Clinical outcome of hypertension management

Abstract

Objective: This study aimed to investigate the association between inadequately controlled hypertension, medication use pattern and the incidence of cardiovascular events (CVEs) over 5 years. The cost of hospitalisation associated with inadequately controlled hypertension was estimated.

Methods: This was a retrospective observational study of patients who were diagnosed with hypertension with at least one blood pressure reading recorded at baseline (year 2000), reviewed from 2001–2005. Primary endpoints were the extent, duration and prevalence of blood pressure deviation from treatment goals. Secondary endpoints were incidence of CVEs and the number of days of hospitalisation associated with inadequately controlled hypertension.

Results: In total, the medical records of 210 patients were reviewed. Statistically significant positive associations were found between inadequately controlled systolic blood pressure with the incidence of CVEs (p=0.018) and increased cardiovascular drugs (p<0.001), non-cardiovascular drugs (p=0.019) and total drug usage (p<0.001). Similar observations were observed between inadequately controlled diastolic blood pressure and increased cardiovascular (p<0.05) and total (p=0.007) drug usage. Patients with uncontrolled blood pressure at all times during the study period were associated with a higher incidence of CVEs (p=0.026). The mean cost of hospitalisation due to CVEs was estimated to be HK$42,584.0 ± 36,670.0 (US$1 = HK$7.8) and it accounted for 3.3% of the total healthcare expenditure during 2005.

Conclusion: Inadequately controlled blood pressure is positively associated with an increased incidence of CVEs and polypharmacy in Hong Kong Chinese hypertensive patients.

• hypertension
• cardiovascular events
• association
• cost

Introduction

Hypertension is one of the most important modifiable risk factors for cardiovascular disease (CVD) and renal disease globally1–3, especially in the Asian population4,5. In the US, hypertension is the most common primary diagnosis, with 35 million office visits in 20006. Despite efforts made by Healthy People 2010, which is managed by the US Department of Health and Human Services, to achieve a goal of 50% control rate for the entire hypertensive population, the current control rate is still far below such a goal. According to the guidelines established by the US Joint National Committee on Prevention, Detection, Evaluation and Treatment of High Blood Pressure (JNC-7), in the years 1999–2000 approximately 30% of hypertensive patients were still unaware of their diagnosis, with a control rate of only approximately 34%7. In Hong Kong, according to the Population Health Survey conducted by the Department of Health in 2003/2004, 27.2% of the study population had hypertension, but less than one-half of them were diagnosed with hypertension by health professionals8. Among those who were diagnosed with hypertension, the control rate was no more than 50%9. Improperly controlled hypertension increases the chance of CVDs, including myocardial infarction (MI), chronic heart failure (CHF), cerebrovascular disease (CBVD) and renal diseases7. In clinical trials, antihypertensive therapy has been associated with 35–40% mean reduction in stroke, 20–25% reduction in MI and >50% reduction in heart failure10. Failure to control blood pressure could lead to undesirable clinical outcomes and increase the economic burden on the health system. In 1998, it was estimated that US$108.8 billion in healthcare spending was attributed to hypertension, 12.6% of total national spending that could be allocated to diagnoses including US$22.8 billion for hypertension, US$29.7 billion for cardiovascular complications and US$56.4 billion for other diagnoses11. In Hong Kong, the cost of acute MI, one of the complications of uncontrolled hypertension, was US$45 million per year12.

Hypertension has been defined as blood pressure ≥140/90 based on the World Health Organization–International Society of Hypertension13 and the JNC-77. For patients with diabetes mellitus or renal diseases, the blood pressure goal is ≤130/80 mmHg for tighter blood pressure control14,15. According to the British Hypertension Society16, the main determinant of benefit from blood pressure lowering is the achieved blood pressure, rather than the choice of therapy. The objectives of this study were to investigate the prevalence, extent and duration to which the systolic blood pressure (SBP) and diastolic blood pressure (DBP) of patients with hypertension deviates from the predefined treatment goals and the pattern of medication use of hypertensive patients with reference to JNC-77, as well as to observe the association of blood pressure control with cardiovascular events (CVEs) and cerebrovascular events (CBVEs) and to estimate the average cost for inpatient care to manage CVEs and CBVEs related to uncontrolled hypertension. The study hypothesis was that the higher and longer the extent and duration blood pressures deviated from the predefined treatment goals, the higher the incidence rates of CVEs and CBVEs.

