Exploring the paradigm of robotic surgery and its contribution to the growth of surgical volume.

Background: Robotic surgery is an appealing option for both surgeons and patients. The question around the introduction of new surgical technology, such as robotics, with the potential link to increased procedure-specific volume has not been addressed. We hypothesize that hospital adoption of robotic technology increases the total volume of specific procedures as compared to nonrobotic hospitals.
Methods: The 2010-2020 Florida Agency for Health Care Administration Inpatient database was queried for open, laparoscopic, and robotic colectomy, lobectomy, gastric bypass, and antireflux procedures. International Classification of Diseases, 9th and 10th Revisions, codes were used. Difference in difference method was used to evaluate the impact of robotics on total procedure-specific volume of robotic hospitals versus nonrobotic hospitals before and after adopting robotic technology. Incident rate ratios from the difference in difference analysis determined the significance of adding robotics. Patient demographics were evaluated using χ2 test.
Results: A total of 291,826 procedures were performed at 217 hospitals, 151 with robotic capabilities. Robotic hospitals experienced a 37% increase in surgical volume due to robotic technology (incident rate ratio 1.37, P < .05), which was consistent for each surgery except antireflux procedures (incident rate ratio 0.95). Robotic procedures had significantly higher charges for medical/surgical supplies; however, the mean length of stay for robotic procedures was significantly shorter than that of laparoscopic and open cases.
Conclusion: Hospital adoption of robotic technology significantly increases surgical volume for select procedures. Hospitals should consider the benefits of introducing robotic technology which leads to higher volume and decreased length of stay, benefitting both hospital systems and patients.

INTRODUCTION

The utilization of robotic surgery is growing; however, its impact on hospital systems and patient care is still being established. Research has that shown select robotic procedures (as compared to laparoscopic and/or open) carry a shorter length of stay (LOS) but higher cost [1]. The upfront cost of purchasing the robot and accoutrement can also not be ignored. One study suggested that the cost of adding a robot to a hospital was more than US$2.6 million [2].

Knowing that there is a high startup cost, hospital systems want to ensure a return on investment. As one could postulate that increasing surgical volume would increase revenue, our study sought to evaluate the effect of adding robotic technology to hospitals in terms of change in overall surgical volume. We hypothesized that the addition of robotic technology would increase hospital surgical volumes when studying select surgical procedures.

METHODS

This study was exempt from our institutional review board given that it was querying a deidentified database and did not contain HIPAA-protected information.

The 2010–2020 Florida Agency for Health Care Administration Inpatient database was queried for open, laparoscopic, and robotic colectomy, pulmonary lobectomy (lobectomy), gastric bypass, and antireflux surgeries [3]. These 4 procedures were chosen because they are very common operations performed in all 3 procedure types: open, laparoscopic (or thoracoscopic), and robotic. International Classification of Diseases, 9th and 10th Revisions (ICD9 and ICD10), codes were used to capture the 3 procedure types (open, laparoscopic, and robotic) using the primary procedure code. Open and laparoscopic procedures were coded based on their primary procedure code. The procedure was labeled "robotic" if a robotic qualifier appeared with the primary procedure code. A total of 257 procedure codes were used including the robotic qualifiers: 63 ICD9 codes (17 robotic qualifiers) and 192 ICD10 codes (234 robotic qualifiers; see Supplementary Material).

Patient demographics including sex, age, race, ethnicity, payer types, and Charlson comorbidity index (CCI) were studied. Stata software version 16 was used for all the data preparation steps and computing the descriptive statistics. R Studio was used to implement all the machine learning models using packages and libraries including "readstata13," "tableone," "MatchIT," "Matching," and "ICD." χ tests were performed to quantify the statistical significance among the 3 types of procedures and the descriptive categorical variables.

As our dataset did not include information on cost to hospital, we used the available data on charges to formulate a comparison between the different procedure types (open/laparoscopic/robotic). Analysis of similarities and differences between the procedure types (open/laparoscopic/robotic) for gross total charges, operating room charges, medical/surgical supply charges, anesthesia charges, and recovery room charges was completed with ANOVA with post hoc analysis; these analyses were risk-adjusted using CCI and separately by procedure type (open/laparoscopic/robotic). Data are presented as mean ± standard deviation.

Propensity score matching was used for a more comparative analysis between the 3 procedure types [4,5]. One-to-one matching was used to find the best and closest match for each robotic procedure in the open/laparoscopic groups based on age, sex, and CCI. One-to-one matching helped capture the closest match precisely based on defined factors and reduced effect of confounding. After propensity one-to-one matching, 32,144 robotic cases were compared with 32,144 open and 32,144 laparoscopic cases, which enabled us to perform a comparison between like cases.

For each hospital in each year, the total volume of each procedure type was calculated by adding the open, laparoscopic, and robotic cases together. Hospital classification as "robotic" or "nonrobotic" was done by flagging hospitals that performed at least 1 procedure (within our set procedure types) labeled with a robotic qualifier. We used a difference in difference (DID) method to test our hypothesis by analyzing the difference of total procedure volume in robotic hospitals pre- and postadoption of robotic technology and compared this to the difference in total procedure volume for nonrobotic hospitals across the same years using our propensity-matched cohorts. The DID is a causal analysis that uses longitudinal data from a treatment and a control group to estimate a causal effect of a specific intervention/treatment. The DID method is based on Poisson regression and is used to predict a response (a dependent variable in the form of "count data") that is impacted by 1 or more independent variables. DID was used to assess the relationship between total procedure volumes of robotic versus nonrobotic hospitals (the dependent variable) before and after adopting robotics (the independent variable). This DID analysis was performed for all procedures together and then individually for colectomy, lobectomy, gastric bypass, and antireflux. Incident rate ratios (IRRs) from the DID analysis determined the size of the effect adding robotics to a hospital had on surgical volume [[6], [7], [8]]. We regressed total procedure volume for robotic versus nonrobotic hospitals in addition to time when a hospital started performing robotic procedures. All data preparation methods and modeling codes used can be accessed electronically [9].

