Diabetic retinopathy incidence, progression, and health-related quality of life from the ACCORD trial

Abstract

Objective:

We performed a retrospective cohort study to evaluate the association between health-related quality-of-life (HRQOL) measures and diabetic retinopathy (DR) progression among participants of the Action to Control Cardiovascular Risk in Diabetes (ACCORD) clinical trial and the ACCORD Eye Study.

Design:

Retrospective cohort

Participants:

ACCORD (n=6219) and ACCORD Eye (n=1687) participants with type 2 diabetes mellitus with a glycated hemoglobin level of at least 7.5%.

Methods:

HRQOL measures included the feeling thermometer (general health), Health Utilities Index Mark (HUI) 2 (aggregate score of sensation, mobility, cognition, self-care, emotion, pain, and fertility), HUI-3 (aggregate score of vision, hearing, speech, ambulation, dexterity, emotion, cognition, and pain), Patient Health Questionnaire-9 (PHQ-9) for depression, and Short Form-36, including the Aggregate Interference score derived from physical health and emotional problems (Brazier Index). Across all measures, a higher score indicated better HRQOL. Multivariate linear regression adjusted for baseline HRQOL scores as well as demographic and clinical covariates was used to compare the difference in final HRQOL scores after 4 years of follow-up between those who did and did not experience DR progression.

Main Outcome Measures:

DR development was self-reported by participants in the ACCORD study. DR progression in the ACCORD Eye study was defined as a ≥3-step person scale increase on the Early Treatment of Diabetic Retinopathy Study scale or progression to proliferative DR requiring laser photocoagulation or vitrectomy.

Results:

Among ACCORD participants, those self-reporting DR development had a significantly lower score on the EuroQoL feeling thermometer by 3.68 points [(−4.91, −2.45), p<0.001], HUI-2 by 0.02 points [(−0.03, −0.01), p=0.005], HUI-3 by 0.04 points [(−0.06, −0.02), p<0.001], and Brazier Index by 0.03 points [(−0.04, −0.01), p=0.001] compared with those who did not report developing DR. Among ACCORD Eye participants, those with DR progression had a significantly lower score on HUI-3 by 0.05 points [(−0.09, −0.01), p=0.022] compared with those who did not experience DR progression.

Conclusions:

DR development was associated with worse subjective general health, worse HUI-2, worse HUI-3, as well as more interference from physical health and depression. DR progression was only associated with a decrease in HUI-3 by year 4.

Table of Contents

Participants of the Action to Control Cardiovascular Risk in Diabetes trial with type 2 diabetes with diabetic retinopathy development or progression had worsened health-related quality of life over a 4-year follow-up period compared to those without development or progression. However, visual acuity, routinely used in clinical practice to monitor disease progression, was not associated with quality-of-life scores. This highlights the importance of patient-reported outcomes in understanding the impact of diabetic retinopathy on patients’ daily lives.

1. Introduction

Diabetic retinopathy (DR) represents one of the most common microvascular complications of diabetes mellitus, affecting over 9.6 million individuals in the United States based on an estimate provided by the US Centers for Disease Control and Prevention’s Vision and Eye Health Surveillance System.¹ Worldwide, DR is projected to impact 366 million by 2030.² This condition more frequently affects underrepresented populations, including African American and Hispanic/Latino groups.³ As DR progresses, it can lead to vision loss, and can manifest as blurry, double, or distorted vision, refractive error changes, and noticeable floaters in the visual field.⁴ The impact of these visual changes on quality of life (QOL) has been previously reported among various geographical regions and ethnic groups.

While visual acuity may be used in clinical and research settings to monitor the development, progression, and treatment response of DR, patient-reported outcomes are stronger measures of the impact of DR on activities of daily living.⁵ The National Eye Institute 25-item Visual Function Questionnaire (NEI VFQ-25) and Short Form Health Survey are commonly administered to assess vision-related QOL (VRQOL) and health-related QOL (HRQOL), respectively. Recently, Zayed et al. conducted a meta-analysis and systematic review of 93 original research studies investigating the association between DR and QOL.⁵ They confirmed the strong relationship between DR and worse QOL, additionally reporting that the impact of early DR can be better assessed with VRQOL scales than HRQOL scales.⁵ Notably, their analysis was limited by variation in DR grading between studies.

