- Email: [email protected]
The incidence of cancer deaths among hypertensive patients in a large Chinese population: A cohort study
International Journal of Cardiology 179 (2015) 178–185
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
The incidence of cancer deaths among hypertensive patients in a large Chinese population: A cohort study Martin C.S. Wong a,b,1, Wilson W.S. Tam a,b,1, X.Q. Lao a,b,1, Harry H.X. Wang a,b,1, Mandy W.M. Kwan a,1, Clement S.K. Cheung c,1, Ellen L.H. Tong c,1, N.T. Cheung c,1, Sian M. Griffiths a,b,⁎,1, Andrew J.S. Coats d,e,1 a
JC School of Public Health and Primary Care, Faculty of Medicine, Chinese University of Hong Kong, Shatin, NT, Hong Kong CUHK Shenzhen Research Institute, Chinese University of Hong Kong, No. 10, 2nd Yuexing Road, Nanshan District, Shenzhen, PR China Hospital Authority Information Technology Services — Health Informatics Section, Suite Nos. 2–15, 7/F Exchange Tower, 33 Wang Chiu Road, Kowloon Bay, Hong Kong d Department of Medicine and Therapeutics, Faculty of Medicine, The Chinese University of Hong Kong, Hong Kong e Monash University, Australia and the University of Warwick, UK b c
a r t i c l e
i n f o
Article history: Received 25 May 2014 Received in revised form 4 September 2014 Accepted 14 October 2014 Available online 20 October 2014 Keywords: Cancer mortality Hypertensive patients Antihypertensive agents Associated factors Chinese population
a b s t r a c t Current evidence is mixed regarding the association between antihypertensive prescriptions and cancer mortality. We evaluated this association in a large Chinese hypertensive population. We followed for five years all patients who were prescribed their first-ever antihypertensive agents between 2001 and 2005 in a public healthcare sector of Hong Kong. The association between antihypertensive drug class and cancer mortality was evaluated by Cox proportional hazard models with propensity score matching. Age, gender, socioeconomic status, service settings, district of residence, proportion of days covered reflecting medication adherence, and the number of comorbidities were adjusted. From 217,910 eligible patients, 9500 (4.4%) died from cancer within five years after their first-ever antihypertensive prescription. Most cancer deaths occurred in the digestive (38.9%) and respiratory system (30.4%); the breast (6.2%); and the lympho-hematopoietic tissues (5.3%). The proportion of patients who died from cancer was the highest in the calcium channel blocker (CCB) group (6.5%), followed by thiazide diuretics (4.4%), angiotensin converting enzyme inhibitors (4.2%) and β-blockers (2.6%). When compared with β-blockers, patients prescribed CCBs (Adjusted Hazard Ratio [AHR] = 1.406, 95% C.I. 1.334–1.482, p < 0.001) were more likely to die from cancer. Thiazide users were also more likely to suffer from cancer deaths (AHR = 1.364, 95% C.I. 1.255–1.483, p < 0.001), but became insignificant in stratified analysis. The association between cancer mortality and use of CCB, and perhaps thaizide, may alert physicians to the need for more meticulous and comprehensive care of these patients in clinical practice. We recommend prospective studies to evaluate cause-and-effect relationships of these associations. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Hypertension is the biggest contributor to the burden of cardiovascular disease, accounting for 54% of stroke and 47% of ischemic heart disease worldwide [1]. Its prevalence is growing in various regions [2, 3] whilst control rates remain poor in many countries, including China [4]. Antihypertensive therapies have been recognized as effective agents to control blood pressure [5,6]. Most international guidelines recommend any major antihypertensive drug classes as suitable first-line medications for management of arterial hypertension, including the
⁎ Corresponding author at: 2/F, School of Public Health and Primary Care, Prince of Wales Hospital, Chinese University of Hong Kong, Shatin, NT, Hong Kong. Tel.: + 852 2252 8702; fax: +852 2606 3500. E-mail address: siangriffi[email protected] (S.M. Griffiths). 1 Statement of authorship: All authors take responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation.
http://dx.doi.org/10.1016/j.ijcard.2014.10.028 0167-5273/© 2014 Elsevier Ireland Ltd. All rights reserved.
Joint National Committee seventh report [7] and the 2009 reappraisal of the hypertension European guidelines [8]. The majority of the scientific literature on the subject has compared the different antihypertensive drug classes according to blood pressure lowering efficacy and cardiovascular events and mortality [5,9] from randomized trials and cohort studies, and most have found little difference between various antihypertensive classes. However, in the most recent decade studies have emerged based on other outcomes, including cancer. For instance, Pahor and colleagues [10,11] found that users of Calcium Channel Blockers (CCBs) had 72% higher risks of developing cancer than non-users among more than 5000 elderly patients in the US. These findings were echoed by Fitzpatrick and colleagues who reported a higher risk of incident invasive breast carcinoma among CCB users from the US Cardiovascular Health Study (CHS) — showing a clear dose response relationship between CCB and cancer [12]. In addition, the combined use of Angiotensin Converting Enzyme Inhibitors (ACEIs) and Angiotensin Receptor
M.C.S. Wong et al. / International Journal of Cardiology 179 (2015) 178–185
Blockers (ARBs) have been shown to have higher cancer risk than other classes from a network meta-analysis of 70 randomized trials, involving more than 320,000 patients [13]. Nevertheless, these findings were not supported by subsequent observational studies [14–23], and a mixed treatment comparison meta-analysis [24] also confirmed that no excess incidence of cancer could be imputed to CCBs and other antihypertensive classes. Interestingly, however, a recent population-based case–control study in the three-county Seattle–Puget Sound metropolitan area conducted among women aged 55 to 74 years found that current use of CCBs was associated with higher risks of both ductal and lobular breast cancer, whilst other antihypertensive drug classes were not associated with risk [25]. Therefore, current evidence is rather mixed on the association between antihypertensive drug class and cancer as an important outcome. Also, these studies included only modest numbers of individuals and might not have been adequately powered to capture mortality due to cancer. Besides, they were almost exclusively conducted among Western populations, limiting generalizability to other patient groups such as the Chinese. It is well recognized that the pharmacological action of antihypertensive drugs differs according to ethnicity, and even small ethnic differences in its optimum management could have large implications for health resources [26]. Chinese people make up one-fifth of the world's population, residing in different parts of the globe. Hence, the objectives of this study are to test the a priori hypothesis that the incidence of deaths due to cancer was similar among the major antihypertensive drug classes when prescribed as first-line agents in large Chinese population.
179
2.2. Data source The present study retrieved data from a comprehensive computerized database known as electronic Clinical Management System adopted by the Hospital Authority in Hong Kong. It captures patients' demographic and prescription profiles, and clinical diagnoses in the forms of the International Classification of Diseases (ICD-9) or the International Classification of Primary Care (ICPC-2). The database has been previously reported with detailed descriptions on its validity [27–34], which demonstrated a high level of completeness with respect to the socio-demographic information (100%) and prescription profiles (99.98%) [29]. All drug prescriptions are entered by the attending physicians and are double-checked by the dispensing pharmacists using standardized procedures, and any subsequent changes in patients' prescription after the initial consultations are also entered into the system. This computerized record system serves as the sole portal which allows cross-referencing of information by physicians at patient visits to all public clinical settings in different districts. The present study employed a sampling frame which included the entire Hong Kong population, i.e., more than 7,000,000 as of the year 2013. Hong Kong is divided into three regions, namely the Hong Kong Island, Kowloon, and the New Territories, from the most urbanized to the most rural regions, respectively. 2.3. Study population
2. Methods
Data of all patients who attended any public practice and who received their first-ever antihypertensive prescription during the calendar years 2001 to 2005 (the index date) from the database were retrieved. Patients who were prescribed with any antihypertensive agents prior to the index date were excluded. We treated deaths due to medical conditions other than cancer within the observation period as being censored. Each patient was classified into one drug group according to the initial prescription of antihypertensive agents, which include β-blockers, thiazide diuretics, CCB and ACEIs. It was found in the database that patients who received ARBs (n = 434, 0.3%) and combination therapy (n = 74, 0.0%) were relatively rare so for that reason such patients were excluded in the data analysis. Concomitant comorbidities include cardiovascular risk factors and conditions which might confound the choice of antihypertensive medication, as indicated by the respective ICD-9 or ICPC-2 codes. Patients were followed up until the occurrence of mortality within the study period or at the end of five years, whichever is earlier. The original dataset consisted of 223,257 subjects and hence 5347 subjects were excluded due to the exclusion criteria mentioned above. The flow of participants was shown in Fig. 1.
2.1. Ethics statement
2.4. Outcome measures and covariates
The ethics clearance of the study was obtained from the Clinical Ethics Research Committee of the Hospital Authority, and the Survey and Behavioral Research Ethics Committee of The Chinese University of Hong Kong. Informed consent was not necessary as all subjects were anonymized with unique identity numbers. The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki.
The primary outcome variables were: (1) All-cause mortality (ICD-10 C00-D49) and (2) deaths due to cancers affecting the gastrointestinal tract or the lungs, respectively. The vast majority of all deaths in Hong Kong occurred in the hospitals, and it allows accurate case ascertainment [35]. Death records in hospitals have been utilized as a valid proxy of patient mortality in a number of previous studies, especially for Chinese populations
Fig. 1. Flow of participants.
180
M.C.S. Wong et al. / International Journal of Cardiology 179 (2015) 178–185
Table 1 Baseline characteristics of patients (N = 217,910) by 5-year cancer mortality.
Gender Male Female Age <50 50–59 60–69 ≥70 Public assistance Yes No Service type on first visit In-/day-patient clinic Special out-patient clinic Accident and emergency General outpatient clinics Others District of residence Hong Kong Kowloon New Territories First prescription ACEIs Alpha blocker Beta blocker CCB Thiazide Proportion Days Covered (PDC) <40% ≥40% Co-morbidity 0 1 2 ≥3
N
All cancer deaths (N = 9500)
Digestive cancer deaths (N = 3697)
Respiratory cancer deaths (N = 2885)
98,192 119,717
5854 (6.0%) 3646 (3.0%)
2412 (2.5%) 1285 (1.1%)
1926 (2.0%) 959 (0.8%)
61,343 42,018 41,602 72,928
562 (0.9%) 1070 (2.5%) 1986 (4.8%) 5882 (8.1%)
217 (0.4%) 465 (1.1%) 824 (2.0%) 2191 (3.0%)
87 (0.1%) 228 (0.5%) 598 (1.4%) 1972 (2.7%)
32,800 183,045
7361 (4.0%) 1970 (6.0%)
2856 (1.6%) 771 (2.4%)
2213 (1.2%) 617 (1.9%)
65,725 64,679 16,898 64,715 5885
6243 (9.5%) 1510 (2.3%) 306 (1.8%) 1298 (2.0%) 143 (2.4%)
2415 (3.7%) 631 (1.0%) 119 (0.7%) 472 (0.7%) 60 (1.0%)
1935 (2.9%) 415 (0.6%) 77 (0.5%) 410 (0.6%) 48 (0.8%)
38,301 73,769 105,840
1844 (4.8%) 3691 (5.0%) 3965 (3.7%)
731 (1.9%) 1400 (1.9%) 1566 (1.5%)
541 (1.4%) 1190 (1.6%) 1154 (1.1%)
22,970 13,452 98,586 65,465 17,437
961 (4.2%) 903 (6.7%) 2588 (2.6%) 4278 (6.5%) 770 (4.4%)
363 (1.6%) 293 (2.2%) 1283 (1.3%) 1468 (2.2%) 290 (1.7%)
291 (1.3%) 308 (2.3%) 559 (0.6%) 1503 (2.3%) 224 (1.3%)
71,598 14,6312
3212 (4.5%) 6288 (4.3%)
1261 (1.8%) 2436 (1.7%)
942 (1.3%) 1943 (1.3%)
131,006 70,872 14,574 1458
5236 (4.0%) 3487 (4.9%) 717 (4.9%) 60 (4.1%)
2244 (1.7%) 1192 (1.7%) 237 (1.6%) 24 (1.6%)
1375 (1.0%) 1222 (1.7%) 268 (1.8%) 20 (1.4%)
All proportions were across rows. Chi-square tests among each categories of all variables showed p < 0.001. PDC is defined as the number of days when medication is supplied in a specified period divided by the total number of days within the observation period. ACEI: Angiotensin converting enzyme inhibitors; CCB: calcium channel blockers. Patients who visited the clinic for once only were not included.
