Author Response: Aspirin Use and Risk of Age-Related Macular Degeneration

We are very appreciative that one of our studies, “Association Between Aspirin Use and Age-Related Macular Degeneration: A Meta-Analysis,” ¹ has been accepted by Investigative Ophthalmology and Visual Science, in which we detected nonsignificant association between aspirin use and age-related macular degeneration (AMD) but that aspirin marginally increases the risk of neovascular AMD, based on an analysis of 2 randomized controlled trials, 3 cohort studies, and 4 case-control studies. We thank Wang and Zhang ² for constructive opinions, and we will try our best to explain. 

• 1.  
  Due to limited space, we provide only the outline of our search strategy, as follows.

Medline (1950 to April 2013): PubMed Advanced Search Builder search items include the following terms ([macular degeneration OR age-related maculopathy OR macular dystrophy OR age-related macular degeneration OR drusen OR retinal degeneration OR choroidal neovascularization OR choroidal neovascularization (CNV) OR geographic atrophy] AND [aspirin OR acetylsalicylic acid OR acylpyrin OR anticoagulant OR nonsteroidal anti-inflammatory OR fibrinolytic OR antiplatelet OR platelet aggregation inhibitor OR cyclooxygenase inhibitor OR antipyretic OR NSAIDs]); search in “All fields.” 

Web of Science (1900 to April 2013): Basic Search included the search items ([macular degeneration OR age-related maculopathy OR macular dystrophy OR age-related macular degeneration OR drusen OR retinal degeneration OR choroidal neovascularization OR CNV OR geographic atrophy] AND [aspirin OR acetylsalicylic acid OR acylpyrin OR anticoagulant OR nonsteroidal anti-inflammatory OR fibrinolytic OR antiplatelet OR platelet aggregation inhibitor OR cyclooxygenase inhibitor OR antipyretic OR NSAIDs]); search in “Topic.” 

Cochrane Library (up to April 2013): Advanced Search included the search items ([macular degeneration OR age-related maculopathy OR macular dystrophy OR age-related macular degeneration OR drusen OR retinal degeneration OR choroidal neovascularization OR CNV OR geographic atrophy] AND [aspirin OR acetylsalicylic acid OR acylpyrin OR anticoagulant OR nonsteroidal anti-inflammatory OR fibrinolytic OR antiplatelet OR platelet aggregation inhibitor OR cyclooxygenase inhibitor OR antipyretic OR NSAIDs]); search in “Search All Text.” 

Embase (1980 to April 2013): Multi-Field Search covered the search items ([macular degeneration OR age-related maculopathy OR macular dystrophy OR age-related macular degeneration OR drusen OR retinal degeneration OR choroidal neovascularization OR CNV OR geographic atrophy] AND [aspirin OR acetylsalicylic acid OR acylpyrin OR anticoagulant OR nonsteroidal anti-inflammatory OR fibrinolytic OR antiplatelet OR platelet aggregation inhibitor OR cyclooxygenase inhibitor OR antipyretic OR NSAIDs]); search in “All Fields.” 

Language was restricted to English in the above 4 databases. 

