VA LU E I N H EA LTH ] (2017) ]]] – ]]]

Available online at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/jval

Development and Validation of the AFImpact: An Atrial

Fibrillation–Specific Measure of Patient-Reported Health-Related
Quality of Life
Karin S. Coyne, MPH, PhD1, Nils Edvardsson, MD, PhD2, Anna Rydén, PhD3,*
1
Outcomes Research, Evidera, Bethesda, MD, USA; 2Sahlgrenska Academy at the Sahlgrenska University Hospital,
Göteborg, Sweden; 3AstraZeneca Gothenburg, Mölndal, Sweden

ABSTRACT

Background: Improvement in health-related quality of life is a key
therapeutic goal of disease management in atrial fibrillation (AF).
Objectives: To describe the development of the AFImpact, an AF-
specific health-related quality-of-life patient-reported outcome meas-
ure. Methods: Development and validation of the AFImpact com-
prised a qualitative stage, consisting of a literature review and
concept elicitation interviews (91 patients with AF), item generation,
and cognitive debriefing (30 patients with AF), and a quantitative
stage, consisting of evaluation of the instrument’s psychometric
properties (313 patients with AF). Preliminary responsiveness to
change was assessed in 118 patients undergoing cardioversion.
Results: On the basis of the literature review and concept elicitation
interviews, 75 items were generated. Factor analyses guided a reduc -
tion to 18 items. Three domains were confirmed: vitality (7 items),
emotional distress (8 items), and sleep (3 items). The 18-item AFIm-
pact demonstrated high item convergent and discriminant validity.
Cronbach α coefficients showed high internal consistency reliability.
Test–retest reliability of individual items in stable patients (n¼ 33)
was satisfactory, with intraclass correlation coefficients ranging from
0.61 to 0.86. All three AFImpact domain scores differentiated patients
who reported different levels of overall health, thereby supporting
known-groups validity. Scores for each item improved after cardio-
version, with effect sizes ranging from —0.19 to —0.65. Conclusions:
Psychometric evaluations support the reliability and validity of the
AFImpact as a patient-reported outcome instrument to measure the
impact of AF, with preliminary results in patients undergoing cardi -
oversion supporting responsiveness to change.
Keywords: atrial fibrillation, health-related quality of life, patient-
reported outcome measure, validation.
Copyright & 2017, International Society for Pharmacoeconomics and
Outcomes Research (ISPOR). Published by Elsevier Inc.

more than 1 week, and permanent if treatment to restore normal

Introduction
Atrial fibrillation (AF) is the most common sustained heart
arrhythmia, with a reported incidence of between 5 and 28 per
1000 person-years in individuals aged 65 years and older in
Europe and the United States [1,2]. Although patients with AF
can have relatively disease-specific symptoms such as heart
palpitations and chest pain, most AF-related symptoms, includ-
ing fatigue and anxiety, are not disease-specific [3,4]. A large
minority of patients with AF either report no symptoms or report
symptoms without realizing that these are due to heart arrhyth-
mia [3,4]. Patients with AF, including those who report no
symptoms, have an impaired health-related quality of life
(HRQOL) compared with the general population, including feeling
worried about their condition, treatment side effects, and wor-
sening health [5,6]. AF is classified as paroxysmal if it stops
spontaneously within 1 week, persistent if it is continuous for
sinus rhythm has either failed or been forgone. Progression from
paroxysmal to sustained AF occurs over time in a substantial
proportion of patients. Although symptom patterns may differ to
some extent, in part because of adaptation to long-lasting AF,
impairments in HRQOL are generally similar across the three
types of AF [7]. Treatment of AF is associated with an improve-
ment in symptoms and HRQOL [8].
Patients with AF may attribute their symptoms and impaired
HRQOL to increasing age and lifestyle-related causes such as
stress or not getting enough sleep, and they may accept the
symptoms unquestioningly, possibly even adapting their lifestyle
(e.g., by reducing levels of their daily activities). Recognizing
symptoms of AF and their impact on HRQOL by means of an AF-
specific patient-reported outcome (PRO) measure may help
ensure appropriate disease management. The AF-specific
symptom questionnaire (AFSymp) was developed and

