Psychometric Development of the Individualized Retinopathy-Dependent Quality of Life Questionnaire (RetDQoL)

Authors


Clare Bradley, Department of Psychology, Royal Holloway, University of London, Egham, Surrey TW20 0EX, UK. Email: c.bradley@rhul.ac.uk

ABSTRACT

Objectives:  Psychometric development of the Retinopathy-Dependent Quality of Life (RetDQoL) questionnaire in a cross-sectional study of 207 German patients with diabetic retinopathy. Forty patients (19%) also had clinically significant macular edema.

Methods:  Principal component analyses identified factor structure, and Cronbach's alpha assessed internal consistencies. Construct validity was examined by testing the additional impact of macular edema and expected relationships of RetDQoL scores with visual impairment, stage of diabetic retinopathy, subscales of the SF-12, and scores of the Retinopathy Treatment Satisfaction Questionnaire (RetTSQ). Analyses were conducted using the RetDQoL's AWI score (average weighted impact of diabetic retinopathy on 26 life domains) and its two overview items (present QoL in general and retinopathy-specific QoL). Content validity was investigated using an open-ended question to identify any additional items needed.

Results:  A forced one-factor solution of the 26 specific weighted impact ratings showed all items except working life (applicable to 27%) to load >0.55, and Cronbach's alpha was 0.96, showing very high reliability. Greater impairment, worse diabetic retinopathy, and macular edema were associated with greater negative impact on scores. AWI correlated as expected more highly with retinopathy-specific QoL (r = 0.71, P < 0.01) than with present QoL (r = 0.28, P < 0.01). RetDQoL scores correlated moderately with SF-12 subscales (r = 0.22–0.51, P < 0.01) and RetTSQ scores (r = 0.27–0.51, P < 0.01). For six domains, >60% of patients reported no impact. No additional domains were needed.

Conclusions:  The RetDQoL is valid and reliable for patients with diabetic retinopathy with or without macular edema. It may be shortened if findings are confirmed cross-culturally.

Introduction

Diabetic retinopathy is one of the three leading causes of visual impairment and blindness in developed countries [1], and is the main reason for blindness in the working age population in Germany and other developed countries [2]. It is the most common microvascular complication of diabetes [3]; almost all patients with type 1 diabetes and over 60% of patients with type 2 diabetes develop retinopathy. The development can be classified into stages; nonproliferative or background retinopathy is characterized by retinal vascular abnormalities including microaneurysms and intraretinal hemorrhages, which at first do not have much effect on vision. Nonproliferative diabetic retinopathy can be classified as mild, moderate, or severe according to the extent of these abnormalities. If retinopathy advances further, it is classified as proliferative diabetic retinopathy, which is characterized by the growth of new blood vessels on the retina or into the vitreous cavity. These vessels are weak and may bleed, causing a sudden deterioration of vision. At any stage of the condition, macular edema can occur. Macular edema is characterized by a thickening of the retina due to leaky blood vessels, and impairs central vision [4,5]. Currently, the main treatments are laser photocoagulation or vitrectomy with efforts to improve glycemic control and blood pressure control. Treatment reduces the risk of progression of the condition and may increase the chance of a small improvement in visual acuity [3,5]. Newer treatment approaches include intraocular injections to inhibit vascular endothelial growth factors (VEGF) [5].

Evaluation of new treatments requires not only assessment of their impact on visual function, but also assessment of patients' satisfaction with the treatments and the impact of the condition and its treatment on individuals' quality of life (QoL). Individual QoL is different from health status and visual function. In asking about health rather than the eye condition, scores on health status measures such as the SF-12 [6] will be affected by comorbid conditions, while respondents may not even consider their eye condition to be an aspect of their health. Similarly, visual function measures such as the NEI-VFQ [7] will be influenced by ocular comorbidities. Health status and visual function measures ask about the impact of health or vision on daily activities. They do not take account of the relevance or importance of these activities to the individual. When researching the impact of eye conditions and treatments on QoL, both the impact of being unable to perform specific tasks and the importance of the tasks to the person need to be considered. Although it is reasonable to assume that some aspects of life are universally relevant to QoL, the weights attached to these have been shown to differ between individuals and within the same person over time following increased severity of the condition or onset of complications [8,9]. Some aspects such as work or family life may only be relevant to some people. Individualized measures of QoL such as the measure of Retinopathy-Dependent QoL (RetDQoL) are designed to measure the impact of a condition on aspects of life relevant to the individual; relevant aspects are weighted by the individual's ratings of the importance of these aspects of life to their QoL.

The objectives of the current analyses were to evaluate the psychometric properties and to determine optimal scoring of the individualized RetDQoL and to explore QoL in diabetic retinopathy.

Materials and Methods

Procedures

The data reported in this article were collected as part of the multicenter, retrospective “Cost of Illness Study for Diabetic Microvascular Complications (DIMICO)” in 2002 and 2003. The objectives of the main study phase were to assess the prevalence of stages of diabetic complications, analyze the resource utilization due to these complications, and estimate the total annual cost of diabetic microvascular complications in Germany. Health status and QoL were assessed. Data from over 500 patients were collected after obtaining their informed consent. Participants were adults with diabetes and retinopathy, neuropathy, or nephropathy. The present article focuses exclusively on the subgroup with diabetic retinopathy (n = 207). Demographic information and medical data on diabetes and the history and course of microvascular complications were collected from medical records and an interview with the patient conducted by the physician. Patients completed questionnaires during a visit to the physician before any treatment or examinations, and physicians were asked to check questionnaires for completeness. The following questionnaire measures were used:

  • 1Health status was measured using the SF-12 [6]. The SF-12 consists of 12 items which can be summarized into a physical health summary score and a mental health summary score. Higher scores represent better health.
  • 2Treatment satisfaction: The Retinopathy Treatment Satisfaction Questionnaire (RetTSQ) consists of 13 items asking participants to rate different aspects of treatment. It can be scored as total score or as two subscales, one covering negative experiences such as side effects and pain, and the other one covering positive aspects of treatment such as safety or efficacy. Higher scores represent more satisfaction [10,11].
  • 3Quality of Life was measured using the RetDQoL [12].

