Accepted Manuscript Factors influencing the provision of adherence support by community pharmacists: A structural equation modelling approach Ms Sarab M. Mansoor, BPharm, Dip Clinical Pharm, MSc, PhD candidate, Ines Krass, BPharm, Dip Hosp Pharm, Grad Dip Educ Studies (Health Ed), PhD, Professor, Dr Daniel Costa, BSc(Hons), PhD, Parisa Aslani, BPharm(Hons), MSc, PhD, Grad Cert Ed Stud (Higher Ed), Associate Professor PII:
S1551-7411(15)00031-5
DOI:
10.1016/j.sapharm.2015.01.004
Reference:
RSAP 591
To appear in:
Research in Social & Administrative Therapy
Received Date: 10 December 2014 Revised Date:
23 January 2015
Accepted Date: 23 January 2015
Please cite this article as: Mansoor SM, Krass I, Costa D, Aslani P, Factors influencing the provision of adherence support by community pharmacists: A structural equation modelling approach, Research in Social & Administrative Therapy (2015), doi: 10.1016/j.sapharm.2015.01.004. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
1
Factors influencing the provision of adherence support by community
2
pharmacists: A structural equation modelling approach
4
Author details
5
Ms Sarab M Mansoor
6
BPharm, Dip Clinical Pharm, MSc, PhD candidate
7
Address:
RI PT
3
SC
Faculty of Pharmacy, Pharmacy and Bank Building, A15 The University of Sydney
9
Sydney, NSW 2006, Australia
M AN U
8
10
E-mail:
[email protected]; [email protected]
11
Telephone:
(+61 2) 9351 6066
12
Professor Ines Krass
14
BPharm, Dip Hosp Pharm, Grad Dip Educ Studies (Health Ed), PhD
15
Address:
TE D
13
Faculty of Pharmacy, Pharmacy and Bank Building, A15 The University of Sydney
17
Sydney, NSW 2006, Australia
AC C
EP
16
18
E-mail:
[email protected]
19
Telephone:
(+61 2) 9351 3507
20
Fax:
(+61 2) 9351 4391
21 22
Dr Daniel Costa
23
BSc(Hons), PhD
ACCEPTED MANUSCRIPT
School of Physiology, Lifehouse Building (C39Z)
25
The University of Sydney
26
Sydney, NSW 2006, Australia
27
E-mail:
[email protected]
28
Telephone:
(+61 2) 9351 6304
29
Fax:
(+61 2) 9036 5223
30
Associate Professor Parisa Aslani
32
BPharm(Hons), MSc, PhD, Grad Cert Ed Stud (Higher Ed)
33
Address:
M AN U
31
Faculty of Pharmacy, Pharmacy and Bank Building, A15
34
The University of Sydney
35
Sydney, NSW 2006, Australia E-mail:
[email protected]
37
Telephone:
(+61 2) 9036 6541
38
Fax:
(+61 2) 9351 4391
EP
40
TE D
36
39
RI PT
Address:
SC
24
Conflict of interest disclosure: The author(s) declare no potential conflicts of interest with
42
respect to the research, authorship, and/or publication of this article.
43
AC C
41
ACCEPTED MANUSCRIPT
Factors influencing the provision of adherence support by community pharmacists: A structural equation modelling approach
4 5 6
Abstract
7
optimum patient outcomes. Community pharmacists are well placed to deliver interventions to
8
support adherence.
9
Aims: To investigate community pharmacists’ activities in supporting patient adherence; and
RI PT
1 2 3
SC
Background: Non-adherence to medication represents an important barrier to achieving
identify factors influencing pharmacists’ provision of adherence support.
11
Method: A random sample of 2,020 Australian community pharmacies was mailed a
12
questionnaire investigating their provision of adherence support. The self-completed, structured
13
questionnaire consisted of eight sections, five of which were relevant to this study: strategies
14
used to identify non-adherent patients, strategies used to support patients’ adherence to
15
medications, pharmacists’ attitudes towards provision of adherence support, perceived barriers to
16
provision of adherence support, and demographics. Structural equation modelling (SEM) was
17
used to determine potential influencing factors.
18
Results: A response rate of 31% was achieved (n=627). Pharmacists reported using strategies to
19
identify non-adherent patients for less than half (45%) of the prescriptions dispensed. A mean of
20
8.4±14.9 (Mean± SD) strategies was used by respondents in the 7 days prior to survey
21
completion. Dose administration aids was the most commonly used strategy (provided by 96.5%
22
of respondents). Time pressure for patients (68%) was perceived by pharmacists as the main
23
barrier to adherence support. SEM identified “stakeholders/skills” and “number of full time
24
equivalent staff” as influencing provision of adherence support strategies.
