The Health Belief Model and Preventive Health Behavior Irwin M. Rosenstock, Ph.D. School of Public Health, University of Michigan

The Health Belief Model was originally formulated to explain (preventive) health behavior. As defined by Kasl and Cobb,’ health behavior is “any activity undertaken by a person who believes himself to be healthy for the purpose of preventing disease or detecting disease in a n asymptomatic stage.” This is in contrast with illness behavior, defined as “any activity undertaken by a person who feels ill, for the purpose of defining the state of his health and of discovering suitable remedy,” and sick-role behavior, “the activity undertaken by those who consider themselves ill for the purpose of getting well.” The present paper is confined to the first of these areas - health behavior. I t should first be noted that the three modes of behavior are not discontinuous. Hardly anyone can be found who, upon intensive questioning, would report himself free of all symptoms. Similarly, the edges between illness behavior and sick-role behavior are blurred. Nevertheless, the distinctions are valuable because they refer to modal mental states which help to account for behavior. STUDIES OF HOW PEOPLE USE HEALTH SERVICES

Consideration may first be given to the relationship between studies of how preventive health services are used and a n understanding of why they are used. Do studies of how people use services explain why people use health services? I n approaching a n answer to this question, a careful distinction should be drawn between studies of utilization whose findings are intended to have immediate application and studies of utilization which are intended to serve as means to still other research ends. In the first case, information is sought to serve as a basis for formulating and implementing public policy in the health area. Utilization data obtained for such purposes have proved invaluable in the health Utilization studies undertaken as means to achieve the broader aim of increased understanding of why services are used, however, have generally failed to accomplish their purpose. Little can be learned from these studies about why people use or fail to use certain services. Evidence in support of this conclusion has been drawn from studies of 354

Health Education Monographs VOL. 2, NO. 4

high a n d low users of free medical examinations,J detection tests for cervical cancer,6 polio i m m ~ n i z a t i o n ,dental ~ services,*>.'Jphysicians' services,J.Luhospital services,1i and from studies of the characteristics of those who do and those who do not delay in seeking diagnosis a n d treatment of cancer.1."

Dem ographic Varia tio m Analyzing the major findings of studies on the patterns of use of preventive a n d detection services permits c e r t a i n s u m m a r y generalizations about the association of personal characteristics with the use of services. I n general, such services a r e used most by younger or middle-aged people, by females, by those who a r e relatively better educated and have higher income (though perhaps not the very highest levels of education a n d income). Striking differences may nearly always be found in acceptance rates between whites a n d non-whites, with whites generally showing higher acceptance rates, although occasional exceptions occur. A review of the previously cited d a t a on utilization of diagnostic a n d treatment services suggests a pattern quite similar to t h a t obtained in connection with preventive a n d detection services. I n general, more females than males visit the physician a n d the dentist a n d incur hospitalization, even when hospitalization for pregnancy is excluded. Higher socioeconomic groupings (defined in terms of educational a n d income level) a r e also more likely to obtain medical, dental, a n d hospital services, although the associations between income a n d utilization a r e becoming less marked.'..$ With reference to race, whites show much higher utilization rates than non-whites in all three utilization categories (physician visits, dental visits, a n d hospitalization). T h e nature of the association between age a n d utilization of treatment services is generally different from t h a t found between age and seeking preventive a n d detection services, probably reflecting the effect of objective medical a n d dental need. With respect to characteristics of those who delay in seeking diagnosis a n d treatment of cancer, similar patterns emerge. I n general, persons who delay a r e older, of low educational status and, a t least in some studies, males." I t h a s recently been suggested t h a t there is a n increasing rate of health supervision visits; t h a t is, visits to practitioners i n the absence of symptoms. A question may be raised as to whether such increases, if indeed they a r e occurring, show the typical social class gradient, with those of higher income accounting for most of the increase. There a r e no definitive d a t a on the subject but inferences can be drawn from a combination of findings from several sources. Herman 1J notes, a s have many others, t h a t while higher rates of disease a n d mortality still Health Education Monographs Winter I974

355

persist among those with very low incomes, their frequency of visiting the physician is considerably lower than for the more affluent, healthier group. I t is commonly known that even when immunizations are free, higher income families show a much better rate of protection than do poorer families; moreover, while ambulatory services generally show a lower utilization rate by lower income households, poor people are overrepresented among hospital patients; their hospitalization rates are as high as, or higher than, those of upper income levels and their length of stay longer on the average. This probably reflects the failure to receive treatment a t earlier stages of disease and disability.

Attributing Cause to Demographic Correlates While some of these data are not as recent as one would like, there seems to be no reason for concluding that the poor, as yet, are showing any marked increase in their likelihood of seeking early health care compared to their prior behavior or to the behavior of the more affluent. This is not to say that the removal of economic and social barriers will not increase the use of health services; indeed, they may well do so. In a n experiment reported by Alpert, et aI,l4 it was shown that after exposure to comprehensive, personalized health care, low income families (i.e., median income of $4,100) became more satisfied with the services received, reported a n increased likelihood of using a family doctor or pediatrician for selected medical problems of children, and reported a greater likelihood of using the telephone as a first contact. Nevertheless, it is still questionable whether such attitudinal changes will result in patterns of use of health services that are like those of the more affluent. Several groups deny that there is any question; for them differences in utilization patterns are entirely explainable by income differences. A report by Colombo et al, is a good e ~ a m p l e . ~Ins 1967, a poverty group, supported by the Office of Economic Opportunity, was admitted into the Portland Region of the Kaiser Foundation. During a period of nearly a year of observation, their use of medical services (3.9 encounters per person) was reportedly remarkably close to that of a sample of the overall health plan membership (4.1 encounters per person), and, with minor exceptions, the nature of the visits were quite similar a s indicated by the most frequent diagnoses made by physicians. The authors conclude that a strong emphasis on preventive care exists both in the general health plan population and in the O.E.O. population. The basis for this important conclusion is worthy of detailed examination. A close examination of methods of selection of poor families into the plan and the mode of analysis casts some doubt on the conclusion that the poor do in fact exhibit a strong emphasis on preventive care. The 356

Health Education Monographs VOL. 2, NO. 4

target area specified “was composed of neighborhoods that had a majority of families who were not defined as poor. These areas contained approximately 4,000 poor families.” In short, the poor population selected represented a minority of the residents of the neighborhoods from which they were drawn. The authors also indicate that since many of the neighbors of the poor families selected were already members of the Kaiser Health plan, the program was already familiar to the indigent residents. We do not know whether the desire for and utilization of health services by poor, urban people living in predominantly non-poor neighborhoods are typical of the nation’s poverty population. No information is available on the prior utilization rates of the poor families. Furthermore, since only 1,200 of the 4,000 identified poor families could be served, a number of selection priorities were imposed of which the two most important were (1) large families with small children and ( 2 ) families with known acute health problems, but with no existing medical care source. One wonders whether these selection priorities did not almost guarantee greater utilization of services than had all 4,000 poor families been included. Selection priorities to include those in ereatest need are quite sensible in terms of health need, but the procedure casts doubt on the comparability of the O.E.O. group with a random sample of the total health plan membership. If the random sample of health plan membership included persons less in need, then their slight superiority over the O.E.O. population in total patient encounters masks a tendency for the general population to seek preventive services much more frequently than the poverty population. In the Portland-Kaiser study, children under six, from O.E.O. families, had 4.2 encounters with physicians while in the general membership the number of such encounters was 6.1. Colombo et a1 report a utilization rate of wellbaby and child care of 235.1 per thousand for the O.E.O. population and only 159.3 per thousand for the general health plan membership, but these figures are very misleading as there were nearly twice as many children under six in the O.E.O. population a s in the general health plan membership. Taking this discrepancy into account, one might have anticipated a utilization rate in the O.E.O. population for well child care of more than 300 per thousand. The authors also present data on utilization of immunizations which apparently shows substantially more utilization by the O.E.O. population than by the health plan membership. Once again these conclusions are derived from total patient encounters without allowing for the overweighting of children in the O.E.O. group. Since the four immunizations reported are primarily intended for those under six years of age, recomputation of the data using as the denominator the total number of children under age six in each of the two groups Health Education Monographs Winter I974