The RetDQoL

The RetDQoL is designed to measure individualized QoL in people with diabetic retinopathy and is modeled on the widely used Audit of Diabetes-Dependent QoL (ADDQol) [9,13,14]. These instruments measure individualized QoL by allowing the participant to indicate where items are not applicable to them, and, for applicable items, rate not only the impact of their condition (retinopathy or diabetes) on each aspect of life, but also the importance of each aspect of life to their QoL.

The RetDQoL was designed simultaneously in UK English and German for Germany. Content, wording, and format were established in in-depth qualitative interviews with 44 patients in four hospitals, two in the UK and two in Germany. All patients were diagnosed with diabetic retinopathy; 31 also had macular edema. Methods and findings from these interviews are reported elsewhere [12]. The RetDQoL is written in large font with a layout designed to facilitate reading by those with visual impairment.

Overview items.  The measure starts with two overview items. Overview item I (“present QoL”) asks participants to complete the statement “In general, my present quality of life is:” using a 7-point scale from “excellent,” scored as 3 through “neither good nor bad,” scored as 0 to “extremely bad,” scored as −3. Overview item II (“retinopathy-specific QoL”) asks how QoL is affected by diabetic eye problems: “If I did not have diabetic eye problems, my quality of life would be:” with the response options: “very much better” (scored −3), “much better” (−2), “better” (−1), “the same” (0), and “worse” (1).

Domain-specific items.  The RetDQoL further includes 26 items covering different domains of life. The domain-specific items each consist of one part to measure the impact of diabetic eye problems on this domain and a second part about the importance of this domain of life to the individuals' QoL. For wording and scoring of items 1 to 26 as used in this study and the English equivalents, see Table 1. Some items such as “working life” include a preliminary question to determine the applicability of the domain to the individuals' life and allow for a “not applicable” answer. The impact and importance ratings for each applicable item are multiplied to obtain a weighted impact score with a range from −9 to 3.

Table 1.  Wording of Retinopathy-Dependent Quality of Life (RetDQoL) questionnaire items 1–26 (2001 version)
Item Wording of part a (impact)Response options*Wording of part b (importance)
  • *

    Response options for part a: A) UK: very much better (−3), much better (−2), better (−1), the same (0), worse (1). DE: sehr viel besser (−3), viel besser (−2), besser (−1), genauso (0), schlechter (1). Scoring is the same for all of the following response options for part a; B) UK: very much easier–more difficult. DE: sehr viel leichter–schwieriger; C) UK: very much more–less. DE: sehr viel mehr–weniger; D) UK: very much less–more. DE: sehr viel weniger–mehr.

  • Wording of item has been changed in more recent versions of the questionnaire.

  • Preliminary yes/no question allowing “not applicable” answer.

  • §

    Item has been deleted in more recent versions.

  • RetDQoL© Prof Clare Bradley 29.11.01.