AC C
EP
TE D
M AN U
10
1
ACCEPTED MANUSCRIPT
25
Conclusion: Provision of adherence support by pharmacists was episodic and infrequent,
26
impeded by a number of barriers. By addressing barriers, it is possible to enable pharmacists to
27
become more proactive and effective in supporting patient adherence.
29
RI PT
28
Keywords: Adherence to medications, pharmacists, barriers, attitudes, Australia.
AC C
EP
TE D
M AN U
SC
30
2
ACCEPTED MANUSCRIPT
Introduction
32
Medication non-adherence has been widely recognized as a continuing problem among patients
33
with chronic diseases, including hypertension, diabetes and lipid disorders.1-4 Non-adherence has
34
been associated with higher mortality and morbidity as well as increased health care costs.5,6 In
35
developed countries, among patients with chronic diseases on long-term therapy, the average
36
medication adherence rate is approximately 50%, with some variation between conditions.1,2
37
A large body of research has focused on the common barriers to medication adherence.1,7,8 These
38
include: patient-related factors such as knowledge, beliefs, attitudes and/or expectations; social
39
and economic factors such as lack of social support and/or poor socioeconomic status; health
40
care team and system-related factors such as poorly developed or fragmented health care
41
systems and/or lack of knowledge and training of health care providers; condition-related factors
42
such as severity of symptoms, severity of disease and/or rate of progression; and therapy-related
43
factors such as complexity of the medical regimen, duration of treatment and/or side effects.
44
Recognition of these barriers has influenced the development of a variety of targeted strategies
45
intended to enhance adherence and persistence to therapy. For more than three decades,
46
researchers and healthcare professionals have sought to promote and improve patients’
47
adherence to medication by interventions which primarily focus on fostering behavioural
48
change.9 Interventions aimed at promoting adherence to therapy have resulted in improvements
49
in adherence and in some cases, improvements in clinical outcomes.2,10 The goal is to encourage
50
and empower the individual patient to learn, adopt, and sustain a regular pattern of medication-
51
taking behaviour.1
AC C
EP
TE D
M AN U
SC
RI PT
31
3
ACCEPTED MANUSCRIPT
As healthcare professionals, community pharmacists are well placed to identify non-adherence
53
and offer support to patients to help them take their medications as prescribed, as part of routine
54
pharmacy practice.11 Indeed, monitoring and improving patients’ adherence to medications
55
should be a routine component of patient care in community pharmacy. Patients must regularly
56
visit the pharmacy to collect their medications offering pharmacists a unique opportunity to
57
assess, identify and resolve problems concerning the chronic use of medications.11,12 At each
58
visit, they also have the opportunity to reinforce information already provided by other
59
healthcare professionals, provide additional information and to continue monitoring adherence.13
60
As a first step towards improving medication adherence, there needs to be broader recognition of
61
the problem, and once identified, several simple strategies may be used to support and improve
62
medication adherence.3 Pharmacists can track medication adherence by pill counts, review of
63
dispensing records and by asking patients directly how they are managing their medications. In
64
terms of supporting adherence, pharmacists can provide advice on the use of tools, such as pill
65
boxes and medication calendars to promote adherence, supply unit dose packaging or dose
66
administration aids, and/or offer to measure blood pressure and cholesterol levels as a way of
67
encouraging self-management.1 Thus supporting patients’ adherence to medication should be a
68
key professional responsibility of community pharmacists.
69
To date, little is known about how and to what extent Australian pharmacists engage with their
70
clients to address issues related to medication adherence and the factors shaping their practice.
71
Therefore, the objectives of this study were to:
72
1) investigate the frequency and types of strategies used by Australian community pharmacists to
73
identify, address and improve patients’ adherence to medication;
AC C
EP
TE D
M AN U
SC
RI PT
52
4
ACCEPTED MANUSCRIPT
2) explore pharmacists’ attitudes and barriers to the provision of adherence support activities;
75
and
76
3) develop a structural model of the factors influencing pharmacists’ provision of medication
77
adherence support.
RI PT
74
78
AC C
EP
TE D
M AN U
SC
79
5
ACCEPTED MANUSCRIPT
Methods
81
A cross-sectional mail survey was used to address the study aims. This study was approved by
82
the institution’s Human Research Ethics Committee.
83
Questionnaire
84
The survey comprised a self-completion, structured questionnaire consisting of eight sections.