whose deaths almost always occur in hospitals [35–37]. The cause of death for each patient was defined according to the primary cause of mortality for each patient, as determined by the physician-in-charge when death was registered in the death certificate. The predictor variable in this study was the major antihypertensive drug classes on its initial prescription. To take into account the influence of medication adherence as a potential confounder, the study included the measure of interval-based proportion of days covered (PDC) as a covariate, which has been widely used in large database study as an internationally accepted metric for drug adherence assessment [38–40]. The interval-based PDC refers to the number of days when medication is supplied in a specified period divided by the total number of days within the period. The PDC was treated as a continuous variable in the analysis and was estimated using a five-year interval, i.e., for patients who died during the five-year period, the PDC was estimated using the time period between the index date and the incidence of mortality. Similar to the internationally recognized approach [41], drug adherence in the present study was divided into two levels, namely: high or intermediate (PDC ≥ 0.40) and low (PDC < 0.40) [40,42]. In addition, we also evaluated the rates of medication discontinuation and switching among these patients. Drug switching is defined as the absence of a refill prescription in all subsequent clinic visits combined with the prescription of another antihypertensive drug of a different class since the date of the first prescription, whereas drug discontinuation refers to the absence of repeat prescription of the same drug class after the first-ever antihypertensive prescription date [43–45]. The independent variables that were controlled for included patients' age, gender, payment status (recipients vs non-recipients of comprehensive social security assistance, CSSA; each consultation costs approximately US $5.77, including investigation and prescription fees), clinic types visited (general outpatient clinic [GOPC], specialist outpatient clinics [SOPCs], in- and day-patient clinics, Accident and Emergency Departments [AEDs], and other clinic types), their districts of residence (Hong Kong Island vs Kowloon vs the New Territories; from the most urbanized to the most rural regions, respectively), and the total number of comorbidities (as reflected by the number of ICD-9 codes). The same list of comorbidities reflected by their ICD-9 and ICPC-2 codings was used in the present study as published elsewhere [28], which were categorized under “Cardiovascular diseases” (24.3%), “Diabetes or impaired glucose tolerance” (23.0%), “Respiratory diseases” (14.6%) and “Renal diseases” (11.0%).
2.5. Statistical analysis The Statistical Package for Social Sciences version 16.0 (SPSS, Inc.) was used for all data analysis. For descriptive analysis, Student's t tests were used to compare continuous variables, and chi-square tests of heterogeneity were used to compare categorical variables. Three separate Cox-proportional hazard models were developed to explore the association between initial antihypertensive drug class and mortality due to cancer at any site, at the gastrointestinal tract, and at the lungs, respectively. To account for the confounding effects of medication change, all study participants were censored if they switched study medications, discontinued treatment (in which case they were followed for 100 days from their last prescription to identify events that may have precipitated discontinuation), commenced treatment with a different antihypertensive agent, or reached the end of the study. This is according to a methodology adopted by Dhalla et al. [46], and we followed patients for a maximum of 5 years. Also, we conducted a sensitivity analysis where patients whose antihypertensive drug class has been switched or added on by another drug class were excluded, and a similar regression model was constructed to detect if there were any changes in the associated factors identified. In addition, a propensity score model was developed to estimate weightings for the selection of the two prescriptions, given the possible existence of treatment indication bias due to the observational nature of the study. The variables used in the calculation of the propensity scores consisted of all the covariates, namely patients' age, gender, receipt of public assistance, service type on first visit, district of residence, the drug class of first prescription, the Proportion Days Covered, and the number of comorbidities. Therefore, the analysis was performed with adjusted estimates weighted by the inverse estimated propensity scores in three Cox proportion hazard models according to standard statistical methodology [47,48]. To further minimize the influence of indication bias due to different baseline characteristics of patients, a high-dimensional propensity score matching to compare patients with similar observed characteristics was adopted using standardized methodology as utilized in other literature [46,49]. All the variables listed above were entered unconditionally into the model. Since patients with prostate cancer might be prescribed with alpha-blockers for prostate symptoms before or after diagnosis, we conducted two separate analyses where patients received alpha blockers as first-line antihypertensive agents were included and excluded, separately, and the significant associated factors were compared between the two analyses. Since the study involved multiple
M.C.S. Wong et al. / International Journal of Cardiology 179 (2015) 178–185
181
Table 2 Baseline Characteristics of Patients (N = 217,910) by First Prescription.
Gender Male Female Age <50 50–59 60–69 ≥70 Public assistance Yes No Service type on first visit In-/day-patient clinic Special out-patient clinic Accident and emergency General outpatient clinics Others District of residence Hong Kong Kowloon New Territories Proportion Days Covered (PDC) <40% ≥40% Co-morbidity 0 1 2 ≥3
N
ACEI N = 22,970
Alpha-blockers N = 13,452
Beta-blockers N = 98,585
CCB N = 65,465
Thiazide N = 17,437
98,192 119,717
11,694 (50.9%) 11,276 (49.1%)
13,131 (97.6%) 321 (2.4%)
37,669 (38.2%) 60,916 (61.8%)
29,394 (44.9%) 36,071 (55.1%)
6304 (36.2%) 11,133 (63.8%)
61,343 42,018 41,602 72,928
4598 (20.0%) 4232 (18.4%) 4530 (19.7%) 9607 (41.8%)
951 (7.1%) 2233 (16.6%) 4041 (30.0%) 6227 (46.3%)
43,740 (44.4%) 21,693 (22.0%) 15,454 (15.7%) 17,695 (17.9%)
9318 (14.2%) 10,377 (15.9%) 13,712 (20.9%) 32,047 (49.0%)
2736 (15.7%) 3483 (20.0%) 3865 (22.2%) 7352 (42.2%)
32,800 183,045
3798 (16.7%) 18,940 (83.3%)
2635 (19.9%) 10,599 (80.1%)
11,000 (11.3%) 86,647 (88.7%)
12,585 (19.4%) 52,320 (80.6%)
2782 (16.1%) 14,539 (83.9%)
65,725 64,679 16,898 64,715 5885
8803 (38.3%) 8092 (35.2%) 227 (1.0%) 5164 (22.5%) 683 (3.0%)
3200 (23.8%) 3820 (28.4%) 635 (4.7%) 5263 (39.1%) 533 (4.0%)
22,783 (23.1%) 37,633 (38.2%) 9124 (9.3%) 26,342 (26.7%) 2702 (2.7%)
27,188 (41.5%) 11,580 (17.7%) 5417 (8.3%) 19,858 (30.3%) 1418 (2.2%)
3751 (21.5%) 3554 (20.4%) 1495 (8.6%) 8088 (46.4%) 549 (3.1%)
38,301 73,769 105,840
5542 (22.8%) 8069 (35.1%) 9659 (42.1%)
1509 (11.2%) 5973 (44.4%) 5970 (44.4%)
16,964 (17.2%) 32,299 (32.8%) 49,323 (50.0%)
11,078 (16.9%) 23,073 (35.2%) 31,314 (47.8%)
3508 (20.1%) 4355 (25.0%) 9574 (54.9%)
71,598 146,312
4298 (18.7%) 18,762 (81.3%)
5040 (37.5%) 8412 (62.5%)
37,068 (37.6%) 61,518 (62.4%)
19,614 (30.0%) 45,851 (70.0%)
5578 (32.0%) 11,859 (68.0%)
131,006 70,872 14,574 1458
8067 (35.1%) 11,578 (50.4%) 2969 (12.9%) 356 (1.5%)
4550 (33.8%) 6474 (48.1%) 2228 (16.6%) 200 (1.5%)
73,559 (74.6%) 21,906 (22.2%) 2906 (2.9%) 215 (0.2%)
34,438 (52.6%) 24,955 (38.1%) 5479 (8.4%) 593 (0.9%)
10,392 (59.6%) 5959 (34.2%) 992 (5.7%) 94 (0.5%)
All proportions were across columns. PDC is defined as the number of days when medication is supplied in a specified period divided by the total number of days within the observation period. For all covariates, p < 0.01 from Chi-square test for association. ACEI: Angiotensin converting enzyme inhibitors; CCB: calcium channel blockers. Patients who visited the clinic for once only were not included.
comparisons, a “definite” association between an antihypertensive drug class with cancer deaths was established only if the drug class was found to be significantly associated with all three cancer mortality. All p values b0.05 were considered statistically significant.
3. Results 3.1. Patient characteristics From a total of 217,910 eligible patients included in the present study (Table 1), more than half were female patients (54.9%) and were aged 60 years or older (52.6%). Only a small proportion (15.2%) of all patients received public assistance and the majority (59.8%) was followed up at in- or day-patient clinics and specialized out-patient
clinics. The majority of patients lived in the more rural regions (48.6% and 33.9% in the New Territories and Kowloon, respectively). The most commonly prescribed antihypertensive drugs were β-blockers (45.2%) and CCBs (30.0%), followed by ACEIs (10.5%) and thiazide diuretics (8.0%). The average PDC was 0.73 (SD 0.90) which implied intermediate to good medication adherence overall, and most patients had one comorbidity or less (92.6%). Except α-blockers which might be prescribed to patients with prostatism, the proportion of cancer deaths was the highest among CCB users (6.5%) when compared with other medication users (2.6%, 4.2% and 4.4% for β-blockers, ACEIs, and thiazide diuretics). This was similar for digestive and respiratory cancer mortality. Table 2 shows the baseline characteristics of the patients according to
Fig. 2. The distribution of cancer mortality according to body systems.
182
M.C.S. Wong et al. / International Journal of Cardiology 179 (2015) 178–185
Table 3 Hazard ratios of cancer mortality by Cox proportional hazards regression analysis.
Gender Male Female Age <50 50–59 60–69 ≥70 Public assistance Yes No Service type on first visit In-/day-patient clinic Special out-patient clinic Accident and emergency General outpatient clinics District of residence Hong Kong Kowloon New Territories First Prescription ACEIs Beta blocker CCB Thiazide Proportion Days Covered (PDC) <40% ≥40% Co-morbidity No Yes
All cancer (N = 8473)
Digestive cancer (N = 3349)
Respiratory cancer (N = 2538)
1.000 0.537 (0.514, 0.561)
1.000 0.437 (0.407, 0.470)
1.000 0.435 (0.401, 0.473)
1.000 2.938 (2.644, 3.263) 4.839 (4.386, 5.340) 7.460 (6.798, 8.186)
1.000 3.538 (2.998, 4.174) 5.879 (5.027, 6.876) 8.970 (7.729, 10.41)
1.000 3.607 (2.797, 4.652) 7.909 (6.261, 9.991) 13.22 (10.55, 16.56)
0.930 (0.880, 0.983) 1.000
1.017 (0.933, 1.110) 1.000
0.865 (0.783, 0.955) 1.000
5.426 (5.075, 5.800) 1.643 (1.514, 1.783) 1.129 (0.986, 1.293) 1.000
5.645 (5.069, 6.286) 1.760 (1.547, 2.003) 1.114 (0.894, 1.387) 1.000
4.943 (4.385, 5.573) 1.546 (1.328, 1.799) 1.025 (0.792, 1.327) 1.000
1.000 0.983 (0.926, 1.043) 0.884 (0.834, 0.937)
1.000 0.937 (0.853, 1.030) 0.892 (0.814, 0.978)
1.000 1.090 (0.978, 1.215) 0.886 (0.795, 0.987)
0.952 (0.881, 1.029) 1.000 1.406 (1.334, 1.482) 1.364 (1.255, 1.483)
0.758 (0.671, 0.856) 1.000 0.994 (0.917, 1.078) 1.048 (0.918, 1.197)
1.136 (0.978, 1.319) 1.000 2.041 (1.841, 2.262) 1.710 (1.457, 2.006)
1.000 0.852 (0.814, 0.893)
1.000 0.844 (0.784, 0.908)
1.000 0.889 (0.816, 0.968)
1.000 0.791 (0.756, 0.828)
1.000 0.647 (0.601, 0.696)
1.000 0.940 (0.866, 1.019)
PDC is defined as the number of days when medication is supplied in a specified period divided by the total number of days within the observation period. ACEI: Angiotensin converting enzyme inhibitors; CCB: calcium channel blockers. Patients who were prescribed with alpha-blockers or seen in service settings other than those listed were excluded from the analysis.
the first antihypertensive prescriptions. CCB users were older, and higher proportions attended in- or day-patient clinics. The number of comorbidites was also substantially different among the various drug classes. 3.2. Profiles of mortality due to cancer and respiratory disease A total of 9500 patients (4.4%) died from cancer within five years after their first-ever antihypertensive prescription. Most cancer deaths occurred in the digestive (38.9%) and respiratory system (30.4%); the breast (6.2%); and the lymphatic and hematopoietic tissue (5.3%) (Fig. 2).
