diff --git a/ehrql/main.py b/ehrql/main.py
index 5df83fd25..95782094c 100644
--- a/ehrql/main.py
+++ b/ehrql/main.py
@@ -133,11 +133,8 @@ def generate_dataset_with_dummy_data(
         query_engine = LocalFileQueryEngine(dummy_tables_path)
         results = query_engine.get_results(variable_definitions)
     else:
-        generator = get_dummy_data_class(dummy_data_config)(
-            variable_definitions,
-            population_size=dummy_data_config.population_size,
-            timeout=dummy_data_config.timeout,
-        )
+        generator_class, kwargs = get_dummy_data_class(dummy_data_config)
+        generator = generator_class(variable_definitions, **kwargs)
         results = generator.get_results()
 
     log.info("Building dataset and writing results")
@@ -150,11 +147,8 @@ def create_dummy_tables(definition_file, dummy_tables_path, user_args, environ):
     variable_definitions, dummy_data_config = load_dataset_definition(
         definition_file, user_args, environ
     )
-    generator = get_dummy_data_class(dummy_data_config)(
-        variable_definitions,
-        population_size=dummy_data_config.population_size,
-        timeout=dummy_data_config.timeout,
-    )
+    generator_class, kwargs = get_dummy_data_class(dummy_data_config)
+    generator = generator_class(variable_definitions, **kwargs)
     table_data = generator.get_data()
 
     directory, extension = split_directory_and_extension(dummy_tables_path)
@@ -168,9 +162,14 @@ def create_dummy_tables(definition_file, dummy_tables_path, user_args, environ):
 
 def get_dummy_data_class(dummy_data_config):
     if dummy_data_config.legacy:
-        return DummyDataGenerator
+        kwargs = {
+            "population_size": dummy_data_config.population_size,
+            "timeout": dummy_data_config.timeout,
+        }
+        return DummyDataGenerator, kwargs
     else:
-        return NextGenDummyDataGenerator
+        kwargs = {"configuration": dummy_data_config}
+        return NextGenDummyDataGenerator, kwargs
 
 
 def dump_dataset_sql(
diff --git a/tests/acceptance/test_embedded_study.py b/tests/acceptance/test_embedded_study.py
index 8a6212d77..567d3711c 100644
--- a/tests/acceptance/test_embedded_study.py
+++ b/tests/acceptance/test_embedded_study.py
@@ -186,16 +186,20 @@ def test_generate_dummy_data_with_dummy_tables(study, tmp_path):
 
 
 @pytest.mark.parametrize(
-    "dataset_definition_fixture",
+    "dataset_definition_fixture,expected_columns",
     (
-        trivial_dataset_definition,
-        trivial_dataset_definition_legacy_dummy_data,
+        # this dataset definition includes date_of_death
+        # in the additional population constraints only
+        (trivial_dataset_definition, "patient_id,date_of_birth,date_of_death"),
+        (trivial_dataset_definition_legacy_dummy_data, "patient_id,date_of_birth"),
     ),
 )
-def test_create_dummy_tables(study, tmp_path, dataset_definition_fixture):
+def test_create_dummy_tables(
+    study, tmp_path, dataset_definition_fixture, expected_columns
+):
     dummy_tables_path = tmp_path / "subdir" / "dummy_data"
     study.setup_from_string(dataset_definition_fixture)
     study.create_dummy_tables(dummy_tables_path)
     lines = (dummy_tables_path / "patients.csv").read_text().splitlines()
-    assert lines[0] == "patient_id,date_of_birth"
+    assert lines[0] == expected_columns
     assert len(lines) == 11  # 1 header, 10 rows