There is limited evidence regarding the association between DR progression and change in QOL over time. In a population-based cohort study in Finland, Wirkkala et al. reported that QOL related to sleeping, daily activities, discomfort, and sexual activity worsened over a 35-year follow-up period among participants with type 1 diabetes mellitus since childhood with proliferative DR (PDR) at the initial visit.⁶ However, their analysis was vulnerable to small-sample bias, with only 29 participants in the follow-up period despite 216 participants at baseline. Other studies have used change in QOL scores as a proxy for treatment effectiveness. Mitchell et al. reported that participants with diabetic macular edema treated with ranibizumab alone or ranibizumab combined with laser had a significantly higher increase in NEI VFQ-25 scores compared to those treated with laser monotherapy.⁷

The Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial included over 10,000 participants with type 2 diabetes mellitus followed over a 5-year study period.⁸ This randomized clinical trial established the importance of optimal glucose control, blood pressure control, and the use of fenofibrate in the treatment of diabetic retinopathy.⁸ The objective of the present study was to evaluate the association between HRQOL measures and DR incidence and progression among participants of the ACCORD clinical trial and its substudy, the ACCORD Eye Study, respectively.

2. Methods

This study is a post hoc analysis of the ACCORD (NCT00000620) and ACCORD Eye (NCT00542178) studies. These studies are registered on Clinicaltrials.gov and located at 77 clinical sites in the United States and Canada. The studies were approved by the Institutional Review Board of the National Heart, Lung, and Blood Institute (as well as the review boards of each clinical site) and adhere to the tenets of the Declaration of Helsinki. Written informed consent to participate in the research was obtained from all participants of the ACCORD and ACCORD Eye trials.

2.1. Study Population

The ACCORD multicenter study included 10,251 participants with type 2 diabetes mellitus with a glycated hemoglobin level of at least 7.5%.⁸ Participants in ACCORD had either established cardiovascular disease or known cardiovascular risk factors at baseline.⁸ The ACCORD Eye consisted of a subset of ACCORD participants without a history of PDR that had previously been treated with focal laser photocoagulation or vitrectomy.⁹

2.2. QOL Measurement and DR Assessment

Participants in the ACCORD study self-reported presence of DR with the following yes/no question: “Has the participant ever experienced any of the following vision problems?”, with vision problems including retinopathy and vision loss. Participants in ACCORD Eye study underwent clinical eye examinations and imaging including visual acuity measurement, dilated fundus exams, and color fundus photographs.⁹ Trained graders masked to the participants’ medical history graded the fundus photographs using the Early Treatment of Diabetic Retinopathy Study (ETDRS) severity scale for DR.⁹ Participants underwent these examinations, along with questionnaires for ocular surgery history, at annual study visits.⁹

HRQOL measures administered in the ACCORD study included the EuroQoL feeling thermometer (general health), Health Utilities Index Mark (HUI)-2 (aggregate score of sensation, mobility, cognition, self-care, emotion, pain, and fertility), HUI-3 (aggregate score of vision, hearing, speech, ambulation, dexterity, emotion, cognition, and pain), Patient Health Questionnaire-9 (PHQ-9) for depression, Short Form-36 (SF-36) aggregate mental health, SF-36 aggregate physical health, and the SF-36D Aggregate Interference score deriving from physical health and emotional problems (Brazier Index).¹⁰ The PHQ-9 and SF-36 were administered to a select subgroup of ACCORD participants. Across all measures, a higher score indicated better HRQOL. The sample size of participants completing each measure varies in part because of an HRQOL substudy that was conducted among a sample of 2053 participants who were asked to complete the SF-36 and PHQ-9.^10,11 In contrast, the EuroQOL feeling thermometer, HUI-2, and HUI-3 were administered to all participants as part of their physical exam. Additionally, missing covariate data further reduced the final sample size of all analyses. The EuroQOL feeling thermometer was administered at baseline, annually over 7 years, and at the exit visit. HUI-2 and HUI-3 were administered at baseline, at years 1, 3, and 4, and at the exit interview. The PHQ-9 and SF-36 were administered at baseline and during visits at years 1, 3, and 4. We used the HRQOL measures from baseline and at year 4 in these analyses.

2.3. Statistical Analysis

We compared the year 4 HRQOL score between participants with and without DR progression among ACCORD Eye participants. DR progression was defined as: (1) ≥3-step person scale increase on the ETDRS person scale or (2) progression to PDR requiring focal laser photocoagulation or vitrectomy. We also compared HRQOL scores at year 4 between participants with and without self-reported DR development among ACCORD participants. Each HRQOL was analyzed separately.