1.334–1.482; p < 0.001) and respiratory cancer (AHR = 2.041, 95% C.I. 1.841–2.262, p < 0.001). In addition, thiazide users were also more likely to suffer from any cancer mortality (AHR = 1.364, 95% C.I. 1.255–1.483, p < 0.001) and respiratory cancer (AHR = 1.710, 95% C.I. 1.457–2.006, p < 0.001). Owing to the substantial difference in the distribution of service setting according to drug class, a stratified analysis according to service setting was conducted (Table 4). It was shown that CCB users were significantly more likely to suffer from mortality due to all cancers in all the settings (AHR ranged from 1.179 to 1.458). Thiazide diuretics was associated with higher mortality from all cancers in the in- or day-patient clinics, SOPCs but not GOPCs. 3.4. Sensitivity analyses
3.3. Factors associated with cancer mortality From Cox proportional hazard analysis with all cancer mortality as the outcome measure, (Table 3), it was found that advanced age (≥50 years; adjusted hazard ratio [AHR] ranged from 2.938 to 7.460, p < 0.001; referent <50 years), male gender (AHR for female = 0.537, 95% C.I. 0.514–0.561, p < 0.001), non-recipients of public assistance (AHR for receiver = 0.930, 95% C.I. 0.880–0.983, p < 0.001), attendance in in-patient or day-patient service (AHR = 5.426, 95% C.I. 5.075–5.800, p < 0.001), residence in more urbanized regions (AHR for the New Territories = 0.884, 95% C.I. 0.834–0.937, p < 0.001), lower levels of medication adherence (AHR = 0.852, 95% C.I. 0.814–0.893, p < 0.001 for PDC < 40%) and the absence of comorbidity (AHR for at least one comorbidity = 0.791, 95% C.I. 0.756–0.828, p < 0.001) were positively associated with mortality from all cancers. These findings were similar for digestive cancers and respiratory cancer (Table 3). When compared with patients who were prescribed with β-blockers, users of CCBs were more likely to die from any cancer (AHR = 1.406, 95% C.I.
An additional analysis where patients whose antihypertensive drug class has been switched or added on by another drug class were excluded reported the same findings (Table 5). In addition, a propensity score model matched for drug class and a high-dimensional propensity score matching methodology yielded similar results. Two separate analyses where patients received α-blockers as first-line antihypertensive agents were included and excluded also gave similar findings. 4. Discussion 4.1. Statement of principal findings From this observational study involving 217,910 Chinese hypertensive patients, we found that first-line prescription of CCBs as a firstline antihypertensive agent was associated with a higher likelihood of death from any cancer and respiratory cancer, when compared with users of β-blockers. This is consistently true from various sensitivity
M.C.S. Wong et al. / International Journal of Cardiology 179 (2015) 178–185
183
Table 4 Hazard ratios of cancer mortality by Cox proportional hazards regression analysis: sensitivity analysis according to the service setting. In-/day-patient clinic
All cancer (N = 5789)
Digestive cancer (N = 2265)
Respiratory cancer (N = 1768)
First prescription ACEIs Beta blocker CCB Thiazide
0.860 (0.781, 0.948) 1.000 1.458 (1.369, 1.553) 1.352 (1.209, 1.511)
0.616 (0.528, 0.718) 1.000 0.938 (0.852, 1.032) 0.990 (0.831, 1.181)
1.128 (0.935, 1.360) 1.000 2.383 (2.049, 2.711) 1.823 (1.468, 2.265)
Specialist out-patient clinics
All cancer (N = 1310)
Digestive cancer (N = 563)
Respiratory cancer (N = 351)
First prescription ACEIs Beta blocker CCB Thiazide
1.122 (0.950, 1.324) 1.000 1.179 (1.029, 1.351) 1.440 (1.192, 1.738)
0.904 (0.701, 1.168) 1.000 0.877 (0.710, 1.082) 1.100 (0.815, 1.484)
1.517 (1.097, 2.099) 1.000 1.780 (1.363, 2.324) 1.999 (1.396, 2.863)
General out-patient clinics
All cancer (N = 1103)
Digestive cancer (N = 418)
Respiratory cancer (N = 347)
First prescription ACEIs Beta blocker CCB Thiazide
1.265 (1.016, 1.574) 1.000 1.209 (1.046, 1.398) 1.136 (0.939, 1.376)
1.677 (1.198, 2.348) 1.000 1.224 (0.966, 1.553) 1.113 (0.813, 1.524)
0.962 (0.634, 1.461) 1.000 1.217 (0.942, 1.573) 1.041 (0.733, 1.478)
ACEI: Angiotensin converting enzyme inhibitors; CCB: calcium channel blockers.
analyses. Thiazide diuretics were not associated with cancer mortality outcomes in the GOPC setting, and hence we did not establish a definite association with cancer mortality for this drug class. Owing to the retrospective nature of this study, we did not conclude a cause-and-effect association between CCBs and cancer mortality.
settings. Since it is not an evaluation of efficacy, caveats should be observed when direct comparisons on the association between antihypertensive drug class and the mortality outcomes were made.
4.2. Study limitations
Previous studies reporting positive [10–12,25] or negative [13–24] associations between CCB use and cancer were mostly observational studies, and they were relatively small-scaled, except two systematic reviews of randomized controlled trials [13,24]. The study by Coleman and colleagues [24] involved more than 126,000 patients by mixed treatment comparison meta-analysis and showed no differences in cancer among different antihypertensive drug classes. The odds ratio of cancer incidence in the CCB group was 0.95 (95% C.I. 0.79–1.13) when compared with placebo or the untreated group. Another large-scale network meta-analysis conducted by Bangalore et al. [13] involving more than 324,000 patients from 70 RCTs also did not show any significant difference — and the odds ratio of cancer development for the CCB group was 1.05 (95% C.I. 0.96–1.13) when compared with placebo. As these studies included different population sub-groups, there exists a possibility that some ethnic groups could be under-represented. Our study is focused on “first-line prescription” as an independent variable without examining concomitant use of other antihypertensive agents and the switching or discontinuation of CCBs in the period of observation. Therefore, direct comparison of this study with previous literature is not appropriate [10–25]. However, it implied that among patients who received CCBs as first-line antihypertensive agents, there exists an association with cancer mortality, irrespective of the patients' subsequent prescription journey. The same can be said for thiazide diuretics in certain service settings — but the mechanism why it was associated with greater cancer mortality remains to be explored. It is also noteworthy that those with comorbidities had lower risk of cancer mortality. We speculated that patients with comorbid conditions may be more health-
To our knowledge this is the largest study which compared the various antihypertensive drug classes with respect to the controversial outcomes of mortality due to cancer. The highly complete, accurate territory-wide database which captured clinical information from over 218,000 patients observed for more than five years, in addition to the good dispensing practice of the clinics in the public healthcare sector further enhanced the reliability of the findings. However, some of the limitations should be addressed here. Firstly, we assume that patients prescribed with the medications were actually taking them, although the analysis has controlled for medication adherence using internationally recognized metrics. In addition, there were some confounders which could not be taken into account, such as smoking, drinking, significant family history and other lifestyle risk factors for both outcomes; and that the number of comorbidities was used as a proxy measure for patients' general well-being. There exists indication bias affecting the choice of antihypertensive drug class in observational studies, and critics might argue that not all patients prescribed with antihypertensive agents had hypertension but may have had other medical conditions. These include diseases like heart failure and coronary heart diseases, as well as complaints due to symptoms like tachycardia, prostatism and peripheral edema. This study relied completely on disease coding as the only criteria to define patient comorbidities, and physicians might miss the entry of these variables during clinical consultations. Also, this is an effectiveness study observing mortality outcomes according to first-line prescription of drug class in real-life clinical
4.3. Relationship with existing literature and explanation of findings
Table 5 Hazard ratios of cancer mortality by Cox proportional hazards regression analysis: sensitivity analysis according to the first prescription. First prescription
All cancer
Digestive cancer
Respiratory cancer
ACEIs (N = 14,496) Beta blocker (N = 61,061) CCB (N = 40,052) Thiazide (N = 8069)
0.849 (0.775, 0.929) 1.000 1.306 (1.228, 1.388) 1.396 (1.206, 1.547)
0.653 (0.567, 0.751) 1.000 0.853 (0.777, 0.937) 1.009 (0.856, 1.188)
1.093 (0.917, 1.302) 1.000 2.153 (1.902, 2.437) 2.065 (1.701, 2.507)
ACEI: angiotensin converting enzyme inhibitors; CCB: calcium channel blockers.
184
M.C.S. Wong et al. / International Journal of Cardiology 179 (2015) 178–185
conscious, in terms of the need for self-management of those chronic conditions and implementation of more intensive lifestyle measures. In addition, the comorbidities themselves could act as competing causes of patient deaths — and hence reducing mortality attributable to cancer. A further interesting observation from this study included an increased risk of cancer mortality with a lower level of medication adherence. We are of the view that patients having low adherence with drugs might have poorer lifestyle habits, like smoking and alcohol drinking which were associated with high cancer mortality rates. Yet why some firstline antihypertensive medications were associated with higher cancer mortality is still a knowledge gap.
[7]
[8]
[9]
[10]
[11]
4.4. Conclusions: implications to clinical practice and future research This study proposed that CCB users, and possibly patients prescribed with thiazide diuretics where a less definite association was observed, might represent a group of patients with higher cancer mortality rate five years within the first prescription. Our interpretation is that the drug class of the first-line prescription is not necessarily a causative factor to mortality, but patients who received CCBs that had intrinsic factors were associated with higher risks for cancer deaths — thus they will need closer monitoring when encountered in clinical practice. We suggested that this study should be substantiated by more larger-scale randomized trials in Chinese patients, as mortality outcome needs large sample size and longer duration to capture. The cause-and-effect association between first-line antihypertensive drug class and cancer mortality will need to be examined in future research.
[12]
[13]
[14] [15] [16] [17]
[18]
[19]
Conflict of interest The authors report no relationships that could be construed as a conflict of interest.
[20]
[21]
Grant support None. Disclosures
[22]
[23] [24]
None declared. Acknowledgments We express our gratitude for all the healthcare professionals who entered the data into the clinical database. We thank the Hospital Authority of the Hong Kong Government for allowing our research team to use the database.
