Sources of data

In France, the general health insurance scheme covers 70% of the population (i.e. 48 million people in 2006) and its information system ( système national d’informations inter-régimes de l’assurance maladie [SNIIRAM]) contains individualized, anonymous and exhaustive data on all health-spending reimbursements. Further information is recorded, such as full refund for people with one of 30 long-term diseases (LTDs), including heart disease, and the existence of full healthcare coverage for low earners (below an income ceiling of €7500 per year). Moreover, vital status (all-cause mortality) stemming from the French National Institute of Statistics and Economic Studies is available in the SNIIRAM system. This information can be linked to the French hospital discharge database ( programme de médicalisation des systèmes d’information [PMSI]), which provides medical information for all patients discharged from both private and public hospitals, including the World Health Organization International Statistical Classification of Diseases and Related Health Problems 10th Revision (ICD-10) diagnostic codes.

Patient population

All hospitalizations from January to June 2006 with a diagnosis-related group of MI were selected from the PMSI data on short-term hospital stays. We selected all patients covered by the general health insurance scheme. For each patient, the first hospital admission for MI in the first half of 2006 was considered as the index hospital admission. These data were linked, using a common, anonymous patient number, to the corresponding records in the reimbursement database. Patients who died or were lost to follow-up during the first 6 months were excluded from the analysis, because their adherence behaviour could not be determined accurately. Possible reasons for loss to follow-up included a move abroad and a change in social security number (e.g. for widowed women). If a patient was lost to follow-up beyond 6 months, they were included in the analysis as a censored observation, with the date of the last refund as the censoring point. All patients were followed for up to 30 months. For each EBT class (beta-blockers, statins, ACEIs/ARBs, aspirin/clopidogrel), the analysis included all patients who had at least one refund for this medication class after index admission. For the medication combination, all patients with at least one refund for each class were included.

Variables studied

For each index hospital admission, the following data were collected: sex, age, use of angioplasty, stent implantation, and coronary artery bypass graft procedures during the index hospital admission or within 30 days after discharge, according to the classification of medical procedures used in France. Prior hospitalization for cardiovascular events (coronary artery disease [acute MI, angina pectoris, coronary atherosclerosis], coronary bypass surgery and stent implantation procedures) were identified using a specific diagnosis-related group notified for hospital stays in the 6 months preceding index hospital admission. The presence of a cardiovascular LTD 6 months prior to index hospital admission and full healthcare coverage for low earners were also recorded.

Comorbid conditions were sought and defined in several ways, taking into account the entire period between 2005 and 2008. Patients with depression or Parkinson’s disease were identified through reimbursements of specific medications by the presence of two dispensations at most, within a 6-month interval. Diabetes and pulmonary diseases were characterized by the reimbursement of indicator medications at least twice over one calendar year. Inflammatory symptoms were identified by regular reimbursements of steroidal or non-steroidal anti-inflammatory drugs (i.e. the proportion of days covered by a filled prescription greater than 80% in the 30-month follow-up period). For psychiatric disorders and chronic liver disease, the occurrences of specific LTDs were used. For neoplasia, the following indicators were used: radiotherapy or chemotherapy sessions, hospital admission with a main diagnosis of cancer, or a specific cancer LTD beginning in 2004 or later. Alzheimer’s disease was defined by the specific LTD or by reimbursements of specific medications issued at least twice within one calendar year or by hospital admission with a specific dementia diagnosis-related group. For chronic renal disease, patients with a ‘chronic renal’ LTD and/or treated with dialysis and/or with immunosuppressive drugs were selected.

For each EBT class, drug adherence was assessed by a medication possession ratio during the 30-month follow-up period after hospitalization for MI, i.e. between the day after index hospital discharge and the last day of presence in the study (end of follow-up or lost to follow-up). More specifically, we used two different measures depending on the medication class. For beta-blockers, statins and ACEIs/ARBs, we calculated the proportion of days on which a patient had pills available (considering one pill per day as necessary). For aspirin/clopidogrel, where combined treatment is not unusual, adherence was measured by the ratio of the number of dates of supply to the number of months in the study. For each medication class, patients were classified in two groups: adherent patients (adherence measure > 80%) and nonadherent patients (otherwise). A patient was considered to be adherent to the combined treatment if they were adherent to each of the four medication classes. Moreover, consumption during the 6 months before the index hospital admission was considered to be ‘regular’ after at least three medication refunds. Health outcome was assessed as death or hospital readmission for acute coronary syndrome (ACS) within the follow-up period.

