Abstract
Aims:
We investigated long-term adherence to renin–angiotensin system inhibitors (RASIs) and β-blockers, and associated predictors, in senior patients after hospitalization for heart failure (HF).
Methods:
A population-based data set identified 4488 patients who survived 60 days following their index hospitalization for HF in Western Australia from 2003 to 2008 with a 3-year follow-up. Their person-linked Pharmaceutical Benefits Scheme records identified medications dispensed during follow-up. Drug discontinuation was defined as the first break ≥90 days following the previous supply. Medication adherence was calculated using the proportion of days covered (PDC), with PDC ≥ 80% defined as being adherent. Multivariable logistic regression models were used to identify predictors of PDC < 80%.
Results:
In the cohort (57% male, mean age: 76.6 years), 77.4% were dispensed a RASI and 52.7% a β-blocker within 60 days postdischarge. Over the 3-year follow-up, 28% and 42% of patients discontinued RASI and β-blockers, respectively. Only 64.6% and 47.5% of RASI and β-blocker users, respectively, were adherent to their treatment over 3 years, with adherence decreasing over time (trend P < .0001 for RASI and trend P = .02 for β-blockers). Older age, increasing Charlson comorbidity score, chronic kidney disease, and chronic obstructive pulmonary disease were independent predictors of PDC < 80% for both drug groups.
Conclusion:
Among seniors hospitalized for HF, discontinuation gaps were common for RASI and β-blockers postdischarge, and long-term adherence to these medications was suboptimal. Where appropriate, strategies to improve long-term medication adherence are indicated in HF patients, particularly in elderly patients with comorbidities.
Keywords
Introduction
Heart failure (HF) is a major public health problem in senior populations aged 65 years and older. It is associated with high mortality, increased hospitalization rates, and rising health care expenditure. 1,2 Effective evidence-based medicines (EBMs), principally renin–angiotensin system inhibitors (RASIs) and β-blockers, are available, 1 and their extensive use and continuance can substantially improve clinical outcomes. 3 -5 However, nonadherence to and discontinuation within 1 year of these medications are common 6,7 and associated with increased mortality, hospitalization risk, and health care costs in HF patients. 8 -12 Hence, medication adherence is a major concern for clinicians, health care systems, and other stakeholders (eg, payers) that seek to improve outcomes for HF patients. Estimates of adherence and persistence to prescribed HF pharmacotherapies vary considerably in the literature, ranging from 40% to 90%, depending on the method of measuring adherence, observation period, and cohort characteristics. 6,13 -18 Almost all adherence studies limited their observation period to around 1 year, 19 but a recent European study has shown that around 64% of patients with HF with reduced ejection fraction remained adherent to dual therapy (RASI and β-blocker) over a 3-year period after hospital discharge. 20 However, little is known about the long-term adherence to EBMs among Australian patients with hospitalized HF, which predominantly affects older patients with a high comorbidity burden that may impact adherence. 21,22 An understanding of the long-term adherence to EBMs and their clinical predictors in a “real-world” population-based HF cohort is necessary to inform targets and potential need for strategies to improve adherence and expected HF outcomes.
Accordingly, we used a population-based administrative database to identify seniors, aged 65 to 84 years, who survived an index hospitalization for HF in Western Australia (WA) between 2003 and 2008 and who were dispensed a RASI and/or a β-blocker within 60 days posthospital discharge. 23,24 We restricted the study to seniors with a health concession card, comprising 90% to 95% of all Australians aged ≥65 years, as this entitles them to inexpensive subsidized medicines under the Pharmaceutical Benefits Scheme (PBS). 23,24 Our study objectives in this HF cohort were to investigate (1) the discontinuation of and levels of adherence to RASI and β-blockers over a 3-year follow-up period after leaving hospital and (2) the independent predictors of lower adherence (PDC < 80%) to these 2 drug groups over the same 3-year period.
