A survey of physician beliefs and self-reported practices concerning screening for early detection of cancer

Pergamon

0277-9536(93)EOO81-0

Sot. Sci. Med. Vol. 39, No. 6, pp. 841-849, 1994 Copyright 0 1994 Elsevier Science Ltd Printedin Greal Britain.All rights reserved 0277-9536/94 57.00 + 0.00

A SURVEY OF PHYSICIAN BELIEFS AND SELF-REPORTED PRACTICES CONCERNING SCREENING FOR EARLY DETECTION OF CANCER CARLA M. CLASEN,’ SALLY W. VERNON, ‘* PATRICIA D. MULLEN’ ‘The University of Texas Health Science Center

Houston,

and

GILCHRIST L. JACKSON*

at Houston, School of Public Health, TX 77225, U.S.A. and *The Cancer Prevention Center of the Kelsey-Seybold Medical Research and Education, Houston, TX 77030, U.S.A.

P.O. Box 20186, Foundation for

Abstract-Cancer is the second leading cause of death in the United States. Early detection of cancer greatly improves 5-year survival for many sites, and in 1980 the American Cancer Society (ACS) published recommendations for performing cancer screening with the goal of promoting early cancer detection in asymptomatic persons. This cross-sectional survey examined beliefs and practices related to six cancer screening tests and procedures in a group of 68 primary care physicians in a multi-specialty group practice in Houston, Texas. Constructs from the Health Belief Model and Social Cognitive Theory were used to identify factors that might influence performance of cancer screening. Physicians in this study reported greater compliance with ACS recommendations for performance than has been found in other studies, and there is an indication that some screening tests may be performed even when not indicated based on age-specific criteria. Respondents reported performing digital rectal examination, stool occult blood testing, and sigmoidoscopy more frequently in men than in women. No belief factor emerged as being associated with performance of all screening procedures, and associations that were noted for some procedures were not consistent across patient age and gender groups. Some possible directions for further research and development of programs to promote the appropriate and cost effective use of cancer screening are physician education to include information about age and gender appropriate guidelines for screening and opportunities for skills training and practice workshops for some procedures. Key words+ancer

screening,

physician

beliefs and behavior

Cancer is the second leading cause of death in the United States after heart disease [ 11. For many cancer sites detection of the cancer early in the course of the disease, when treatment is more likely to be successful, improves survival [2]. Regular use of screening methods can detect cancers of the breast, colon, rectum, cervix, and prostate at an early stage. In 1980 the American Cancer Society (ACS) published recommendations for performing cancer screening tests with the goal of promoting early cancer detection in asymptomatic persons [3] (Table 1). Despite these recommendations and the ease of administration of most screening procedures, numerous surveys of physicians and reviews of medical records have found that cancer screening is not consistently performed as recommended. Interventions to improve physician compliance with screening recommendations have met with varying degrees of success but have not increased compliance with guidelines above 50% [4-131. One possible explanation for this failure to greatly affect compliance is that the interventions chosen do not address all the

*To whom correspondence

salient barriers and beliefs that prevent ance with guidelines.

full compli-

AIMS OF THE STUDY

This cross-sectional, descriptive study was undertaken to gather information about cancer screening beliefs and behaviors among primary care physicians in a multispecialty group practice in Houston, Texas. The main questions of interest included: 1. What are physicians’ self-reported frequencies of performing cancer screening tests and procedures? What are physicians’ beliefs that may either impede or facilitate cancer screening, including outcome expectancies for cancer screening tests and procedures, and perceived barriers to performance of the tests? What are physicians’ self-effic-cy beliefs about their ability to perform or recommend screening and obtain patient compliance? To what extent are physicians’ beliefs associated with self-reported performance of cancer screening tests and procedures?

should be addressed. 841

842

CARLA M. CLASEN et ut. CONCEPTUAL

MODELS

In an attempt to construct a relevant and parsimonious conceptual model of physician behavior in relation to cancer screening, constructs from both the Health Belief Model [ 141 and Social Cognitive Theory (formerly called Social Learning Theory) [IS, 161 were used to delimit factors that may affect physicians’ performance of cancer screening. The Health Belief Model (HBM) hypothesizes that health-related behavior depends on the value an individual places on a particular goal and the expectation that a given action will help to achieve that goal. Specifically, the HBM postulates that health-related behavior depends on (1) the individual’s estimate of his perceived susceptibility to a condition, (2) the perceived severity of the condition, (3) the perceived benefits of a course of action in reducing the threat, weighed against (4) the perceived barriers to that course of action, and that furthermore, (5) a ‘cue to action’, either internal or external, is necessary to trigger the appropriate health behavior [l7]. The HBM has been used extensively to explain and predict laypersons’ actions in relation to primary prevention. screening, and sick-role behaviors [18]. As yet, there have been no attempts to assess physicians’ behavior relative to providing preventive health care by using concepts from the HBM. It would appear to be possible to do so, however, by approaching each construct from the perspective of the physician, that is, the physician’s estimate of a patient’s susceptibility to a condition. the physician’s estimate of the severity of the condition, and the benefits of and barriers to a course of action available to the physician in meeting the threat. Social Cognitive Theory (SCT) explains human behavior, including health behavior, as a reciprocal situation in which behavior, environmental factors, and personal factors all interact and influence each Table

