Evidence on Robotic Prostatectomy: Discussing the Limitations of

Real-World Data

Maciej Dzik

, Kacper Mucha and Monika Zaleska

Agency for Health Technology Assessment and Tariff System, Warsaw, Poland

Keywords: Prostate Cancer, Radical Prostatectomy, Real-World Evidence.

Abstract: INTRODUCTION: The aim of this research is to evaluate Real-World Data (RWD) related to prostatectomy

outcomes for prostate cancer with a focus on identifying potential biases and data limitations. METHODS:

This study was based on the financial records collected in the database the Polish National Health Fund. The

sample included 14,376 patients who underwent robot-assisted radical prostatectomy (RARP), laparoscopic

radical prostatectomy (LRP) or conventional radical prostatectomy (CRP) between 20 September 2022 and

31 December 2023. Comparative analysis focused exclusively on the duration of hospitalisation. Additional

outcomes included mortality. RESULTS: In total 6,609 patients had RARP. RARP compared to both CRP

and LRP was associated with a reduction in inpatient days by 2.81 (95% CI: -2.98, -2.65; p<0.0001) and 0.91

(95% CI: -1.02, -0.8; p<0.0001) respectively. Patient admitted as emergencies had statistically longer hospital

stays by 1.03 days (p<0.0001). CONCLUSIONS: The overall length of hospitalization has been reduced, but

interpreting the results obtained from RWD in terms of relative benefits is challenging. The analysis faced

several challenges, including interpreting outcome measures and validating their clinical significance,

handling outliers, addressing non-random assignment, and accounting for unobserved covariates. These

limitations underscore the need for further research to enhance the quality of comparisons.

1 INTRODUCTION

The aim of this research is to evaluate Real-World Data

(RWD) related to prostatectomy outcomes for prostate

cancer with a focus on identifying potential biases and

data limitations. This study builds on the previous

work in the field of robotic prostatectomy by Dzik et

al. (2024). Early data from the first few months of

financial records from the Polish National Health Fund

indicated the Robot-Assisted Radical Prostatectomy

(RARP) was associated with shorter hospital stays and

fewer transfusions of blood products than conventional

prostatectomy. In this study we focus on the duration

of hospitalization to highlight challenges in the

interpretation of the outcomes.

2 BACKGROUND

RWD is defined as data related to patient health,

experiences, or care delivery that is collected outside

https://orcid.org/0000-0001-9069-7566

of controlled clinical trials (NICE, 2022). RWD is of

significant interest in the regulatory context, as

evidenced by the efforts of multiple state institutions

to publish guidance on the use of RWD (NICE, 2022;

CADTH, 2023; EMA, 2023) and international

initiatives aimed generation, gathering and sharing

RDW such as European Health Data Space and

DARWIN-EU. Unlike the rigorously controlled

clinical trials, the quality of evidence derived with

RWD can be a cause of concern. Guidance

extensively mentions limitations of the RWD,

including biases due to non-random assignment to

treatment, unblinded ascertainment of outcomes and

errors in the data generation and entry (NICE, 2022;

CADTH, 2023; EMA, 2023).

To better understand the aforementioned

limitations, we conducted an analysis of evidence

derived from the financial records of the Polish

National Health Fund regarding the Robotic-Assisted

Radical Prostatectomy (RARP), a procedure that has

been publicly funded since April 2022. In our previous

work, using data from part of 2022, we found that

694

Dzik, M., Mucha, K. and Zaleska, M.

Evidence on Robotic Prostatectomy: Discussing the Limitations of Real-World Data.

DOI: 10.5220/0013248100003911

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 18th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2025) - Volume 2: HEALTHINF, pages 694-697

ISBN: 978-989-758-731-3; ISSN: 2184-4305

RARP, compared to both Conventional Radical

Prostatectomy (CRP) and Laparoscopic Radical

Prostatectomy (LRP), was associated with a reduction

in inpatient days by 1.13 (95% CI: -1.27, -0.99;

p<0.0001) and 0.83 (95% CI: -1.02, -0.64; p<0.0001),

respectively (Dzik et al., 2024). In this study, we aim

to review the outcomes with an expanded sample and

discuss the limitations evident in the data.

3 MATERIALS AND METHODS

This study was based on the financial records

collected in the database the Polish National Health

Fund. Comparative analyses used records on patients

who had RARP, laparoscopic radical prostatectomy

(LRP) or open (conventional) radical prostatectomy

(CRP) between 20 September 2022 and 31 December

2023. The cut-off date for mortality data was 1

September 2024. All patients were diagnosed

with malignant neoplasm of the prostate (ICD-10

C61) and had to meet the following clinical criteria to

be eligible for the procedure:

 Gleason Score: 6-10 (ISUP 1-5), disease

limited to the prostate, staged as cT1-2 N0 M0;

 Locally advanced disease, staged as cT3a-b

N0-1 M0;

 No distant metastases (M0), confirmed by

negative bone scintigraphy or whole-body

magnetic resonance imaging;

 Erectile function score (IIEF-5) greater than 21.

Patients were considered outliers and excluded if

they had more than one record of radical

prostatectomy.

Following outcomes were evaluated: duration of

hospitalisation and the number of deaths.

To identify potential confounders, both

continuous and categorical covariates were used.

Exact age was calculated as the number of days

between the date of admission and the date of birth,

divided by 365.25. To test for provider-related

effects, patients were divided into subgroups based on

whether they were treated in hospitals equipped with

a robot. A hospital was considered to have a robot if

at least one robotic-assisted radical prostatectomy

(RARP) had been reported. Additionally, patients

were stratified based on whether they had an

anaesthesiologic consultation prior to hospitalization

and whether they were admitted as an emergency.

We employed a variety of statistical tests to ensure

robust and comprehensive results. The T test was used

to compare the means between two groups, while

the Chi-Squared Pearson test assessed the association

between categorical variables. For comparing means

across multiple groups, we utilized ANOVA (Analysis

of Variance). To control for potential confounding

variables, ANCOVA (Analysis of Covariance) was

applied. Finally, post hoc comparisons were conducted

using the Tukey HSD Test to identify specific group

differences following a significant ANOVA result. The

threshold for statistical significance was set at 0.05.

Statistical analyses were performed in R version 4.4.

4 RESULTS

The sample included 14,376 patients who underwent

radical prostatectomy. Six patients were excluded

from the comparison due to having undergone more

than one radical prostatectomy. Specifically, two

patients had LRP followed by CRP, two had RARP

followed by LRP, one had two RARPs, and another

had two LRPs.

Of the remaining 14,370 patients 64.6% were

treated in hospitals equipped with a robot.

Anaesthesiologic consultations were provided to 24%

of patients. 97% of patients were admitted as

scheduled or through a fast track route, while only 3%

were admitted as emergencies.

Patients ranged in age from 40 to 80 years, with

an average age of 66.4 years. The ANOVA, followed

by the Tukey’s HSD test revealed that the patients in

RAPR group were statistically younger than those in

the LRP and CRP groups with adjusted p-values less

than 0.0001 for both comparisons. The age difference

between CRP and LRP was not statistically

significant (p-value = 0.1816).

Table 1: Baseline characteristics.

Category CRP LRP RARP Total

N 1,951 5,810 6,609 14,370

Mean age (SD) 67.3 (5.8)

67.0

(6.1)

65.7 (6.8) 66.4 (6.4)

Emergency

admission (%)

(3.5%)

191

(3.5%)

235

(3.6%)

494

(3.4%)

Anaesthesio-

logic

consultation

before

hospitaliza-tion

(%)

415 (21%)

1,134

(20%)

1,910

(29%)

3,459

(24%)

Hospital

equipped with a

robot (%)

433

(22%)

2,246

(39%)

6,609

(100%)

9,288

(65%)

The groups differed significantly in the

proportion of anaesthesiologic consultations received

(p-value < 0.0001). However, there were no

significant differences between the groups regarding

the proportion of emergency admissions (p-value =

Evidence on Robotic Prostatectomy: Discussing the Limitations of Real-World Data

695

0.7096). Baseline characteristics are summarized in

the Table 1.

Average duration of hospitalization was 4.8

(ranging from 1 to 128) days. The length of

hospitalization in RARP group was nearly one day

shorter than in LRP and three days shorter than CRP

(Figure 1.). The differences were statistically

significant. Results are summarized in the Table 2.

Table 2: Duration of hospitalization across treatments.

Comparison

Difference

(

)

95% confidence

interval

LRP vs. CRP -1.91* -2.07; -1.74

RARP vs.CRP -2.81* -2.98; -2.65

RARP vs. LRP -0.91* -1.02; -0.80

*p-value=0.0

Figure 1: Mean duration of hospitalization.

The impact of age on the length of hospitalization

was tested using ANCOVA, but it was found to be

non-significant (p-value=0.3286).

The average length of stay was shorter in

hospitals equipped with a robot compared to other

hospitals (4.3 vs. 5.7 respectively; p-value<0.0001).

Notably, for the same procedures, hospitals with

robotic equipment reported shorter hospital stays for

LRP (MD=-0.7, p-value<0.0001) but not CRP (MD=-

0.2, p-value = 0.3950).

Regarding other confounders, emergency

admissions were associated with longer hospital stays

overall and for all procedures, whereas

anesthesiological consultations were not (Table 3.,

Table 4.). Impact of the confounders on the

hospitalization duration is presented on Figure 2.

Table 3: Duration of hospitalization for emergency

admissions.

Comparison CRP LRP RARP Overall

Emergency

admission

8.32 5.75 5.02 5.76

MD versus no

emer

enc

1.55* 0.86* 1.05* 1.03*

*p-value<0.0001

Table 4: Duration of hospitalization Anaesthesiologic

consultations.

Comparison CRP LRP RARP Overall

Anaesthesiologic

consultation

6.33 4.82 4.32 4.73

MD vs no

consultation

-0.63 -0.12 0.43 -0.04

Figure 2: Effect of Treatment and Covariates on

Hospitalization Duration.

As of 1 September 2024, a total of 127 deaths

were recorded, representing 0.9% of the sample. The

data was deemed insufficiently mature for performing

survival analysis. Specifically, 45 deaths were

reported in the RARP group, 61 in the LRP group,

and 21 in the CRP group. The RARP group had the

lowest proportion of deaths at 0.7%. These

differences were not statistically significant (p-

value=0.06).

5 DISCUSSION

The data demonstrated a substantial and continuous

reduction in the length of hospital stays associated

with radical prostatectomy within less than two years

of the introduction of public funding for RARP in the

Polish healthcare system. Compared to early data

from the first year (Dzik et al., 2024) average length

of hospitalization has decreased by nearly half a day

(4.8 vs 5.2). However, interpretation of the relative

benefit of RARP in comparison to LRP and CRP

remains complex.

Firstly, outcome measure itself is subject to

misinterpretation. We must acknowledge that we

have been operating under the assumption that shorter

hospital stays equate to better performance and

HEALTHINF 2025 - 18th International Conference on Health Informatics

696

subsequently outcomes. In our example the clinical

evidence is abundant. The length of hospitalization

has been already a subject of multiple prior studies

(Ma et al., 2023) which indicates its clinical

importance. In our previous analysis (Dzik et al.,

2024) we have found that patients with RARP had

fewer transfusions than CRP, which could translate to

less invasive operations or fewer complications.

Another RWD study of Swiss medical records

reported that the most common complications of the

radical prostatectomy were bleeding, hematoma and

acute bleeding anaemia (von Ahlen et al, 2024).

