Purpose: To determine the longterm risk of pseudophakic retinal detachment (PRD) in a defined population undergoing standardized phacoemulsification surgery at a single centre.
Design: We carried out a register-based, retrospective, consecutive, uncontrolled study. All 6352 eyes that underwent cataract surgery at our institution during the years 1996–1998 were included in the study. The main outcome measure was the incidence of surgery for PRD.
Methods: Cataract-operated eyes were identified by a search in the local hospital registry. Eyes that subsequently underwent surgery for PRD were identified by a search in the Danish Patients Registry (LPR). Follow-up ended by 31 December 2003. Eyes in patients who died before that date were censored.
Results: The 8-year cumulated incidence of PRD after phacoemulsification was 0.93 per eye (95% CI 0.65–1.33). This is 8.77 (95% CI 7.12–10.72) times higher than expected in eyes that do not undergo cataract surgery. The incidence rate for retinal detachment was significantly higher in pseudophakic eyes than in the background population for at least 6 years after surgery. Patient mortality was slightly increased for men, but no different for women from that of the background population.
Conclusion: For eyes undergoing phacoemulsification surgery, the cumulated incidence of PRD continues to increase for at least 8 years after surgery. Comparison of the cumulated incidence of PRD between different studies requires equal lengths of follow-up.
The present study was undertaken to evaluate the longterm risk of PRD in a defined population undergoing standardized phacoemulsification surgery at a single centre. We have previously reported on the results of RD surgery in an incomplete sample of the PRD subjects included in the present study (Boberg-Ans et al. 2003).
Materials and Methods
The patients were sampled from the registry in the Eye Department, Copenhagen University Hospital, Herlev, Denmark. We included all eyes that underwent phacoemulsification surgery for cataract between 1 January 1996 and 31 December 1998 in this institution. We excluded eyes in which phacoemulsification was part of a posterior segment procedure and eyes with a prior record of RD.
The end-point of the study was surgery for rhegmatogenous RD performed in a hospital in Denmark between the time of phacoemulsification surgery and 31 December 2003. The end-point was assessed by accessing the Danish National Patient Registry (NPR), which, since 1995, has recorded all diagnoses and surgical procedures on patients treated in public hospitals in Denmark. It is a close-to-complete register and only patients operated outside Denmark could have escaped registration.
For all included eyes we accessed the Danish civil person registry to assess whether the patient had died before 1 January 2004 and, if so, to register the time of death. In patients who underwent bilateral cataract surgery during the study period, survival was defined from the time of the last operation.
The eye department in which the study was conducted has standardized its surgical approach to cataract surgery. In almost all cases operated within the study period, phacoemulsification was performed using two-handed, down-slope sculpting, nuclear cracking and phacoaspiration of the fragments using a Storz Premiere unit (Storz, now Bausch & Lomb, Rochester, NY, USA) with the standard setting as follows: energy 40%; aspiration 40 mmHg during sculpting and 120 mmHg with a 7-Hz pulse during nuclear aspiration. Irrigation/aspiration was used to polish the capsular bag if necessary. The infusion bottle was placed 40–95 cm above the eye. Very few patients underwent surgery by other techniques, but 65 had conventional ECCE. All IOLs used were approved by the US Food and Drug Administration and included the 809P polymethylmethacrylate (PMMA) lens (Pharmacia, now AMO, Santa Ana, CA, USA), the Hydroview H60M foldable lens (Storz, now Bausch & Lomb, Rochester, NY, USA), the SI-40 foldable silicone lens (Allergan, now AMO, Santa Ana, CA, USA), and the AcrySof MA30 (Alcon, Fort Worth, TX, USA).