[25]
[26] [27]
[28]
References [29] [1] C.M.M. Lawes, S. Vander Hoorn, A. Rodgers, I.S. Hypertens, Global burden of bloodpressure-related disease, 2001, Lancet 371 (2008) 1513–1518. [2] M.C.S. Wong, H.H.X. Wang, M.C.M. Leung, C.S.H. Tsang, S.V. Lo, S.M. Griffiths, The rising prevalence of self-reported hypertension among Chinese subjects: a populationbased study from 121 895 household interviews, QJM (2014), http://dx.doi.org/10. 1093/qjmed/hcu129. [3] K. Wolf-Maier, R.S. Cooper, J.R. Banegas, S. Giampaoli, H.W. Hense, M. Joffres, et al., Hypertension, prevalence and blood pressure levels in 6 European countries, Canada, and the United States, JAMA 289 (2003) 2363–2369. [4] T. Dorjgochoo, X.O. Shu, X. Zhang, H. Li, G. Yang, L. Gao, et al., Relation of blood pressure components and categories and all-cause, stroke and coronary heart disease mortality in urban Chinese women: a population-based prospective study, J Hypertens 27 (2009) 468–475. [5] Blood Pressure Lowering Treatment Trialists C, F. Turnbull, B. Neal, T. Ninomiya, C. Algert, H. Arima, et al., Effects of different regimens to lower blood pressure on major cardiovascular events in older and younger adults: meta-analysis of randomised trials, BMJ 336 (2008) 1121–1123. [6] G. Mancia, R. Fagard, K. Narkiewicz, J. Redon, A. Zanchetti, M. Bohm, et al., 2013 ESH/ ESC guidelines for the management of arterial hypertension: the Task Force for the Management of Arterial Hypertension of the European Society of Hypertension
[30] [31] [32]
[33]
[34]
[35]
(ESH) and of the European Society of Cardiology (ESC), Eur Heart J 34 (2013) 2159–2219. A.V. Chobanian, G.L. Bakris, H.R. Black, W.C. Cushman, L.A. Green, J.L. Izzo, et al., The seventh report of the joint national committee on prevention, detection, evaluation, and treatment of high blood pressure — the JNC 7 report, JAMA 289 (2003) 2560–2572. G. Mancia, S. Laurent, E. Agabiti-Rosei, E. Ambrosioni, M. Burnier, M.J. Caulfield, et al., Reappraisal of European guidelines on hypertension management: a European Society of Hypertension Task Force document, J Hypertens 27 (2009) 2121–2158. M.R. Law, J.K. Morris, N.J. Wald, Use of blood pressure lowering drugs in the prevention of cardiovascular disease: meta-analysis of 147 randomised trials in the context of expectations from prospective epidemiological studies, BMJ 338 (2009). M. Pahor, J.M. Guralnik, L. Ferrucci, M.C. Corti, M.E. Salive, J.R. Cerhan, et al., Calciumchannel blockade and incidence of cancer in aged populations, Lancet 348 (1996) 493–497. M. Pahor, J.M. Guralnik, M.E. Salive, M.C. Corti, P. Carbonin, R.J. Havlik, Do calcium channel blockers increase the risk of cancer? Am J Hypertens 9 (1996) 695–699. A.L. Fitzpatrick, J.R. Daling, C.D. Furberg, R.A. Kronmal, J.L. Weissfeld, Use of calcium channel blockers and breast carcinoma risk in postmenopausal women, Cancer 80 (1997) 1438–1447. S. Bangalore, S. Kumar, S.E. Kjeldsen, H. Makani, E. Grossman, J. Wetterslev, et al., Antihypertensive drugs and risk of cancer: network meta-analyses and trial sequential analyses of 324168 participants from randomised trials, Lancet Oncol 12 (2011) 65–82. H.J. Cohen, C.F. Pieper, J.T. Hanlon, W.E. Wall, B.M. Burchett, R.J. Havlik, Calcium channel blockers and cancer, Am J Med 108 (2000) 210–215. H. Jick, S. Jick, L.E. Derby, C. Vasilakis, M.W. Myers, C.R. Meier, Calcium-channel blockers and risk of cancer, Lancet 349 (1997) 525–528. L. Rosenberg, S. Rao, J.R. Palmer, B.L. Strom, P.D. Stolley, A.G. Zauber, et al., Calcium channel blockers and the risk of cancer, JAMA 279 (1998) 1000–1004. J.H. Olsen, H.T. Sorensen, S. Friis, J.K. McLaughlin, F.H. Steffensen, G.L. Nielsen, et al., Cancer risk in users of calcium channel blockers, Hypertension 29 (1997) 1091–1094. A.F. Lever, D.J. Hole, C.R. Gillis, I.R. McCallum, G.T. McInnes, P.L. MacKinnon, et al., Do inhibitors of angiotensin-I-converting enzyme protect against risk of cancer? Lancet 352 (1998) 179–184. C.R. Meier, L.E. Derby, S.S. Jick, H. Jick, Angiotensin-converting enzyme inhibitors, calcium channel blockers, and breast cancer, Arch Intern Med 160 (2000) 349–353. T.L. Vaughan, D.C. Farrow, P.D. Hansten, W.H. Chow, M.D. Gammon, H.A. Risch, et al., Risk of esophageal and gastric adenocarcinomas in relation to use of calcium channel blockers, asthma drugs, and other medications that promote gastroesophageal reflux, Cancer Epidemiol Biomarkers Prev 7 (1998) 749–756. K.B. Michels, B.A. Rosner, A.M. Walker, M.J. Stampfer, J.E. Manson, G.A. Colditz, et al., Calcium channel blockers, cancer incidence, and cancer mortality in a cohort of US women — the Nurses' Health Study, Cancer 83 (1998) 2003–2007. H.T. Sorensen, J.H. Olsen, L. Mellemkjaer, A.M. Thulstrup, F.H. Steffensen, J.K. McLaughlin, et al., Cancer risk and mortality in users of calcium channel blockers — a cohort study, Cancer 89 (2000) 165–170. R.M. Vezina, S.M. Lesko, L. Rosenberg, S. Shapiro, Calcium channel blocker use and the risk of prostate cancer, Am J Hypertens 11 (1998) 1420–1425. C.I. Coleman, W.L. Baker, J. Kluger, C.M. White, Antihypertensive medication and their impact on cancer incidence: a mixed treatment comparison meta-analysis of randomized controlled trials, J Hypertens 26 (2008) 622–629. C.I. Li, J.R. Daling, M.T.C. Tang, K.L. Haugen, P.L. Porter, K.E. Malone, Use of antihypertensive medications and breast cancer risk among women aged 55 to 74 years, JAMA Intern Med 173 (2013) 1629–1637. M.J. Brown, Hypertension and ethnic group, Br Med J 332 (2006) 833-6. M.C.S. Wong, W.W.S. Tam, C.S.K. Cheung, H.H.X. Wang, E.L.H. Tong, A.C.H. Sek, et al., Drug adherence and the incidence of coronary heart disease- and stroke-specific mortality among 218,047 patients newly prescribed an antihypertensive medication: a five-year cohort study, Int J Cardiol 168 (2013) 928–933. M.C.S. Wong, W.W.S. Tam, C.S.K. Cheung, E.L.H. Tong, A.C.H. Sek, N.T. Cheung, et al., Medication adherence to first-line antihypertensive drug class in a large Chinese population, Int J Cardiol 167 (2013) 1438–1442. M.C.S. Wong, J.Y. Jiang, J.L. Tang, A. Lam, H. Fung, S.W. Mercer, Health services research in the public healthcare system in Hong Kong: an analysis of over 1 million antihypertensive prescriptions between 2004–2007 as an example of the potential and pitfalls of using routinely collected electronic patient data, BMC Health Serv Res 8 (2008). M.C.S. Wong, J.Y. Jiang, S.M. Griffiths, Factors associated with compliance to thiazide diuretics among 8551 Chinese patients, J Clin Pharm Ther 36 (2011) 179–186. M.C.S. Wong, J.Y. Jiang, S.M. Griffiths, Adherence to combination therapy among ethnic Chinese patients: a cohort study, Hypertens Res 33 (2010) 416–421. M.C.S. Wong, J.Y. Jiang, S.M. Griffiths, Short-term adherence to beta-blocker therapy among ethnic Chinese patients with hypertension: a cohort study, Clin Ther 31 (2009) 2170–2177. M.C.S. Wong, J.Y. Jiang, S.M. Griffiths, Antihypertensive drug adherence among 6408 Chinese patients on angiotensin-converting enzyme inhibitors in Hong Kong: a cohort study, J Clin Pharmacol 50 (2010) 598–605. M.C.S. Wong, J.Y. Jiang, S.M. Griffiths, Factors associated with compliance, discontinuation and switching of calcium channel blockers in 20,156 Chinese patients, Am J Hypertens 22 (2009) 904–910. W.J. Sun, C.M. Schooling, W.M. Chan, K.S. Ho, T.H. Lam, The association between depressive symptoms and mortality among Chinese elderly: a Hong Kong cohort study, J Gerontol A Biol Sci Med Sci 66 (2011) 459–466.
M.C.S. Wong et al. / International Journal of Cardiology 179 (2015) 178–185 [36] C.M. Schooling, T.H. Lam, Z.B. Li, S.Y. Ho, W.M. Chan, K.S. Ho, et al., Obesity, physical activity, and mortality in a prospective Chinese elderly cohort, Arch Intern Med 166 (2006) 1498–1504. [37] T.H. Lam, S.Y. Ho, A.J. Hedley, K.H. Mak, R. Peto, Mortality and smoking in Hong Kong: case–control study of all adult deaths in 1998, Br Med J 323 (2001) 361–362. [38] N.K. Choudhry, W.H. Shrank, R.L. Levin, J.L. Lee, S.A. Jan, M.A. Brookhart, et al., Measuring concurrent adherence to multiple related medications, Am J Manag Care 15 (2009) 457–464. [39] B.C. Martin, E.K. Wiley-Exley, S. Richards, M.E. Domino, T.S. Carey, B.L. Sleath, Contrasting measures of adherence with simple drug use, medication switching, and therapeutic duplication, Ann Pharmacother 43 (2009) 36–44. [40] S.E. Andrade, K.H. Kahler, F. Frech, K.A. Chan, Methods for evaluation of medication adherence and persistence using automated databases, Pharmacoepidemiol Drug Saf 15 (2006) 565–574 [discussion 75–7]. [41] G. Mazzaglia, E. Ambrosioni, M. Alacqua, A. Filippi, E. Sessa, V. Immordino, et al., Adherence to antihypertensive medications and cardiovascular morbidity among newly diagnosed hypertensive patients, Circulation 120 (2009) 1598–1605. [42] J.N. Rasmussen, A. Chong, D.A. Alter, Relationship between adherence to evidencebased pharmacotherapy and long-term mortality after acute myocardial infarction, JAMA 297 (2007) 177–186.
185
[43] T.A. Burke, M.C. Sturkenboom, S.E. Lu, C.E. Wentworth, Y. Lin, G.G. Rhoads, Discontinuation of antihypertensive drugs among newly diagnosed hypertensive patients in UK general practice, J Hypertens 24 (2006) 1193–1200. [44] G. Mazzaglia, L.G. Mantovani, M.C.J.M. Sturkenboom, A. Filippi, G. Trifiro, C. Cricelli, et al., Patterns of persistence with antihypertensive medications in newly diagnosed hypertensive patients in Italy: a retrospective cohort study in primary care, J Hypertens 23 (2005) 2093–2100. [45] M.C. Wong, W.W. Tam, C.S. Cheung, E.L. Tong, A.C. Sek, G. John, et al., Initial antihypertensive prescription and switching: a 5 year cohort study from 250,851 patients, PLoS One 8 (2013) e53625. [46] I.A. Dhalla, T. Gomes, Z. Yao, J. Nagge, N. Persaud, C. Hellings, et al., Chlorthalidone versus hydrochlorothiazide for the treatment of hypertension in older adults: a population-based cohort study, Ann Intern Med 158 (2013) 447–455. [47] L.H. Curtis, B.G. Hammill, E.L. Eisenstein, J.M. Kramer, K.J. Anstrom, Using inverse probability-weighted estimators in comparative effectiveness analyses with observational databases, Med Care 45 (2007) S103–S107. [48] R.B. D'Agostino, Propensity scores in cardiovascular research, Circulation 115 (2007) 2340–2343. [49] S. Schneeweiss, J.A. Rassen, R.J. Glynn, J. Avorn, H. Mogun, M.A. Brookhart, Highdimensional propensity score adjustment in studies of treatment effects using health care claims data, Epidemiology 20 (2009) 512–522.