The limited scope of this study did not allow to

look into other indicators of performance, but

extensions to this ongoing analysis are possible. The

financial records can be studied to identify

information on other procedures performed after the

prostatectomy, as mentioned transfusions, but also

parenteral nutrition and readmissions.

Secondly, regarding readmissions, we have

excluded 6 patients who had two separate records of

radical prostatectomy. This small number could be

considered an outlier. It is possible that these were

data entry errors or failed operations. The latter is

supported by the fact that four patients were reported

to had undergone a different surgical procedure the

second time. For the future investigations it will be

necessary to monitor for more cases like that.

Finally, the interpretation of a relative benefit of

RARP to other treatments remains a concern. We

have demonstrated that hospitals with robotic

equipment reported shorter hospitalizations overall

and for LRP but not for CRP. Concern number one is

existence of hospital-related effects due to differences

in capacity, skills of the personnel and overall

standards of care. These effects can be factored into

analysis with a care selection of additional covariates.

Concern number two is selection bias, which leads

to non-homogeneous groups. Unlike our previous

analysis, our sample predominantly consisted of

RARP cases due to the rapid rise in the popularity of

this procedure in clinical practice. Nevertheless, some

patients continued to undergo CRP even in hospitals

equipped with a robot. There may be unobserved

factors influencing the choice of CRP over RARP for

certain patients. Supporting this reasoning, we

observe that while hospitals with RARP reported

shorter hospitalizations overall and for LRP, there

was no evidence of such a difference for CRP.

Therefore, to estimate relative outcomes further,

analysis including more advanced techniques such as

regression and adjusted matched comparisons may be

necessary.

6 CONCLUSIONS

The RWD indicated that RARP is associated with

shorter hospital stays than LRP and CRP. The

challenges in the analyses included interpreting the

outcome measures and validating their clinical

significance, handling outliers, addressing non-random

assignment, and accounting for unobserved covariates.

These limitations highlight the need for further

research to enhance the quality of comparisons.

ACKNOWLEDGEMENTS

The authors of this study would like to thank

Dr. hab. Anna Kowalczuk of the Agency for Health

Technology Assessment for the support necessary to

develop this study.

REFERENCES

Dzik, M., Płusa, A., Zaleska M., (2024). Real world

evidence on the effectiveness of robotic prostatectomy

in Poland. [Poster Presentation]. Health Technology

Assessment International 2024 Annual Meeting,

Seville, Spain

National Institute for Health and Care Excellence.

(2022). NICE real-world evidence

framework. https://www.nice.org.uk/corporate/ecd9

CADTH. (2023). Guidance for reporting real-world

evidence. https://www.resilienthealthcare.ca/publicatio

ns/cadth-guidance-for-reporting-real-world-evidence

European Medicines Agency. (2023). Real-world evidence

framework to support EU regulatory decision-making:

Report on the experience gained with regulator-led

studies from September 2021 to February 2023.

https://www.ema.europa.eu/en/documents/report/real-

world-evidence-framework-support-eu-regulatory-

decision-making-report-experience-gained-regulator-

led-studies-september-2021-february-2023_en.pdf

Ma, J., Xu, W., Chen, R., Zhu, Y., Wang, Y., Cao, W., Ju,

G., Ren, J., Ye, X., He, Q., Chang, Y., & Ren, S. (2023).

Robotic-assisted versus laparoscopic radical

prostatectomy for prostate cancer: The first separate

systematic review and meta-analysis of randomised

controlled trials and non-randomised studies.

International Journal of Surgery, 109(5), 1350-1359.

https://doi.org/10.1097/JS9.0000000000000193

Von Ahlen, C., Geissler, A., & Vogel, J. (2024).

Comparison of the effectiveness of open, laparoscopic,

and robotic-assisted radical prostatectomies based on

complication rates: A retrospective observational study

with administrative data from Switzerland. BMC

Urology, 24(215). https://doi.org/10.1186/s12894-024-

01597-3.

Evidence on Robotic Prostatectomy: Discussing the Limitations of Real-World Data

697