- To examine the association between digoxin exposure and mortality in men with prostate cancer using linked Irish National Cancer Registry and pharmacy claims data.
Preclinical studies have shown that the cardiac glycoside, digoxin, inhibits hypoxia-inducible factor 1α (HIF-1α) protein synthesis and the expression of HIF-1 target genes in prostate cancer cells . Overexpression of HIF-1α in prostate cancer is associated with larger tumour size , increased angiogenesis , treatment resistance and poorer prognosis [2, 4]. The down regulation of HIF-1α signalling by digoxin in prostate cancer models has been shown to inhibit tumour growth  and tumour vascularisation . Digoxin and other cardiac glycosides have also been shown to reduce the development of metastases in prostate  and breast tumour models ; and to sensitise cancer cells to radiation . These preclinical results suggest that digoxin exposure may be associated with reduced mortality in men with prostate cancer.
To date, observational research has focused on associations between digoxin exposure and cancer incidence. In a recent observational study, regular digoxin exposure was associated with a 23% reduction in the risk of prostate cancer . The results from that study indicated the presence of an exposure–response effect, with longer duration of digoxin use (≥10 years) associated with greater reductions in risk . Inhibition of HIF-1α by digoxin was proposed as one of the potential mechanisms for this risk reduction. Conversely, digoxin exposure has been associated with an increased risk of oestrogen-dependent cancers [10-14]. It has been suggested that this increased risk may be due to phyto-oestrogenic properties of digoxin and its ability to bind with oestrogen receptors . Targeting oestrogen receptors in prostate cancer has been explored as a treatment for the disease and may represent an alternative potential mechanism by which digoxin could influence prostate cancer outcomes [15, 16].
There is little information available on whether digoxin exposure is associated with longer survival in men with prostate cancer. The aims of the present study were to investigate: (i) associations between digoxin exposure at diagnosis and mortality in men with prostate cancer; (ii) whether associations between digoxin exposure and mortality are modified by receipt of radiation therapy or androgen-deprivation therapy (ADT).
Patient records from the National Cancer Registry Ireland (NCRI), linked to Ireland's Health Services Executive (HSE) – Primary Care Reimbursement Service (PCRS) pharmacy claims database, were used to conduct this study. The NCRI uses active registration methods to collect detailed information on all incident cancers in the population usually resident in Ireland. Trained, hospital-based tumour registration officers collect information on patient characteristics, tumour details, treatment received and death from multiple sources including pathology laboratories, radiology departments, oncology departments, hospital administrative systems, individual medical records and death certificates. The HSE-PCRS General Medical Services (GMS) scheme provides state-funded universal healthcare, including medicines, to about one third (1.4 million) of the Irish population . Eligibility for the GMS scheme is assessed through means test and age. The GMS database records full details of all prescription drugs dispensed from community pharmacies to eligible patients. Drugs are coded according to the WHO Anatomical Therapeutic Chemical Classification (ATC) system . The use, for research, of anonymous data held by the NCRI is covered by the Health (Provision of Information) Act 1997.
Men were eligible for inclusion in the study if they had a diagnosis of prostate cancer (International Classification of Diseases [ICD-O], C61.9)  of any stage between 1st January 2001 and 31st December 2006, and eligibility for the GMS scheme from ≥1 year before diagnosis. Men with prior invasive tumours other than non-melanoma skin cancer or diagnosed with prostate cancer at the time of death were excluded.
Men exposed to digoxin at the time of prostate cancer diagnosis were identified from prescription claims for digoxin (ATC: C01AA05) in the 90 days before diagnosis. The date, dose and number of days' supply on each prescription were recorded. Digoxin exposure in the 90 days before diagnosis was stratified by dosing intensity, defined as the proportion of days in the 90 days before diagnosis that a man has a supply of digoxin available, divided into tertiles (low, intermediate, high).
Deaths from prostate cancer (ICD9: 185; ICD10: C61), from other causes, and associated dates of death were identified from the NCRI database. Patients were followed from date of diagnosis until either death or 31 December 2009.
