|Home | About | Journals | Submit | Contact Us | Français|
Patients' knowledge concerning high blood pressure (HBP) is a useful outcome measure in HBP education programs. However, valid and easy-to-use HBP knowledge assessment tools are scarce. The purpose of the study was to validate the High blood pressure Knowledge Test (HKT) in two independent samples of Korean Americans with HBP (N=885); 61% women; mean age about 61 years (SD=11.0); 44% some college education or greater. Psychometric and item response theory analyses identified 21 items fitting a unidimensional model to form the HKT. Internal consistency was acceptable: Cronbach's alpha=0.70. Construct validity testing revealed that those with controlled BP had significantly higher HKT scores than those with uncontrolled BP (p=.012). The HKT is a simple questionnaire for assessing and monitoring HBP knowledge. It is sensitive to differences in BP control status and should provide a valid, reliable and standardized measure of HBP knowledge with wide relevance.
The burden of high blood pressure (HBP) in the US is tremendous, affecting more than one in every four adults in the US.1 Assessing HBP-related knowledge is an integral part of the overall care of patients with HBP. In particular, HBP knowledge has been used in intervention studies as outcomes of patient education programs.2,3 Higher HBP knowledge has been associated with lower drop-out rates among participants in a behavioral intervention program,4 greater medication adherence,5 and better HBP control.6,7
To assess individuals' knowledge about HBP, several instruments have been used (Table 1). However, valid and easy-to-use knowledge assessment tools are scarce. For example, of the five HBP knowledge instruments identified (Table 1), only one study provided information about the validity of the instrument.8 Internal consistency reliability (i.e., coefficient alpha) was tested for four of the five instruments (range=.70 to .85) in samples as small as 26.4,6,8,9 While most instruments used correct/incorrect response format, one instrument included open-ended questions with answers individually recorded verbatim and scored.6 Only three instruments provided specific content areas (e.g., definition of HBP, risk factors, treatment, consequences) covered by the instrument.6,8,10 These knowledge instruments have been used in a limited number of settings, which may be due, in part, to a lack of rigorous validation of the instruments and a lack of clarity in terms of its content or scoring instruction.
Available epidemiological studies indicate that ethnic minorities, in particular, Korean Americans (KAs), bear a considerable burden of HBP, as they have been shown to have a higher prevalence and lower rate of treatment and control of HBP when compared to other age-matched ethnic groups.11-14 In order to address the lack of adequate HBP care and control among KAs, an academic-community partnership was established. Together, the team conducted community-based intervention studies aimed at promoting HBP control in KAs.3,15,16 While the scope and design of each of those studies differ, information about a few key psychosocial variables such as HBP knowledge has been collected using the same instruments. In particular, data on HBP knowledge has been collected with a Hypertension Knowledge Test (HKT) from two independent samples of nearly 900 KAs with HBP. The HKT is a tool that was designed to assess not only general knowledge about HBP but also Korean cultural practices that could affect HBP care. Cumulatively, the resulting data may provide a unique opportunity to systematically assess the reliability and validity of the HKT in KAs, one of the hardest-to-reach ethnic minority groups.17 The purpose of this paper is to assess the utility, reliability and validity of the HKT. Using a rigorous psychometric evaluation approach (item response theory [IRT]), item characteristics including item difficulty and item discrimination were evaluated for each item on the HKT.
Data from two samples of KA immigrant patient populations were the basis for the current analysis (Table 2). The first data set (Study 1) was obtained from 445 middle-aged (range=40-64 yrs, mean=51.9±5.7 yrs), first-generation Korean immigrants in the Baltimore-Washington metropolitan area. They completed the HKT as part of an intervention trial for HBP control designed to test the effects of a culturally-tailored education program combined with bilingual nurse telephone counseling (see 3,12,16 for details about study design and methods). Eligibility criteria for entry into the trial were: 40 to 64 years of age; systolic BP (SBP)> 140 mm Hg and/or diastolic BP (DBP)> 90 mm Hg on two separate occasions or being on antihypertensive medication; and self-identification as a KA. At baseline, more than half of participants were women (52.3%), had resided in the United States for 16.2 (SD=8.9) years, and had at least some college education (54.6%). The majority of participants was married (93.0%), and reported working full- or part-time (81.8%), with only 40.4% being covered by any form of health insurance at the time of the study. Nearly three-quarter (72.9%) reported a family history of HBP with a mean years of HBP being 4.3 (SD=6.3) years. While 26.3% had a co-morbid condition (e.g., diabetes), 54.6% were on anti-hypertensive medication, with less than one third (29.7%) having BP control.1
The second data set (Study 2) was obtained from 440 elderly, first-generation hypertensive KA immigrants (range=60-89 yrs, mean=70.9±5.5 yrs) in a prospective intervention trial for HBP control designed to test the effects of a health literacy-focused behavioral intervention. The on-going trial is being conducted in the Baltimore-Washington metropolitan Korean community. Since the majority of KAs attend ethnic churches, churches were used as the randomization unit, after considering several key factors such as denomination, geographic location, and the size of membership of each church. This group also completed the HKT at baseline. Eligibility criteria were the same except for the age being 60 years of age or older. Slightly more than two thirds of the sample included women (69.5%), mean years in the United States were 24.2 (SD=11.3) years and 58.9% reported living with a spouse at the time of the study. About one third completed some college or more education (33.9%). About 73% had worked full- or part-time before retirement, with 80.2% having health insurance. Additionally, 46.6% reported having a co-morbid condition. In the sample, 52.2% had a family history of HBP with a mean years of having HBP being 9.6 years (SD=8.9). The majority of participants reported being on a HBP medication regimen (84.3%), but only 38.6% had their BP controlled.