• 2.  
  Due to limited space, we only displayed the summary scores of Downs and Black instrument. The scores in all 5 columns (reporting, external validity, bias, confounding, and power) can be seen in the Table.
• 3.  
  The exact mechanism of aspirin in AMD pathogenesis is unclear. The immediate and long-term effects of aspirin may differ among various stages and subtypes of AMD. Defining a person as an aspirin user who probably only took aspirin once or twice in a lifetime may sound arbitrary, but there is also no solid evidence to exclude these participants as eligible samples. Although some studies used “Never/Ever” to define aspirin exposure, they did have different standards to guarantee the validity. For instance, Douglas et al. ³ defined aspirin exposure based on the prescription records from General Practice Research Database (GPRD) of the United Kingdom (originally called the Value Added Medical Systems Research Bank, which was established in 1987 and has been used successfully to conduct numerous pharmacoepidemiologic studies ^4–6 ), as taken at least twice per week for at least 6 months before AMD was diagnosed; or as a drug used regularly in the study by DeAngelis et al. ⁷ ; use of antiplatelet therapy was defined as any dose of aspirin and doctor-prescribed medications taken by patients identified and coded using standard procedures (British National Formulae) in the study by Rudnicka et al. ⁸ In order to prevent misleading conclusions, we thought it was necessary to pool as many pertinent studies as possible.
• 4.  
  Generally the randomized controlled trial (RCT) is the most advanced study design, technically. Nevertheless, the two RCTs conducted by Christen et al. ^9,10 included in our meta-analysis recruited health care professionals who would likely be more conscientious about maintaining a healthy lifestyle, which would inevitably cause the so-called volunteer bias. This bias might dilute the association between aspirin use and AMD toward null, if any association existed. Conversely, in some of the observational studies, participants were from clinic populations and were likely to have other diseases, which would enhance the association. We have already mentioned this in the discussion section of the article.
• 5.  
  There is no reason to ignore these unadjusted or residual confounding factors. We have checked the outcomes of smoking-adjusted groups and smoking-unadjusted groups and found no significant differences. Most of the included studies have adjusted or matched risk factors like age, sex, smoking, alcohol intake, and treatment assignment. The study by Liew et al. ¹¹ had the longest follow-up, which covered risk factors such as smoking, cardiovascular diseases, blood pressure, and body mass index and also detected a significant association between aspirin use and neovascular AMD. Efforts were made to do subgroup analyses based on every factor, but we recognized it was impractical because some investigators would have less than 1 study. We can not deny the possibility of causal phenomenon at all. Further high-quality-designed studies are warranted to confirm our findings.
• 6.  
  We failed to evaluate the dose-response association due to insufficient data for aspirin dosage and different durations. Most of the studies included in our meta-analysis reported nonsignificant outcomes, except for the studies by Liew et al. ¹¹ and Klein et al. ¹² A nearly 2-fold increase in the incidence of neovascular AMD was detected in those 2 studies. Both of those studies had a long observational time period (no less than 10 years) and fully stratified AMD stage, suggesting cumulative dosage may be important in AMD pathogenesis.

Table

 

Quality Evaluation of Included Studies

Source (Publication Year, Country)	Design	Reporting	External Validity	Bias	Confounding	Power	Total
Liew et al.11 (2013 Australia)	Cohort	5	1	3	3	4	16
Klein et al.12 (2012 US)	Cohort	6	1	4	3	5	19
Rudnicka et al.8 (2010 UK)	Case-control	6	1	3	3	4	17
Christen et al.9 (2009 US)	RCT	7	1	6	5	5	24
Douglas et al.3 (2007 UK)	Case-control	6	0	2	2	5	15
Clemons et al.13 (2005 USA)	Cohort	7	1	2	3	4	17
DeAngelis et al.7 (2004 US)	Case-control	6	1	2	3	3	15
Christen et al.10 (2001 US)	RCT	8	1	4	5	5	23
Blumenkranz et al.14 (1986 US)	Case-control	5	1	3	2	4	15

 1. Reporting (9 items): which assessed whether the information provided in the paper was sufficient to allow a reader to make an unbiased assessment of the findings of the study. 2. External validity (3 items): which addressed the extent to which the findings from the study could be generalized to the population from which the study subjects were derived. 3. Bias (7 items): which addressed biases in the measurement of the intervention and the outcome. 4. Confounding (6 items): which addressed bias in the selection of study subjects. 5. Power (1 item): which attempted to assess whether the negative findings from a study could be due to chance. Answers were scored 0 or 1, except for one item in the Reporting subscale, which scored 0 to 2 and the single item on power, which was scored 0 to 5. The more scores a study gets, the higher quality the study has. RCT, randomized controlled trial.

We agree that further long-term population-based prospective studies with validated quantitative exposure ascertainment and use of standardized diagnostic methods and fully stratified grading criteria are warranted to confirm our findings. Thank you again for your pertinent opinions, which contributed to improving our previous work.