35I

revealed that the immunization experience of the two groups was virtually identical. However, since it is reasonable to believe that the O.E.O. population had been under-doctored prior to their enrollment in the Kaiser plan, if finances and organization were all important one might have expected a much higher rate of immunization among the poverty group than in the health plan membership whose children had been seen more or less regularly during the period preceding the study. In short, the O.E.O. group probably did not make preventive visits as frequently as the more affluent, when differences in the numbers of children are taken into account and when the apparently greater need in the O.E.O. population is considered. In summary, this report that purports to show that utilization rates of health services by poor people increases to the levels of the non-poor under proper organizational and prepayment patterns is unconvincing. In fact, reexamination and reinterpretation of the data and the sampling procedures lead to the conclusion that improved delivery and payment mechanisms - however desirable and necessary they may be for the poor - do not result in the extent of preventive health behavior exhibited by the more affluent. A recent article by Monteiro presents arguments similar to those of Colombo et al.lS The author points out that, through 1968, physician visits reported in national data no longer seemed directly dependent on income. The lowest (under $3,000) and highest (above $10,000) income groups made the most visits with intermediate incomes exhibiting fewer visits. Monteiro reports a study in Rhode Island, conducted in 1967, 1968, 1969, with a follow-up in 1971, which confirms the national data. The author herself notes that there may be wide differences as to what is defined as illness, a topic she did not study. Thus, the poor may tolerate more severe morbidity before restricting their activity and in fact go to the doctor with much more serious illness than the non-poor. But, since it has been shown that the poor have substantially higher levels of morbidity than the non-poor, if income were unrelated to utilization one might expect that they should visit the physician not only as often as the non-poor, but much more often. Equally crucial criticisms can be leveled a t Monteiro’s interpretations, for she merely used gross counts of doctor visits, a measure that has always been regarded a s fairly crude. If she had studied the purposes of the visits or the use of other professional providers, for example dentists, she would have found what has so often been reported in the national health survey and by other investigators, namely that the poor show substantially lower utilization than the non-poor of each of the following kinds of preventive and curative services. 1. Percent under age 17 receiving routine physical examinations. 2. Percent under 17 visiting the pediatrician. 358

Health Education Monographs VOL, 2, NO. 4

3. Percent of women visiting a n Ob/Gyn. 4. Percent visiting the dentist within the last year and, within that

category, percent obtaining dental restorations vs. extractions. 5. Percent of women voluntarily seeking Pap smears. 6. Percent of adults seeking regular chest X-rays. 7. Percent obtaining polio immunizations. 8. Percent brushing their teeth regularly. What Monteiro's study seems to show is that publicly financed care will increase utilization among poor people experiencing substantial medical need but the level of utilization for less disabling conditions, and the use of preventive services, remains highly correlated with income; the poor receive far less of such services than the non-poor. I t may also be emphasized that, within comparable income categories, utilization of services tends to be substantially higher for whites than non-whites.

Perception of Symptoms as an Intermediate Variable in Utilization Although most studies of utilization do not throw light on why people use health services, one area of research can be identified in which quite sophisticated efforts have been made to understand health and illness behavior as a function of personal characteristics; a n area described by Kasl and Cobb as "variables affecting the perception of symptoms." I Several other workers attempt to link personal and subcultural variables to the individual's likelihood of perceiving a n event as a symptom or to his mode of responding to a symptom. For instance, Koos found a social class gradient in terms of the likelihood of interpreting a particular sign as a symptom.17 Stoeckle, Zola,and Davidson and Zola studied the effects of ethnic values upon the specific decision to seek medical attention and on the differential interpretation of objectively similar symptoms.l"1t' Freidson illustrated the different processes through which members of different social groups move in obtaining diagnosis (lay and professional) and in seeking care." Suchman attempted a n interesting and promising approach which links demographic factors to social structure, both of these to medical orientation and in turn to health and medical care.21 Studies of the kinds performed by Koos, Stoeckle, Zoia, Freidson, and Suchman are far superior in their ability to explain than are the more traditional analyses of relationships between demographic factors and the utilization of services. This superiority lies in the proposed linking mechanisms between personal characteristics and behavior. These studies also demonstrate that health decision making is a process in which the individual moves through a series of stages or phases. Interactions with persons or events a t each of these stages influence the individual's decisions and subsequent behavior. Health Education Monographs Winter I974

359

Yet, even these more sophisticated approaches limit their focus to illness behavior; that is, to behavior undertaken in response to symptoms. The findings are, thus, of unknown relevance to the situation confronting the person who must decide whether to seek preventive or detection services before the appearance of events that he interprets as symptoms. Suchman explicitly notes the failure to his concepts of social structure and health orientation to account for preventive health actions." Stimulating the development of a preventive orientation in the public is a t the heart of most educational programs in public health. EVIDENCE FOR AND AGAINST THE MODEL

Although many recent investigations have identified explanatory variables which are similar to one or another variable contained in the model, seven major projects were originally undertaken whose design was largely or entirely determined by the behavioral model.2"z&"Of these, four were retrospective while three were prospective studies."&"' The retrospective research projects have in common the crucial characteristic that data about respondents' beliefs and behavior are gathered during the same interview, and the beliefs are assumeci to have existed in a point in time prior to the behavior. That assumption is a questionable one at best and will be considered after a review of the retrospective research. One other problem in the interpretation of the studies should be noted. With the exception of the Hochbaum study 54 and the National Study of Health Attitudes and the research has been based on quite small samples. Sometimes sample size has been limited by financial or other insuperable obstacles. However, in some cases difficulties in categorizing responses or in obtaining responses to every necessary item have reduced samples to dangerously low proportions. Finally, the concept of motivation, now considered central to the Health Belief Model, was not specifically studied in these seven early investigations. The best documented of the early retrospective studies were performed by Hochbaum and Kegeles,zs and these will be reviewed in some detail.

The Retrospective Studies Hochbaum clearly originated research on the Health Belief Model. Beginning in 1952, he studied more than 1200 adults in three cities in a n attempt to identify factors underlying the decision to obtain a chest X-ray for the detection of tuberculosis. He tapped beliefs in susceptibility to tuberculosis and beliefs in the benefits of early detection. Perceived susceptibility to tuberculosis contained two elements. It included, first, the respondent's beliefs about whether tuberculosis was a real possibility in his case, and second, the extent to which he 360

Health Education Monographs VOL. 2, NO. 4

accepted the fact that one may have tuberculosis in the absence of all symptoms. Consider first the findings for the group of persons that exhibited both beliefs, that is, belief in their own susceptibility to tuberculosis and the belief that overall benefits would accrue from early detection. In that group 82 percent had had a t least one voluntary chest X-ray during a specified period preceding the interview. On the other hand, of the group exhibiting neither of these beliefs, only 21 percent had obtained a voluntary X-ray during the criterion period. Thus, four out of five people who exhibited both beliefs took the predicted action, while four of five people who accepted neither of the beliefs had not taken the action. Thus, Hochbaum appears to have demonstrated with considerable precision that a particular action is a function of the two interacting variables - perceived susceptibility and perceived benefits. The belief in one’s susceptibility to tuberculosis appeared to be the more powerful variable studied. For the individuals who exhibited this belief without accepting the benefits of early detection, 64 percent had obtained prior voluntary X-rays. Of the individuals accepting the benefits of early detection without accepting their susceptibility to the disease, only 29 percent had prior voluntary X-rays. Hochbaum failed to show clearly that perceived severity plays a role in the decision-making process. This may be due to the fact that his study was not designed to identify perceived severity with any high degree of accuracy and his measures of severity proved not to be sensitive. However, he did identify 16 individuals who seemed intensely afraid of TB. None of these had had a single voluntary X-ray during the preceding eight-year period. In addition, those respondents who appeared indifferent to the disease were among those who tended not to feel susceptible, and consequently not to take X-rays. Finally, those who exhibited some “mid-range’’ level of fear participated to a slightly greater extent than those a t the very high or low end of the scale. Kegeles 85 dealt with the conditions under which members of a prepaid dental care plan will come in for preventive dental check-ups or for prophylaxis in the absence of symptoms. He attempted to measure the respondent’s perceived susceptibility to a variety of dental diseases, the perceived severity of these conditions, his beliefs about the benefits of preventive action and his perceptions of barriers to those actions. While findings usually support the importance of the model variables, their general applicability is greatly limited by a n unusually large loss in the sample. The study was initiated with a sample of 430, but those without teeth, those for whom information was not available to determine whether past dental visits had been made for preventive purposes or for treatment of symptoms and those whose positions Health Education Monographs Winter I974

361

could not be coded on all three belief variables were excluded. The crucial analysis could thus be made only on 77 individuals. Within the major limitations implied by the small sample size and by their likely nonrepresentativeness, Kegeles showed that with sucessive increases in the number of beliefs exhibited by respondents from none to all three, their frequency of making preventive dental visits also increased. The actual findings show that: ( 1 ) of only three persons who were low on all three variables none made such preventive visits; ( 2 ) of 18 who were high on any one variable but low on the other two, 61 percent made such visits; ( 3 ) of 38 persons high on two beliefs and low on one, 66 percent made preventive visits; and, finally, ( 4 ) of 18 persons who were high on all three variables, 13 or 78 percent made preventive dental visits. Similar patterns of findings based on much larger samples were obtained in a n analysis of relationships between behavior and each of a series of single variables, that is, susceptibility, severity, benefits and barriers. The findings ot tne two remaining retrospective studies will not be reviewed in detail but are in most respects quite similar to the two that have been r e v i e ~ e d . ~ eIn; ~each ~ case evidence that supports the model has been obtained, although the sample sizes were not large. In summary, while no one of the original studies provided convincing confirmation of the model variables, each produced internally consistent findings which were in the predicted direction. Taken together they thus provided reasonably strong support for the model. As indicated, any interpretations made of the findings of the retrospective studies are based on a n assumption. The hypothesis that behavior is determined by a particular constellation of beliefs can only be tested adequately where the beliefs are known to have existed prior to the behavior that they are supposed to determine. However, the retrospective projects were undertaken in situations which necessitated identifying the beliefs and prior behavior a t the same point in time. This approach has always been known to be quite dangerous. Work on cognitive dissonance:2 supported these suspicions and suggested that the decision to accept or reject a health service may in and of itself modify the individual’s perceptions in areas relevant to that health action. Obviously, a two-phase study was needed in which beliefs would be identified at one point in time, and behavior measured later.