Methodology

The study adopted a retrospective observational design. All hypertensive patients seen in the hypertensive clinic, which was the primary care treatment centre in the Prince of Wales Hospital (PWH), during the study period of January 2001 to December 2005 were included. The hypertension clinic is a specialised outpatient clinic (SOPC) that all patients are referred to by physicians in the generalised outpatient clinic (GOPC). Care prior to the SOPC was carried out in the GOPC. Care under the SOPC is more focused on the target disease that was not controlled in the GOPC. The study site is an acute care, 1200-bed, public teaching hospital providing medical services to patients in the New Territories East region of Hong Kong and is one of the major cardiovascular centres in Hong Kong.

Patients and classifications

Patients were eligible if they were >18 years of age at baseline and had a primary diagnosis of hypertension with SBP ≥140 mmHg and/or DBP ≥90 mmHg with or without hypertensive treatment. The blood pressure goal for each patient was set based on the JNC-7 recommendations (i.e. blood pressure should be ≤140/90 mmHg for patients without diabetes or blood pressure should be ≤130/80 mmHg for diabetic patients). Subjects were classified into different groups based on their SBP or DBP as follows:

• Patients who had their blood pressure well controlled at or below their respective treatment goals (100%) throughout the 5-year study period based on case note review (Group 1S (SBP) or Group 1D (DBP));

• Patients who had partially reached their respective blood pressure treatment goals (1–99%) from the case note review of the study period (Group 2S (SBP) or Group 2D (DBP)); and

• Patients who did not reach their blood pressure treatment goal (0%) throughout the entire study period (Group 3S (SBP) or Group 3D (DBP)). Patients must have both SBP and DBP at the target goal in order to be considered blood pressure goal attained.

Patients were excluded if they had secondary hypertension such as hypertension due to Cushing's syndrome, chronic renal diseases at baseline, primary hyperaldosteronism, pregnancy, sleep apnoea, drug-induced or drug-related hypertension and pheochromocytoma; history of CVD or CBVD at baseline; and no clinical blood pressure measurement available for the 12-month period immediately preceding the study period. Medical records of the patients who met the inclusion criteria were reviewed.

Medical information

Demographics, clinical characteristics, drug therapies, laboratory tests, diagnostic procedures (including echocardiography, angiography, stress test, magnetic resonance imaging and computed tomography), clinic visits, emergency department admissions and hospitalisations were identified through a review of patients' medical and pharmacy records. The following data were collected at time points 1, 2, 3, 4 and 5 years from baseline:

• Average SBP and DBP from the first and last two office visits within the 12-month period during the corresponding time point;

• Any addition/removal/alteration of antihypertensive therapy;

• Incidence of CVEs including fatal or non-fatal MI, CHF, angina pectoris, percutaneous coronary intervention or other coronary procedures, CBVEs, mortality from all cardiovascular or cerebrovascular causes and total mortality; and

• Management of CVEs in hospital including laboratory tests, diagnostic and therapeutic procedures, radiology services, drug therapy and duration of hospitalisation.

Outcome measurement

The primary outcome measurements were the prevalence, duration and extent of SBP and DBP deviation from the treatment goals at baseline and throughout the study period. The secondary outcome measurements included the number of incidents of defined and undefined CVEs requiring coronary procedures and CBVEs, the number of coronary procedures performed, mortality from all cardiovascular or cerebrovascular causes, and the total mortality throughout the study period. The number of hospitalisations and the medication prescribed for each hospitalised patients were documented for each hospitalisation due to CVEs or CBVEs.