Stepwise and lasso regression were used to evaluate these associations. Covariates examined included: age, body mass index, year of diabetes diagnosis, low density lipoprotein, total cholesterol, hemoglobin A1c, visual acuity in the worse eye at baseline, glomerular filtration rate, sex (referring to biological characteristics associated with physiology), race, highest level of education, smoking status, aspirin use, history of clinical cardiovascular events, neuropathy/nerve problems, history of cataract removal or Yag laser for cataract capsule in either eye, history of coronary artery bypass graft, history of percutaneous transluminal coronary angioplasty, other revascularization procedure, protein in urine and the score at baseline. The model with the higher adjusted R² was used to select the subset of covariates. We reported estimates with 95% confidence intervals for each analysis, and p-values less than 0.05 were considered significant.

Pearson’s chi-squared tests were used to compare the rate of DR incidence among ACCORD participants who were and were not included in the analysis (due to drop out). Pearson’s chi-squared tests were also used to compare the distribution of DR severity among ACCORD Eye participants who were and were not included in the analysis (due to drop out). For ACCORD Eye participants we present contingency tables to evaluate the concordance between self-reported DR and DR severity from graded fundus photography. We evaluated the association between change in VA in the worse eye with DR progression among ACCORD Eye participants using multivariable regression analysis adjusted for age and sex. We used the Spearman correlation coefficient to evaluate the association between change in VA and change in HRQOL scores over time among ACCORD participants. Statistical analysis was conducted using SAS v. 9.4.

3. Results

3.1. Cohort demographics and HRQOL scores

ACCORD participant demographic, clinical, and ophthalmic characteristics are shown in Table 1. Characteristics for ACCORD Eye participants are shown in Supplementary Table 1. Table 1 and Supplementary Table 1 are based on the participants that were included in at least one HRQOL analysis. Table 2 shows the ranges for each HRQOL measures^12–14, minimal important differences,^12,15–19 as well as baseline and year 4 scores for ACCORD and ACCORD Eye participants. Of the 10,251 participants recruited in the ACCORD trial, 9,434 (92.0%) completed follow up through the end of the study.²⁰ Of the 10,251 participants recruited in the ACCORD trial, 3,472 were enrolled in the ACCORD Eye study. Of these 3,472 participants, 2,856 (86% of surviving participants) returned for their second eye examination and fundus photographs.⁹

Table 1.

Demographic, clinical, and ophthalmic characteristics of ACCORD participants included in at least one HRQOL analysis.

	N=6219
Age, mean ± SD, years	62.6 ± 6.6
Body mass index, mean ± SD, kg/m2	32.1 ± 5.4
Years of diabetes, mean ± SD	10.7 ± 7.5
Hemoglobin A1c, mean ± SD, %	8.3 ± 1.0
VA in the worse eye, mean ± SD, letters	69.8 ± 16.5
Glomerular filtration rate, mean ± SD, mL/min/m2	91.6 ± 28.5
Sex, no. (%)
Male	3815 (61.3)
Female	2404 (38.7)
Race, no. (%)
White	3910 (62.9)
Black	1135 (18.3)
Hispanic	413 (6.6)
Other	761 (12.2)
Education, no. (%)
Less than high school graduate	855 (13.7)
High school grad (or GED	1661 (26.7)
Some college or technical school	2054 (33.0)
College graduate or more	1649 (26.5)
Smoker, no. (%)
Yes	691 (11.1)
No	5528 (88.9)
Prior Cardiovascular Disease, no. (%)
Yes	2126 (34.2)
No	4093 (65.8)
Neuropathy, no. (%)
Yes	1657 (26.6)
No	4562 (73.4)
Cataract removal/Yag laser, no. (%)
Yes	715 (11.5)
No	5504 (88.5)
VA in the Worse Eye, no. (%)
20/20+	705 (11.3)
<20/20–20/40	3538 (56.9)
<20/40–20/200	1663 (26.7)
<20/200	313 (5.0)
Self-Reported DR at Year 4, no. (%)
Yes	577 (9.3)
No	5642 (90.7)

Abbreviations: ACCORD, Action to Control Cardiovascular Risk in Diabetes; DR, diabetic retinopathy; GED, general education diploma; VA, visual acuity.

Table 2.

HRQOL scores at baseline and year 4 among ACCORD and ACCORD Eye participants.

