rather low response rate of 39% among the physicians; 75% of the dietitians and 56% of the nurses participated. All dietitians and nurses were women. Bagwell et al. performed a one-way analysis of variance on the aggregate knowledge scores across the professions and then on the aggregate attitude scores across the professions. Out of 100 the physicians', nurses' and dietitians' scores were 75.5, 73.0 and 79.6, respectively for knowledge and 70.2, 74.5 and 78.6 respectively for attitude. Physicians had significantly lower knowledge scores than dietitians and lower attitude scores than both nurses and dietitians (p<.05). Several studies measure counselling behaviour based on physician self-reports. Selfreports of counselling behaviour seem to vary according to physician specialty. Reames (1985) reported that only 44 percent of obstetricians considered breastfeeding to be very important, compared with 74 percent of pediatricians and 65 percent of family practitioners. Lawrence (1982) found that pediatricians are more likely than obstetricians and family practitioners to advocate breastfeeding to the undecided mother. Ironically, pediatricians were

25 15 also the group most likely to recommend supplementation with prepared formula sometimes or always. Michelman et al. (1990) investigated both the degree to which pediatricians promote breastfeeding in their practices, and the attitudes and beliefs about breastfeeding that are associated with their breastfeeding promotion activities. The sample size was small: 59. The gender split among the study subjects is not indicated. Respondents with more supportive beliefs more frequently reported recommending breastfeeding. Seventy-five percent of pediatricians who had highly supportive beliefs usually recommended breastfeeding, whereas only 18.2 percent of pediatricians who had low level beliefs did so (Michelman et al, 1990). No differences in support for breastfeeding among pediatricians according to gender were found and gender failed to differentiate breastfeeding counselling behaviour of physicians in two other major studies: Lawrence (1982) and Reames (1985). In a study of 155 family practice residents (response rate: 69%), Goldstein and Freed (1993) found three predictors of residents' counselling women more than 50% of the time: 1) confidence in counselling abilities; 2) perceived adequacy of training; and 3) female gender. Reames (1985) looked at personal or spousal experience (i.e., having children who were breastfed) and found that physicians who had breastfed a child (or physicians whose spouses had breastfed) were more likely to promote breastfeeding and were more convinced of the beneficial properties of breastmilk. Goldstein and Freed (1993), also found that residents with a personal or spousal experience with breastfeeding displayed greater knowledge and confidence and perceived effectiveness in counselling. They emphasize, however, that personal or spousal experience did not lead to significantly greater reports of breastfeeding counselling, suggesting that such experiences alone do not substitute for actual training.

26 16 Research objectives In reviewing the studies, strengths and limitations were apparent. The aim of this thesis was to overcome limitations by including measures of counselling behaviour rather than looking just at cognitive constructs (e.g. knowledge and attitudes). The principle goals of Goldstein and Freed's study and the Michelman study reflects the focus of the thesis: to determine the counselling practices as well as several predisposing factors among the physicians: knowledge, beliefs, attitudes and self-efficacy. Like in the Goldstein and Freed study, the thesis will attempt to determine the degree to which the predisposing factors predict counselling behaviours. The thesis, however, will make use of data from several specialties and a range of years in practice in attempt to add to the knowledge base. Table 2 summarizes the components of the thesis. Table 2. Components to be included in the thesis Physician behavior: the degree to which physicians promote breastfeeding Beliefs concerning breastfeeding Knowledge of breastfeeding Attitudes concerning breastfeeding Self-efficacy in breastfeeding counselling Several physician specialties Gender Years in practice Physician ethnicity

