diff --git a/eds_scikit/period/stays.py b/eds_scikit/period/stays.py
index 13f76033..5d3532e4 100644
--- a/eds_scikit/period/stays.py
+++ b/eds_scikit/period/stays.py
@@ -7,7 +7,9 @@
 from eds_scikit.utils.datetime_helpers import substract_datetime
 from eds_scikit.utils.framework import get_framework
 from eds_scikit.utils.typing import DataFrame
+from eds_scikit.utils.sort_first_koalas import sort_values_first_koalas
 
+import pandas as pd
 
 def cleaning(
     vo,
@@ -69,7 +71,6 @@ def cleaning(
 
     return vo[~mask], vo[mask]
 
-
 @concept_checker(concepts=["STAY_ID", "CONTIGUOUS_STAY_ID"])
 def merge_visits(
     vo: DataFrame,
@@ -290,28 +291,23 @@ def get_first(
             right_index=True,
             how="inner",
         )
-
-        # Getting the corresponding first visit
-        first_visit = (
-            merged.sort_values(
-                by=[flag_name, "visit_start_datetime_1"], ascending=[False, False]
-            )
-            .groupby("visit_occurrence_id_2")
-            .first()["visit_occurrence_id_1"]
-            .reset_index()
-            .rename(
-                columns={
-                    "visit_occurrence_id_1": f"{concept_prefix}STAY_ID",
-                    "visit_occurrence_id_2": "visit_occurrence_id",
-                }
-            )
-        )
-
+                        
+        first_visit = sort_values_first_koalas(merged, 
+                                               by_cols=["visit_occurrence_id_2"], 
+                                               cols=[flag_name, "visit_start_datetime_1"],
+                                               ascending=False
+                                              ).rename(
+            columns={
+                "visit_occurrence_id_1": f"{concept_prefix}STAY_ID",
+                "visit_occurrence_id_2": "visit_occurrence_id",
+            })[[f"{concept_prefix}STAY_ID", "visit_occurrence_id"]]
+        
+                
         return merged, first_visit
 
-    merged, first_contiguous_visit = get_first(merged, contiguous_only=True)
+    merged, first_contiguous_visit = get_first(merged, contiguous_only=True)    
     merged, first_visit = get_first(merged, contiguous_only=False)
-
+    
     # Concatenating merge visits with previously discarded ones
     results = fw.concat(
         [
diff --git a/eds_scikit/utils/sort_first_koalas.py b/eds_scikit/utils/sort_first_koalas.py
new file mode 100644
index 00000000..22355982
--- /dev/null
+++ b/eds_scikit/utils/sort_first_koalas.py
@@ -0,0 +1,6 @@
+def sort_values_first_koalas(dataframe, by_cols, cols, ascending=True):
+    for col in cols:
+        dataframe_min_max = dataframe.groupby(by_cols, as_index=False)[col]
+        dataframe_min_max = dataframe_min_max.min() if ascending else dataframe_min_max.max() 
+        dataframe = dataframe.merge(dataframe_min_max, on=[*by_cols, col], how="right")
+    return dataframe
diff --git a/tests/test_sort_first_koalas.py b/tests/test_sort_first_koalas.py
new file mode 100644
index 00000000..bdb4cdef
--- /dev/null
+++ b/tests/test_sort_first_koalas.py
@@ -0,0 +1,37 @@
+import pandas as pd
+from eds_scikit.utils.sort_first_koalas import sort_values_first_koalas
+from numpy import array
+import pytest
+from eds_scikit.utils import framework
+from eds_scikit.utils.test_utils import assert_equal_no_order
+
+data = {
+    'A': array(['X', 'Y', 'X', 'Y', 'Y', 'Z', 'X', 'Z', 'X', 'X', 'X', 'Z', 'Y', 'Z', 'Z', 'X', 'Y', 'Y', 'Y', 'Y'], dtype='<U1'),
+    'B': array([9, 5, 4, 1, 4, 6, 1, 3, 4, 9, 2, 4, 4, 4, 8, 1, 2, 1, 5, 8]),
+    'C': array([4, 3, 8, 3, 1, 1, 5, 6, 6, 7, 9, 5, 2, 5, 9, 2, 2, 8, 4, 7]),
+    'D': array([8, 3, 1, 4, 6, 5, 5, 7, 5, 5, 4, 5, 5, 9, 5, 4, 8, 6, 6, 1]),
+    'E': array([2, 6, 4, 1, 6, 1, 2, 3, 5, 3, 1, 4, 3, 1, 8, 6, 1, 3, 8, 3])
+}
+
+all_inputs = [
+    pd.DataFrame(data)
+]
+
+all_params = dict(ascending=[True, False], cols=["A", ["A", "B"]], by_cols=[["B", "C", "D"], ["C", "D"]])
+all_params = pd.DataFrame(all_params).to_dict("records")
+
+@pytest.mark.parametrize("module", ["pandas", "koalas"])
+@pytest.mark.parametrize(
+    "params, inputs",
+    [(params, inputs) for params, inputs in zip(all_params, all_inputs)],
+)
+def test_visit_merging(module, params, inputs):
+    print(params)
+    expected_results = inputs.sort_values(params["cols"]).groupby(params["by_cols"]).first().reset_index()
+    inputs = framework.to(module, inputs)
+    results = sort_values_first_koalas(inputs, **params)
+    results = framework.pandas(results)
+
+    assert_equal_no_order(
+        results, expected_results
+    )