Conflicts of interest: K. S. Coyne is employed by Evidera, which provides consulting and other research services to pharmaceu tical,
device, government, and nongovernment organizations, and has received consultancy fees from AstraZeneca. N. Edvardsson was a
consultant for AstraZeneca at the time the AFImpact was developed. A. Rydén is an employee of AstraZeneca.
* Address correspondence to: Anna Rydén, AstraZeneca Gothenburg, Pepparedsleden 1, Mölndal 43150, Sweden. E-
mail: anna.ryden@astrazeneca.com.
1098-3015$36.00 – see front matter Copyright & 2017, International Society for Pharmacoeconomics and Outcomes Research (ISPOR).
Published by Elsevier Inc.
http://dx.doi.org/10.1016/j.jval.2017.06.005
2 VA LU E I N H EA LT H ] (2 0 1 7 ) ]]] – ]] ]
psychometrically evaluated on the basis of guidance on PRO
measures published by the US Food and Drug Administration
[9,10]. During the development of the AFSymp, patients were also
asked about the impact of their AF on HRQOL, including on
energy levels and activities of daily living [10].
Improvement in HRQOL is a key therapeutic goal in AF manage-
ment and has been identified as a relevant clinical end point in
trials of AF therapies [5]. Regulatory guidelines recommend that
symptoms and impacts of diseases or conditions be measured
using direct reporting from patients themselves and that HRQOL
should be clearly differentiated from the core symptoms of a
disease [9,11]. There is a concern regarding the use of generic
HRQOL measures, because they are deemed not sufficiently sensi-
tive or specific to detect changes in AF-specific HRQOL [6]. Several
AF-specific HRQOL instruments are available [12–17]. Nevertheless,
their use is limited by uncertain validity [14–17] and limited
generalizability [12,13]. For the Atrial Fibrillation Effect on Quality
of Life questionnaire (AFEQT, also known as AF-QoL-18), rigorous
validation analyses were conducted on the preliminary long
version of the PRO; similar analyses still remain to be conducted
on the final condensed version that is being used today [15,17]. The
HeartQoL was psychometrically evaluated in patients with AF; it
was, however, developed in a broad group of patients with
ischemic heart disease, and its validity specifically in patients with
AF remains to be established [16]. The European Heart Rhythm
Association score was recommended in European Society of
Cardiology guidelines for assessing symptoms related to AF on
the basis of their impact on daily activity, but it has not been
validated specifically as an HRQOL instrument [18]. These scores
correlate negatively with HRQOL, although they do not discrim-
inate sufficiently between patients with no symptoms and those
with mild symptoms (a modified version has been proposed to
improve discrimination) and they have not been assessed by AF
subtype [19].
The aim of this article was to describe the development
of the AFImpact, an AF-specific PRO measure that was
developed and evaluated in accordance with current regulatory
guidelines.
Methods
The AFImpact was developed and validated in two stages: stage I
was qualitative and consisted of concept elicitation, item gen-
eration, and cognitive interviews; stage II was quantitative and
involved the evaluation of the instrument’s psychometric
properties.
Stage I: Development of the AFImpact (Qualitative)
Concept elicitation was conducted jointly for the AFImpact and
the AFSymp [10]. After conducting a literature review, concepts
about AF symptoms and impacts were elicited during qualitative
interviews among 91 patients from the United States, Europe, and
Japan (n ¼ 30, 46, and 15, respectively) who had a confirmed
diagnosis of paroxysmal, persistent, or permanent AF (n ¼ 30, 30,
and 31, respectively). Interviews were conducted by trained
interviewers following a standardized interview guide. During
the interviews, patients were asked to describe the impact of AF
and its treatment on their daily routine, including physical
functioning, emotional well-being, and work. Interviews were
transcribed verbatim. Transcripts were analyzed qualitatively to
identify participants’ expressions of concepts.
After the concept elicitation interviews, draft items for the
AFImpact were generated in US English on the basis of an
analysis of the qualitative interview information about the
impact of AF. The draft items were then evaluated for content
validity and comprehension through cognitive interviews with 30
patients with AF (10 with paroxysmal, 10 with persistent, and 10
with permanent AF) in the United States. The draft questionnaire
was completed in an electronic format by 20 patients and in a
pen–paper format by 10 patients. A seven-point Likert scale was
used to provide a wide-enough range of response options to
enable detection of small but important differences in scores [20].
Stage II: Validation of the AFImpact (Quantitative)
Item reduction, identification of domains, instrument reliability,
and construct validity
A prospective, observational psychometric evaluation study was
conducted at 32 sites in the United States to evaluate the
AFImpact, administered as an electronic version [10]. The study
recruited patients with paroxysmal, persistent, or permanent AF
with symptoms in the past 7 days and consisted of a baseline visit
(visit 1) and a follow-up visit (visit 2; 14 ± 3 days later). There were
no investigational interventions. The AFImpact was completed at
both visits. Patients also completed the 36-item short form health
survey (SF-36) [21], the University of Toronto Atrial Fibrillation
Severity Scale (AFSS) [22], and the Medical Outcomes Study (MOS)-
Sleep Scale [23] at visit 1 and the Overall Treatment Effect (OTE)
[20] at visit 2. Construct validity of the AFImpact was assessed by
examining the association of AFImpact scores with those of well-
established HRQOL questionnaires (i.e., the generic SF-36, the
disease-specific AFSS, and the MOS-Sleep Scale that measures
sleep problems), and positive correlations were hypothesized.
Preliminary responsiveness
Preliminary responsiveness to change was assessed in a pro-
spective, observational, multicenter study of US patients with AF
who were scheduled to undergo either direct-current or pharma-
cological cardioversion. Patients completed the AFImpact at
baseline (before cardioversion) and the AFImpact and the OTE
at follow-up (14 ± 2 days after cardioversion).
Statistical Analyses
Item reduction, identification of domains, instrument reliability,
and construct validity
Analyses were conducted using Statistical Analysis System ver-
sion 8 (SAS Institute, Cary, NC), Multitrait Analysis Program-
Revised version 1.0 (RAND Corp., Santa Monica, CA), and Rasch
Unidimensional Measurement Models software version 3.1
(RUMM Laboratory, Perth, WA). Item reduction and identification
of domains were informed by using frequency distributions of
response options, exploratory and confirmatory factor analyses
using varimax rotation and oblimin rotation, and Rasch analysis.
Items were deleted if more than 30% of respondents gave the
lowest possible score or the highest possible score.
Cronbach α was used to assess internal consistency
reliability. Test–retest reliability (intraclass correlation coeffi-
cient) was assessed in patients whose status was considered
stable (defined as patients who reported that their health was
“the same”: a score of “0” on the OTE at the second study visit).
Construct validity (convergent and discriminant validity) was
examined by using Pearson correlations between the AFImpact
domain scores and the SF-36, the MOS-Sleep Scale, and the AFSS
Wellbeing, Total AF Burden, and Total Symptom Severity domain
scores at visit 1. Known-groups validity was assessed by compar-
ing AFImpact scores between groups of patients who self-
reported different levels of overall health on item 1 of the SF- 36,
patients who reported being currently in AF and those who
reported not being currently in AF (as reported on item 3 of the
AFSS: “Are you in atrial fibrillation currently?”), and patients with
different levels of clinician-reported AF severity (physicians were
 VA LU E I N H EA LT H ] (2 0 1 7 ) ]]] – ] ]] 3