 1UKIf I did not have diabetic eye problems, I could handle my household tasks:AHandling my household tasks is:
DEWenn ich keine diabetischen Augenprobleme hätte, könnte ich meine Aufgaben im Haushalt . . . . . . . . erledigen:ADie Erledigung meiner Aufgaben im Haushalt ist für mich:
1a–26a all begin with:
UK: “If I did not have diabetic eye problems,”
DE: “Wenn ich keine diabetischen Augenprobleme hätte,”
 Response options for 1b–26b:
UK: very important (3), important (2), somewhat important (1), not at all important (0). DE: sehr wichtig (3), wichtig (2), relativ wichtig (1), gar nicht wichtig (0)
 2UKI could handle my personal affairs (letters, bills, etc.):AHandling my personal affairs is:
DEkönnte ich meine persönlichen Angelegenheiten (Briefe, Rechnungen usw.) . . . . . . . . erledigen:ADie Erledigung meiner persönlicher Angelegenheiten ist für mich:
 3UKmy experience of shopping would be:AMy experience of shopping is:
DEwären meine Erfahrungen beim Einkaufen:AMeine Erfahrungen beim Einkaufen sind für mich:
 4UKmy feelings about the future (e.g., worries, hopes) would be:AMy feelings about the future are:
DEwären meine Gefühle in Bezug auf die Zukunft (z.B. Sorgen, Hoffnungen):AMeine Gefühle in Bezug auf die Zukunft sind für mich:
 5UKmy feelings about the past (e.g., anger or regret about earlier diabetes care) would be:AMy feelings about the past are:
DEwären meine Gefühle in Bezug auf die Vergangenheit (z.B. Ärger oder Bedauern wegen der früheren Behandlung oder meiner früheren Lebensweise):AMeine Gefühle in Bezug auf die Vergangenheit sind für mich:
 6†‡UKmy working life and work-related opportunities would be:AMy working life is:
DEwären mein Berufsleben und meine beruflichen Möglichkeiten:AMein Berufsleben ist für mich:
 7UKmy close personal relationship (e.g., marriage, living companion, steady relationship), now or in the future, would be:AFor me, having a close personal relationship is:
DEwäre meine feste Beziehung (z. B. Ehepartner, Lebensgefährte, feste Freundschaft) jetzt oder in Zukunft:AEine feste Beziehung zu haben ist für mich:
 8UKmy family life would be:AMy family life is:
DEwäre mein Familienleben:AMein Familienleben ist für mich:
 9UKmy friendships and social life would be:AMy friendships and social life are:
DEwären meine Freundschaften und meine sozialen Kontakte:AMeine Freundschaften und sozialen Kontakte sind für mich:
10UKI could do things for others as I wish:AFor me, doing things for others is:
DEkönnte ich . . . . . . . . etwas für andere tun wie ich es möchte:ADinge für andere zu tun ist für mich:
11UKI could get out and about (e.g., on foot, or by car, bus, or train):AFor me, getting out and about is:
DEkönnte ich . . . . . . . . raus- und herumkommen (z.B. zu Fuß, mit dem Auto, dem Bus oder mit der Bahn):ARaus- und herumzukommen ist für mich:
12‡§UKI would find long journeys:BFor me, long journeys are:
DEfände ich weite Reisen:BWeite Reisen sind für mich:
13UKmy holidays would be:AFor me, holidays are:
DEwäre mein Urlaub:AUrlaub ist für mich:
14UKmy financial situation would be:AMy financial situation is:
DEwäre meine finanzielle Situation:AMeine finanzielle Situation ist für mich:
15UKthe way people in general react to me would be:AThe way people in general react to me is:
DEwäre die Art und Weise, wie andere auf mich reagieren:AWie andere auf mich reagieren ist für mich:
16UKmy physical appearance (including clothes and grooming) would be:AMy physical appearance is:
DEwäre meine äußere Erscheinung (einschließlich Kleidung und Gepflegtheit):AMeine äußere Erscheinung ist für mich:
17UKphysically, I could do:CFor me, how much I can do physically is:
DEwäre ich körperlich zu . . . . . . . . in der Lage:CWozu ich körperlich in der Lage bin, ist für mich:
18UKI could pursue or enjoy my leisure activities (e.g., reading, TV, radio, cinema:AMy leisure activities are:
DEkönnte ich meine Freizeitaktivitäten (z. B. Lesen, Fernsehen, Radio hören, Kino) . . . . . . . . verfolgen oder genießen:AMeine Freizeitaktivitäten sind für mich:
19§UKI could pursue or enjoy my hobbies and interests:AMy hobbies and interests are:
DEkönnte ich meine Hobbys und Interessen . . . . . . . . verfolgen oder genießen:AMeine Hobbys und Interessen sind für mich:
20UKmy self-confidence would be:AMy self-confidence is:
DEwäre mein Selbstvertrauen:AMein Selbstvertrauen ist für mich:
21UKmy motivation to achieve things would be:AMy motivation is:
DEwäre meine Motivation, etwas zu erreichen:AMeine Motivation, etwas zu erreichen, ist für mich:
22UKI would have to depend on others when I do not want to:DFor me, not having to depend on others is:
DEwäre ich . . . . . . . . auf andere angewiesen:DNicht auf andere angewiesen zu sein ist für mich:
23UKI would have mishaps or would lose things:DFor me, not having mishaps or losing things is:
DEwürde ich . . . . . . . . Missgeschicke erleben oder Dinge verlieren:DKeine Missgeschicke zu erleben und keine Dinge zu verlieren ist für mich:
24UKthe time it takes me to do things would be:DThe time it takes me to do things is:
DEwäre die Zeit, die ich zur Erledigung von Dingen brauche:DDie Zeit, die ich zur Erledigung von Dingen brauche, ist für mich:
25UKI would find taking care of my diabetes (e.g., self-testing, medication, diet, exercise):BTaking care of my diabetes is:
DEkönnte ich mich . . . . . . . . um meinen Diabetes kümmern (z.B. Selbstkontrollen, Medikamente einnehmen, Ernährung und körperliche Aktivität beachten):BMich um meinen Diabetes zu kümmern ist für mich:
26UKI could enjoy nature:CMy enjoyment of nature is:
DEkönnte ich die Natur . . . . . . . . genießen:CDie Natur zu genießen ist für mich:

Average weighted impact (AWI) score.  A total score, the AWI of the condition on QoL, can be obtained by summing the weighted impact scores of all applicable domain-specific items and dividing the result by the number of applicable domains. This is justified if factor analysis indicates that all domain-specific items measure one overall construct.

The RetDQoL finishes with an open-ended question that asks whether diabetic eye problems affect QoL in any way not already covered by the questionnaire. The data analyzed here were obtained using the 2001 German for Germany version of the questionnaire; a few changes have been made to the questionnaire since then (see Discussion). Nevertheless, structure, layout, and general content have stayed the same.

Analyses

Statistical analyses were conducted using SPSS 12.01 and 14. Principal component analyses with Varimax rotation were carried out to identify possible subscales. The 26 domain-specific items were included, and “not applicable” answers for items 6, 8, 12, and 13 were replaced with zeros to allow inclusion of all participants [8,9]. Internal consistencies were assessed with Cronbach's alpha. Corrected item–total correlations and alpha-if-item-deleted statistics indicated the strength of each item's association with the construct.