Statistical analysis

For each recommended class of medication, as well as for their combined use, we compared adherent and nonadherent patients according to their baseline characteristics, in-hospital management and use of concomitant medications after index hospital admission, by calculating the proportions of nonadherent patients for each characteristic and the crude odds ratios (adherent versus nonadherent patients). To explore the crude relationship between age and adherence in more detail, we used locally-weighted linear regression, highlighting non-linear trends graphically. Adjusted odds ratios were determined using multivariable logistic regression models as a function of age, sex, full healthcare coverage for low earners, comorbidities, prior hospitalization for cardiovascular condition, ‘cardiovascular’ LTD, use of the considered medication class in the 6 months before index admission, myocardial revascularization procedures during and 1 month after the index admission, deliveries and adherence to the other medication classes.

For each medication class, crude ACS-free survival rates in adherent and in nonadherent patients were calculated using the Kaplan-Meier method. For further comparisons, hazard ratios were derived from Cox proportional hazards models, both univariate and multivariable (using the above-mentioned characteristics as covariates). In addition, adjusted ACS-free survival rates were calculated with their 95% confidence interval (CI) .

A P -value less than 0.05 was considered significant. All analyses were performed using SAS software (SAS version 9.1.3, SAS Inc., Cary, NC, USA).

Determinants of adherence

The proportions of patients with an insufficient adherence measure (< 80% during the observation period) were 32.0% for beta-blockers, 24.0% for statins, 22.7% for ACEIs/ARBs, 18.3% for aspirin/clopidogrel and 50.0% for combined EBT. Tables 1 and 2 show the relationships between drug adherence and patient characteristics. After adjustment for these patient characteristics, adherence to EBT was found to be decreased significantly by the following factors: age greater than 74 years; comorbidities (neoplasia, renal disease, chronic obstructive pulmonary disease, Alzheimer’s disease, Parkinson’s disease and depression); and full healthcare coverage for low earners. Conversely, prior combination refund and stent implantation before or during index hospitalization increased adherence. For each separate medication class, adherence was associated positively with adherence to the three other classes as well as with prior use of the medication class under consideration. Moreover, adherence to beta-blockers was decreased in patients with renal disease, chronic obstructive pulmonary disease or depression. Adherence to statins was decreased by presence of diabetes mellitus, Alzheimer’s disease, depression or full refund for cardiovascular disease. For ACEIs/ARBs, adherence was decreased by renal disease or depression.

Table 1

Univariate analysis of factors associated with adherence to evidence-based treatment in the 30-month follow-up period after hospitalization for myocardial infarction in France.