asked to grade symptoms at baseline as very mild, mild, moder-

ate, severe, or very severe). Analysis of variance was used for

Proportion of patients giving

highest possible score (%)‡
between-group comparisons.

* The following ancillary measures were used to assess convergent and discriminant validity: 36 -item short form health survey, University of Toronto Atrial Fibrillation Severity Scale, and
Preliminary responsiveness

Table 1 – Convergent validity, discriminant validity, and internal consistency of items and floor and ceiling analyses of domains of the AFImpact (psychometric
Effect size was calculated as the difference in mean scores (i.e.,

22.0
7.3
7.7
follow-up mean score – baseline mean score) divided by the SD of
the baseline score. Standardized response mean was calculated
as the difference in mean scores (i.e., follow-up mean score –
baseline mean score) divided by the SD of the score change.

Domain analyses
Results

Stage I: Development of the AFImpact (Qualitative)

Proportion of patients giving

lowest possible score (%)‡
The concept elicitation interviews included 60 men and 31
women, most of whom (73%) were aged 60 years or older and
all of whom were currently receiving medication for their AF.
Interviewees reported a considerable impact of AF on their daily

0.6
0.0

0.0
lives and HRQOL. The greatest impact was on patients’ daily
living and physical functioning, primarily in terms of reduced
motility, low vitality, and sleep problems. Emotional well-being
was also considerably affected, with patients feeling anxious and

Proportion of item-domain correlations that were higher with the item’s own domain than with the other AFImpact domains.
worried about their AF. Results were generally similar across
groups. Some subtle variations included that more patients with
paroxysmal AF spoke about depression than those with persis-
tent or permanent AF, and that patients from Europe spoke in
more detail about the impacts of their AF on daily living than did
patients from the United States and Japan. On the basis of these consistency,
interviews, 75 items were generated that included all the HRQOL Cronbach α
Internal

concepts that were mentioned by patients. Care was taken to

0.96
0.96

0.90
include all HRQOL concepts patients specifically noted were
impacted by AF. The term “because of my heart condition” was
included in each question instead of “because of my atrial
fibrillation,” because it was found that some patients were not
familiar with the term “atrial fibrillation.”
Of the 30 patients who took part in the cognitive interviews,
Discriminant validity*,

26 reported that the draft questionnaire was easy to understand

success rate (%)†

and answer, and 25 agreed that it was clear. A 1-week recall was
supported by half the patients (the other half suggested that this
Item analyses

was too short of a recall and wanted a longer period). The draft
100
100

100

questionnaire used a seven-point Likert scale, with response

options initially ranging from “1” representing “all of the time”
to “7” representing “none of the time.” This was subsequently
amended such that increases in numerical values indicated
increases in AF impact (i.e., 1 ¼ none of the time and 7 ¼ all of
the time).
Convergent validity *,
correlation range

Stage II: Validation of the AFImpact (Quantitative)

0.86–0.90
0.76–0.89

0.77–0.83

In total, 313 patients took part in the psychometric evaluation

Less than 20% was considered acceptable.

study. Patients’ demographic and clinical characteristics have

been reported previously [10]. Factor analyses using varimax
Medical Outcomes Study-Sleep Scale.

