Accuracy of self‐reported hypertension: Effect of age, gender, and history of alcohol dependence

Abstract

Patient awareness of medical conditions may influence treatment seeking and monitoring of these conditions. Accurate awareness of hypertension reported to clinicians evaluating patients for whom clinical history is limited, such as in emergency care, can aid in diagnosis by revealing whether measured hypertension is typical or atypical. Measurement of blood pressure in a laboratory study was assessed at rest, immediately before phlebotomy, and within 10 minutes after. The resting measure was used to determine the accuracy of self‐reported hypertension in 283 adults. Parametric analyses were conducted to identify potential variables influencing accuracy of self‐reported hypertension. Sensitivity, specificity, and the kappa coefficient of agreement were calculated to determine the influence of alcohol dependence (AD), sex, age, and cigarette smoking on hypertension awareness. Self‐report was mildly sensitive, correctly identifying individuals with hypertension in approximately 37% of the cases, but was highly specific (95%) in identifying individuals without hypertension. Similar sensitivities were found in analyses separated by sex. Sensitivity was greater in those over age 55 (53%) in comparison with those <54, as well as in those who were not smoking. Comparison of those with and without a history of AD revealed that both groups show similar accuracy in reporting hypertension. Absence of hypertension can be accurately determined with self‐report data in those without hypertension. A significant proportion of those with measured hypertension report an absence of hypertension.

1. INTRODUCTION

Cardiovascular disease is a worldwide, chronic health condition. One treatable risk factor leading to cardiovascular disease is hypertension, which has been reported to contribute to over 7 million deaths annually. ¹ Assessing the accuracy of self‐reported hypertension is needed for evaluating results of health survey data.

Several studies have examined self‐reported hypertension status. Multiple factors appear to influence accuracy of self‐report, including age, sex, and education. In middle‐aged adults (mean = 45.5 years), Bowlin et al ² reported that hypertension tends to be underreported. Half of the hypertensive cases were misclassified as not being hypertensive by self‐report, and additionally, 82% of those who reported having hypertension at the time of the interview stated their blood pressure was controlled. In a large sample of Koreans over the age of 50, Chun et al ³ found the highest sensitivity (77.5%) in those between the ages of 60 and 69. Goldman et al ⁴ also reported the highest accuracy of self‐report in individuals over the age of 60. A recent meta‐analysis has shown an association between age and sensitivity. ⁵ While self‐report can be considered reliable to rule out hypertension, the probability of self‐report correctly identifying patients with clinically diagnosed hypertension was quite diverse, ranging from 13% to 92% across the studies analyzed in the meta‐analysis. ⁵

Sex has also been reported to have an influence on the accuracy of hypertension self‐report. Tolonen et al ⁶ found underreporting of hypertension, particularly in men, while women tend to over‐report hypertension. However, Chun et al ³ noted that females are more aware of their hypertensive status.

An individual's level of education may have an influence on their awareness of hypertensive status. A recent study has shown a gradient effect of hypertensive awareness and level of education, with individuals who attained at least a college education showing the highest sensitivity for hypertension that gradually decreased with educational level. ³

Alcohol dependence (AD) is associated with chronic medical problems requiring continued medical care. Heavy and moderate drinkers are hospitalized for a greater number of days than abstainers. ⁷ Currently, there is a paucity of literature on the accuracy of self‐reported medical conditions in alcohol‐dependent individuals. Because cognitive deficits have been found to be a consequence of long‐term and excessive drinking in those with an alcohol use disorder, ⁸ , ⁹ , ¹⁰ we hypothesized that self‐report accuracy may be diminished in alcohol‐dependent individuals relative to those in the general population.

Hypertension is a common alcohol‐related health problem in heavy users of alcohol and those who are alcohol dependent. ¹¹ Some of the hypertension seen in these individuals occurs in association with alcohol withdrawal. ¹¹ Elevations associated with withdrawal were reported to dissipate by the end of an 18‐day observation period. Maheswaran et al ¹² assessed non‐dependent users of alcohol as part of an occupational survey in which respondents were asked to recall use of alcohol in the previous 7 days, finding effects limited to the previous 3 days but not all 7 days.

The present analyses compare self‐reported hypertension obtained in a laboratory research setting against blood pressure measured on the same day as the self‐report.

