**3. Results**

#### *3.1. Search Results*

The PRISMA flow diagram is shown in Figure 1. A total of 67 studies were identified initially. After eliminating duplicated articles (*n* = 8), we removed non-relevant studies by screening titles and abstracts (*n* = 52). Then, a full-text review was performed. Conference abstracts were excluded (*n* = 1). Finally, six studies were enrolled in our meta-analysis [22–27].

**Figure 1.** Preferred reporting items for systemic reviews and meta-analyses (PRISM) flow diagram for searching and identifying included studies.

#### *3.2. Evaluation of the Quality of Included Studies*

Risk of bias for each study assessed by the ROBINS-I tool is presented in Table 1. The overall results showed that one study (Schargus) had low risk of bias, four studies (Yu, Shao, Zhou, and Xu) had moderate risk of bias, and one study (Ju) had severe risk of bias. None of them had critical risk of bias.


**Table 1.** Risk of bias assessment for the individual studies included in the meta-analysis.

D1 = Bias due to confounding; D2 = bias in selection of participants into the study; D3 = bias in classification of interventions; D4 = bias due to deviations from intended interventions; D5 = bias due to missing data; D6 = bias in measurement of outcomes; D7 = bias in selection of the reported result; Low = low risk of bias; Moderate = moderate risk of bias; and Severe = severe risk of bias.

#### *3.3. Characteristics of Included Studies*

The characteristics of the studies included in the meta-analysis are presented in Table 2. A total of 678 eyes from 611 patients were enrolled in six studies, with 359 eyes receiving FLACS and 319 eyes receiving MCS. Of the included studies, two were randomised controlled trials and four were prospective cohort studies. Five studies were conducted in China, whereas one study was performed in Germany. The mean age of the participants was 60 to 70 years in most studies.



Num= number; PCS= prospective cohort study; RCT= randomised controlled trial randomised control trial; FLACS= femto-second laser cataract surgery; MCS = manual cataract surgery; NS = nuclear sclerotic cataract; NR = not reported; and Phaco = phacoemulsification.

#### *3.4. Outcome Assessment of FLACS Group*

Table 3 presents the three parameters (OSDI, tear meniscus height, and Schirmer's test) at baseline and postoperative time points. Table 4 shows the values of the other three parameters (fluorescein staining, first tear breakup time, and average tear breakup time). The postoperative time points include day one, week one, one month, and three months.



cataract surgery; and NR = not reported.



The FLACS group pooled analyses comparing the postoperative and baseline values of the six parameters are presented in Figures 2 and 3. The overall SMDs showed increased values at postoperative day one in four of the six parameters (OSDI, tear meniscus height, Schirmer's test, and fluorescein staining). The increase was statistically significant in tear meniscus height (SMD: 0.456, 95% confidence interval (CI): 0.257 to 0.655), Schirmer's test (SMD: 0.132, 95% CI: 0.037 to 0.226) and fluorescein staining (SMD: 3.550, 95% CI: 0.354 to 6.747), but was not statistically significant in OSDI (SMD: 5.610, 95% CI: −2.191 to 13.411). Subsequently, tear meniscus height and Schirmer's test scores decreased to a level lower than baseline, while OSDI and fluorescein staining scores remained higher than baseline. TheSMDsofeachparameterhadatendencytowardzeroovertime.



**Figure 2.** *Cont*.


**Figure 2.** Overall effect of femtosecond laser-assisted cataract surgery (FLACS) on (**a**) ocular surface disease index (OSDI), (**b**) tear meniscus height, and (**c**) Schirmer's test. The square represents the standardised mean difference of each study. The size of square stands for the relative weight of each study. The lozenge represents the overall standardised mean difference.


**Figure 3.** *Cont*.



**Figure 3.** Overall effect of femtosecond laser-assisted cataract surgery (FLACS) on (**a**) fluorescein staining, (**b**) first tear breakup time (fBUT), and (**c**) average tear breakup time (avBUT). The square represents the standardised mean difference of each study. The size of square stands for the relative weight of each study. The lozenge represents the overall standardised mean difference.

Regarding the first and average tear breakup times, both had lower values than baseline from postoperative day one to the first month. The decreased values were only significant in the first tear breakup time at postoperative week one and the first month. The SMDs of the first and average tear breakup times trended towards zero with time. Finally, at postoperative three months, the six parameters were nearly similar to their baseline values except for tear meniscus height, which was significantly lower than at baseline (SMD: −0.172, 95% CI: −0.328 to −0.015).