Methods
Data Sources
This retrospective cohort study used statutory government-held administrative and drug claims data of person-linked health information as previously described. 23,24 Briefly, the Hospital Morbidity Data Collection from the WA Data Linkage System 25 was used to identify patients admitted to hospital for HF from 2003 to June 30, 2008, with hospitalizations linked to the WA death registry. Pharmaceutical Benefits Scheme claims data 26 were used to identify matching records for the dispensing of RASI or approved HF β-blockers (bisoprolol, carvedilol, metoprolol tartrate, metoprolol succinate, and nebivolol) using their Anatomical Therapeutic Chemical (ATC) codes between June 1, 2002, and June 30, 2011. While there are no validation studies of the accuracy of PBS data, we expect these data to be accurate due to the high standards of pharmacy practice in Australia, electronic data collection by the Australian Department of Health, and accurate record linkage within the PBS data set using the unique Medicare number that all Australian residents have.
Study Cohort
We identified a cohort of 4488 patients, aged 65 to 84 years, with an index (first in period) hospitalization between 2003 and mid-2008 for HF as principal discharge diagnosis or HF as a secondary discharge diagnosis with ischemic heart disease (IHD, International Classification of Diseases-Tenth Edition-AM codes I20-I25) as principal discharge diagnosis and who survived 60 days after leaving hospital. 23,24 Patients with a history of valvular heart disease or renal dialysis were excluded. 23 The coding of HF admissions and acute myocardial infarction in administrative hospital morbidity data has been previously validated. 27,28
Data Collection
Demographic data were identified from the index HF admission. 23 Individual comorbidities and surgical histories were identified from prior hospital admissions using a fixed 20-year look-back period for all sampled patients prior to the index admission. 23,24 The Charlson comorbidity score, excluding HF, was derived from the identified comorbidities. 29 All relevant prior medications were identified from the PBS data from at least 2 supplies for the 6 months prior to the index HF admission. The number of drugs supplied in the 6 months prior to the index HF admission was derived from drugs with at least 2 dispensings identified from the top level of ATC codes for medications used in chronic diseases.
Medication Discontinuation and Adherence
Medication adherence to RASI and β-blockers was calculated by using the proportion of days covered (PDC) method. 30,31 The meaning of medication adherence here applies only to electronic administrative data and does not attempt to capture behavioral processes involved in maintaining a treatment plan according to the prescriber’s instructions. The denominator of the PDC was the number of days from first supply to the end of 1, 2, and 3 years post-HF discharge or the date of death (whichever came first), and the numerator was the number of days covered by the medications during the denominator period.
While the PBS database includes the quantity dispensed for a drug, it does not include a “days supply” field to indicate the number of days the medication is expected to last. Hence, the expected duration of supply was estimated separately for RASI and β-blockers 31 from the 75th percentile of the distribution of time to next supply date 32 in our PBS data (ie, the time in days from one supply date to the next). The 75th percentile estimate of drug duration for all RASI and β-blockers was 35 days, excluding metoprolol tartrate which was 50 days. This was consistent with PBS prescriptions which are intended to approximate monthly supplies for most drugs, with some exceptions such as metoprolol tartrate for which there was an extended supply. 33 In further confirmation, we also checked whether these estimates of duration were reasonable by comparing with the values calculated from dosage. However, dosage information is not currently available from PBS data, so we reviewed the registered product information for each drug and assumed that RASI were 1 pill per day, except for enalapril (2 per day) and captopril (3 per day), and 1 pill per day for β-blockers, except for metoprolol tartrate and carvedilol which were assumed to be 2 pills per day. 31 Based on these dosage assumptions, as well as actual values for quantity and strength from the PBS data, the assumed duration was calculated as (quantity dispensed × drug strength)/assumed dose. This showed consistency between the estimated duration calculated from the 75th percentile and duration calculated from the assumed dose. Hence, we proceeded with the 75th percentile method as our estimate of duration of cover as this was based on actual data (supply dates) rather than assumed doses.