For the purposes of this study, the two Korean samples were combined to create a new sample pool. The rationale for the data aggregation was three-folds: (i) the primary purpose of this analysis was a psychometric assessment of the HKT using IRT methods, which required a large sample size; (ii) the KA participants were all first-generation immigrants who were born, educated, and socialized in Korea and therefore assumed to share the same language and cultural backgrounds; and (iii) each sample was recruited independently, each representing different age ranges, and hence fulfilling the independent observations assumption in IRT-based analysis.18 As a result, we were able to achieve a sample of KA participants that was diverse in age, gender, education, and medical history (Table 2).
All study procedures were approved by the Johns Hopkins Hospital Institutional Review Board, and every participant provided written informed consent. Massive outreach efforts, in collaboration with our community partners (e.g., the Korean Resource Center, a community-based non-profit organization), were made during each enrollment period of October 2003-December 2004 for Study 1 and December 2007-June 2009 for Study 2, respectively. Detailed outreach effort and recruitment success have been published elsewhere.12,17 In brief, program brochures and fliers were distributed at ethnic churches, groceries, and social clubs in the target community. Additionally, intensive media campaigns announcing the project were made in ethnic newspapers, weekly magazines, and several ethnic radio stations. During the baseline evaluation for each study, trained bilingual research staff collected physiologic data, including BP and body weight, while demographic and psychological data, height, and antihypertensive medication use were all obtained via self-report. Furthermore, trained bilingual research staff and nurses were available at every recruitment event held in the community to help with the explanation and completion of the baseline survey.
The HKT consisted of 12 items developed by the National HBP Education Program, the National Heart, Lung, and Blood Institute (Check Your High Blood Pressure IQ),19 which assessed: knowledge of causes, symptoms and dangers of HBP and ways to prevent or control it. The IQ quiz was used in earlier studies of patients with HBP in the US and South Africa.20-22 In addition, 14 items were generated by the bilingual investigative team (HRH, JEL, MTK) based on a literature review and community input (involving lay health workers with several years of working experience in the KA community and hypertensive patients) which reflected KA health practice related to HBP (e.g., common Korean diet based on high-salt food). This resulted in a total of 26 items written at the 4th grade reading level. The 26-item knowledge test has been used in our previous studies of KAs.5,12,23,24 For the purpose of validating the knowledge items, expert consensus approach (HRH, MTK, TN, JEL, and HS) was taken to identify red herring items (e.g., “Bald people are more likely to have HBP”). Red herring items are those which do not directly relate to HBP knowledge. The resulting 21 item on the HKT measure individuals' level of understanding of HBP, its treatment, and follow-up care. Participants are asked to respond true or false to statements such as “If your mother or father has HBP, your chance of getting it is higher,” “Koreans eat two or three times more sodium than they need,” or “Regular exercise can help reducing BP,” or several multiple choice items (Table 3). HBP knowledge scores are calculated by counting the number of items with correct responses with a possible range from 0 to 21. This knowledge quiz was originally developed in English and back-translated.
Measurements of DBP and SBP were obtained by trained research staff using the A&D UA-767 (A&D Company, Ltd, Tokyo, Japan), a fully automatic device based on the oscillometric method. The A&D UA-767 device had been validated against a mercury sphygmomanometer.25 BP measurements were taken with appropriately sized cuffs after the participant was seated for 5 minutes.26 Baseline BP was measured by averaging the second and third BP readings, recorded in mmHg.