I. American

Test or procedure

Cancer

other simultaneously. Important concepts of SCT are (1) expectancies about environmental cues, (2) outcome expectancy, or the expectation that a particular behavior will lead to a certain outcome and the desirability of that outcome, and (3) self-efficacy expectations, or the person’s belief about his/her ability to perform the behavior [15, 191. Similarities exist between the two models. and two of the HBM’s chief proponents have proposed that the HBM and SCT be integrated by including selfefficacy as a separate independent variable in the HBM [20]. We suggest that a further refinement and integration of the two theories occurs when outcome expectancies (which may hold either positive or negative valences) are seen as functioning as either benefits of or barriers to a course of action. A separate concept of benefits and barriers needs to be retained in a combined model, however, since some potential barriers or benefits may exist independently of expected outcomes. The HBM constructs of perceived susceptibility and severity are hypothesized to be less useful than barriers and benefits in explaining a physician’s ordering or performing cancer screening tests or procedures because it is assumed that physicians believe their adult patients are susceptible to cancer given appropriate risk profiles and that most types of cancer pose a significant threat to well-being. METHODS Study

population

The target population consisted of 68 physicians engaged in adult primary care medicine at the Kelsey Seybold Clinic (KSC), P.A., in Houston, Texas. The KSC is a multi-site, multispecialty clinic that has been in operation since 1949. Until 1983, the clinic was a multispecialty private practice with no relationship to managed care. Since 1983. however, the clinic has

Society recommendations tomatic persons

for cancer screening

m asymp-

Gender

Age

Digltal rectal exammation

Male and Female

40 and over

Every year

stool occult blood test

Male and Female

50 and over

Every year

Sigmoidoscopy

Male and Female

50 and over

Every 3-5 years

Mammography

Female

35-39 40-49 50 and over

BaselIne Every l-2 years Every year

Clinical breast examination

Female

20-40 owzr 40

Every 3 years Every year

Pap test

Female

I8 and over, or sexually active.

If 3 consecutive normal annual exams, may be done less often at physictan’s discretion.

Frequency

Physician beliefs about cancer screening contracted with health maintenance organizations (HMOs) and other types of managed care plans including preferred provider organizations, point-ofservice plans, and exclusive provider organizations. In 1992 approx. 65% of patient visits were by HMO members. The same health care providers serve patients regardless of the type of medical insurance and the same screening policies and coverage, procedures are followed for all patients. Physician specialties included in the survey were internal medicine, family medicine, occupational medicine, and obstetrics/gynecology. At the time of the survey in 1992,50 of the physicians were male and 18 were female. The racial/ethnic composition was: 53 white, 2 black, 2 Hispanic, and 11 other (e.g. Asian, Egyptian, Indian). Physicians perform all cancer screening tests and procedures, and recommendations concerning cancer screening are at the discretion of the physician. The nursing staff does not perform any of the tests or procedures; however, they may provide patient education when patients call for information about the frequency of or need for these tests. Definitions and measures of the variables The survey instrument was a ‘I-page self-administered questionnaire with a fixed choice format. One series of questions asked whether the physician usually performs each test as part of a complete medical examination of asymptomatic male and female patients in the 20-39 year, 4&49 year, 50 years and older age groups. A performance variable for each procedure was defined by whether or not the physician reported he/she would perform the test or procedure as part of a general check-up examination. Sociodemographic variables included gender, year of birth, year of graduation from medical school, and medical specialty. Physicians’ beliefs about factors related to performance of cancer screening were assessed using constructs from SCT and the HBM as described below. Outcome expectancies. According to the HBM, performance of cancer screening, like any health-related behavior, depends on the physician’s belief that there is some benefit to recommending or performing it. Since all possible benefits of cancer screening appear to be related to expected outcomes, the construct of benefits was subsumed under outcome expectancies. For each screening procedure, respondents were asked to agree or disagree on a 5-point Likert-type scale with a series of statements intended to reflect outcome expectancies specific to that test or procedure, including that early detection of cancer at a particular site is effective in increasing length of survival and chances for cure; that performance would lead to a decrease in mortality from cancer of specific sites; that performance would fulfil patients’ expectations for medical care; and that performance