The NCRI database was used to identify the following tumour details: tumour stage (I, II, III, IV, unspecified) and tumour grade (grade I, Gleason score <5; grade II, Gleason score 5–7; grade III and IV, Gleason score >7; unspecified) . Treatment received in the year after diagnosis (surgery, radiation, ADT, chemotherapy); age at diagnosis (years, continuous) and smoking status at diagnosis (current, never, former, unspecified) were also identified from this database.
Prescription dispensing data from the GMS database was used to identify the use of other co-prescribed medication (exposed, unexposed) before diagnosis (Appendix 1, Table A1). The number of distinct medication classes (five character ATC code) dispensed in the year before diagnosis was used as a measure of comorbidity (number of medication classes, continuous). This medication-based comorbidity score is a validated predictor of healthcare usage and mortality in an older adult population .
The frequency and proportion of digoxin exposed and unexposed men were tabulated by clinical and demographic variables. Unadjusted all-cause and prostate cancer-specific mortality (PCSM) rates were calculated for digoxin exposed and unexposed men. Adjusted cumulative probabilities for PCSM and all-cause mortality were estimated for digoxin exposed and unexposed men . Univariate and multivariate hazard ratios and 95% CIs for associations between digoxin exposure and (i) all-cause mortality and (ii) PCSM were estimated using Cox proportional hazards models (SAS®, PROC PHREG). Covariates were assessed for inclusion in the multivariate model based on prior knowledge of potential predictors of prostate cancer mortality (age ; comorbidity score ; smoking status [24, 25]; tumour stage ; tumour grade ; diabetes ; and exposure to aspirin , β-blockers [28, 29], warfarin , statins , NSAIDS  or drugs used for the treatment of BPH ). The year of incidence, and treatments received in the year after diagnosis (surgery, radiation, ADT; time-dependent covariates) were also assessed for inclusion. Backwards elimination of variables in a stepwise manner up to a 10% maximum cumulative change in the effect component of the fully adjusted HR, was used to select the final multivariate model .
Exposure response analyses were conducted by tertiles of digoxin dosing intensity (low, intermediate, high). The presence of effect modification by radiation therapy or ADT received in the year after diagnosis was assessed. Measures of interaction were estimated on a multiplicative scale (ratio of HRs, rHRs) with 95% CIs. SAS Version 9.2 was used for all analyses (SAS Institute, Cary NC). Results were considered statistically significant at a two-sided α-level of 0.05.
The high cardiovascular comorbidity associated with indications for digoxin use may confound associations between digoxin exposure and prostate cancer outcomes through differential effects on the selection and use of prostate cancer treatments [34-37]. Secondary analyses of all-cause mortality and PCSM were carried out using propensity score trimmed and matched cohorts . A propensity score model was developed to predict digoxin exposure at the time of prostate cancer diagnosis as follows: covariates were assessed for inclusion in the propensity score model based on prior knowledge of demographic covariates associated with cardiovascular comorbidity (age, comorbidity score) and exposure to cardiovascular medications commonly co-prescribed with digoxin (Appendix 2, Table A2) . Logistic regression models (SAS®, PROC LOGISTIC) were used to estimate propensity scores for digoxin exposure using these covariates. Main effects, interaction terms and quadratic or cubic terms were included as appropriate . Covariate balance within propensity score quintiles was assessed by standardised differences (d), with a d < 0.1 being the desired limit . The multivariate propensity score model that achieved the optimal balance of matched covariates between digoxin exposed and unexposed men was selected. Men with a propensity score outside the first to 99th percentile for digoxin exposed men were excluded  (trimmed cohort) and the propensity score was re-estimated in this population . Digoxin exposed and unexposed men were then matched (1:1) within a calliper of 0.2 standard deviations of the propensity score logit [41, 43] using greedy matching without replacement . Covariate balance between digoxin exposed and unexposed men in the matched cohort was assessed by standardised differences (d < 0.1).