rotation and oblimin rotation yielded similar results (data not

shown). Item deletion was based on the varimax results. High
evaluation study, N ¼ 313).

ceiling effects, high interitem correlations (indicating redun-

dancy), and items loading on more than one factor (indicating
that the item assesses more than one concept) guided a reduction
No. of
items

in the number of items to 18. Rasch analysis confirmed that all 18

3
7
8

selected items were able to discriminate patients at all severity

levels and that patients were able to differentiate the response
options.
distress
Emotional
Domain

Results of the factor analysis for the final 18-item version of

Vitality

the AFImpact identified and confirmed the appropriateness of

Sleep

three domains: vitality (7 items), emotional distress (8 items), and

sleep (3 items). The 18-item AFImpact demonstrated both high
†

‡
4 VA LU E I N H EA LT H ] (2 0 1 7 ) ]]] – ]] ]

item convergent validity and high item discriminant validity
(Table 1). Mean item scores of the AFImpact ranged from 2.5 ±
1.4, for item 15 (“I felt depressed because of my heart condition”),
and 2.5 ± 1.5, for item 9 (“I woke up during the night because of
my heart condition”), to 3.2 ± 1.7, for item 17 (“I could not be
as physically active as I wanted to be because of my heart
condition”).
Reliability
Cronbach α coefficients demonstrated high internal consistency
reliability (Table 1). Domain analyses showed no floor effects for
any of the three domains and no ceiling effects for the vitality
and the emotional distress domains. There was, however, a
ceiling effect for the three-item sleep domain, with 22% of
patients scoring the highest possible score (Table 1).
Interscale correlations (r) were 0.71, 0.64, and 0.70 for vitality
versus emotional distress, emotional distress versus sleep, and
sleep versus vitality, respectively, indicating that the three
domains measure related concepts but are not so highly corre-
lated as to indicate redundancy. Construct validity findings in
patient subgroups with paroxysmal, persistent, or permanent AF
were similar to those in the overall study population (data not
shown).
In total, 152 of the patients who took part in the psychometric
evaluation study had a follow-up visit within 11 to 17 days after
their baseline visit, of whom 111 reported that their health was
about the same (OTE score —1, 0, or 1) and 33 reported that their
health was the same (OTE score 0). Test–retest reliability of the
AFImpact individual items in stable patients (n ¼ 33; defined by
OTE score 0) was satisfactory, with intraclass correlation coef-
ficients ranging from 0.61, for item 12 (“I felt worn out because of
my heart condition”), to 0.86, for item 15 (“I felt depressed
because of my heart condition”) (Table 2).
Domain convergent and discriminant validity
We hypothesized that AFImpact domains would correlate pos-
itively with SF-36 and AFSS domains, which measure similar
concepts, and the MOS-Sleep Scale. Correlations between the
AFImpact domains and the AFSS Wellbeing, Total AF Burden, and
Total Symptom Severity domains were statistically significant (r,
range —0.48 to 0.70; all P o 0.0001). Correlations between the
AFImpact domains and the SF-36 domains were statistically
significant (all r 4 0.40; all P o 0.0001), with domains measuring
similar concepts correlating more highly (convergent validity)
than those measuring dissimilar concepts (discriminant validity).
The AFImpact sleep domain correlated with the MOS-Sleep Scale
domains and composite scores (r, range —0.22 to 0.78; all P o
0.005), further supporting convergent validity.
Known-groups validity
All three AFImpact domain scores differentiated between
patients who reported different levels of overall health on item
1 of the SF-36 (all P o 0.0001), with a clear pattern of lower scores
for patients reporting better general health (Fig. 1). For all three
domains, scores were significantly higher for patients who
reported being currently in AF than for patients who reported
not being currently in AF (Fig. 2), providing further evidence of
the known-groups validity.
The AFImpact domain scores also differentiated between
different levels of clinician-reported AF severity (all P o 0.0001).
The severity of AF symptoms was reported by physicians as very
mild, mild, moderate, and severe for 12.8%, 47.3%, 36.1%, and
3.8% of patients, respectively. Mean scores for each of these four

Table 2 – Test–retest reliability in stable patients between visit 1 and visit 2 in the psychometric evaluation
study (n ¼ 33).
Item Item score, mean ± SD t value P value Spearman r‡ ICC