843

would provide opportunities for patient education and for detection of cancer and other pathology. Physician agreement with these statements would presumably be seen as benefits of the screening test, and should therefore be associated with high levels of performance. Another outcome expectancy, that a patient will suffer pain or discomfort as a result of performance of a screening test, would presumably be seen as a deterrent or barrier to performance, and agreement with this statement should be associated with lower rates of performance. For analysis, responses were compressed into 3 categories: agree, disagree, and neither/not sure. Self-eficacy. Self-efficacy measures included beliefs about the respondent’s ability to perform correctly a given screening test or procedure; to instruct patients about procedures; and/or to persuade patients to have procedures. It was expected that high selfefficacy beliefs would be associated with high levels of reported performance. Conversely, low self-efficacy for a particular test would be associated with low levels. Self-efficacy for each procedure was assessed by questions using a 5-point Likert-type scale. For analysis, responses were compressed into 3 categories: confident, not confident, and neither. Barriers. It was hypothesized above that negative outcome expectancies and low self-efficacy would function as barriers to performance of a screening test or procedure. In addition, seven other barriers were measured for each screening test or procedure: cost of the procedure, risk associated with the procedure, time, equipment, and experience required for performance, degree of patient discomfort, and inconvenience of scheduling the test. Physicians were asked to indicate whether the item might discourage their performance of the test or procedure. A barrier score was calculated for each test or procedure by counting the number of items indicated. Barrier scores were dichotomized into 2 or fewer vs 3 or more barriers. In addition, barriers for each test or procedure were assessed separately. Other potential barriers related to performance of cancer screening in general were: physician forgetfulness that a test is due; patients’ unwillingness or inability to comply; patient embarrassment about the procedure; physician discomfort performing a cancer screening procedure during a patient visit for an unrelated complaint; and lack of reimbursement for the procedure. These barriers were measured with a 5-point Likert-type scale and were compressed into 3 categories for analysis. Administration and response The survey instrument, with a personalized cover letter and return envelope, was distributed at the regular staff meeting of each specialty department by the University researchers during the summer of 1992. Physicians not attending the meeting were sent a survey instrument with a cover letter from the investigators explaining the study and a return envel-

CARLA M. CLASEN Ed al. Table

2. Demographic

Demographic

description

of target

Respondents

characteristics

populatmn

(n = 53)

and survey Target

respondents

populatlo”

(n = 68)

n

%

n

%

40

75

50

74

Gender Male Specialty Fanuly

medtcine

Internal

medicine

Obstetrics/gynecology Occupational Age

medicine

14 24 9 6

IX

26

32

47

17

II

II

I6

I0

7

(in years)

under

II 21 4 x 6

35

3544 45-54 55-64 65 and over

?I

Data

40

available

not

8 I5 II 6

Missme Mean

26 45

= 45 years.

SD = 13.2 years.

ope. Physicians who had not returned questionnaires after 2 weeks received a replacement questionnaire and a cover letter from the chief of their department emphasizing the importance of the study and requesting their cooperation. At the staff meetings and in each letter the confidentiality of the responses was emphasized, and completed questionnaires were returned directly to the University investigators. The response rate was 53% (36 respondents) after the first contact and 78% (53 respondents) after follow-up. Data analysis Response frequencies for reported performance of screening procedures and response categories for belief variables were computed using the SPSSjPC statistical package. The self-reported performance of each cancer screening test or procedure for patients in each of three age groups was examined in relation to beliefs using cross-classification. The chi-square test for independence was used in the bivariate analyses to test for statistically significant associations between performance and belief variables. Lambda was calculated for associations that were statistically significant in the chi-square analysis at P < 0.05. Lambda, a measure based on proportional reduction in error and appropriate for nominal data, was used as a measure of the strength of the association. Values of lambda range from 0.0 to 1.0 [21].

RESULTS

Demographic characteristics of the survey respondents are compared with the target population in Table 2. Respondents did not differ significantly from non-respondents on these variables. Nearly all physicians reported at least some familiarity with ACS guidelines for mammography, clinical breast examination, and Pap test, and more than 85% were at least somewhat familiar with the guidelines for performance of digital rectal examination, stool occult blood testing, and sigmoidoscopy.