Sensitivity analyses were conducted to assess the possibility that PCSM was misclassified on death certificates. Firstly, mortality from prostate cancer was defined using ICD mortality site codes for ill-defined cancer sites, secondary cancer sites, cancers of uncertain or unknown behaviour (Appendix 3, Table A3.1) . Secondly, an analysis was carried out in which deaths where prostate cancer was identified as a secondary or contributory cause of death on the death certificate were defined as prostate cancer deaths.
A flow diagram outlining the study cohort selection is presented in Fig. 1. The characteristics of digoxin exposed and unexposed men in the full cohort and the propensity score-matched cohort are presented in Table 1 . In the full cohort, digoxin exposed men (391 men) were older and had a higher comorbidity score than unexposed men (5341). Digoxin exposed men were also more likely to have stage IV disease and less likely to have received radiation. Men in the low, intermediate and high dosing intensity tertiles had mean post-diagnostic digoxin exposures in the year after diagnosis of 53.8%, 70.8% and 80.6% respectively. The median follow-up was 4.3 years.
|Characteristic||Full cohort||Propensity score-matched cohort|
|Unexposed (N = 5341)||Exposed (N = 391)||Unexposed (N = 387)||Exposed (N = 387)|
|Mean (sd) age, years||73.1 (7.9)||77.5 (7.2)**||77.8 (7.0)||77.5 (7.2)|
|Mean (sd) Comorbidity score||8.9 (6.2)||13.2 (6.6)**||13.0 (6.3)||13.2 (6.5)|
|Smoking status, n (%):|
|Never||1712 (32.1)||134 (34.3)||143 (37.0)||131 (33.9)|
|Former||1020 (19.1)||85 (21.7)||93 (24.0)||85 (22.0)|
|Current||853 (16.0)||55 (14.1)||47 (12.1)||55 (14.2)|
|Unspecified||1756 (32.9)||117 (29.9)||104 (26.9)||116 (30.0)|
|Stage, n (%)‡:|
|I||166 (3.1)||12 (3.1)*||19 (4.9)||12 (3.1)|
|II||2589 (48.5)||147 (37.6)||151 (39.0)||145 (37.5)|
|III||410 (7.7)||22 (5.6)||17 (4.4)||22 (5.7)|
|IV||761 (14.2)||84 (21.5)||86 (22.2)||82 (21.2)|
|Unspecified||1415 (26.5)||126 (32.2)||114 (29.5)||126 (32.6)|
|Grade, n (%):|
|I||334 (6.3)||30 (7.7)*||23 (5.9)||30 (7.8)|
|II||2769 (51.8)||155 (39.6)||167 (43.2)||152 (39.3)|
|III/IV||1061 (19.9)||70 (17.9)||75 (19.4)||70 (18.1)|
|Unspecified||1177 (22.0)||136 (34.8)||122 (31.5)||135 (34.9)|
|Treatment, n (%)§:|
|Surgery/RP||1339/181 (25.1)/(3.4)||98/5 (25.1)/(1.3)||87/3 (22.5)/(0.8)||97/5 (25.1)/(1.3)|
|Radiotherapy||1509 (28.3)||52 (13.3)*||66 (17.1)||51 (13.2)†|
|ADT||2617 (49.0)||198 (50.6)||202 (52.2)||195 (50.4)|
|Chemotherapy||111 (2.1)||6 (1.5)||7 (1.8)||6 (1.6)|
|Medication exposures, n (%):|
|Aspirin||1986 (37.2)||194 (49.6)*||197 (50.9)||194 (50.1)|
|β-blocker||1158 (21.7)||97 (24.8)||122 (31.5)||95 (24.5)†|
|Statin||1293 (24.2)||95 (24.3)||91 (23.5)||95 (24.5)|
|Warfarin||214 (4.0)||150 (38.4)*||91 (23.5)||148 (38.2)†|
|Anti-diabetic||417 (7.8)||48 (12.3)*||32 (8.3)||48 (12.4)†|
|NSAID||1833 (34.3)||139 (35.5)||151 (39.0)||136 (35.1)|
|BPH medicines||1393 (26.1)||111 (28.4)||125 (32.3)||109 (28.2)|
|Digoxin exposure details (90 days before diagnosis)|
|No. of prescriptions dispensed||–||1030||–||1011|
|Median (IQR) dosing intensity, %||–||84.6 (75.6, 100)||–||84.5 (75.6, 100)|
|Digoxin exposure details (1 year after diagnosis)|
|No. of prescriptions dispensed||567||3374||99||3336|
|Median (IQR) dosing intensity, %||0.01 (0.0, 0.0)||69.6 (36.4, 99.2)||1.8 (0.0, 0.0)||69.5 (36.4, 99.2)|
In the trimmed cohort, differences between exposed and unexposed men were reduced, but remained significant for some covariates, including age and comorbidity score. In the propensity score-matched cohort, acceptable balance for matched covariates was achieved between digoxin exposed (387 men) and unexposed men (387). Tumour stage, tumour grade and treatment received in the year after diagnosis were also comparable between the matched groups, although digoxin exposed men were marginally less likely to have been treated with radiation (13.2% vs 17.1%, d = 0.11).