#### *3.5. Outcome Assessment Comparing FLACS and MCS Group*

Figures 4 and 5 compare the postoperative change in six parameters between FLACS and MCS at various postoperative time points. The FLACS group had a higher reduction in tear meniscus height, Schirmer's test, fBUT, and avBUT. In addition, it had a higher increase in OSDI and fluorescent staining than the MCS group at every postoperative time point. In addition, the FLACS group showed less tear secretion postoperatively. However, most differences between FLACS and MCS were becoming less from postoperative day one to three months. Further, the differences were only significant at the following three time points: Schirmer's test at postoperative day one (SMD: −0.208, 95% CI: −0.397 to −0.020), one month (SMD: −0.309, 95% CI: −0.534 to −0.085), and first tear breakup time at postoperative week one (SMD: −0.685, 95% CI: −1.058 to −0.311).


**Figure 4.** *Cont*.


**Figure 4.** Comparison of (**a**) ocular surface disease index (OSDI), (**b**) tear meniscus height, and (**c**) Schirmer's test between the femtosecond laser-assisted cataract surgery (FLACS) group and manual cataract surgery (MCS) group. *I*2 represents heterogeneity. The square represents the standardised mean difference of each study. The size of square stands for the relative weight of each study. The lozenge represents the overall standardised mean difference.


**Figure 5.** *Cont*.


**Figure 5.** Comparison of (**a**) fluorescein staining, (**b**) first tear breakup time (fBUT), and (**c**) average tear breakup time (avBUT) between the femtosecond laser-assisted cataract surgery (FLACS) group and manual cataract surgery (MCS) group. *I*2 represents heterogeneity. The square represents the standardised mean difference of each study. The size of square stands for the relative weight of each study. The lozenge represents the overall standardised mean difference.

#### *3.6. Heterogeneity and Publication Bias*

Most analyses showed high between-study heterogeneity when evaluating the SMDs of six parameters (*I*2 > 75%). Concerning publication bias, Figure 6 demonstrates the funnel plots of studies regarding the post-FLACS effects. Regarding OSDI, tear meniscus height and Schirmer's test, the *p*-values of the Egger's test were 0.31, 0.94, and 0.65, respectively—revealing no significant publication biases. Significant publication biases were noted regarding post-FLACS effects corresponding to fluorescent staining, first tear breakup time, and average breakup time (all Egger's tests *p* < 0.01).

**Figure 6.** Funnel plots evaluating the publication biases regarding post-FLACS impacts on the six dry eye parameters (**a**) OSDI, (**b**) tear meniscus height, (**c**) Schirmer's test, (**d**) fluorescein staining, (**e**) fBUT, and (**f**) avBUT. The lozenge stands for overall standardised mean difference.

Funnel plots of the studies comparing postoperative effects between FLACS and MCS are presented in Figure 7. They exhibited no significant publication biases in all six parameters of dry eye symptoms/signs (all Egger's tests *p* > 0.1).

Since the publication bias is statistically significant regarding post-FLACS impacts on fluorescein staining, fBUT, and avBUT, we applied the trim-and-fill method to deal with the publication biases. After trimming the studies that caused a funnel plot's asymmetry and filling imputed missing studies in the funnel plot based on the bias-corrected overall estimate, the funnel plots were adjusted and are presented in Figure 8. The direction and significance of SMD did not change after adjusting the publication biases. Therefore, our previous statistical analyses regarding SMD were convincible.

**Figure 7.** Funnel plots evaluating the publication biases regarding the comparison between FLACS and MCS on the six dry eye parameters (**a**) OSDI, (**b**) tear meniscus height, (**c**) Schirmer's test, (**d**) fluorescein staining, (**e**) fBUT, and (**f**) avBUT. The lozenge stands for overall standardised mean difference.

**Figure 8.** Funnel plots after using trim-and-fill method to adjust the publication biases regarding post-FLACS impacts on the dry eye parameters (**a**) fluorescein staining, (**b**) fBUT, and (**c**) avBUT. The lozenge stands for overall standardised mean difference. The data points for imputed studies are highlighted in black.

