Upon reading expert and participant comments about the scales, the authors simplified and modified some wordings of items during the modification phase [28,33,36,39]. This strategy was frequently used in the reviewed studies (eg, some words in the items were replaced by other words, and medical terms were replaced by generally known terms).

To reach cultural accuracy of items, some words in the items were replaced by other words. Following one participant’s suggestion, in the study to adapt a Korean version of Champion’s Health Belief Model Scale, the word “hok” was changed into “meongwooli.” Even though both words meant lump or mass in Korean, the authors asserted the modification helped the women in the study to better understand the content of the instrument [35]. To accurately measure Turkish women’s health belief of breast cancer screening, the word “komik” (meaning funny) was changed into “tuhaf” (another expression of funny), and “gizlilik” (meaning privacy) was changed into “mahremiyet” (another expression of privacy) per expert suggestions. This change increased the consistency between the translated version and original version of the instrument [25]. The item “When I do breast screening examination, I feel good about myself” was changed to “I feel self-satisfied” as it was closer to the Iranian meaning than “feel good” [29].

Another strategy used to reach cross-cultural semantic equivalence was rephrasing medical terms to generally known terms. This strategy was used in the study conducted to measure health beliefs of colorectal cancer screening among Korean Americans [30]. The medical term “fecal occult blood test” was replaced with lay language “stool blood test,” as the medical term might be difficult to understand for participants not employed in a health-related field [30]. In the Turkish Health Belief Model Scale [26], the term “lump” was translated to “kitle” initially and changed to “sert yumrubeze,” a lay-language word that has a similar meaning to “kitle” and would be understood by Turkish women [26]. In the Maltese Health Belief Model Scale [34], “mammografija” was changed to “mammogram” and “nipil” was translated to “nipple” because “mammogram” and “nipple” are generally known terms with similar meanings to “mammografija” and “nipil” in Maltese. A similar strategy was used in the study conducted among Hispanic women [31]. The authors changed the original term of breast, “mama,” to “seno,” a word with similar meaning, since “seno” would be most understood in all Hispanic groups.

To avoid causing misunderstanding, confusion, and anxiety, some instrument items were changed by reversing direction of the meaning. This strategy was used in one study conducted with Maltese women [34]. The item “...will last for a short time” with reverse scoring was replaced with “...will last for a long time.” Then scores on the item did not have to be reversed.

To increase clarity and achieve semantic equivalence, wordings of response options in the instruments may need to change. This strategy was used in 3 studies [29,30,32]. To adequately measure Mexican women’s health belief toward breast cancer screening, the response options were amended to 4=yes, 3=I think so, 2=I don’t think so, and 1=no, since the original response options were found to be problematic for the participants [32]. Also, instead of using the original response options (from “strongly disagree” to “strongly agree”), the Farsi Health Belief Model Scale used “not at all true” to “very true” for the perceived severity, susceptibility, benefits, and barriers subscales and “never” to “always” for the health motivation subscale, since most of the participants in the pretest phase reported problems with the format of response options [29]. Furthermore, the response option in the scale to measure Korean Americans’ health beliefs about colorectal cancer screening, “neutral” was changed to “so-so” per expert advice and suggestions in the published literature [30].

This strategy was used to achieve content equivalence by making the meanings of the items relevant or appropriate. It could avoid arousing fatalistic thoughts that were commonly present among minority populations. This strategy was used in the study conducted with Maltese women [34]. The item “My illness has serious...consequences” was replaced with “Breast cancer has serious...consequences.” The participants were asked to report their personal views about breast cancer instead of an illness personally affecting them [34]. Another strategy used in the study was using the third person pronoun instead of the first person pronoun to avoid arousing fatalistic thought. In some cultures, people believe that expressing an ominous event in the first person indicates that the event will occur [29]. In the Iranian Health Belief Model Scale [29], the first person pronoun in the item “If I developed breast cancer, I would not live longer than 5 years” was changed to the third person pronoun: “If someone developed breast cancer, she would not live longer than 5 years.” Similarly, changes were also made to another 3 items in the perceived severity scale in this study [29].

To modify items measuring the sizes of breast lumps in Champion’s Health Belief Model Scale (“I am able to find a breast lump which is the size of a quarter/dime/pea”), different items or coins familiar to the target population were used. In the Indonesian Health Belief Model Scale, “quarter” and “dime” were changed to “walnut” and “hazelnut,” respectively, because the sizes of a walnut and hazelnut are commonly known in Indonesia [33]. To find equal sizes to “quarter, dime, and pea,” the words were translated into “chickpea, hazelnut, and walnut” in the Turkish Health Belief Model Scale [25]. To represent dime and quarter in Iranian culture, the authors used “filbert” and “rather greater than filbert,” since “filbert” is commonly known in Iran culture [29]. In addition, the authors used different sizes of Turkish coins equalized to the sizes of the American quarter and dime, and the sizes of the original quarter and dime were given in centimeters in another study with Turkish women [26]. Similarly, wordings about the size of a palpable lump were also changed according to the sizes of currencies in Kaunas [38] and South Korea [35] in 2 other studies.

To increase the cultural sensitivity of the adapted instrument, thus increasing the scale’s content equivalence, entire items or parts of items were added to the adapted instrument.