Discontinuation of RASI and β-blockers was defined as the first break (stop) in therapy of ≥90 days following a previous supply within the period from first supply date to 3-year postdischarge or date of death. 30 There are 4 possible scenarios to separate discontinued from persistent medication users: (1) if a patient dies prior to the end of the study period and is persistent up to 90 days before that point, this patient was coded as a persistent user, (2) if a patient experiences a stop gap of ≥90 days prior to death, they were coded as a discontinued user, (3) if a patient was still alive and persistent up to the end of study period, this patient was coded as a persistent user, and (4) if a patient was still alive and experienced a stop gap prior to the end of study period, they were coded as a discontinued user.
Statistical Methods
Categorical variables were presented as proportions and continuous variables as mean and standard deviation or median and interquartile range, as appropriate. The proportion of patients who discontinued medications in each drug class was determined by the Kaplan-Meier estimator with time measured from first supply date to the first break in therapy with censoring for death or end of follow-up. Medication adherence is presented as mean, median, and categorical PDC levels (low adherence: 1%-50%, medium adherence: 51%-79%, and high adherence: ≥80%) at 1, 2, and 3 years following HF discharge. We transformed the adherence measure as
Sensitivity Analysis
Stratified analyses investigated adherence patterns in patients with (vs without) CKD and COPD, in view of the precautions related to RASI and β-blocker use in these conditions, respectively. 1 In addition, we performed a separate subgroup analysis in patients who survived 1 year post-HF discharge to examine whether their prior adherence to RASI or β-blockers predicted their subsequent adherence to the same drug class. The sensitivity analyses are reported in the Supplementary File.
Results
Baseline Characteristics
The characteristics of the study cohort and for RASI and β-blocker user subgroups stratified by 3-year adherence level (PDC < 80% vs PDC ≥ 80%) are shown in Table 1. The study cohort comprised 4488 patients, mean age 76.6 years, with 57% males. Prevalent comorbidities (>30%) in the cohort included IHD, hypertension, atrial fibrillation (AF), diabetes, COPD, and CKD. Over 50% of patients had a Charlson comorbidity score ≥3 and patients were taking an average of 6 main classes of medications in the 6-month period prior to the index HF admission. Within this cohort, 77.4% used a RASI and 52.7% used a β-blocker within 60 days upon leaving hospital. Among RASI users, 71.8% and 35.1% were on a RASI and β-blocker, respectively, in the 6-month period prior to index admission, and conversely, 68.4% and 52.1% of β-blocker users were on an RASI and a β-blocker, respectively, prior to index admission.
Characteristics of Patients Dispensed RASI and β-Blockers Post-HF Discharge, Stratified by their Adherence Level (PDC < 80% and PDC ≥ 80%) to RASI and β-Blockers at End of the 3-Year Follow-Up.
Abbreviations: ARIA, Accessibility and Remoteness Index of Australia; CCB, calcium channel blockers, HF, heart failure; MRA, mineralocorticoid receptor antagonists; PDC, proportion of days covered; RASI, renin–angiotensin system inhibitors.
a Identified with a fixed 20-year look-back period prior to the index HF admission.
b Identified as at least 2 scripts within a fixed 6-month look-back period prior to the index HF admission.
c Drugs from the top level of Anatomical Therapeutic Chemical code (A, B, C, G, H, L, M, N, R).
Among users, 64.6% and 47.5% were adherent (PDC ≥ 80%) to RASI and β-blockers, respectively, over a 3-year period. Negative associations of adherence to RASI were age ≥75 years, COPD, CKD, higher Charlson comorbidity score (>4), and prior use of diuretics and antipsychotic/psychoanaleptic agents (all P < .05). Negative associations of adherence to β-blockers were age ≥75 years, earlier years of index admission, COPD, CKD, and higher Charlson comorbidity score (>4; all P < .05). Positive associations of both RASI and β-blocker adherence were age <75 years, low Charlson comorbidity score and prior RASI, β-blocker, or statin use (all P < .05). There was no association between indigenous status and residential location with adherence to either medication group, although there were only a small number of Indigenous patients and patients living in remote and very remote areas (Table 1).