Kaplan−Meier analysis was used for calculating the cumulative risk of RD in the population (Collett 2003). In the Kaplan−Meier analysis, and all subsequent analyses, a patient was censored by time of death or by end of follow-up (31 December 2003). The relative importance of age and sex as risk factors for pseudophakic RD were evaluated using Cox proportional hazard regression analysis (Collett 2003). For each year after phacoemulsification surgery we calculated the standardized incidence ratio (SIR), i.e. the ratio between the observed and expected number of RDs in that year. For each year, we used the SIR and its confidence intervals to evaluate whether more RDs had occurred than expected. The calculation and estimation of confidence intervals for SIR followed Sahai & Khurshid 1996). The expected number of RDs was calculated using the incidence density for RD obtained previously in the Danish population with a similar age profile to the cataract population (Norregaard et al. 1996). A similar calculation was used to calculate the standardized mortality ratio (SMR) between the observed and expected numbers of deaths in the population. The expected number of deaths was calculated from official Danish life tables for the years 1998–1999 (Danmarks Statistik 2005). SAS Version 9.01 was used for the calculations (SAS Institute, Cary, North Carolina, USA). Throughout the study, a significance level of 5% was used.
Internal review board or ethics committee approval was not required for this register-based study.
The material consists of 6352 operations performed in 5092 individuals. Bilateral cataract surgery was performed in 1260 patients during the study period. A total of 1782 (40%) of the operated patients were male. The age of the patients ranged from 2 years to 98 years; the age distribution is shown in Fig. 1. The median age at the time of surgery of patients who underwent surgery on only one eye during the study period was 76 years. The median age at the time of surgery on the second eye of patients who underwent surgery in two eyes during the study period was also 76 years; the interval between the two operations ranged from 1 to 1043 days, with a median of 252 days.
During the follow-up period, 1774 patients died and 43 patients underwent at least one surgical procedure for rhegmatogenous RD in the cataract-operated eye. One patient who was cataract-operated in both eyes subsequently underwent RD surgery in both eyes, bringing the total number of RD procedures performed on eyes included in this study to 44.
Figure 2 shows the Kaplan−Meier plot of the cumulated incidence of RD. The 95% confidence intervals of these estimates are listed in Table 1. It is noted that cumulated incidence continues to increase over the 8 years of follow-up. In order to compare our results with previous studies, Table 1 also shows the cumulated incidence of PRD calculated per patient. Table 2 shows the observed incidence of RD as it relates to the expected incidence as calculated from a previously reported Danish control group of phakic individuals with a similar age profile to that of our cataract patients (Norregaard et al. 1996). In this study the incidence rate of (phakic) RD was estimated to be 0.00029 RDs per person-year. In the present study we estimate incidence of RDs per eye-year, and have therefore used an expected incidence rate of 0.000145 (=0.00029/2) RDs per eye-year. Table 2 also shows the calculated SIR (i.e. the ratio between the observed and expected numbers of RDs). The SIR is significantly higher than unity in the years immediately after phacoemulsification surgery: years 1, 2, 3, 4 and 6 (p < 0.01). In the last 2 years, years 7 and 8, the SIR estimate is higher than unity, but not significantly so (p > 0.1). At 8 years postoperatively the ratio of the observed versus the expected cumulated number of PRDs was 8.77 (95% CI = 7.12–10.72).
Table 1. Kaplan-Meier estimates of the cumulated incidence of retinal detachment (RD) and their 95% confidence intervals.
Cumulated incidence of RD per eye (95% CI)
Cumulated incidence of RD per patient (95% CI)
Table 2. Standardized incidence ratio (SIR, i.e. the ratio between the observed and expected numbers of retinal detachment) in the years after phacoemulsification.
Figure 3 shows the empirical incidence rate, or empirical hazard, as the number of observed RDs in each year after phacoemulsification surgery divided by the number of eye-years at risk in that year. The calculation is based on the numbers from Table 3. There was no significant correlation between year after surgery and empirical hazard (Spearman's ρ = − 0.57, p = 0.14).
Table 3. Result of COX proportional hazard analysis. Relative risk (RR) of retinal detachment between different levels of age and sex.
The results of a Cox proportional hazard analysis of age and sex as risk factors for PRD are shown in Table 3. Patients in their ninth decade were used as a reference group, and we found a progressively increased risk of RD with decreasing age, with a more than 10-fold increased risk among patients younger than 60 years (Table 3). We also found that male sex was associated with a statistically significant 2.5-fold increased risk of PRD (Table 3).
