🏥 Clinical Data Analysis: Robotic vs. Laparoscopic Hysterectomy¶
Comparative Surgical Outcomes — Panama 2023¶
Author: Edwin Berbey · Senior Data Engineer
Certification: IBM Data Science Professional Certificate
Dataset: Clinical registry of 122 patients who underwent minimally invasive hysterectomy
Tools: Python · Pandas · SciPy · Statsmodels · Plotly
Data provided with institutional authorization. Hospital identities anonymized per standard research ethics practice.
Clinical Background¶
Hysterectomy is the most common major gynecological procedure worldwide. Two minimally invasive techniques have dominated modern surgical practice:
- Laparoscopic Hysterectomy: Standard technique, widely available in public health systems.
- Robotic Hysterectomy (Da Vinci System): Robot-assisted technology, available at high-complexity private centers.
This analysis compares the clinical and operative outcomes of both techniques using real-world 2023 data from two Panamanian hospitals — one private (Hospital A) and one public reference center (Hospital B).
Research Question¶
Are there statistically significant differences in surgical time, intraoperative blood loss, and hospital stay between robotic and laparoscopic hysterectomy?
0. Environment Setup¶
✅ Environment ready Pandas 3.0.2 | NumPy 2.4.4
1. Data Loading & Cleaning¶
📊 Dataset loaded: 122 rows × 23 columns Robotic (Hospital A): 63 patients Laparoscopic (Hospital B): 59 patients Missing values: 0
2. Descriptive Statistics — Clinical Comparison Table¶
Equivalent to a Table 1 in peer-reviewed medical publications
Table 1 — Baseline characteristics and surgical outcomes by technique
Significance: *** p<0.001 | ** p<0.01 | * p<0.05 | ns = not significant
Variable Robotic (n=63) median [IQR] Laparoscopic (n=59) median [IQR] p-value (Mann-Whitney U) Sig.
Age (years) 48.0 [43.0–52.0] 45.0 [39.5–49.0] 0.0185 *
BMI (kg/m²) 27.0 [25.0–30.5] 31.0 [27.0–34.5] 0.0003 ***
Weight (kg) 70.0 [62.0–80.5] 76.0 [68.5–88.5] 0.0221 *
Height (cm) 159.0 [156.0–164.5] 159.0 [154.0–163.0] 0.3706 ns
Uterus weight (g) 161.0 [120.0–256.5] 210.0 [110.0–350.0] 0.1353 ns
Uterus size (cm) 12.0 [10.0–13.5] 12.0 [8.0–14.0] 0.6312 ns
Surgery time (min) 76.0 [63.0–94.0] 117.0 [100.5–141.0] 0.0000 ***
Blood loss (mL) 30.0 [15.0–50.0] 200.0 [100.0–300.0] 0.0000 ***
Hospital stay (days) 2.0 [2.0–2.0] 2.0 [2.0–2.0] 0.0015 **
3. Exploratory Data Analysis¶
4. Hypothesis Testing — Statistical Inference¶
The Mann-Whitney U test (non-parametric) was selected because:
- Normality cannot be assumed with n < 200 per group
- Surgery time and blood loss exhibit positive skewness
- It is the standard test in clinical comparative studies of this type
H₀: No significant difference between the medians of both techniques
H₁: A significant difference exists (α = 0.05, two-tailed)
Variable Robotic median Laparosc. median U statistic p-value Result ------------------------------------------------------------------------------------------------------------ Surgery Time (min) 76.0 117.0 646 0.0000 SIGNIFICANT ✅ Blood Loss (mL) 30.0 200.0 164 0.0000 SIGNIFICANT ✅ Hospital Stay (days) 2.0 2.0 1516 0.0015 SIGNIFICANT ✅ Uterus Weight (g) 161.0 210.0 1566 0.1353 Not significant BMI (kg/m²) 27.0 31.0 1150 0.0003 SIGNIFICANT ✅ Age (years) 48.0 45.0 2318 0.0185 SIGNIFICANT ✅ Significance level α = 0.05 | Test: Mann-Whitney U (two-tailed)
5. Correlation Analysis¶
6. Regression Analysis — Predictors of Surgical Outcomes¶
=================================================================
MODEL 1 — OLS Regression: Predictors of Surgery Time (min)
=================================================================
Coef. Std.Err. t P>|t| [0.025 0.975]
Intercept 83.1214 24.0136 3.4614 0.0008 35.5595 130.6834
bmi 0.5862 0.4042 1.4503 0.1497 -0.2144 1.3867
uterus_weight_g 0.0845 0.0141 5.9931 0.0000 0.0566 0.1125
uterus_size_cm 0.3061 1.1955 0.2561 0.7983 -2.0617 2.6739
age -0.2291 0.3318 -0.6907 0.4911 -0.8862 0.4280
technique_bin -27.2412 5.5359 -4.9208 0.0000 -38.2056 -16.2767
R² = 0.6229
Adj. R² = 0.6067
F-statistic p-value = 0.000000
=================================================================
MODEL 2 — OLS Regression: Predictors of Blood Loss (mL)
=================================================================
Coef. Std.Err. t P>|t| [0.025 0.975]
Intercept 3.7960 76.3974 0.0497 0.9605 -147.5187 155.1108
bmi 2.2444 1.2859 1.7454 0.0836 -0.3025 4.7913
uterus_weight_g 0.0559 0.0449 1.2457 0.2154 -0.0330 0.1448
uterus_size_cm 4.4989 3.8033 1.1829 0.2393 -3.0341 12.0319
age 1.2454 1.0555 1.1800 0.2404 -0.8451 3.3359
technique_bin -160.1828 17.6119 -9.0951 0.0000 -195.0655 -125.3001
R² = 0.5369
Adj. R² = 0.5169
F-statistic p-value = 0.000000
7. Summary Dashboard¶
8. Conclusions¶
Key Findings¶
| Outcome | Robotic (median) | Laparoscopic (median) | p-value | Interpretation |
|---|---|---|---|---|
| Surgery time | 76 min | 117 min | <0.001 | Robotic significantly faster (−35%) |
| Blood loss | 30 mL | 200 mL | <0.001 | Robotic ~6.7× less blood loss |
| Hospital stay | 2 days | 2 days | 0.002 | Distribution differs despite equal median |
| BMI | 27 kg/m² | 31 kg/m² | <0.001 | Different patient profiles by center |
| Uterus weight | 161 g | 210 g | ns | No significant difference |
Interpretation¶
Robotic technique shows significant intraoperative advantages in surgical time (−35%) and blood loss (−85%) compared to laparoscopic, with high statistical evidence (p<0.001).
Hospital stay did not differ meaningfully (median 2 days in both groups), suggesting that intraoperative benefits do not translate into shorter postoperative recovery at this sample size.
Laparoscopic patients presented higher BMI (median 31 vs. 27 kg/m²), likely reflecting referral differences between public and private centers rather than a technique effect.
Surgical technique is the dominant predictor of both surgery time (R²=0.62) and blood loss (R²=0.54) in the regression models, outperforming uterus weight, BMI, and age.
Limitations¶
- Retrospective observational design: causal inference is not possible
- Patients come from two centers with different profiles (private vs. public)
- Near-zero complication rate limits inferential analysis of that variable
- n=122 is adequate for descriptive and basic inferential statistics, but insufficient for meaningful subgroup analysis
Edwin Berbey (Timmy) · Senior Data Engineer
IBM Data Science Professional Certificate
🌐 Portfolio · LinkedIn
Data provided with institutional authorization. Hospital identities anonymized per standard research ethics practice.