2. METHODS

2.1. Study sample

Multiplex AD families (high‐risk) were identified through the presence of a same‐sex sibling pair, each with a diagnosis of AD. First‐degree family members were also recruited for involvement in the study. Control families were recruited through responses to advertisements and random digit dialing procedures. All participants were asked to refrain from use of alcohol and street drugs for 48 hours before coming to the laboratory. Smokers were not required to stop smoking before the laboratory session. Accordingly, blood pressure measurement was not done within 30 minutes of the last cigarette in most cases. All participants provided consent. The study has ongoing approval from the University of Pittsburgh Institutional Review Board.

2.2. Self‐report questionnaire

A total of 1010 adult participants were administered a health and medication questionnaire by a trained interviewer during an in‐person interview. Positive or negative responses were coded for the presence of several diseases and conditions, including whether or not the individual had been diagnosed with hypertension. Use of medications in the past month was queried, including use of anti‐hypertensive medications. Also, participants were asked to provide the number of cigarettes smoked daily. Smokers were identified as those who reported any daily smoking.

2.3. AD assessment

A structured, psychiatric interview (Diagnostic Interview Schedule [DIS]) ¹³ was administered by Masters level clinicians. This provided a diagnosis of AD by DSM‐III (diagnostic system in place when the study was initiated). Feighner Criteria ¹⁴ for AD were administered as a validity check on the DIS diagnosis and to provide detailed information on problems associated with alcohol use. Four symptom categories were assessed in the Feighner interview: (a) medical consequences; (b) loss of control; (c) legal/social problems; and (d) report of excessive drinking as judged by participant or others. The earliest age the participant had a symptom in each category was determined. The age at first symptom in any of the four categories was considered the age of the first problem of alcohol‐dependent use.

Validity of Clinical Data: The interview data were supplemented with an open‐ended clinical interview by a second clinician in order to provide reliability of the DIS and Feighner Criteria interviews. A best‐estimate diagnosis of AD was made using the information from the DIS, the Feighner Criteria, a second open‐ended interview, and any family history information provided by the participating relatives.

2.4. Assessment of drinking history

All subjects were administered a lifetime drinking instrument ¹⁵ which assessed usual and maximum quantity and frequency of alcohol use over four time periods when drinking habits changed, corresponding to the occurrence of significant life events. Additionally, current drinking was queried to determine the quantity and frequency of alcohol used in the past 7 days and the past 30 days. Also obtained was the date of their last drink of alcohol. These instruments were used in combination with the DIS to obtain a complete drinking history over the lifetime.

2.5. Physiological assessment of blood pressure

Blood pressure was measured in 162 high‐risk subjects from the multiplex for AD families and 121 low‐risk control subjects, evaluated as part of the family studies. Arterial blood pressure was measured in the upper arm in mm Hg with a sphygmomanometer and stethoscope during the test session. Both the self‐report and the blood pressure were obtained on the same day by a trained, MA level clinician, who was blind to presence or absence of an AD diagnosis.

Blood pressure measurement was obtained on three occasions. (a) The first measure was obtained in a resting state after the participant had been seated for at least 15 minutes while informed consent procedures were conducted. Participants were then transferred to a phlebotomy laboratory a few blocks away from our laboratory. (b) After the participant was seated and immediately before blood was drawn, a second measure was obtained. (c) At the conclusion of the phlebotomy while the participant remained seated, a third measure was obtained.

2.6. Statistical analysis

Validity of self‐reported hypertension was determined by comparing the self‐report to the measurement of resting arterial blood pressure in two ways. First, a comprehensive parametric analysis was conducted to identify potential variables that influenced concordance of reported hypertension and measured blood pressure. Analyses were conducted to determine the influence of an AD diagnosis, sex, age, use of anti‐hypertensive medication, and cigarette smoking.

Because the heritability of hypertension has been reported to be between 30% and 60%, ¹⁶ the presence of siblings or parent‐child pairs from the same family was statistically controlled in all parametric data analyses. Ninety‐seven pedigrees were represented in the data (52 from high‐risk families and 45 from low‐risk control families). From the total of 97 pedigrees, 27 contributed only 1 individual, 21 contributed 2 relatives, and 16 pedigrees contributed 3 relatives. Some pedigrees were quite large: 18 contributed 4 relatives, while 15 pedigrees contributed 5 or more relatives. To control for possible effects of larger families contributing a greater portion of outcome to our analysis, the mixed model analyses of variance used family ID as a random effect and the binary phenotype, hypertension (yes/no), was evaluated against measured blood pressure, the dependent variable as a fixed effect.