Self-reported

performance

of cuncer screening

In response to questions about which procedures they would perform during a complete medical examination of patients in each of 3 age groups, physicians in the study population reported high levels of use of all screening tests and procedures, particularly in the 2 oldest age groups. Physicians were somewhat more likely to report performing digital rectal examination, stool occult blood test, and sigmoidoscopy on male than on female patients in all age groups (Table 3). There were no significant differences in the patterns of performance between male and female physicians. The ACS does not recommend screening with digital rectal examination in persons under the age of 40 years. However, 36% of respondents reported performing digital rectal examination in females under age 40, and 49% reported performing the examination in males of this age group. The ACS also does not recommend routine stool occult blood testing in asymptomatic persons under the age of 50, but again a substantial proportion of physicians reported that they performed this test in the 2 younger age groups (Table 3). If age-appropriate performance of cancer screening is defined as performance only in persons in the age groups for which the test or procedure is recommended by the ACS, then only about half of the study population reported performing digital rectal examination and sigmoidoscopy appropriately in both genders, and less than 10% of the study group reported performing stool occult blood testing appropriately (Table 4). Beliefi related to cancer screening Outcome expectancies. Physicians responding to the survey generally indicated that most types of cancer screening were effective and that early detection improved survival. The Pap test, mammography, and sigmoidoscopy were thought to be most effective, with over 90% of the physicians agreeing that they are very or somewhat effective in detecting cancer at specific sites. The procedure thought to be least effective was digital rectal examination for the detec-

Physician

beliefs about

cancer

845

screening

Table 3. Proportion

of physicians performing cancer screening as part of a complete medical examination in asvmDtomatic patients bv gender and age group Females” Age categories

Procedure Digital rectal examination Stool occult blood test Sigmoidoscopy Mammography Clinical breast examination Pap test

Males Age categories

2G-39

40-49

over 50

20-39

4wl9

%

%

%

%b

%’

over 50

%’

(36)d

86

92

(49)

98

100

(36) (2) 47

(86) (25) 91

94 79 100

(51) (4) -

(91) (30) --

98 86

100 100

100 100

100 100

%I = 53 for clinical breast exam, Pap test, and mammography; n = 52 for digital rectal exam, stool occult blood test, and sigmoidoscopy. %I = 45. ?I = 44. dProportions in parentheses () indicate that the screening test or procedure is not recommended by the American Cancer Society for that age group.

tion of rectal pathology, with only 30% agreeing that this procedure is very or somewhat effective. Most physicians (6698%) agreed that performance of each of the tests or procedures would lead to a decrease in mortality from the specific cancer tested for. Over 90% thought that their patients expected them to perform or order mammography, Pap test, and clinical breast exam, while less than half (48%) thought that performance of digital rectal exam was expected. Most physicians (over 900/,) agreed that stool occult blood testing and clinical breast exam provide an opportunity for patient education; these were the only tests for which that question was asked. Self-efficacy beliefs. Physicians were most confident about performing Pap tests (91%) persuading patients to have mammography (90%), and detecting breast masses during physical examination (78%) and least confident about being able to perform sigmoidoscopy safely and correctly (44%). In addition, 90% felt confident that they would be able to persuade a reluctant patient to have a mammogram, compared with 73% who felt confident about persuading a reluctant patient to undergo sigmoidoscopy. Barriers to performance. Barrier scores were highest for sigmoidoscopy (mean = 3.47, SD = 2.53) followed by mammography (mean = 2.49, SD = 2.16), and lowest for clinical breast examination (mean = 0.09, SD = 0.30). Barrier scores for stool occult blood test, digital rectal examination, and Pap test ranged from 0.43 to 0.93.

Table 4. Proportion of physicians reporting performance of colorectal cancer screening in male and female patients in accordance with ACS recommendations Screening Patient gender Males Females

Digital rectal examination 50 48

procedure

Stool occult blood test I 8

Sigmoidoscopy 51 54

For barriers related to cancer screening in general, only 10% reported that they sometimes omit screening because they are unaware that it is due; however, 24% reported that they would probably perform more tests or procedures if reminded when patients were due to receive them. More than half agreed that they usually recommended or performed screening during a general physical examination or check-up visit. Only 16% reported that performing a screening test or procedure during a patient visit for an unrelated matter would make them uncomfortable. Most physicians disagreed that lack of reimbursement for a procedure might affect their ordering or performance. Association of model variables with test performance Reported rates of performance of clinical breast examination, Pap test, and mammography did not vary enough for meaningful cross-classification of performance with belief variables (Table 3). Thus, analysis was performed only for digital rectal examination for colorectal and prostate cancer and for stool occult blood testing and sigmoidoscopy for colorectal cancer. Tables 5 and 6 summarize outcome expectancies, self-efficacy, and barriers that were examined in relation to performance of those procedures in male and female patients in three age groups. Only associations where lambda was greater than zero are reported in the tables. For male patients, two outcome expectancies that can be interpreted as benefits to screening were associated with performance of digital recta1 examination, believing that the procedure is effective in detecting prostate cancer and that it is justified even in younger age groups than recommended. Among the barriers, physician discomfort in performing cancer screening tests during patient visits for unrelated matters was associated with performance of digital rectal examination. Significant associations of these variables with test performance were found only in the youngest patient age group (Table 5). No variables showed significant associations for female patients (Table 6).