The cost of hospitalisation was calculated as the sum of the charges for inpatient maintenance, laboratory tests, radiology services, diagnostic and therapeutic procedures and drug therapy. The unit cost of drug and other service charges was based on the PWH drug formulary17 in the year 2005 and the Hong Kong Government Gazette 200318, respectively. The base year for the cost calculations was 2005 and a 3% discount rate was used in each year prior to 2005. The direct medical cost was projected to the entire Hong Kong population based on the prevalence rate of uncontrolled hypertensive patients who would require hospital management in Hong Kong. The total cost of hospital management of hypertensive patients in Hong Kong was then estimated by multiplying the cost calculated in this study by the total number of hypertensive patients in Hong Kong using the prevalence rate and assuming that among those hypertensive patients no more than 50% would reach their target blood pressure goals.

Statistical analysis

The sample size was estimated by the McNemar's test19. A sample size of 170 would give a power of 80% and a significance level of 95% to detect a true difference of the study hypothesis. It was assumed that uncontrolled patients had 25% more CVEs. The continuous variables are described as mean and standard deviation (sd). The average SBP or DBP control over the 5-year follow-up was expressed as a percentage rate. Univariate analysis was performed using χ² test, Fisher's exact test, and analysis of variance and Kruskal–Wallis tests where appropriate. Cox proportional hazard regression analysis was performed to estimate the hazard ratio with corresponding 95% confidence intervals (CIs) for comparison of time to first CVE in the three different groups of SBP and DBP. The grouping of SBP and DBP is based on the percentage of failure to achieve blood pressure goals from the baseline to the event year. Poorly controlled groups for both SBP and DBP as well as female sex were selected as the reference group. The statistical analysis was performed in an intention-to-treat (ITT) population with the last observation carried forward. The ITT population had a baseline measurement and at least one post-baseline measurement for SBP and DBP variables. The primary CVE variable was a composite of fatal MI, non-fatal MI, congestive heart failure, angina, stroke, coronary procedures, other CVEs including arrhythmia and cerebrovascular death. The single-term models for investigation of the relationship between either SBP groups or DBP groups and the event were initially constructed. In a multivariate model for both blood pressure groups, adjustments were made for the age group at baseline (≤50, 51–60, 61–70 and >70 years) and hypercholesterolaemia (Yes/No). The dual-term model for investigation of the relationship between both blood pressure components and the event status with the mentioned factors was assessed simultaneously. Examination of log-minus-log survival plots and testing age group × time interaction term showed no apparent violation of the underlying assumption of proportional hazards on which the Cox regression model is predicated. A two-sided p-value of 0.05 was considered statistically significant. Cost of hospitalisation was expressed as mean ± sd and median with interquartile range between the 25th and 75th percentiles. All statistical analyses were performed using SPSS package version 14 (SPSS Inc., Chicago, IL).

Results

All patients who were seen in the hypertension clinic of PWH at baseline year 2000 were screened. A total of 1,331 patients were identified. Patient recruitment is summarised in Figure 1. In total, 210 patients met the inclusion criteria and their medical charts were reviewed. The demographic data for the study population is summarised in Table 1. The study population had an average of 6.3 years of hypertension history at baseline. The average SBP/DBP values at baseline and at the end of the study period were 149.3/82.7 mmHg and 137.3/78.7 mmHg, respectively. The total incidence of CVEs in the study population was 16.7% for the 5-year study period. The average total number of cardiovascular drugs taken per person was 4.3 and the average total number of non-cardiovascular drugs taken per person was 1.2. The average total number of drugs taken per person in 5 years was 5.4. The percentages of patients in each group of cardiovascular medications are summarised in Table 1. The top seven classes of drugs prescribed for the study population for management of hypertension and CVDs were β-blockers (70%), calcium channel blockers (68%), angiotensin-converting enzyme inhibitors (50%), statins (28%), indapamide (20%), diuretic combinations (20%) and thiazide diuretics (20%).