RESULTS

A total of 291,826 surgical cases of our selected types were performed at 217 hospitals within the database: 139,796 open, 119,886 laparoscopic, and 32,144 robotic cases. Of these 217 hospitals, 151 had robotic capabilities. Our analysis was performed on a propensity-matched cohort to the robotic cases such that we had a total N of 96,432 with 32,144 each of robotic, laparoscopic, and open cases. Most patients were female (57%), white (84%), non-Hispanic or Latino (83%), and ages 51–70 (45%). Overall, 64% of procedures were elective; 9%, urgent; and 28%, emergent. Most patients (51%) fell into the severe CCI (CCI 3; all P < .001; Table 1).

Table 1

Patient characteristics

	Open		Lap		Robotic		Total
	n = 139,796		n = 119,886		n = 32,144		n = 291,826
	n	%	n	%	n	%	n	%
Patient sex
Female	74,446	53%	73,028	61%	19,714	61%	167,188	57%
Male	65,350	47%	46,858	39%	12,430	39%	124,638	43%
Age range
≤ 30	3663	3%	7217	6%	835	3%	11,715	4%
31–50	18,490	13%	27,126	23%	5505	17%	51,121	18%
51–70	62,278	45%	53,534	45%	15,470	48%	131,282	45%
71–90	53,035	38%	31,203	26%	10,231	32%	94,469	32%
90 +	2330	2%	806	1%	103	0%	3239	1%
Race
White	118,253	85%	100,116	84%	27,389	85%	245,758	84%
Black	14,044	10%	12,541	10%	2596	8%	29,181	10%
Asian	942	1%	707	1%	256	1%	1905	1%
American Indian	162	0%	154	0%	48	0%	364	0%
Hawaiian Pacific Islander	55	0%	46	0%	14	0%	115	0%
Other	6340	5%	6322	5%	1841	6%	14,503	5%
Ethnicity
Hispanic or Latino	20,029	14%	22,790	19%	6866	21%	49,685	17%
Non-Hispanic or Latino	119,767	86%	97,096	81%	25,278	79%	242,141	83%
Admission source
Non–health care facility point of origin	104,816	75%	86,513	72%	22,900	71%	214,229	73%
Clinic or physician's office	26,650	19%	30,184	25%	8910	28%	65,744	23%
Transfer from a hospital	1846	1%	1070	1%	173	1%	3089	1%
Transfer from a skilled nursing facility (SNF) or intermediate care facility (ICF)	858	1%	184	0%	21	0%	1063	0%
Transfer from another health care facility	393	0%	154	0%	22	0%	569	0%
Other transfers	5233	4%	1781	1%	118	0%	7132	2%
Admission priority
Emergency	58,905	42%	19,800	17%	1894	6%	80,599	28%
Urgent	13,684	10%	9165	8%	2680	8%	25,529	9%
Elective	67,207	48%	90,921	76%	27,570	86%	185,698	64%
Payer types
Commercial health insurance	37,943	27%	44,346	37%	11,720	36%	94,009	32%
Medicaid	9409	7%	8652	7%	1317	4%	19,378	7%
Medicare	81,898	59%	56,756	47%	17,515	54%	156,169	54%
Other govt	3112	2%	3786	3%	854	3%	7752	3%
Self-pay	4764	3%	4646	4%	462	1%	9872	3%
All others	2670	2%	1700	1%	276	1%	4646	2%
Charlson comorbidity index
Low	38,905	28%	45,334	38%	9583	30%	93,822	32%
Moderate	19,859	14%	24,372	20%	4939	15%	49,170	17%
Severe	81,032	58%	50,180	42%	17,622	55%	148,834	51%
Discharge status
Discharged to home or self-care	67,143	48%	93,671	78%	24,134	75%	184,948	63%
Discharged or transferred to a skilled nursing facility	17,545	13%	4970	4%	899	3%	23,414	8%
Discharged or transferred to a short-term care facility	1301	1%	423	0%	83	0%	1807	1%
Discharged or transferred to home under care of home health care organization	38,613	28%	17,817	15%	6433	20%	62,863	22%
Expired	5701	4%	920	1%	185	1%	6806	2%
Discharged or transferred to a long-term care facility	9493	7%	2085	2%	410	1%	11,988	4%

All P values < .001 within category by χ.

Altogether, robotic hospitals had a 37% increase in procedure volume (IRR 1.37, P < .0001). This significant increase held true for all procedure types except antireflux procedures where there was no such significant increase in surgical volume at robotic hospitals (IRR 0.959, P = .079). The largest increase in volume due to robotics (73% increase) was seen in gastric bypass surgeries (IRR 1.735, P < .0001), which were performed at 82 robotic hospitals and 110 nonrobotic hospitals (Table 2).