Adding entire items to the modified scale was a common strategy used in the reviewed studies. In the Iranian Health Belief Model Scale, the item “I am more likely than the average woman to get breast cancer” was added to the adapted instrument because this item was maintained in the previous version of Champion’s scale and also because of the special features of the participants in the study (who had a family history of breast cancer) [27]. Furthermore, 4 items, (I don’t know where to go for mammography; I don’t have any problem with my breasts, I don’t need mammography; I do self-examination of the breasts, so there is no need for mammography; and I don’t have enough money for mammography) were added to the subscale of perceived barriers per participant discussion [27]. In addition, per expert suggestions, 2 items concerning awareness of the age and frequency at which mammograms should be undertaken were added to the self-efficacy subscale of the Mexican Health Belief Model Scale. Two items concerning myths were added to the barrier subscale. Two items concerning risk factors for cancer were added to the susceptibility subscale, and one item on drug use avoidance was added to the health motivation subscale [32].

To make the meaning of the items clear and easily understandable, additional words of explanation were added as part of some items. In the study conducted with Korean Americans, the explanation “not wanting to let other people know that you are doing the stool blood test or handling stool for the test” was added to the item “not having privacy would keep you from having a stool blood test,” since privacy could be interpreted as several different Korean words depending on the context [30]. Also, the item, “I have other problems more important than having a stool blood test,” was expanded to “Having a stool blood test is not the most urgent and important problem I have, which keeps me from doing it” because several participants reported they did not understand its meaning [30]. In the Turkish Health Belief Model Scale [26], the term “radyasyon” was explained as “radiation, x-ray/in other words radyasyon-rontgen” because some women in the target population would not know the word radiation since general understanding of the word was “x-ray.” Similarly, in the same study, “mammography” was expanded to “mammography” and “breast x-ray” in the modified scale because many women in Turkey do not know about early diagnostic methods for breast cancer, especially mammography [26]. Furthermore, in the translated Kaunas instrument, after pretesting with 10 women, 2 alternatives of the original word meaning “privacy,” solitude and severalty, were written with the original meaning next to the item “I don’t have enough privacy to do breast examination” [38].

Deleting entire items or parts of items was a common strategy used in the studies to achieve content equivalence. Entire items in the reviewed studies were removed due to redundancy, irrelevance, or inaccuracy or a low content validity index at the item level. In one study, the item “I am too old to need a routine mammogram” was deleted because the Iranian women (n=200) in the study were younger (mean age 46.15 [SD 7.26] years, range 28 to 69 years) than the participants in the original Champion study (aged 50 years and older) [27]. Furthermore, the item “I don’t know how to go about getting a mammogram” was eliminated because the city was quite small [27]. Similarly, 2 items, “Breast cancer will last for a long time” and “I expect to have breast cancer for the rest of my life,” were removed in another study because the 2 items were found to confuse the Maltese participants and cause consistent heightened anxiety in responders [34]. Finally, the item “receiving a mammogram prior to breast screening” was deleted from a study to avoid overlap [34].

To reach cross-cultural content equivalence, parts of sentences in items from the original instruments were deleted. The terms “boyfriend” and “partner” in the item “Breast cancer would threaten a relationship with my boyfriend, husband, or partner” were deleted in 2 studies with Iranian women because sexual relationship outside the marriage is forbidden by Islamic rules and religious norm [29]. Furthermore, the word “blood” in the item describing “stool blood test” was deleted to emphasize that a stool sample was needed because some Korean American participants wrongly thought the test required blood to be taken [30].

In the Maltese Health Belief Model Scale, the authors added subscales concerning the impact of sociodemographic and socioeconomic factors (eg, items about education level and income) on women’s breast screening behavior to acknowledge the contributions of those criteria to breast cancer screening [34]. A cues to action subscale (such as physician recommendations and family history), often omitted from empirical studies using Champion’s Health Belief Model Scale, was added because the authors thought cues to action were important criteria through which to examine the health belief of Maltese toward breast cancer screening [34]. In the Turkish Health Belief Model Scale, 4 items (cost, fatalism, preference for female health care professionals, and distance from the health center) thought to be appropriate to Turkish culture were added to the barrier subscale [23]. Furthermore, to test Jordan women’s fatalistic beliefs about breast cancer, the item “If I get sick with breast cancer, I believe this is my fate and practicing breast screening examination will not change my fate regardless of when the tumor is detected” was added to the barrier subscale [36].

In the reviewed studies, some subscales in the Health Belief Model Scale were deleted to increase cultural sensitivity. When the literature search indicated that most Maltese women perceived breast cancer to be a serious threat, the perceived severity scale was removed from the Maltese Health Belief Model Scale because the authors determined that perceived severity was not a criterion for examining the health belief of Maltese women toward breast cancer screening [34]. In addition, the item “people who perform mammograms are rude to women” was eliminated because participants believed that a sense of shame prevents them from receiving mammography instead of the issue of obscenity, and the statement lacked compatibility with Iranian culture [27].

This strategy was used in one study aiming to measure the health belief of Korean Americans toward colorectal cancer screening [30]. When the primary investigator and translation committee members encountered difficulty reaching a consensus on translation, they either sought guidance from an expert or literature published in both Korean and English to solve the dispute [30].

This strategy was used in one study conducted with Hispanic women. The terms used for marital status aroused a disagreement over the comment from an expert panel member. The expert did not believe that every Hispanic would understand estado civil to mean marital status. However, a consensus was reached by the majority of the panel and the term estado civil was used in the translated instrument [31].

When meanings of the items in the original scale were not stated clearly, consulting with the author who developed the scales may provide clarification. In a study conducted with Korean Americans, some participants did not understand the meaning of the term “privacy” in the barrier items. Hence, the primary investigator consulted with the author who developed the barrier scale. The author clarified the meaning of privacy, and the item was rephrased accordingly [30].