85
The five relevant sections were: strategies used to identify non-adherent patients, strategies used
86
to support patients’ adherence to medications, pharmacists’ attitudes towards provision of
87
adherence
88
pharmacy/pharmacists’ demographic characteristics. The questionnaire took approximately 10 to
89
15 minutes to complete.
90
The questionnaire, including the attitudinal scales, was informed by the medication adherence
91
literature,2,14-21 and was pretested for content validity, face validity, and clarity of the questions
92
and instructions prior to administration.22
93
Study sample and sample size
94
A random sample of Australian community pharmacies was selected (through a random number
95
generator process) from the lists of pharmacies in obtained from a range of state and territory
96
based sources. The list for the state of New South Wales (NSW) was obtained from the NSW
97
Pharmacy Council; Victoria obtained from Victorian Pharmacy Authority; Western Australia
98
(WA) obtained from Pharmacy Registration Board of WA; South Australia (SA) obtained from
99
C.M Coffy & CO. PTY.LTD (Publishers of Industry Related Directories); and Australian Capital
100
Territory, Tasmania and Queensland obtained from the Yellow Pages (Telephone and Address
barriers
to
provision
of
adherence
M AN U
perceived
support,14-16
and
AC C
EP
TE D
support,
SC
RI PT
80
6
ACCEPTED MANUSCRIPT
Directory). A list of pharmacies was used as we did not have access to a list of pharmacists
102
employed in community pharmacies. In Australia, community pharmacies are privately owned
103
by registered pharmacists. They are independently run and may be part of banner groups. There
104
are no multinational chains of community pharmacies.
105
The sample size was calculated using the standard error of proportions equation,23,24 and based
106
on a previous study measuring provision of adherence support strategies by community
107
pharmacists.22 An average provision rate of 40% was assumed, 22 and at a 4% precision, a sample
108
size of 576 community pharmacies was required.23 The pilot study achieved a 27% response
109
rate.22 Therefore, based on this response rate, a national random sample of 2020 community
110
pharmacies, stratified by state, was invited to participate to obtain the sample size needed for our
111
data analysis. 23
112
Study process
113
In 2012, each pharmacy was mailed the questionnaire with an explanatory cover letter, prepaid
114
return envelope, and completion survey card with envelope for an incentives draw (to increase
115
response rate). Completion and return of the questionnaire indicated consent and willingness to
116
participate in the study. Since a code number was used for tracking non-respondents, study
117
participants were assured in the cover letter that the information they provided would remain
118
confidential, and that only group data would be presented.
119
To increase the response rate, reminder letters were also sent to non-respondents at three time
120
points, at 2, 4 and 8 weeks after initial mail-out. After three reminder letters, a follow-up
121
telephone survey of a sample of non-respondents was conducted. A list of non-respondents was
122
prepared after removing responding pharmacies and those that had returned questionnaires
AC C
EP
TE D
M AN U
SC
RI PT
101
7
ACCEPTED MANUSCRIPT
unanswered. A random sample of non-respondents was selected from this list, and contacted by
124
telephone. Telephone interviews were conducted until a total of 10% of the non-respondents25
125
had responded (n=62). The content of the telephone survey was based on, and similar to the mail
126
survey, but focused on specific questions that addressed the aim of the study, and where it was
127
postulated that differences between respondent and non-respondent behaviour with regards to
128
adherence support services may exist. These questions investigated strategies used to identify
129
non-adherent patients; strategies used to support patients’ adherence to medication; and rate of
130
provision of adherence support, by the pharmacist. Data on pharmacy and pharmacist
131
characteristics, including age, gender, and years of experience, position and number of
132
prescriptions dispensed /week were also collected from the non-respondents.
133
Data Analysis
134
Completed questionnaires were coded, reviewed for accuracy, entered into a database in the
135
Statistical Package for the Social Sciences (Version 19.0; IBM Corporation, Somers, NY). Data
136
were analysed using descriptive statistics. Univariate analysis was used to test for any differences
137
in the community pharmacist and pharmacy demographics between respondents and non-
138
respondents.
139 140
1. Principal Components Analysis of attitudes and barriers to provision of adherence support
141
Pharmacists` attitudes towards provision of adherence support was measured using 7 attitudinal
142
items on a 5-point Likert scale (from 1= Strongly Agree to 5= Strongly Disagree). Perceived
143
barriers to provision of adherence support was measured using 13 attitudinal items, also on a 5-
144
point Likert scale. Responses to pharmacists’ attitudes towards, and barriers to, provision of
145
adherence support were analysed separately using principal components analysis (PCA) with an
AC C
EP
TE D
M AN U
SC
RI PT
123
8
ACCEPTED MANUSCRIPT
oblique (Varimax) rotation with Kaiser Normalization. An eigenvalue greater than 1 rule, visual
147
inspection of the scree plot, and the number of items with high loadings on the factor (factor
148
loadings >0.3) were used to determine the final factor solution.
