Abstract
Purpose:
To quantify the association between visit adherence and visual acuity (VA) in diabetic macular edema (DME).
Methods:
This secondary analysis of the 2-year DRCRnet Protocol T study of 656 patients required one visit every 4 weeks in the first year, then at variable 4-16 week intervals in the second year. Visit adherence measured as: number of missed visits, average (avg days) and longest (max days) visit interval, average (avg missed days) and longest (max missed days) unintended visit interval, and visit constancy (percentage of 3-month periods with at least 1 visit). Avg and max missed days were categorized as on time(0 days), late(>0-60 days), and very late(>60 days). Primary outcome was change in ETDRS VA between baseline study visit and last attended visit, using multivariate linear regression models controlling for age, gender, race, ethnicity, treatment arm, baseline VA, hemoglobin A1c, insulin use and number of lasers and injections.
Results:
Mean number of missed visits was 1.7. 616(94%) patients had 100% visit constancy. 331(51%) patients were on time, 171(26%) late, and 154(23%) very late in avg missed days. Max missed days ranged 0-696 days. Adjusted, each missed visit was associated with 0.3 letter decrease(95%CI: −0.6, −0.1, p=0.02); being very late in avg and max missed days saw −4.2 letters(95%CI: −6.4, −2.0, p<0.001) and −4.0 letters (95%CI: −6.1, −1.9, p<0.001), respectively, than on time. Those that averaged >4 days missed per attended visit saw 4.6 letters worse(95%CI:−7.3, −2.0, p<0.001).
Conclusions:
Visit adherence is associated with visual acuity outcomes in DME patients.
Keywords: diabetic macular edema, adherence, Protocol T, DRCRnet, anti-VEGF
Introduction
Currently, patients receiving anti-vascular endothelial growth factor (anti-VEGF) treatment for diabetic macular edema (DME) attend considerably fewer visits, receive fewer injections and have much worse visual acuity (VA) outcomes when compared to their clinical trial counterparts [1–4]. While several factors are likely contributing to the reduced outcomes, visit adherence may play a significant role. Adherence in a medical setting can broadly be defined as the extent to which a patienťs behavior matches with medical advice, whether it is diet alterations, taking prescribed medications or attending scheduled office visits [5,6]. While visit adherence is a newer concept within ophthalmology, we recently reported on numerous visit adherence metrics showing worse adherence is associated with worse visual outcomes in neovascular age-related macular degeneration (nAMD)[7].
Although it has not always been labeled visit adherence, keeping diabetic patients engaged with their health care appointments has been a focus for decades. Specific to eye care, diabetics have been shown to have suboptimal eye care utilization, with only 50-65% of diabetic Medicare beneficiaries receiving an annual eye exam [8]. Within those already with diabetic retinopathy, early identification and treatment has been proven to be able to prevent at least 50% of severe vision loss [9], but of course, this is reliant on patients attending their visits. Unfortunately, patients with the most sight threatening forms of the disease, DME and proliferative diabetic retinopathy (PDR), have high rates of loss to follow up, ranging from 25-44% [10–12]. In PDR specifically, this has resulted in worse visual acuity outcomes in patients who receive anti-VEGF treatment and are lost to follow up for over a year compared to those who are treated with laser [13,14].
While it is clear that following up appropriately matters in caring for diabetic retinopathy, few attempts have been made to quantify visit adherence while also assessing its impact on visual acuity outcomes in DME. DME is the most frequent cause of vision loss in diabetic retinopathy and anti-VEGF injections have become the first-line treatment over focal/grid lasers or steroid injections [15,16]. Indeed, clinical trials indicate that 50% of DME patients treated with anti-VEGF should expect a 2-line gain and 30% a 3-line gain in visual acuity [17]. However, these results come at the cost of more frequent visits to the ophthalmologis’s office to receive regular anti-VEGF injections; therefore, unlike medication adherence in glycemic control where the burden is largely shifted to a patien’s residence, in DME, medication adherence to anti-VEGF therapy is largely predicated on office visits. To date, only a single-center study with 156 DME patients assessed a single binary visit adherence metric (a visit interval >56 days over the first year) and found nonadherence was more likely to lose 3 lines of vision [12]. To better elucidate the association between visit adherence and visual acuity outcomes in DME, we performed a secondary analysis of the multi-centered, randomized DRCRnet Protocol T study.