The Prospectioe Studies Such a study was attempted in the fall of 1957, around the topic of the impact of Asian influenza on American community As one of a series of related studies, Leventhal, Hochbaum, and Rosenstock investigated the impact of the threat of influenza on families through the use of a design that was intended to permit a test of the model in a 362

Health Education Monographs VOL. 2, NO. 4

prospective manner. In this phase of the study, 200 randomly selected respondents in each of two medium-size cities in the United States were interviewed twice. The first interview was intended to be made before most people had the opportunity to seek vaccination or to take any other preventive action and before much influenza-like illness had occurred in the communities. The second interview w a s to be made after all available evidence indicated that the epidemic had subsided. In fact, only partial success was achieved in satisfying these conditions because community vaccination programs as well as the spread of the epidemic moved much faster than had been anticipated. For these reasons the sample on which the test could be made was reduced to 86. This sample of 86 respondents had, at the time of initial interview, neither taken preventive action relative to influenza nor had they experienced influenza-like illness in themselves or in other members of their families. Twelve of the 86 scored relatively high on a combination of beliefs in their own susceptibility to influenza and the severity of the disease.36 Five of these 12 subsequently made preventive preparations relative to influenza. On the other hand, at the time of the first interview, the remaining 74 persons rejected either their own susceptibility to the disease or its severity or both. Of these, only eight, or 11 percent, subsequently made preparations relative to Although the samples on whom comparable data could be obtained were very small and possibly not representative, the differences were statistically significant beyond the one percent level of significance. Analyses of the available data thus suggest that prior beliefs in susceptibility and severity are instrumental in determining subsequent action. A second prospective study- was a follow-up by Kegeles 28 on the study reported earlier.’6 Three years after the initial collection of data on a sample of more than 400 in 1958, a mail questionnaire was sent to each person in the sample as well as to a comparable control group to obtain information about the three most recent dental visits. The objective of the follow-up was to determine whether the beliefs identified during the original study were associated with behavior during the subsequent three-year period. Kegeles found that perceptions of seriousness, whether considered independently or together with other variables, were not at all associated with subsequent behavior. Perceptions of benefits taken alone were not related to subsequent behavior. However, the perception of susceptibility did show a correlation with making subsequent preventive dental visits. Of those who had earlier seen themselves as susceptible, 58 percent made subsequent preventive dental visits while 42 percent who had not accepted their susceptibility made such visits. When beliefs about susceptibility and benefits were combined, a more accurate prediction was possible of who would or Health Education Monographs Winter I974

363

would not make preventive dental visits. Considering only those who scored high on susceptibility, and cross-tabulating against beliefs in benefits, 67 percent of those high on both beliefs made subsequent preventive visits while only 38 percent low in benefits made such visits. Thus, the combination of susceptibility and benefits was demonstrated to be important in predicting behavior. The results of the six studies cited above lend support to the importance of several of the variables in the model as explanatory or predictive variables. However, a seventh major investigation conflicted in most respects with the findings of earlier studies.a",al The study included analyses of beliefs and behavior of a probability sample of nearly 1500 American adults studied in 1963, and the subsequent behavior of a 50 percent subsample studied 15 months later. Perceived susceptibility, severity, and benefits, whether taken singly or in combination, did not account for a major portion of the variance in subsequent preventive and diagnostic behavior, in regard to dental visits, X-rays for TB, or check-ups for cancer, although predictions based on the belief in benefits taken alone frequently approached significance. The study findings did not disclose any explanation for the failure to obtain findings similar to those of the earlier described studies, but the more recent national study was conducted in a setting which distinguished it from all the other reported studies in one respect that may have been crucial.

Health Education as an Independent Variable I n the earlier described studies, the settings were such that the population in each case had been offered the opportunity to take action through directed messages and circumstances that could have served as cues to stimulate action. In Hochbaum's study,%4mass media had been used in three study cities to urge the population to obtain chest X-rays. I n the Kegeles studies,"*.Joevery member of the population was offered free or inexpensive dental treatment and was urged to use it. I n the Leventhal, Hochbaum, and Rosenstock study,zm the population had been alerted by newspapers and by public health officials to the desirability of obtaining influenza immunizations. In the Heinzelmann study,Ze the patients had been urged to use penicillin prophylactically. The Flach study 27 offered the population a free test for cervical cancer. I n short, in all the prior studies the population had been exposed to information which both indicated the availability of a health procedure and which, in most cases, urged them to avail themselves of that procedure. In contrast, such conditions did not occur for the national sample in the most recent national study. With respect to the several health problems covered in the study, neither the sample nor the United 364

Health Education Monographs VOL. 2, NO. 4

States adult population which it represents, had been uniformly exposed to intensive campaigns to inform them about available services and to persuade them to use such services. Nor can the assumption reasonably be made that preventive and diagnostic services were equally available to all. The absence of clear-cut cues to stimulate action as well as unequal opportunity to act, may, in large measure account for the failure to replicate the earlier results. However, those possibilities must be treated as hypotheses which will need to be tested in new research. Two more recent studies lend further support to the general explanatory model of health behavior although its specific variables have been treated more innovatively than in earlier research. Haefner and Kirscht attempted experimentally to increase people’s readiness to follow preventive health practices by presenting them with messages about selected health problems that were intended both to increase their perceived susceptibility and/or severity regarding the health problems and their beliefs in the efficacy of professionally recommended behavior. Significantly more persons exposed to such messages visited a physician for a check-up in the eight months following the experimental manipulation than in a control group not exposed to the messages. This significant difference held only for visits made in the absence of symptoms, i.e., preventive health behavior. For individuals reporting actual symptoms during the interval, the rate of physician visits was the same in the experimental and control group. While income as such was not treated, the sample represented a universe of non-academic University employees, a group above the poverty level but, in general, far from affluent. This study, incidentally, provided evidence that it is possible to modify the perceived threat of disease; that is, the combination of perceived susceptibility to and severity of diseases as well as the perceived efficacy of professional intervention, and that such modification leads to predictable changes in health behavior. Still more recently, Becker, Kaback, Rosenstock, and Ruth applied the Health Belief Model to the area of genetic ~creening.3~ Beginning in 1971, a n identified Jewish population in the Baltimore-Washington area was invited to participate in screening for the Tay-Sachs trait which has a frequency of about one in 30 Jews of Ashkenazi ancestry (compared with one in 300 among non-Jews). About one in 900 Jewish couples could thus be expected to be a t risk of having a child with Tay-Sachs Disease, a n incurable condition which is invariably fatal in early childhood. Of course the probability of having such a diseased child for at-risk couples is one in four. The disease can, however, be diagnosed in utero through amniocentesis at a stage when abortion is feasible. Thus, the situation presents all the conditions for observing the role of the components of the Health Belief Model in predicting JJ

Health Education Monographs Winter I974

365

preventive health behavior. Furthermore, since the relatively rare disease and the diagnostic test were largely unknown to the lay puolic, it is a reasonable inference that the majority had had little contact with the disease, with screening or with amniocentesis and that they had few relevant beliefs about it in advance of the program. The education of the target community began six to eight weeks before initiation of mass screening. Multiple educational approaches were used to saturate the communities with accurate and clear information. These included the press, T V , radio, letters from Rabbis, fliers from community organizations, medical presentations to the community, telephone calls from trained volunteers, brochures from physicians and other special mailings. Since lists of the target population were available, it could be assured that all members of the target group - couples of childbearing age - were exposed to a t least some of these educational activities. As applied to the Tay-Sachs situation, the explanatory variables were defined as follows: Health motive was for the first time explicitly introduced into the model to explain health behavior. In the present case, motive included two components: (1) a positive response indicating a desire to have (additional) children and ( 2 ) a set of generalized items about typical health behavior, such as the frequency with which the person ’thinks about his own health and whether he generally goes to the physician right away if he feels sick. Perceived susceptibility included the person’s belief that he could carry the Tay-Sachs gene and transmit it to his progeny. Severity was interpreted as the individual’s views of the potential impact of learning that he was a carrier, especially as regards future family planning. The definition of perceived benefits was in terms of a personal evaluation of how much good it would do the potential carrier to be screened for the trait. Did he really need to know or want to know his carrier status? Costs or barriers to action were not directly measured in this study. They might include, however, usual monetary or convenience factors as well as threats we currently know very little about, for example, the impact on an individual of learning that he is a carrier of some recessive trait. How does it affect his self-image, his perception of his health and of his well being? Does it affect his marriage? How does it influence future family planning? I n all, nearly 7,000 adults were screened during the first year of the study, all drawn from lists of synagogue membership, and names in predominantly Jewish neighborhoods. All adults who appeared for screening were asked to complete a brief questionnaire just before going through the screening process; 500 of these were selected a t random as the participant sample. In addition, 500 questionnaires were mailed to a random sample of nonparticipants who had been invited in for screening; here the response rate was 82 percent. I t should be noted that non-respondents as well a s respondents had 366