Adjusted cumulative probability plots and HRs for all-cause mortality and PCSM in the full cohort are presented in Fig. 2 and Table 2, respectively. In the full cohort digoxin use was associated with a 24% increase in the risk of all-cause mortality (multivariate HR 1.24, 95% CI 1.07, 1.43) and a statistically non-significant 13% increase in the risk of PCSM (HR 1.13, 95% CI 0.91, 1.42). Adjusted estimates were not appreciably different in the propensity score trimmed (Table 2: all-cause mortality HR 1.23, 95% CI 1.07, 1.43; PCSM HR 1.12, 95% CI 0.90, 1.41) or matched populations (Table 2: all-cause mortality HR 1.20, 95% CI 1.00, 1.49; PCSM HR 1.17, 95% CI 0.88, 1.57). Adjusted cumulative probability plots indicate that associations between digoxin exposure and PCSM did not vary considerably over time.
|Digoxin use||N||Person years||All-cause mortality||PCSM|
|No. of deaths (rate)†||Univariate HR (95%CI)||Multivariate HR (95%CI)‡||No. of deaths (rate)†||Univariate HR (95%CI)||Multivariate HR (95%CI)‡|
|Digoxin unexposed||5341||22 774||2096 (92.0)||Ref||Ref||995 (43.7)||Ref||Ref|
|Digoxin exposed||391||1 277||253 (198.1)||2.11 (1.86, 2.41)||1.24 (1.07, 1.43)||103 (80.7)||1.77 (1.45, 2.17)||1.13 (0.91, 1.42)|
|Exposure response: dosing intensity|
|Dosing intensity 0–85%||117||319||89 (279.2)||2.94 (2.38, 3.63)||1.59 (1.27, 1.97)||33 (103.5)||2.22 (1.57, 3.14)||1.18 (0.83, 1.68)|
|Dosing intensity 86–99%||120||413||78 (188.9)||2.02 (1.61, 2.54)||1.33 (1.05, 1.67)||33 (79.6)||1.78 (1.26, 2.52)||1.39 (0.97, 1.98)|
|Dosing intensity 100%||154||544||86 (157.9)||1.69 (1.36, 2.10)||0.93 (0.74, 1.18)||37 (67.9)||1.50 (1.08, 2.08)||0.93 (0.65, 1.32)|
|Propensity score trimmed cohort analysis|
|Digoxin unexposed||3940||15 938||1780 (111.7)||Ref||Ref||833 (52.3)||Ref||Ref|
|Digoxin exposed||389||1 272||252 (198.1)||1.75 (1.53, 2.00)||1.23 (1.07, 1.43)||102 (80.2)||1.49 (1.21, 1.82)||1.12 (0.90, 1.41)|
|Propensity score matched cohort analysis|
|Digoxin unexposed||387||1 339||234 (174.8)||Ref||Ref||105 (78.4)||Ref||Ref|
|Digoxin exposed||387||1 269||250 (197.1)||1.13 (0.94, 1.35)||1.20 (1.00, 1.45)||101 (79.6)||1.02 (0.77, 1.34)||1.17 (0.88, 1.57)|
In multivariate exposure response analyses (Table 2) there was no trend for associations between PCSM and increasing dose intensity (P-trend 0.59). Analyses of interaction between digoxin use and the receipt of radiation with respect to PCSM (Table 3, P-interaction 0.14), or ADT (Table 4, P-interaction 0.35) were also non-significant. Within-strata of men who received radiation and ADT adjusted HRs suggested the possibility of increased PCSM for digoxin exposed men compared with unexposed men, (Table 3, HR 1.77, 95% CI 0.95, 3.30 and Table 4, HR 1.22, 95% CI 0.93, 1.59, respectively). However, the risk estimates did not reach formal statistical significance. These results were unchanged in sensitivity analyses for misclassification of PCSM (Appendix 3, Tables A3.2 and A3.3).