Construct validity was assessed by examining expected relationships between questionnaire scores and clinical data, using correlation indices (Pearson's r and Spearman's rho), t tests, and one-way or two-way independent analyses of variance (ANOVA) with post hoc tests. It was expected that RetDQoL scores would have significant relationships with level of visual impairment and stage of retinopathy, as well as with the occurrence of macular edema regardless of the stage of the disease, although the subgroups with macular edema were very small. Greater visual impairment and advanced stages of disease, as well as the additional impact of macular edema, were expected to lead to more negative impact on QoL. It was expected to find significant associations with clinical variables for both the overview items and the domain-specific items with stronger associations for the retinopathy-specific overview item II and AWI score than for present QoL (overview item I). When the stage of retinopathy differed between the eyes of individual participants or data were only available for one eye (n = 19, 9.2% of participants), the stage of the better eye or the available data, respectively, were used for subgroup categorization. It was also expected to find significant correlations between AWI and overview items I and II, with the strongest positive relationship between AWI and overview item II (retinopathy-specific QoL). Smaller significant correlations with subscales of the SF-12 and treatment satisfaction as measured by the RetTSQ were also expected. No significant relationships with sociodemographic variables were expected; however, these were also explored. Item distributions and total scores were nonnormally distributed; therefore, nonparametric tests of relationships between variables were performed to check parametric results. When Levene's test for equality of variances indicated unequal variances for an ANOVA, an approximation to a permutation test was performed. Neither result altered the conclusion reached from parametric results, thus they are not reported. Answers to the open-ended question were assessed to see if additional items or modifications were needed.

Results

Sample

Data for 207 participants were available. Mean age was 60.94 years with a range from 18 to 92 years. The majority (73.9%) had type 2 diabetes. Participants had been diagnosed with diabetes between 1 and 51 years ago (mean [M] = 19.89 years, standard deviation [SD] = 10.33). For further sociodemographic and condition-related details, see Table 2. Of the 157 participants not in employment at the time of the survey, 121 were retired; 26 of those had retired early, for 18 of them diabetes or its complications was a reason for early retirement. Visual acuity in the better eye (decimal) ranged from 0.01 to 1.25. The participants were classified in five groups from lowest visual acuity (≤0.2) to good vision (>0.8). A high proportion of participants had little or no loss of visual acuity in their better eye; 33 had visual acuities under 0.3, classified as low vision by the World Health Organization [15]. Stage of retinopathy was categorized by one or more methods using direct fundoscopy, stereobiomicroscopic examination of the fundus, or fluorescein angiography. The participants had experienced a wide range of ophthalmic treatments during the previous year.

Table 2a.  Sample characteristics
 FrequencyPercent
  1. A total <100% indicates missing data.

SexFemale10450.2
Male10349.8
Total207100.0
Marital statusSingle188.7
Married/Partnered14871.5
Divorced167.7
Widowed2311.1
Total20599.0
Living situationAlone3617.4
With partner/family14469.6
Other10.5
Total18187.4
Employment statusEmployed5024.2
Not employed15775.8
Total207100.0
Visual acuity≤0.2 (Low acuity)2411.6
0.21–0.40157.2
0.41–0.603516.9
0.61–0.804521.7
>0.8 (Good vision)6531.4
Total18488.8
Table 2b.  Sample characteristics, stage of diabetic retinopathy (better eye if different in both eyes), and presence of clinically significant macular edema per group
Stage of diabetic retinopathyFrequency (%)Significant macular edema frequency
Mild nonproliferative46 (22.2)1
Moderate nonproliferative56 (27.1)8
Severe nonproliferative50 (24.2)9
Proliferative55 (26.6)22
Total207 (100.0)40 (19.3%)

Descriptives

The participants on average rated present QoL in general (overview item I) as approaching “good” (M = 0.70; SD = 0.82), but indicated QoL would be better without diabetic eye problems (overview item II, M = −1.43; SD = 1.07; Table 3). Items were missed by very few participants; with 10 (4.8%) answers missing, the item asking for any other ways in which the condition affects their QoL (item 27) was the only one missed by more than 4% of participants. Observed scores included the minimum possible score for all domains; positive scores were infrequent (Tables 4 and 5).

Table 3.  Descriptives for overview items
Overview itemnMissingMinimumMinimum scored by (%)MaximumMaximum scored by (%)Mean (SD)SkewKurtosis
(I) Present QoL2025−30.531.00.70 (0.82)−0.572.04
(II) Retinopathy-specific QoL2043−320.1024.0−1.43 (1.07)−0.08−1.20
Table 4.  Descriptives: impact and importance of domain-specific items
ItemNMissingImpact scoreImportance score
part a/b or n.a.*Mean (SD)MinimumMaximumMean (SD)MinimumMaximum
  • *

    Preliminary yes/no question allowing “not applicable” answer, leading to smaller n.

  • n.a., not applicable.

 (1) Household tasks2043/2−1.05 (1.09)−302.19 (0.68)03
 (2) Personal affairs2052/1−1.07 (1.13)−302.51 (0.56)13
 (3) Shopping2043/2−1.05 (1.09)−302.12 (0.72)03
 (4) Feelings about future2034/3−1.44 (1.03)−302.37 (0.61)03
 (5) Feelings about past2016/3−0.94 (1.00)−311.75 (0.73)03
 (6) Working life*555−0.77 (0.97)−302.27 (0.79)03
 (7) Close relationship2007/7−0.43 (0.86)−312.37 (0.83)03
 (8) Family life*1973−0.60 (0.92)−302.66 (0.49)13
 (9) Friendships/social life2043/2−0.52 (0.87)−312.34 (0.58)13
(10) Do things for others2043/2−0.82 (0.97)−302.11 (0.59)13
(11) Get out and about2052/2−1.08 (1.17)−302.29 (0.59)13
(12) Journeys*995−1.03 (1.28)−311.66 (0.84)03
(13) Holidays*1555−0.91 (1.05)−312.07 (0.75)03
(14) Finances2052/2−0.48 (0.84)−302.32 (0.56)03
(15) People react to me2043/3−0.33 (0.67)−301.80 (0.76)03
(16) Physical appearance2043/2−0.30 (0.68)−302.16 (0.57)03
(17) Do physically2024/4−0.98 (1.02)−302.31 (0.55)03
(18) Leisure activities2052/1−1.14 (1.14)−302.15 (0.62)03
(19) Hobbies2007/6−1.06 (1.10)−302.20 (0.63)03
(20) Self-confidence2034/3−0.86 (1.04)−302.38 (0.55)13
(21) Motivation2052/2−0.83 (1.00)−312.10 (0.61)03
(22) Depend on others2043/2−0.82 (1.07)−312.43 (0.61)03
(23) Mishaps/lose things2034/4−0.72 (0.95)−302.08 (0.68)03
(24) Time it takes2043/3−0.96 (1.06)−311.92 (0.65)03
(25) Care of diabetes2043/2−0.77 (1.04)−302.59 (0.52)13
(26) Enjoy nature2052/1−0.97 (1.12)−302.42 (0.61)13
(27) Any other ways19710n.a.n.a.n.a.n.a.n.a.n.a.
Table 5.  Domain-specific items in order of weighted impact (WI); average weighted impact (AWI)
Item (item number)Mean WI (SD)Min.Max.WI = −9 (%)WI = 0 (%)WI > 0 in (%)SkewKurtosisRank impact
  • *