Self‐reported hypertension was examined by calculating sensitivity and specificity. Sensitivity or the percentage of true positives was calculated by comparing those reporting hypertension to those who measured in the hypertensive range (≥140 mm Hg systolic or ≥90 mm Hg diastolic). ¹⁷ While guidelines and recommendations for hypertension control have been evolving through the years, ¹⁸ the guidelines for hypertension were a blood pressure equal to or greater than 140/90 mm Hg at the time the participants were first studied. Specificity or the percentage of true negatives was calculated for those who reported normal blood pressure among those classified as normotensive. Moreover, positive predictive (rate of individuals having a disorder among all positive test results) and negative predictive (rate of individuals having no disorder among all negative test results) values were computed. Κappa scores were calculated to determine the agreement between self‐report of high blood pressure and measured blood pressure. Values over 0.40 were considered to have moderate agreement. Receiver operating characteristic (ROC) curves were graphed, and the area under the curve (AUC) was computed for the different cohorts. ROC curves allow for the exploration of the sensitivities and specificities by showing sensitivity on the y‐axis and 1‐specificity on the x‐axis. SPSS Statistics, version 20, was used for all analyses.

3. RESULTS

3.1. Demographics

At the time of the assessment, the overall sample age was 40.90 ± 13.4 years: males (41.49 ± 13.6 years) and females (40.56 ± 13.3 years). Because two generations were included in the family study reported here, the ages ranged from 18 to 81 years. Calculations of socioeconomic status (SES) ¹⁹ showed that 52.9% of the sample were in the highest two SES levels (Professional and Technical) with 90.5% of the sample having a minimum of a high school education or GED. Those who were classified as daily smokers (N = 113) were smoking an average of 20.32 ± 9.8 cigarettes daily. While both the AD and non‐AD individuals were drinking for a similar number of years, the AD individuals were drinking an average of 15.7 ± 7.8 years after meeting the first Feighner Criteria problem category for AD. Demographic characteristics of the sample by AD status can be seen in Table 1.

TABLE 1.

Demographic characteristics

	Alcohol dependent N = 87 Mean (SD)	Not dependent N = 196 Mean (SD)
Sex	41 M; 46 F	61 M; 135 F
Age	37.05 (10.3)	42.61 (14.2)
Education	12.64 (1.7)	13.74 (2.6)
Socioeconomic status (SES) a	33.22 (11.25)	41.55 (12.5)
Years since began drinking 1×/mo	20.25 (8.9)	21.24 (12.9) b
Years drinking after first problem c	15.65 (7.8)
Days since last drink	131.33 (382.6)

^aSES, hollingshead four factor index of social status (N = 278).

^b36 cases were abstinent or drinking <1×/mo.

^cYears drinking determined from the age the first Feighner Criteria problem occurred.

3.2. Parametric statistical analysis

3.2.1. Systolic blood pressure

A comprehensive mixed model analysis of variance was conducted to assess the effects of sex, age, diagnosis of AD (positive/negative diagnosis), smoking status (daily smokers/non‐smokers), and use of anti‐hypertensive medication in the past month on the concordance between reported hypertension (yes/no) and systolic blood pressure. Significant effects were seen for sex (F = 19.64, df = 1,275, P < .0001) and age (F = 54.00, df = 1,274, P < .0001). Significant effects for the concordance of measured hypertension and self‐reported hypertension were not found in association with the individual's current smoking status (F = 0.18, df = 1,275, P = ns), lifetime AD diagnosis (F = 0.28, df = 1,273, P = ns), or current use of anti‐hypertensive medications (F = 0.44, df = 1,261, P = ns).

3.2.2. Diastolic blood pressure

Using a comprehensive mixed model analysis of variance, similar results were found for diastolic blood pressure and self‐reported hypertension. Sex influenced the relationship between measured and self‐reported hypertension (F = 12.18, df = 1,269, P < .001) while no effect was found for smoking status (F = 0.29, df = 1,271, P = ns), AD diagnosis (F = 0.58, df = 1,275, P = ns), or use of anti‐hypertensive medications (F = 1.74, df = 1,245, P = ns). Unlike the results obtained for systolic concordance, there was no effect of age on the concordance of self‐reported hypertension and diastolic blood pressure (F = 2.66, df = 1,266, P = ns).