149
2. Internal consistency
150
Internal consistency of the components of attitude and barriers scales established by PCA was
151
assessed by Cronbach’s alpha. A reliability coefficient of ≥0.7 was considered acceptable.26
SC
RI PT
146
152
3. Structural equation modelling
154
Structural equation modelling (SEM) was conducted using maximum likelihood in AMOS 18.0
155
to test the hypothesized structural model (Figure 1).
156
[INSERT FIGURE 1 HERE]
158
TE D
157
M AN U
153
Dependent Variables:
160
The dependent (endogenous) variables of this model were:
162 163 164
1. Identifying non-adherence (Strategy 1): Defined as the rate of use of strategies to identify
AC C
161
EP
159
non-adherent patients. This rate was computed as a summative score based on pharmacists` responses to three items (in the questionnaire) on a 5-point Likert type scale (1= Always, 2= 75% of prescriptions dispensed, 3= 50% of prescriptions dispensed,
165
4=25% of prescriptions dispensed, 5=Never). The possible score range was 3-15. These
166
strategies were (reproduced from the questionnaire):
167
“Q1. Assessing how frequently patients collect repeat prescriptions using computerized records. 9
ACCEPTED MANUSCRIPT
168
Q2. Checking with patients about their response to medications.
169
Q3. Asking about barriers to taking medications without being confrontational.” 2. Supporting adherence (Strategy 2): Defined as the rate of provision of strategies to
171
support medication adherence in the 7 days prior to questionnaire completion.
172
Participants indicated the number of strategies they had provided in the week prior to
173
completing the questionnaire.
SC
RI PT
170
Independent variables:
175
Based on the information from the literature on improving patients’ adherence to medications,14-
176
16
177
(defined as Strategy 1 and 2 from above) were: attitudes and barriers, modelled as latent
178
variables represented by the items identified in principal components analysis; number of
179
prescriptions dispensed weekly; number of full time equivalent staff; and total scores of
180
enhanced services provided in pharmacy.
M AN U
174
TE D
the independent variables hypothesised to influence the overall provision of adherence support
1. Pharmacists’ attitudes towards provision of adherence support: this was measured by
182
listing 7 items and asking respondents to indicate their agreement on a 5-point Likert
183
scale (1= Strongly Agree, 2= Agree, 3= Neither Agree nor Disagree, 4= Disagree, 5=
185 186
AC C
184
EP
181
Strongly Disagree). As a result of PCA, the constructs of the attitude scale consisted of two latent (unobserved) variables. These were 1) Pharmacists’ role orientation to medication adherence monitoring (Pharmacists’ role), and 2) Stakeholders` role. The
187
latent construct of Pharmacists’ role composed of four observed variables, while the
188
second latent variable labelled Stakeholder`s role, comprised two observed variables.
10
ACCEPTED MANUSCRIPT
2. Perceived barriers to provision of adherence support: this was measured using 13 items
190
on a 5-point Likert scale (1= Strongly Agree, 2= Agree, 3= Neither Agree nor Disagree,
191
4= Disagree, 5= Strongly Disagree). Three components were extracted in the final factor
192
solution. The constructs of the barriers scale composed of three latent (unobserved)
193
variables. These were: 1) Stakeholder/Skills which was reflected by six observed
194
variables; 2) Patient’s literacy which comprised two observed variables; and 3) Logistics,
195
which was composed of five observed variables.
SC
RI PT
189
3. Others: number of prescriptions dispensed weekly, number of full time equivalent staff,
197
and total scores of enhanced services provided in pharmacy. All of these were considered
198
as observed variables.
M AN U
196
The fit of model was assessed using four indices:27,28
200
1). Relative Chi-square (χ2/df): Used to assess the overall fit and parsimony of the model.
201
Adequate model fit is obtained when χ2/df value is 1.0 to 2.0.
202
2). Adjusted Goodness-of-Fit Index (AGFI): Absolute fit, based on the predicted and observed
203
covariances. Good model fit is obtained when AGFI > 0.95.
204
3). Comparative Fit Index (CFI): Used to assess incremental fit. Values close to 0 indicate poor
205
fit where adequate model fit is obtained when CFI >0.90.