CARLA M. CLASEN et ui.

846 Table 5. Self-reported

physician

performance

of cancer screenmg tests and procedures endorsing belief variables (Lambda) Digital

Outcome

rectal exam

in men by age group

Stool occult blood -___ Ages Ages Ages 2&39 4&49 50+

Ages 2&39

Ages 4&49

Ages 50+a

“b

“S

ns

ns

“s

ns “b

“s “s

na

“b not asked

“s

“s “s

90’ (0.43)

ns

“s

“s ns

“b “s

“3 ns

ns

“s

among

physicians

Sigmoidoscopy Ages 2&39b

Ages 4@49

Ages 50+

“S

94’ (0.33)

expecluncies

Performance Effectively Colon Prostate

decreases mortality

ns

“b

detects cancer

Test is reliable Patients will comply Performance justified under a certain age

(Z, “s

ns ns not applicable ns “s

“s “s not apphcable

“5 ns

“S ns

*s “s

60’ (0.23)

ns

even

Self e#icuq Colon Prostate Barrrers 10 perfirmoncc~ Uncomfortable performing during unrelated visits

not asked

“s

$73, Too costly in time and effort Barrier score. 3 or “lore

“S “S not applicable

“s ns not applicable

ns

ns

“s

“s

ns

ns

“S

ns

“s

ns

“s

“s

IS”

7n**

(zY5, not asked

_/

_c

_c

ns = not significant. “All respondents report performance of this test in this age group. bOnly two respondents report performing this test in this age group. lP
For stool occult blood testing, physician discomfort in performing screening during unrelated patient visits was significantly associated with reported performance among males in the 20-39 year old age group (Table 5). Two variables were associated with reported performance of sigmoidoscopy in both men and women

Table 6. Self-reported

physician

performance

of cancer screening tests and procedures endorsing belief variables (Lambda) Digital

Outcome expectancies Performance decreases mortality Effectively detects colon cancer Test is reliable Patients will comply Performance justified even under a certain age

aged 40-49 years: the expectancy that sigmoidoscopy is justified in terms of disease detection even in persons under the age of 50 and belief that convincing patients to undergo the procedure takes too much time and effort (Tables 5 and 6). One variable, the expectancy that performance of sigmoidoscopy will lead to a decrease in mortality from colorectal cancer,

rectal exam

in women by age group among physicians

Stool occult blood

Sigmoidoscopy

Ages 2&39

Ages 4wI9

Ages 50+”

Ages 20-39

Ages 40-49

Ages 50+

ns

ns

“S

“S

“s

“h

ns

“s

ns

ns

ns

“s

ns

“b

“S

“S

ns

“S

“S

*s

“b

“s

ns

“S

ns

“s

“S

not asked “s

“S

not asked

ns

Ages 20-39h

Ages 4w9

Ages 50f

CtK, ns

53”

ns (0.0X)

sel/ eJicl7c.I Colon Barriers

ns

ns

“s

“S

“S

“s

“s

“b

ns

“s

ns

“S

“S

“S

“S

“S

ns

“s

“S

15’ (0.23)

“b

ns

ns

ns

16’

“h

IO performancr

Uncomfortable performing unrelated visits Too costly in time and effort Barrier score: 3 or more

during

not asked

_b

_b

-b

“Only one respondent reported performance of this test in this age group bAll respondents reported 2 or fewer barriers for this test. *P < 0.05. *r/J < 0.01.

Physician beliefs about cancer screening was associated with reported performance of the procedure in both men and women over the age of 50, but to a much greater extent in men (Tables 5 and 6). Sigmoidoscopy was the only test for which the composite barrier score was associated with performance of the procedure. For male patients aged 4049 years, only 18% who named 3 or more barriers to sigmoidoscopy reported performing the procedure, compared with 47% of the physicians who named 2 or fewer barriers. For men over age 50, 78% of the physicians who named 3 or more barriers reported performing the procedure, while 100% who named 2 or fewer barriers so reported (Table 5). For female patients aged 4049, a significantly lower proportion of physicians who named 3 or more barriers to sigmoidoscopy reported performing the procedure (Table 6); however, for these associations lambda was 0.0. DISCUSSION