			ACCORD		ACCORD Eye
			Baseline	Year 4	Baseline	Year 4
	Range	Minimally Important Difference	N, Mean ± SD	N, Mean ± SD	N, Mean ± SD	N, Mean ± SD
EuroQOL Feeling Thermometer	0–100	7–10	6181, 76.3 ± 16.3	6163, 79.2 ± 15.3	1675, 75.8 ± 16.0	1675, 79.8 ± 15.4
HUI-2	−0.03–1.00	0.045	6175, 0.8 ± 0.2	6032, 0.8 ± 0.2	1675, 0.8 ± 0.2	1675, 0.8 ± 0.2
HUI-3	−0.36–1.00	0.032	6182, 0.7 ± 0.3	6040, 0.7 ± 0.3	1676, 0.7 ± 0.2	1649, 0.7 ± 0.3
PHQ-9	0–27	5	1106, 5.1 ± 4.7	1109, 4.5 ± 4.4	396, 5.2 ± 4.9	397, 4.4 ±4.5
SF-36 Mental Health Score	0–100	5	1096, 67.0 ± 22.8	1098, 66.4 ± 24.1	394, 66.8 ± 23.3	393, 66.1 ± 24.1
SF-36 Physical Health Score	0–100	5	1096, 50.5 ± 19.7	1098, 49.0 ± 20.4	394, 51.3 ± 20.1	393, 49.8 ± 20.2
SF-36D Brazier Index	0–1	0.041	1104, 0.7 ± 0.1	1110, 0.7 ± 0.1	395, 0.7 ± 0.1	397, 0.7 ± 0.1

Abbreviations: ACCORD, Action to Control Cardiovascular Risk in Diabetes; HRQOL, health-related quality-of-life; HUI, Health Utilities Index Mark; PDR, proliferative diabetic retinopathy; PHQ-9, Patient Health Questionnaire-9; SD, standard deviation. SF-36, Short Form-36.

3.2. Association Between HRQOL at Year 4 and DR Development/Progression

In ACCORD participants, 577 of 6219 participants (9.3%) reported developing DR over the 4-year follow-up period. Those who self-reported developing DR had a significantly lower score on the EuroQoL feeling thermometer by 3.68 points [(−4.91, −2.45), p<0.001], HUI-2 by 0.02 points [(−0.03, −0.01), p=0.005], HUI-3 by 0.04 points [(−0.06, −0.02), p<0.001], and Brazier Index by 0.03 points [(−0.04, −0.01), p=0.001] compared with those who did not report developing DR (Table 3 and Supplementary Table 2). Additionally, those who reported developing DR had a significantly higher PHQ-9 score by 0.94 points [(0.23, 1.64), p=0.009] compared with participants who did not report developing DR. There was no significant difference in the rate of self-reported DR at baseline between ACCORD participants who were and were not included in each of the HRQOL analyses at year 4 (Supplementary Table 3).

Table 3.

Regression estimates for the 4-year score in the health-related quality-of-life (HRQOL) scores between ACCORD participants who did and did not self-report DR development; and for ACCORD Eye participants who did and did not experience DR progression (≥3-step person scale increase on the Early Treatment of Diabetic Retinopathy Study scale or progression to proliferative DR requiring laser photocoagulation or vitrectomy). HRQOL measures with a significant regression estimate (p<0.05) are bolded.

	ACCORD				ACCORD Eye
HRQOL Measure	N	N (%) with DR Progressiona	Estimate [95% CI]	P-value	N	N (%) with DR Developmentb	Estimate [95% CI]	P-value
EuroQOL Feeling thermometer	6125	567 (9.3%)	−3.68j [−4.91, −2.45]	<0.001	1663	158 (9.5%)	−0.65c [−2.99, 1.69]	0.503
HUI-2	5995	560 (9.3%)	−0.02 k [−0.03, −0.01]	0.005	1636	154 (9.4%)	−0.01d [−0.03, 0.02]	0.452
HUI-3	6006	562 (9.4%)	−0.04 l [−0.06, −0.02]	<0.001	1639	154 (9.4%)	−0.05e [−0.09, −0.01]	0.022
PHQ-9	1065	118 (11.1%)	0.94 m [0.23, 1.64]	0.009	385	33 (8.6%)	−0.18f [−1.48, 1.12]	0.786
SF-36 Mental Health	1046	116 (11.1%)	−3.27 n [−7.28, 0.74]	0.110	379	33 (8.7%)	4.03g [−3.40, 11.46]	0.287
SF-36 Physical Health	1046	116 (11.1%)	−1.60o [−5.33, 2.12]	0.399	379	33 (8.7%)	−1.66h [−8.52, 5.20]	0.635
Brazier Index	1064	119 (11.2%)	−0.03 p [−0.04, −0.01]	0.001	384	33 (8.6%)	0.01i [−0.02, 0.03]	0.676