3.3. Sensitivity and specificity—non‐parametric analysis

Non‐parametric tests were then conducted to assess sensitivity and specificity of self‐reported hypertension and measured hypertension status (defined as systolic pressure ≥140 mm Hg or diastolic blood pressure ≥90 mm Hg) using the first measure of resting blood pressure. As each of the three measures of blood pressure were significantly different from the other two, an analysis combining the two resting measures could not be done (post‐consent resting/post‐phlebotomy resting t = 3.74; df = 262; P<.001, post‐consent resting/pre‐phlebotomy t = −5.14; df = 281; P<.001, pre‐phlebotomy/post‐phlebotomy resting t = 9.38; df = 262; P<.001). Only the first measure of blood pressure obtained at rest was analyzed. Concordance of self‐report and measured hypertension status in the full sample of 283 individuals showed a statistically significant association (Pearson r ² = .387, P < .001). Table 2 shows the sensitivity and specificity test statistics using the first resting measure of blood pressure. Among those classified as hypertensive based on measured blood pressure, only 37.2% reported having hypertension while 62.8% reported they were not hypertensive. However, the specificity was quite high at 95.0%, indicating that self‐report can accurately identify individuals who are not hypertensive. The positive predictive value or the probability of being classified as hypertensive upon measurement among those who reported being hypertensive was 57.1%. The negative predictive value or the probability of not being classified as hypertensive by direct measurement among those who reported not being hypertensive was quite high at 89.4%. Prevalence, or the percentage of the sample classified hypertensive, was calculated to be 15.2%. The accuracy, or probably of correctly classifying true positives and true negatives, was 86.2%. The Κ value for the whole sample was 0.376, indicating a somewhat moderate agreement between the self‐report and hypertensive status.

TABLE 2.

Comparison of self‐reported hypertension and measured resting blood pressure

		Measured blood pressure >140/90 mm Hg			Test statistic
	Self‐report	Yes	No	Total	Sensitivity [95% CI]	Specificity [95% CI]	Positive predictive value [95% CI]	Negative predictive value [95% CI]	Prevalence [95% CI]	Kappa [95% CI]
Whole sample	Yes	16	12	28
	No	27	228	255	37.2%	95.0%	57.1%	89.4%	15.2%	0.376
	Total	43	240	283	[23.0‐53.3]	[91.4‐97.4]	[40.5‐72.4]	[87.0‐91.4]	[11.2‐19.9]	[0.22‐0.53]
Male	Yes	9	5	14
	No	14	74	88	39.1%	93.7%	64.3%	84.1%	22.6%	0.381
	Total	23	79	102	[19.7‐61.5]	[85.8‐97.9]	[40.1‐82.9]	[79.1‐88.1]	[14.9‐31.9]	(0.16‐0.60]
Female	Yes	7	7	14
	No	13	154	167	35.0%	95.7%	50.0%	92.2%	11.1%	0.353
	Total	20	161	181	[15.4‐59.2]	[91.3‐98.2]	[28.1‐71.9]	[89.6‐94.2]	[6.9‐16.6]	[0.13‐0.57]
Alcohol	Yes	4	5	9
Dependent	No	7	71	78	36.4%	93.4%	44.4%	91.0%	12.6%	0.323
	Total	11	76	87	[10.9‐69.2]	[85.3‐97.8]	[20.2‐71.7]	[86.6‐94.1]	[6.5‐21.5]	[0.03‐0.62]
Not Alcohol	Yes	12	7	19
Dependent	No	20	157	177	37.5%	95.7%	63.2%	88.7%	16.3%	0.397
	Total	32	164	196	[21.1‐56.3]	[91.4‐98.3]	[42.2‐80.1]	[85.7‐91.1]	[11.4‐22.3]	[0.21‐0.58]
Over age 55	Yes	10	5	15
	No	9	26	35	52.6%	83.9%	66.7%	74.3%	38.0%	0.381
	Total	19	31	50	[28.9‐75.6]	[66.3‐94.6]	[44.6‐83.2]	[63.7‐82.6]	[24.7‐52.8]	[0.12‐0.65]
Under age 55	Yes	6	7	13
	No	18	202	220	25.0%	96.7%	46.2%	91.8%	10.3%	0.272
	Total	24	209	233	[9.8‐46.7]	[93.2‐98.6]	[23.9‐70.1]	[89.9‐93.4]	[6.7‐14.9]	[0.07‐0.47]