Table 2

Difference in difference analysis and incident rate ratios

	n (# hospitals)	IRR	95% CI	P value
Colectomy
Robotic hospitals	133	1.38	1.36–1.40	<.0001
Nonrobotic hospitals	78
Lobectomy
Robotic hospitals	124	1.11	1.07–1.14	<.0001
Nonrobotic hospitals	46
Gastric bypass
Robotic hospitals	82	1.73	1.67–1.79	<.0001
Nonrobotic hospitals	110
Antireflux
Robotic hospitals	141	0.95	0.91–1.00	NS
Nonrobotic hospitals	50
All procedures combined
Robotic hospitals	151	1.37	1.36–1.39	<.0001
Nonrobotic hospitals	66

IRRs shown for the overall difference in difference analysis and for each procedure. CI, confidence interval; NS, not significant.

Hospital charges were reviewed as well among our propensity-matched cohort of 32,144 of each procedure type. Overall, the mean total charge was $122,141 for open surgery, $90,178 for laparoscopic, and $125,998 for robotic. On the whole, charges for robotic surgeries were statistically significantly higher than charges for open or laparoscopic surgeries except for total charges for robotic lobectomy ($119,301), gastric bypass ($112,411), and antireflux surgery ($121,383), which were statistically significantly less than open lobectomy ($122,283; P = .041), gastric bypass ($135,094; P < .0001), and antireflux surgery ($133,372; P < .0001), respectively. These findings held true when risk-adjusted for CCI except for CCI 3 patients where robotic cost was not statistically significantly less than open (Table 3a, Table 3b, Table 3c). Across all procedure types, total charges for robotic surgery were higher than those for laparoscopic, which remained true when risk-adjusted for CCI (Table 3a, Table 3b, Table 3c). When the data were separated out by procedure type (open/laparoscopic/robotic) and then within those data risk-adjusted by CCI, CCI 3 patients had higher total charges than CCI 1 patients except in open lobectomy cases (Table 3d, Table 3e, Table 3f).

Table 3a

Charges risk-adjusted by procedure type for Charlson comorbidity index 1

	Open	Laparoscopic	Robotic	ANOVA
Colectomy				Lap vs open	Rob vs open	Rob vs lap
Total charges	94,655.3 ± 88,353	76,514.33 ± 55,136.63	127,846.5 ± 84,364.81	<.001	<.001	<.001
OR	30,595.23 ± 26,390.61	26,763.32 ± 20,082.41	60,259.28 ± 49,747.44	<.001	<.001	<.001
Med/surg supply	12,218.95 ± 11,752.53	12,978.54 ± 9165.513	19,799.81 ± 14,281.85	.016	<.001	<.001
Anesthesia	7468.17 ± 7103.78	6776.745 ± 6044.73	12,045.52 ± 10,739.19	.001	<.001	<.001
PACU	3538.424 ± 3065.859	3264.919 ± 2581.371	4356.761 ± 5260.719	.005	<.001	<.001
Lobectomy				Lap vs open	Rob vs open	Rob vs lap
Total charges	115,739.1 ± 133,786.3	80,050.61 ± 58,726.24	96,493.48 ± 58,177.91	<.001	<.001	<.001
OR	30,173.05 ± 23,166.78	26,626.55 ± 19,925.29	33,472.99 ± 29,094.21	.003	.008	<.001
Med/surg supply	15,858.9 ± 15,926.35	13,191.21 ± 12,295.53	20,595.75 ± 15,991.59	<.001	<.001	<.001
Anesthesia	7777.866 ± 7176.86	5929.962 ± 5322.601	6348.524 ± 5751.171	<.001	<.001	.244
PACU	2654.527 ± 3067.164	2926.067 ± 2911.612	3424.7 ± 3134.339	.105	<.001	<.001
Gast bypass				Lap vs open	Rob vs open	Rob vs lap
Total charges	124,560.6 ± 136,137	63,949.3 ± 33,410.63	105,120 ± 52,473.49	<.001	<.001	<.001
OR	39,071.56 ± 39,179.89	23,545.11 ± 17,324.75	52,867.78 ± 36,980.87	<.001	<.001	<.001
Med/surg supply	15,554.51 ± 13,620.4	17,150.05 ± 13,086.07	22,820.6 ± 15,008.82	.002	<.001	<.001
Anesthesia	8571.887 ± 8306.492	5509.143 ± 5646.901	8318.006 ± 8544.746	<.001	.576	<.001
PACU	3872.312 ± 3327.858	3899.316 ± 3318.818	3180.95 ± 2497.934	.963	<.001	<.001
Antireflux				Lap vs open	Rob vs open	Rob vs lap
Total charges	119,073.9 ± 155,409.1	67,875.53 ± 66,148.05	105,970.4 ± 76,132.32	<.001	<.001	<.001
OR	31,627.63 ± 31,163.36	26,518.71 ± 19,863.28	52,371.57 ± 35,721.58	<.001	<.001	<.001
Med/surg supply	11,720.7 ± 17,223.87	11,681.67 ± 11,154.39	14,376.24 ± 11,737.29	.995	<.001	<.001
Anesthesia	8365.69 ± 10,183.23	6724.682 ± 6113.568	9421.881 ± 9044.877	<.001	<.001	<.001
PACU	3586.774 ± 3222.567	3299.32 ± 2690.251	4014.039 ± 4009.335	.012	<.001	<.001