Figure 1.  Summary of patient recruitment.

Table 1.  Demographic summary of study subjects.

Characteristic
No. of subjects	210
Male	94 (44.8%)
Female	116
Baseline age (years) (mean ± sd)	53.4 ± 12.9
Positive family history of hypertension or CVEs (%)	28.6
Positive smoking history (%)	10.5
Positive alcohol history (%)	12.9
Length of diagnosed hypertension at baseline (years) (mean ± sd)	6.3 ± 5.7
Diagnosed with diabetes (%)	19.0
Diagnosed with obesity (%)	35.2
Diagnosed with renal disease (%)	7.1
Diagnosed with dyslipidaemia (%)	33.8
Attending compliance clinic (%)	22.4
Average SBP at baseline (year 2000) (mmHg) (mean ± sd)	149.3 ± 15.9
Average DBP at baseline (year 2000) (mmHg) (mean ± sd)	82.7 ± 10.3
Drug class/agent (n (%))
b-blockers	146 (69.5)
Calcium channel blockers	142 (67.6)
Angiotensin-converting enzyme inhibitors	106 (50.5)
Statins	59 (28.1)
Indapamide	42 (20.0)
Diuretic combinations	42 (20.0)
Thiazides	42 (20.0)
Aspirin	37 (17.6)
Methyldopa	21 (10.0)
a-blocker	21 (10.0)
Fibrates	20 (9.5)
ARBs	18 (8.6)
Sublingual nitroglycerin	12 (5.7)
Long-acting nitrates	9 (4.3)
Loop diuretics	7 (3.3)
Warfarin	6 (2.9)
Antiarrhythmics	6 (2.9)
ARBs + diuretic combinations	5 (2.4)
Digoxin	4 (1.9)
Trimetazidine	1 (0.5)

sd, standard deviation; CVE, cardiovascular event; SBP, systolic blood pressure; DBP, diastolic blood pressure; ARB, angiotensin receptor blocker.

There was a statistically significant difference between groups for age, average SBP and DBP at baseline and at the end of study, the percentage of times SBP and DBP were out of control, and the average SBP and DBP deviations from treatment goals (Table 2). The clinical characteristics of the study population categorised according to SBP and DBP control are summarised in Tables 3 and 4, respectively. Cox regression analysis was performed to assess the development of CVEs. Compared with those with best SBP control (Group 1S), Group 3S was independently associated with the development of CVEs (relative risk (RR) = 4.38, 95% CI 1.29–14.86). Subjects in Group 2S also had a higher risk of developing CVEs but this was not statistically significant (RR = 1.27, 95% CI 0.36–4.41).

Table 2.  Blood pressure characteristics of three blood pressure-controlled groups.

Characteristic	Total	Well controlled	Partially controlled	Poorly controlled	p-value
No. of cases for SBP	210	24	133	53
Average SBP at baseline (mmHg)	149.26 ± 15.924	129.44 ± 7.415	149.03 ± 14.358	158.81 ± 14.038	<0.0001*
Average SBP from 2001–2005 deviation from goal (mmHg)	3.75 ± 12.386	–13.31 ± 7.190	1.47 ± 8.680	17.20 ± 8.394	<0.0001^†
% SBP out of control	57.79 ± 33.951	0 ± 0	51.4 ± 21.480	100 ± 0	<0.0001*
No. of cases for DBP	210	99	102	9
Average DBP at baseline (mmHg)	82.67 ± 10.344	77.05 ± 7.957	87.47 ± 9.894	90.28 ± 5.826	<0.0001*
Average DBP from 2001–2005 deviation from goal (mmHg)	–7.50 ± 9.096	–14.66 ± 6.433	–2.04 ± 5.269	7.67 ± 5.314	<0.0001^†
% DBP out of control	25.52 ± 31.06	0 ± 0	43.7 ± 23.0	100 ± 0	<0.0001^†

SBP, systolic blood pressure; DBP, diastolic blood pressure; ANOVA, analysis of variance.