27 17 The purpose of the thesis was not only to describe the practices of the physicians (dependent variables) and select cognitive (independent) variables, but also to find evidence for the relationships between the independent and dependent variables. Martin and Bateson (1993) suggest that the following conditions be satisfied: 1. a theoretical, conceptual or practical basis for the hypothesized relationship, 2. statistical association, 3. the independent variable precedes the dependent variable, 4. the hypothesis rules out other possible explanations. The statements listed below were the hypotheses for the thesis. These were tested using correlational analysis where a relationship between variables was said to present with an rof at least 0.2 (Martin & Bateson, 1993). 1. Physicians with higher self-efficacy in breastfeeding counselling (higher scores on the two self-efficacy questions) will report counselling more often about breastfeeding. 2. Physicians with stronger beliefs regarding breastfeeding (higher scores on the two belief questions) will report counselling more often about breastfeeding. 3. Female physicians will report counselling more often about breastfeeding. 4. Female physicians will report higher belief, self-efficacy and knowledge scores. 5. Physicians with personal or spousal experience with breastfeeding will have higher belief, self-efficacy and knowledge scores.

28 18 CHAPTER 2: METHODOLOGY, VARIABLES AND DATA ANALYSIS METHODOLOGY Instrument development and pilot-testing A confidential questionnaire was constructed. An opening paragraph explained the purpose of the survey, requested response and assured respondents of confidentiality. Respondents were asked about attitudes, beliefs, knowledge and self-efficacy concerning breastfeeding. There were also questions regarding their practices and their interest in continuing education on breastfeeding. Content validation focussed on whether questionnaire items were based on documented aspects of breastfeeding practices. The first step was a review of the literature to identify substrata of the concept of support for breastfeeding: counselling of the pregnant woman or new mother, referral to a lactation consultant or a community breastfeeding support group, assistance during engorgement or mastitis. The literature was complemented by dialogue and exchange with practitioners. The drafted questionnaire was reviewed by the directors of both the Department of Health Care and Epidemiology (Dr. Sam Sheps) and the Institute of Health Promotion Research (Dr. Lawrence Green) at UBC as well as by the B.C. Women's Baby-Friendly Hospital Committee (BFHC). Dr. Lorie A. Smith (the principal investigator) and lactation consultants are members of the BFHC. The questionnaire was pilot-tested by a sample of physicians providing obstetrical care at another community hospital. The purpose of the pilot test was to ensure that the questions were clear and to determine the time required to complete the questionnaire. The pilot test investigated what questionnaire items appear to measure based on the actual reading of the measure; i.e. face validity.

29 19 Mailing of the questionnaire B.C. Women's Hospital intends to provide an educational program to the family practitioners, general practitioners, obstetricians, and pediatricians providing care to their maternity patients. This project aimed to survey each physician. The sample was, therefore, all 325 family practitioners, general practitioners, obstetricians and pediatricians working out of B.C. Women's Hospital. They were identified through the hospital's mailing list. The principal investigator was not included in the survey. The questionnaires were coded to ensure confidentiality. The identity code allowed the researchers to keep track of those who had not yet responded. The questionnaire was promoted at rounds but this was limited to the Family Practice rounds for reasons that will later be discussed. A postcard reminder was sent to non-respondents three weeks after the first mailing. Three weeks after the postcard reminder a second mailing was made. In the last month of the data collection, follow-up phone calls were made to non-respondents. They were notified of the upcoming deadline for completing the questionnaire and asked whether they required another copy.