Table 3b

Charges risk-adjusted by procedure type for Charlson comorbidity index 2

	OPEN	LAPAROSCOPIC	ROBOTIC		ANOVA
Colectomy				Lap vs open	Rob vs open	Rob vs lap
Total charges	116,726.8 ± 127,949.1	87,195 ± 66,374.52	145,269.6 ± 98,417.39	<.001	<.001	<.001
OR	31,195.01 ± 25,793.25	28,158.21 ± 21,154.61	64,582.55 ± 52,536.78	.054	<.001	<.001
Med/surg supply	13,476.14 ± 13,700.66	13,628.27 ± 9897.232	21,573.38 ± 15,857.79	.949	<.001	<.001
Anesthesia	7779.025 ± 7234.903	7176.901 ± 6338.175	13,075.71 ± 10,580.75	.118	<.001	<.001
PACU	3918.997 ± 3791.265	3500.623 ± 2883.185	4487.421 ± 3658.178	.003	<.001	<.001
Lobectomy				Lap vs open	Rob vs open	Rob vs lap
Total charges	119,043.7 ± 129,993.5	90,642.78 ± 73,291.32	106,638.6 ± 69,672.1	<.001	.025	.001
OR	29,376.23 ± 29,251.75	25,076.32 ± 17,000.02	36,076.22 ± 30,924.79	.001	<.001	<.001
Med/surg supply	15,335.22 ± 15,560.47	14,126.52 ± 14,577.15	21,001.52 ± 16,114	.213	<.001	<.001
Anesthesia	7150.308 ± 6690.582	5756.216 ± 4830.622	6677.189 ± 5969.829	<.001	.248	.002
PACU	2870.329 ± 3370.97	3160.028 ± 2922.892	3513.992 ± 3293.657	.125	<.001	.047
Gast bypass				Lap vs open	Rob vs open	Rob vs lap
Total charges	136,978.3 ± 148,591.1	66,958.55 ± 37,563.16	110,261.6 ± 61,864.17	<.001	<.001	<.001
OR	36,904.11 ± 36,789.47	23,987.67 ± 18,813.92	53,697.37 ± 36,038.01	<.001	<.001	<.001
Med/surg supply	16,553.82 ± 14,974.09	17,957.27 ± 12,676.55	22,903 ± 14,285.01	.01	<.001	<.001
Anesthesia	8658.287 ± 9076.686	5984.756 ± 4918.212	9728.854 ± 10,212.45	<.001	.001	<.001
PACU	4206.803 ± 4235.797	3894.175 ± 3470.711	3388.174 ± 2517.72	.022	<.001	<.001
Antireflux				Lap vs open	Rob vs open	Rob vs lap
Total charges	129,643 ± 152,427.1	82,237.19 ± 89,221.29	130,900.1 ± 109,153.7	<.001	.97	<.001
OR	31,440.65 ± 33,740.25	29,438.11 ± 25,948.48	57,494.21 ± 42,963.13	.402	<.001	<.001
Med/surg supply	11,015.4 ± 12,687.59	12,355.68 ± 11,048.87	14,679.51 ± 11,805.91	.03	<.001	<.001
Anesthesia	8294.456 ± 9168.844	7304.118 ± 6700.476	11,167.05 ± 10,608.35	.035	<.001	<.001
PACU	3691.341 ± 3833.096	3572.346 ± 3083.741	4456.894 ± 4070.033	.749	<.001	<.001

Table 3c

Charges risk-adjusted by procedure type for Charlson comorbidity index 3

	Open	Laparoscopic	Robotic		ANOVA
Colectomy				Lap vs open	Rob vs open	Rob vs lap
Total charges	121,332.9 ± 150,154.1	99,170.39 ± 86,102.06	154,037.1 ± 115,172.1	<.001	<.001	<.001
OR	32,168.89 ± 30,385.56	29,735.09 ± 24,490.73	64,474.09 ± 51,866	.003	<.001	<.001
Med/surg supply	13,387.41 ± 12,195.15	13,702.23 ± 10,430.27	20,701.21 ± 16,771.37	.467	<.001	<.001
Anesthesia	7881.825 ± 7796.974	7569.164 ± 6904.681	13,488.04 ± 11,299.49	.181	<.001	<.001
PACU	3781.377 ± 3457.973	3661.935 ± 3397.939	4605.33 ± 4165.976	.236	<.001	<.001
Lobectomy				Lap vs open	Rob vs open	Rob vs lap
Total charges	123,053.9 ± 117,855.8	108,845 ± 88,317.8	122,476.1 ± 82,267.99	<.001	.903	<.001
OR	33,472.23 ± 29,066.97	32,825.86 ± 24,277.44	43,288.45 ± 38,464.6	.299	<.001	<.001
Med/surg supply	17,030.9 ± 15,469.1	18,102.37 ± 16,023.67	23,987.32 ± 17,297.48	<.001	<.001	<.001
Anesthesia	8278.606 ± 7268.962	7062.693 ± 6300.246	7755.483 ± 6707.74	<.001	<.001	<.001
PACU	2967.321 ± 3219.008	2991.406 ± 3001.685	3661.673 ± 3760.589	.863	<.001	<.001
Gast bypass				Lap vs open	Rob vs open	Rob vs lap
Total charges	139,395.7 ± 156,939.1	84,956.56 ± 70,259.85	130,049.5 ± 109,754	<.001	.088	<.001
OR	29,586.87 ± 26,458.96	27,623.76 ± 23,316.61	54,290.94 ± 39,256.42	.109	<.001	<.001
Med/surg supply	15,473.54 ± 15,031.02	20,590.08 ± 14,436.88	24,583.38 ± 17,193.01	<.001	<.001	<.001
Anesthesia	7573.191 ± 6981.913	6912.194 ± 6105.275	11,702.16 ± 12,566.22	.049	<.001	<.001
PACU	4024.848 ± 3690.569	3612.774 ± 3056.891	3667.342 ± 2694.047	.001	.005	.909
Antireflux				Lap vs open	Rob vs open	Rob vs lap
Total charges	172,160 ± 162,225.3	100,524.2 ± 104,854.5	157,878.2 ± 139,023.8	<.001	.108	<.001
OR	34,281.75 ± 29,831.23	29,685.62 ± 21,350.46	60,532.18 ± 42,258.57	.015	<.001	<.001
Med/surg supply	13,558.15 ± 17,008.62	13,181.86 ± 10,980.11	17,156.97 ± 14,556.07	.871	<.001	<.001
Anesthesia	9445.393 ± 10,327.4	7606.396 ± 6524.486	12,616.67 ± 12,692.79	.001	<.001	<.001
PACU	3813.73 ± 3510.809	3478.583 ± 2976.502	4793.128 ± 4549.322	.187	<.001	<.001