This group of physicians reported higher levels of performance for six cancer screening tests and procedures than has been found in other studies [22-391. The high self-reported performance of cancer screening by this group of physicians may represent an acquiescent response bias, may reflect a tendency of physicians to err on the side of caution in recommending or performing a test or procedure when not sure whether or not it is indicated, or may be related to the relatively large proportion of HMO patients in this practice. One study found that physicians tend to overreport their performance of screening by about 10% [lo]. However. if reported performance in this group of physicians exceeds actual performance of screening by only lo%, it still indicates over-performance of screening procedures in certain age and gender subgroups and consequently increased medical care costs. A review of medical records for this group of physicians would be useful to determine how closely actual performance of procedures matches selfreported performance. If actual performance supports the data based on self-reported performance, physician education about age-appropriate cancer screening would be important to decrease unneeded testing as well as to increase screening when indicated. Another finding of this survey was that physicians reported performing digital rectal examination, stool occult blood test, and sigmoidoscopy more frequently in males than in females for all patient age groups, although differences in reported performance by gender were not statistically significant. It would be worthwhile to examine whether or not a difference in performance is observed in a larger sample of physicians. The gender difference in reported performance of colorectal cancer screening in this study might reflect the pattern of gender differences in health care that has been reported in recent studies of cardiovascular disease and other conditions [40-42].

847

Most respondents to this survey agreed that cancer screening is effective in detecting disease, although this belief was not significantly associated with performance of any test or procedure except digital rectal examination in males aged 20-39 years. Most also agreed that performance of these tests or procedures would ultimately lead to decreased mortality from cancer for specific sites, and this belief was associated with performance of sigmoidoscopy in persons over age 50. The performance of sigmoidoscopy, followed by mammography, were most affected by barriers of cost, risk, time, inconvenience, patient discomfort, and experience and equipment required. Since mammography was reportedly performed as recommended more often than sigmoidoscopy, it seems that other factors, such as self-efficacy in performing or recommending the procedure, may be involved. A higher proportion of physicians were confident that they could persuade a reluctant patient to have a mammogram than believed they could persuade a reluctant patient to undergo sigmoidoscopy; an even smaller proportion were confident about their ability to perform sigmoidoscopic examination. This lower level of self-efficacy probably results from a lack of training and experience in performing the procedure. Skills training directed at instruction in and practice of sigmoidoscopy technique would provide an opportunity for increasing self-efficacy. It is expected that greater self-efficacy will lead to the performance of more procedures, thus to more experience and increasing expertise. Physicians were far less likely to report that patients expect them to perform digital rectal examination than Pap tests, mammography, and clinical breast examinations. This perceived patient expectation may be one reason why virtually all physicians reported that they recommend mammography and perform clinical breast examination and Pap tests as frequently or more frequently than recommended by the ACS. Physicians’ perceptions of patient expectations for sigmoidoscopy and stool occult blood testing were not explored in this study. It is possible that education of patients about less familiar screening tests and procedures would lead to greater patient expectation that they be performed, which would lead to increased performance if physicians are influenced by what they believe patients expect. Although only 16% of physicians reported that they would be uncomfortable performing screening during a patient visit for an unrelated complaint, most reported that they usually perform screening during a general physical exam or check-up visit. One quarter of the physicians felt that they would probably perform more tests if reminded that a test was due. Since most screening apparently takes place during general check-up visits, appropriate performance of cancer screening might be increased if general screening visits were systematically scheduled and combined with a method of reminding physicians

CARLA M. CLASEN 6’1ul

848

when screening was due. A reminder system also would help physicians remember to perform screening in patients who normally see a physician only with a specific complaint. For procedures where there was sufficient variability of performance to examine associations with beliefs (i.e. digital rectal examination, stool occult blood test, and sigmoidoscopy), no belief was consistently associated with performance across screening procedures or across patient age and gender groups. As noted, this sample of physicians reported greater than expected compliance with screening guidelines based on published literature, and the lack of clear and consistent associations between reported performance and beliefs may be due in part to the lack of variation in reported performance in some patient age/gender categories. Also, variance in belief variables may have been too small in the context of the size of the study population. Another possible reason for the lack of consistency of associations may be that beliefs related to constructs of the HBM and SCT were not correctly identified and tapped in this survey. Still another possibility is that beliefs about screening interact with other factors in a more complex way to influence behavior. This study was undertaken to describe beliefs and practices in a specific group of primary care physicians in a multispecialty group practice. The overall response rate was 78% of those surveyed, and respondents did not differ demographically from the target population. Thus it appears that physicians were representative of this group of physicians. However. since the study group was limited to a relatively small number of physicians in an urban group practice in a sophisticated medical environment, results of this survey may not be generalizable to primary care physicians in other geographic areas or practice settings.