30 20 DEPENDENT AND INDEPENDENT VARIABLES The three dependent variables of the regression models were the physicians' self reports of their counselling behaviours or practices. Each was measured by a question and five-point Likert scale of never, infrequently, sometimes, usually and always. The main practice question asked; "How often do you discuss breastfeeding with your patients in the prenatal period?" The two other practice questions were "If a patient is planning to bottlefeed do you attempt to convince her to breastfeed?" and "In the presence of breastfeeding problems, how often do you encourage your patients to continue breastfeeding?". Independent variables included cognitive variables and other variables related to counseling behavior in univariate analysis. There were eight cognitive independent variables to be investigated: two self-efficacy questions, two belief questions, three knowledge questions and one attitude question. The self-efficacy questions asked physicians to rank their confidence on a scale of one to ten concerning; 1) positioning a new mother for breastfeeding, and 2) assisting mothers experiencing common breastfeeding problems. These questions were used in the Michelman study (1990). The self-efficacy variables were on an interval scale and considered to be a continuous variables. The two belief questions were "Exclusive breastfeeding provides all the nutrition required by a healthy newborn up to the age of four to six months with the possible exception of Vitamin D" and "Exclusively breastfed babies have fewer Gl infections, respiratory illnesses, eczema and/or allergic reactions than formula fed babies". These were measured on five-point Likert scales; strongly agree, agree, neither agree nor disagree, disagree and strongly disagree. The scale for each belief variable will be considered an interval scale.

31 21 Three knowledge questions were analysed: 1) a diagram of an infant at the breast, 2) cereals help the infant sleep at night and 3) awareness of Motilium as a means to increase breastmilk supply. Each of the questions has one correct and one incorrect answer and are therefore dichotomous nominal variables. The attitude question asked "In general, how do you feel about the WHO - UNICEF policy to enhance physicians' skills in breastfeeding promotion and breastfeeding support". The responses are on a five-point Likert scale from very favourable to very unfavourable which will be considered to be an interval scale. Controlling variables were included in the statistical model. To identify controlling variables, statistical tests were performed to determine which of the following variables were associated with the independent or dependent variables: specialty, gender, years in practice, ethnic origin and personal breastfeeding experience. The demographics page of the questionnaire provided the necessary information. Dichotomous (dummy) variables were made for each of the physicians' specialties: obstetrician, pediatrician, family practitioner or general practitioner. The "obstetrician" variable, for example, takes the value of 1 for the case where the physician is an obstetrician. Otherwise, its value is 0. Gender was recoded into the variable "female". Females received a code of 1 and males a code of 0. Ethnic origin was a categorical variable with the categories: Caucasian, Asian, Indo- Canadian, Native Indian, African-Canadian, other. These were recoded into a dichotomous

32 22 variable "Caucasian" where Caucasian physicians were coded "1" and non-caucasian were coded "0". Personal or spousal breastfeeding experience was coded into a dichotomous variable "Children". Physicians who indicated that their own children were breastfed were coded as "1", otherwise the value was "0". The physicians were asked to indicate their number of years in practice. This was a continuous variable. It was of interest to obtain data on the demographics of the nonrespondents. The researchers knew the gender and ethnicity of most of the physicians. In a few cases the information was confirmed during the follow-up phone calls to physicians' offices. Years in practice were estimated by finding the year of graduation in the College of Physicians and Surgeons of B.C. directory. METHODS OF ANALYSIS The analyses were concerned with: 1) assessing the reliability of the measures of knowledge, self-efficacy, beliefs, attitudes and counselling behaviour; 2) using the reliable measures in the regression model. The purpose of the regression analyses was to predict the amount of variance accounted for in self-reported counselling behaviour using measures of knowledge, self-efficacy, beliefs and attitudes. Data were analyzed using SPSS-PC at the Institute of Health Promotion Research.

33 23 The following section will discuss the use of reliability analysis and other criteria for the selection of items to be used in the regression model. A brief description of logistic regression will also be provided. Internal consistency reliability analysis methods Ideally, reliability testing of an instrument assesses the extent to which the measures yield the same results on repeated trials (Carmines & Zeller, 1979). An assessment of reliability should consist of testing for internal consistency, stability and equivalence. As the test-retest method was not performed it was not possible to perform tests for stability. Likewise, the study did not include tests for equivalence. The reliability analyses were, therefore, limited to testing internal consistency. Internal consistency refers to the extent to which a set of items measuring the same concept are actually homogeneous. In conditions of internal consistency a respondent will give a similar response to each item measuring a given concept. Cronbach's Alpha, the most widely used method of internal consistency is computed by the following formula: Np/ [1 + p(n -1)] where N is equal to the number of items in the scale and p is equal to the mean inter-item correlation (Carmines & Zeller, 1979).