Health Education Monographs VOL. 2, NO. 4

received intensive informational material on Tay-Sachs disease and screening. T h e analysis showed t h a t the participants, compared to nonparticipants, were significantly younger, had had fewer children, were less likely to have completed their families, and were slightly better educated. Turning to the health belief variables, the participants differed sharply in perceived susceptibility, the first component of health motivation - 82 percent of those who expressed the desire to have future children participated in the screening program while less than 19 percent who did not desire future children participated; there was no significant difference in participation according to perceived severity, the second, less direct motivational measure used. Mean score on perceived susceptibility to being a carrier was highly correlated with participating in the screening program while perceived severity was negatively associated with participation. When the three foregoing variables were combined it became apparent that while each of the three is associated with participation, perceived susceptibility interacted with the desire to have future children while perceived severity played a n independent, additive role; for people who desire additional children, moderate perceived susceptibility and low perceived severity best explains participation in the program. Among those who are not motivated to have additional children, high susceptibility and low severity best explains participation. Irrespective of motivation, the combination of high perceived susceptibility and low perceived severity best accounts for participation. Benefits-to-barriers ratio. Among those individuals who indicated that they planned to have more children, the non-participants more than the participants indicated that the discovery that either or both husband and wife were carriers would change their future child planning behavior; frequently they reported they would have no additional children. One possible interpretation of this finding related to beliefs exhibited by participants and non-participants about the transmission and detection of Tay-Sachs disease and about reproductive alternatives. A question on the impact of learning that one member of a married couple was a carrier obtained quite different answers from participants and non-participants. Participants were much less likely than non-participants to alter their plans. More of the participants had apparently learned that carrier status in only one member of the couple poses no dangers. However, in response to the question on the impact if both parents were found to be carriers, while participants were again less likely to change their reproductive plans than nonparticipants, they did indicate they would reduce the number of children they would have or that they would use “other” 5

Health Education Monographs Winter 19 74

367

approaches. In nearly every case where the “other” category was used, participants went on to explain that they would elect to use amniocentesis (fetal diagnostic test) in order to continue to have children. Very few of the non-participants displayed knowledge of the availability of amniocentesis; rather, they tended to indicate that, in the event either member of both members of a couple were found to be carriers, they would not have further children. Since more participants than non-participants learned about the “fetal diagnostic test,” it may be inferred that screening conferred considerable benefits for participants: (1) they could rule out the possibility that both parents carried the recessive gene or ( 2 ) if both proved to be carriers, amniocentesis could rule out the possibility that the fetus had the disease, or (3) if the child were diseased, they could elect to abort it. On the other hand, while nearly all the study respondents held attitudes favoring abortion in the event that a fetus had TSD, the non-participants could not have seen as much benefit in screening, since they did not give evidence of having learned about amniocentesis. Perceived barriers. Barriers to screening, though not studied explicitly, were minimized in the present case by offering the test a t low cost to a relatively affluent group and a t convenient times and locations. Such financial and situational factors could, however, be important for other target groups. I n summary, more of those who agreed to participate, compared to non-participants, (1) believed they could be Tay-Sachs carriers (high perceived susceptibility), (2) stated that learning that they were carriers would not affect their family planning (low perceived severity), and (3) abortion was appropriate if the fetus should prove to be diseased (high perceived benefits). Fewer of the non-participants, on the other hand, believed that they were susceptible. Although they favored abortion to the same extent as the participants, more of the non-participants indicated that learning they were carriers would be quite disruptive of their lives and that it would seriously affect their future family planning. Limitations in the length of the questionnaire precluded studying why the non-participants felt that learning of their carrier status would be considerably more disruptive than was true for participants. But, given t h a t orientation, their behavior is understandable. One final consideration should be emphasized. I t is believed that in this case perceived severity associated with the Tay-Sachs trait reached such high levels as to become dysfunctional. ( I t will be recalled that a similar phenomenon had been observed in Hochbaum’s original study.) Although the nature of the data in previous studies have precluded the use of standard parametric analytic techniques, it has always been believed that what is needed for behavior is “an 368

Health Education Monographs VOL. 2, NO. 4

optimal” balance of perception of health motive, vulnerability, severity, and the psychological benefitlcost ratio. Where the balance among these is either quite ‘‘low’’ or quite “high,” professionally recommended behavior is not to be expected. The truth of this assertion can only come out of the studies which use measures sensitive to variations in the degree to which each variable is present.

Additional Evidence As indicated earlier, various studies of preventive health behavior have dealt with one or another component of the Health Belief Model as a more or less incidental part of their design. In a 1959 analysis of public response to polio vaccinations, drawing on a number of independent investigations, Rosenstock, Derryberry, and Carriger 7 concluded that the variables: perceived susceptibility, perceived severity, perceived benefits (safety and effectiveness of the vaccine) accounted for major portions of the variation in seeking polio vaccination, though the authors of the several studies had not been guided in their designs by the Health Belief Model. The same review confirmed the importance of social pressures in the decision to seek vaccination a s Hochbaum had earlier shown was the case for the decision to seek chest X-rays. Heinzelmann and Bagley 35 reported on reasons for participation in physical activity programs and perceived outcomes. The two most important reasons given for participation were the desire to feel better and to lessen the chance of a heart attack. And one of the major perceptual outcomes of participation w a s a feeling of decreased vulnerability to specific health threats, including heart attacks. Fink et a1 36 provide data that suggested that the perception of personal vulnerability to cancer and a concern with its severity distinguish participants from non-participants in a breast cancer screening program. Battistella 37 purports to show that a measure of readiness to initiating physician visits and to obtaining checkups, presumably adapted from Hochbaum’s concepts a s well as others, yielded only slight relationships. Since Battistella’s work is criticized in some detail in Kirscht’s chapter, only a few brief summary statements are needed here. Despite the fact that an inappropriate statistical test was chosen, the relationships he showed in the Hochbaum data were substantial; variables preferred by the author turned out to be less predictive. In a second paper 38 Battistella ruled out a whole host of typical sociological variables (age, economic status, etc.) a s explaining delay in seeking care. Only two variables were promising - both health beliefs. The first was the perceived efficacy of care and the second was worry about health. Health Education Monographs Winter 1974

369

Suchman agrees that the “motivation to change one’s health practices depends, to a large extent, upon the individual’s feelings of personal vulnerability and the seriousness with which he views the health hazard.” Ogionwo attempted to influence response to a cholera immunization and educational program in Nigeria and showed health concern (motivation), perceived vulnerability to cholera, and knowledge of preventive measures highly associated with response, as were also several attitudes to prevention and correct knowledge of cholera . Gochman 41 has shown that where health-motivation is high children’s perceptions of vulnerability to dental disease and perceived benefits of dental treatment predict their intentions to make a future dental visit better than where health motivation is low. Health motivation is thus seen to be a n important organizing factor in health beliefs and intentions. I n a n experimental attempt to change beliefs and behavior of women in a n urban ghetto concerning screening for cervical cancer, Kegeles 42 demonstrated that women with relatively high beliefs in their vulnerability to cervical cancer and in the effectiveness of cytology made more visits than their counterparts. Antonovsky and Kats 4J have developed a n integrated model of the determinants of health behavior which is acknowledged to be in many respects similar to the Health Belief Model and which explains much of the variation in behavior of a sample of more than 500 employees of the Hadassah Medical Organization in Jerusalem. The authors raise, however, three points of difference between their model and the Health Belief Model, two of which are of considerable significance. These should be considered in some detail. The first criticism is of our concept of “cues.” For Antonovsky and Kats “cue” is a superfluous concept - the individual otherwise prepared to take action will “create his own cues.”4J There may be no more than a semantic difference here; nevertheless, it seems useful to retain the concept of cues which serve as triggers to initiate a train of events. It has already been suggested that Zola’s “critical incidents” may be identical with cues, and the concept seems to fit well with general psychological theory. Of greater moment is the criticism that the early Health Belief Model failed to include any frank motivational concept. This criticism is quite well taken. For Antonovsky and Kats, motivation is goal oriented behavior and the relevant goal is that of maintaining health. Other relevant goals might include achieving approval by significant others or achievement of self-approval. As indicated in several chapters, we have independently come to recognize the need to include a motivational variable, and recent J@

