Gender disparity and the impact of COVID-19 on surgical training in New Zealand ophthalmology

The COVID-19 pandemic profoundly disrupted healthcare systems worldwide, including postgraduate medical education and training.^1,2 In New Zealand strict measures implemented in late March 2020 to control the virus, such as mandatory self-isolation, border closures and lockdowns, led to the cancellation of elective surgeries and the reprioritisation of healthcare resources.^3,4 The pandemic had a significant impact on surgical training programmes, including ophthalmology, where trainees rely heavily on hands-on experience to develop essential surgical skills.^5–7

While previous research has explored the broader effects of the pandemic on surgical education, including the increased use of virtual learning platforms and simulation training, potential gender disparities in the impact on surgical training remain under-explored.⁸ Studies in other surgical specialties have suggested that female trainees may face unique challenges and biases in the allocation of surgical opportunities, potentially exacerbated during times of disruption.^9,10 In addition, the influence of geographic location on training experiences during such disruptions requires further examination. Differences in patient demographics, case mix and access to resources between urban and provincial centres may result in varied impacts on surgical training.

This study analyses New Zealand ophthalmology trainee logbook data from 2017 to 2022 to assess the impact of the COVID-19 pandemic on surgical training volume, case mix and trainee involvement, with a focus on identifying any gender disparities and differences between urban and provincial training locations. Understanding these nuanced effects is crucial to ensuring equitable and resilient surgical training programmes in the face of future healthcare disruptions.

Methods

This study received approval from the Auckland Health Research Ethics Committee (AHREC reference: AH24820). The study involved retrospective longitudinal analysis of all de-identified New Zealand based ophthalmology trainee logbook data from 1 January 2017 to 31 December 2022 provided by the Royal Australian and New Zealand College of Ophthalmologists (RANZCO). The dataset contained information on all surgical procedures recorded by trainees including name of procedure, the trainee’s gender and location (urban or provincial) as well as the trainee’s role in the procedure (observed, assisted or performed). All logged procedures regardless of whether the trainee observed, assisted in or performed the procedure were defined as “trainee-involved” surgery while cases specifically performed by the trainee were defined as “trainee-performed” cases. Surgical procedures were further categorised for analysis by surgery type into seven categories: anterior segment, cataract, glaucoma, oculoplastic, vitreoretinal, strabismus, and other.

Data were examined for any variations in surgical volume, case complexity and case mix during this time frame, with specific attention given to differences between pre-pandemic (1 January 2017 to 31 December 2019) and pandemic data (1 January 2020 to 31 December 2022). Sub-group analysis was conducted based on gender and geographic location to identify groups that may have experienced a disproportionate impact on their surgical training due to the COVID-19 pandemic. Surgical numbers were normalised for trainee clinical full-time equivalents (FTE) per year. Research and non-clinical FTE were not included in FTE calculations. Annual FTE calculations involved simple multiplication of each trainee’s clinical FTE fraction by the proportion of the year spent in that role. For example, a female trainee in a 0.6 FTE role in an urban centre for the first 4 months of the year then moves to a 1.0 FTE role in a provincial centre for the remaining 8 months of the year. This would be accounted for as a total of 0.87 female FTE for the year (consisting of 0.67 provincial FTE [1x8/12] and 0.2 urban FTE [0.6x4/12]). Clinical FTE normalisation was used to control for differences in total surgical numbers that could be explained purely by differences in group size and number of theatre sessions; thus, more accurately reflecting differences in surgical opportunities for trainees in different groups.

Initially, Levene’s test for homogeneity of variance and the Shapiro–Wilk test for normality were performed. Analysis of variance (ANOVA) testing and Cohen’s effect size testing, where appropriate, were performed in cases where homogeneity and normality criteria were satisfied. For the ANOVA and related procedures, the primary response variable was the mean number of surgeries per FTE per year, calculated for each of the 6 years of the study period across the relevant subgroups (gender and location). To account for potential non-independence of data across the study period, a linear mixed-effects model with “procedure year” as a random effect was also fitted. Post hoc analysis of significant interactions was conducted using estimated marginal means. The results were consistent with the primary linear model, justifying its use.