Discontinuation of and Adherence to Drug Groups
Among RASI and β-blocker users, 17% and 29% of patients, respectively, first discontinued (stop gap ≥90 days) their medication group within 1-year follow-up, and the proportions increased to 28% and 42% at 3-year follow-up (both P trend <.0001; Figure 1). Among those who had a first discontinuation gap in their RASI and β-blocker treatment, 44.3% and 59.2%, respectively, reinitiated their treatment within the subsequent 12 months.
Proportion of RASI and β-blocker users who discontinued medication for ≥90 days from first drug supply to 3 years post-HF hospital discharge. HF indicates heart failure; RASI, renin-angiotensin system inhibitors.
Figure 2 shows long-term trends for PDC adherence to RASI and β-blockers at 1, 2, and 3 years postdischarge. The adherence distributions for both drug groups are left skewed so that the median is consistently higher than the mean adherence values. The median adherence to RASI and β-blockers declined to 87.9% and 77.9%, respectively, at 3-year follow-up (both trend P < .05), with the largest falls in adherence occurring within the first year for both drug groups. At the end of the first year, 72.0% and 53.8% of RASI and β-blocker users, respectively, were still adherent (PDC ≥ 80%) and the proportions declined further to 64.6% and 47.5%, respectively, at 3-year follow-up (both trend P < .0001).
Long-term trends of medication adherence for RASI and β-blockers from heart failure discharge presented as categorical variables (PDC adherence of 1%-50%, 51%-79%, and ≥80%) and mean and median adherence. Trend test for mean and median medication adherence to RASI (both P < .0001) and β-blockers (mean: P = .02, median: P < .0001). Trend tests for medication adherence to RASI in 1% to 50% (P < .0001), 51% to 79% (P = .037), ≥80% (P < .0001); trend tests for medication adherence to β-blockers in 1% to 50% (P < .0001), 51% to 79% (P = .29), ≥80% (P < .0001). IQR indicates interquartile range; PDC, proportion of days covered; RASI, renin-angiotensin system inhibitors.
The discontinuation and adherence patterns over 3 years in subgroups of HF patients with and without CKD and COPD are shown in Supplementary Tables 1 and 2 and Figures 1 and 2. As expected, the proportion of patients who discontinued RASI was higher and adherence levels were lower in patients with coexistent CKD versus no CKD (Supplementary Table 1 and Figure 1). The same observation held for discontinuation and adherence for β-blockers in patients with coexistent COPD versus no COPD (Supplementary Table 2 and Figure 2).
Independent Predictors of Lower Adherence to Medications
Table 2 shows the independent predictors of lower adherence (PDC < 80%) to RASI and β-blockers at 3-year follow-up. The independent predictors of RASI PDC < 80% were older age, CKD, higher Charlson comorbidity score, and prior use of diuretics or antipsychotic/psychoanaleptic agents, while prior use of a RASI or statin was a positive predictor of adherence. The independent predictors of β-blocker PDC < 80% were older age, COPD, higher Charlson comorbidity score, and prior use of antiarrhythmic agents. Conversely, a more recent year of index HF admission and prior use of a RASI or β-blocker were positive predictors of adherence. The c-statistics for the logistic regression models for predictors of PDC < 80% was 0.61 for RASI and 0.65 for β-blockers. The significant predictors of discontinuation were essentially the same as those for lower adherence (PDC < 80%) to RASI or β-blockers (Supplementary Table 3). With interaction analysis, only the COPD × age-group term was borderline significant for discontinuation and lower adherence to β-blockers at 3 years (both P = .04), and further sensitivity analysis for these outcomes was not considered justified.