206
4). Root Mean-Square Error of Approximation (RMSEA) with 90% confidence level: this index
207
is a measure of the discrepancy per degree of freedom which indicated absolute fit of the model.
208
Good model fit is obtained when RMSEA
S1551-7411(15)00031-5
DOI:
10.1016/j.sapharm.2015.01.004
Reference:
RSAP 591
To appear in:
Research in Social & Administrative Therapy
Received Date: 10 December 2014 Revised Date:
23 January 2015
Accepted Date: 23 January 2015
Please cite this article as: Mansoor SM, Krass I, Costa D, Aslani P, Factors influencing the provision of adherence support by community pharmacists: A structural equation modelling approach, Research in Social & Administrative Therapy (2015), doi: 10.1016/j.sapharm.2015.01.004. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
1
Factors influencing the provision of adherence support by community
2
pharmacists: A structural equation modelling approach
4
Author details
5
Ms Sarab M Mansoor
6
BPharm, Dip Clinical Pharm, MSc, PhD candidate
7
Address:
RI PT
3
SC
Faculty of Pharmacy, Pharmacy and Bank Building, A15 The University of Sydney
9
Sydney, NSW 2006, Australia
M AN U
8
10
E-mail:
[email protected]; [email protected]
11
Telephone:
(+61 2) 9351 6066
12
Professor Ines Krass
14
BPharm, Dip Hosp Pharm, Grad Dip Educ Studies (Health Ed), PhD
15
Address:
TE D
13
Faculty of Pharmacy, Pharmacy and Bank Building, A15 The University of Sydney
17
Sydney, NSW 2006, Australia
AC C
EP
16
18
E-mail:
[email protected]
19
Telephone:
(+61 2) 9351 3507
20
Fax:
(+61 2) 9351 4391
21 22
Dr Daniel Costa
23
BSc(Hons), PhD
ACCEPTED MANUSCRIPT
School of Physiology, Lifehouse Building (C39Z)
25
The University of Sydney
26
Sydney, NSW 2006, Australia
27
E-mail:
[email protected]
28
Telephone:
(+61 2) 9351 6304
29
Fax:
(+61 2) 9036 5223
30
Associate Professor Parisa Aslani
32
BPharm(Hons), MSc, PhD, Grad Cert Ed Stud (Higher Ed)
33
Address:
M AN U
31
Faculty of Pharmacy, Pharmacy and Bank Building, A15
34
The University of Sydney
35
Sydney, NSW 2006, Australia E-mail:
[email protected]
37
Telephone:
(+61 2) 9036 6541
38
Fax:
(+61 2) 9351 4391
EP
40
TE D
36
39
RI PT
Address:
SC
24
Conflict of interest disclosure: The author(s) declare no potential conflicts of interest with
42
respect to the research, authorship, and/or publication of this article.
43
AC C
41
ACCEPTED MANUSCRIPT
Factors influencing the provision of adherence support by community pharmacists: A structural equation modelling approach
4 5 6
Abstract
7
optimum patient outcomes. Community pharmacists are well placed to deliver interventions to
8
support adherence.
9
Aims: To investigate community pharmacists’ activities in supporting patient adherence; and
RI PT
1 2 3
SC
Background: Non-adherence to medication represents an important barrier to achieving
identify factors influencing pharmacists’ provision of adherence support.
11
Method: A random sample of 2,020 Australian community pharmacies was mailed a
12
questionnaire investigating their provision of adherence support. The self-completed, structured
13
questionnaire consisted of eight sections, five of which were relevant to this study: strategies
14
used to identify non-adherent patients, strategies used to support patients’ adherence to
15
medications, pharmacists’ attitudes towards provision of adherence support, perceived barriers to
16
provision of adherence support, and demographics. Structural equation modelling (SEM) was
17
used to determine potential influencing factors.
18
Results: A response rate of 31% was achieved (n=627). Pharmacists reported using strategies to
19
identify non-adherent patients for less than half (45%) of the prescriptions dispensed. A mean of
20
8.4±14.9 (Mean± SD) strategies was used by respondents in the 7 days prior to survey
21
completion. Dose administration aids was the most commonly used strategy (provided by 96.5%
22
of respondents). Time pressure for patients (68%) was perceived by pharmacists as the main
23
barrier to adherence support. SEM identified “stakeholders/skills” and “number of full time
24
equivalent staff” as influencing provision of adherence support strategies.