Visit 1 Visit 2* Difference†

1. Did not sleep well 3.1 ± 1.6 3.2 ± 1.9 0.0 0.12 0.9035 0.66§ 0.66
2. Worried might shorten my life 3.3 ± 1.9 3.1 ± 1.9 —0.2 —1.04 0.3039 0.77§ 0.81
3. Felt anxious 3.5 ± 1.8 3.1 ± 1.9 —0.4 —1.51 0.1421 0.65§ 0.71
4. Less energy 3.4 ± 1.5 3.4 ± 1.7 0.0 —0.14 0.8890 0.67§ 0.72
5. Annoyed 2.7 ± 1.9 2.7 ± 1.8 0.0 0.00 1.0000 0.77§ 0.85
6. Stressed 3.3 ± 1.8 3.1 ± 1.9 —0.2 —0.94 0.3531 0.71§ 0.75
7. Walk slowly 3.3 ± 1.8 3.1 ± 1.9 —0.2 —0.86 0.3947 0.75§ 0.79
8. Tired 3.2 ± 1.7 3.4 ± 1.6 0.2 1.14 0.2632 0.81§ 0.85
9. Woke up during the night 3.1 ± 1.7 3.0 ± 1.7 —0.1 —0.27 0.7902 0.69§ 0.72
10. Performed daily activities slowly 3.1 ± 1.7 3.1 ± 2.0 —0.1 —0.24 0.8126 0.63§ 0.69
11. Worried about my health 3.2 ± 2.0 2.8 ± 1.8 —0.4 —1.74 0.0908 0.74§ 0.76
12. Felt worn out 3.2 ± 1.6 3.2 ± 1.8 0.0 0.12 0.9089 0.56§ 0.61
13. Difficulty falling asleep 3.0 ± 1.7 3.4 ± 1.9 0.4 1.45 0.1566 0.64§ 0.62
14. Nervous 3.1 ± 1.9 2.8 ± 1.8 —0.2 —0.88 0.3860 0.65§ 0.72
15. Depressed 2.7 ± 1.6 2.4 ± 1.8 —0.3 —1.79 0.0831 0.86§ 0.86
16. Had to rest during the day 3.3 ± 1.7 3.3 ± 1.8 0.0 0.12 0.9089 0.55§ 0.64
17. Not as physically active 3.5 ± 1.9 3.4 ± 1.9 —0.1 —0.49 0.6245 0.73§ 0.73
18. Worried might get worse 3.3 ± 1.8 3.2 ± 1.7 —0.1 —0.30 0.7628 0.72§ 0.79
Note. The evaluation was done in 33 stable patients, defined as having a score of 0 (“about the same”) on the OTE.
ICC, intraclass correlation coefficient; OTE, Overall Treatment Effect.
* Window of 11–14 d.
†
Visit 2 mean score minus visit 1 mean score. Note. Values do not always equal the difference between the values presented in the 'Visit 1'
and 'Visit 2' columns because of rounding.
‡
Spearman rank-order correlations.
§
P o 0.001.
 VA LU E I N H EA LT H ] (2 0 1 7 ) ]]] – ] ]] 5

Excellent scores in the subgroup of patients with sinus rhythm at the

Very good
7.0 Good follow-up visit (n ¼ 46) (data not shown).
Fair
Poor
6.0
P<0.0001 P<0.0001 P<0.0001
5.0 Discussion
Mean score

4.0 An AF-specific HRQOL instrument developed and evaluated in

3.0
2.0
1.0
0.0
Vitality Emotional Distress
Fig. 1 – AFImpact domain scores by patient-rated general
health status. General health status was determined from
item 1 of the 36-item short form health survey (“In general,
would you say your health is excellent/very good/good/fair/
poor?”). Data represent mean and SD. Comparisons were
conducted using a parametric analysis of variance.
severity categories were 2.3, 2.9, 3.5, and 3.8 for vitality; 2.1, 2.8,
3.2 , and 3.6 for emotional distress; and 1.6, 2.4, 3.1, and 3.0 for
sleep. Very severe symptoms were reported for no patients.
Preliminary responsiveness to change
Table 3 presents the comparison of domain scores between
baseline and follow-up for patients who were undergoing car-
dioversion. Scores for each of the three domains of the AFImpact
decreased statistically from baseline to the follow-up visit at 2
weeks after cardioversion (Table 3). Significant changes from
baseline to follow-up were also observed for each of the 18 items
(Table 3), with effect sizes ranging from —0.19, for item 15
(“I felt depressed because of my heart condition”), to —0.65, for
item 4 (“I had less energy because of my heart condition”).
A significant decrease (P o 0.05) from baseline was also
observed for all three domain scores and for most (17 of 18) item
Not currently in AF (n=118)
Currently in AF (n=151)
6.0
P<0.0001 P=0.0356
5.0
4.0
Mean score
accordance with the Food and Drug Administration guidance has
long been an unmet clinical need. Results of the analyses
presented here support the psychometric validity of the AFIm-
pact as a PRO instrument to measure the impact of AF. The
AFImpact demonstrated satisfactory reliability and construct
Sleep
validity. Responsiveness to change was supported by preliminary
data from patients undergoing cardioversion. The instrument
includes 18 items covering three domains (vitality, emotional
distress, and sleep), with a 1-week recall period.
The AFImpact was developed with patients and in accordance
with current regulatory guidelines [9], in parallel with the
AFSymp [10]. Clinical trials in AF have used generic and/or
disease-specific instruments when measuring HRQOL, and sev-
eral PROs in the last decade or so having been developed with the
aim of being AF-specific [5,24]. Nevertheless, additional work is
required to support their psychometric validity for regulatory
purposes in the clinical trial setting. For example, additional
evidence is required for the AFEQT and the HeartQoL to support
their content validity [15–17] and for the Spanish AF-QoL to
support generalizability [12–14,25]. A similar number of patients
with paroxysmal, persistent, and permanent AF were included in
the development and validation of the AFImpact. This approach
ensured reliability and validity of the instrument in all three AF
subgroups, which is an issue that has not been dealt with
adequately during the development of other AF-specific instru-
ments. AF is a heterogeneous and progressive condition, and it
cannot be taken for granted that the symptomatology would be
the same among patients with paroxysmal, persistent, or perma-
nent AF. Palpitations, for example, occur more frequently among
patients with paroxysmal AF than among patients with perma-
nent AF, whereas fatigue is more common in patients with
permanent AF than in patients with paroxysmal AF. The AFIm-
pact also underwent cross-cultural development as part of con-
cept elicitation, thereby minimizing cultural bias and increasing
its applicability. Although about half the patients supported
recall periods of longer than 1 week, a 1-week recall period was
P=0.0204
chosen for the AFImpact in accordance with regulatory guidance,
which recommends that PRO instruments use short recall
periods [9].
HRQOL, using either generic or disease-specific instruments,
has been measured in clinical trials of interventions in AF,