370

Health Education Monographs VOL. 2, NO. 4

studies using the Health Belief Model have incorporated such a component. Unfortunately, we cannot go as far as the authors do in regarding the principal roles of the perception of susceptibility and severity as that of increasing the salience of the motive. We believe that the cognitive factors play a somewhat independent role in influencing behavior, though motivation just as clearly is important (see papers by Recker and Kirscht). One final important point is the introduction by Antonovsky and Kats of the concept of a threshold level rather than a linear, monotonic relationship between a given variable and preventive health behavior. There is some reason to believe they are correct. Difficulties in measurement to date, to be described subsequently, have precluded testing the hypothesis that the motive and the perceptions of susceptibility, severity, benefits, and costs may each have cut-off points for a given individual below which the variable will have no effect and above which it will have effect, with additional increments making no difference. While this is a most useful concept, data already presented, especially as regards perceived severity, suggest that the threshold concept may need to be applied cautiously, since a t very high levels, anxiety and maladaptive behavior may result. Even with this caution the notion of threshold levels, rather than that of monotonic relationships, seems well worth pursuing. CRITIQUE OF THE MODEL AND NEEDED ADDITIONAL RESEARCH

The Place of the Model in the Hea lth Decision-Mak ing Process Health decision-making, including health behavior, illness behavior, and sick-role behavior, is a process in which the individual moves through a series of stages or phases in each of which he interacts with individuals and events. The nature of the interactions a t any one of these stages may increase or decrease the probability that a particular subsequent response will be made. Freidson ZIJ and Zola 19 have illustrated some of these stages. The individual’s relevant health beliefs as described in this paper are presumed to serve a s a setting for his subsequent responses at other stages in the decision process. For example, individuals who accept their susceptibility to a particular condition and are aware of actions that might be beneficial in reducing their susceptibility may well be the same persons who exhibit what Freidson terms “cosmopolitan” rather than “parochial” orientations toward health services. They may be more prone to learn about and seek out professional diagnosis rather than using the ”lay referral” system. In such a case the initial set of beliefs would itself determine subsequent choices in the decision-making process. Health Education Monographs Winter 19 74

371

W h a t have here been termed “cues” a r e probably identical with Zola’s “critical incidents.” 18 One cannot b u t agree heartily with his recommendation t h a t the role of such triggers to action be much more thoroughly investigated t h a n h a s previously been done. T h i s is urged despite the forbidding difficulties in identifying cues t h a t have already been described.

Operational Definitions of the Variables

No two studies of the model’s variables have used identical questions for determining the presence o r absence of each belief. T h i s raises the possibility t h a t t h e concepts being measured m a y also vary from study to study. It m a y be helpful to quote the questions actually asked (where available) in the first five studies guided by the H e a l t h Belief Model.

PERCEIVED SUSCEPTIBILITY

Hochbaum (1958) (1) If you were some day to get very seriously sick what do you think it would be? (a) What else could it be? (2) Did you ever think you might some day get TB? ( a ) Why do you feel that way? (3) Please look a t this card. (SHOW CARD A ) Which choice comes closest to your feeling about how likely it is that you may get TB some day? ( a ) What is your reason for picking that one? ( 4 ) (USING SAME CARD) How likely do you think it is that you may get TB in the near future? ( a ) Why do you pick that one? Kegeles (1963) (1)How likely do you think it will be that (worst dental problem previously experienced by respondent) will happen to you again? ( a ) Why do you think that? (2) How likely do you think it will be that (worst dental problem he could imagine) will ever happen to you?

Fhch (1960) The specific questions used are not available but the report notes that respondents were asked: (1) Whether single women, married women with children, or married women without children “are most likely to get uterine cancer.” Women choosing their own group were coded as “susceptible.” (2) Each woman was also asked whether she had ever thought it possible that she could have cancer.

Heinzelmann (1962) Susceptibility was an index score derived from the following two questions: (1) Taking all possible factors into consideration, what do you think your 312

Health Education Monographs VOL. 2, NO. 4

own chances are of getting rheumatic fever again? ( a ) Very Likely, ( b ) Likely, (c) Unlikely, ( d ) Very Unlikely? ( 2 ) Taking all possible factors into consideration, what do you think your chances of getting it again are in comparison to otherpeople who have had a n attack of rheumatic fever? ( a ) Much More, ( b ) More, (c) Same, ( d ) Less, (e) Much Less?

Leventhal ( 1 960)

a9

Questions are not available but the report indicates t h a t each respondent was asked whether he thought he might contract influenza. PERCEIVED SEVERITY

Hoch baum No measures Kegeles ( 1 ) If any of the dental problems (previously experienced) happened, how serious would i t be? ( 2 ) If any of the dental problems (imagined but not experienced) happened, how serious would i t be? I n addition, a respondent was categorized as believing in severity if he indicated anywhere in the interview t h a t such problems could detract from one’s aesthetic appearance.

Flach No measures. It was presumed that all or virtually all women believed cancer to be serious. Heinzelmann Severity was a n index score derived from the following four items: (1) How would you estimate the seriousness of your attack of rheumatic fever? ( a ) Serious, (b) Fairly serious, (c) Fairly mild, ( d ) Very mild. ( 2 ) Please check all of the following situations which apply to your attack: (a) Missed more than two months of schooling. (b) After the first few days, didn’t feel bad a t all. (c) Didn’t have any long-lasting after effects. ( d ) Was not permitted to take part in school gym or athletics. (e) Was not in bed more than a week. (f) Compared to others I know, my attack was mild. (g) Had a heart condition. (3) Rank the following illnesses in order of seriousness from most serious to least serious. Use “1” for most serious. (a) Measles, (b) Pneumonia, (c) Rheumatic fever, ( d ) Polio, (e) Tuberculosis, (f) Chicken pox. ( 4 ) Have you had any other diseases or illnesses, which you think were more serious than your attack of rheumatic fever? (a) No, (b) Yes.

Leventhal (1) “Do you think it would be worse or not as bad if you had the Asian flu compared to how i t usually is when you have the flu (grippe, bad cold)?” (2) “How would the Asian flu be different from the usual flu or grippe?” Health Education Monographs Winter I974

313

PERCEIVED BENEFITS

Hochbaum (1)If you should happen to get T B some day, how do you think you would be most likely to find out you had it? ( a )Why do you think you would find out that way (those ways)? (b) What other ways are there for discovering that a person has T B ? (2) You’ve just mentioned X-rays as one of the ways of finding out that a person has TB: In your opinion could an X-ray discover T B before or after the person himself could notice that something is wrong? (a) Why do you say that? (3)Suppose X-rays show that a person has TB, would it make any difference whether he starts treatment immediately or waits about six months to a year? ( 4 ) If “YES”: What difference would it make? (5) Some people think it is a good idea to have chest X-rays and some may not think so. Why do some think it is a good idea? (Else?) ( a ) How do you feel about these reasons? (6)Some may not think it is a good idea to have chest X-rays. Why do they feel t h a t way? (Else?) (a) How do you feel about these reasons? Kegeles The following questions were asked separately for ( a ) the worst problem experienced and (b) the worst problem anticipated: (1) Do you know of anything a person could do that would make it less likely t h a t he would get (worst dental problem)? What could he do? (2) If “NO” - “Do you mean that there is absolutely nothing a person can do to make it less likely t h a t (worst dental problem) would happen? Flach Questions are not available but the report indicates that respondents were asked about (a) their beliefs about whether early detection makes a great deal of difference so far as chances of cure are concerned, and (b) their degree of optimism concerning effects of cancer surgery upon marital relations and on physical energy.

Heinze lmann Benefits is an index score derived from following four items of knowledge and beliefs: (1) Please check below what you believe is the most important cause of rheumatic fever. (a) Overweight condition, (b) Heart weakness, (c) Bacteria, viruses, or other infections, ( d ) Typhoid fever, (e) Runs in the family, (f) Sore throats, (g) R u n down physical condition, (h) Improper diet. (2) A person who has a lot of colds and sore throats is more likely to get rheumatic fever again than a person who never has a cold or sore throat. ( a ) Agree very much (b) Agree, (c) Disagree, ( d ) Disagree very much. (3) Rheumatic Fever: (a) Can lead to rheumatoid arthritis, (b) Has no after effects, (c) Can lead to heart disease, ( d ) Usually weakens a person for a t least 10 years after, (e) None of these. ( 4 ) Which of the following do you think would be most important in keeping you from getting rheumatic fever again? ( a ) Keep from getting tired, (b) Take 374

Health Education Monographs VOL. 2, NO. 4

care of sore throats, (c) Eat proper foods, (d) Get regular checkups from doctor, (e) Take medicine, (f) Stay in good physical condition.