In order to calculate the risk of PRD we assessed the survival in our population. From the time of surgery to the end of follow-up at 31 December 2003, 1774 patients died. Based on official Danish life expectancy figures for 1998−1999, we calculated the expected number of deaths, as well as the SMR in our population (Sahai & Khurshid 1996). The results of this calculation are shown in Table 4, and a Kaplan−Meier plot of patient survival is shown in Fig. 4. The mortality of the study population as a whole did not differ significantly from that of the background population. However, there was a slight but significant increased mortality among operated men in the population. Interestingly, we found patients operated in both eyes during the period had more than twice the chance of survival than the normal population. This was highly statistically significant and true for both men and women (Table 4, Fig. 4). It should be noted that, while the analysis of PRD was conducted per eye, the analysis of mortality was conducted per individual; therefore the number of eye-years in Table 2 is larger than the number of person years in Table 4.
Table 4. Standardized mortality rate (SMR, i.e. the ratio between expected deaths and observed deaths).
In the present study 25% of patients were operated in both eyes. We calculated the incidence of PRD per operated eye and per patient (Table 1). The incidence of PRD in our population is comparable with those found in other studies. Four-year cumulated PRD incidences per individual of 0.9% were found in Danish and US populations of ECCE-operated cases (Norregaard et al. 1996). In Minnesota, an 8-year cumulated incidence of 1.21% per individual was found in a mixed ECCE and phacoemulsification population (Rowe et al. 1999). In both these studies, the observed incidence of PRD was compared with that of a control population and it was found that the incidence rate of RD remained higher in the pseudophakic population than in the background population for at least 6–10 years after cataract surgery (Norregaard et al. 1996; Rowe et al. 1999). In the present study, we found that the incidence of PRD was significantly higher than expected 1, 2, 3, 4 and 6 years postoperatively (Table 2). More than 6 years postoperatively this incidence was no longer significantly elevated, but as we had complete follow-up for only 6 years, the number of operated eyes declined beyond the sixth postoperative year. There was no significant trend towards a decreased incidence with increased time after surgery. At 8 years postoperatively, cataract-operated eyes had an 8.7 times higher risk of RD than non-operated eyes. We do not know how much, if at all, the cumulated excess risk will continue to increase with longer follow-up. However, the result warrants caution when considering clear lens extraction in young individuals, as well, perhaps, as vitrectomy for floaters in young individuals due to the subsequent higher risk of cataract development.
The mortality of cataract patients has been reported to be increased, particularly for younger patients and diabetes patients (Ninn-Pedersen & Stenevi 1995; Knudsen et al. 1999; Christen et al. 2000; Wang et al. 2001; Reidy et al. 2002; Borger et al. 2003; Clemons et al. 2004). In our population this was only true for men and for patients undergoing surgery on only one eye within the study period (Table 4). Surprisingly, for patients who underwent bilateral cataract surgery in the study period, we found a marked increase in survival (Table 4, Fig. 4) The median age in our population at the time of surgery was 76 years and 89% of the population was 60 years or older. It has previously been found that older cataract patients and cataract patients with good living conditions do not have increased mortality rates (Ninn-Pedersen & Stenevi 1995; Christen et al. 2000). During the study period there were waiting lists for cataract surgery in Denmark, and we believe that the observed association between increased survival and bilateral cataract surgery is due to ‘healthy patient’ selection bias.
It is an advantage of the present study that the end-point was determined by use of the Danish Patient Registry. As RD surgery is not performed outside public hospitals in Denmark, only patients treated abroad will have escaped registration. The diagnoses of all PRD cases identified were verified by obtaining the patients' charts. It is, however, theoretically possible that erroneous coding of the PRD might have allowed some cases to escape identification in our study. Moreover, using the Danish Civil Person Registry allowed complete control of the survival status of the population. We believe that the use of the two registers allows close to 100% control of the study population, despite the long follow-up.
The study is limited by the data available in the registers employed. We did obtain patient charts in all cases where the recording seemed erroneous or ambiguous, as well as in all cases where surgery for a PRD was recorded. However, we did not systematically extract information from the patient's charts. Hence we were not able to estimate the significance of important risk factors such as axial length, complicating eye diseases, peroperative complications or YAG laser capsulotomy.
The authors thank the Hede Nielsens Fond and the Dandy Foundation for their generous support.