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
The incidence of cancer deaths among hypertensive patients in a large Chinese population: A cohort study Martin C.S. Wong a,b,1, Wilson W.S. Tam a,b,1, X.Q. Lao a,b,1, Harry H.X. Wang a,b,1, Mandy W.M. Kwan a,1, Clement S.K. Cheung c,1, Ellen L.H. Tong c,1, N.T. Cheung c,1, Sian M. Griffiths a,b,⁎,1, Andrew J.S. Coats d,e,1 a
JC School of Public Health and Primary Care, Faculty of Medicine, Chinese University of Hong Kong, Shatin, NT, Hong Kong CUHK Shenzhen Research Institute, Chinese University of Hong Kong, No. 10, 2nd Yuexing Road, Nanshan District, Shenzhen, PR China Hospital Authority Information Technology Services — Health Informatics Section, Suite Nos. 2–15, 7/F Exchange Tower, 33 Wang Chiu Road, Kowloon Bay, Hong Kong d Department of Medicine and Therapeutics, Faculty of Medicine, The Chinese University of Hong Kong, Hong Kong e Monash University, Australia and the University of Warwick, UK b c
a r t i c l e
i n f o
Article history: Received 25 May 2014 Received in revised form 4 September 2014 Accepted 14 October 2014 Available online 20 October 2014 Keywords: Cancer mortality Hypertensive patients Antihypertensive agents Associated factors Chinese population
a b s t r a c t Current evidence is mixed regarding the association between antihypertensive prescriptions and cancer mortality. We evaluated this association in a large Chinese hypertensive population. We followed for five years all patients who were prescribed their first-ever antihypertensive agents between 2001 and 2005 in a public healthcare sector of Hong Kong. The association between antihypertensive drug class and cancer mortality was evaluated by Cox proportional hazard models with propensity score matching. Age, gender, socioeconomic status, service settings, district of residence, proportion of days covered reflecting medication adherence, and the number of comorbidities were adjusted. From 217,910 eligible patients, 9500 (4.4%) died from cancer within five years after their first-ever antihypertensive prescription. Most cancer deaths occurred in the digestive (38.9%) and respiratory system (30.4%); the breast (6.2%); and the lympho-hematopoietic tissues (5.3%). The proportion of patients who died from cancer was the highest in the calcium channel blocker (CCB) group (6.5%), followed by thiazide diuretics (4.4%), angiotensin converting enzyme inhibitors (4.2%) and β-blockers (2.6%). When compared with β-blockers, patients prescribed CCBs (Adjusted Hazard Ratio [AHR] = 1.406, 95% C.I. 1.334–1.482, p < 0.001) were more likely to die from cancer. Thiazide users were also more likely to suffer from cancer deaths (AHR = 1.364, 95% C.I. 1.255–1.483, p < 0.001), but became insignificant in stratified analysis. The association between cancer mortality and use of CCB, and perhaps thaizide, may alert physicians to the need for more meticulous and comprehensive care of these patients in clinical practice. We recommend prospective studies to evaluate cause-and-effect relationships of these associations. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Hypertension is the biggest contributor to the burden of cardiovascular disease, accounting for 54% of stroke and 47% of ischemic heart disease worldwide [1]. Its prevalence is growing in various regions [2, 3] whilst control rates remain poor in many countries, including China [4]. Antihypertensive therapies have been recognized as effective agents to control blood pressure [5,6]. Most international guidelines recommend any major antihypertensive drug classes as suitable first-line medications for management of arterial hypertension, including the
⁎ Corresponding author at: 2/F, School of Public Health and Primary Care, Prince of Wales Hospital, Chinese University of Hong Kong, Shatin, NT, Hong Kong. Tel.: + 852 2252 8702; fax: +852 2606 3500. E-mail address: siangriffi[email protected] (S.M. Griffiths). 1 Statement of authorship: All authors take responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation.
http://dx.doi.org/10.1016/j.ijcard.2014.10.028 0167-5273/© 2014 Elsevier Ireland Ltd. All rights reserved.
Joint National Committee seventh report [7] and the 2009 reappraisal of the hypertension European guidelines [8]. The majority of the scientific literature on the subject has compared the different antihypertensive drug classes according to blood pressure lowering efficacy and cardiovascular events and mortality [5,9] from randomized trials and cohort studies, and most have found little difference between various antihypertensive classes. However, in the most recent decade studies have emerged based on other outcomes, including cancer. For instance, Pahor and colleagues [10,11] found that users of Calcium Channel Blockers (CCBs) had 72% higher risks of developing cancer than non-users among more than 5000 elderly patients in the US. These findings were echoed by Fitzpatrick and colleagues who reported a higher risk of incident invasive breast carcinoma among CCB users from the US Cardiovascular Health Study (CHS) — showing a clear dose response relationship between CCB and cancer [12]. In addition, the combined use of Angiotensin Converting Enzyme Inhibitors (ACEIs) and Angiotensin Receptor
M.C.S. Wong et al. / International Journal of Cardiology 179 (2015) 178–185
Blockers (ARBs) have been shown to have higher cancer risk than other classes from a network meta-analysis of 70 randomized trials, involving more than 320,000 patients [13]. Nevertheless, these findings were not supported by subsequent observational studies [14–23], and a mixed treatment comparison meta-analysis [24] also confirmed that no excess incidence of cancer could be imputed to CCBs and other antihypertensive classes. Interestingly, however, a recent population-based case–control study in the three-county Seattle–Puget Sound metropolitan area conducted among women aged 55 to 74 years found that current use of CCBs was associated with higher risks of both ductal and lobular breast cancer, whilst other antihypertensive drug classes were not associated with risk [25]. Therefore, current evidence is rather mixed on the association between antihypertensive drug class and cancer as an important outcome. Also, these studies included only modest numbers of individuals and might not have been adequately powered to capture mortality due to cancer. Besides, they were almost exclusively conducted among Western populations, limiting generalizability to other patient groups such as the Chinese. It is well recognized that the pharmacological action of antihypertensive drugs differs according to ethnicity, and even small ethnic differences in its optimum management could have large implications for health resources [26]. Chinese people make up one-fifth of the world's population, residing in different parts of the globe. Hence, the objectives of this study are to test the a priori hypothesis that the incidence of deaths due to cancer was similar among the major antihypertensive drug classes when prescribed as first-line agents in large Chinese population.
179
2.2. Data source The present study retrieved data from a comprehensive computerized database known as electronic Clinical Management System adopted by the Hospital Authority in Hong Kong. It captures patients' demographic and prescription profiles, and clinical diagnoses in the forms of the International Classification of Diseases (ICD-9) or the International Classification of Primary Care (ICPC-2). The database has been previously reported with detailed descriptions on its validity [27–34], which demonstrated a high level of completeness with respect to the socio-demographic information (100%) and prescription profiles (99.98%) [29]. All drug prescriptions are entered by the attending physicians and are double-checked by the dispensing pharmacists using standardized procedures, and any subsequent changes in patients' prescription after the initial consultations are also entered into the system. This computerized record system serves as the sole portal which allows cross-referencing of information by physicians at patient visits to all public clinical settings in different districts. The present study employed a sampling frame which included the entire Hong Kong population, i.e., more than 7,000,000 as of the year 2013. Hong Kong is divided into three regions, namely the Hong Kong Island, Kowloon, and the New Territories, from the most urbanized to the most rural regions, respectively. 2.3. Study population
2. Methods
Data of all patients who attended any public practice and who received their first-ever antihypertensive prescription during the calendar years 2001 to 2005 (the index date) from the database were retrieved. Patients who were prescribed with any antihypertensive agents prior to the index date were excluded. We treated deaths due to medical conditions other than cancer within the observation period as being censored. Each patient was classified into one drug group according to the initial prescription of antihypertensive agents, which include β-blockers, thiazide diuretics, CCB and ACEIs. It was found in the database that patients who received ARBs (n = 434, 0.3%) and combination therapy (n = 74, 0.0%) were relatively rare so for that reason such patients were excluded in the data analysis. Concomitant comorbidities include cardiovascular risk factors and conditions which might confound the choice of antihypertensive medication, as indicated by the respective ICD-9 or ICPC-2 codes. Patients were followed up until the occurrence of mortality within the study period or at the end of five years, whichever is earlier. The original dataset consisted of 223,257 subjects and hence 5347 subjects were excluded due to the exclusion criteria mentioned above. The flow of participants was shown in Fig. 1.
2.1. Ethics statement
2.4. Outcome measures and covariates
The ethics clearance of the study was obtained from the Clinical Ethics Research Committee of the Hospital Authority, and the Survey and Behavioral Research Ethics Committee of The Chinese University of Hong Kong. Informed consent was not necessary as all subjects were anonymized with unique identity numbers. The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki.
The primary outcome variables were: (1) All-cause mortality (ICD-10 C00-D49) and (2) deaths due to cancers affecting the gastrointestinal tract or the lungs, respectively. The vast majority of all deaths in Hong Kong occurred in the hospitals, and it allows accurate case ascertainment [35]. Death records in hospitals have been utilized as a valid proxy of patient mortality in a number of previous studies, especially for Chinese populations
Fig. 1. Flow of participants.
180
M.C.S. Wong et al. / International Journal of Cardiology 179 (2015) 178–185
Table 1 Baseline characteristics of patients (N = 217,910) by 5-year cancer mortality.
Gender Male Female Age <50 50–59 60–69 ≥70 Public assistance Yes No Service type on first visit In-/day-patient clinic Special out-patient clinic Accident and emergency General outpatient clinics Others District of residence Hong Kong Kowloon New Territories First prescription ACEIs Alpha blocker Beta blocker CCB Thiazide Proportion Days Covered (PDC) <40% ≥40% Co-morbidity 0 1 2 ≥3
N
All cancer deaths (N = 9500)
Digestive cancer deaths (N = 3697)
Respiratory cancer deaths (N = 2885)
98,192 119,717
5854 (6.0%) 3646 (3.0%)
2412 (2.5%) 1285 (1.1%)
1926 (2.0%) 959 (0.8%)
61,343 42,018 41,602 72,928
562 (0.9%) 1070 (2.5%) 1986 (4.8%) 5882 (8.1%)
217 (0.4%) 465 (1.1%) 824 (2.0%) 2191 (3.0%)
87 (0.1%) 228 (0.5%) 598 (1.4%) 1972 (2.7%)
32,800 183,045
7361 (4.0%) 1970 (6.0%)
2856 (1.6%) 771 (2.4%)
2213 (1.2%) 617 (1.9%)
65,725 64,679 16,898 64,715 5885
6243 (9.5%) 1510 (2.3%) 306 (1.8%) 1298 (2.0%) 143 (2.4%)
2415 (3.7%) 631 (1.0%) 119 (0.7%) 472 (0.7%) 60 (1.0%)
1935 (2.9%) 415 (0.6%) 77 (0.5%) 410 (0.6%) 48 (0.8%)
38,301 73,769 105,840
1844 (4.8%) 3691 (5.0%) 3965 (3.7%)
731 (1.9%) 1400 (1.9%) 1566 (1.5%)
541 (1.4%) 1190 (1.6%) 1154 (1.1%)
22,970 13,452 98,586 65,465 17,437
961 (4.2%) 903 (6.7%) 2588 (2.6%) 4278 (6.5%) 770 (4.4%)
363 (1.6%) 293 (2.2%) 1283 (1.3%) 1468 (2.2%) 290 (1.7%)
291 (1.3%) 308 (2.3%) 559 (0.6%) 1503 (2.3%) 224 (1.3%)
71,598 14,6312
3212 (4.5%) 6288 (4.3%)
1261 (1.8%) 2436 (1.7%)
942 (1.3%) 1943 (1.3%)
131,006 70,872 14,574 1458
5236 (4.0%) 3487 (4.9%) 717 (4.9%) 60 (4.1%)
2244 (1.7%) 1192 (1.7%) 237 (1.6%) 24 (1.6%)
1375 (1.0%) 1222 (1.7%) 268 (1.8%) 20 (1.4%)
All proportions were across rows. Chi-square tests among each categories of all variables showed p < 0.001. PDC is defined as the number of days when medication is supplied in a specified period divided by the total number of days within the observation period. ACEI: Angiotensin converting enzyme inhibitors; CCB: calcium channel blockers. Patients who visited the clinic for once only were not included.