a

Leventhal

Questions are not available but the report indicates that respondents were asked ( a ) whether vaccination provides protection against Asian influenza; and, (b) what actions families could take to prepare for an epidemic of Asian flu. I n a n effort to bring some stability into the area of measures of health belief variables, the National Study of Health Beliefs described earlier was undertaken.Ju7Jl I n t h a t study alternative methods were used to identify beliefs about the severity of a n d susceptibility to four diseases: dental decay, gum trouble, tuberculosis, a n d cancer. Four different question formats were developed, differing simultaneously on two dimensions: ( 1) "self-reference" versus "reference to men-women your age" a n d ; ( 2 ) fixed-alternative versus more open items. A two-bytwo design was used with approximately one-quarter of the total sample randomly assigned to each of the four question formats. T h e findings demonstrated t h a t the question types obtained different distributions of responses.Ju However, since in t h a t study no clear relationship was demonstrated between possession of the beliefs, however measured, and health behavior, no decision can be made on which method of questioning is most valid. Clearly, more standardized ways of asking questions will have to be developed. Quantification T h e model implies t h a t certain levels of readiness a r e optimal in stimulating behavior but neither theory nor research have disclosed what the levels are. I n most of the studies limitations in sample size have necessitated dichotomizing scores on the variables into categories of "high" a n d "low." Until d a t a can be collected on a t least a n ordinal scale the problem of determining optimal quantities will not be solved and the dispute between threshold effects a n d linear relationships as predictors of behavior will not be resolved. Stability and Reliability of the Beliefs Little is known about the stability of the beliefs, although they may well vary from time-to-time as a function of situational changes. Learning t h a t a friend or a president h a s suffered a serious illness may raise personal levels of motivation a n d readiness to act. Research is needed to determine how stable the beliefs are. Similarly, little information is available on the reliability of the measures of beliefs. More work is also needed in this area. The appropriate approach to testing reliability depends on the stability of the beliefs. If the beliefs do change from time to time, test-retest Health Education Monographs Winter I 9 74

315

measures of reliability would not be as appropriate as split-half measures of reliability.

Perceived Seriousness Hochbaum,zd Kegeles,’u and Kirscht et al,Y” failed to demonstrate the importance of perceived severity in determining behavior. Flach z 7 did not measure severity since she assumed that cancer was universally seen as severe in the group of women studied. On the other hand. the studies by Heinzelmann,z6 the first Kegeles study,’Z and Leventhal et al,” did support the importance of perceived severity. Subsequently, Haefner and Kirscht 33 and Becker et a1 34 again showed the importance of perceived severity although, in the latter case, the relationship was complex. Greater doubt must be maintained about the importance of perceived severity as a n explanatory factor in health behavior than about the other variables. I t should be noted that while doubt remains concerning the role of perceived severity in stimulating preventive health behavior, its role, while complex, is clearer in studies of illness and sick-role behavior (see the chapters by Kirscht and Becker in this issue). Severity would appear to be related to behavior in a curvilinear manner. Where it is very low or very high, maladaptive behavior would seem to result.12,44 Where it is “moderate,” more adaptive behavior results. Clearly, more research on this variable is needed.

Genesis of the Beliefs Not enough is known about the genesis of the beliefs, nor of the conditions under which they are acquired. Moreover, no research has been done on how a n individual’s position on the three health beliefs is related to other comparable beliefs he holds. The potential value of the model would be greatly enhanced if the origins and development of the health beliefs were specified and if the beliefs were placed within a broader theoretical framework that would account for fesponses to a wide variety of stimuli. Gochman 4 1 9 4 6 9 4 7 has initiated efforts to study the genesis of the determinants of the health beliefs and their correlates. Haefner et al,J1 have made a beginning toward identifying the correlations among preventive orientations toward three health conditions. More work in both these areas is needed.

The Need for Additional Experimental Studies Convincing demonstrations of cause and effect can rarely, if ever, be provided through cross-sectional surveys of the kind typically employed to study the model. This is true because of the survey’s susceptibility to errors in judging which of two associated factors 316

Health Education Monographs VOL. 2, NO.4

preceded the other in time and because the possibility is great that apparent relationships may be spurious. For these reasons, experimental studies must be undertaken to determine the causal role of the relevant health beliefs. For example, a n effort could be made to modify the health beliefs of a randomly assigned experimental group while holding constant the beliefs of a comparable control group. Both groups would then be offered a particular health service and observations taken of the relative responsiveness of the groups to the health appeal. A variety of specific experiments could be devised to assess the contribution of the health beliefs to behavior. As indicated earlier, Haefner and Kirscht 9J did conduct such a n experiment and showed positive findings. But such research must be repeated.

Susceptibility of the Beliefs to Modification Even if the model did predict behavior, its ultimate usefulness would depend upon the extent to which the health beliefs can be modified in a planned way. Three efforts to attempt such change have been reported. Guskin,'* through the use of a film succeeded in modifying the reported beliefs of fifth and sixth grade students relative to their perceived susceptibility to and severity of tuberculosis, although no changes in perceived benefits took place. In a study of fear arousal and persuasion, which will be discussed in some detail in a subsequent chapter, Haefner '9 obtained data which showed that the health beliefs of ninth graders could be modified. High fear messages tended to have more favorable effects on beliefs about severity and preventibility (benefits) than did low fear messages. One of two effects was observed: (1) high fear messages led to a greater increase in each of the two beliefs than low fear messages, or (2) high fear meassages led to a smaller reduction in the beliefs than did low fear messages. Results for perceived susceptibility were not clear; in one experimental treatment a high fear message led to a greater increase in perceived susceptibility than did a low fear message while in a second experimental treatment, a high fear message resulted in no change or even led to a reduction in perceived susceptibility. Finally, a s indicated, Haefner and Kirscht 93 were able to modify beliefs and concomitant preventive health behavior. UNIVERSALITY OF MODEL

Voluntary, Symptom-free Health Behavior The model has been applied largely in situations in which the behavior in question is purely voluntary and the individuals studied do not believe themselves to have disabling symptoms. These criteria Health Education Monographs Winter 19 74

311

are not met in a variety of situations in which people obtain health services. For instance, social pressures may be effective in stimulating action. Legal compulsion and job requirements also account for much “health behavior.” Finally, the appearance of clear symptoms is a most frequent instigator to health action. The likelihood is, therefore, that only a minority of the population currently takes voluntary preventive action or action to detect disease in the absence of distinct symptoms. Despite these facts, continued work with the model may have great ultimate benefit. The aim in public health education is to increase the proportion of people who consistently, rationally and freely take preventive actions or actions to check on the presence of disease while free of symptoms. Careful analysis of the health decision processes in what is currently a small group of people may well be useful in subsequent planning efforts to modify the behavior of very large groups of people. Studying the exceptional case may have vast practical implications for working with the more typical. Moreover, work cited in the following chapters by Becker and by Kirscht suggest the Health Belief Model may have direct applicability to illness behavior and to sick-role behavior.

Health Beliefs and Social Class The Health Belief Model would seem to have greater applicability to middle class groups than to lower status groups since possession of the health beliefs implies a n orientation toward the future, toward deliberate planning, toward deferment of immediate gratification in the interest of long-run goals. The fact has frequently been debated 6 O ~ 6 1 whether lower status people accord greater priority to immediate rewards than to long-range goals. This difference in the time orientation of the different social classes may well have implications for the planning of preventive health programs. But these implications are far from obvious ones. Hochbaum and Kegeles, in earlier cited studies, have indeed shown that social classes differ in the frequency with which the beliefs are held. But they have also shown that where the proper constellation of beliefs exists, the probability is greater that the recommended behavior will occur irrespective of social class. Thus, public health workers must recognize that members of the lower social classes are not as prone to accept health beliefs of the kind described as are members of the higher classes. But they must also recognize that many members of the lower classes do accept such beliefs, indicating their ability to adopt a long-range perspective. Subjective time horizons are thus not immutable.

Health Habits Another possible limitation in the ultimate applicability of the model is in the case of habitual behaviors and in styles of behavior. 318

Health Education Monographs VOL. 2, NO. 4

Patterns of behavior that are developed in early life most likely are not motivated by the kinds of health concerns that may guide the adult’s behavior. During the socialization process, children learn to adopt many health related habits and practices which will permanently influence their adult behavior, e.g., brushing teeth, visiting the physician or dentist regularly, and adopting unique nutritional practices. Yet, these patterns of adjustment cannot be explained by applying the Health Belief Model. Clearly, the entire area of the determinants of health related habits is worthy of detailed investigation. THE RELATIONSHIP BETWEEN HEALTH BELIEFS AND DEMOGRAPHIC FACTORS

Typical demographic analysis of utilization rates was previously criticized, partially on the grounds that few attempts have been made to show the mechanisms that link behavior with fixed, personal characteristics. However, two published studies are relevant in this connection. Kegeles e t al, investigated relationships among the use of Papanicolaou tests, demographic factors and beliefs in the benefits of early detection of cancer. Beliefs in benefits were measured by responses to questions on the perceived importance of early versus delayed treatment for cancer and on opinions a s to whether medical check-ups or tests could detect cancer before the appearance of symptoms. An analysis of the findings discloses that personal characteristics and beliefs each make independent contributions to the understanding of behavior. Tests were much more likely to have been taken by women who were relatively young, age 35-44,white, of higher income, married, relatively well educated, and who reported higher occupational levels (using husband’s occupation in the case of married women). The study also showed that accepting the benefits of early professional detection and treatment was highly associated with having taken the test. However, the joint analysis is of most interest. Within every demographic grouping those who held a belief in benefits were much more likely to have taken the test than those not holding that belief. Similarly, within each of the belief categories those with the appropriate demographic characteristics were much more likely to have taken the action than those who did not. Clearly, the joint effects of the beliefs and the personal characteristics is much greater than the effects of either alone. In Hochbaum’s earlier study 94 a similar finding was obtained. Socioeconomic status (education and income) and the combination of beliefs in susceptibility and benefits were independently associated with having taken voluntary chest X-rays in the absence of symptoms. Within each socioeconomic status category, however, those who scored Health Education Monographs Winter I974