A multiple analytic approach was employed. First, to evaluate the internal consistency reliability and item homogeneity of the HKT, item-total correlations and Cronbach alphas were calculated for the HKT total scores. Item-total correlations above 0.15 and alpha coefficients above 0.70 are considered adequate.27 Then, Item Response Theory (IRT) analysis of the entire set of HBP knowledge items was conducted to further describe item characteristics. Under IRT, the likelihood that a subject responds correctly to an item, or endorses a response category, depends on the amount of the underlying construct, being measured by the item, that a respondent possesses.28 The discrimination (a) parameter reflects the ability of an item to discriminate between different levels of underlying trait, with higher a values indicating better discrimination. The location or difficulty (b) parameter indicates the location of the item on the underlying construct, reflecting where along the measured construct the item is most discriminating. In a simple binary case, such as with correct and incorrect responses to knowledge questions, the location parameter would be defined as the location on the underlying construct (HBP knowledge) where the probability is 50% for answering the item correctly. In this study, higher b estimates indicate that greater HBP knowledge is needed to answer the question correctly (i.e., harder questions). Using Multilog, we tested the model fit of the 2-parameter and 3-parameter logistic models to generate item parameters. Item characteristic curves based on these estimates were created to provide graphic representation of item performance. The 3-parameter model, which adds a guessing parameter to the model, did not improve model fit (2 parameter model: -2LL=5446.9; 3 parameter model: -2LL=5446.3). Furthermore, the estimated guessing parameter was negligible for 18 of the 21 items (c ≤0.03) and small for the remaining three (c ≤0.16). Therefore, we choose to present findings from the more parsimonious 2-parameter model.
Finally, construct validity of the HKT was tested by examining the relationship between scores on the HKT and BP. Based on a theoretical proposition by the Learned Resourcefulness Model29 which guided both studies, one of the key objectives was to improve participants' knowledge related to HBP, and hence improving HBP control. In accordance with the definition used in the seventh report of the Joint National Committee on Prevention, Detection, and Treatment and Control of High Blood Pressure,1 BP control was defined as BP <140/90 mm Hg (130/80 mm Hg for patients with diabetes). Thus, for the KA samples, it was hypothesized that those with BP control would have higher HBP knowledge than those without BP control. This hypothesis was tested using independent samples T-test. Statistical significance was defined as p ≤0.05.
Table 4 summarizes means, standard deviations, item-total correlations, and Cronbach's alpha reliability coefficients for the HKT for two independent samples. Mean scores for both samples were similar (15.0 for Study 1 vs. 15.8 for Study 2), with a mean score of 15.8 for the combined sample (about 75 based on 100 as a perfect score). All item-total correlations were above the cutoff set a priori at 0.15, ranging from 0.18 to 0.42 for Study 1 and 0.18 to 0.34 for the total sample. For Study 2, two items (Q3 and Q6) did not meet the cutoff with item-total correlations of 0.11 and 0.14, respectively; item-total correlations for items Q3 and Q6 were 0.30 and 0.40 for Study 1 and 0.21 and 0.30 for the total sample. Alpha coefficients of the 21-item HKT in the Korean samples yielded 0.74 for Study 1, 0.62 for Study 2, and 0.70 for the combined sample.
Table 5 shows the estimated parameters from the IRT models. The discrimination (a) parameter suggested that all items on the HKT were generally indicative of their corresponding scale, although some items had better discriminability than others. In particular, items such as “HBP is a man's problem” (a = 1.38), “If you're overweight, you're two to six times more likely to develop HBP” (a = 1.24), and “Regular exercise can help reducing blood pressure” (a = 1.10) had the best discrimination among the items examined. In contrast, items on symptoms of HBP (Q3, a = 0.62), sodium consumption by Koreans (Q8, a = 0.65), cold weather effects on BP (Q12, a = 0.65), and diet changes that lower BP (Q20, a = 0.64) had lower discriminability. “Regular exercise can help reduce BP” was identified as an easy item (b = -3.52), which most respondents answered correctly. This was consistent with the high percentage (96.5%) of participants who answered the question correctly. On the other hand, the items about pregnancy related HBP (b = 0.29), whether HBP always has symptoms (b = 0.67) or whether BP gets lower in cold weather (b = -0.29) were substantially harder questions to answer correctly. Those location estimates were also consistent with the lower percentages, 45%, 41%, and 54% respectively, of participants who responded correctly to those items. A review of the item characteristic curves for all 21 questions on the HKT revealed a clustering of items at the lower end of the HBP knowledge trait (Figure 1).