For analysis of data which deviated from normality (Shapiro–Wilk test, p=0.01), a non-parametric bootstrap analysis was employed to ensure robust confidence intervals. The procedure involved resampling the observed data with replacement across 1000 iterations to generate a distribution of the mean difference in surgery volumes between the pre-pandemic and pandemic periods. The 95% confidence intervals were then derived from this distribution using the percentile method.

Statistical analysis was completed using the R statistical software package.¹¹

Results

During the study period there were 41,370 total trainee-involved surgeries logged and 190.7 trainee FTE-years (216.93 surgeries/FTE/year). The distribution of FTE for male and female trainees in urban and provincial placement types during pre-pandemic and pandemic periods is outlined in Table 1. The data indicates a relatively balanced distribution of trainees, suggesting that differences in FTE are not the primary driver of the observed outcomes.

View Table 1–4, Figure 1–2.

Levene’s test and the Shapiro–Wilk testing confirmed that the annual FTE-adjusted data met the assumptions of homogeneity and normality (p>0.05) when stratified by trainee gender and placement type, allowing for parametric testing. Subsequent two-way ANOVA testing of FTE-adjusted data demonstrated significant main effects for both training location and gender. Over the entire study period, provincial trainees had significantly higher surgical volumes per FTE than urban trainees for both total trainee-involved (+105.4%, p<0.001, Cohen’s d=3.55) and trainee-performed surgeries (+94.3%, p<0.001, Cohen’s d=2.876). Regarding gender, while there was no significant difference in trainee-involved surgeries per FTE (p=0.141), male trainees performed 17.7% more surgeries than their female counterparts over the entire study period which was statistically significant (p=0.022) with a modest effect size (Cohen’s d=-1.215). These overview and FTE-adjusted figures are displayed in Table 2.

The impact of the pandemic on surgical volumes was then examined with a linear mixed-effects model. While the volume of total trainee-involved surgeries remained stable, the number of trainee-performed surgeries per FTE decreased significantly by 10.84% (p=0.05) showing that the pandemic was associated with a significant reduction in trainee-performed ophthalmic surgery in New Zealand. Trainee-involved surgeries per FTE decreased by 11.6% for female trainees and increased by 7.2% for male trainees during the pandemic period, though this difference was not statistically significant (p=0.157). However, trainee-performed surgeries per FTE decreased by 24.94% for female trainees and increased by 0.74% for male trainees during the pandemic and this difference was statistically significant (p=0.045). Pre-pandemic and pandemic period figures are detailed in Table 3.

Further sub-group analysis was performed to ascertain if specific trainee subgroups were disproportionately affected. This demonstrated a statistically significant two-way interaction between gender and placement type (p=0.018) where the only placement-gender subgroup showing a substantial change during the pandemic was urban female trainees, who experienced a reduction of 28.03 performed surgeries/FTE/year, a trend that approached statistical significance (p=0.079). In contrast, the surgical volumes for all other placement-gender subgroups remained stable, with small, non-significant increases observed for urban male trainees (+4.91 surgeries/FTE/year, p=0.738), provincial female trainees (+3.23 surgeries/FTE/year, p=0.826) and provincial male trainees (+7.73 surgeries/FTE/year, p=0.600). These trends are evident in the annual FTE-adjusted surgical volumes illustrated in Figure 1.

Additional analysis of the subspecialty case mix revealed that cataract surgery was the most common trainee-involved surgery (59.7%), followed by oculoplastic (12.8%), vitreoretinal (9.2%) and anterior segment (7.9%) surgeries. As shown in Figure 2, cataract surgery is more prominent in provincial trainee-involved caseloads while other surgeries (particularly oculoplastic and vitreoretinal surgeries) comprise a greater proportion of urban trainee-involved caseloads.