* One-way ANOVA using log SBP at baseline.

^† One-way ANOVA.

^† Kruskal–Wallis test.

Table 3.  Clinical characteristics of study population categorised according to SBP control.

	n (N=210)	SBP control			p-value
		Well	Partially	Poorly
		controlled (1S)	controlled (2S)	controlled (3S)
Baseline age (years)	210	52.08 ± 12.420	51.83 ± 12.846	54.28 ± 12.608	0.01*
Age group	210	24	133	53
≤50	87	11	61	15	0.046^†
51–60	56	6	39	11
61–70	49	5	23	21
≥71	18	2	10	6
Gender	210	24	133	53
Female	116	12	73	31	0.779^†
Male	94	12	60	22
Alcohol	187	24	120	43
Current or regular	27	3	18	6	0.630^†
Former	9	1	4	4
Never	151	20	98	33
Smoking	197	24	125	48
Current	22	3	14	5	0.427^†
Former	23	1	13	9
Never	152	20	98	34
History of baseline hypertension	210	24	133	53
Yes	200	23	127	50	0.936^†
No	10	1	6	3
History of diabetes	210	24	133	53
Yes	40	2	24	14	0.154^†
No	170	22	109	39
History of hypercholesterolaemia	209	24	132	53
Yes	66	5	41	20	0.328^†
No	143	19	91	33
History of renal disease	209	24	132	53
Yes	15	3	4	8	0.0009^†
No	194	21	128	45
History of obesity	210	24	133	53
Yes	74	6	47	21	0.461^†
No	136	18	86	32
History of dyslipidaemia	210	24	133	53
Yes	71	6	43	22	0.306^†
No	139	18	90	31
Attendance of compliance clinic	210	24	133	53
Yes	56	4	37	15	0.499^†
No	154	20	96	38

SBP, systolic blood pressure.

* Kruskal–Wallis test.

^† Fisher's exact test or χ² test where appropriate.

Table 4.  Clinical characteristics of study population categorised according to DBP control.

	n (N=210)	DBP control			p-value
		Well	Partially	Poorly
		controlled (1D)	controlled (2D)	(3D)
Baseline age (years)	210	56.56 ± 12.364	50.96 ± 12.997	45.22 ± 9.602	0.016*
Age group	210	99	102	9
≤50	87	28	52	7	0.009^†
51–60	56	30	24	2
61–70	49	30	19	0
≥71	18	11	7	0
Gender	210	99	102	9
Female	116	66	48	2	0.003^†
Male	94	33	54	7
Alcohol†	187	24	120	43
Current or regular	27	18	6	3	0.630^†
Former	9	4	4	1
Smoking†	197	94	94	9
Current	22	7	13	2	0.331^†
Former	23	10	11	2
History of baseline hypertension	210	99	102	9
Yes	200	95	97	8	0.632^†
No	10	4	5	1
History of diabetes	210	99	102	9
Yes	40	10	27	3	0.007^†
No	170	89	75	6
History of hypercholesterolaemia	209	98	102	9
Yes	66	33	28	5	0.183^†
No	143	65	74	4
History of renal disease	209	99	101	9
Yes	15	6	7	2	0.197^†
No	194	93	94	7
History of obesity	210	99	102	9
Yes	74	25	42	7	0.001^†
No	136	74	60	2
History of dyslipidaemia	210	99	102	9
Yes	71	37	29	5	0.151^†
No	139	62	73	4
Attendance of compliance clinic	210	99	102	9
Yes	56	30	22	4	0.176^†
No	154	69	80	5

DBP, diastolic blood pressure; ANOVA, analysis of variance.

* One-way analysis of variance (ANOVA).

^† Fisher's exact test or χ² test where appropriate.

^† The summation did not add up to total sample size in each category due to missing data in the medical chart.

On the other hand, compared with those with best DBP control (Group 1D), Group 2D was independently associated with a lower risk of developing CVEs (RR = 0.43, 95% CI 0.20–0.91). Subjects in Group 3D did not show any significant association with the development of CVEs (RR = 2.01, 95% CI 0.69–5.86).

