Key result
Correction for cuff size measurement error reduced the apparent decline in US hypertension prevalence between 1976-80 and 1999-02 to -4.8 percentage points (95% CI -7.5 to -2.0).
Why the study?
Between 1976-80 and 1999-02, US obesity prevalence doubled while hypertension declined by 25 percent, prompting evaluation of the effects of measurement error and risk factors on this difference.
Observational
Yes
Measurement errors, specifically zero end digit preference and inappropriate cuff size, contributed to the apparent decline in hypertension prevalence between 1976-80 and 1999-02, which was not explained by changes in BMI or other demographics.
Answers are grounded in this paper and cite their sources.
Measurement error correction attenuates observed US hypertension prevalence declines; leaves open the true magnitude of secular trends.
The National Health and Nutrition Examination Survey (NHANES) is conducted to monitor the health and nutrition status of the United States (US) population. Results from the NHANES indicate that between 1976-80 and 1999-02 the prevalence of obesity doubled in the US while hypertension declined by 25 percent. Much attention has been focused on obesity as a risk factor for hypertension and on programs to reduce obesity prevalence in the population. The objective of the first paper was to estimate the effect of two sources of measurement error on the estimated prevalences of hypertension in the two surveys. In the second paper I estimated the effect of measurement error and selected risk factors on the difference in hypertension prevalence between 1976-80 and 1999-02. Blood pressure measurements were adjusted for use of inappropriate cuff size using previously published correction equations. I modeled the effect of zero end digit preference using different sets of assumptions. The adjustment for cuff size changed the age-adjusted prevalence of hypertension in 1976-80 from 34 to 31% and in 1999-02 from 25 to 26%. After the adjustment for cuff size, adjustment for end digit preference reduced the prevalence in 1976-80 slightly more from 31 to 30% and in 1999-02 the prevalence was unchanged at 26%. The prevalence difference was -8.6 percentage points (-11.4, -5.8) after standardization for age alone compared with -8.0 percentage points (-10.8, -5.1) after standardization for age, body mass index (BMI), sex, race-ethnicity, smoking, and education. This should not be interpreted as meaning that none of these factors affected the difference. The difference decreased to -4.8 percentage points (-7.5, -2.0) after correction for cuff size and increased to -14.7 percentage points (-17.6, -11.7) after standardization for BMI and age. Using observed blood pressure prevalence differences were greater in overweight and obese persons than in normal weight persons but were similar after adjustment for measurement error. Zero end digit preference and lack of adjustment for cuff size contributed to the apparent decline in hypertension prevalence. Standardization for BMI, age, sex, race-ethnicity, education, and smoking did not explain the decline in hypertension seen between 1976-80 and 1999-02.
No takes yet. Share an insight, caveat, or question.
Jacqueline D. Wright (2019) conducted an observational in Hypertension. Adjustment for measurement error (cuff size and zero end digit preference) vs. Unadjusted measurements was evaluated on Difference in hypertension prevalence between 1976-80 and 1999-02 after correction for cuff size (Difference -4.8 percentage points, 95% CI -7.5 to -2.0). Correction for cuff size measurement error reduced the apparent decline in US hypertension prevalence between 1976-80 and 1999-02 to -4.8 percentage points (95% CI -7.5 to -2.0).
Synapse has enriched 10 related papers. Consider them for comparative context: