Trend and cost analysis comparing laparoscopic versus open liver resections at a Tertiary Referral hospital in Australia

Abstract

Background

Minimally invasive methods for liver resection operations have increased throughout the world. We aimed to determine the trends and cost differences in liver resections conducted by open and laparoscopic approaches at a teaching hospital in South Australia.

Methods

We conducted a retrospective analysis of 284 liver resections (247 open (OLR); 37 laparoscopic (LLR); subgroup analysis of 135 major and 149 minor) at Flinders Medical Centre between the 2013/14 and 2021/22 financial years. Trends were calculated using Mann Kendall and statistical analysis was conducted using the Welch t-test.

Results

87% of all resections, 98% major, and 77% of minor resections were OLRs. There was a significant decrease in the number of liver resections in total (tau = -0.686, p$<$0.05) and OLR (tau = -0.817, p < 0.05). OLR had a significantly higher length of stay ($8.17\pm 6.34$days) vs. LLR$(4.35\pm 3.4$days) (p$<$0.001) in total and for minor resections ($4.30\pm 3.43$ days vs $7.45\pm 5.05$ days, p$<$0.001). OLR also had a significantly higher total cost ($\$42,873\pm 20182\mathit{vs}.\$23,958\pm 14,700,$p$<$0.001) in total and minor resections ($\$21,982\pm 10,344$ vs.$\$38,210\pm 16,936$, p$<$0.001). Operating room (OR) was the largest cost bucket in total and subgroup analysis. LLR had a significantly lower OR cost in total ($\$9,079\pm 5,160\mathit{vs}\$14,238\pm 5,739$, p$<$0.001) and in minor resections ($$8,891\pm 3,671$ VS. $12,225\pm 4,794$, p$<$0.001)

Conclusions

Over the study period there has been a decrease in the total number of liver resections, especially OLR. LLR can reduce costs in appropriately selected patients.

Acronyms

EMR

Electronic medical records

HCC

Hepatocellular Carcinoma

ICU

Intensive Care Unit

FMC

Flinders Medical Centre

LLR

Laparoscopic Liver Resection

LOS

Length of Stay

MBS

Medicare Benefit Schedule

OLR

Open Liver Resection

OACIS

Open Architecture Clinical Information System

OR

Operating room

ORMIS

Operating Room Management Information System

SPS

Special Procedural Suite

URN

Unit Record Number

Introduction

Over time hepatobiliary surgery has evolved with the introduction of minimally invasive approaches to a variety of surgeries including liver resections. Though the initial uptake was slow, benefits of laparoscopic operations over the conventional open approach including reduced intraoperative blood loss, improved cosmetic healing, and reduced post operative complications, the number of laparoscopic liver resections (LLR) rose with a subsequent extension to more complex operations including major resections which were previously left for open liver resections (OLR) [[1], [2], [3], [4], [5], [6]].

While the safety and benefits of LLR have been confirmed, uncertainty persists over cost effectiveness of LLR over OLR [7,8]. A 2021 meta-analysis comparing OLR and LLR looked at thirty-eight studies from across the globe and showed that LLR had significantly higher operative costs (i.e. theatre time, instrumentation, staff costs, etc.) when compared to OLR. However, when analysing hospitalisation costs (i.e. intensive care unit, high-dependency unit, ward, medication, etc.) LLR was shown to be significantly lower in cost compared to OLR. When the two costs were added, Ziogas et al. (2021) found no statistically significant difference between the two approaches [9]. In this meta-analysis only one out of the thirty-eight studies were Australian.

To understand patterns of OLR and LLR use and its costs over time, out study aimed to look at the trends and cost analysis of liver resections in a medium volume public hospital, Flinders Medical Centre (FMC), in South Australia (SA).

Methods

This study was a retrospective analysis of data collected from 1 January 2013 to 31 December 2022 at FMC, part of the South Adelaide Local Health Network (SALHN) which services a population of >380,000 in the southern Adelaide metropolitan region [10]. Data was extracted through the Operating Room Management Information System (ORMIS) for the appropriate Medicare Benefits Schedule (MBS) item codes. After which a combination of Open Architecture Clinical Information System (OACIS) and Sunrise Electronic Medical Record (Sunrise EMR) was used to obtain clinical information (Fig. 1).