AC C
EP
TE D
M AN U
10
1
ACCEPTED MANUSCRIPT
25
Conclusion: Provision of adherence support by pharmacists was episodic and infrequent,
26
impeded by a number of barriers. By addressing barriers, it is possible to enable pharmacists to
27
become more proactive and effective in supporting patient adherence.
29
RI PT
28
Keywords: Adherence to medications, pharmacists, barriers, attitudes, Australia.
AC C
EP
TE D
M AN U
SC
30
2
ACCEPTED MANUSCRIPT
Introduction
32
Medication non-adherence has been widely recognized as a continuing problem among patients
33
with chronic diseases, including hypertension, diabetes and lipid disorders.1-4 Non-adherence has
34
been associated with higher mortality and morbidity as well as increased health care costs.5,6 In
35
developed countries, among patients with chronic diseases on long-term therapy, the average
36
medication adherence rate is approximately 50%, with some variation between conditions.1,2
37
A large body of research has focused on the common barriers to medication adherence.1,7,8 These
38
include: patient-related factors such as knowledge, beliefs, attitudes and/or expectations; social
39
and economic factors such as lack of social support and/or poor socioeconomic status; health
40
care team and system-related factors such as poorly developed or fragmented health care
41
systems and/or lack of knowledge and training of health care providers; condition-related factors
42
such as severity of symptoms, severity of disease and/or rate of progression; and therapy-related
43
factors such as complexity of the medical regimen, duration of treatment and/or side effects.
44
Recognition of these barriers has influenced the development of a variety of targeted strategies
45
intended to enhance adherence and persistence to therapy. For more than three decades,
46
researchers and healthcare professionals have sought to promote and improve patients’
47
adherence to medication by interventions which primarily focus on fostering behavioural
48
change.9 Interventions aimed at promoting adherence to therapy have resulted in improvements
49
in adherence and in some cases, improvements in clinical outcomes.2,10 The goal is to encourage
50
and empower the individual patient to learn, adopt, and sustain a regular pattern of medication-
51
taking behaviour.1
AC C
EP
TE D
M AN U
SC
RI PT
31
3
ACCEPTED MANUSCRIPT
As healthcare professionals, community pharmacists are well placed to identify non-adherence
53
and offer support to patients to help them take their medications as prescribed, as part of routine
54
pharmacy practice.11 Indeed, monitoring and improving patients’ adherence to medications
55
should be a routine component of patient care in community pharmacy. Patients must regularly
56
visit the pharmacy to collect their medications offering pharmacists a unique opportunity to
57
assess, identify and resolve problems concerning the chronic use of medications.11,12 At each
58
visit, they also have the opportunity to reinforce information already provided by other
59
healthcare professionals, provide additional information and to continue monitoring adherence.13
60
As a first step towards improving medication adherence, there needs to be broader recognition of
61
the problem, and once identified, several simple strategies may be used to support and improve
62
medication adherence.3 Pharmacists can track medication adherence by pill counts, review of
63
dispensing records and by asking patients directly how they are managing their medications. In
64
terms of supporting adherence, pharmacists can provide advice on the use of tools, such as pill
65
boxes and medication calendars to promote adherence, supply unit dose packaging or dose
66
administration aids, and/or offer to measure blood pressure and cholesterol levels as a way of
67
encouraging self-management.1 Thus supporting patients’ adherence to medication should be a
68
key professional responsibility of community pharmacists.
69
To date, little is known about how and to what extent Australian pharmacists engage with their
70
clients to address issues related to medication adherence and the factors shaping their practice.
71
Therefore, the objectives of this study were to:
72
1) investigate the frequency and types of strategies used by Australian community pharmacists to
73
identify, address and improve patients’ adherence to medication;
AC C
EP
TE D
M AN U
SC
RI PT
52
4
ACCEPTED MANUSCRIPT
2) explore pharmacists’ attitudes and barriers to the provision of adherence support activities;
75
and
76
3) develop a structural model of the factors influencing pharmacists’ provision of medication
77
adherence support.
RI PT
74
78
AC C
EP
TE D
M AN U
SC
79
5
ACCEPTED MANUSCRIPT
Methods
81
A cross-sectional mail survey was used to address the study aims. This study was approved by
82
the institution’s Human Research Ethics Committee.
83
Questionnaire
84
The survey comprised a self-completion, structured questionnaire consisting of eight sections.
85
The five relevant sections were: strategies used to identify non-adherent patients, strategies used
86
to support patients’ adherence to medications, pharmacists’ attitudes towards provision of
87
adherence
88
pharmacy/pharmacists’ demographic characteristics. The questionnaire took approximately 10 to
89
15 minutes to complete.