Further analysis of trainee-involved surgeries by subspecialty during the pre-pandemic and pandemic periods was conducted. Levene’s test confirmed homogeneity of variance for all subspecialties (p>0.05); however, the Shapiro–Wilk test revealed significant deviations from normality (p=0.01). As a result, non-parametric bootstrap analysis was employed, with statistical significance defined as a 95% confidence interval (CI) that did not include zero.¹² The results are outlined in Table 4 below. Of note, a 27.6% rise in trainee-involved glaucoma surgeries/FTE/year in the pandemic period compared to the pre-pandemic period was noted and was statistically significant (95% CI: 0.03–3.67). Concurrently, a 20.8% reduction in trainee-involved oculoplastic surgery was noted and was also statistically significant (95% CI: -9.49–-1.48). No statistically significant differences between mean number of trainee-involved surgeries/FTE/year between pre-pandemic and pandemic periods were noted for other subspecialties. Data for cataract surgery, the largest subspecialty in the dataset, were further analysed, specifically to investigate for changes in the number of trainee-performed cataract surgeries/FTE/year during the pandemic. Although the mean number of trainee-performed cataract surgeries/FTE/year was 14.0% lower during the pandemic (83.7 surgeries/FTE/year) compared to pre-pandemic (97.3 surgeries/FTE/year), this difference was not statistically significant by bootstrap analysis (95% CI: -24.1–0.84).

Discussion

The COVID-19 pandemic has had a lasting effect on healthcare and will continue to influence medical practice.^8,13 It has also changed modern ophthalmic training and education. The missed learning opportunities resulting from cancelled surgeries prompted various educational advancements, including a significant increase in virtual resources such as Surgical Education and Training (SET) modules, virtual case discussions via video conferencing, and live-streamed surgeries.¹⁴ Our study examined the volume and types of surgical cases trainees were exposed to and their level of involvement in cases in both provincial and urban centres in New Zealand and analysed for changes between the pre-pandemic and pandemic periods.

The analysis of surgical logbook data demonstrated that the number of total trainee-involved surgeries normalised for FTEs did not significantly differ between pre-pandemic and pandemic years. However, analysis of trainee-performed surgeries showed a statistically significant 10.84% reduction in surgeries actually performed by all trainees during the pandemic, highlighting a clear impact on direct trainee involvement in surgical procedures. This trend suggests that although trainees are still exposed to a similar number of cases as they were before COVID-19, a greater proportion of this exposure was in the role of observer or assistant rather than actively performing surgery.

Potential reasons for this include increasing service provision demands after extended theatre shutdowns, healthcare funding constraints and increasing outsourcing of cases deemed suitable for trainees to perform resulting in a greater proportion of cases being performed by consultant ophthalmologists.

Interestingly, our detailed analysis revealed this effect was highly specific, being almost exclusively felt by female trainees in urban centres. This subgroup was the only one to experience a notable reduction in performed surgeries (-28.03 surgeries/FTE/year), a trend that approached statistical significance, while their male counterparts in urban centres and all trainees in provincial settings were relatively unaffected. This finding suggests that access to theatre sessions may be similar between groups, but that female trainees in urban centres in particular spend comparatively more cases observing or assisting. Reasons for this may include institutional factors, implicit bias, behavioural differences and gender normative stereotypes which have previously been reported elsewhere.⁹ While self-reported logbook data may introduce reporting bias between genders (e.g. potential bias introduced by male trainees being more likely to record their role as primary surgeon than female trainees for cases in which they had similar levels of involvement), this is unlikely to fully explain the significant decrease observed, as any systemic gender-based differences in reporting would have been present prior to the pandemic and would not account for the specific decline in cases affecting urban female trainees during the pandemic period.