Methods
Data Source
Data were obtained from the publicly available data files of the DRCRnet Protocol T study [18]. The Protocol T study has been described elsewhere [19,20]. Briefly, DRCRnet Protocol T was a 2-year study evaluating the comparative effectiveness of aflibercept, bevacizumab, and ranibizumab in patients with diabetic macular edema. 660 subjects were enrolled from 89 clinical sites in the United States from August 2012 to October 2014. Institutional review board approval was not required for this secondary analysis as the data are freely available online. The study adhered to the tenets of the Declaration of Helsinki. Analysis occurred from April 2019 to November 2019. Patients were excluded from our analysis for not having a second visual acuity during the study more than 3 days after the baseline acuity taken (4 patients).
Measuring Visit Adherence
Several parameters were explored to measure visit adherence (Table 1). The study protocol required that after randomization, patients had follow up visits every 4 weeks (±7 days) in the first year, then at variable intervals of 4 to 16 weeks in the second year, depending on treatment response. Study coordinators indicated at each visit interval whether a visit was missed. Per study protocol, patients who missed a visit were then required to be seen within the next 4 weeks, and therefore, for our analysis, patients who were lost to follow up had every 4-week interval following a missed visit counted as an additional missed visit with the last visual acuity being carried forward. For patients who withdrew from the study, visit adherence and VA were assessed up until the final visit for which data were collected. For patients who died during the study period, visit adherence was evaluated up to the time of death and visual acuity at the last visit was carried forward.
Table 1.
Description of visit adherence metrics
| Metric | Description |
|---|---|
| Total number of missed visits | Tally of unattended visits in study period. If patients were listed as lost to follow up, every 4-week interval from the last attended visit was counted |
| Visit constancy | Percentage of 3-month periods (quarters) with at least 1 attended visit |
| Avg days | Average interval between attended visits (in days) |
| Max days | Maximum interval between attended visits (in days) |
| Avg missed days | Average unintended interval between attended visits (in days). This only includes visit intervals with at least one missed visit. |
| Max missed days | Maximum unintended interval between attended visits (in days). This reflects the longest single interval with at least one missed visit. |
| Miss index | Avg missed days / total number of attended visits (In missed days/visit) |
The first and most basic metric was the total number of missed visits. Another metric assessed was visit constancy, the tally of 3-month periods (quarters) with at least 1 visit attended, which has been proven to be associated with mortality and viral loads in patient with HIV [21,22]. We previously reported an association between visit constancy and visual acuity in nAMD [7], and so considered it worthy of assessment in this study. To account for patients who died or withdrew from the study before the 2-year period, visit constancy was only counted until the quarter at which the patient dropped from the study, and then reported as a percentage of total quarters. For analysis, visit constancy was tested as a binary variable of 100% visit constancy or less than 100% constancy, as only a small subset of patients had less than perfect visit constancy.
Next, the average (avg days) and longest single interval (max days) in days between visits were calculated. Avg and max days were categorized as monthly (28-35 days), long (36-60), and very long (>60 days) intervals. Since the protocol allowed for variable visit intervals up to 16 weeks in the second year, avg days and max days metrics were not necessarily reflective of intervals where visits were missed. To more directly account for this variability, two other metrics were created that evaluated the average unintended interval (avg missed days) and longest unintended interval (max missed days) in days that occurred between two completed visits surrounding a missed visit(s). Avg missed and max missed days were categorized as on time (0 missed days), late (>0-60 missed days), and very late (>60 missed days). Finally, since avg missed days metric does not account for the number of times a visit was missed, we created a last metric called the miss index which divided the avg missed days by the total number of attended visits, which is reported as number of missed days/ completed visit. The miss index was categorized as 0, between >0 and 4, and greater than 4 missed days/visit. (See Figure 1 for diagrammatic description of each metric).
Figure 1. Measuring visit adherence metrics in an example patient.