Abbreviations: ACCORD, Action to Control Cardiovascular Risk in Diabetes; BMI, body mass index; CI, confidence interval; DR, diabetic retinopathy; ETDRS, Early Treatment of Diabetic Retinopathy Study; GFR, glomerular filtration rate; HRQOL, health-related quality-of-life; HUI, Health Utilities Index Mark; PDR, proliferative diabetic retinopathy; PHQ-9, Patient Health Questionnaire-9; SF-36, Short Form-36; VA, visual acuity.

^aDR progression in the ACCORD Eye trial was defined as: (1) ≥3-step person scale increase on the ETDRS scale or (2) progression to PDR requiring focal laser photocoagulation or vitrectomy.

^bDR development in the ACCORD trial was self-reported.

^cAdjusted for years of diabetes and baseline EuroQOL feeling thermometer score

^dAdjusted for years of diabetes, highest level of education, BMI, smoking status, neuropathy, and baseline HUI-2 score

^eAdjusted for years of diabetes, BMI, and baseline HUI-3 score

^fAdjusted for cataract removal/Yag laser history, BMI, and baseline PHQ-9 score

^gAdjusted for baseline SF-36 mental health score

^hAdjusted for smoking status, neuropathy, and baseline SF-36 physical health score

ⁱAdjusted for highest level of education and baseline Brazier index

^jAdjusted for BMI, GFR, race, neuropathy, and baseline EuroQOL feeling thermometer score

^kAdjusted for age, BMI, years of diabetes, VA in the worse eye, sex, race, highest level of education, smoking status, clinical history of cardiovascular events, neuropathy, and baseline HUI-2 score

^lAdjusted for age, BMI, years of diabetes, hemoglobin A1c, neuropathy, and baseline HUI-3 score

^mAdjusted for BMI, VA in the worse eye, history of cataract removal/Yag laser, and baseline PHQ-9 score

ⁿAdjusted for baseline SF-36 mental health score

^oAdjusted for baseline SF-36 physical health score

^pAdjusted for baseline Brazier Index

Among ACCORD Eye participants, 160 of 1687 (9.5%) had DR progression over the 4-year follow-up period. Those with progression had a significantly lower score on HUI-3 by 0.05 points [(−0.09, −0.01), p=0.022] compared with those who did not experience DR progression (Table 3 and Supplementary Table 4). There were no other significant differences in HRQOL scores at year 4 between ACCORD Eye participants who did and did not experience DR progression. There was no significant difference in the distribution of DR presence/severity at baseline between ACCORD participants who were and were not included in each of the HRQOL analyses at year 4 (Supplementary Table 5).

3.3. Association Between Visual Acuity, HRQOL at Year 4, and DR Incidence/Progression

Among both ACCORD and ACCORD Eye participants, baseline VA was not significantly associated with the score at year 4 for any HRQOL measure (p>0.05), and therefore not selected as a covariate in any longitudinal regression model (Supplementary Tables 2 and 4). After adjusting for age and sex, the change in VA in the worse eye was significantly associated with self-reported DR development among ACCORD participants (p<0.001) but not with photo-graded DR progression among ACCORD Eye participants (p=0.058) (Supplementary Table 6). Among ACCORD participants, the change in VA was significantly associated with change in HUI-2 and HUI-3 (p<0.001 for both, Supplementary Table 7).

Table 4.

Concordance Between Self-Reported DR and Photo-Based Grading of DR Status Among ACCORD Eye Participants at Baseline and Year 4

	Self-Reported Presence of DR
Diabetic Retinopathy Presence/Severity, no. (%)	Baseline		Year 4
	Yes	No	Yes	No
No DR	14 (0.9)	1588 (99.1)	15 (2.1)	713 (97.9)
Mild DR	13 (1.8)	703 (98.2)	18 (3.6)	476 (96.4)
Moderate NPDR	84 (8.2)	941 (91.8)	76 (13.9)	472 (86.1)
Severe NPDR	0 (0)	10 (100)	0 (0)	0 (0)
PDR	3 (8.6)	32 (91.4)	20 (38.5)	32 (61.5)
Total	114	3274	129	1693

Abbreviations: ACCORD, Action to Control Cardiovascular Risk in Diabetes; DR, diabetic retinopathy; NPDR, nonproliferative diabetic retinopathy; PDR, proliferative diabetic retinopathy.