## **4. Discussion**

This meta-analysis included six studies focusing on dry eye after FLACS. Six parameters (OSDI, tear meniscus height, Schirmer's test, fluorescein staining, first breakup time, and average breakup time) were used to evaluate dry eye symptoms/signs, which were also compared between FLACS and MCS groups. On postoperative day one, eyes receiving FLACS had transiently increased dry eye symptoms (OSDI) and tear secretion (tear meniscus height and Schirmer's test) but then decreased. Microscopic ocular surface damage (fluorescein staining) was significantly increased on postoperative day one and week one but improved after one month. Tear film instability (first breakup time and average breakup time) lasted for one month after surgery and then returned to the baseline level. Three months after surgery, only tear meniscus height was significantly decreased, while all the other parameters were similar to baseline. Compared with MCS, FLACS had a greater tendency towards dry eye in the early postoperative stage. However, the dry eye symptoms/signs between FLACS and MCS showed no significant differences three months after surgery.

This study is the first meta-analysis to compare the impact on postoperative dry eye between FLACS and MCS, to the best of our knowledge. In our study, a transient increase in tear secretion on postoperative day one may be related to surgical-induced pain. One possible explanation for the tear film instability presenting itself immediately after surgery is inflammation. Wound epithelial cells secrete inflammatory factors that accumulate in tears. The bandage of the eye decreases the tear removal rate and aggravates the inflammatory reaction, hyperosmolarity in tears, and subjective discomfort.

Regarding microscopic ocular surface damage, multiple reasons are responsible, including preoperative instillation of povidone-iodine and local anaesthesia [28,29], intraoperative irrigation, and light exposure [30]. Dry eye symptoms improved, but signs were worse at postoperative week one, implying more cytokines were released from the wound in order to induce inflammation. In addition, our study found that FLACS had a more severe effect on dry eye than MCS. This effect may be due to the suction ring in FLACS, injuring the limbal stem cells, conjunctival epithelium, and goblet cells. It is similar to the dry eye mechanism after laser-assisted in situ keratomileusis [31]. In addition, the extra laser procedure in FLACS leads to prolonged light exposure, thereby deteriorating tear film stability.

Fortunately, in our study, the symptoms/signs of dry eye immediately following FLACS almost returned to baseline within three months postoperatively. This result might be explained by the anti-inflammatory effects of postoperative eye drops. Previous studies have revealed that neuroregeneration occurs 25 days postoperatively [32], supporting our finding that postoperative dry eye tends to improve. Furthermore, the differences in dry eye parameters between FLACS and MCS mainly have no significant difference and have a decreasing trend. However, Yu et al. have found that FLACS causes more ocular surface damage than MCS in patients with pre-existing dry eye [22]. Therefore, preoperative screening and postoperative treatment for dry eye should be performed meticulously for those receiving FLACS with a pre-existing unhealthy ocular surface.

The main limitation of our meta-analysis is the heterogeneity among the included studies. The between-study variations may arise from differences in surgical machines, study protocols, inclusion criteria, and perioperative use of topical medication. Five of the six enrolled studies used the LenSx femtosecond laser system (Alcon Laboratories, Fort Worth, TX, USA). Only Schargus et al. used the CATALYS laser system (Johnson and Johnson, New Brunswick, NJ, USA). Different docking devices used in the laser platforms may cause different effects on the ocular surface [24,33]. Another limitation is that most included studies have a non-randomised design, increasing bias.

Moreover, the parameters used in our meta-analysis (OSDI, tear meniscus height, Schirmer's test, fluorescein staining, and tear breakup time) are not objective enough and are prone to observers' errors. Previous studies have suggested that tear film osmolarity and matrix metalloproteinase levels are more reliable dry eye tests and correlate well with dry eye severity [20,34,35]. In addition, meibomian gland dysfunction, lipid layer thickness,

inflammatory levels, and goblet cell densities also play an important role in dry eye [36,37]. These parameters should be assessed in further studies. Still another limitation is that we cannot perform subgroup analyses according to cataract grading or phacoemulsification time, which are relevant with post-operative dry eye. We have extracted data of cataract grading from three studies (Yu, Zhou, and Xu) and phacoemulsification time from two studies (Yu and Xu). However, the information was presented as overall proportion or mean, without mentioning the individual dry eye symptoms/signs corresponding to each category of cataract grading or phacoemulsification time. The lack of details and the too few study numbers makes subgroup analyses infeasible.

The strength of our study is that our results provide an evaluation of dry eye symptoms/signs following FLACS and include comparisons with those following MCS. Therefore, we could have a better understanding of postoperative dry eye risk. More comprehensive studies will need to be conducted, thereby supplying evidence for further meta-analyses.