- Number of missed visits (×) is 4
- Visit constancy (percentage of 3-monthly intervals with at least 1 attended visit) is 100%
- Avg days between visits (calculated from the gray intervals) = 45.5 (classified as ‘long’)
- Max days between visits (the longest gray interval) = 112 (‘very long’)
- Avg missed days between visits (calculated from the striped intervals) = 74.7 (‘very late’)
- Max missed days between visits (the longest striped interval) = 84 (‘very late’)
- Miss index = 74.7 avg missed days / 17 completed visits = 4.4
Outcomes and Statistical Analysis
The primary outcome was change in Early Treatment Diabetic Retinopathy Study (ETDRS) VA between the baseline study visit and the last visit recorded for each patient. Linear univariate and multivariate regression models were used to analyze the association between all visit adherence parameters and change in VA. Covariates controlled for included age, sex, race, ethnicity, diabetes type, insulin use, baseline hemoglobin A1c, baseline VA, anti-VEGF agent (aflibercept, bevacizumab, ranibizumab), number of injections and number of lasers. Six patients did not have a baseline hemoglobin A1c recorded, so they were excluded from the adjusted model. Statistical analyses were conducted using SAS, version 9.4 (SAS Institute Inc., Cary, NC). All P values were 2-sided, with significance set at 0.05.
Results
656 patients from the Protocol T study had complete visit data and were included in this analysis. During the study period, 28 patients (4.3%) died, 31 patients (4.7%) withdrew from the study, and 32 patients (4.9%) were lost to follow up. Mean (SD) age was 60.0 (10.4) years, and 305 patients (46.5%) were women. 593 (90.4%) had type 2 diabetes. 331 patients (50.5%) did not miss a single visit; the mean (SD) number of missed visits was 1.7 (3.4). 616 patients (93.9%) had 100% visit constancy. The mean (SD) days between appointments was 35.0 (16.9) days, and the longest duration between 2 appointments ranged from 28 to 696 days (mean [SD] 97.0 [81.7] days). The mean (SD) interval of missed visits was 42.2 (72.9) missed days, with a mean (SD) miss index of 3.3 (13.2) missed days/visit. The longest interval of missed days ranged from 0 to 696 missed days (mean [SD] 51.2 (89.5) missed days) (Table 2).
Table 2.
Visit adherence in patients with diabetic macular edema in DRCRnet Protocol T
| Characteristic | No. (%) |
|---|---|
| Total No. | 656 |
| Death | 28 (4.3) |
| Withdrew from study | 31 (4.7) |
| Lost to follow up | 32 (4.9) |
| No. of missed visits | |
| Mean (SD) | 1.7 (3.4) |
| Median (IQR) | 0.0 (0.0-2.0) |
| Visit constancy | |
| <100% | 40 (6.1) |
| 100% | 616 (93.9) |
| Average interval between visits, days | |
| Mean (SD) | 35.0 (16.9) |
| Median (IQR) | 31.7 (28.9-36.4) |
| Maximum interval between visits, days | |
| Mean (SD) | 97.0 (81.7) |
| Median (IQR) | 77.0 (53.0-112.0) |
| Average interval of missed visits, days | |
| Mean (SD) | 42.2 (72.9) |
| Median (IQR) | 0.0 (0.0, 57.8) |
| Maximum interval of missed visits, days | |
| Mean (SD) | 52.1 (89.5) |
| Median (IQR) | 0.0 (0.0, 63.5) |
| Miss index, missed days/visit | |
| Mean (SD) | 3.3 (13.2) |
| Median (IQR) | 0.0 (0.0, 2.8) |
In univariate analysis, total number of missed visits was not significantly associated with change in visual acuity. However, when controlling for age, sex, race, ethnicity, diabetic status, baseline VA, insulin use, baseline hemoglobin A1c, anti-VEGF treatment, number of injections, and number of lasers in multivariate analysis, each missed visit was associated with −0.3 letter decline in final visual acuity after 2 years (95% CI: −0.6, −0.1, P=0.02). In terms of average days between visits, 471 patients (71.8%) were classified as having monthly intervals, 168 (25.6%) had long intervals, and 17 (2.6%) had very long intervals. These groups averaged +11.3, +10.2, and +11.5 letters from baseline vision, respectively. Maximum days between visits were segmented into 54 (8.2%) monthly, 197 (30.0%) long, and 405 (61.8%) very long intervals and achieved +10.9, +12.2, +10.5 letters in final vision, respectively. There was no significant relationship between avg and max days and visual acuity in either unadjusted or adjusted analysis (p≥0.34 for both comparisons). Similarly, no significant association was found between visit constancy and final visual acuity (p≥0.06for all comparisons). (Table 3 for all univariate and multivariate results.)
Table 3.