For Table 3a, Table 3b, Table 3c: charges (mean ± standard deviation), risk-adjusted per procedure type, after separating by CCI. Post-Hoc ANOVA with pairwise comparison of the means, P values presented with significant values in italics. PACU, recovery room charges; Gast bypass, gastric bypass; Lap, laparoscopic.

Table 3d

Charges risk-adjusted by Charlson comorbidity index for open procedures

	CCI 1	CCI 2	CCI 3	ANOVA
Colectomy				CCI 2 vs CCI 1	CCI 3 vs CCI 1	CCI 3 vs CCI 2
Total charges	94,655.3 ± 88,353	116,726.8 ± 127,949.1	121,332.9 ± 150,154.1	<.001	<.001	.462
OR	30,595.23 ± 26,390.61	31,195.01 ± 25,793.25	32,168.89 ± 30,385.56	.78	.025	.5
Med/surg supply	12,218.95 ± 11,752.53	13,476.14 ± 13,700.66	13,387.41 ± 12,195.15	.003	<.001	.97
Anesthesia	7468.17 ± 7103.78	7779.025 ± 7234.903	7881.825 ± 7796.974	.383	.026	.894
PACU	3538.424 ± 3065.859	3918.997 ± 3791.265	3781.377 ± 3457.973	.001	.002	.374
Lobectomy				CCI 2 vs CCI 1	CCI 3 vs CCI 1	CCI 3 vs CCI 2
Total charges	115,739.1 ± 133,786.3	119,043.7 ± 129,993.5	123,053.9 ± 117,855.8	.843	.189	.648
OR	30,173.05 ± 23,166.78	29,376.23 ± 29,251.75	33,472.23 ± 29,066.97	.841	.003	<.001
Med/surg supply	15,858.9 ± 15,926.35	15,335.22 ± 15,560.47	17,030.9 ± 15,469.1	.775	.079	.01
Anesthesia	7777.866 ± 7176.86	7150.308 ± 6690.582	8278.606 ± 7268.962	.186	.118	<.001
PACU	2654.527 ± 3067.164	2870.329 ± 3370.97	2967.321 ± 3219.008	.366	.015	.703
Gast bypass				CCI 2 vs CCI 1	CCI 3 vs CCI 1	CCI 3 vs CCI 2
Total charges	124,560.6 ± 136,137	136,978.3 ± 148,591.1	139,395.7 ± 156,939.1	.069	.008	.871
OR	39,071.56 ± 39,179.89	36,904.11 ± 36,789.47	29,586.87 ± 26,458.96	.183	<.001	<.001
Med/surg supply	15,554.51 ± 13,620.4	16,553.82 ± 14,974.09	15,473.54 ± 15,031.02	.164	.985	.059
Anesthesia	8571.887 ± 8306.492	8658.287 ± 9076.686	7573.191 ± 6981.913	.954	<.001	<.001
PACU	3872.312 ± 3327.858	4206.803 ± 4235.797	4024.848 ± 3690.569	.046	.437	.292
Antireflux				CCI 2 vs CCI 1	CCI 3 vs CCI 1	CCI 3 vs CCI 2
Total charges	119,073.9 ± 155,409.1	129,643 ± 152,427.1	172,160 ± 162,225.3	.183	<.001	<.001
OR	31,627.63 ± 31,163.36	31,440.65 ± 33,740.25	34,281.75 ± 29,831.23	.987	.091	.126
Med/surg supply	11,720.7 ± 17,223.87	11,015.4 ± 12,687.59	13,558.15 ± 17,008.62	.491	.013	.002
Anesthesia	8365.69 ± 10,183.23	8294.456 ± 9168.844	9445.393 ± 10,327.4	.981	.019	.034
PACU	3586.774 ± 3222.567	3691.341 ± 3833.096	3813.73 ± 3510.809	.709	.23	.724