When examining differences between urban and provincial placements, provincial trainees logged significantly more surgeries, both total and performed, compared to their urban counterparts when normalised for FTE. This difference may be attributed to factors such as lower patient volume and fewer specialists in provincial settings, leading to greater opportunities for trainees to gain hands-on experience. Conversely, urban trainees may have faced more competition for surgical opportunities due to higher patient volumes, presence of senior fellows and a greater concentration of specialists, which could explain the differences in surgical exposure. These findings underscore the influence of placement type on the amount and quality of surgical training opportunities available to ophthalmology trainees and suggest that a period of provincial placement is likely beneficial for trainee surgical experience. This highlights that while a provincial placement is beneficial for acquiring surgical volume for all trainees, other factors within the urban training environment appear to be the primary drivers of the observed gender disparity.

Although the number of trainee-involved surgeries remained relatively stable, a significant shift in case mix was noted with a significant reduction in oculoplastic surgeries and a significant increase in glaucoma surgeries. It is unclear why this occurred but may be due to factors such as preferential substitution of oculoplastic surgeries for other surgical procedures on trainee lists such as cataracts and glaucoma surgery. Also, oculoplastic surgeries are more likely to be elective and require general anaesthesia, making them more susceptible to cancellation during periods of resource limitation compared to more urgent glaucoma surgery. The rise in glaucoma surgery may also be in part due to the rising popularity of minimally invasive glaucoma surgery (MIGS) or due to delayed diagnosis and treatment of glaucoma during the pandemic which necessitated greater rates of surgical intervention. Additionally, in a small workforce, changes in subspecialist consultant staff at particular training centres can dramatically alter the case mix available to trainees.

The impact of COVID-19 and the changes in surgical activity were not unique to RANZCO training during the pandemic. Multiple retrospective studies conducted in various countries have highlighted the significant disruption caused by the COVID-19 pandemic on clinical activities, impacting not only ophthalmology but also other surgical specialties. The availability of operative theatre opportunities was notably reduced as a result of multiple waves of cancellations of elective cases in various surgical fields.^15–17 However, despite these challenges, results of the current study suggest that ophthalmology services in New Zealand adapted reasonably well during this time.

It is important to note the limitations of this study. Potential variations in data logging practices between trainees could introduce inconsistencies in data interpretation. However, the recordings are likely to be generally accurate. Guidelines are available to trainees to ensure consistency, and trainees are motivated to capture every case to meet training requirements and demonstrate their comprehensive experience. Other limitations include the analysis of data on an annual rather than a monthly basis, which may not fully capture the nuances of changes over the COVID-19 period. Furthermore, the study cannot isolate the effects of COVID-19 from other effects such as healthcare budget changes, the impact of a growing population or the increased outsourcing of trainee-appropriate cases to the private sector. Additionally, our aggregated data did not allow us to track individual trainees through the training program. As the analysis was conducted on de-identified and aggregated FTE data provided by RANZCO, the exact number of unique trainees, overlap of trainees between the two periods, or variability in surgical volume between individual trainees could not be determined. A model accounting for repeated measures on individual participants would be more robust, though our mixed-effects model accounting for procedure year did not show a significant random effect.

Due to significant deviations from normality in the subspecialty data, identified by the Shapiro–Wilk test, a non-parametric bootstrap analysis with 1000 resamples was utilised to generate robust confidence intervals. This method was preferred over parametric tests such as ANOVA, which assume data normality, and the Mann–Whitney U test, which is less suitable for small sample sizes and does not provide confidence intervals to quantify the magnitude of any change. The bootstrap approach, which involves resampling the observed data, provided a more reliable and interpretable assessment of the changes in surgery volumes, thereby strengthening the validity of these findings despite the data’s limitations.¹²

In conclusion, this study identified that the COVID-19 pandemic was associated with the emergence of a significant gender disparity in surgical training in New Zealand, driven specifically by a reduction in surgical procedures performed by female trainees in urban centres. Our findings also confirm the substantial influence of geography on training, with provincial placements offering a much higher volume of surgical experience. Furthermore, the pandemic induced a tangible shift in case mix, with an increase in glaucoma and a decrease in oculoplastic surgeries. These results underscore the urgent need for training programs to develop strategies that ensure equitable and resilient surgical training pathways, particularly within urban environments.