The effects of a hypertension diagnosis on health behaviors: A two‐dimensional regression discontinuity analysis

Abstract

This paper explores how a diagnosis of hypertension might affect a person's health‐related behaviors. The analysis uses a two‐dimensional regression discontinuity design because hypertension is diagnosed when a person's systolic or diastolic blood pressure (SBP or DBP) surpasses a pre‐established threshold. We find that those closely above the SBP threshold significantly adjusted their lifestyle, such as reducing daily fat intake and quitting smoking, while those just surpassing the DBP cutoff did not. Further mechanism analysis suggests that the possibility of constraints, rather than education and income gradients, does more to explain the disparate behaviors of subjects near the SBP and DBP thresholds. Those around the DBP threshold generally have tighter work schedules and undertake more competitive jobs, which hinder them from improving their lifestyle. Overall, our findings complement the existing literature by posing a new perspective for understanding people's potential reluctance to adjust their behavior.

The effect of population-based blood pressure screening on long-term cardiometabolic morbidity and mortality in Germany: A regression discontinuity analysis

“…The density plots for diastolic BP, systolic BP, and the BPS are reported in Fig 1 . The graphical inspection of these plots confirmed the absence of manipulation around the respective cutoffs, since the density curve runs smoothly through the cutoff (if a manipulation had occurred, we would have seen a jump in the distribution of these variables close to the cutoff). The same result applies to the density plots for the reduced sample for the analysis of the secondary 18) 135 ( 8) 130 ( 6) 140 (7) Diastolic BP (mm Hg, SD) 80 ( 11) 84 ( 7) 82 ( 6) 87 (7) N events 275 (5%) 122 (4%) 153 (6%) BMIAU : AbbreviationlistsinTables1and2havebeenupdated:Pleasecheckandverifythatalladdedentriesarecorrectlyabbreviated: , body mass index; BP, blood pressure; CI, confidence interval; CVD, cardiovascular disease; SD, standard deviation. � Within the optimal respective bandwidths.…”

“…The current evidence on the effect of hypertension diagnosis differs from the evidence on the effect of the other abovementioned diseases because of the mixed nature of the results hitherto reported. Studies using an RD design in samples from China and South Africa have shown an significant decrease in BP 2 years after diagnosis [ 4 , 5 ] and a significant improvement in diet 3 to 4 years after diagnosis [ 6 , 7 ]. Conversely, studies on the US population have showed no improvement in dietary patterns after a diagnosis of hypertension within 2 years of diagnosis [ 8 , 9 ].…”

“…Population-based screening is typically a "light touch" intervention, in which individuals are informed of their hypertension status and encouraged to consult a physician for treatment. However, the available evidence has shown mixed results regarding the effect of simply communicating a new diagnosis on medium-term behavioral changes and health improvements, both in the case of hypertension [4][5][6][7][8][9] and other diseases [10][11][12][13][14][15][16][17][18]. Studies in China and South Africa found that hypertension screening and counseling reduced BP and improved lifestyle outcomes at 2 years [4][5][6][7].…”

“…However, the available evidence has shown mixed results regarding the effect of simply communicating a new diagnosis on medium-term behavioral changes and health improvements, both in the case of hypertension [4][5][6][7][8][9] and other diseases [10][11][12][13][14][15][16][17][18]. Studies in China and South Africa found that hypertension screening and counseling reduced BP and improved lifestyle outcomes at 2 years [4][5][6][7]. However, studies conducted in the United States showed only modest changes in dietary patterns up to 2 years after hypertension diagnosis [8,9].…”

“…This study evaluates the effect of a population-based BP screening intervention on longterm CVD morbidity and mortality in Germany. Like other studies in this context [4][5][6][7], we used a quasi-experimental approach to evaluate the causal effect of the screening using data from a population-based epidemiological study. These methods offer the possibility of deriving estimates of causal impact from observational data in substantially less time and using substantially fewer resources [19], in a context where experimental studies would not be feasible due to ethical and economic reasons [20,21].…”