Table 3e

Charges risk-adjusted by Charlson comorbidity index for laparoscopic procedures

	CCI 1	CCI 2	CCI 3	ANOVA
Colectomy				CCI 2 vs CCI 1	CCI 3 vs CCI 1	CCI 3 vs CCI 2
Total charges	76,514.33 ± 55,136.63	87,195 ± 66,374.52	99,170.39 ± 86,102.06	<.001	<.001	<.001
OR	26,763.32 ± 20,082.41	28,158.21 ± 21,154.61	29,735.09 ± 24,490.73	.106	<.001	.045
Med/surg supply	12,978.54 ± 9165.513	13,628.27 ± 9897.232	13,702.23 ± 10,430.27	.081	.002	.965
Anesthesia	6776.745 ± 6044.73	7176.901 ± 6338.175	7569.164 ± 6904.681	.111	<.001	.1
PACU	3264.919 ± 2581.371	3500.623 ± 2883.185	3661.935 ± 3397.939	.031	<.001	.169
Lobectomy				CCI 2 vs CCI 1	CCI 3 vs CCI 1	CCI 3 vs CCI 2
Total charges	80,050.61 ± 58,726.24	90,642.78 ± 73,291.32	108,845 ± 88,317.8	.007	<.001	<.001
OR	26,626.55 ± 19,925.29	25,076.32 ± 17,000.02	32,825.86 ± 24,277.44	.239	<.001	<.001
Med/surg supply	13,191.21 ± 12,295.53	14,126.52 ± 14,577.15	18,102.37 ± 16,023.67	.311	<.001	<.001
Anesthesia	5929.962 ± 5322.601	5756.216 ± 4830.622	7062.693 ± 6300.246	.768	<.001	<.001
PACU	2926.067 ± 2911.612	3160.028 ± 2922.892	2991.406 ± 3001.685	.138	.749	.169
Gast bypass				CCI 2 vs CCI 1	CCI 3 vs CCI 1	CCI 3 vs CCI 2
Total charges	63,949.3 ± 33,410.63	66,958.55 ± 37,563.16	84,956.56 ± 70,259.85	.088	<.001	<.001
OR	23,545.11 ± 17,324.75	23,987.67 ± 18,813.92	27,623.76 ± 23,316.61	.741	<.001	<.001
Med/surg supply	17,150.05 ± 13,086.07	17,957.27 ± 12,676.55	20,590.08 ± 14,436.88	.12	<.001	<.001
Anesthesia	5509.143 ± 5646.901	5984.756 ± 4918.212	6912.194 ± 6105.275	.015	<.001	<.001
PACU	3899.316 ± 3318.818	3894.175 ± 3470.711	3612.774 ± 3056.891	.999	.031	.045
Antireflux				CCI 2 vs CCI 1	CCI 3 vs CCI 1	CCI 3 vs CCI 2
Total charges	67,875.53 ± 66,148.05	82,237.19 ± 89,221.29	100,524.2 ± 104,854.5	<.001	<.001	<.001
OR	26,518.71 ± 19,863.28	29,438.11 ± 25,948.48	29,685.62 ± 21,350.46	.001	.003	.972
Med/surg supply	11,681.67 ± 11,154.39	12,355.68 ± 11,048.87	13,181.86 ± 10,980.11	.241	.006	.293
Anesthesia	6724.682 ± 6113.568	7304.118 ± 6700.476	7606.396 ± 6524.486	.041	.004	.603
PACU	3299.32 ± 2690.251	3572.346 ± 3083.741	3478.583 ± 2976.502	.029	.322	.785

Table 3f

Charges risk-adjusted by Charlson comorbidity index for robotic procedures

	CCI 1	CCI 2	CCI 3	ANOVA
Colectomy				CCI 2 vs CCI 1	CCI 3 vs CCI 1	CCI 3 vs CCI 2
Total charges	127,846.5 ± 84,364.81	145,269.6 ± 98,417.39	154,037.1 ± 115,172.1	<.001	<.001	.017
OR	60,259.28 ± 49,747.44	64,582.55 ± 52,536.78	64,474.09 ± 51,866	.024	<.001	.997
Med/surg supply	19,799.81 ± 14,281.85	21,573.38 ± 15,857.79	20,701.21 ± 16,771.37	.001	.02	.179
Anesthesia	12,045.52 ± 10,739.19	13,075.71 ± 10,580.75	13,488.04 ± 11,299.49	.01	<.001	.453
PACU	4356.761 ± 5260.719	4487.421 ± 3658.178	4605.33 ± 4165.976	.647	.028	.684
Lobectomy				CCI 2 vs CCI 1	CCI 3 vs CCI 1	CCI 3 vs CCI 2
Total charges	96,493.48 ± 58,177.91	106,638.6 ± 69,672.1	122,476.1 ± 82,267.99	.022	<.001	<.001
OR	33,472.99 ± 29,094.21	36,076.22 ± 30,924.79	43,288.45 ± 38,464.6	.312	<.001	<.001
Med/surg supply	20,595.75 ± 15,991.59	21,001.52 ± 16,114	23,987.32 ± 17,297.48	.874	<.001	<.001
Anesthesia	6348.524 ± 5751.171	6677.189 ± 5969.829	7755.483 ± 6707.74	.551	<.001	<.001
PACU	3424.7 ± 3134.339	3513.992 ± 3293.657	3661.673 ± 3760.589	.868	.116	.54
Gast bypass				CCI 2 vs CCI 1	CCI 3 vs CCI 1	CCI 3 vs CCI 2
Total charges	105,120 ± 52,473.49	110,261.6 ± 61,864.17	130,049.5 ± 109,754	.051	<.001	<.001
OR	52,867.78 ± 36,980.87	53,697.37 ± 36,038.01	54,290.94 ± 39,256.42	.746	.516	.9
Med/surg supply	22,820.6 ± 15,008.82	22,903 ± 14,285.01	24,583.38 ± 17,193.01	.983	.003	.007
Anesthesia	8318.006 ± 8544.746	9728.854 ± 10,212.45	11,702.16 ± 12,566.22	<.001	<.001	<.001
PACU	3180.95 ± 2497.934	3388.174 ± 2517.72	3667.342 ± 2694.047	.021	<.001	.008
Antireflux				CCI 2 vs CCI 1	CCI 3 vs CCI 1	CCI 3 vs CCI 2
Total charges	105,970.4 ± 76,132.32	130,900.1 ± 109,153.7	157,878.2 ± 139,023.8	<.001	<.001	<.001
OR	52,371.57 ± 35,721.58	57,494.21 ± 42,963.13	60,532.18 ± 42,258.57	.001	<.001	.25
Med/surg supply	14,376.24 ± 11,737.29	14,679.51 ± 11,805.91	17,156.97 ± 14,556.07	.793	<.001	<.001
Anesthesia	9421.881 ± 9044.877	11,167.05 ± 10,608.35	12,616.67 ± 12,692.79	<.001	<.001	.011
PACU	4014.039 ± 4009.335	4456.894 ± 4070.033	4793.128 ± 4549.322	.013	<.001	.222