	Beta-blockers ( n = 9937)		Statins ( n = 10,501)		ACEIs/ARBs ( n = 9823)		Aspirin/clopidogrel ( n = 1056)		Combination ( n = 8249)
	Nonadherent (%)	OR a	Nonadherent (%)	OR a	Nonadherent (%)	OR a	Nonadherent (%)	OR a	Nonadherent (%)	OR a
Total	32.0		24.0		22.7		18.3		50.0
Age (years)
≤ 44	37.9	0.70 e	29.0	0.60 e	26.7	0.64 e	24.3	0.51 e	52.4	0.85 b
45–54	33.1	0.86 b	23.9	0.77 d	23.4	0.76 d	17.7	0.76 d	47.7	1.02
55–64	29.2	1.03	18.5	1.08	19.4	0.96	13.2	1.08	45.4	1.12
65–74	29.9	1	19.6	1	18.8	1	14.1	1	48.3	1
75–84	30.7	0.96	27.7	0.64 e	24.6	0.71 e	21.2	0.61 e	55.6	0.74 e
≥ 85	40.3	0.63 e	42.6	0.33 e	33.3	0.46 e	30.9	0.37 e	67.1	0.46 e
Sex
Female	31.7	1	27.4	1	25.5	1	22.2	1	53.3	1
Male	32.1	0.98	22.6	1.29 e	21.6	1.24 e	16.7	1.42 e	48.9	1.19 d
Full healthcare coverage for low earners
No	31.7	1	23.0	1	22.1	1	17.5	1	49.6	1
Yes	35.5	0.85	37.7	0.49 e	33.4	0.56 e	30.6	0.48 e	55.9	0.78 c
Comorbidities
Neoplasia	32.5	0.98	26.9	0.84 b	25.8	0.83 a	20.7	0.84 b	54.8	0.81 b
Diabetes mellitus	29.0	1.20 d	25.2	0.91	21.8	1.08	17.1	1.12	48.7	1.07
Renal disease	43.0	0.62 c	36.6	0.54 d	45.5	0.35 e	20.7	0.86	67.3	0.48 d
COPD	36.3	0.80 d	25.4	0.91	23.0	0.98	17.7	1.05	57.9	0.70 e
Alzheimer’s disease	39.7	0.71 b	41.6	0.43 e	30.4	0.67 c	31.9	0.47 e	67.6	0.47 e
Parkinson’s disease	38.2	0.76	36.0	0.56 c	30.5	0.67	25.6	0.65 b	62.9	0.59 b
Depression	34.8	0.84 d	28.0	0.76 e	26.4	0.77 e	20.2	0.85 c	54.9	0.78 e
Steroidal or non-steroidal anti-inflammatory use	29.0	1.15	21.2	1.18	26.3	0.82	14.5	1.33	55.5	0.80
Chronic hepatic disease	33.3	0.94	33.3	0.63	33.3	0.59	36.1	0.39 d	64.3	0.55
Psychotic disorder	36.0	0.83	31.9	0.67 c	30.6	0.66 c	24.1	0.70 b	56.7	0.76
Six months prior to index admission
Cardiovascular disease LTD	29.2	1.18 c	27.9	0.77 e	23.2	0.97	20.5	0.83 c	51.1	0.95
Prior admission for cardiovascular reason
None	32.1	1	23.4	1	22.5	1	18.1	1	49.8	1
Stent implantation	23.1	1.58 c	17.6	1.43 b	18.3	1.29	12.7	1.52 b	40.5	1.46 c
Other diagnoses	33.8	0.92	32.6	0.63 e	27.7	0.75 d	22.7	0.75 c	56.7	0.76 c
Prior use of medications
Beta-blockers	22.9	1.83 e		–		–		–	45.8	–
Statins		–	17.9	1.61 e		–		–	44.6	–
ACEIs/ARBs		–		–	17.4	1.65 e		–	47.0	–
Aspirin/clopidogrel		–		–		–	16.1	1.22 d	49.3	–
Combination		–		–		–		–	39.0	1.62 e
Index admission for MI
No procedures	34.7	1	31.9	1	28.7	1	28.4	1	57.8	1
Coronary artery bypass graft	29.2	1.29	14.9	2.67 e	24.3	1.25	16.9	1.95 d	52.2	1.25
Angioplasty without stent implantation	28.7	1.32	22.0	1.66 d	21.6	1.46 b	17.2	1.91 e	48.2	1.47 c
Stent implantation	31.1	1.17 d	21.3	1.73 e	20.2	1.59 e	14.1	2.41 e	47.5	1.51 e
Concomitant medication use after index admission
Beta-blockers
No use		–	33.8	0.33 e	28.4	0.43 e	29.4	0.24 e	–	–
Nonadherent use		–	41.3	0.24 e	38.6	0.27 e	33.5	0.20 e	–	–
Adherent use		–	14.4	1	14.7	1	9.0	1	–	–
Statins
No use	45.9	0.37 e		–	37.8	0.26 e	39.8	0.11 e	–	–
Nonadherent use	56.5	0.24 e		–	48.6	0.17 e	49.1	0.07 e	–	–
Adherent use	23.7	1		–	13.8	1	6.7	1	–	–
ACEIs/ARBs
No use	40.8	0.46 e	31.1	0.39 e		–	25.2	0.33 e	–	–
Nonadherent use	53.9	0.27 e	51.2	0.17 e		–	42.9	0.15 e	–	–
Adherent use	24.2	1	15.1	1		–	9.9	1	–	–
Aspirin/clopidogrel
No use	40.8	0.49 e	37.0	0.29 e	40.5	0.27 e	–		–	–
Nonadherent use	63.4	0.20 e	69.5	0.07 e	55.5	0.15 e	–		–	–
Adherent use	25.4	1	14.5	1	15.4	1	–		–	–

Only gold members can continue reading. Log In or Register to continue

Share this:

• Click to share on Twitter (Opens in new window)
• Click to share on Facebook (Opens in new window)

Related

Related posts:

1. Cryoballoon for pulmonary vein isolation: Is it better tolerated than radiofrequency? Retrospective study comparing the use of analgesia and sedation in both ablation techniques
2. Factors influencing the level of circulating procoagulant microparticles in acute pulmonary embolism
3. Right pneumothorax resulting from an atrial lead inserted through the left cephalic vein
4. Unusual evolution of an aneurysm of the proximal left anterior descending coronary artery 15 years after a Simpson atherectomy

Stay updated, free articles. Join our Telegram channel

Join