whose deaths almost always occur in hospitals [35–37]. The cause of death for each patient was defined according to the primary cause of mortality for each patient, as determined by the physician-in-charge when death was registered in the death certificate. The predictor variable in this study was the major antihypertensive drug classes on its initial prescription. To take into account the influence of medication adherence as a potential confounder, the study included the measure of interval-based proportion of days covered (PDC) as a covariate, which has been widely used in large database study as an internationally accepted metric for drug adherence assessment [38–40]. The interval-based PDC refers to the number of days when medication is supplied in a specified period divided by the total number of days within the period. The PDC was treated as a continuous variable in the analysis and was estimated using a five-year interval, i.e., for patients who died during the five-year period, the PDC was estimated using the time period between the index date and the incidence of mortality. Similar to the internationally recognized approach [41], drug adherence in the present study was divided into two levels, namely: high or intermediate (PDC ≥ 0.40) and low (PDC < 0.40) [40,42]. In addition, we also evaluated the rates of medication discontinuation and switching among these patients. Drug switching is defined as the absence of a refill prescription in all subsequent clinic visits combined with the prescription of another antihypertensive drug of a different class since the date of the first prescription, whereas drug discontinuation refers to the absence of repeat prescription of the same drug class after the first-ever antihypertensive prescription date [43–45]. The independent variables that were controlled for included patients' age, gender, payment status (recipients vs non-recipients of comprehensive social security assistance, CSSA; each consultation costs approximately US $5.77, including investigation and prescription fees), clinic types visited (general outpatient clinic [GOPC], specialist outpatient clinics [SOPCs], in- and day-patient clinics, Accident and Emergency Departments [AEDs], and other clinic types), their districts of residence (Hong Kong Island vs Kowloon vs the New Territories; from the most urbanized to the most rural regions, respectively), and the total number of comorbidities (as reflected by the number of ICD-9 codes). The same list of comorbidities reflected by their ICD-9 and ICPC-2 codings was used in the present study as published elsewhere [28], which were categorized under “Cardiovascular diseases” (24.3%), “Diabetes or impaired glucose tolerance” (23.0%), “Respiratory diseases” (14.6%) and “Renal diseases” (11.0%).
2.5. Statistical analysis The Statistical Package for Social Sciences version 16.0 (SPSS, Inc.) was used for all data analysis. For descriptive analysis, Student's t tests were used to compare continuous variables, and chi-square tests of heterogeneity were used to compare categorical variables. Three separate Cox-proportional hazard models were developed to explore the association between initial antihypertensive drug class and mortality due to cancer at any site, at the gastrointestinal tract, and at the lungs, respectively. To account for the confounding effects of medication change, all study participants were censored if they switched study medications, discontinued treatment (in which case they were followed for 100 days from their last prescription to identify events that may have precipitated discontinuation), commenced treatment with a different antihypertensive agent, or reached the end of the study. This is according to a methodology adopted by Dhalla et al. [46], and we followed patients for a maximum of 5 years. Also, we conducted a sensitivity analysis where patients whose antihypertensive drug class has been switched or added on by another drug class were excluded, and a similar regression model was constructed to detect if there were any changes in the associated factors identified. In addition, a propensity score model was developed to estimate weightings for the selection of the two prescriptions, given the possible existence of treatment indication bias due to the observational nature of the study. The variables used in the calculation of the propensity scores consisted of all the covariates, namely patients' age, gender, receipt of public assistance, service type on first visit, district of residence, the drug class of first prescription, the Proportion Days Covered, and the number of comorbidities. Therefore, the analysis was performed with adjusted estimates weighted by the inverse estimated propensity scores in three Cox proportion hazard models according to standard statistical methodology [47,48]. To further minimize the influence of indication bias due to different baseline characteristics of patients, a high-dimensional propensity score matching to compare patients with similar observed characteristics was adopted using standardized methodology as utilized in other literature [46,49]. All the variables listed above were entered unconditionally into the model. Since patients with prostate cancer might be prescribed with alpha-blockers for prostate symptoms before or after diagnosis, we conducted two separate analyses where patients received alpha blockers as first-line antihypertensive agents were included and excluded, separately, and the significant associated factors were compared between the two analyses. Since the study involved multiple
M.C.S. Wong et al. / International Journal of Cardiology 179 (2015) 178–185
181
Table 2 Baseline Characteristics of Patients (N = 217,910) by First Prescription.
Gender Male Female Age <50 50–59 60–69 ≥70 Public assistance Yes No Service type on first visit In-/day-patient clinic Special out-patient clinic Accident and emergency General outpatient clinics Others District of residence Hong Kong Kowloon New Territories Proportion Days Covered (PDC) <40% ≥40% Co-morbidity 0 1 2 ≥3
N
ACEI N = 22,970
Alpha-blockers N = 13,452
Beta-blockers N = 98,585
CCB N = 65,465
Thiazide N = 17,437
98,192 119,717
11,694 (50.9%) 11,276 (49.1%)
13,131 (97.6%) 321 (2.4%)
37,669 (38.2%) 60,916 (61.8%)
29,394 (44.9%) 36,071 (55.1%)
6304 (36.2%) 11,133 (63.8%)
61,343 42,018 41,602 72,928
4598 (20.0%) 4232 (18.4%) 4530 (19.7%) 9607 (41.8%)
951 (7.1%) 2233 (16.6%) 4041 (30.0%) 6227 (46.3%)
43,740 (44.4%) 21,693 (22.0%) 15,454 (15.7%) 17,695 (17.9%)
9318 (14.2%) 10,377 (15.9%) 13,712 (20.9%) 32,047 (49.0%)
2736 (15.7%) 3483 (20.0%) 3865 (22.2%) 7352 (42.2%)
32,800 183,045
3798 (16.7%) 18,940 (83.3%)
2635 (19.9%) 10,599 (80.1%)
11,000 (11.3%) 86,647 (88.7%)
12,585 (19.4%) 52,320 (80.6%)
2782 (16.1%) 14,539 (83.9%)
65,725 64,679 16,898 64,715 5885
8803 (38.3%) 8092 (35.2%) 227 (1.0%) 5164 (22.5%) 683 (3.0%)
3200 (23.8%) 3820 (28.4%) 635 (4.7%) 5263 (39.1%) 533 (4.0%)
22,783 (23.1%) 37,633 (38.2%) 9124 (9.3%) 26,342 (26.7%) 2702 (2.7%)
27,188 (41.5%) 11,580 (17.7%) 5417 (8.3%) 19,858 (30.3%) 1418 (2.2%)
3751 (21.5%) 3554 (20.4%) 1495 (8.6%) 8088 (46.4%) 549 (3.1%)
38,301 73,769 105,840
5542 (22.8%) 8069 (35.1%) 9659 (42.1%)
1509 (11.2%) 5973 (44.4%) 5970 (44.4%)
16,964 (17.2%) 32,299 (32.8%) 49,323 (50.0%)
11,078 (16.9%) 23,073 (35.2%) 31,314 (47.8%)
3508 (20.1%) 4355 (25.0%) 9574 (54.9%)
71,598 146,312
4298 (18.7%) 18,762 (81.3%)
5040 (37.5%) 8412 (62.5%)
37,068 (37.6%) 61,518 (62.4%)
19,614 (30.0%) 45,851 (70.0%)
5578 (32.0%) 11,859 (68.0%)
131,006 70,872 14,574 1458
8067 (35.1%) 11,578 (50.4%) 2969 (12.9%) 356 (1.5%)
4550 (33.8%) 6474 (48.1%) 2228 (16.6%) 200 (1.5%)
73,559 (74.6%) 21,906 (22.2%) 2906 (2.9%) 215 (0.2%)
34,438 (52.6%) 24,955 (38.1%) 5479 (8.4%) 593 (0.9%)
10,392 (59.6%) 5959 (34.2%) 992 (5.7%) 94 (0.5%)
All proportions were across columns. PDC is defined as the number of days when medication is supplied in a specified period divided by the total number of days within the observation period. For all covariates, p < 0.01 from Chi-square test for association. ACEI: Angiotensin converting enzyme inhibitors; CCB: calcium channel blockers. Patients who visited the clinic for once only were not included.
comparisons, a “definite” association between an antihypertensive drug class with cancer deaths was established only if the drug class was found to be significantly associated with all three cancer mortality. All p values b0.05 were considered statistically significant.
3. Results 3.1. Patient characteristics From a total of 217,910 eligible patients included in the present study (Table 1), more than half were female patients (54.9%) and were aged 60 years or older (52.6%). Only a small proportion (15.2%) of all patients received public assistance and the majority (59.8%) was followed up at in- or day-patient clinics and specialized out-patient
clinics. The majority of patients lived in the more rural regions (48.6% and 33.9% in the New Territories and Kowloon, respectively). The most commonly prescribed antihypertensive drugs were β-blockers (45.2%) and CCBs (30.0%), followed by ACEIs (10.5%) and thiazide diuretics (8.0%). The average PDC was 0.73 (SD 0.90) which implied intermediate to good medication adherence overall, and most patients had one comorbidity or less (92.6%). Except α-blockers which might be prescribed to patients with prostatism, the proportion of cancer deaths was the highest among CCB users (6.5%) when compared with other medication users (2.6%, 4.2% and 4.4% for β-blockers, ACEIs, and thiazide diuretics). This was similar for digestive and respiratory cancer mortality. Table 2 shows the baseline characteristics of the patients according to
Fig. 2. The distribution of cancer mortality according to body systems.
182
M.C.S. Wong et al. / International Journal of Cardiology 179 (2015) 178–185
Table 3 Hazard ratios of cancer mortality by Cox proportional hazards regression analysis.
Gender Male Female Age <50 50–59 60–69 ≥70 Public assistance Yes No Service type on first visit In-/day-patient clinic Special out-patient clinic Accident and emergency General outpatient clinics District of residence Hong Kong Kowloon New Territories First Prescription ACEIs Beta blocker CCB Thiazide Proportion Days Covered (PDC) <40% ≥40% Co-morbidity No Yes
All cancer (N = 8473)
Digestive cancer (N = 3349)
Respiratory cancer (N = 2538)
1.000 0.537 (0.514, 0.561)
1.000 0.437 (0.407, 0.470)
1.000 0.435 (0.401, 0.473)
1.000 2.938 (2.644, 3.263) 4.839 (4.386, 5.340) 7.460 (6.798, 8.186)
1.000 3.538 (2.998, 4.174) 5.879 (5.027, 6.876) 8.970 (7.729, 10.41)
1.000 3.607 (2.797, 4.652) 7.909 (6.261, 9.991) 13.22 (10.55, 16.56)
0.930 (0.880, 0.983) 1.000
1.017 (0.933, 1.110) 1.000
0.865 (0.783, 0.955) 1.000
5.426 (5.075, 5.800) 1.643 (1.514, 1.783) 1.129 (0.986, 1.293) 1.000
5.645 (5.069, 6.286) 1.760 (1.547, 2.003) 1.114 (0.894, 1.387) 1.000
4.943 (4.385, 5.573) 1.546 (1.328, 1.799) 1.025 (0.792, 1.327) 1.000
1.000 0.983 (0.926, 1.043) 0.884 (0.834, 0.937)
1.000 0.937 (0.853, 1.030) 0.892 (0.814, 0.978)
1.000 1.090 (0.978, 1.215) 0.886 (0.795, 0.987)
0.952 (0.881, 1.029) 1.000 1.406 (1.334, 1.482) 1.364 (1.255, 1.483)
0.758 (0.671, 0.856) 1.000 0.994 (0.917, 1.078) 1.048 (0.918, 1.197)
1.136 (0.978, 1.319) 1.000 2.041 (1.841, 2.262) 1.710 (1.457, 2.006)
1.000 0.852 (0.814, 0.893)
1.000 0.844 (0.784, 0.908)
1.000 0.889 (0.816, 0.968)
1.000 0.791 (0.756, 0.828)
1.000 0.647 (0.601, 0.696)
1.000 0.940 (0.866, 1.019)
PDC is defined as the number of days when medication is supplied in a specified period divided by the total number of days within the observation period. ACEI: Angiotensin converting enzyme inhibitors; CCB: calcium channel blockers. Patients who were prescribed with alpha-blockers or seen in service settings other than those listed were excluded from the analysis.
the first antihypertensive prescriptions. CCB users were older, and higher proportions attended in- or day-patient clinics. The number of comorbidites was also substantially different among the various drug classes. 3.2. Profiles of mortality due to cancer and respiratory disease A total of 9500 patients (4.4%) died from cancer within five years after their first-ever antihypertensive prescription. Most cancer deaths occurred in the digestive (38.9%) and respiratory system (30.4%); the breast (6.2%); and the lymphatic and hematopoietic tissue (5.3%) (Fig. 2).