although historically always as a secondary rather than a (co-)

3.0 primary end point [5]. A key reason for their not being used as
primary end points may have been the lack of comprehensively
evaluated, disease-specific measures to quantify HRQOL param-
2.0
eters objectively and to assess the effect of treatment reliably [5].
Preliminary results from the current analyses support respon-
1.0 siveness to change of the AFImpact after cardioversion. It
remains to be explored whether similar results would be obtained
0.0 after pharmacological intervention. Pharmacological therapy
Vitality Emotional Sleep tends to lead only to small to moderate improvements in HRQOL
Distress in patients with AF as measured with existing disease-specific
PROs [24], and even asymptomatic patients with AF have a lower
Fig. 2 – AFImpact domain scores for patients not currently in HRQOL than do their counterparts without AF [5]. Responsiveness
AF versus those currently in AF. Item 3 of the University of to change in clinical status has been shown for the AFEQT and
Toronto Atrial Fibrillation Severity Scale (“Are you in atrial the Quality of Life in AF questionnaires, both of which, however,
fibrillation currently?”) was used to determine whether still require further psychometric assessment fully to meet
patients were currently in AF. Data represent mean and SD. regulatory requirements for the clinical trial setting [14,17].
Comparisons were conducted using the Kruskal–Wallis test. Doubts about whether available PROs could measure HRQOL
AF, atrial fibrillation. sufficiently reliably in AF to support regulatory approvals led to
6 VA LU E I N H EA LT H ] (2 0 1 7 ) ]]] – ]] ]

Table 3 – Change in domains and item scores from baseline to follow-up in patients undergoing cardioversion
(N ¼ 118).
Domain/item Item score, mean ± SD P value‡ Effect size§ Standard
response mean||
Baseline Follow-up* Difference†

Domain
Vitality 3.1 ± 1.5 2.5 ± 1.4 —0.6 ± 1.3 o0.0001 —0.41 —0.50
Emotional distress 3.9 ± 1.8 3.0 ± 1.7 —1.0 ± 1.5 o0.0001 —0.53 —0.62
Sleep 2.6 ± 1.5 1.9 ± 1.2 —0.7 ± 1.3 o0.0001 —0.45 —0.51
Item
1. Did not sleep well 2.7 ± 1.7 1.9 ± 1.3 —0.8 ± 1.6 o0.0001 —0.45 —0.49
2. Worried might shorten my life 3.0 ± 1.8 2.4 ± 1.6 —0.6 ± 1.7 0.0002 —0.33 —0.35
3. Felt anxious 3.1 ± 1.7 2.5 ± 1.6 —0.6 ± 1.7 o0.0001 —0.38 —0.38
4. Less energy 4.5 ± 2.0 3.2 ± 2.0 —1.3 ± 1.9 o0.0001 —0.65 —0.68
5. Annoyed 3.5 ± 2.1 2.7 ± 1.8 —0.8 ± 1.9 o0.0001 —0.37 —0.40
6. Stressed 3.3 ± 1.8 2.5 ± 1.6 —0.8 ± 1.6 o0.0001 —0.43 —0.49
7. Walk slowly 3.5 ± 2.2 2.7 ± 1.9 —0.8 ± 1.8 o0.0001 —0.38 —0.47
8. Tired 4.3 ± 2.0 3.1 ± 1.9 —1.2 ± 1.7 o0.0001 —0.62 —0.69
9. Woke up during the night 2.6 ± 1.8 1.8 ± 1.4 —0.8 ± 1.8 o0.0001 —0.42 —0.42
10. Performed daily activities slowly 3.6 ± 2.1 2.8 ± 1.9 —0.9 ± 1.9 o0.0001 —0.41 —0.45
11. Worried about my health 3.3 ± 1.8 2.7 ± 1.7 —0.6 ± 1.7 0.0003 —0.32 —0.34
12. Felt worn out 3.9 ± 2.0 3.0 ± 2.0 —0.9 ± 2.0 o0.0001 —0.44 —0.43
13. Difficulty falling asleep 2.4 ± 1.6 1.9 ± 1.4 —0.5 ± 1.5 0.0003 —0.32 —0.35
14. Nervous 3.0 ± 1.7 2.3 ± 1.6 —0.7 ± 1.5 o0.0001 —0.43 —0.49
15. Depressed 2.4 ± 1.7 2.1 ± 1.5 —0.3 ± 1.5 0.0284 —0.19 —0.21
16. Had to rest during the day 3.4 ± 1.9 2.7 ± 1.8 —0.7 ± 1.8 o0.0001 —0.38 —0.40
17. Not as physically active 4.2 ± 2.2 3.3 ± 2.1 —0.9 ± 2.0 o0.0001 —0.41 —0.45
18. Worried might get worse 3.5 ± 1.9 2.8 ± 1.8 —0.7 ± 1.5 o0.0001 —0.35 —0.44
* Window of 2 wk ± 2 d.
†
Follow-up mean score minus baseline mean score. Note. Values do not always equal the difference between the values presented in the
‘Baseline’ and ‘Follow-up’ columns because of rounding.
‡
Matched t test.
§
Difference in mean scores±i.e., follow-up mean score – baseline mean score) divided by the SD of the baseline score.
||
Difference in mean scores±i.e., follow-up mean score – baseline mean score) divided by the SD of the score change.

