Predicting Cataract Surgery Time Based on Preoperative Risk Assessment

Introduction

Cataract extraction is the most common operation performed in the United States (1). Accurate prediction of and control for variation in the procedure's duration may improve operating room scheduling, reduce the unit cost per surgery and waiting list times, and increase the number of operations performed (2-4). Predicting a cataract surgery's duration mainly depends on estimates for similar cases and on the surgeon's skill level (5). Prediction accuracy may be improved by applying statistical models to the assessment of surgery complexity made by the surgeon (6).

Muhtaseb et al (7) introduced a preoperative risk (PORS) assessment method based on analysis of intraoperative complications in 1441 cataract surgeries. They devised a scoring system placing patients into 1 of 4 risk groups according to anatomical and demographic parameters such as age, ametropia, previous vitrectomy, corneal scarring, pupil size, anterior chamber depth, pseudoexfoliation (PXF), cataract staging, and phacodonesis. This system enables uniform and objective risk assessment and better preparation of a patient's expectations.

As operating room (OR) time is an expensive resource that should be optimized, we aim to evaluate whether the PORS system, which predicts surgical complications, will also aid in predicting surgery time. We developed dedicated software (EyeDB) to conduct the semiautomatic calculation of a patient's PORS. In this study, we examine the feasibility of integrating the EyeDB into our daily practice and using it to predict surgery time for a patient based on individual risk and designated surgeon.

Methods

The study was conducted at the Edith Wolfson Medical Center, Holon, Israel, adhered to the tenets of the Declaration of Helsinki, and was approved by the local ethics committee. Patients were consecutively recruited from November 2013 to December 2014.

Risk scoring

The PORS were assessed at the preoperative examination for each patient. One point of risk was assigned to a patient with the following parameters: previous vitrectomy, corneal scarring, small pupil (<3 mm), shallow anterior chamber (<2.5 mm), age >88 years, high ametropia (>6 D or <-6D), posterior capsule plaque, or posterior polar cataract. Three points of risk were assigned to a patient with mature or brunescent cataract, phacodonesis, or PXF. Miscellaneous factors for increasing surgical risk were assessed and risk points were allocated as fit by the physician. For example, deep-set eyes contribute 1 risk point and α-antagonist use (risk of intraoperative floppy iris syndrome) contributes 2 risk points (Fig. 1).

OPEN IN VIEWER

Fig. 1

Surgery time by risk group. CI = confidence interval.

Risk score was calculated by accumulation of all risk points and patients were divided into low-risk group (PORS ≤2) and high-risk group (PORS >5). This categorization is based on ­intraoperative complications rate (low risk [4.3%] vs high risk [32.0%]) (7).

Results

Using the EyeDB, we calculated the PORS for 178 patients undergoing cataract surgery. Analyses were conducted on 150 surgeries after 28 teaching cases were excluded. Demographic data are summarized in Table I. Patients in the high-risk group were older (81.3 vs 73.1 years, p = 0.008) and had longer surgery (37.6 vs 19.6 minutes, p<0.001) (Fig. 1).

OPEN IN VIEWER

Table I

Patient demographics

	Cohort	Low risk	High risk	p a
No.		114	13
Age, y, mean ± SD (range)	74.0 ± 11.3 (37-93)	73.1 ± 10.5 (44-92)	81.3 ± 10.7 (55-92)	0.009
Male, %	48.4	50.0	61.5
OD, %	50.3	50.0	53.8
Risk score, mean ± SD (range)	1.45 ± 1.95 (0-9)	0.48 ± 0.70	6.23 ± 1.48	<0.001
Surgery time, min, mean ± SD (range)	22.0 ± 13.4 (6.7-109.1)	19.6 ± 9.3 (7-46)	37.6 ± 24.4 (12-109)	<0.001

a

Comparison between the low-risk and high-risk groups.

Although surgeries were conducted by 9 different surgeons (supplementary Fig. 1, available online at www.eur-j-ophthalmol.com), regression analysis showed that risk scores for all patients were positively correlated with surgery time (r = 0.30, p<0.001).

In order to further control for variance in surgeon skills, we calculated the mean surgery time for each surgeon and identified cases more than 1 SD from each surgeon's mean time. These cases (n = 18) representing the outliers for each surgeon's mean surgical time had more than twice the risk score of cases within 1 SD from the mean (3.16 vs 1.22, p<0.001).

In addition, we compared surgery length of the high-risk patients to the low-risk patients for the 4 surgeons with the larger number of cases (n = 70, n = 27, n = 19, n = 18). A difference of 61.0 minutes (p<0.001), 24.4 minutes (p<0.001), 12.3 minutes (p<0.001), and 18.0 minutes (p<0.001) could be demonstrated between high-risk and low-risk patients (Tab. II). Regression analysis for each surgeon demonstrated that PORS accounted for significant variability in OR time for 2 surgeons (29% and 51%) and a prediction equation for the OR time could be calculated for each surgeon (every increase in 1 risk point added 2.2 or 3.3 minutes to the OR time, respectively). For the 2 other surgeons, PORS accounted for a nonsignificant 5% (p = 0.07) and 3% (p = 0.46) of the variability in OR time (Tab. III). We observed complications in 4 cases (Tab. IV). These cases had a mean PORS of 3.2 and included 2 high-risk patients and 1 low-risk patient.