|Radiation||Digoxin unexposed||Digoxin exposed||Exposed vs unexposed|
|Multivariate HR (95%CI)||1.00||1.08 (0.86, 1.37) P = 0.51||1.08 (0.86, 1.37) P = 0.51|
|Multivariate HR (95%CI)||0.95 (0.79, 1.15) P = 0.62||1.69 (0.92 3.10) P = 0.09||1.77 (0.95, 3.30) P = 0.07|
|Multiplicative scale: rHR (95%CI) yes vs no||1.64 (0.85, 3.14) P = 0.14|
|ADT||Digoxin unexposed||Digoxin exposed||Exposed vs unexposed|
|Multivariate HR (95%CI)||1.00||1.00 (0.69, 1.43) P = 0.98||1.00 (0.69, 1.43) P = 0.98|
|Multivariate HR (95%CI)||1.06 (0.92, 1.21) P = 0.46||1.29 (0.97, 1.70) P = 0.08||1.22 (0.93, 1.59) P = 0.14|
|Multiplicative scale: rHR (95%CI) yes vs no||1.23 (0.80, 1.89) P = 0.35|
Preclinical evidence has suggested a possible role for digoxin in the treatment of prostate cancer. In the present study of 5732 men with prostate cancer, digoxin use was not associated with a reduction in PCSM. These results were unchanged in matched analyses of men with similar cardiovascular comorbidities, suggesting that the lack of observed effect is not confounded by associations between high cardiovascular comorbidity and less aggressive treatment of prostate cancer in digoxin-treated men. Additionally, there was no trend for increasing digoxin dosing intensity in exposure–response analyses. Digoxin dose dispensed (low, ≤125 μg; high, >125 μg) was not associated with any trend in all-cause mortality or PCSM either (results not presented).
In a recent observational study of men with metastatic prostate cancer treated with docetaxel, digoxin exposure was associated with an increased risk of all-cause mortality (HR 1.43, 95% CI 1.01, 2.03) . The authors concluded that this increased risk was due to higher levels of cardiovascular comorbidity in the digoxin-exposed group. Similarly, the present results also show an increased risk of all-cause mortality among digoxin-exposed men; this is most probably also due to increased cardiovascular comorbidity in this group. In the propensity score-matched cohort associations with all-cause mortality were reduced although these approached significance after adjustment.
Several preclinical studies have indicated a role for digoxin in prostate cancer through the inhibition of HIF-1α. However, it has been suggested that digoxin plasma levels achievable in humans may not be sufficient to effectively inhibit HIF-1α and prostate cancer progression [47, 48]. In pre-clinical studies, digoxin has been shown to inhibit HIF-1α in prostate cancer cell lines at concentrations of 100 nm and prostate cancer cell proliferation at 23–255 nm . However, these digoxin concentrations are considerably higher than the therapeutic plasma concentrations normally tolerated in humans, 1.6 ± 1.0 nm . The possibility that typical levels of digoxin exposure in humans may not adequately inhibit HIF-1α may explain why digoxin exposure was not associated with reduced PCSM in the present study. It has been suggested that prolonged exposure to digoxin, even at normal therapeutic concentrations, may successfully inhibit HIF-1α in humans . However, there was no evidence in the present study of a dose-response trend with increasing digoxin dosing intensity. Additionally, it should be noted that the clinical benefits of HIF-1α inhibition as a therapeutic target in prostate cancer, have yet to be shown in a randomised setting .