    Preliminary yes/no question allowing “not applicable” answer.

Feelings about future (4)−3.64 (2.99)−9015.322.7−0.51−0.821
Personal affairs (2)−2.80 (3.18)−9013.243.4−0.83−0.664
Leisure activities (18)−2.62 (2.91)−908.842.0−0.86−0.392
Get out and about (11)−2.59 (3.02)−9010.245.9−0.89−0.423
Hobbies (19)−2.56 (2.98)−9011.043.0−1.00−0.145
Enjoy nature (26)−2.53 (3.14)−9011.750.2−0.94−0.4610
Shopping (3)−2.50 (2.91)−909.842.6−1.03−0.046
Journeys* (12)−2.48 (3.25)−9113.147.51.0−1.02−0.448
Household tasks (1)−2.48 (2.86)−908.342.2−0.992.047
Do physically (17)−2.37 (2.75)−908.441.6−1.140.429
Self-confidence (20)−2.15 (2.80)−907.452.7−1.150.2713
Depend on others (22)−2.10 (2.98)−9310.352.91.0−1.230.4515
Holidays* (13)−2.10 (2.72)−906.549.7−1.190.4117
Working life* (6)−2.09 (2.79)−9010.947.3−1.571.6819
Care of diabetes (25)−2.02 (2.90)−908.857.8−1.300.4718
Time it takes (24)−2.01 (2.49)−933.944.11.0−1.080.5511
Motivation (21)−1.88 (2.53)−935.948.81.0−1.351.3814
Feelings about past (5)−1.88 (2.42)−924.545.80.5−1.351.3112
Do things for others (10)−1.84 (2.40)−903.950.0−1.351.1916
Mishaps/lose things (23)−1.73 (2.50)−904.955.2−1.511.7820
Family life* (8)−1.61 (2.55)−904.164.5−1.471.0921
Friendships/social life (9)−1.25 (2.27)−933.466.70.5−1.923.2822
Finances (14)−1.22 (2.26)−903.470.7−1.973.2823
Close relationship (7)−1.06 (2.27)−924.074.50.5−2.294.5124
People react to me (15)−0.71 (1.62)−901.576.0−3.0410.5825
Physical appearance (16)−0.68 (1.69)−901.580.4−3.039.6126
AWI−2.05 (1.97)−7.330.08n.a.n.a.n.a.−0.90−0.21n.a.

As shown in Tables 4 and 5, on average, the least impacted domains were “the way people in general react to me” (item 15, impact = −0.33; weighted impact = −0.71) and “physical appearance” (item 16, impact = −0.30; weighted impact = −0.68). The most impacted domain was “feelings about the future” (item 4, impact = −1.44; weighted impact = −3.64). For six items, over 60% of respondents indicated no weighted impact. The weighted impact for these items was zero for all but two of these participants because they reported no impact on these domains of life, not because these domains were unimportant to them. Unweighted and weighted impact scores were highly correlated (r = 0.9). Nevertheless, weighting had considerable effects on domain scores. In total, the 207 participants responded 5010 times to domain-specific items. Of these, 1051 (21.0%) indicated little negative impact by ticking −1 in response to part a of a domain-specific item. Only a small minority of 119 (11.3% of 1051) indicated 1, “somewhat important” in the importance rating for the same domain, which means that weighting does not change the impact, while 666 (63.4%) indicated 2 (“important”) and 255 (24.3%) indicated 3 (“very important”). The remaining 11 (1%) scores were 0 (“not at all important”). Weighting by importance also changed the ranking of 16 of 26 domains; it changed by three or more places for nine domains.

Several items correlated significantly (P < 0.01) and highly with at least one other item, most notably item 12, “journeys,” which correlated >0.75 with “personal affairs,”“working life,”“get out and about,”“holidays,”“leisure activities,” and “hobbies.”“Leisure activities” also correlated highly with “hobbies” (r = 0.83). “Working life” correlated >0.75 with “get out and about” and “finances.”“Personal affairs” and “household tasks” were highly correlated (r = 0.83).

Factor Structure

An unforced solution resulted in items 1 to 26 splitting onto four factors with no clear structure and several items having substantial double loadings. In a forced one-factor solution, all items loaded >0.55, except “working life” (0.22). Because of its low loading and the small number of participants to whom it was applicable (n = 55, 26.6%), “working life” has been excluded from principal component analyses.