Change in visual acuity of the study eye and visit adherence
| Characteristic | N (%) | Mean (SD) VA change | Univariate analysis | Adjusted analysisb,c | ||
|---|---|---|---|---|---|---|
| VA Difference (95% CI) | P valuea | VA Difference (95% CI) | P valuea | |||
| Total number of missed visits | N/A | N/A | −0.3 (−0.5, 0.03) | 0.08 | −0.3 (−0.6, −0.1) | 0.02 |
| Visit constancy | ||||||
| 100% | 616 (93.9) | 11.3 (12.9) | REF | 0.09 | REF | 0.06 |
| < 100% | 40 (6.1) | 7.7 (11.2) | −3.6 (−7.7, 0.5) | −3.6 (−7.3, 0.1) | ||
| Average interval between visits, days | ||||||
| Monthly (28-35) | 471 (71.8) | 11.3 (13.6) | REF | 0.46 | REF | 0.55 |
| Long (>35-60) | 168 (25.6) | 10.2 (10.7) | −1.1 (−3.4, 1.1) | 0.9 (−1.4, 3.3) | ||
| Very long (>60) | 17 (2.6) | 11.5 (11.2) | −0.2 (−6.0, 6.4) | 0.2 (−5.7, 6.1) | ||
| Maximum interval between visits, days | ||||||
| Monthly (28-35) | 54 (8.2) | 10.9 (10.6) | REF | 0.34 | REF | 0.96 |
| Long (>35-60) | 197 (30.0) | 12.2 (14.5) | 1.2 (−2.7, 5.1) | 0.8 (−2.7, 4.2) | ||
| Very long (>60) | 405 (61.8) | 10.5 (12.3) | −0.4 (−4.1, 3.3) | 0.4 (−3.0, 3.8) | ||
| Average interval of missed visits, days | ||||||
| On time (0) | 331 (50.5) | 12.2 (12.4) | REF | 0.01 | REF | <0.001 |
| Late(>0-60) | 171 (26.1) | 10.6 (13.1) | −1.7 (−4.0, 0.7) | −2.1 (−4.1, 0.0) | ||
| Very late (>60) | 154 (23.4) | 9.1 (13.3) | −3.1 (−5.6, −0.7) | −4.2 (−6.4, −2.0) | ||
| Maximum interval of missed visits, days | ||||||
| On time (0) | 331 (50.5) | 12.2 (12.4) | REF | 0.01 | REF | <0.001 |
| Late(>0-60) | 142 (21.6) | 10.7 (13.2) | −1.5 (−4.1, 1.0) | −1.9 (−4.1, 0.3) | ||
| Very late (>60) | 183 (27.9) | 9.3 (13.2) | −3.0 (−5.3, −0.7) | −4.0 (−6.1, −1.9) | ||
| Miss index, missed days/visit | ||||||
| 0 | 331 (50.5) | 12.2 (12.4) | REF | <0.001 | REF | <0.001 |
| >0-4 | 224 (34.1) | 10.9 (13.7) | −1.3 (−3.4, 0.8) | −2.4 (−4.3, −0.5) | ||
| >4 | 101(15.4) | 7.7 (11.8) | −4.6 (−7.4, −1.7) | −4.6 (−7.3, −2.0) | ||
Linear trend p value
Adjusted by baseline VA, age, gender, race (modeled as white/black/other), ethnicity, anti-VEGF agent, diabetes type, insulin use, baseline hemoglobinA1c, number of lasers and number of injections
Baseline hemoglobinA1c is missing for 6 patients, so the final sample size for the adjusted model is 650
When considering average missed days, 331 patients (50.5%) were categorized as on time, 171 (26.1%) late, and 154 (23.4%) very late. Unadjusted, these groups averaged +12.2, +10.6, +9.1 letters in final VA, respectively. After controlling for covariates, the late group performed −2.1 letters worse (95% CI: −4.1, +0.01) and the very late group −4.2 letters worse (95% CI: −6.4, −2.0), compared to the on-time group (P<0.001). After accounting for total number of completed visits, a miss index of between >0-4 missed days/visit led to −2.4 letter vision loss (95% CI: −4.3, −0.5) and >4 missed days/visit led to −4.6 letter vision loss (95% CI: −7.3, −2.0), when compared to 0 missed days/visit (P<0.001). The longest unintended interval, max missed days, showed similar results to avg miss days. 331 (50.5%) patients were on time, 142 (21.6%) were late, and 183 (27.9%) were very late; these groups averaged +12.2, +10.7, +9.3 letters, respectively. In multivariate analysis, being late was associated with seeing 1.9 less letters (95% CI: −4.1, +0.3) and very late 4.0 less letters (95% CI: −6.1, −1.9), compared to being on time (P<0.001) (Table 2).