3.4. Concordance Between Self-Reported DR and photographically graded DR

Table 4 demonstrates the concordance between self-reported DR and graded fundus photos at baseline and year 4 among ACCORD Eye participants. At baseline, 1686 of 3388 participants (49.8%) stated they did not have DR despite a diagnosis of mild, moderate, or severe NPDR or PDR based on photo grading (Table 3). At year 4, 980 of 1822 participants (53.8%) stated they did not have DR despite having a diagnosis of mild or moderate NPDR, or PDR.

4. Discussion

Visual impairment can negatively influence physical, emotional, and social functioning, reducing patients’ quality of life.²¹ Quality-of-life measures are widely used in ophthalmic clinical trials, recognizing the importance of the participants’ experience in evaluating the effectiveness of diagnostic and treatment options.²¹ Although DR can be a progressive disease depending on blood glucose control, there exists a paucity of research evaluating change in QOL over time.

The objective of this study was to compare quality of life scores between patients with type 2 diabetes mellitus with versus without diabetic retinopathy development or progression in a post hoc analysis of the ACCORD and ACCORD Eye study over a 4-year follow-up period, respectively. We found that DR development (self-reported) was associated with worse subjective general health, worse HUI-3 (aggregate self-assessment of vision, hearing, speech, ambulation, dexterity, emotion, cognition, and pain), worse HUI-2 (aggregate self-assessment of aggregate score of sensation, mobility, cognition, self-care, emotion, pain, and fertility), as well as more interference from physical health and depression. We also found that DR progression (from fundus photography grading and report of treatment for PDR) (ACCORD Eye study) was associated with a decrease in HUI-3 score. To our knowledge, this is one of the most comprehensive longitudinal assessments evaluating the impact of DR on health-related quality of life. We also found that visual acuity at baseline was not associated with quality-of-life measures at year 4, demonstrating the weakness of this objective measurement in assessing participants’ well-being.

Our analysis of ACCORD participants found that participants with a self-reported DR development over four years had significantly lower scores on the feeling thermometer, HUI-2, HUI-3, and Brazier Index, after adjusting for demographic, clinical, and ophthalmic covariates as determined by the best-fit multivariable regression model. The feeling thermometer is a measure of general health status, and our results indicated that participants with DR development had a worse perception of their general health compared to those without DR. This is consistent with Mahobia et al., who reported significantly lower patient-reported general health among participants with DR compared to controls in a case-control study.²² To our knowledge, we are the first to report that participants with DR development had a significantly lower Brazier Index than those without DR development. This suggests that individuals with DR development may be perceiving higher levels of interference in their daily life from their physical and emotional problems and may benefit from counseling and other forms of psychosocial support.

Although ACCORD participants with self-reported DR development had a small difference in HUI-2 and HUI-3 scores at year 4 compared to those without DR, these differences were statistically significant and changes as low as 0.01 can be clinically meaningful.²³ Maddigan et al. reported on the association between diabetes and worse HUI-2, and our study further adds that worse HUI-2 scores are related to DR, a microvascular complication of diabetes.²⁴ Furthermore, existing literature (primarily cross-sectional) confirms a significant association between worse HUI-3 and DR.^25,26 HUI-3 includes a vision attribute, suggesting that DR development negatively impacts vision-related QOL. However, we were unable to analyze the association between DR development or progression and vision scores separately due to data availability issues. Future research could investigate the association between DR development and change in vision-related QOL over time in a longitudinal cohort study. Among ACCORD Eye participants, HUI-3 was the only HRQOL significantly associated with DR progression. While ACCORD participants self-reported DR development, ACCORD Eye participants received standardized eye examinations, which may explain the difference in results between analyses, as the eye examinations included fundus photography and standardized DR severity grading using the ETDRS scale. Additionally, individuals who self-report DR development may be likely to report worse QOL because of concerns regarding possible future vision loss.