34 24 Descriptive statistics and univariate and correlational analyses Analyses were performed to describe the data, including measures of central tendency and standard deviation. Correlations were used to show the associations between counselling behaviours and physician factors. The strength of association is indicated by the size of the correlation coefficient. Coefficients range from -1.0 to A coefficient of zero indicates that there is no linear association between the two variables. Generally a significant correlation represents a correlation that differs significantly from zero (Martin & Bateson, 1993). The use of the parametric test of correlation, the Pearson correlation, requires that these assumptions are met: normality, homogeneity of variance and measurement on an interval or ratio scale. "The Pearson correlation is, however, reasonably robust when there is departure from normality" (Martin & Bateson, 1993, p. 137). The point biserial correlation is the appropriate test for testing the association between an interval variable and a dichotomous variable. The correlational analyses were carried out in the following manner. First, a series of analyses assessed whether gender, specialty, years in practice, ethnicity or personal breastfeeding experience were correlated with counselling behaviours. Secondly, correlations looked at the associations between counselling behaviour and the predisposing variables (beliefs, self-efficacy and knowledge). Finally, a series of analyses detected the presence of gender, specialty, years in practice, ethnicity or personal breastfeeding experience differences in the predisposing variables.

35 25 Regression Analyses Regression analyses were performed to identify which variables, including selfefficacy, beliefs and knowledge might be independently predictive of physicians' self-reported counselling behaviours. In regression analysis, each independent variable (X) is given a regression coefficient as an estimate of how much of the changes in Y (slope) are attributable to changes in X. Regression is generally more restrictive than correlation tests. The assumption that must accompany the information obtained from regression is that the dependent variable is in fact a linear function of the independent variable. It is not necessary to designate which is the dependent and independent variable when obtaining the correlation coefficient (Schroeder, Sjoquist & Stephan, 1988). The original intent was to use a counselling behaviour composite score as the dependent variable of the regression model. As will be discussed later, the Pearson correlation coefficients indicated that the individual self-reports of counselling behaviours were only moderately associated with each other which made it impractical to construct a composite score. Therefore, each of the three counselling behaviours were analyzed individually. The five-point Likert scales of the individual counselling behaviour items limit their variability. Logistic regression was used as this method is compatible when the dependent variable displays limited variability. In logistic regression the dependent variable must be dichotomized. Thus, the counselling behaviour variables were dichotomized with "usually" and "always" grouped together and "sometimes", "infrequently" and "never" grouped together. This classification scheme reflects the goal of an education intervention; to target the factors that are associated with lower rates of discussing and advocating breastfeeding to patients, implied by the terms "sometimes", "infrequently" and "never".

36 26 The following procedure was performed for each of the three counselling behaviour variables. Construction of the regression model was achieved through a model-building approach. Figure 2 outlines potential components of the model. The controlling variables were the independent variables such as years in practice, specialty, personal or spouse's personal breastfeeding experience and gender that were found to be associated with counselling behaviour in the univariate analysis. The cognitive variables were then entered into the model first individually and then with each other. All of the significant variables were entered into a final model which was assessed for significance. The regression method was forward stepwise regression. Figure 2: Hypothesized regression model and potential components Effect of Components on Counselling Behaviour: Counselling behaviour 1 = C1 + C2 + C Knowledge1/knowledge2/knowledge3 +attitude + beliefl/ belief2 +self-efficacy composite score dependent variable controlling variables cognitive variables added stepwise 1 Regressions were run for each of the three counselling behaviours The significance level for the beta weights in the regression analyses was set as 0.10 as this level is generally seen to be a trend towards significance in behavioural research. The effects that are significant up to the 0.10 level may be important for health promotion programming. These trends may also identify areas requiring further research.