Discussion
In this study we assessed seven visit adherence metrics with their association to visual outcomes. We found multiple significant results, the largest of which was seen in the miss index metric, which divided the amount of time missed by the number of completed visits, lowering the score for those that attended more visits. Using this metric, we found that patients who missed on average 4 or more days for every visit attended saw 4.6 letters worse at the end of the study.
We previously reported an association of visit adherence with visual acuity outcomes in neovascular AMD [7]. In that study, we found a linkage between avg and max days and visual outcomes, an association that was not replicated in this study. Nor were the magnitudes of association as strong in this study (12 letters in the AMD study vs. 4.5 in this study). The critical factor in considering these discrepancies can be found in the natural course of DME and by proxy, the design of Protocol T, which allowed variable visit intervals in the second year. This protocol difference had major implications for this study. First, new onset neovascular AMD tends to be much more visually destructive than new onset DME over the first two years of the disease course. Next, per the Protocol T study design, each patient had individualized visit intervals making a comparison against a single standard visit interval (i.e. every 28 days as was seen in the AMD trial) less likely to show an association. To account for the protocol flexibility, the average and maximum day metrics were modified into average and maximum missed days, which allowed for only assessing intervals where an appointment was missed making the duration between visits unintended (compared to some of the intentionally prolonged visit intervals seen during year 2). The last way this protocol likely caused deviation from our AMD results was that those that did have a protocol-mandated prolonged interval, did so because they proved they were doing well without frequent treatment. This contrasts with those patients who had maintained a monthly schedule throughout the study and likely did so because they were not faring as well. Within this context, it is not surprising that metrics based on solely on days between visits, which were significant for AMD, would be inconclusive in this study. Interestingly, visit adherence in its simplest measure (tallying each missed visit) was significantly associated with worse visual outcomes in both AMD and DME.
As with our AMD study, one interpretation for our findings could argue that results are simply due to patients who followed up more receiving more injections and subsequently had better final acuity. Certainly, the more often a patient was seen implies that injections also occurred more often; however, our visit adherence metrics were associated with visual outcomes even after controlling for the number of injections and lasers each patient received. This suggests a more nuanced association between treatments and visits. One likelihood is that missed visits were missed opportunities to assess early disease activity, prior to lasting VA damage occurred. Clearly injections are an integral and mandatory step for maintaining the best VA; however, while the debate continues on whether monthly, treat and extend, or pro re nata injection protocols provide the best outcomes, based on these results, more emphasis should likely be placed on follow up in these DME patients.
As with every study, the results must be considered within the context of the study design. First, the dataset analyzed was that of a randomized clinical trial, not a typical clinical setting. Using a clinical trial protocol allowed for easily defined visit intervals and assessing when visits were missed. Second, visual acuity was measured by ETDRS letters, which is more sensitive in assessing differences less than 5 letters (1 Snellen line) of vision. Contrasting these benefits, however, is the fact that at least half of the study cohort did not miss a single visit in 2 years, and only 5% were lost to follow up. This is dramatically better visit adherence than would be expected in a typical clinical setting as noted previously from “real-world” studies [2,4,10,11,12]. While the reasons for visit nonadherence were not evaluated in this study, it is likely because clinical trial populations tend to be younger and healthier than 'real world' patients and are supported by study coordinators who follow up with patients to ensure visit completion. Cost of treatment is also less likely to be a prohibitive factor, since the clinical trial covered the cost of medications patients received in order to maintain masked treatment arms. Despite the relatively small N in the study and the low variability in visit adherence, significant associations were still elucidated between the adherence metrics and VA. Nevertheless, it is unclear whether these associations would remain in a real-world population, particularly in those with VA better than 20/40 which was not tested in Protocol T. Next, the analysis is also only limited to the 2 years of the study duration, and it is unknown if these associations would hold over longer follow up periods. Lastly, although our model controlled for significant confounding factors in visual acuity outcomes, including baseline VA, diabetic status, anti-VEGF treatment, number of injections and lasers, the association found with adherence may suggest, but should not imply, causation.