For Table 3d, Table 3e, Table 3f: charges (mean ± standard deviation), risk-adjusted by CCI, after separating by procedure type (open, laparoscopic, or robotic). Post-hoc ANOVA with pairwise comparison of the means, P values presented with significant values in italics.

LOS was statistically significantly shorter for robotic surgery when compared to open and laparoscopic (P < .0001) except when comparing robotic versus laparoscopic gastric bypass (2.56 vs 2.47 days; P = .788) and antireflux surgery (3.86 vs 3.55 days; P = .104), where there was no statistically significant difference. When risk-adjusted for CCI, this held true except for CCI 1 and CCI 2 colectomy patients, and CCI 1 lobectomy patients where robotic versus laparoscopic LOS was not statistically different (Table 4a). Table 4b displays the LOS risk-adjusted by procedure type; in general, the more severe CCI (2 or 3), the longer the length of stay regardless of procedure type (open/laparoscopic/robotic).

Table 4a

Length of stay risk-adjusted per Charlson comorbidity index

	Open	Lap	Robotic	ANOVA
Colectomy				Lap vs open	Rob vs open	Rob vs lap
CCI 1	6.66 ± 5.38	4.67 ± 4.02	4.47 ± 3.51	<.001	<.001	.114
CCI 2	8.17 ± 8.45	5.47 ± 4.72	5.30 ± 5.03	<.001	<.001	.754
CCI 3	8.67 ± 11.47	6.51 ± 5.97	5.98 ± 5.73	<.001	<.001	.003
Lobectomy
CCI 1	6.89 ± 7.03	4.07 ± 3.70	3.66 ± 3.51	<.001	<.001	.107
CCI 2	8.44 ± 7.89	5.58 ± 6.06	4.34 ± 3.95	<.001	<.001	<.001
CCI 3	7.73 ± 7.28	5.83 ± 5.57	4.78 ± 4.98	<.001	<.001	<.001
Antireflux
CCI 1	7.15 ± 8.70	2.83 ± 4.86	3.02 ± 3.89	<.001	<.001	.553
CCI 2	8.57 ± 9.05	3.79 ± 5.27	4.32 ± 5.34	<.001	<.001	.192
CCI 3	12.03 ± 10.31	5.63 ± 7.21	5.95 ± 7.02	<.001	<.001	.763
Gast bypass
CCI 1	7.54 ± 10.61	2.06 ± 1.97	2.12 ± 1.98	<.001	<.001	.907
CCI 2	8.78 ± 11.40	2.27 ± 2.05	2.32 ± 2.44	<.001	<.001	.96
CCI 3	11.16 ± 12.17	3.47 ± 4.87	3.79 ± 6.62	<.001	<.001	.678

Length of stay (mean ± standard deviation), risk-adjusted per CCI, comparing laparoscopic (Lap) versus open, robotic (Rob) versus open, and Rob versus Lap. Post-hoc ANOVA with pairwise comparison of the means, P values presented with significant values in italics.

Table 4b

Length of stay risk-adjusted per procedure type

	CCI 1	CCI 2	CCI 3	ANOVA
Colectomy				CCI 2 vs CCI 1	CCI 3 vs CCI 1	CCI 3 vs CCI 2
Open	6.66 ± 5.38	8.17 ± 8.45	8.67 ± 11.47	<.001	<.001	.182
Laparoscopic	4.67 ± 4.02	5.47 ± 4.72	6.51 ± 5.97	<.001	<.001	<.001
Robotic	4.47 ± 3.51	5.30 ± 5.03	5.98 ± 5.73	<.001	<.001	<.001
Lobectomy
Open	6.89 ± 7.03	8.44 ± 7.89	7.73 ± 7.28	<.001	.003	.026
Laparoscopic	4.07 ± 3.70	5.58 ± 6.06	5.83 ± 5.57	<.001	<.001	.306
Robotic	3.66 ± 3.51	4.34 ± 3.95	4.78 ± 4.98	.009	<.001	.043
Antireflux
Open	7.15 ± 8.70	8.57 ± 9.05	12.03 ± 10.31	<.001	<.001	<.001
Laparoscopic	2.83 ± 4.86	3.79 ± 5.27	5.63 ± 7.21	<.001	<.001	<.001
Robotic	3.02 ± 3.89	4.32 ± 5.34	5.95 ± 7.02	<.001	<.001	<.001
Gast bypass
Open	7.54 ± 10.61	8.78 ± 11.40	11.16 ± 12.17	.012	<.001	<.001
Laparoscopic	2.06 ± 1.97	2.27 ± 2.05	3.47 ± 4.87	.061	<.001	<.001
Robotic	2.12 ± 1.98	2.32 ± 2.44	3.79 ± 6.62	.184	<.001	<.001

Length of stay (mean ± standard deviation), risk-adjusted per procedure type—open, laparoscopic, and robotic—comparing CCI for each patient in the cohort. Post hoc ANOVA with pairwise comparison of the means, P values presented with significant values in italics.