CONCLUSION

The theoretic framework provided by adding the constructs of self-efficacy and a specific type of barrier or benefit, outcome expectancy, from SCT to the HBM was useful in guiding the development of a questionnaire for assessing physicians’ beliefs about the use of screening tests and procedures. Although specific beliefs were not associated consistently with self-reported performance, it is possible that these constructs may be useful in predicting actual behavior. The data presented here suggest that education of physicians about the recommendations for cancer screening needs to focus on eliminating or reducing the inappropriate performance of tests and procedures as well as on increasing performance when indicated. Physicians also may need to have attention called to the apparent difference in frequency of colorectal testing between men and women. For

procedures such as sigmotdoscopy and digital rectal examination for which self-efficacy is lower, opportunities for skills training and practice workshops would be one way to increase physicians’ self-efficacy with performance of those procedures. .4~,kno~~,/r~~c,mr,,,.v~~We thank

Dr Michael Condit, Chief of lnteral Medicine. Dr George Coale. Chief of Obstetrics and Gynecology, Dr Alfred0 Czerwinski. Medical Director of the Kelsey-Seybold Clinic. P. A.. Dr Charles Smith. Chief of Family Medicine. and Dr T. J. Trumble, Chief of Occupational Medicine for their assistance in conducting this survey. We also thank the physicians who took the time to complete the survey. We appreciate the assistance of the staff of-the Joe and Jesse Grump Center for Clinical Cancer Research of the Kelsev-Sevbold Foundation for Medical Research and Education with many tasks related to the project. We are grateful to Vicki Buxton for providing information about the historical development of the KelseySeybold Clinic. PA This research was supported m part though a Biomedical Research Support Grant S07RRO5828-07

REFERENCES

I. Centers for Disease Control. ~Morhid. Mortrrl. IVeek!,~ Rep. 41. 121. lY92. 2. .i\merican Cancer Society. C’tmc,er FO~I.F and Fi,~ures, IYYL? American Cancer Society. Inc., 1992. 3. American Cancer Society. Guidelines for the cancer-related checkup: recommendations and rationale. CA--.4 C‘c07ccr J. Clinic~ium 30, 193, 1980. 4. Cheney C. and Ramsdell J. W. Effect of medical records of periodic health checklists on implementation measures. 401. J. ,Wrc/. 83, 129, 19X7. 5. Frame P. S.. Kowalich B. A. and Llcwellyn A. M. Improving physician compliance with a health maintenance protocol. J. Funr. Prwr. 19, 341, 1984. 6. Harris R. P., O’Mallcy M. S.. Fletcher S. W. and Knight B. P. Prompting physicians for preventive procedures: A live-year study of manual and computer reminders. Am. .I. Prermr. Med. 6, 145. 1990. 7. Lane D. S.. Polednak A. P. and Burg M. A. Effect of continuing medical education and cost reduction on physician compliance with mammography screening guidelines. J Fun,. Prtrc I. 33. 35Y. 1991. C. J.. Hui S. L.. Smith D. M. cr (I/. Re8. McDonald minders to physicians from an introspective computer medical record. A two-year randomized trial. Ann. ;~lrcv-flulMed. loo, IN. 19x4. 9. McDowell 1.. Newell C‘. and Rosser W. Computerized reminder to encourage cervical screening in family practice. J. Funl Prwf. 28, 420, 1989. IO. McPhee S. J., Bird J. A.. Jenkins c‘. N. H. and Fordham D. Promoting cancer screening. A randomized. controlled trial of three interventions ,4rr,/i.v inrc~mtrl Mel/. 149, 1X66, 1989 I I. Nattinger A. B.. Panzer R J. and Janus J. Improving the utilization of screening mammography in primary care practices. Ad?.\ irrrcmnl Med. 149, 2087. 1989. 12. Shank J. C., Powell T. and Llewelyn J. A five-year demonstration proJect associated with improvement in physician health maintenance behavior. Fam. Med. 21, 173. 1989. 13. Wayland M. T., Cuhk D.. Goldsby C. et u/. Screening by family practice and internal medicine residents and the effect of an intervention program. J. Med. E&c/. 62, 519. 19x7. 14. Becker M. H. (ed). The Health Behef Model and personal health behavior. H//h Educe/. Monogr. 2, 236, I974