In contrast to our expectations, ACCORD participants with self-reported DR development had a significantly higher PHQ-9 at year 4 compared with those without DR development. This also contradicts existing literature, which found that depression is more common among those with DR compared with diabetes only.²⁷ Valluru et al. also reported that patients with moderate to severe NPDR more commonly had depression compared to those with no DR, mild NPDR, or PDR.²⁸ Only 11.1% of ACCORD participants who completed the PHQ-9 self-reported DR development, and this small sample size likely limited our analysis. Our analysis of ACCORD participant data was also limited using self-reported DR status, rather than the use of standardized eye examinations.

Our analysis also showed that visual acuity at baseline was not significantly associated with any HRQOL measures despite visual acuity in the worse-seeing eye previously having been shown to have a better correlation with QOL measures than the visual acuity of the better-seeing eye.²⁹ Additionally, change in VA was not significantly associated with change in HRQOL measures over time among ACCORD participants (Supplementary Table 7). While change in HUI-2 and HUI-3 scores were significantly associated with change in VA, these analyses were unadjusted and subject to confounding. Furthermore, change in VA was not significantly associated with the distribution of DR severity in the ACCORD Eye analyses (which used fundus photos to grade severity) (Supplementary Table 6). Previous studies have reported inconsistent associations of visual acuity with various types of vision functioning: Chen et al. reported that visual acuity detected only 43% and 45% of post-panretinal photocoagulation participants with subnormal performance on visual function testing at the baseline and final visits;²⁹ in contrast, the Low Luminance Questionnaire, a validated tool designed to evaluate patient-reported functional impairments in dim lighting and at night, detected 81–90% of participants with abnormal visual function testing.²⁹ Overall, this indicates that visual acuity may not capture patients’ psychosocial distress and impairments in daily functioning that may be secondary to DR development and progression. Providers should instead rely on the results of subjective, validated surveys to understand the social and emotional impact of DR on their patients, as well as identify patients who may benefit from additional emotional or psychiatric management.²⁹ That being said, our study may have been inadequately powered to detect differences in HRQOL scores by VA, as the majority of included participants in both the ACCORD and ACCORD Eye analyses had relatively good vision (better than 20/40).

The strengths of this study lie in the comprehensiveness of the HRQOL tools administered to ACCORD and ACCORD Eye participants. With five validated surveys, we were able to assess physical, emotional, and social wellbeing in our analysis. Furthermore, our longitudinal analysis spanned four years, in contrast to existing literature that investigates the association between DR and QOL with cross-sectional study designs. Additionally, we utilized the stepwise and lasso regression techniques to determine the appropriate covariates for the analysis of the year 4 score of each QOL measure. However, there were also several limitations inherent to the post hoc study design. To begin with, ACCORD participants self-reported the presence of DR, and their responses were not confirmed by any record review. Therefore, the analysis of this cohort was susceptible to recall bias. When evaluating the concordance between self-reported DR and DR diagnosed by fundus photograph grading, over half of ACCORD Eye participants reported no DR despite presence of DR on photographic grading, demonstrating the limitations in accuracy of self-reported DR. This may also reflect the absence of subjective symptoms during the early stages of DR, when pathologic changes in the eye are not vision-threatening and apparent to the patient. Furthermore, less than 12% of participants in both the ACCORD and ACCORD Eye cohorts who completed each HRQOL measure experienced DR progression or development, making our analysis susceptible to small-sample bias. Also, approximately 8% of ACCORD participants and 14% of ACCORD Eye participants were lost to follow up and 3% died. While there was no significant difference in the DR rate/severity at baseline among ACCORD and ACCORD Eye participants included and excluded (due to drop out) from analyses at year 4, the number of participants completing the HRQOL measures at baseline is not same as the number of participants completing the HRQOL measures at year 4. This limits the conclusions that can be drawn about participants who were included relative to those who were excluded from analysis at year 4, and it is possible that those who were excluded had a significantly different DR rate/severity at baseline than those who were not excluded. Finally, some significant associations between DR progression/development and HRQOL scores at year 4 may not be clinically significant.

In conclusion, participants with type 2 diabetes mellitus with DR development or progression over a 4-year period had worsened health-related quality of life compared with those who remain stable over time. Furthermore, quality of life at year 4 was unrelated to visual acuity, demonstrating the importance of patient-reported outcomes in clinical research and practice.

Highlights.

• Diabetic retinopathy development was associated with worse quality-of-life scores

• Diabetic retinopathy progression was associated with worse quality-of-life scores

• Visual acuity was not associated with any quality-of-life measures