3.4. Predictive probabilities

Further analyses were conducted to examine whether the predictive probabilities would differ in the sub‐samples (Table 2). Separate tests were conducted by sex (male/female), individual diagnosis of AD (positive/negative diagnosis), age category (over/under age 55), and by smoking status (daily smokers/non‐smokers). Similar results were seen in separate tests, indicating there were no differences in accuracy of report of hypertension in individuals diagnosed with and without AD or in analyses by sex. AD individuals were, therefore, included in further analyses since differences were not found as a function of AD status. As expected, the highest sensitivity was seen in those over the age of 55 (52.6%), with agreement measured by Κ to be 0.381. Additionally, in this older cohort, 66.7% were correctly classified as having hypertension (positive predictive value). Higher sensitivity (42.9%) was also seen in the non‐smokers with a moderate level of agreement (Κ = 0.420). All of the samples tested had a specificity of 84% or higher, or a low false‐positive rate.

Of the 28 individuals who reported hypertension on the self‐report questionnaire, 25 cases reported using blood pressure medication in the past month. In half of the cases, blood pressure measured in the normal range (N = 12) while the other half measured in the hypertensive range (N = 13).

3.4.1. ROC curves

To evaluate the sensitivity and specificity of hypertension, ROC curves were graphed in the over 55‐year‐old sample where the highest sensitivity (0.526) was found (Figure 1). The AUC is 0.683 (95% CI = 0.523, 0.842; asymptotic significance = 0.03) indicating that self‐report is a reasonably accurate method for revealing hypertension in the older sample.

FIGURE 1.

The receiver operating characteristic curve analysis of hypertension in the study population over age 55 is shown. The x‐axis denotes 1‐specificity, while the y‐axis denotes sensitivity. The area under curve is 0.683 (95% CI = 0.523, 0.842; asymptotic significance = 0.03)

To determine the effect of using 130/80 mm Hg as the cutoff for hypertensive status, sensitivity, specificity, and kappa values were re‐calculated using the lower cutoff. Results comparing the two cutoffs can be seen in Table 3. As expected, sensitivities are decreased in each of the three measures of blood pressure (post‐consent resting, prior to phlebotomy, and post‐phlebotomy) using the more stringent definition of hypertensive status, irrespective of which measure of blood pressure was used.

TABLE 3.

Comparison of test statistics across hypertension definitions

Blood pressure	Hypertension defined as ≥140/90 mm Hg			Hypertension defined as ≥130/80 mm Hg
	Sensitivity %	Specificity %	Kappa	Sensitivity %	Specificity %	Kappa
Resting	37.2	95.0	0.376	25.5	98.4	0.287
Preceding blood draw	37.7	96.5	0.418	23.1	98.3	0.249
Post‐blood draw	36.1	94.2	0.344	28.8	97.4	0.326

Significant changes in blood pressure occur in those with AD during withdrawal with elevations seen in the first 2 weeks. ¹¹ Therefore, data were analyzed separately in AD and non‐AD individuals by the length of time since their last drink of alcohol (Table 4). Similar sensitivities were seen in the AD group who were not drinking for under 2 weeks compared to those who had not been drinking for over 2 weeks. However, in the non‐AD group who had their last drink over 2 weeks prior to testing, sensitivity increased to 50.0% with a strong moderate agreement (Kappa = 0.589).

TABLE 4.

Comparison of resting blood pressure by days abstinent from alcohol use

	Abstinent—less than 2 wk			Abstinent—longer than 2 wk
	Sensitivity %	Specificity %	Kappa	Sensitivity %	Specificity %	Kappa
Alcohol dependent	37.5	91.2	0.317	33.3	94.7	0.281
Not alcohol dependent	27.3	100.0	0.396	50.0	98.4	0.589

4. DISCUSSION

It can be concluded that the self‐report is highly specific in identifying individuals without hypertension. However, it is only moderately sensitive, correctly identifying individuals with hypertension in approximately 37% of the entire sample. Conducting separate analyses by sex and diagnosis yielded similar results for sensitivity while differentiation was seen in the age and smoking status groups. As expected, the highest sensitivities were seen in the over age 55 group (53%), as prevalence of hypertension increases with age. Interestingly, one of the higher levels of sensitivity (43%) was seen in individuals who were not daily cigarette smokers, though agreement was in the moderate range (Κ = 0.420). It is possible that current smoking exerted a poorer recognition of health status. Cigarette smoking has been associated with impaired memory performance. ²⁰ , ²¹ Therefore, it is possible that cigarette smoking may be one factor contributing to impaired memory for interactions with medical professionals where they may have been told they had hypertension.