Physician

beliefs about

15. Bandura A. Social Learning Theory. Englewood Cliffs, Prentice-Hall, NJ, 1977. 16. Bandura A. Social Foundations of Thought and Action. A Social Cognitive Theory. Prentice-Hall, Englewood Cliffs, NJ, 1986. 17. Rosenstock I. M. Historical origins of the Health Belief Model. Hlth Educt. Monogr. 2, 328, 1974. 18. Janz N. K. and Becker M. H. The health belief model: A decade later. Hlth Educt. Q. 11, 1, 1984. 19. Mullen P. D., Gottlieb N. H., Biddle A. K. et al. Predictors of safety belt initiative by primary care physicians. A social learning theory perspective. Med. Care 26, 373, 1988. 1. M.. Strecher V. J. and Becker M. H. 20. Rosenstock Social learning theory and the health belief model. Hlth Educt. Q. 15, 175, 1988. 21. Norusis M. J. SPSS Base S~xrrm User’s Guide. SPSS Inc., Chicago, IL, 1990. 22. Schwartz J. S., Lewis C. E., Clancy C. er ctl. Internist’s practices in health promotion and disease prevention: A survey. Ann. internal Med. 114, 46. 1991. in primary 23. Gemson D. H. and Elinson J. Prevention care: Variability in physician practice patterns in New York City. Am. J. Prwenr. Med. 2, 226, 1986. 24. Battista R. N. Adult cancer prevention in primary care: Patterns of practice in Quebec. Am. J. publ. Hlth 13, 1036, 1983. 25. American Cancer Society. Survey of physicians’ attitudes and practices in early cancer detection (1984). CA-A Cancer J. Clinicians 35, 197, 1985. atti26. American Cancer Society. Survey of physicians’ tudes and practices in early cancer detection (1989). CA-A Cancer J. Clinicians 40, 71. 1990. D. H. and Elinson J. Cancer screening 27. Gemson and prevention. Knowledge, attitudes, and practices of New York City physicians. NY St. J. Med. 87, 643, 1987. R. J. and Solkowitz S. N. 28. McPhee S. J., Richard Performance of cancer screening in a university general internal medicine practice: Comparison with the 1980 American Cancer Society guidelines. J. Gen. inrernal Med. 1, 275, 1986. 29. Bergner M. A., Allison C. J.. Diehr P. et al. Early detection and control of cancer in clinical practice. Archs infernal Med. 150, 431, 1990.

cancer

screening

849

30. Sandler R. S., Holland K. L., Brooks E. et al. Physicians’ opinions and use of colorectal cancer screening procedures. N.C. Med. J. SO, 452, 1989. 31. Cummings K. M., Mettlin C., Lazar L. and Frisof K. B. Family physicians’ beliefs about cancer screening tests. In Progress in Cancer Control IV: Research in the Cancer Center, p. 361. Alan R. Liss, New York, 1983. 32. Albanes D., Weinberg G. B., Boss L. and Taylor P. R. A survey of physicians’ breast cancer early detection practices. Prevent. Med. 17, 643, 1988. 33. Slenker S. E. and Wright J. M. A survey of physician beliefs and practices concerning routine mammography. Ohio Med. 84, 416, 1988. 34. Weinberger M., Saunders A. F., Samson G. P. et al. Breast cancer screening in older women: Practices and barriers reported by primary care physicians. J. Am. Geriafr. Sot. 39, 22, 199 1. 35. Warner S. L., Worden J. K., Soloman L. J. and Wadland W. C. Physician interest in breast cancer screening education. A survey of Vermont family physicians J. Fam. Pratt. 29, 281, 1989. 36. Selinger H. A., Goldfarb N. I., Perkel R. L. and Carlson B. L. Physician compliance with mammography guidelines: A retrospective chart review. Fam. Med. 21, 56, 1989. 37. Cummings K. M., Jaen C. R. and Funch D. P. Family physicians’ beliefs about screening for colorectal cancer using the stool guaiac slide test. Publ. Hlth Rep, 99, 307, 1984. 38. Cummings K. M., Funch D. P., Mettlin C. and Jennings E. Family physicians’ beliefs about breast cancer screening by mammography. J. Fam. Pratt. 17, 1029, 1983. 39. Fox S., Klos D. S. and Tsou C. V. Underuse of screening mammography by family physicians. Radiology 166, 431, 1988. 40. Steingart R. M., Packer M.. Hann P. er al. Sex differences in the management of coronary artery disease. _ _ N. Engl. J. Med. 325, 221, 1991. 41. Hannan E. L.. Bernard H. R.. Kilburn H. C. Jr and O’Donnell J. F. Gender differences in mortality rates for coronary artery bypass surgery. Am. Heart J. 123, 866, 1992. 42. Clancy C. M. and Massion C. T. American women’s health care: a patchwork quilt with gaps. J. Am. med. Assoc. 268, 1918, 1992.