their not being recommended as a primary outcome measure in
clinical trials in the last decade [26].
Another potential reason for HRQOL measures traditionally
not being used as primary end points is that trials of interven-
tions for AF, both pharmacological and nonpharmacological,
have focused much on measuring the effect on AF per se, such
as time to first recurrence, AF burden, and progression from
paroxysmal to persistent or permanent AF, and on correlating the
results to the risk of stroke or other outcomes. Nevertheless,
it seems sensible to use the effects on symptoms and the
impact of symptoms on HRQOL as a patient end point, because
symptoms are a major reason why patients with AF seek care [3].
This approach will become more feasible with the availability of
an AF-specific instrument developed to regulatory guidelines and
when trials are designed to detect a difference in severity of
symptoms as well as to assess the effect on symptom-related
HRQOL between a new treatment and placebo/no treatment or a
comparator treatment. AF is a long-term, chronic condition that
is generally symptomatic and it is thus useful to include a PRO to
monitor patient status over time. In future, AF-specific HRQOL
assessment may become a more important measure when differ-
entiating between treatments in clinical trials or considering
patient preferences for therapy options in clinical practice.
The development and psychometric evaluation of the AFIm-
pact support its reliability and validity. Nevertheless, although
the qualitative development of the questionnaire was based on
cross-cultural patient input, the quantitative development and
psychometric evaluation were conducted in the United States
and still need cross-cultural validation. Also, the assessments of
its responsiveness to change are based on preliminary data and
require confirmation. Furthermore, some patients may have
found it difficult to respond to item 3 of the AFSS, about whether
they were currently in AF. Last, the 1-week recall of the AFImpact
has potential drawbacks for patients who experience impacts of
AF less frequently than weekly.
Conclusions
The AFImpact was developed with patient input and in accordance
with current regulatory guidelines. Psychometric evaluations sup-
port the reliability and validity of the AFImpact as a PRO instru-
ment to measure the impact of AF, with preliminary results in
patients undergoing cardioversion supporting the instrument’s
responsiveness to change.
Acknowledgments
We thank Jennie Medin and Karoly Kulich for their contributions
to the study.
Source of financial support: This study was funded by Astra-
Zeneca Gothenburg (Mölndal, Sweden). Writing support was
 VA LU E I N H EA LT H ] (2 0 1 7 ) ]]] – ] ]]
provided by Dr Anja Becher from Oxford PharmaGenesis Ltd
(Oxford, UK) and was funded by AstraZeneca Gothenburg
(Mölndal, Sweden).
R EF ER ENCES
[1] Chugh SS, Roth GA, Gillum RF, Mensah GA. Global burden of atrial
fibrillation in developed and developing nations. Glob Heart
2014;9:113–9.
[2] Zoni-Berisso M, Lercari F, Carazza T, Domenicucci S. Epidemiology
of atrial fibrillation: European perspective. Clin Epidemiol
2014;6:213–20.
[3] Rienstra M, Lubitz SA, Mahida S, et al. Symptoms and functional status
of patients with atrial fibrillation: state of the art and future research
opportunities. Circulation 2012;125:2933–43.
[4] Nabauer M, Gerth A, Limbourg T, et al. The Registry of the German
Competence NETwork on Atrial Fibrillation: patient characteristics and
initial management. Europace 2009;11:423–34.
[5] Savelieva I, Kirchhof P, Danchin N, et al. Regulatory pathways for
development of antiarrhythmic drugs for management of atrial
fibrillation/flutter. Europace 2011;13:1063–76.
[6] Calkins H, Kuck KH, Cappato R, et al. 2012 HRS/EHRA/ECAS Expert
Consensus Statement on Catheter and Surgical Ablation of Atrial
Fibrillation: recommendations for patient selection, procedural