Multivariable Logistic Regression Models for Predictors of 3-Year Low Adherence (PDC < 80%) to RASI and β-Blockers Post Heart Failure Discharge.
Abbreviations: AF, atrial fibrillation; CCB, calcium channel blockers; CKD, chronic kidney disease, COPD, chronic obstructive pulmonary disease, HF, heart failure, IHD, ischemic heart disease, MRA, mineralocorticoid receptor antagonists, OR, odds ratio; PAD, peripheral artery disease, RASI, renin–angiotensin system inhibitors.
a C-statistic: 0.61 for RASI, 0.65 for β-blockers.
b Used univariate logistic regression model to test the interaction terms.
In a sensitivity analysis, PDC < 80% for RASI and β-blockers in the year prior to the index HF admission was associated with lower adherence to the corresponding drug groups in the year after HF discharge, with an adjusted OR of 2.08 (95% CI: 1.83-2.25) and OR 2.73 (95% CI: 2.39-3.10), respectively (Supplementary Table 4).
Discussion
Our population-based study addressed the knowledge gap in long-term (3 years) adherence to first-line HF pharmacotherapies, notably RASI and β-blockers, among senior patients following hospitalization for HF. Only 65% of RASI and 47% of β-blocker users remained adherent (PDC ≥ 80%) to these medications at 3-year follow-up. Lower adherence (PDC < 80%) and first discontinuation gaps ≥90 days for both pharmacotherapies were highest in the first year after hospital discharge, indicating the difficulty in maintaining continuous use of these drugs during this period. We identified several important clinical predictors of long-term adherence although only a modest proportion of the variability in adherence could be explained by these patient-related factors. These results have clinical implications for quality care programs after HF hospitalization, especially if long-term outcomes for HF are to improve.
Our population-based study represents a “real-world” cohort of hospitalized HF patients who are generally older, have multiple comorbidities, and include a higher proportion of females than HF patients usually included in clinical trials. 35 -37 However, even under ideal circumstances of a clinical trial with frequent patient monitoring, drug discontinuation of 20% to 30% has been reported. 35 -37 The clinical characteristics of our senior cohort are comparable to that reported in a recent representative cross-sectional study of consecutive patients admitted with HF in another Australian state. 38 Compared to the latter representative cohort, our study population has a similar mean age (76 years), proportion of males (56%), Charlson comorbidity score >3, and similar RASI and β-blocker usage at the time of discharge from hospital.
Discontinuation Gaps and Long-Term Adherence
Suboptimal adherence to effective medications remains one of the most important obstacles to successful chronic disease management. 39 We have previously shown in this same HF cohort that patients dispensed a RASI and/or a β-blocker within 60 days posthospital discharge had a higher propensity-adjusted survival at 1 year. 24 It is crucial to understand their long-term adherence to these survival-influencing pharmacotherapies and barriers to continuing adherence. Our results demonstrated that discontinuation gaps in the use of RASI and β-blockers were frequent and increased in frequency over time. Although breaks in therapy are common, 13,14 this is a dynamic process, and even for breaks of 90 days or longer, 40% to 60% of our patients reinitiated their therapy within 1 year. Hence, discontinuation gaps cannot be directly equated to permanent discontinuation or nonpersistence. 30 Some gaps may have been due to hospitalization, which we could identify, and for these, we assumed that patients continued to use their current treatment. Hence, hospital length of stay was not counted as part of a gap in treatment.
Measuring medication adherence in this study accounts for stop–restart patterns over the entire follow-up period. We used the PDC because it is a widely accepted method to measure adherence in population-based administrative data sets. 19,30 The observed low to moderate levels of 1-year adherence (PDC ≥ 80%) to β-blocker therapy (54%) and RASI (72%) in our HF cohort are within the range reported for these drug groups in other cohort studies using a comparable measure. 15 -18 We have extended the adherence observation period out to 3 years compared to the usual shorter term studies reported elsewhere. It was reassuring that in our “real-world” cohort, there were only small, albeit significant, absolute declines in the mean/median adherence to RASI and β-blockers, as well as the proportions remaining adherent, after the first year of follow-up. Conversely, we found that HF patients who had PDC < 80% to these 2 therapies in the year prior to the index HF admission were also twice as likely to have PDC < 80% in the subsequent year. This highlights the value of interventions to optimize treatment adherence in the early months after leaving hospital to maintain good adherence over the longer term.