Elevated HIF-1α expression in prostate tumours has been associated with increased resistance to radiation  and it has been suggested that digoxin may have synergistic activity in combination with radiation therapy . Analyses of interaction between digoxin use and receipt of radiation in the year after prostate cancer diagnosis did not indicate that digoxin use was associated with additional clinical benefit in men receiving radiation therapy. There was, instead, the suggestion that digoxin use in men receiving radiation therapy may be associated with increased PCSM. The reasons for this are unclear. However, it should be noted that the number of men, receiving both digoxin and radiation in this subgroup analysis was small and these results will require confirmation in larger studies.
Digoxin has also been reported to have phyto-oestrogenic properties . Increased endogenous oestrogen levels, and alterations in the testosterone–oestrogen ratio have been weakly associated with prostate cancer risk . Oestrogens have also been used in the treatment of prostate cancer . More recently, it has been shown in prostate cancer that oestrogen signalling is mediated through two oestrogen receptor (ER) subtypes, ERα and ERβ, with opposing effects [56, 57]. ERα signalling in prostate cancer has been associated with increased tumour cell proliferation; ERβ signalling has been reported to have anti-proliferative effects that balance the proliferative action of androgens in prostatic tissue ; ERβ can also be associated with a more aggressive prostate cancer phenotype . Concerns have been raised about the proliferative effects of agents with ERα agonist activity in prostate cancer [54, 56, 57]. However, the exact ER-subtype that digoxin is proposed to act on has not been identified. The results presented here do not exclude the possibility that digoxin exposure may be associated with an increased risk of PCSM.
The strengths of the present study include the cohort design, high-quality outcome data and the availability of detailed digoxin prescription histories. The study also has some limitations. High levels of comorbidity associated with digoxin use may have limited the ability to detect small or longer term benefits from digoxin exposure. Subgroup analyses, stratified by treatment receipt, were limited by small numbers and the results from these should be interpreted with caution. Digoxin use was based upon prescriptions dispensed and non-compliance with received treatment will have resulted in exposure misclassification. Additionally, digoxin exposure groups were defined at diagnosis and post-diagnostic treatment crossover will also have resulted in exposure misclassification; such misclassifications will usually bias results towards the null. In the recent study by Platz et al. , digoxin use for >10 years was associated with a significantly reduced incidence of prostate cancer. Prescription histories of this duration were not available for analysis in the present study.
In conclusion, the results from this analysis do not suggest that digoxin use is associated with a reduction in PCSM.
We would like to thank the NCRI and the Irish HSE Primary Care Reimbursement Service for providing access to the data upon which this study was based. The interpretation and reporting of these data are the responsibility of the authors and should in no way be seen as the official policy or interpretation of the NCRI or the Irish HSE Primary Care Reimbursement Service.
E.M.F. is supported by a PhD studentship from the Irish Cancer Society (CRS10FLA). T.I.B. is supported by the Health Research Board Ireland (HRA-2009-221, ICE-2011-9). The Irish Cancer Society and the Health Research Board Ireland had no role in the study design; collection, analysis, and interpretation of data; writing of the report; or the decision to submit for publication.
LS has received an unrestricted project grant from Sanofi-Aventis (2011–2012).