When the remaining 25 items were forced on one factor (Table 6), this solution explained 51.7% of variance with loadings from 0.84 for “get out and about” to 0.56 for “close relationship.” This structure allows computing a total score, the AWI for all domain-specific items (including the “working life” item if applicable) with a possible range from −9 to 3. In this sample, AWI ranged from −7.33 to 0.08 with a mean of −2.05 (SD = 1.97; n = 206).

Table 6.  Retinopathy-Dependent Quality of Life component matrix
Item (item number)Component 1
  1. Extraction method: principal component analysis. Items with “not applicable” answers scored as 0.

Household tasks (1)0.782
Personal affairs (2)0.777
Shopping (3)0.739
Feelings about future (4)0.585
Feelings about past (5)0.580
Close relationship (7)0.561
Family life (8)0.705
Friendships/social life (9)0.659
Do things for others (10)0.783
Get out and about (11)0.844
Journeys (12)0.611
Holidays (13)0.581
Finances (14)0.569
People react to me (15)0.595
Physical appearance (16)0.574
Do physically (17)0.793
Leisure activities (18)0.841
Hobbies (19)0.833
Self-confidence (20)0.769
Motivation (21)0.802
Depend on others (22)0.770
Mishaps/lose things (23)0.793
Time it takes (24)0.772
Care of diabetes (25)0.674
Enjoy nature (26)0.816

Reliability and Implications for Missing Values

Internal consistency of weighted impact scores for all domains was excellent at α = 0.958 if all items were included and “not applicable” answers substituted with 0. With “working life” excluded, it rose marginally to α = 0.960. This makes the measure suitable for application both at group and individual level [16]. Internal consistency was very robust against omissions; it stayed above 0.9 with up to the 12 strongest items omitted. Except for “working life” (0.21), corrected item–total correlations ranged from 0.53 (“close relationship”) to 0.83 (“get out and about”), thus far exceeding the recommended minimum value of 0.2 [17]. Internal consistency for the unweighted impact and the importance ratings alone was high (α = 0.96 and 0.84, respectively). Reliability in the form of reproducibility (test–retest) could not be assessed using the present cross-sectional data.

Validity

Overview items I and II.  Participants with different levels of visual impairment showed significant differences in their rating of overview item I, present QoL (F[4,175] = 8.75, P < 0.001). Participants with good vision reported significantly better present QoL than those in the three groups with visual acuities of 0.6 and worse. There were no significant differences between other groups. Level of visual impairment was also significantly associated with overview item II, retinopathy-specific QoL (F[4,177] = 10.20, P < 0.001). Participants in the two groups with the lowest visual acuities scored significantly more negatively than participants in the other three groups. Other group differences were not significant.

Stage of diabetic retinopathy had no significant association with present QoL (F[3,198] = 1.322, n.s.), but was related to significant differences in retinopathy-specific QoL; participants with proliferative retinopathy reported worse retinopathy-specific QoL than those with nonproliferative retinopathy (t = 3.33, P < 0.01). Participants with mild nonproliferative retinopathy reported significantly better retinopathy-specific QoL than those with moderate or severe nonproliferative or proliferative retinopathy (F[3,200] = 11.22, P < 0.001). Differences between other groups were not significant.

Macular edema was associated with significantly worse present QoL (F[1,194] = 11.61, P < 0.01) and retinopathy-specific QoL (F[1,196] = 6.86, P < 0.05), regardless of stage of diabetic retinopathy.

AWI correlated much more strongly with retinopathy-specific QoL (r = 0.71, rho = 0.75, both P < 0.001) than with present QoL (r = 0.28, rho = 0.27, both P < 0.001).

The subscales of the SF-12 correlated significantly with the RetDQoL overview items. Present QoL correlated more strongly with the physical subscale than with the mental subscale, while retinopathy-specific QoL correlated more strongly with the mental subscale. The overview items of the RetDQoL also correlated significantly with RetTSQ scores of treatment satisfaction (Table 7).

Table 7.  Correlations with SF-12 and Retinopathy Treatment Satisfaction Questionnaire (RetTSQ) scores
 RetDQoL scores
Present QoLRetinopathy-specific QoLAWI
rrhorrhorrho
  • *

    P < 0.01.

  • P < 0.05.

  • RetTSQ subscale 1 consists of items covering negative experiences such as side effects and pain; subscale 2 covers positive aspects of the treatment such as efficacy and safety.

  • AWI, average weighted impact.

SF-12 physical0.51*0.54*0.22*0.22*0.33*0.33*
SF-12 mental0.38*0.37*0.330.320.34*0.34*
RetTSQ total score0.51*0.45*0.43*0.46*0.46*0.52*
RetTSQ subscale 10.46*0.41*0.46*0.46*0.48*0.54*
RetTSQ subscale 20.44*0.37*0.28*0.28*0.27*0.31*

Some demographic variables showed associations with present QoL. Those living with a partner or family reported better present QoL (t = 2.54, P < 0.05) than those living alone, as did those in employment compared to people not in employment (t = 2.22, P < 0.05). These differences lost significance when visual acuity was taken into account. Present QoL showed a tendency to decrease with age (r = −0.12, n.s, rho = −0.18, P < 0.05), which can also be explained by a significant correlation of visual acuity (ungrouped) and age (r = −0.29, P < 0.01; rho = −0.33, P < 0.01). Men reported better present QoL than women (t = 2.74, P < 0.01). As women were older, more likely not to be employed and to live alone than men in this sample, these variables and visual acuity were entered into a regression to explain present QoL, which showed visual acuity to be the only significant predictor of present QoL.