379

high on the combination of beliefs were much more likely to have taken the X-ray than those scoring medium or low. An interpretation of the findings of the two studies suggests that certain of the beliefs may be necessary for taking preventive or screening tests, but that they are distributed unevenly in the population, tending to be more prevalent among whites, females, those of higher socioeconomic status, and the relatively young. Why this is so is not yet known. INDUCING BEHAVIORAL CHANGE

The major focus in this paper has been on identifying factors that help to explain why people use health services. Since, however, the ultimate aim of understanding behavior in the health area is a n applied one, the problem of persuading people to use health services may appropriately be considered. Material presented earlier indicates that a decision to take a health action is influenced by the individual’s motivation, his perceived susceptibility to illness, the perceived severity of illness, socially and individually determined beliefs about the efficacy of alternative actions, psychological barriers to action, interpersonal influences, and one or more cues or critical incidents which serve to trigger a response. No a priori reason may be found to indicate that action directed toward any one of these will in the long run prove more effective than action directed toward the others. Therefore, action programs to modify behavior could legitimately focus on any one or more of the determinants. Only systematic investigation will demonstrate the conditions under which one or another of the determinants is most susceptible to effective manipulation. Despite the lack of definitive research findings, a few practical considerations may clarify the problem. Ordinarily, to change people is much more difficult than to change their environment (though the latter may itself represent no simple task). Therefore efforts to increase public response should always aim a t minimizing the barriers to action, increasing the opportunities to act (which will increase perceived benefits), and providing cues to trigger responses. Some simple but important environmental features may be modified with good effect, e.g., minimizing inconvenience by reducing financial costs of services and distances that have to be traveled to obtain them, and setting hours for service that are convenient. Moreover, cues may frequently be arranged to trigger respones, e.g., reminders from dentists and physicians, spot announcements in the mass media. Fairly simple situational changes of the kinds described may well increase the rate of preventive and diagnostic behavior. However, their effect is probably limited, if current views of the determinants of health behavior are a t all correct. Probably, after all situational 380

Health Education Monographs VOL. 2, NO. 4

improvements are made, a large number of individuals remain who are not in a state of readiness to act, and, other things being equal, Nil1 not act. Concerning such people, one must ask whether a direct effort to increase their readiness can be successful and efficient or whether success is more likely through a n indirect effort to stimulate the behavior as, for example, through the use of social pressures. Again, the question is empirical; research cited is only suggestive. Definitive research has not been performed. As a concluding section we may outline a program of research that would seem to have great potential pay-off in answering many unresolved questions a bout individual beliefs and behavior affecting the public's health, and how beliefs and behavior may be modified. The Study Population

A population group of family units in a selected geographic area should be studied for a period of a t least ten or fifteen years. While no "typical" population exists, the group selected should exhibit heterogeneity with respect to economic and educational status and racial and ethnic background. The group should be large enough to permit definitive conclusions to be drawn about the three study topics described below, after due allowance is made for inevitable sample losses due to mobility, mortality and dropouts. At least 15,000 to 20,000 families should constitute the population of concern. This, of course, means a community of 40,000 to 75,000 individuals. The studies would have three major sets of objectives and activities, each to be phased in gradually. Descriptive Studies Descriptive studies would focus on the nature of health motives, beliefs, and practices of a population over the study period with attention to changes that occur in beliefs and behavior as a consequence of purely natural events including changes in age, changes in health status, and changes in fads and fashions in medicine. ( a ) The distribution and changes over time in preventive practices should be studied, including such personal practices as exercise, smoking, diet, toothbrushing, etc., as well as professional practices, including frequency of visits to various professionals for check-ups in the absence of symptoms. The importance of habits vs conscious decision processes should also be investigated. ( b ) In addition, behavior in response t'o symptoms should be investigated to identify pathways through which various groups travel between initial experience of a symptom and the ultimate seeking of help from others. Of special interest would be the use of self medication - vitamins, tonics, tranquilizers, etc. - and the use of Health Education Monographs Winter I 9 74

381

“others,” including non-professional, para-professional, a n d professional. (c) Finally, studies are needed of behavior undertaken in response to perceived sickness. Included here might be the components of the ”sick role” (avoidance of certain social role responsibilities, acceptance of the idea that one needs help, a desire to get well, and a willingness to seek and follow the advice of competent help). Of particular importance in this connection would be studies of compliance with health recommendations. We would also wish to describe how individuals are currently involved in planning for and operating health delivery systems and how such involvement changes over time. The descriptive studies are necessary for two reasons: (1) to provide a baseline of data on what various sub-groups of people currently believe and how they behave, and ( 2 ) to provide a beginning for the study of the natural history of health beliefs and practices, that is, toward a n epidemiology of health behavior.

Explanatory Studies

It is important not merely to describe but to explain health behavior. We need to learn first how much of behavior that affects health and health practice is attributable to long standing habits which may have been acquired for reasons unrelated to health. And we need to learn when and under what circumstances these habits begin to be “stamped in.” Concerning behavior that cannot be classified as purely or primarily habitual we need to test hypotheses to account for practices. I n the area of health behavior there is no absence of hypotheses which attempt to link health practices to certain predisposing sociopsychological conditions. The following eclectic list of classes of variables included theories proposed by a wide variety of social psychologists and sociologists. The general hypothesis to be tested is that people will undertake action to prevent illness, will respond to symptoms in specified ways and/or will follow professional recommendations if ( a ) they exhibit a t least moderate health motivation, ( b ) they believe they are vulnerable to or currently suffering from serious problems, ( c ) they believe the potential or existing problem to be preventable or controllable, ( d ) their subjective time horizons are long enough for them to justify taking immediate action to ward off future health threats or impacts, (e) their social groups sanction the use of particular health treatments or providers, (f) they are willing to overcome the barriers involved in taking certain personal health actions, entering the professional delivery system and remaining in it, and (g) they are willing to follow professional advice. We know that people vary widely in their positions on each of the 382

Health Education Monographs VOL. 2, NO. 4

foregoing seven classes of predisposing psychological factors and more often than not their behavior will reflect a particular position on one or several of the factors. We do not know, however, how these psychological factors change over time and how such naturally occurring changes in the factors are reflected in subsequent behavior. I t is also necessary to learn how these social psychological factors are influenced by cues coming from the social or physical environment, such as symptoms, the occurrence of disease in significant others, the mass media, and personal influence and group dynamics. Longitudinal studies, using families a s units, would permit a test of the hypothesis that these seven classes of belief or orientation are associated with the kinds of health practices people engage in and, if so, how naturally occurring events influence those beliefs and subsequent practices. The design would permit retesting of segments of the same sample and of testing equivalent (randomly drawn) samples of the population. This feature will permit the development of reliable measures of sociopsychological factors associated with health behavior, using such approaches as test-retest, split-half and comparable forms.

Experimental Studies While descriptive and explanatory studies of health behavior are important they fail to be completely persuasive since the causal association between beliefs and behavior is not fully tested. Furthermore, even if we were persuaded that these beliefs do partly cause subsequent behavior it would still be important to know whether they can be modified by outside intervention and whether habits and practices can be taught or modified directly without resorting to efforts to change people's beliefs. Therefore cooperative field experiments with service providers should be conducted on selected sub-groups in the population directed toward teaching or modifying selected health practices, for example, dietary practices, smoking practices, volunteering for multiphasic screening, increasing compliance with health regimens and the like. Two distinct approaches should be tried. The first would attempt to teach or modify practice in children and adults by altering the relevant sociopsychological factors. Persuasive techniques would be used, delivered both through the mass media and personal and group influence. Among the persuasive techniques that seem worthy of experimentation are a number which have from time to time been shown to have some impact on opinions and behavior. These could include such variables as credibility of the communicator, the presentation of two sides vs one side of a message, the selective use of print and broadcast media for different topics and audiences, and the degree of specificity of the '

Health Education Monographs Winter I9 74

383

action recommended. In addition, we need to experiment with cumulative effects of exposure to a number of messages. A second effort would concern direct attacks on the practices themselves, both through the use of behavior modification techniques, where appropriate and ethical, and through social engineering, that is, modifying the social or physical environment to increase the probability of desired behavior. Behavior modification techniques may be particularly useful for those who are motivated to acquire or give up a particular behavior but who lack the ability to do it themselves, for example, smokers who wish to quit, obese persons who wish to lose weight, sedentary individuals who wish to exercise. Such techniques would also seem appropriate for use with school children where the cooperation of the parent and the school can be obtained in the interests of teaching or eliminating some behavior in the child. Such experimentation would also permit the test of the hypothesis that learning new complex practices will be enhanced if the desired outcome behavior is subdivided into a series of small, manageable steps, each of which is reinforced as it occurs. This, of course, reflects a fairly basic learning principle most systematically used in programmed learning but it appears not to have been systematically employed in health education programs.