Fig. 1.

The final data set analysed included 247 OLR and 37 LLR from a starting dataset of 460 extracted from ORMIS.

This project was exempt from requiring ethics review or approval as it was deemed a quality improvement project. The study was registered under the SALHN quality Register (reference number 4682).

Each operation was classified into OLR or LLR following the operation details or operation note. The Brisbane 2000 terminology for liver anatomy and resections was used to determined how many Couinaud segments were removed using the operation details or by accessing operation notes [11]. Each surgery was then classified as minor ($\le 2$Couinaud segments) or major ($\ge 3$Couinaud segments) resections as per the second international consensus conference in Morioka [12]. The reason for resection was then determined using the primary diagnosis and if needed secondary diagnosis and pathology reports.

Cost data was obtained from SALHN Patient Costing Team. Data extracted included the total cost and individual cost buckets (see Supplementary Table 1 for description). All in hospital costs including direct and indirect costs were assessed in Australian Dollar (AUD) indexed to the year 2021/22. Patients whose admission spread across two financial years were combined into the financial year at admission.

Statistical analyses was conducted using R version 2023.12.1 + 402 [13]. Mann Kendall was used to determine trends between the two groups. Welch t-test was used to determine differences between the two groups due to the small sample size in this study, with results represented as mean$\pm$SD. Statistical significance was set as a p $\le$0.05. Medial values and IQR was also used for age and cost data due to the large spread.

Results

Demography

The mean age of patients undergoing OLR was $63.16\pm 12.81$ (median age =66 years, IQR =15). Which was non-significantly higher than the mean age for LLR which was $56.73\pm 19.36$ (median age = 59, IQR =30) (95 CI –0.21 to 13.07, p = 0.06). Over time OLR showed a significant increase in the mean age of patients having liver resections per financial year (tau =0.944, p$<$0.001). While LLR had a non-significant increase in the mean age of patients over time (tau = 1 and p > 0.05).

Pathology

The most common pathology for OLR was malignant neoplasms of the liver and intra hepatic bile duct (65.59%) followed by HCC (11.34%). There was a similar pattern in the LLR group with the most common pathology being malignant neoplasm of the liver and intrahepatic bile duct (29.73%) and HCC and benign neoplasm of the liver being the second most common cause at 16.22%. The number of OLR for malignant neoplasms of the liver and intra hepatic bile duct significantly decreased over time (tau = −0.92, p < 0.05). All other pathologies showed no significant trends over time. Total number of operations

In total 284 patients underwent a liver resection between the 2013/14 and 2021/22 FY at FMC. Of the 284 patients, the majority had OLR (86.97%) vs. LLR (13.03%). When comparing the number of operations conducted by the two modes of surgery there was a significant difference in the mean number of operations conducted via OLR ($27.44\pm 8.13$) vs. LLR ($4.11\pm 2.42$) over time (95% CI 17.13 to 30.20, p$<$0.001). Over time there was a significant decrease in the total (OLR + LLR) number of liver resections (tau = −0.686, p$<$0.05) and those conducted via OLR (tau = −0.817, p$<$0.05) (Fig. 2). There was an increase in the number of liver resections conducted laparoscopically from 4 in the 2014/15 financial year to 6 in 2021/22, however, this did not yield statistical significance (tau = 0.34 p$<$0.05).

Fig. 2.

Total of 284 liver resections - 247 OLR and 37 LLR showing a significant decrease in both the total number of liver resections (OLR + LLR – grey) and those conducted via OLR (blue). The increase in LLR (orange) was non-significant. 21.

Minor vs major liver resections

Out of the total 284 liver resections between 2013/14 – 2021/22 financial period, 135 were major resections (47.5%) (98% OLR) and 149 (52.5%) were minor (77% OLR). There was a statistically significant decrease in the number of major liver resections conducted via OLR (tau = −0.61, p$<$0.05) and a non-significant decrease in the number of minor liver resections via OLR (tau = −0.31, p > 0.05). Both major (tau=−0.31, p > 0.05) and minor (tau=0.51, p > 0.05) LLRs showed no significant trends over time.

Length of stay

OLR ($8.17\pm 6.34$ days) had a significantly higher LOS compared to LLR ($4.35\pm 3.41$ days) (95% CI 22.23 to 5.73, p$<$0.001). OLR showed a non-significant decrease in mean LOS (tau = −0.5, p < 0.05) while LLR showed a non-significant increase in LOS over time (tau = 0.11, p > 0.05). Subgroup analysis showed a significantly lower LOS for minor resections of $4.30\pm 3.43$ days for LLR vs $7.45\pm 5.05$ days for OLR (95% CI 1.65 to 4.65, p$<$0.001). There was no significant difference in LOS for major resections via OLR (LOS $8.79\pm 7.25$ days) vs LLR ($8.07\pm 6.49$ days) (95% −14.80 to 16.25, p > 0.05). Statistical significance was not met when removing the 2017/18 outlier patient with a significant LOS (OLR mean LOS $8.79\pm 7.25$ days vs. LLR mean LOS $5.05\pm 5.44$, 95%CI −40.18 to 47.66, p > 0.05).

Total cost

The mean total cost was significantly higher for OLR ($\$42,872.89\pm 20182.33)$ vs. LLR $(\$23,957.88\pm 14700.16$) (95% CI $13,437.95 to $24,392.06, p$<$0.001) when looking at all resections together. There was a nonsignificant decrease in cost for both OLR (tau = −0.5, p = 0.076) and LLR (tau = −0.143 and p > 0.05) over time.

There were no significant differences in the mean cost of major liver resection done via LLR ($\$46351.87\pm 36131.65$) compared to OLR $(\$46935.55\pm 21898.98$) (95% CI $−88,136.86 to $89,204.23, p > 0.05). However, minor resections conducted via LLR ($\$21,981.94\pm 10344.36)$ was significantly lower when compared to OLR ($\$38,209.66\pm 16936.38$) (95% CI $11,508.79 to $20,946.63, p$<$0.001) There was a significantly high mean cost for the LLR group in the 2017/18 FY corresponding to the major resection with long LOS occurring in that year. This resulted in major resections via LLR to have a high IQR of $35,787.98 vs. OLR IQR of $14,656.63. However, statistical significance was not met when removing this outlier (mean LLR_outlier_removed $\$29,893\pm 31394.86$, p > 0.05).

Cost buckets

The highest cost bucket for both OLR and LLR was the operating room (OR) cost bucket (Supplementary figure 1,see Supplementary Table 1 [[24], [25], [26]] for description of cost buckets). OLR showed a non-significant downward trend in OR cost over time (tau = −0.111, p > 0.05) while LLR had a significant increase over time (tau=0.556, p < 0.05) (Supplementary Table 2). OLR showed a significant decrease in costs for hotel, prosthetics, and ward supplies (Supplementary Table 2). Trends in all other cost buckets were non-significant. There was a significant difference in the mean OR cost between OLR and LLR with LLR having a lower mean cost of $\$9078.51\pm 5160.34$ compared to OLR mean cost of $\$14237.95\pm 5739.18$ (95%CI 3304.60 to 7014.27, p$<$0.001). Allied health, critical care, depreciation, imaging, non-clinical, on costs, prosthetics, ward medicine, ward nurse, and ward supplies all showed significant differences with OLR having a higher mean cost over LLR. ED, hotel, pharmacology, and SPS showed non-significant cost differences between OLR and LLR (Supplementary figure 1, Supplementary Table 3)

Subgroup analysis of major resections showed that prosthetics had a significantly higher mean cost for OLR $(\$1712.12\pm 1598.42$) vs. LLR ($\$0.00\pm 0.00$) (95% CI $1436.90 to $1987.35, p$<$0.001) All other cost buckets showed no significant cost difference for major resections (Supplementary Table 4). Subgroup analysis of minor resections showed that OR, allied health, depreciation, hotel, imaging, non-clinical, on-costs, pharmacology, ward medicine, ward nursing, and ward supplies were more expensive for patients undergoing OLR vs LLR (Supplementary Table 5).

Discussion

Our research looked at the trends and cost analysis of OLR and LLR at FMC – a medium volume university hospital in Adelaide, Australia. Our dataset included 284 patients who underwent a liver resection between the 2013/14 – 2021/22 FY. Most were OLR (87%) likely reflecting the lower likelihood of LLR in lower volume centres [4,14].

Malignancy was the most common reason for liver resection within our cohort with malignant neoplasm of the liver and intra hepatic duct being the most common reason for both OLR (65.59%) and LLR (29.73%). HCC was the second most common pathology for OLR (11.34%). While for LLR, HCC together with benign neoplasm of the liver was the second most common cause at 16.22% each. This is in keeping with global research showing metastasis and HCC as the most common causes for liver resection [15,16].

We showed a significant decrease in the total (OLR + LLR) and OLR over time, which may reflect the decreasing incidence of colorectal cancer in Australia with an estimated incidence of 58 cases per 100,00 persons in 2023 vs. an incidence rate of 65 cases per 100,000 persons in 2019 [17]. As well as effective screening programs that diagnose cancer at an early stage and improvements in therapeutic options that mitigate the risk of colorectal cancer metastasis. Including cytotoxic chemotherapy which has shown to reach overall survival of 10–10–12% without resection at 5 years, pre-operative chemotherapy showing 5–38% of patients having disappearing liver metastases, and neoadjuvant chemotherapy options such as anti- vascular endothelial growth factor receptor antibodies and increased use of trans arterial radioembolization for primary liver cancers including HCC and intrahepatic cholangiocarcinoma [[18], [19], [20], [21], [22]].

LLR was shown to have a significantly reduced LOS ($4.35\pm 3.41$ days) compared to OLR ($8.17\pm 6.34$ days). With subgroup analysis showing a significantly lower LOS for minor resections for LLR ($4.30\pm 3.43$ days) vs. OLR ($7.45\pm 5.05$ days). The reduced LOS may reflect quicker return to oral intake, earlier ambulation time requiring reduced allied health input, and reduced number of days for narcotic pain medication [23]. The lack of statistical significance for major resections may reflect the low number of laparoscopic major resections with variable LOS.

LLR had a significantly lower mean total cost overall (OLR $\$42,872.89\pm 20182.33$ vs. LLR $\$23,957.88\pm 14700.16$) and for minor resections (OLR =$\$38,209.66\pm 16936.38$ vs. LLR $\$21,981.94\pm 10344.36)$. No significant difference between OLR and LLR was observed for major resections likely due the small sample number and significant outlier. When the outlier was removed, we showed that the mean LLR cost was lower, though we were unable to meet significance which may correspond to the small sample size which reduces power to detect statistical significance. OR was the highest cost bucket in both OLR and LLR with a significantly increasing trend in cost over time for LLR. LLR had a significantly lower costs in allied health, depreciation, imaging, non-clinical, ward medicine, ward nurse, and ward supplies overall and in minor resections. This may reflect the reduced LOS we saw in LLR vs OLR which reduces the associated ward costs, allied health input due to reduced post-op complications, decreased use of high value items such as the iron intern, and larger volume of minor resections requiring less pre-operative investigations.

Our study was limited by its analysis of one hospital with low to medium volume of liver resections. Evidently, we only had 3 major LLR making statistical analysis difficult and limits the use of our findings externally. Though increasing the number of major resections in the future would allow a larger sample size, this may require capital investments for further laparoscopic equipment and training. As our study did not analyse such costs, further research is required to determine if this is feasible. The comparison of open and laparoscopic surgical methods is also rife with limitations that warrant cautious interpretation of the reported cost differences. It is noted that there are a multitude of factors including the size, type, location, and physiological status of the patient that must be considered when deeming the appropriateness of choosing laparoscopic vs. open surgical approaches. Though our research did not explicitly analyse the differences in the complexity of surgery undertaken by OLR vs LLR, we had a large skew for minor resections via LLR indicating a potential selection bias for lower complexity surgeries performed using this approach. Further research incorporating a larger data set is hence required to provide power to analyse subgroups for surgical complexity.

Disclosure statement

The authors declare that we have no relevant or material financial interests related to the research described in this paper

CRediT authorship contribution statement

K. Kariyawasam: Writing – original draft, Methodology, Investigation, Formal analysis. Jonathan Karnon: Writing – review & editing, Supervision, Project administration, Methodology, Investigation, Conceptualization. Mark Brooke-Smith: Writing – review & editing, Conceptualization. Savio George Barreto: Writing – review & editing, Supervision, Project administration, Methodology, Data curation, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.