techniques, patient management and follow-up, definitions, endpoints,
and research trial design. Europace 2012;14:528–606.
[7] Peinado R, Arribas F, Ormaetxe JM, Badia X. Variation in quality of life
with type of atrial fibrillation. Rev Esp Cardiol 2010;63:1402–9.
[8] January CT, Wann LS, Alpert JS, et al. 2014 AHA/ACC/HRS guideline for the
management of patients with atrial fibrillation: a report of the American
College of Cardiology/American Heart Association Task Force on Practice
Guidelines and the Heart Rhythm Society. J Am Coll Cardiol 2014;64:e1–76.
[9] US Food and Drug Administration. Guidance for Industry. Patient-
Reported Outcome Measures: Use in Medical Product Development to
Support Labeling Claims. Rockville, MD: US Food and Drug
Administration, 2009.
[10] Medin J, Arbuckle R, Abetz L, et al. Development and validation of the
AFSymp: an atrial fibrillation-specific measure of patient-reported
symptoms. Patient 2014;7:319–27.
[11] European Medicines Agency. Reflection Paper on the Regulatory Guidance
for the Use of Health-Related Quality of Life (HRQL) Measures in the
7
Evaluation of Medicinal Products. London: Committee for Medicinal
Products for Human Use of the European Medicines Agency, 2005.
[12] Arribas F, Ormaetxe JM, Peinado R, et al. Validation of the AF-QoL, a
disease-specific quality of life questionnaire for patients with atrial
fibrillation. Europace 2010;12:364–70.
[13] Badia X, Arribas F, Ormaetxe JM, et al. Development of a questionnaire
to measure health-related quality of life (HRQoL) in patients with atrial
fibrillation (AF-QoL). Health Qual Life Outcomes 2007;5:37.
[14] Braganca EO, Filho BL, Maria VH, et al. Validating a new quality of life
questionnaire for atrial fibrillation patients. Int J Cardiol 2010;143:391–8.
[15] Dorian P, Burk C, Mullin CM, et al. Interpreting changes in quality of life
in atrial fibrillation: How much change is meaningful? Am Heart J
2013;166:381–7.
[16] Kristensen MS, Zwisler AD, Berg SK, et al. Validating the HeartQoL
questionnaire in patients with atrial fibrillation. Eur J Prev Cardiol
2016;23:1496–503.
[17] Spertus J, Dorian P, Bubien R, et al. Development and validation of the
Atrial Fibrillation Effect on QualiTy-of-Life (AFEQT) Questionnaire in
patients with atrial fibrillation. Circ Arrhythm Electrophysiol 2011;4:15–25.
[18] Camm AJ, Kirchhof P, Lip GY, et al. Guidelines for the management of
atrial fibrillation: the Task Force for the Management of Atrial Fibrillation
of the European Society of Cardiology (ESC). Europace 2010;12:1360–420.
[19] Wynn GJ, Todd DM, Webber M, et al. The European Heart Rhythm
Association symptom classification for atrial fibrillation: validation and
improvement through a simple modification. Europace 2014;16:965–72.
[20] Jaeschke R, Singer J, Guyatt GH. Measurement of health status:
ascertaining the minimal clinically important difference. Control Clin
Trials 1989;10:407–15.
[21] Ware JE Jr, Sherbourne CD. The MOS 36-item short-form health survey
(SF-36), I: conceptual framework and item selection. Med Care
1992;30:473–83.
[22] Dorian P, Jung W, Newman D, et al. The impairment of health-related
quality of life in patients with intermittent atrial fibrillation:
implications for the assessment of investigational therapy. J Am Coll
Cardiol 2000;36:1303–9.
[23] Hays RD, Stewart AL. Sleep measures. In: Stewart AL, Ware JEJ, eds.
Measuring Functioning and Well-Being: The Medical Outcomes Study
Approach. Durham, NC: Duke University Press, 1992. p. 235–59.
[24] Mark DB. Assessing quality-of-life outcomes in cardiovascular clinical
research. Nat Rev Cardiol 2016;13:286–308.
[25] Moreira RS, Bassolli L, Coutinho E, et al. Reproducibility and reliability
of the Quality of Life Questionnaire in patients with atrial fibrillation.
Arq Bras Cardiol 2016;106:171–81.
[26] Kirchhof P, Auricchio A, Bax J, et al. Outcome parameters for trials in
atrial fibrillation: executive summary. Eur Heart J 2007;28:2803–17.