Construct validity for the HKT was assessed by testing the hypothesized relationship between HBP knowledge and HBP control. The scores on the HKT were correlated with SBP values for both samples but not with DBP (Table 6). For the combined sample, a statistically significant inverse relationship was obtained between HBP knowledge and SBP (r=-.11, p=0.001). As hypothesized, those with controlled BP had significantly higher HBP knowledge scores than those with uncontrolled BP (t=-2.52, p=.012).
Accurate assessment of HBP knowledge is an important first step to identify individuals in need of HBP education, since knowledge is often a prerequisite for a patient to perform appropriate HBP self-care as shown in previous research.5 HBP knowledge may be particularly critical in the management of HBP among non-English speaking immigrants such as KAs who often have very limited resources in health information.17 Based on the descriptive statistics (Table 3), KAs were generally knowledgeable about etiology and common risk factors (e.g., drinking, obesity, lack of exercise, salt) of HBP and correctly identified “high” versus “good” BP values. On the contrary, a significant proportion of KAs answered incorrectly items about HBP symptoms and diagnosis, HBP medications, harmful effects of HBP over time, and the relationship between BP and cold weather. KAs' lower knowledge about these items, particularly HBP medications is of concern in view of their high prevalence of uncontrolled HBP and less access to medications.12,14 These results suggest and support continuing need for educational interventions to address their insufficient knowledge in key areas of HBP management.
The reliability and validity of the HKT were supported in the combined sample of KA patients with adequate differentiation between those with and without HBP control. Item-total correlations for the 21 items on the HKT were all within the acceptable range of 0.1527 for the total sample with internal consistency coefficient of 0.70, meeting the minimum acceptable level of reliability. The result may have to do with the item heterogeneity in that the HKT addresses knowledge of different HBP risk factors (e.g., obesity, diet, drinking) or various symptoms and consequences of HBP. Nevertheless, the HKT yielded a reliability estimate that should allow researchers to use it with some confidence. Indeed, the usefulness of the HKT as an outcome measure has been reported in a separate study with significant changes in HKT scores at 3 months follow-up (p=0.000) after an educational intervention for KAs.16
IRT findings offer insights on the overall and specific performance of the HKT items. In general, the information curves for the HKT items revealed that the 21 items performed in a similar pattern as a cluster at the lower end of the HBP knowledge trait. Location parameters (b) of the 21 items were between 0 and -3. The results suggest that the HKT is best at quantifying differences among respondents who are less knowledgeable about HBP, particularly within 3 standard deviations below the group mean. All the items on the HKT were found to be at least moderately discriminating with the discriminating (a) parameters ranging from 0.62 to 1.38, though several items were particularly good at distinguishing between respondents at different levels of HBP knowledge. Taken together, the current set of items on the HKT will perform best among quantifying differences amongst individuals with lower HBP knowledge and hence, can be useful for identifying individuals who are in need of further education about HBP.
While this study offers great opportunities to measure HBP knowledge among recent KA immigrants in the Baltimore-Washington metropolitan area, an understudied and vulnerable population, there are a number of limitations to consider. The two samples included in the analysis were participants in educational intervention trials for HBP control. People who volunteered to be in those trials were likely to be more interested in their health and HBP management than those who did not. Therefore, the level of HBP knowledge reported in this study might have been an overestimate of the HBP knowledge in the larger KA population, limiting the generalizability of the findings. Additionally, the HKT was administered in the two behavioral intervention trials along with a number of other study questionnaires. Therefore, participant fatigue might have been introduced, influencing participant performance on the measure.
In conclusion, while there is no standard to measure the level of HBP knowledge required for adequate management of HBP, we developed the HKT, a measure that was modified from the National Heart Blood and Lung Institute hypertension knowledge questionnaire19 as well as a literature review and community input, to assess patients' general knowledge of HBP. The reliability and validity of the HKT was supported in this sample of KAs. The HKT was also sensitive in detecting differences among those with and without adequate HBP control. Furthermore, given that there are 21 items written at the 4th grade level, the HKT can be easily self-administered, and may be particularly helpful in identifying patients with low HBP knowledge. With continued validation, clinicians can use the knowledge gaps demonstrated on the HKT to provide the basis for additional hypertensive management education for KAs with HBP. Future research is warranted to evaluate the HKT in other ethnic samples of hypertensive patient populations.
This study was supported by grants from the Agency for Healthcare Research and Quality (R01 HS13160, Clinical Trials Registry NCT00473681) and the National Heart, Lung, and Blood Institute (R01 HL085567, Clinical Trials Registry NCT00406614). Editorial support was provided by the Johns Hopkins University School of Nursing Center for Excellence for Cardiovascular Health in Vulnerable Populations (P30 NR011409). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.