Overall, our results suggest that adherence with scheduled visits is associated with better final visual acuity. Complex relationships likely exist between the physician-patient relationship, clinical disease activity, and treatment efficacy affecting adherence and these need to be explored. Further work should expand on applying these metrics to real-world clinical practice. The results suggest that targeting visit adherence as a therapeutic strategy may improve patient outcomes.
Key Messages.
Visit adherence has been shown in medicine to play a a significant role in patient health outcomes. There are high rates of loss to follow up in diabetic macular edema patients. The impact on visual acuity outcomes has yet to be well characterized.
Missing scheduled office visits by more than 60 days on average is associated with 4 letter decline in final visual acuity compared to those who attended each visit.
Adherence to scheduled visits for anti-VEGF therapy is associated with better final visual acuity in diabetic macular edema
Acknowledgments
Funding: This work was supported by National Institutes of Health (1K23EY025729 – 01); University of Pennsylvania Core Grant for Vision Research (2P30EY001583); and block research grants to the Scheie Eye Institute from the Research to Prevent Blindness, the Paul and Evanina Mackall Foundation, and the Karen & Herbert Lotman Fund for Macular Vision Research Foundation.
Footnotes
Publisher's Disclaimer: This Author Accepted Manuscript is a PDF file of a an unedited peer-reviewed manuscript that has been accepted for publication but has not been copyedited or corrected. The official version of record that is published in the journal is kept up to date and so may therefore differ from this version.
Conflicts of interest/competing interests: The authors declare that they have no conflict of interest
Ethics approval: This study adhered to the tenets put forth in the 1964 Declaration of Helsinki. Institutional review board by the University of Pennsylvania was not required as it was conducted on freely available public data.
Consent to participate: Informed consent for this secondary analysis was not required as the data are deidentified and freely available online.
Consent for publication: Consent for publication for this secondary analysis was not required as the data are deidentified and freely available online.
Availability of data and material: Data are available in a public, open access repository: The DRCRnet Protocol T Public Dataset, accessible at https://public.jaeb.org/drcrnet/stdy/206
Code availability: Code used for data analysis can be provided by contacting the corresponding author.
References
- 1.Ciulla TA, Bracha P, Pollack J, Williams DF (2018) Real-world outcomes of anti-vascular endothelial growth factor therapy in diabetic macular edema in the United States. Ophthalmol Retina 2:1179–87 [DOI] [PubMed] [Google Scholar]
- 2.VanderBeek BL, Shah N, Parikh PC, Ma L (2016) Trends in the care of diabetic macular edema: analysis of a national cohort. PloS One 11: e0149450. DOI: 10.1371/journal.pone.0149450 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Kiss S, Liu Y, Brown J, Holekamp NM, Almony A, Campbell J, Kowalski JW (2014) Clinical utilization of anti-vascular endothelial growth-factor agents and patient monitoring in retinal vein occlusion and diabetic macular edema. Clin Ophthalmol 8:1611–21 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Holekamp NM, Campbell J, Almony A, Ingraham H, Marks S, Chandwani H, Cole AL, Kiss S (2018) Vision outcomes following anti-vascular endothelial growth factor treatment of diabetic macular edema in clinical practice. Am J Ophthalmol 191:83–91 [DOI] [PubMed] [Google Scholar]
- 5.Haynes RB, Taylor DW, Sackett DL (1979) Compliance in Health Care Johns Hopkins University Press, Baltimore, Maryland [Google Scholar]
- 6.Dunbar-Jacob J, Mortimer-Stephens MK (2001) Treatment adherence in chronic disease. J Clin Epidemiol 54:S57–60 [DOI] [PubMed] [Google Scholar]