Average weighted impact (AWI).  Visual impairment was significantly associated with AWI. The better the vision, the less negative impact participants reported (F[4,179] = 22.83, P < 0.001); differences were significant between all groups except for the ones next to each other as shown in Figure 1. Participants with proliferative diabetic retinopathy reported more negative impact on QoL than those with nonproliferative retinopathy (t = −3.67, P < 0.001). Participants with mild nonproliferative retinopathy reported significantly less negative impact than all other groups (F[3,202] = 11.03, P < 0.001; Fig. 2). Differences between other groups were not significant. Patients with macular edema reported significantly stronger negative impact on their QoL than those without macular edema, regardless of stage of diabetic retinopathy (F[1,198] = 9.08; P < 0.01).

Figure 1.

Average weighted impact (AWI) in groups with different levels of visual impairment (visual acuity, decimal notation). AWI possible scores range from 3 to −9. A more negative AWI score indicates a more negative impact of visual impairment on quality of life (QoL); a positive score would indicate a positive impact of visual impairment on QoL. *P < 0.05; **P < 0.01; ***P < 0.001.

Figure 2.

Average weighted impact (AWI) in different stages of diabetic retinopathy. AWI possible scores range from 3 to −9. A more negative AWI score indicates a more negative impact of diabetic retinopathy on quality of life (QoL); a positive score would indicate a positive impact of diabetic retinopathy on QoL. **P < 0.01; ***P < 0.001.

AWI correlated significantly but not highly with the two subscales of the SF-12 and with RetTSQ scores (Table 7).

Negative impact on QoL showed a tendency to increase with age (r = −0.162, P < 0.05; rho = −0.122, n.s.). This relationship disappeared when controlling for visual acuity (r = −0.13, n.s.). Participants not in employment reported a more negative impact than those in employment (t = 3.86; P < 0.01), and this difference was significant regardless of level of visual impairment (F[1,174] = 5.01, P < 0.05).

Additional aspects described.  The open-ended question in item 27 was answered by 12 participants (5.8%). Problems in recognizing acquaintances or friends on the street were mentioned twice. Some other aspects described appeared to be covered in existing items. For example, “I am always dependent on someone, can no longer go by car, bus or train unaccompanied” underlines the relevance of items 11 (“to get out and about”) and 22 (“depend on others”). “Particular difficulties when completing forms, payments into the bank, etc.” emphasizes difficulties handling personal affairs as described in item 2, which includes letters and bills as examples of personal affairs. Some aspects described did not appear to be directly relevant to the measure, such as “Impaired blood flow in both legs has been improved by venous catheter.” This appears to be an explanation for recent improvement unrelated to the eye condition rather than a suggestion requiring a new item.

Discussion

The psychometric properties of the measure reported here are excellent. Visual acuity of many participants in this sample was good, reflected in the skewed distribution of scores. Even though the questionnaire showed nonnormally distributed data and some unequal variances across groups, nonparametric tests confirmed all results. Score distribution will be less skewed in samples with more severe visual impairment. The questionnaire showed a high completion rate, although this should be interpreted with caution as physicians were asked to check for completeness.

The overview items of the RetDQoL showed that on average, the participants rated their present QoL as being in between “neither good nor bad” and “good,” with a modal response of “good.” They expected their QoL to be “a little better” to “much better” if they did not have their eye condition (modal response “a little better”). The difference between present QoL and condition-specific QoL shows the importance of using condition-specific measures and not relying on generic QoL measures alone when making statements about the impact of medical conditions on QoL.

By far, the most negatively impacted domain of life was feelings about the future, showing that even a well-monitored and treated eye condition can lead to uncertainty and fear about how one's life will be affected by it in the future. The ranking of domains is likely to be different in samples with more advanced retinopathy or visual impairment.

A highly reliable scale in terms of internal consistency resulted from principal component analysis when all items except “working life” were forced onto a single factor. The “working life” item was omitted because of low loadings and applicability to only few participants in the present sample, which can explain the low loading. Nonetheless, it should not be removed from the measure, as participants who completed it clearly rated it as negatively impacted and important (Table 4), and it may be more applicable in samples with a lower proportion of retired participants. Internal consistency proved to be robust against missing items as it stayed above 0.9 with almost half the items omitted. This indicates clearly that excellent internal consistency reliability can be maintained if the mean scale value for the individual is substituted for up to 12 items. No more than 12 items should be substituted, as to do so would detract from content validity. The range of item–total correlations indicates that the individual items represent the underlying construct well. The measure can be used in individual patient management, for example to identify priorities for rehabilitation, and at group level, for example to compare different treatments. Reliability in the form of reproducibility or stability of QoL ratings over time needs to be assessed using longitudinal data from a sample with stable retinopathy. For the sister measure MacDQoL for people with macular disease, Mitchell et al. [18] reported excellent test–retest reliability, suggesting that RetDQoL scores may be similarly stable.

Good construct validity is indicated by the measure's sensitivity to different levels of visual impairment, different stages of disease progression and macular edema, as the expected relationships were found for both the overview items and AWI. This sensitivity to group differences suggests that the measure will be responsive to changes; however, longitudinal data before and after treatment or rehabilitation are needed to confirm this. The AWI score reflects, as intended, QoL as impacted by retinopathy and not general QoL. This is suggested by a much stronger correlation of AWI with the retinopathy-specific overview item than with the present QoL item. The high correlation between AWI score and the retinopathy-specific overview item makes it possible to use overview item II alone if participant burden is of particular concern and a very brief measure of condition-specific QoL is desirable. Nevertheless, this would lead to a loss of detailed information obtainable by the specific individualized items.

The variability in importance ratings demonstrates that, without weighting by the importance of the domain, the impact of diabetic retinopathy on aspects of life would have been underestimated for many individuals. This effect is masked when correlating average scores because of a high proportion of participants reporting no impact on life domains. Weighting also influenced the ranking of impact on life domains considerably with the rankings of nine domains being changed by three or more places.