We found consistently higher discontinuation and lower long-term adherence levels for β-blockers compared to RASI, which has also been reported elsewhere. 13,14,16,18 This suggests that β-blockers may be less well tolerated than RASI in older HF patients, in whom coexisting COPD and AF might be more prevalent. This highlights the importance of using cardioselective β-blockers in HF patients with COPD because of not only their lower adverse effects than some other β-blockers 40 but also proven prognostic benefit. 41
Predictors Associated With Lower Adherence to Treatment
Nonadherence is a complex multidimensional problem, but some well-known clinical predictors such as older age and a high comorbidity burden were confirmed in our real-world study. 6,22,42 This emphasizes the need for targeted strategies to support drug adherence in older HF patients affected by multiple chronic diseases, polypharmacy, and cognitive impairment. 7 Surprisingly, neither Indigenous status nor remote geographic residence was associated with adherence, although these patients represented only a minority of our study cohort, which would limit our ability to detect an effect for these factors. As expected, patients with CKD were more likely to have lower adherence to RASI due to their risk of renal side effects and hyperkalemia. We also showed that COPD is consistently the most powerful predictor of lower adherence to β-blockers in HF patients, even though cardioselective β-blockers can be safely used in the majority of HF patients with COPD and are associated with the same mortality benefits as patients without COPD. 41,43 Prior use of antipsychotic/psychoanaleptic agents was an independent predictor for lower adherence to RASI probably because depression is frequent in HF patients and has an adverse impact on adherence. 22,42 It has been shown that adherence usually declines most rapidly in new users of medication. 44 Our study confirmed that patients taking a RASI or a β-blocker prior to their index HF hospitalization were more likely to be adherent to their therapies over the long-term. This raises the need for patient education and counseling support when new treatment regimens are initiated. 7,22
We are unable to ascertain the actual reasons for drug discontinuation or nonpersistence from administrative data. Ultimately, the patient-related factors associated with adherence to medications in our logistic regression models explained only a small proportion of the variability in adherence as indicated by the c-statistics (a measure of discrimination for a predictive model). This highlights that nonadherence is a multidimensional problem, including social and economic factors, health care system-related factors, condition-related factors, and patient-related factors in patients with HF. 42 In order to formulate effective and targeted interventions, a more detailed understanding of the generally applicable and locally specific modifiable determinants of medication adherence among HF patients is required. Hence, multimodal approaches and related interventions have to be adopted to improve medication adherence. 11,45 For example, patient and caregiver education programs on the importance of medications and medication adherence, HF self-care and management programs, improved discharge/postdischarge care, and bundled interventions (including telephone monitoring, smoking cession courses, home visits, advisory hotline) were shown to have a positive effect on adherence to medication treatments and on mortality, 11,45 as well as use of cardioselective β-blockers as previously discussed. Patient education and regular contact with health professionals would be important factors in maintaining adequate long-term adherence to medication regimens for improved survival in HF. This may increase the likelihood of a drug being ceased (either by the patient or from professional advice) due to adverse effects, which may then be observed as low adherence in administrative data (depending on available data and method of calculating adherence). However, this would depend on the proportion of patients who experience adverse effects. In clinical practice, there are positive benefits from maintaining patient education and regular contact with health professionals, especially nurses and pharmacists, which improve medication adherence when there are no side effects. 11
Strengths and Limitations
This is a well-characterized population-based “real-world” HF cohort with access to individual-linked medication dispensing data for the cohort over 3 years after index HF hospitalization. The cohort had to survive at least 60 days after hospital discharge to minimize selection biases of ascertainment and survivor treatment by ensuring that patients had an adequate opportunity to fill prescriptions and enable calculation of adherence. The study was limited by a lack of detailed clinical (eg, functional HF status) and echocardiographic data to differentiate HF phenotypes. Although these data may be relevant when estimating the fraction of patients who require therapy according to guidelines, they are less important to a study of treatment adherence because patients started on medications were considered to have an indication for the therapy by the treating physician. Although our study confirmed some important predictors for adherence to RASI and β-blockers, other predictors reported in the literature (eg, clinical depression) were not measured in this study. Our PBS data set contains dispensing data but not the doses prescribed, and we are unable to assess if patients were on the optimal dosage of each medication. Finally, the PDC provides only an estimate of adherence based on prescription refills, and true patient compliance or consumption of the medications cannot be measured from administrative data. We are only seeking to obtain a measure of what we are calling adherence in the context of electronic administrative data, and further work is required to standardize the terminology when applied to such data.