- 7.Ramakrishnan MS, Yu Y, VanderBeek BL (2020) Association of visit adherence and visual acuity in patients with neovascular age-related macular degeneration. JAMA Ophthalmol 138:237–42 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Sloan FA, Yashkin AP, Chen Y (2014) Gaps in receipt of regular eye examinations among Medicare beneficiaries diagnosed with diabetes or chronic eye diseases. Ophthalmology 121:2452–60 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Early Treatment Diabetic Retinopathy Study Research Group (1991) Early photocoagulation for diabetic retinopathy. ETDRS report number 9. Ophthalmology 98:766–85 [PubMed] [Google Scholar]
- 10.Gao X, Obeid A, Aderman CA, Talcott KE, Ali FS, Adam MK, Rovner BW, Hyman L, Ho AC, Hsu J (2019) Loss to follow up after anti-vascular endothelial growth factor injections in patients with diabetic macular edema. Ophthalmol Retina 3:230–6 [DOI] [PubMed] [Google Scholar]
- 11.Best AL, Fajnkuchen F, Nghiem-Buffet S, Nghiem-Buffet S, Grenet T, Quentel G, Delahaye-Mazza C, Cohen SY, Giocanti-Auregan A (2018) Treatment efficacy and compliance in patients with diabetic macular edema treated with ranibizumab. J Ophthalmol DOI: 10.1155/2018/4610129 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Ehlken C, Helms M, Bohringer D (2018) Association of treatment adherence with real-life VA outcomes in AMD, DME, and BRVO patients. Clin Ophthalmol 12:13–20 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Obeid A, Gao X, Ali FS, Talcott KE, Aderman CM, Hyman L, Ho AC, Hsu J (2018) Loss to follow up in patients with proliferative diabetic retinopathy after panretinal photocoagulation or intravitreal anti-VEGF injections. Ophthalmology 125:1386–92 [DOI] [PubMed] [Google Scholar]
- 14.Wubben TJ, Johnson MW, Anti-VEGF Treatment Interruption Study Group (2019) Anti-vascular endothelial growth factor therapy for diabetic retinopathy: consequences of inadvertent treatment interruptions. Am J Ophthalmol 204:204–13 [DOI] [PubMed] [Google Scholar]
- 15.Ding J, Wong TY (2012) Current epidemiology of diabetic retinopathy and diabetic macular edema. Curr Diab Rep 12:346–54 [DOI] [PubMed] [Google Scholar]
- 16.Ho AC, Scott IU, Kim SJ, Brown GC, Brown MM, Ip MS, Recchia FM (2012) Anti-vascular endothelial growth factor pharmacotherapy for diabetic macular edema: a report by the American Academy of Ophthalmology. Ophthalmology 119:2179–88 [DOI] [PubMed] [Google Scholar]
- 17.Diabetic Retinopathy Clinical Research Network, Elman MJ, Aiello LP, Beck RM, Bressler NM, Bressler SB, Edwards AR, Ferris FL 3rd, Friedman SM, Glassman AR, Miller KM, Scott IU Stockdale CR, Sun JK (2010) Randomized trial evaluating ranibizumab plus prompt or deferred laser or triamcinolone plus prompt laser for diabetic macular edema. Ophthalmology 117:1064–77 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.DRCR Retina Network (2012) www.public.jaeb.org/drcrnet/stdy/206. Accessed 25 Apr 2019
- 19.Diabetic Retinopathy Clinical Research Network, Wells JA, Glassman AR, Ayala AR, Jampol LM, Aiello LP, Antoszyk AN, Arnold-Bush B, Baker CW, Bressler NM, Browning DJ, Elman MJ, Ferris FL, Friedman SM, Melia M, Pieramici DJ, Sun JK, Beck RW (2015) Aflibercept, bevacizumab, or ranibizumab for diabetic macular edema. N Engl J Med 372:1193–203 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Wells JA, Glassman AR, Ayala AR, Jampol LM, Bressler NM, Bressler SB, Brucker AJ, Ferris FL, Hampton GR, Jhaveri C, Melia M, Beck RW, Diabetic Retinopathy Clinical Research Network (2016) Aflibercept, bevacizumab, or ranibizumab for diabetic macular edema: two-year results from a comparative effectiveness randomized clinical trial. Ophthalmology 123:1351–9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Mugavero MJ, Lin HY, Willig JH, Westfall AO, Ulett KB, Routman JS, Abroms, Raper JL, Saag MS, Allison JJ (2009) Missed visits and mortality among patients establishing initial outpatient HIV treatment. Clin Infect Dis 48:248–56 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Tripathi A, Youmans E, Gibson JJ, Duffus WA (2011) The impact of retention in early HIV medical care on viro-immunological parameters and survival: a statewide study. AIDS Res Hum Retroviruses 27:751–8 [DOI] [PubMed] [Google Scholar]