REFERENCES 1. Kasl SV, Cobb S: Health behavior, illness behavior and sick-role behavior. Arch Environ Health I 12:246-266, February 1966 and I1 12:534-541, April 1966. 2. Lerner M, Anderson OW: Health Progress in the United States, 1900-1960: A Report of the Health Information Foundation. Chicago, University of Chicago Press, 1963. 3. Somers HM, Somers AR: Doctors, Patients, and Health Insurance: The Organization a n d Financing of Medical Care. Washington, T h e Brookings Institution, 1961. 4. United States Department of Health, Education and Welfare, Office of the Secretary: Health, Education, and Welfare Trends, 1963 Edition. Washington, United States Government Printing Office, 1963. 5. Borsky PN, Sagen OK: Motivations toward health examinations. Am J Public Health 49:514-527, April 1959. 6. Kegeles S S , Kirscht JP, Haefner DP, Rosenstock IM: Survey of beliefs about cancer detection and taking Papanicoloau tests. Public Health Rep 80:815-824, September 1965. 7. Rosenstock IM, Derryberry M, Carriger BK: Why people fail to seek poliomyelitis vaccination. Public Health Rep 74:98-103, February 1959. 8. United States Department of Health, Education and Welfare, Public Health Service: Health Statistics From the U S . National Health Survey: Dental Care, Interval and Frequency of Visits, United States, July 1957-June 1959, Public Health Service Publication Number 584-B14. Washington, United States Government Printing Office, 1960. 9. Kegeles SS, Lotzkar S , Andrews LW: Dental care for the chronically ill and aged. 11. Some factors relevant for predicting the acceptance of dental care by nursing home residents. (Unpublished manuscript), United States Department of Health, Education and Welfare, Public Health Service, 1959. 10. United States Department of Health, Education and Welfare, Public Health Service:

3 84

Health Education Monographs VOL. 2, NO. 4

11.

12. 13. 14.

15.

16. 17. 18.

19.

20. 21. 22. 23. 24.

25. 26. 27.

28.

29. 30.

Health Statistics From the U S . National Health Survey: Volume of Physician Visits, United States, July 1957-June 1959, Public Health Service Publication Number 584B19. Washington, United States Government Printing Office, 1960. United States Department of Health, Education and Welfare, Public Health Service: Health Statistics From the US. National Health Survey, Hospital Discharges and Length of Stay: Short-Stay Hospitals, United States, 1958-1960, Public Health Service Publication Number 584-B32. Washington, United States Government Printing Office, 1962. Antonovsky A, Hartman H: Delay in the detection of cancer: A review of the literature. Health Educ Mongr 2:98-128, Summer 1974. Herman MW: The poor: Their medical needs and the health services available to them. Ann Am Acad Political SOCSci 399:12-21, January 1972. Alpert J J , Kosa J , Haggerty RJ, e t al: Attitudes and satisfactions of low-income families receiving comprehensive pediatric care. Am J Public Health 60:499-506, March 1970. Colombo T J , Saward EW, Greenlick MR: T h e integration of a n OEO health program into a prepaid comprehensive group practice plan. Am J Public Health 59:641-650, April 1969. Monteiro L: Expense is no object Income and physician visits reconsidered. J Health SOCBehav 14:99-115, June 1973. Koos EL: T h e Health of Regionville - What the People Thought and Did About It. New York, Columbia University Press, 1954. Stoeckle JD, Zola IK, Davidson GE: On going to see the doctor, the contributions of the patient to the decision to seek medical aid, a selected review. J Chronic Dis 16:975-989, 1963. Zola IK: Illness behavior of the working class: Implications a n d recommendations. In Shostak A, Gomberg W (eds): Blue Collar World. Englewood Cliffs, New Jersey, Prentice-Hall, Inc., 1964, pp 351-361. Freidson E : Patients’ Views of Medical Practice. New York, Russell Sage Foundation, 1961. Suchman EA: Social patterns of illness and medical care. J Health Hum Behav 6:216, Spring 1965. Rosenstock IM: What research in motivation suggests for public health. Am J Public Health 50:295-302, March 1960. Rosenstock IM: Why people use health services. Milbank Mem Fund Q 44:94-127, July 1966. Hochbaum GM: Public Participation in Medical Screening Programs: A Sociopsychological Study, Public Health Service, Public Health Service Publication Number 572. Washington, United States Government Printing Office, 1958. Kegeles S S : Some motives for seeking preventive dental care. J Am Dent Assoc 67:90-98, July 1963. Heinzelmann F: Determinants of prophylaxis behavior with respect to rheumatic fever. J Health Hum Behav 3:73-81, 1962. Flach E : Participation in case finding program for cervical cancer. Administrative Report, Cancer Control Program, United States Public Health Service, Washington, D.C., 1960. Leventhal H, Hochbaum G, Rosenstock I: Epidemic impact on the general population in two cities. In T h e Impact of Asian Influenza on Community Life: A Study in Five Cities. United States Department of Health, Education and Welfare, Public Health Service, Pub No 766, 1960. Kegeles SS: Why people seek dental care: A test of a conceptual formulation. J Health Hum Behav 4:166-173, Fall 1963. Kirscht JP, Haefner D, Kegeles S , Rosenstock I : A national study of health beliefs. J Health Hum Behav 7:242-254. 1966.

Health Education Monographs Winter I 9 74

385

31. Haefner DP, Kegeles S, Kirscht J, Rosenstock I: Preventive actions in dental disease, tuberculosis and cancer. Public Health Rep 82:451-459, 1967. 32. Festinger L: A Theory of Cognitive Dissonance. Evanston, Illinois, Row Peterson, 1957. 33. Haefner DP, Kirscht JP: Motivational and behavioral effects of modifying health beliefs. Public Health Rep 85:478-484, June 1970. 34. Becker M, Kaback M, Rosenstock I, Ruth M: Some influences on public participation in a genetic screening program. ( I n press) Community Health 1, 1975. 35. Heinzelmann F, Bagley RW: Response to physical activity programs and their effects on health behavior. Public Health Rep 85:905-911, October 1970. 36. Fink R, Shapiro S, Roester R : Impact of efforts to increase participation in repetitive screenings for early breast cancer detection. Am J Public Health 62:328-336, March 1972. 37. Battistella RM: Limitations in use of the concept of psychological readiness to initiate health care. Med Care 6:308-319, July-August 1968. 38. Battistella RM: Factors associated with delay in the initiation of physicians’ care among late adulthood persons. Am J Public Health 61:1348-1361, July 1971. 39. Suchman EA: Health attitudes and behavior. Arch Environ Health 20:105-110, January 1970. 40. Ogionwo W: Socio-psychological factors in health behavior: a n experimental study on methods and attitude change. I n t J Health Educ, Supplement to Vol. 16, No. 2, April-June 1973. 41. Gochman DS: Some correlates of children’s health beliefs and potential health behavior. J Health SOCBehav 12:148-154, June 1971. 42. Kegeles SS: A field experimental attempt to change beliefs and behavior of women in a n urban ghetto. J Health SOCBehav 10:115-124, June 1969. 43. Aiitonovsky A, Kats R: T h e model dental patient: An empirical study of preventive health behavior. SOCSci Med 4:367-380, 1970. 44. Birch HG: T h e role of motivational factors in insightful problem-solving. J Comp Psych 38:295-317, 1945. 45. Gochman DS: Children’s perceptions of vulnerability to illness a n d accidents. Public Health Rep 85:69-73, January 1970. 46. Gochman DS: Children’s perceptions of vulnerability to illness and accidents: A replication, extension, and refinement. HSMHA Health Rep 86:247-252, March 1971. 47. Gochman DS: Children’s perceptions of vulnerability to illness and accidents: a vulnerability to health problems. Health Serv Rep 87:282-288, March 1972. 48. Guskin SL: T h e measurement of change in beliefs about tuberculosis. In Proceedings of the First Annual Conference on Behavioral Aspects of Tuberculosis Control, Atlanta, Georgia. Tuberculosis Branch, Communicable Disease Center, Public Health Service, 1965. 49. Haefner DP: Arousing fear in dental health education. J Public Health Dent 25:140146, 1965. 50. Simmons OG: Social Status and Public Health. Social Science Research Council, Pamphlet number 13, New York, 1958. 51. Green LW: Status Identity and Preventive Health Behavior. University of California, Berkeley, Pacific Health Education Report No. 1, 1970.

386

Health Education Monographs VOL. 2, NO. 4