1.334–1.482; p < 0.001) and respiratory cancer (AHR = 2.041, 95% C.I. 1.841–2.262, p < 0.001). In addition, thiazide users were also more likely to suffer from any cancer mortality (AHR = 1.364, 95% C.I. 1.255–1.483, p < 0.001) and respiratory cancer (AHR = 1.710, 95% C.I. 1.457–2.006, p < 0.001). Owing to the substantial difference in the distribution of service setting according to drug class, a stratified analysis according to service setting was conducted (Table 4). It was shown that CCB users were significantly more likely to suffer from mortality due to all cancers in all the settings (AHR ranged from 1.179 to 1.458). Thiazide diuretics was associated with higher mortality from all cancers in the in- or day-patient clinics, SOPCs but not GOPCs. 3.4. Sensitivity analyses
3.3. Factors associated with cancer mortality From Cox proportional hazard analysis with all cancer mortality as the outcome measure, (Table 3), it was found that advanced age (≥50 years; adjusted hazard ratio [AHR] ranged from 2.938 to 7.460, p < 0.001; referent <50 years), male gender (AHR for female = 0.537, 95% C.I. 0.514–0.561, p < 0.001), non-recipients of public assistance (AHR for receiver = 0.930, 95% C.I. 0.880–0.983, p < 0.001), attendance in in-patient or day-patient service (AHR = 5.426, 95% C.I. 5.075–5.800, p < 0.001), residence in more urbanized regions (AHR for the New Territories = 0.884, 95% C.I. 0.834–0.937, p < 0.001), lower levels of medication adherence (AHR = 0.852, 95% C.I. 0.814–0.893, p < 0.001 for PDC < 40%) and the absence of comorbidity (AHR for at least one comorbidity = 0.791, 95% C.I. 0.756–0.828, p < 0.001) were positively associated with mortality from all cancers. These findings were similar for digestive cancers and respiratory cancer (Table 3). When compared with patients who were prescribed with β-blockers, users of CCBs were more likely to die from any cancer (AHR = 1.406, 95% C.I.
An additional analysis where patients whose antihypertensive drug class has been switched or added on by another drug class were excluded reported the same findings (Table 5). In addition, a propensity score model matched for drug class and a high-dimensional propensity score matching methodology yielded similar results. Two separate analyses where patients received α-blockers as first-line antihypertensive agents were included and excluded also gave similar findings. 4. Discussion 4.1. Statement of principal findings From this observational study involving 217,910 Chinese hypertensive patients, we found that first-line prescription of CCBs as a firstline antihypertensive agent was associated with a higher likelihood of death from any cancer and respiratory cancer, when compared with users of β-blockers. This is consistently true from various sensitivity
M.C.S. Wong et al. / International Journal of Cardiology 179 (2015) 178–185
183
Table 4 Hazard ratios of cancer mortality by Cox proportional hazards regression analysis: sensitivity analysis according to the service setting. In-/day-patient clinic
All cancer (N = 5789)
Digestive cancer (N = 2265)
Respiratory cancer (N = 1768)
First prescription ACEIs Beta blocker CCB Thiazide
0.860 (0.781, 0.948) 1.000 1.458 (1.369, 1.553) 1.352 (1.209, 1.511)
0.616 (0.528, 0.718) 1.000 0.938 (0.852, 1.032) 0.990 (0.831, 1.181)
1.128 (0.935, 1.360) 1.000 2.383 (2.049, 2.711) 1.823 (1.468, 2.265)
Specialist out-patient clinics
All cancer (N = 1310)
Digestive cancer (N = 563)
Respiratory cancer (N = 351)
First prescription ACEIs Beta blocker CCB Thiazide
1.122 (0.950, 1.324) 1.000 1.179 (1.029, 1.351) 1.440 (1.192, 1.738)
0.904 (0.701, 1.168) 1.000 0.877 (0.710, 1.082) 1.100 (0.815, 1.484)
1.517 (1.097, 2.099) 1.000 1.780 (1.363, 2.324) 1.999 (1.396, 2.863)
General out-patient clinics
All cancer (N = 1103)
Digestive cancer (N = 418)
Respiratory cancer (N = 347)
First prescription ACEIs Beta blocker CCB Thiazide
1.265 (1.016, 1.574) 1.000 1.209 (1.046, 1.398) 1.136 (0.939, 1.376)
1.677 (1.198, 2.348) 1.000 1.224 (0.966, 1.553) 1.113 (0.813, 1.524)
0.962 (0.634, 1.461) 1.000 1.217 (0.942, 1.573) 1.041 (0.733, 1.478)
ACEI: Angiotensin converting enzyme inhibitors; CCB: calcium channel blockers.
analyses. Thiazide diuretics were not associated with cancer mortality outcomes in the GOPC setting, and hence we did not establish a definite association with cancer mortality for this drug class. Owing to the retrospective nature of this study, we did not conclude a cause-and-effect association between CCBs and cancer mortality.
settings. Since it is not an evaluation of efficacy, caveats should be observed when direct comparisons on the association between antihypertensive drug class and the mortality outcomes were made.
4.2. Study limitations
Previous studies reporting positive [10–12,25] or negative [13–24] associations between CCB use and cancer were mostly observational studies, and they were relatively small-scaled, except two systematic reviews of randomized controlled trials [13,24]. The study by Coleman and colleagues [24] involved more than 126,000 patients by mixed treatment comparison meta-analysis and showed no differences in cancer among different antihypertensive drug classes. The odds ratio of cancer incidence in the CCB group was 0.95 (95% C.I. 0.79–1.13) when compared with placebo or the untreated group. Another large-scale network meta-analysis conducted by Bangalore et al. [13] involving more than 324,000 patients from 70 RCTs also did not show any significant difference — and the odds ratio of cancer development for the CCB group was 1.05 (95% C.I. 0.96–1.13) when compared with placebo. As these studies included different population sub-groups, there exists a possibility that some ethnic groups could be under-represented. Our study is focused on “first-line prescription” as an independent variable without examining concomitant use of other antihypertensive agents and the switching or discontinuation of CCBs in the period of observation. Therefore, direct comparison of this study with previous literature is not appropriate [10–25]. However, it implied that among patients who received CCBs as first-line antihypertensive agents, there exists an association with cancer mortality, irrespective of the patients' subsequent prescription journey. The same can be said for thiazide diuretics in certain service settings — but the mechanism why it was associated with greater cancer mortality remains to be explored. It is also noteworthy that those with comorbidities had lower risk of cancer mortality. We speculated that patients with comorbid conditions may be more health-
To our knowledge this is the largest study which compared the various antihypertensive drug classes with respect to the controversial outcomes of mortality due to cancer. The highly complete, accurate territory-wide database which captured clinical information from over 218,000 patients observed for more than five years, in addition to the good dispensing practice of the clinics in the public healthcare sector further enhanced the reliability of the findings. However, some of the limitations should be addressed here. Firstly, we assume that patients prescribed with the medications were actually taking them, although the analysis has controlled for medication adherence using internationally recognized metrics. In addition, there were some confounders which could not be taken into account, such as smoking, drinking, significant family history and other lifestyle risk factors for both outcomes; and that the number of comorbidities was used as a proxy measure for patients' general well-being. There exists indication bias affecting the choice of antihypertensive drug class in observational studies, and critics might argue that not all patients prescribed with antihypertensive agents had hypertension but may have had other medical conditions. These include diseases like heart failure and coronary heart diseases, as well as complaints due to symptoms like tachycardia, prostatism and peripheral edema. This study relied completely on disease coding as the only criteria to define patient comorbidities, and physicians might miss the entry of these variables during clinical consultations. Also, this is an effectiveness study observing mortality outcomes according to first-line prescription of drug class in real-life clinical
4.3. Relationship with existing literature and explanation of findings
Table 5 Hazard ratios of cancer mortality by Cox proportional hazards regression analysis: sensitivity analysis according to the first prescription. First prescription
All cancer
Digestive cancer
Respiratory cancer
ACEIs (N = 14,496) Beta blocker (N = 61,061) CCB (N = 40,052) Thiazide (N = 8069)
0.849 (0.775, 0.929) 1.000 1.306 (1.228, 1.388) 1.396 (1.206, 1.547)
0.653 (0.567, 0.751) 1.000 0.853 (0.777, 0.937) 1.009 (0.856, 1.188)
1.093 (0.917, 1.302) 1.000 2.153 (1.902, 2.437) 2.065 (1.701, 2.507)
ACEI: angiotensin converting enzyme inhibitors; CCB: calcium channel blockers.
184
M.C.S. Wong et al. / International Journal of Cardiology 179 (2015) 178–185
conscious, in terms of the need for self-management of those chronic conditions and implementation of more intensive lifestyle measures. In addition, the comorbidities themselves could act as competing causes of patient deaths — and hence reducing mortality attributable to cancer. A further interesting observation from this study included an increased risk of cancer mortality with a lower level of medication adherence. We are of the view that patients having low adherence with drugs might have poorer lifestyle habits, like smoking and alcohol drinking which were associated with high cancer mortality rates. Yet why some firstline antihypertensive medications were associated with higher cancer mortality is still a knowledge gap.
[7]
[8]
[9]
[10]
[11]
4.4. Conclusions: implications to clinical practice and future research This study proposed that CCB users, and possibly patients prescribed with thiazide diuretics where a less definite association was observed, might represent a group of patients with higher cancer mortality rate five years within the first prescription. Our interpretation is that the drug class of the first-line prescription is not necessarily a causative factor to mortality, but patients who received CCBs that had intrinsic factors were associated with higher risks for cancer deaths — thus they will need closer monitoring when encountered in clinical practice. We suggested that this study should be substantiated by more larger-scale randomized trials in Chinese patients, as mortality outcome needs large sample size and longer duration to capture. The cause-and-effect association between first-line antihypertensive drug class and cancer mortality will need to be examined in future research.
[12]
[13]
[14] [15] [16] [17]
[18]
[19]
Conflict of interest The authors report no relationships that could be construed as a conflict of interest.
[20]
[21]
Grant support None. Disclosures
[22]
[23] [24]
None declared. Acknowledgments We express our gratitude for all the healthcare professionals who entered the data into the clinical database. We thank the Hospital Authority of the Hong Kong Government for allowing our research team to use the database.