A total of 48 defined and undefined CVEs requiring coronary procedures were recorded for 35 patients over the 5-year study period, with 43% and 21% of the events being anginal attacks and coronary procedures, respectively. Six patients (17%) had documented stroke during the study period. Other CVEs included CHF (8%), arrhythmia (8%) and non-fatal MI (4%). There was one non-cardiovascular death in the study population. The distribution of the incidence of CVEs according to blood pressure control groups is summarised in Table 5.

Table 5.  Distribution of incidence of CVEs according to blood pressure control group.

First time CVEs*	Total	Well controlled	Partially controlled	Poorly controlled	p-value
SBP groups	210	24	133	53
Yes	34	3	15	16	0.006
No	176	21	118	37
DBP groups	210	99	102	9
Yes	34	21	11	2	0.118
No	176	78	91	7

CVE, cardiovascular event; SBP, systolic blood pressure; DBP, diastolic blood pressure.

* χ² test.

The cumulative survival curves by SBP and DBP over 5 years are shown in Figure 2. The hazard ratios by the three different blood pressure control groups are summarised in Table 6. The multivariate model was adjusted by age group (≤50, 51–60, 61–70 and >70 years) and hypercholesterolaemia both in the single-term and dual-term models.

Table 6.  HR by blood pressure control group.

	No. of cases	Coefficient	HR (95% CI)
Single-term model
Well controlled
DBP	99	–0.137	0.872 (0.204–3.72)
DBP*	98	–0.500	0.606 (0.129–2.853)
SBP	24	–0.962	0.382 (0.111–1.312)
SBP*	24	–0.701	0.496 (0.142–1.733)
Partially controlled
DBP	102	–0.860	0.423 (0.094–1.909)
DBP*	102	–1.082	0.339 (0.071–1.623)
SBP	133	–1.108	0.330 (0.163–0.668)^†
SBP*	132	–0.910	0.420 (0.194–0.835)^†
Poorly controlled
DBP	9		1.00 (ref.)
DBP*	9		1.00 (ref.)
SBP	53		1.00 (ref.)^†
SBP*	53		1.00 (ref.)^†
Dual-term model^¶
Well controlled
DBP	99	0.564	1.758 (0.394–7.834)
DBP*	98	0.160	1.174 (0.224–6.163)
SBP	24	–1.432	0.239 (0.067–0.856)^†
SBP*	24	–1.082	0.339 (0.090–1.276)
Partially controlled
DBP	102	–0.484	0.616 (0.135–2.807)
DBP*	102	–0.684	0.504 (0.102–2.496)
SBP	133	–1.336	0.263 (0.126–0.548)^$
SBP*	112	–1.085	0.338 (0.152–0.751)^†
Poorly controlled
DBP	9		1.00 (ref.)^†
DBP*	9		1.00 (ref.)
SBP	53		1.00 (ref.)^†
SBP*	53		1.00 (ref.)^†

HR, hazard ratio; DBP, diastolic blood pressure; SBP, systolic blood pressure, CI, confidence interval; ref., referent.

* Multivariate model was adjusted by age group (≤50, 51–60, 61–70 and >70 years) and hypercholesterolaemia.

^†p<0.01.

^† p<0.05.

^$ p<0.0001.

^¶ Dual-term model included two blood pressure components to asses the relationship simultaneously.

The cost estimation of the direct expenditure on cardiovascular medical management in hospital included 12 of the 35 patients who experienced CVEs that required hospitalisation. The median cost was HK$32,179 (25–75th percentile range, HK$14,132.8–62,791.8) (US$1 = HK$7.8). The mean annual direct expenditure for hypertensive patients with CVEs that required hospital management was HK$42,584.0 ± 36,670.0. The estimated hospitalisation cost for a single patient ranged from HK$8,177.0 to HK$132,186.0. With a population of 6.81 million in the year 200520 and with 27.2% of the population estimated to have hypertension8, the control rate of hypertensive patients was no more than 50%9. In light of the established association between poorly controlled hypertension and CVD, CBVD and renal disease, inadequately controlled hypertension represents a heavy financial burden to society.