Conclusion
This population-based study documents the 3-year declining pattern of adherence to EBMs in senior HF patients. Furthermore, preventing avoidable discontinuation gaps and obtaining good adherence to prescribed medical therapy in the first year after HF hospitalization may be key to maintaining high adherence longer term. Our findings have important implications for clinicians, health policy makers, and health care providers seeking to optimize late clinical outcomes after discharge from hospital for HF. However, more research is required to devise and test strategies to improve long-term adherence and minimize discontinuation.
Supplemental Material
Supplementary_file_long-term_adherence_HF_1 - Long-Term Adherence to Renin–Angiotensin System Inhibitors and β-Blockers After Heart Failure Hospitalization in Senior Patients
Supplementary_file_long-term_adherence_HF_1 for Long-Term Adherence to Renin–Angiotensin System Inhibitors and β-Blockers After Heart Failure Hospitalization in Senior Patients by Xiwen Qin, Joseph Hung, Tiew-Hwa Katherine Teng, Tom Briffa and Frank M. Sanfilippo in Journal of Cardiovascular Pharmacology and Therapeutics
Footnotes
Acknowledgments
The authors thank the following institutions for providing the data used in this study, the Australian Department of Health for the cross-jurisdictional linked PBS data, and staff at the WA Data Linkage Branch and data custodians of the WA Department of Health Inpatient Data Collections and Registrar General for access to and provision of the State linked data.
Authors’ Note
The authors will consider requests for data sharing on an individual basis, with the aim to share data whenever possible for appropriate research purposes. However, this research project uses data obtained from a third-party source under strict privacy and confidentiality agreements from the Western Australian Department of Health (State) and Australian Department of Health (Federal) databases, which are governed by their ethics committees and data custodians. The data were provided after approval was granted from their standard application processes for access to the linked data sets. Therefore, any requests to share these data with other researchers will be subject to formal approval from the third-party ethics committees and data custodians. Researchers interested in these data should contact the Client Services Team at the Data Linkage Branch of the Western Australian Department of Health (
). This study complies with the Declaration of Helsinki. Ethics approvals were obtained from The University of Western Australia (ref RA/4/1/8065), Western Australian Department of Health (ref 2014/11), Western Australian Aboriginal Health Ethics Committee (ref 572), and the Commonwealth Department of Health (XJ-16). This included a waiver of informed consent.
Author Contributors
XQ was a PhD student during the work and had input in the study design, carried out all the analyses, interpreted the results, wrote the first draft of the manuscript and updated the manuscript. FMS, JH, TB and THT supervised XQ, designed the study and analyses, provided clinical interpretation of the findings, reviewed and revised the manuscript drafts. FMS obtained the study datasets. FMS, JH and TB are chief investigators on the grant and obtained study funding. All authors gave approval for the final version of the manuscript.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by a project grant from the National Health and Medical Research Council of Australia (NHMRC project grant 1066242).
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