[25]
[26] [27]
[28]
References [29] [1] C.M.M. Lawes, S. Vander Hoorn, A. Rodgers, I.S. Hypertens, Global burden of bloodpressure-related disease, 2001, Lancet 371 (2008) 1513–1518. [2] M.C.S. Wong, H.H.X. Wang, M.C.M. Leung, C.S.H. Tsang, S.V. Lo, S.M. Griffiths, The rising prevalence of self-reported hypertension among Chinese subjects: a populationbased study from 121 895 household interviews, QJM (2014), http://dx.doi.org/10. 1093/qjmed/hcu129. [3] K. Wolf-Maier, R.S. Cooper, J.R. Banegas, S. Giampaoli, H.W. Hense, M. Joffres, et al., Hypertension, prevalence and blood pressure levels in 6 European countries, Canada, and the United States, JAMA 289 (2003) 2363–2369. [4] T. Dorjgochoo, X.O. Shu, X. Zhang, H. Li, G. Yang, L. Gao, et al., Relation of blood pressure components and categories and all-cause, stroke and coronary heart disease mortality in urban Chinese women: a population-based prospective study, J Hypertens 27 (2009) 468–475. [5] Blood Pressure Lowering Treatment Trialists C, F. Turnbull, B. Neal, T. Ninomiya, C. Algert, H. Arima, et al., Effects of different regimens to lower blood pressure on major cardiovascular events in older and younger adults: meta-analysis of randomised trials, BMJ 336 (2008) 1121–1123. [6] G. Mancia, R. Fagard, K. Narkiewicz, J. Redon, A. Zanchetti, M. Bohm, et al., 2013 ESH/ ESC guidelines for the management of arterial hypertension: the Task Force for the Management of Arterial Hypertension of the European Society of Hypertension
[30] [31] [32]
[33]
[34]
[35]
(ESH) and of the European Society of Cardiology (ESC), Eur Heart J 34 (2013) 2159–2219. A.V. Chobanian, G.L. Bakris, H.R. Black, W.C. Cushman, L.A. Green, J.L. Izzo, et al., The seventh report of the joint national committee on prevention, detection, evaluation, and treatment of high blood pressure — the JNC 7 report, JAMA 289 (2003) 2560–2572. G. Mancia, S. Laurent, E. Agabiti-Rosei, E. Ambrosioni, M. Burnier, M.J. Caulfield, et al., Reappraisal of European guidelines on hypertension management: a European Society of Hypertension Task Force document, J Hypertens 27 (2009) 2121–2158. M.R. Law, J.K. Morris, N.J. Wald, Use of blood pressure lowering drugs in the prevention of cardiovascular disease: meta-analysis of 147 randomised trials in the context of expectations from prospective epidemiological studies, BMJ 338 (2009). M. Pahor, J.M. Guralnik, L. Ferrucci, M.C. Corti, M.E. Salive, J.R. Cerhan, et al., Calciumchannel blockade and incidence of cancer in aged populations, Lancet 348 (1996) 493–497. M. Pahor, J.M. Guralnik, M.E. Salive, M.C. Corti, P. Carbonin, R.J. Havlik, Do calcium channel blockers increase the risk of cancer? Am J Hypertens 9 (1996) 695–699. A.L. Fitzpatrick, J.R. Daling, C.D. Furberg, R.A. Kronmal, J.L. Weissfeld, Use of calcium channel blockers and breast carcinoma risk in postmenopausal women, Cancer 80 (1997) 1438–1447. S. Bangalore, S. Kumar, S.E. Kjeldsen, H. Makani, E. Grossman, J. Wetterslev, et al., Antihypertensive drugs and risk of cancer: network meta-analyses and trial sequential analyses of 324168 participants from randomised trials, Lancet Oncol 12 (2011) 65–82. H.J. Cohen, C.F. Pieper, J.T. Hanlon, W.E. Wall, B.M. Burchett, R.J. Havlik, Calcium channel blockers and cancer, Am J Med 108 (2000) 210–215. H. Jick, S. Jick, L.E. Derby, C. Vasilakis, M.W. Myers, C.R. Meier, Calcium-channel blockers and risk of cancer, Lancet 349 (1997) 525–528. L. Rosenberg, S. Rao, J.R. Palmer, B.L. Strom, P.D. Stolley, A.G. Zauber, et al., Calcium channel blockers and the risk of cancer, JAMA 279 (1998) 1000–1004. J.H. Olsen, H.T. Sorensen, S. Friis, J.K. McLaughlin, F.H. Steffensen, G.L. Nielsen, et al., Cancer risk in users of calcium channel blockers, Hypertension 29 (1997) 1091–1094. A.F. Lever, D.J. Hole, C.R. Gillis, I.R. McCallum, G.T. McInnes, P.L. MacKinnon, et al., Do inhibitors of angiotensin-I-converting enzyme protect against risk of cancer? Lancet 352 (1998) 179–184. C.R. Meier, L.E. Derby, S.S. Jick, H. Jick, Angiotensin-converting enzyme inhibitors, calcium channel blockers, and breast cancer, Arch Intern Med 160 (2000) 349–353. T.L. Vaughan, D.C. Farrow, P.D. Hansten, W.H. Chow, M.D. Gammon, H.A. Risch, et al., Risk of esophageal and gastric adenocarcinomas in relation to use of calcium channel blockers, asthma drugs, and other medications that promote gastroesophageal reflux, Cancer Epidemiol Biomarkers Prev 7 (1998) 749–756. K.B. Michels, B.A. Rosner, A.M. Walker, M.J. Stampfer, J.E. Manson, G.A. Colditz, et al., Calcium channel blockers, cancer incidence, and cancer mortality in a cohort of US women — the Nurses' Health Study, Cancer 83 (1998) 2003–2007. H.T. Sorensen, J.H. Olsen, L. Mellemkjaer, A.M. Thulstrup, F.H. Steffensen, J.K. McLaughlin, et al., Cancer risk and mortality in users of calcium channel blockers — a cohort study, Cancer 89 (2000) 165–170. R.M. Vezina, S.M. Lesko, L. Rosenberg, S. Shapiro, Calcium channel blocker use and the risk of prostate cancer, Am J Hypertens 11 (1998) 1420–1425. C.I. Coleman, W.L. Baker, J. Kluger, C.M. White, Antihypertensive medication and their impact on cancer incidence: a mixed treatment comparison meta-analysis of randomized controlled trials, J Hypertens 26 (2008) 622–629. C.I. Li, J.R. Daling, M.T.C. Tang, K.L. Haugen, P.L. Porter, K.E. Malone, Use of antihypertensive medications and breast cancer risk among women aged 55 to 74 years, JAMA Intern Med 173 (2013) 1629–1637. M.J. Brown, Hypertension and ethnic group, Br Med J 332 (2006) 833-6. M.C.S. Wong, W.W.S. Tam, C.S.K. Cheung, H.H.X. Wang, E.L.H. Tong, A.C.H. Sek, et al., Drug adherence and the incidence of coronary heart disease- and stroke-specific mortality among 218,047 patients newly prescribed an antihypertensive medication: a five-year cohort study, Int J Cardiol 168 (2013) 928–933. M.C.S. Wong, W.W.S. Tam, C.S.K. Cheung, E.L.H. Tong, A.C.H. Sek, N.T. Cheung, et al., Medication adherence to first-line antihypertensive drug class in a large Chinese population, Int J Cardiol 167 (2013) 1438–1442. M.C.S. Wong, J.Y. Jiang, J.L. Tang, A. Lam, H. Fung, S.W. Mercer, Health services research in the public healthcare system in Hong Kong: an analysis of over 1 million antihypertensive prescriptions between 2004–2007 as an example of the potential and pitfalls of using routinely collected electronic patient data, BMC Health Serv Res 8 (2008). M.C.S. Wong, J.Y. Jiang, S.M. Griffiths, Factors associated with compliance to thiazide diuretics among 8551 Chinese patients, J Clin Pharm Ther 36 (2011) 179–186. M.C.S. Wong, J.Y. Jiang, S.M. Griffiths, Adherence to combination therapy among ethnic Chinese patients: a cohort study, Hypertens Res 33 (2010) 416–421. M.C.S. Wong, J.Y. Jiang, S.M. Griffiths, Short-term adherence to beta-blocker therapy among ethnic Chinese patients with hypertension: a cohort study, Clin Ther 31 (2009) 2170–2177. M.C.S. Wong, J.Y. Jiang, S.M. Griffiths, Antihypertensive drug adherence among 6408 Chinese patients on angiotensin-converting enzyme inhibitors in Hong Kong: a cohort study, J Clin Pharmacol 50 (2010) 598–605. M.C.S. Wong, J.Y. Jiang, S.M. Griffiths, Factors associated with compliance, discontinuation and switching of calcium channel blockers in 20,156 Chinese patients, Am J Hypertens 22 (2009) 904–910. W.J. Sun, C.M. Schooling, W.M. Chan, K.S. Ho, T.H. Lam, The association between depressive symptoms and mortality among Chinese elderly: a Hong Kong cohort study, J Gerontol A Biol Sci Med Sci 66 (2011) 459–466.
M.C.S. Wong et al. / International Journal of Cardiology 179 (2015) 178–185 [36] C.M. Schooling, T.H. Lam, Z.B. Li, S.Y. Ho, W.M. Chan, K.S. Ho, et al., Obesity, physical activity, and mortality in a prospective Chinese elderly cohort, Arch Intern Med 166 (2006) 1498–1504. [37] T.H. Lam, S.Y. Ho, A.J. Hedley, K.H. Mak, R. Peto, Mortality and smoking in Hong Kong: case–control study of all adult deaths in 1998, Br Med J 323 (2001) 361–362. [38] N.K. Choudhry, W.H. Shrank, R.L. Levin, J.L. Lee, S.A. Jan, M.A. Brookhart, et al., Measuring concurrent adherence to multiple related medications, Am J Manag Care 15 (2009) 457–464. [39] B.C. Martin, E.K. Wiley-Exley, S. Richards, M.E. Domino, T.S. Carey, B.L. Sleath, Contrasting measures of adherence with simple drug use, medication switching, and therapeutic duplication, Ann Pharmacother 43 (2009) 36–44. [40] S.E. Andrade, K.H. Kahler, F. Frech, K.A. Chan, Methods for evaluation of medication adherence and persistence using automated databases, Pharmacoepidemiol Drug Saf 15 (2006) 565–574 [discussion 75–7]. [41] G. Mazzaglia, E. Ambrosioni, M. Alacqua, A. Filippi, E. Sessa, V. Immordino, et al., Adherence to antihypertensive medications and cardiovascular morbidity among newly diagnosed hypertensive patients, Circulation 120 (2009) 1598–1605. [42] J.N. Rasmussen, A. Chong, D.A. Alter, Relationship between adherence to evidencebased pharmacotherapy and long-term mortality after acute myocardial infarction, JAMA 297 (2007) 177–186.
185
[43] T.A. Burke, M.C. Sturkenboom, S.E. Lu, C.E. Wentworth, Y. Lin, G.G. Rhoads, Discontinuation of antihypertensive drugs among newly diagnosed hypertensive patients in UK general practice, J Hypertens 24 (2006) 1193–1200. [44] G. Mazzaglia, L.G. Mantovani, M.C.J.M. Sturkenboom, A. Filippi, G. Trifiro, C. Cricelli, et al., Patterns of persistence with antihypertensive medications in newly diagnosed hypertensive patients in Italy: a retrospective cohort study in primary care, J Hypertens 23 (2005) 2093–2100. [45] M.C. Wong, W.W. Tam, C.S. Cheung, E.L. Tong, A.C. Sek, G. John, et al., Initial antihypertensive prescription and switching: a 5 year cohort study from 250,851 patients, PLoS One 8 (2013) e53625. [46] I.A. Dhalla, T. Gomes, Z. Yao, J. Nagge, N. Persaud, C. Hellings, et al., Chlorthalidone versus hydrochlorothiazide for the treatment of hypertension in older adults: a population-based cohort study, Ann Intern Med 158 (2013) 447–455. [47] L.H. Curtis, B.G. Hammill, E.L. Eisenstein, J.M. Kramer, K.J. Anstrom, Using inverse probability-weighted estimators in comparative effectiveness analyses with observational databases, Med Care 45 (2007) S103–S107. [48] R.B. D'Agostino, Propensity scores in cardiovascular research, Circulation 115 (2007) 2340–2343. [49] S. Schneeweiss, J.A. Rassen, R.J. Glynn, J. Avorn, H. Mogun, M.A. Brookhart, Highdimensional propensity score adjustment in studies of treatment effects using health care claims data, Epidemiology 20 (2009) 512–522.