DISCUSSION

This study establishes that robotic surgery increases surgical volume, decreases LOS, and, for select procedures studied, has lower total charges, which may have great benefit for both hospitals and patients. We used propensity matching for comparison of the robotic, laparoscopic, and open procedures to minimize unaccounted for variance in the cohorts [4,5] and a DID analysis to determine what amount of the change in surgical volume can be attributed to the addition of robotic technology. The DID analysis has been used in similar studies to establish a causal relationship between a dependent and independent variable in 2 continuous groups of data that our otherwise similar, ie, propensity-matched cohorts, over time [8,[10], [11], [12]].

One may postulate that the decreased total charges for robotic surgery compared with open surgery (and laparoscopic compared to open) can be attributed to the significant decrease in LOS for robotic and laparoscopic surgeries compared with open operations.

Decreased LOS following robotic surgery has been proven. Several colorectal surgeries have identified decreased LOS with robotic colectomy versus laparoscopic [[13], [14], [15]]. One study demonstrated equivalent overall cost between robotic and laparoscopic colectomy [15], leading to the conclusion that robotic surgery is more valuable to hospitals and patients than previously thought.

In the thoracic surgery arena, it has been shown that although robotic procedure cost was higher, there was no statistically significant difference in overall cost to patients due to lower postoperative costs [16]. Two studies even documented a profit margin with robotic lobectomy [17,18]. Although we did not examine cost, our study demonstrates significantly lower charges for robotic lobectomy versus open but still significantly higher charges for robotic versus laparoscopic. Interestingly, the LOS for robotic lobectomy was statistically significantly less when compared with open and laparoscopic lobectomy, pointing to the fact that decreasing LOS alone does not result in decreased overall charges to patient. Based on our risk-adjusted analysis, there is a strong element to patient severity of illness/comorbidities that contributes to LOS across procedure types. That said, one must also consider that a decreased LOS could mean less complications, risk of hospital-acquired infection, and faster recovery; these should all be further studied.

Our study does show decreased LOS across 4 major surgical procedures and adds the next step of identifying an overall increase in surgical volume in 3 out of 4 procedures. It will be interesting to see the long-term effects of adding robotic technology. The increase in volume may be short-lived because the prevalence of disease is likely not increasing and other hospitals will adopt robotic technology as time goes on. This phenomenon demonstrates 2 of Porter's 3 competitive strategies: cost leadership and differentiation.

Robotic technology does come with a high upfront cost to the hospital (and theoretically explains the relative lack of robotic surgery at ambulatory surgery centers). There is also a higher charge per procedure for robotics, although how each hospital establishes this cost/charge is unknown. Do they add in a base-charge for use of the robot to recuperate cost to purchase said robot? And if they do, for how long will they do that given the fixed startup cost? The answer to these questions is unknown and likely hospital-specific, but these are prudent questions as we move forward in a robot-centric surgical world.

Retrospective database review is an inherent limitation as one's conclusions are limited to the data provided. The data set we queried did not include hospital information that may also have an impact on surgical volumes, such as expansion of surgical space (adding operating rooms), personnel, marketing, and quality measures. The data set did not include information about surgeon training or information about length of surgery and postoperative complications. In the future, a prospective collection of surgical volume data with more hospital-specific data may be warranted to further evaluate the effects of adding robotics, as being able to control for other factors would help narrow the focus and increase the power of the study. Additionally, our data set was limited to the state of Florida, and 84% of the patients in the set were white; this reduces generalizability to the rest of the United States and warrants a larger exploration into similar data in different parts of the country.

Conclusions

By using propensity matching and difference-in-difference method to control for changes over time, we found that hospitals that adopt robotic technology increase their overall surgical volume by 37%. Robotic surgeries had decreased LOS but higher charges than their laparoscopic or open counterparts. Our study is limited by inability to control for all other factors, and a prospective trial or larger database review should be performed to reduce bias and increase generalizability of our findings.

The following are the supplementary data related to this article.

Supplementary Material

ICD9 and 10 codes used for our data search

Author Contribution

Dr. Emily Grimsley: methodology, data curation, writing – original draft, writing – review & editing

Dr. Tara Barry: conceptualization, investigation, methodology, data curation, software

Dr. Haroon Janjua: data curation, formal analysis, methodology, software

Dr. Emanuel Eguia: conceptualization, investigation, methodology, supervision

Dr. Christopher DuCoin: supervision, writing – review & editing

Dr. Paul Kuo: conceptualization, investigation, methodology, supervision, validation, writing – review & editing.

Conflict of Interest

The authors have no related conflicts of interest to declare.

Funding Source

This study was not funded.

Ethics Approval

This study was exempt from the Institutional Review Board given that it was a query of a deidentified database.