OPEN IN VIEWER

Table II

Comparison of low-risk and high-risk cases by surgeon

	Low risk	High risk	p
Surgeon 1
No.	56	2
Age	73.1	85.0	0.1
Risk score	0.47	7.00	<0.001
Surgery time, min	14.9	60.6	<0.001
Surgeon 2
No.	18	4
Age	72.6	74.2	0.8
Risk score	0.52	6.75	<0.001
Surgery time, min	13.3	24.1	0.001
Surgeon 3
No.	16	1
Age	70.8	90	0.1
Risk score	0.62	7.0	<0.001
Surgery time, min	32.6	54.0	0.005
Surgeon 4
No.	13	1
Age	78.2	91.0	0.01
Risk score	0.34	5.0	<0.001
Surgery time, min	31.4	41.6	0.2

OPEN IN VIEWER

Table III

The effect of risk score on the length of cataract surgery for 2 individual surgeons

Surgeon	Prediction equation for the OR time	R	R2	F	p
1	OR time = 1.6 (risk score) + 14.1	0.22	0.05	3.4	0.07
2	OR time = 2.2 (risk score) + 13.6	0.54	0.29	10.4	0.003
3	OR time = 3.3 (risk score) + 30.2	0.71	0.51	17.0	0.001
4	OR time = 0.8 (risk score) + 30.7	0.17	0.03	0.55	0.46

F  = the significance of using this model in predicting OR time, compared with a best guess; OR = operating room; R  = correlation coefficient between the predictor (risk score) and the outcome (OR time); R²  = percentage of variance in the outcome (OR time) explained by the predictor (risk score).

OPEN IN VIEWER

Table IV

Complications

Case	OR time	PORS	Complications
1	41.8	0	Posterior capsular tear, anterior vitrectomy, sulcus-IOL insertion
2	64.2	5	Zonular instability, dropped lens particle, anterior vitrectomy, AC-IOL insertion
3	67.7	3	Posterior capsular tear, anterior vitrectomy, AC-IOL insertion
4	109.1	5	Dropped lens, pars plana vitrectomy, and lensectomy

AC = anterior chamber; IOL = intraocular lens; OR = operating room; PORS = preoperative risk parameters.

Discussion

In this study, we implemented a stratifying risk system to assign preoperative risk to each patient and to calculate OR time using dedicated software. Although the PORS was initially applied to predict complications during cataract surgery, using the PORS may also allow better prediction of operating time as patients in the high-risk group had significantly longer surgeries. This work adds to the observation of Wright et al (9) that an experienced ophthalmic surgeon can predict, with high accuracy, the OR time based on case complexity. In addition, similar to the observation by Strum et al (10), we also noticed that surgeon skill predominated in determining OR time. Using the PORS, we were able to calculate the effect of every risk point on OR time for a specific surgeon, raising the possibility of using this tool for individual OR planning.

We have shown the feasibility of implementing dedicated software that could aid in improving operation time scheduling. The software is free to download and can assist individual surgeons with scheduling with a few steps of configuration.

Operating room time is a critical resource that generates high costs for a medical center (11). Therefore, optimally allocating OR time and planning for staff can maximize OR efficiency (12). As previously estimated, 1 minute of cataract surgery costs $11.24 (13); prediction of OR time may lead to improvement in surgical throughput (number of cases per unit time), reduced hospital cost, and improved patient care and satisfaction (14). Moreover, high-risk patients with pulmonary or orthopedic disorders could consider general anesthesia for longer surgery periods.

For 2 surgeons in our center, the PORS significantly contributed to the variability of OR time (29% and 51%), almost achieved a significant prediction for the third surgeon (p = 0.07), and was not helpful in prediction for the fourth surgeon. This may be due to the small number of cases but also to an inherent limitation of the PORS system, as risks are not fully characterized. For example, PXF on the pupil margin with dilated pupil diameter of 8 mm will receive 3 risk points but may not increase OR time as much as an extensive PXF on the lens capsule with zonular instability, which will receive the same number of risk points. Further amendments to the PORS system could be further incorporated in order to fine-tune this system. As demonstrated in Table IV, complications and longer OR time also were observed in not-high-risk patients, suggesting that the PORS cannot predict exactly how long each case will take. Hoverer, results from this study may serve as a basis for designing a surgery scheduling system based on the clinical skill of a given surgeon and patient individual risk.

In conclusion, this study examined predictors for cataract surgery time. Our results can be considered in surgery planning and preparation in order to allocate OR time considering cases with higher PORS.