This is the first study to examine the validity of self‐reported hypertension against measured blood pressure in alcohol‐dependent individuals. The results are notable in finding that AD individuals are as accurate in their report of hypertensive status as those without an AD diagnosis. However, all participants, irrespective of having an AD diagnosis, showed deficits in accurately reporting hypertension. Moderate agreement was found in the groups tested here, with the exception of individuals who were under age 55 or who were currently smoking. These groups showed poor agreement between the self‐report questionnaire and measured arterial blood pressure (all kappa values were <0.30).

The low sensitivity reported here for the entire sample is consistent with the report of Tenkorang et al ²² who studied a cohort over the age of 18 and found low sensitivity in data from 4 out of the 5 countries studied (China, India, Russia, South Africa, and Ghana). That report suggested the importance of considering sociodemographic factors (access to health care, sex, education, and wealth) when interpreting self‐report data on hypertension. Goldman et al ⁴ reported that those with higher levels of education were more likely to acknowledge their hypertensive condition than uneducated individuals.

The present sample was well educated with 90.5% having a minimum of a high school education or GED, suggesting they were knowledgeable of their hypertensive status. Also, the SES was calculated to be in the Professional and Technical categories for more than half of the sample. While SES and number of years of education were lower in the AD individuals, similar accuracy of reporting was found in the AD and non‐AD groups.

Measurement in a laboratory setting provided estimates of usual blood pressure that showed 43 of 283 individuals (15.2%) in the hypertensive range. This would be expected based on the average age of the sample (mean age = 41.9 ± 13.4 years). Population estimates have been reported to be 7.5% for ages 18‐39 and 33.2% for those 40‐59. ²³ Although three blood pressure measurements in a row were not done due to study design, the accuracy of measured blood pressure is assured by the overall rates seen in the sample. Discordance between measured hypertension and self‐report is most likely the result of lack of knowledge on the part of the participant regarding their hypertensive status.

4.1. Limitations

The current study has some limitations. The assessment was done in the context of an ongoing study of factors associated with the etiology of alcohol use disorders that was based in a pedigree design. A larger scale study that is population‐based might reveal differing results. However, a large representative cohort study in Australia showed similar results with specificity of 98.8% but sensitivity of only 49.0%. ²⁴ These results are similar to the present study in finding overall specificity of 95% but sensitivity of 37%. While a large community sample may be more representative, the benefit of the present sample is that it is highly exposed to alcohol. Additionally, while 25 participants reported the use of anti‐hypertensive medication, regularity of use or accuracy of dosage could not be assessed. By including those cases whose blood pressure was controlled with medication, results were biased toward lower sensitivities.

4.2. Positive features

Although there are limitations, several positive features of our analyses should be mentioned. First, the study provides the first report of the accuracy of self‐reported hypertension in alcohol‐dependent individuals who might be suspected of having less accurate reports. Second, the study was able to account for a number of background variables that might alter the conclusions such as age, SES, and sex. Third, while physicians with an ongoing relationship with a patient may know whether hypertension is usual or unusual for a given patient, others may be unaware of the past history of the individual and cannot use this information to make an informed diagnosis. The present findings lend support to the benefit of obtaining a history of self‐reported hypertension, especially in situations where medical personnel do not have an ongoing relationship with the patient such as in an emergency setting, to determine whether the hypertension seen is out of the ordinary or part of a long‐standing problem. This benefit appears to be present even in alcohol‐dependent individuals.

5. CONCLUSIONS

Absence of hypertension can be accurately determined with self‐report data in those without hypertension. However, a significant proportion of those with measured hypertension report an absence of hypertension. Overall, our results suggest that self‐report of hypertension is reasonably accurate but requires knowledge of the population characteristics that may influence accuracy of report.

CONFLICT OF INTEREST

The author(s) declare that there is no conflict of interest.

AUTHOR CONTRIBUTIONS

Shirley Y. Hill is responsible for study conception, design, and interpretation of data and revision of the initial draft of the manuscript. Jeannette Wellman completed the statistical analysis of the data and drafting of the manuscript. Brian Holmes assisted in the drafting of the manuscript. All of the authors reviewed and approved of the final manuscript.

Wellman JL, Holmes B, Hill SY. Accuracy of self‐reported hypertension: Effect of age, gender, and history of alcohol dependence. J Clin Hypertens. 2020;22:842–849. 10.1111/jch.13854