Anatomical Therapeutic Chemical Classification (system)
General Medical Services
hypoxia-inducible factor 1α
(ratio of) hazard ratio
Health Services Executive (Ireland)
International Classification of Diseases
National Cancer Registry Ireland
Primary Care Reimbursement Service
prostate cancer-specific mortality
|Medication group||WHO-ATC codes|
|Biguanides||A10BA; A10BD01; A10BD02; A10BD03; A10BD05; A10BD07|
|Aspirin||B01AC06; M01BA03; N02BA01; N02BA51; N02BA71|
|Other anti-thrombotic agents (excluding aspirin)||B01A, (excluding B01AC06)|
|Antiarrhythmic agents Class I a||C01BA|
|Antiarrhythmic agents Class I c||C01BC|
|Class III antiarrhythmic agents||C01BD|
|Other Cardiac agents||C01E|
|Calcium Channel Blocker Vascular||C08C|
|Calcium Channel Blocker Cardiac||C08D|
|Angiotensin converting enzyme inhibitors||C09A; C09B|
|Angiotensin II receptor blockers||C09C; C09D|
|Benign Prostatic Hypertrophy Medication||G04C|
|Age at diagnosis (years)|
|Comorbidity score (Distinct medication classes, 5 character ATC)|
|Comorbidity score squared|
|Comorbidity score cubed|
|Other anti-thrombotic agents (excluding aspirin, including warfarin)|
|Class III antiarrhythmic agents|
|Angiotensin converting enzyme inhibitors|
|Angiotensin II receptor blockers|
|Aspirin * other antithrombotic|
|Aspirin * Statins|
|Aspirin * High-Ceiling Diuretics|
|Aspirin * Nitrates|
|Other Antithrombotic * Verapamil|
|Other Antithrombotic * High-Ceiling Diuretics|
|Other Antithrombotic * Aldosterone Antagonist|
|Cancer site||ICD 9 code||ICD 10 code|
|Malignant neoplasm of prostate||185||C61|
|Malignant neoplasm of other male genital organs, site unspecified||187.9||C63.9|
|Malignant neoplasm of pelvis||195.3||C41.4|
|Secondary malignant neoplasm||196–198||C76–C80|
|Malignant neoplasm without specification of site||199||C80.9|
|Benign neoplasm of prostate||222.2||D29.1|
|Benign neoplasm of male genital organs, site unspecified||222.9||D29.9|
|Neoplasm of uncertain behaviour of prostate||236.5||D40.0|
|Neoplasm of uncertain behaviour of other and unspecified male genital organs||236.6||D40.9|
|Neoplasm of uncertain behaviour, site unspecified||238.9||D48.9|
|Neoplasm of unspecified nature of other genitourinary organs||239.5||D40.7, D41|
|Neoplasm of unspecified nature, site unspecified||239.9||D48.9|
|Digoxin use||Prostate cancer-specific mortality|
|N||Person years||No. of deaths (rate)†||Univariate HR (95%CI)||Multivariate HR‡ (95%CI)|
|Digoxin unexposed||5431||22 774||1018||(44.7)||Ref||–||Ref||–|
|Digoxin exposed||391||1 277||106||(83.0)||1.78||(1.46, 2.18)||1.14||(0.91, 1.42)|
|Propensity Score Trimmed Cohort|
|Digoxin unexposed||3940||15 938||852||(53.5)||Ref||–||Ref||–|
|Digoxin exposed||389||1 272||105||(82.5)||1.49||(1.22, 1.83)||1.13||(0.90, 1.41)|
|Propensity Score Matched Cohort|
|Digoxin unexposed||387||1 339||109||(81.4)||Ref||–||Ref||–|
|Digoxin exposed||387||1 269||104||(82.0)||1.01||(0.77, 1.32)||1.14||(0.86, 1.51)|
|Digoxin use||Prostate cancer-specific mortality|
|N||Person years||No. of deaths (rate)†||Univariate HR (95%CI)||Multivariate HR‡ (95%CI)|
|Digoxin unexposed||5431||22 774||1068||(46.9)||Ref||–||Ref||–|
|Digoxin exposed||391||1 277||115||(90.1)||1.83||(1.51, 2.22)||1.19||(0.96, 1.46)|
|Propensity Score Trimmed Cohort|
|Digoxin unexposed||3940||15 938||899||(56.4)||Ref||–||Ref||–|
|Digoxin exposed||389||1 272||114||(89.6)||1.53||(1.26, 1.86)||1.17||(0.95, 1.45)|
|Propensity Score Matched Cohort|
|Digoxin unexposed||387||1 339||113||(84.4)||Ref||–||Ref||–|
|Digoxin exposed||387||1 269||113||(89.1)||1.06||(0.81, 1.37)||1.24||(0.94, 1.63)|