90
The questionnaire, including the attitudinal scales, was informed by the medication adherence
91
literature,2,14-21 and was pretested for content validity, face validity, and clarity of the questions
92
and instructions prior to administration.22
93
Study sample and sample size
94
A random sample of Australian community pharmacies was selected (through a random number
95
generator process) from the lists of pharmacies in obtained from a range of state and territory
96
based sources. The list for the state of New South Wales (NSW) was obtained from the NSW
97
Pharmacy Council; Victoria obtained from Victorian Pharmacy Authority; Western Australia
98
(WA) obtained from Pharmacy Registration Board of WA; South Australia (SA) obtained from
99
C.M Coffy & CO. PTY.LTD (Publishers of Industry Related Directories); and Australian Capital
100
Territory, Tasmania and Queensland obtained from the Yellow Pages (Telephone and Address
barriers
to
provision
of
adherence
M AN U
perceived
support,14-16
and
AC C
EP
TE D
support,
SC
RI PT
80
6
ACCEPTED MANUSCRIPT
Directory). A list of pharmacies was used as we did not have access to a list of pharmacists
102
employed in community pharmacies. In Australia, community pharmacies are privately owned
103
by registered pharmacists. They are independently run and may be part of banner groups. There
104
are no multinational chains of community pharmacies.
105
The sample size was calculated using the standard error of proportions equation,23,24 and based
106
on a previous study measuring provision of adherence support strategies by community
107
pharmacists.22 An average provision rate of 40% was assumed, 22 and at a 4% precision, a sample
108
size of 576 community pharmacies was required.23 The pilot study achieved a 27% response
109
rate.22 Therefore, based on this response rate, a national random sample of 2020 community
110
pharmacies, stratified by state, was invited to participate to obtain the sample size needed for our
111
data analysis. 23
112
Study process
113
In 2012, each pharmacy was mailed the questionnaire with an explanatory cover letter, prepaid
114
return envelope, and completion survey card with envelope for an incentives draw (to increase
115
response rate). Completion and return of the questionnaire indicated consent and willingness to
116
participate in the study. Since a code number was used for tracking non-respondents, study
117
participants were assured in the cover letter that the information they provided would remain
118
confidential, and that only group data would be presented.
119
To increase the response rate, reminder letters were also sent to non-respondents at three time
120
points, at 2, 4 and 8 weeks after initial mail-out. After three reminder letters, a follow-up
121
telephone survey of a sample of non-respondents was conducted. A list of non-respondents was
122
prepared after removing responding pharmacies and those that had returned questionnaires
AC C
EP
TE D
M AN U
SC
RI PT
101
7
ACCEPTED MANUSCRIPT
unanswered. A random sample of non-respondents was selected from this list, and contacted by
124
telephone. Telephone interviews were conducted until a total of 10% of the non-respondents25
125
had responded (n=62). The content of the telephone survey was based on, and similar to the mail
126
survey, but focused on specific questions that addressed the aim of the study, and where it was
127
postulated that differences between respondent and non-respondent behaviour with regards to
128
adherence support services may exist. These questions investigated strategies used to identify
129
non-adherent patients; strategies used to support patients’ adherence to medication; and rate of
130
provision of adherence support, by the pharmacist. Data on pharmacy and pharmacist
131
characteristics, including age, gender, and years of experience, position and number of
132
prescriptions dispensed /week were also collected from the non-respondents.
133
Data Analysis
134
Completed questionnaires were coded, reviewed for accuracy, entered into a database in the
135
Statistical Package for the Social Sciences (Version 19.0; IBM Corporation, Somers, NY). Data
136
were analysed using descriptive statistics. Univariate analysis was used to test for any differences
137
in the community pharmacist and pharmacy demographics between respondents and non-
138
respondents.
139 140
1. Principal Components Analysis of attitudes and barriers to provision of adherence support
141
Pharmacists` attitudes towards provision of adherence support was measured using 7 attitudinal
142
items on a 5-point Likert scale (from 1= Strongly Agree to 5= Strongly Disagree). Perceived
143
barriers to provision of adherence support was measured using 13 attitudinal items, also on a 5-
144
point Likert scale. Responses to pharmacists’ attitudes towards, and barriers to, provision of
145
adherence support were analysed separately using principal components analysis (PCA) with an
AC C
EP
TE D
M AN U
SC
RI PT
123
8
ACCEPTED MANUSCRIPT
oblique (Varimax) rotation with Kaiser Normalization. An eigenvalue greater than 1 rule, visual
147
inspection of the scree plot, and the number of items with high loadings on the factor (factor
148
loadings >0.3) were used to determine the final factor solution.