Figure 2.  Cumulative survival curves by (A) SBP and (B) DBP controlled groups: well controlled, partially controlled and poorly controlled groups.

When projecting the study results to the Hong Kong population in 2005, it was estimated that approximately 21,169 hypertensive patients with CVEs would need hospitalisation. Total incremental annual direct expenditure for cost of hospitalisation is estimated to be more than HK$901 million after adjusting for demographics, obesity, diabetes, dyslipidaemia, renal disease and attendance of the compliance clinic as co-varieties. This amount constituted more than 3.0% of government subvention for the Hospital Authority of Hong Kong in the year 2005.

Discussion

The findings of the current study were comparable with other overseas prospective clinical trials including the Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT)21, the Australian National Blood Pressure (ANBP-2) study22 and the Blood Pressure Lowering Arm of the Anglo-Scandinavian Cardiac Outcomes Trial (ASCOT-BPLA)23. The study population was considerably younger and comprised more females than the other three clinical trials. The mean baseline blood pressure measurement was comparable with that of the ALLHAT study but considerably lower than those of the other two studies. Fewer subjects were smokers or drinkers in the current study. Aspirin was prescribed for one-fifth of the current study population and the ASCOT-BPLA study, whilst 36% of subjects in the ALLHAT trial were prescribed aspirin. In addition, more subjects in the current study were prescribed lipid-lowering drugs than in the other clinical trials (28.1 vs. 10–13%). This may explain the lower CVE rate in the current study (16.7%) than the ALLHAT study (30.9–33.3%). However, the current study may contribute to underestimation of CVEs since many patients were excluded at baseline owing to loss to follow-up, many of which may have CVEs over the study period.

The decreases in blood pressures from baseline were more prominent in the prospective clinical trials than in the current study. There are several reasons to explain this observation. First, there was no study protocol involved in the current study design. Second, the blood pressure readings were extrapolated directly from the medical charts. The technique of blood pressure measurement could not be regulated. Third, the study population in the current study was considerably younger and had more females than the other clinical trials. The mean baseline blood pressure measurement was comparable with the ALLHAT study but considerable lower than those of the other two studies. In addition, the percentages of smokers and drinkers in the present study population were generally lower than the other three studies. Therefore, the impact of change in blood pressure measurements was not as prominent as in the other three clinical trials. This study demonstrated that it was more difficult to control SBP compared with DBP. SBP was not controlled in the current study population for 54.7% of the study period, whilst for DBP it was 25.8%. The average SBP deviation from the treatment goal within the study period was 3.8 mmHg whilst for DBP it was -7.5 mmHg.

Polypharmacy (use of multiple medications) was noted, since each patient required an average of five drugs during the 5-year study period both including CVD management drugs and non-CVD-related drugs. Most drug switching was done before the maximum effective dose for one drug was achieved. Patients were often seen by different physicians during each follow-up visit. The different prescribing preferences of different physicians may have led to possible unnecessary drug changes.

Furthermore, the drug use pattern in the current study was different than the preferred agents in the published guidelines or clinical trials. b-blockers was the most commonly prescribed drug class in the current study for hypertensive management. First-line treatment for hypertension recommended by the JNC-7 guidelines7 and the ALLHAT study21 was thiazide diuretic, which was ranked fourth among all blood pressure lowering medications in this study. Use of angiotensin-converting enzyme inhibitors and calcium channel blockers as preferred treatment in the ANBP-222 and ASCOT-BPLA23 studies was ranked third and second, respectively, among all blood pressure lowering medications in the current study. Change in the local prescribing patterns to increase the proportion of patients using thiazide diuretics as a more cost-effective alternative may potentially reduce the medication cost of managing hypertension.