The RetDQoL and the SF-12 show some overlap, but with correlations of only 0.22 to 0.51 it is clear that the instruments measure very different phenomena. It is to be expected that there will be modest correlations between a health status measure and a measure of the impact of diabetic retinopathy on QoL, particularly as people who have more severe diabetic retinopathy are likely to be more at risk of other microvascular complications of diabetes including nephropathy and neuropathy, which will lead to reduced health scores. That the SF-12 subscales correlate more strongly with the overview item about present QoL than with the retinopathy-specific item or AWI confirms that the SF-12 measures a more generic construct than the condition-specific RetDQoL. Correlations between scores of the RetTSQ and the RetDQoL show that negative impact on QoL is associated with less treatment satisfaction, but the modest size of the correlation indicates that the instruments measure different aspects of the experience of diabetic retinopathy.

Interestingly, at all levels of visual impairment, AWI was significantly less negatively impacted in people in employment than in those who were not in employment. Employment may have a protective effect on QoL via its benefits to well-being. The causality may also be reversed with those who feel less negatively impacted by their retinopathy being more attractive to employers and less likely to seek early retirement, or AWI and employment may be linked via a third variable.

Answers to the open-ended question mainly emphasized aspects already covered in existing items. Nevertheless, similar to the answers given in the current study, when evaluating the MacDQoL, Mitchell et al. [8] also reported an additional aspect mentioned was not being able to recognize people. This is further supported by reports from focus groups on the impact of diabetic retinopathy on life [19], where difficulties recognizing faces were one of the key concerns reported. This aspect was expected to be covered by “friendships/social life” or “people react to me;” however, those adding these comments did indicate little or no impact on these domains. Other key concerns expressed in the focus groups included inability to drive or driving restrictions, decreased mobility, loss of independence, decreased social activities, impact on general day-to-day tasks, inability or difficulties in reading, and difficulties in maintaining diabetes care activities, all of which are reflected in items in the RetDQoL.

Since the start of this study, the number of items in the RetDQoL has been reduced. Following design and psychometric development of the MacDQoL [8,20], “hobbies” and “leisure activities” have been merged; “long journeys” has been deleted because of considerable overlap with “holidays.” This removes most of the high correlations between items; even though some remain, the items involved cover different aspects of life separately impacted by the condition. “Depend on others” now asks about independence instead and includes a preliminary question to establish relevance. Following linguistic validation to produce translations of the measure, “feelings about past,”“working life,”“close relationship,” and “motivation” have been simplified and shortened. A preliminary question has been added to “close relationship” to check applicability. “Diet” in “diabetes care” has been changed to “food” to be relevant to those who do not consider themselves to be on a diet. “Somewhat important” in the importance rating scale is now translated as “etwas wichtig” in German because recent interviews [21] indicated that the previous wording represented a higher importance than intended. These changes are not expected to have a major influence on psychometric properties; small improvements are more likely than detrimental effects.

Removing some items with high proportions of no reported impact from the RetDQoL has been considered. This would be desirable as it would reduce the burden on the participants and it is supported by an alpha that is very robust with a shorter scale. Nevertheless, before removing items, further confirmation of these findings in different populations and cultures is necessary. This is particularly important as a large proportion of participants in this sample had no or little visual impairment, and the sample had access to the comparatively good and reliable health-care system in Germany where the vast majority of people are covered to a great extent by health insurance schemes.

When linguistically validated versions in other languages are used in other countries, the psychometric properties of the RetDQoL will need to be examined for each language version/country. The RetDQoL together with the MacDQoL has been used as the basis for the design of a general Eye-Dependent QoL (EyeDQoL) measure; the EyeDQoL is for use with people who have one or more of a range of eye conditions [21].

Conclusion

The RetDQoL is a valid and reliable individualized measure of QoL for use with people with diabetic retinopathy with or without macular edema.

  • 1Diabetic retinopathy has a strong negative impact on QoL as shown by the overview items and the AWI; feelings about the future are most negatively impacted. In some instances, the overview items could be used as substitutes for the complete measure.
  • 2The RetDQoL domain-specific items form a highly reliable and robust scale. Difficulties with recognizing people may need further attention.
  • 3The high internal consistency allows for the measure to be used both at individual and group levels; stability of scores over time needs to be assessed in future longitudinal studies.
  • 4The measure is sensitive to visual acuity, stage of retinopathy, and occurrence of macular edema.
  • 5QoL is correlated significantly, but not strongly, with health status and treatment satisfaction, but results confirm that these constructs differ.
  • 6The RetDQoL may usefully be shortened further if present findings are confirmed cross-culturally and in other samples.

Acknowledgments

Eli Lilly and Company Limited provided data that were collected by Eli Lilly and their affiliates in Germany and some funding for the analyses reported here. We would particularly like to acknowledge the contribution of Jessamy Watkins and Michael Happich at Eli Lilly who made data and information available and reviewed drafts of this article.

We thank the participants, ophthalmologists, and other investigators who provided data for these analyses.

Access to Questionnaires

The copyright holder of the RetDQoL is Professor Clare Bradley, Health Psychology Research, Department of Psychology, Royal Holloway, University of London, Egham Hill, Egham, Surrey, TW20 0EX, UK; c.bradley@rhul.ac.uk. Visit http://www.healthpsychologyresearch.com for information about language versions available and access to questionnaires.

Source of financial support: Eli Lilly and Company Limited funded the DIMICO study and contributed to the funding of the analyses reported here. The first author, L.B., is supported by a studentship grant from Health Psychology Research Ltd, which has a licence from the senior author, C.B., to sublicense the questionnaires.

Ancillary