diff --git a/pandas/core/frame.py b/pandas/core/frame.py
index 82ae93ad31763..8dc838a5d61a5 100644
--- a/pandas/core/frame.py
+++ b/pandas/core/frame.py
@@ -9523,7 +9523,6 @@ def _append(
         verify_integrity: bool = False,
         sort: bool = False,
     ) -> DataFrame:
-        combined_columns = None
         if isinstance(other, (Series, dict)):
             if isinstance(other, dict):
                 if not ignore_index:
@@ -9536,8 +9535,6 @@ def _append(
                 )
 
             index = Index([other.name], name=self.index.name)
-            idx_diff = other.index.difference(self.columns)
-            combined_columns = self.columns.append(idx_diff)
             row_df = other.to_frame().T
             # infer_objects is needed for
             #  test_append_empty_frame_to_series_with_dateutil_tz
@@ -9563,16 +9560,6 @@ def _append(
             verify_integrity=verify_integrity,
             sort=sort,
         )
-        if (
-            combined_columns is not None
-            and not sort
-            and not combined_columns.equals(result.columns)
-        ):
-            # TODO: reindexing here is a kludge bc union_indexes does not
-            #  pass sort to index.union, xref #43375
-            # combined_columns.equals check is necessary for preserving dtype
-            #  in test_crosstab_normalize
-            result = result.reindex(combined_columns, axis=1)
         return result.__finalize__(self, method="append")
 
     def join(