In addition, more patients had CVEs and polypharmacy problems when their SBP and DPB had higher percentages of time that were out of control. Although the cause and effect relationship could not be established owing to the retrospective nature of this study, there was a clear association between SBP control, polypharmacy and the incidence of CVEs. A similar observation has been reported by the Multiple Risk Factor Intervention Trial (MRFIT)24 where the coronary heart disease death rate was positively associated with SBP and DBP out of their treatment targets. This could be explained by the fact that aging increases arterial stiffness as a result of progressive replacement of elastin by collagen in the walls of large arteries25. This leads to dilatation and lengthening of the aorta and its immediate branches via fibrosis and hypertrophy of the arterial musculatures26. This process, known as arteriosclerosis, is therefore more closely related to SBP but not DBP. Therefore, SBP is a more reliable marker of age-related vascular target organ damage than DBP25. The importance of management of isolated systolic hypertension was also emphasised by the JNC-7 guidelines7 and the Framingham Heart Study27. Both reports stated that SBP is more robust than DBP as a CVD risk factor and that improper management of SBP leads to unacceptable low rates of overall blood pressure control. In the ALLHAT21 and Controlled Onset Verapamil Investigation of Cardiovascular End Points (CONVINCE)28 trials, DBP control rates exceeded 90% but SBP control rates were between 60 and 70%. These blood pressure control patterns coincided with the findings of this study.

In the current study, the focus was on the economic outcomes when a patient required hospitalisation secondary to uncontrolled hypertension. Other treatment outcomes such as indirect medical costs, increased clinic visits or rehabilitation services were not examined. Future studies are necessary to investigate other related economic outcomes in the management of hypertension. It is an inevitable fact that the trend of aging in the Hong Kong population will increase the burden to the healthcare system. The prevalence of hypertension is expected to be higher in the future. Reallocation of resources should be considered to earlier and more effective drug therapy to reduce overall economic costs. More pharmacoeconomic studies would provide current guidance for decision making.

There were several limitations to the current study. First, owing the retrospective observational study design, the study was not able to establish a definite cause and effect relationship between the two variables observed. In addition, multiple factors may account for the incidence of CVEs, CBVEs and mortality. Since this study focused on only one factor, inadequately controlled hypertension, the effects of other risk factors for CVEs, CBVEs and mortality were not addressed. Moreover, blood pressure readings were extrapolated directly from the medical charts. Therefore, the technique of blood pressure measurement could not be regulated. As a result, the accuracy of the data could not be validated. In addition, many patients were lost to follow-up owing to the large number of patients mobilised from district to district in Hong Kong. Once they had moved away to different districts, their medical records could not be retrieved and what had happened to those patients could not be determined. The investigators did not know whether the patients had passed away or whether their blood pressures were controlled. It is quite commonly seen in Hong Kong that patients either do not want to be followed-up or that they have been relocated to another district where the investigators had no assess to their medical records. The study was also limited by the small study population and its retrospective nature which led to the inability to identify detailed medical information. In the future, it would be useful to conduct a more formal study with a larger study population to look into this area in greater detail. Having reliable estimates of the cost of treating hypertension is critical to decision-making regarding the cost effectiveness of treatments and the allocation of resources.

Conclusion

This study confirmed that the incidence of CVEs or CBVEs in patients with inadequately controlled hypertension was positively related to the extent and duration of blood pressure deviation from treatment goals. The association was stronger for SBP than DBP. The number of cardiovascular and non-cardiovascular drugs prescribed was positively associated with the extent and duration of blood pressure deviation from treatment goals and the incidence of CVEs or CBVEs. The study also estimated that managing hypertensive patients with CVE in hospital accounted for 3% of total government subvention in the year 2005. Therefore, a systematic and multidisciplinary approach for proper hypertensive management is necessary to reduce cost and clinical events secondary to hypertension.

Acknowledgements

Declaration of interest: The authors have declared no conflict of interest and received no payment in the preparation of this manuscript.