149
2. Internal consistency
150
Internal consistency of the components of attitude and barriers scales established by PCA was
151
assessed by Cronbach’s alpha. A reliability coefficient of ≥0.7 was considered acceptable.26
SC
RI PT
146
152
3. Structural equation modelling
154
Structural equation modelling (SEM) was conducted using maximum likelihood in AMOS 18.0
155
to test the hypothesized structural model (Figure 1).
156
[INSERT FIGURE 1 HERE]
158
TE D
157
M AN U
153
Dependent Variables:
160
The dependent (endogenous) variables of this model were:
162 163 164
1. Identifying non-adherence (Strategy 1): Defined as the rate of use of strategies to identify
AC C
161
EP
159
non-adherent patients. This rate was computed as a summative score based on pharmacists` responses to three items (in the questionnaire) on a 5-point Likert type scale (1= Always, 2= 75% of prescriptions dispensed, 3= 50% of prescriptions dispensed,
165
4=25% of prescriptions dispensed, 5=Never). The possible score range was 3-15. These
166
strategies were (reproduced from the questionnaire):
167
“Q1. Assessing how frequently patients collect repeat prescriptions using computerized records. 9
ACCEPTED MANUSCRIPT
168
Q2. Checking with patients about their response to medications.
169
Q3. Asking about barriers to taking medications without being confrontational.” 2. Supporting adherence (Strategy 2): Defined as the rate of provision of strategies to
171
support medication adherence in the 7 days prior to questionnaire completion.
172
Participants indicated the number of strategies they had provided in the week prior to
173
completing the questionnaire.
SC
RI PT
170
Independent variables:
175
Based on the information from the literature on improving patients’ adherence to medications,14-
176
16
177
(defined as Strategy 1 and 2 from above) were: attitudes and barriers, modelled as latent
178
variables represented by the items identified in principal components analysis; number of
179
prescriptions dispensed weekly; number of full time equivalent staff; and total scores of
180
enhanced services provided in pharmacy.
M AN U
174
TE D
the independent variables hypothesised to influence the overall provision of adherence support
1. Pharmacists’ attitudes towards provision of adherence support: this was measured by
182
listing 7 items and asking respondents to indicate their agreement on a 5-point Likert
183
scale (1= Strongly Agree, 2= Agree, 3= Neither Agree nor Disagree, 4= Disagree, 5=
185 186
AC C
184
EP
181
Strongly Disagree). As a result of PCA, the constructs of the attitude scale consisted of two latent (unobserved) variables. These were 1) Pharmacists’ role orientation to medication adherence monitoring (Pharmacists’ role), and 2) Stakeholders` role. The
187
latent construct of Pharmacists’ role composed of four observed variables, while the
188
second latent variable labelled Stakeholder`s role, comprised two observed variables.
10
ACCEPTED MANUSCRIPT
2. Perceived barriers to provision of adherence support: this was measured using 13 items
190
on a 5-point Likert scale (1= Strongly Agree, 2= Agree, 3= Neither Agree nor Disagree,
191
4= Disagree, 5= Strongly Disagree). Three components were extracted in the final factor
192
solution. The constructs of the barriers scale composed of three latent (unobserved)
193
variables. These were: 1) Stakeholder/Skills which was reflected by six observed
194
variables; 2) Patient’s literacy which comprised two observed variables; and 3) Logistics,
195
which was composed of five observed variables.
SC
RI PT
189
3. Others: number of prescriptions dispensed weekly, number of full time equivalent staff,
197
and total scores of enhanced services provided in pharmacy. All of these were considered
198
as observed variables.
M AN U
196
The fit of model was assessed using four indices:27,28
200
1). Relative Chi-square (χ2/df): Used to assess the overall fit and parsimony of the model.
201
Adequate model fit is obtained when χ2/df value is 1.0 to 2.0.
202
2). Adjusted Goodness-of-Fit Index (AGFI): Absolute fit, based on the predicted and observed
203
covariances. Good model fit is obtained when AGFI > 0.95.
204
3). Comparative Fit Index (CFI): Used to assess incremental fit. Values close to 0 indicate poor
205
fit where adequate model fit is obtained when CFI >0.90.
206
4). Root Mean-Square Error of Approximation (RMSEA) with 90% confidence level: this index
207
is a measure of the discrepancy per degree of freedom which indicated absolute fit of the model.
208
Good model fit is obtained when RMSEA
