Merge pull request #82 from opensafely/paper-fixes

Paper fixes

analysis/report/panel_plots.py

@@ -426,7 +426,7 @@ def get_group_chart(
             )
             # Save the season table only if there is more than one group
             more_than_one_group = panel_group_data.group.nunique() > 1
-            if (produce_season_table or mark_seasons) and more_than_one_group:
+            if (produce_season_table and more_than_one_group) or mark_seasons:
                 season_table = make_season_table(
                     panel_group_data.set_index("date"),
                     "group",

analysis/report/plot_measures_report.py

@@ -102,7 +102,7 @@ def plot_measures(
                 df_copy, category, column_to_plot, output_dir, filename
             )
             # TODO: check whether this works for weekly
-            if season_table and mark_seasons:
+            if not season_table.empty and mark_seasons:
                 annotate_seasons(season_table, column_to_plot, ax)
 
     elif frequency == "week":

analysis/report/report_utils.py

@@ -356,14 +356,17 @@ def make_season_table(df, category, column_to_plot, output_dir, filename=None):
         table.numerator = pd.to_numeric(table.numerator, errors="coerce")
         table.denominator = pd.to_numeric(table.denominator, errors="coerce")
     except ValueError:
-        return None
-    if filename:
-        # If too much is redacted, we cannot make the table
+        table = pd.DataFrame()
+        formatted = pd.DataFrame()
+    # If too much is redacted, we cannot make the table
+    # But project yaml might expect existance of html
+    if filename and not table.empty:
         try:
             formatted = format_season_table(table, column_to_plot, category)
-        except KeyError:
-            return None
-        formatted = reorder_dashes(formatted, level=0)
+            formatted = reorder_dashes(formatted, level=0)
+        except:
+            formatted = pd.DataFrame()
+    if filename:
         formatted.to_html(output_dir / f"{filename}_table.html")
     return table
 

analysis/report/top_5_report.py

@@ -234,7 +234,7 @@ def get_proportion_of_events(df, code, date):
 
 
 def plot_top_codes_over_time(
-    code_df, top_codes, measure, output_dir, frequency
+    code_df, top_codes, measure, output_dir, frequency, xtick_frequency=1
 ):
     """
     Plots the top 5 codes over time for each measure.
@@ -287,6 +287,9 @@ def plot_top_codes_over_time(
             )
             ax.set_xticks(xticks)
             ax.set_xticklabels([x.strftime("%B %Y") for x in xticks])
+            for index, label in enumerate(ax.xaxis.get_ticklabels()):
+                if index % xtick_frequency != 0:
+                    label.set_visible(False)
         elif frequency == "week":
             xticks = pd.date_range(
                 start=code_proportions["date"].min(),
@@ -319,6 +322,7 @@ def main():
     input_file = args.input_file
     output_dir = args.output_dir
     frequency = args.frequency
+    xtick_frequency = args.xtick_frequency
     base_fontsize = args.base_fontsize
 
     output_dir.mkdir(parents=True, exist_ok=True)
@@ -327,6 +331,11 @@ def main():
     measure_table = get_measure_tables(input_file)
 
     all_codes = {**MEDICATION_TO_CODELIST, **CLINICAL_TO_CODELIST}
+    if frequency == "month":
+        for med, codelist in MEDICATION_TO_CODELIST.items():
+            all_codes[f"{med}_with_clinical_any"] = codelist
+        for clin, codelist in CLINICAL_TO_CODELIST.items():
+            all_codes[f"{clin}_with_medication_any"] = codelist
 
     for key, codelist in all_codes.items():
         measure = f"event_code_{key}_rate"
@@ -404,6 +413,7 @@ def main():
             measure=measure,
             output_dir=output_dir,
             frequency=frequency,
+            xtick_frequency=xtick_frequency,
         )
 
 
@@ -426,6 +436,12 @@ def parse_args():
         choices=["month", "week"],
         help="The frequency of the data",
     )
+    parser.add_argument(
+        "--xtick-frequency",
+        help="Display every nth xtick",
+        type=int,
+        default=1,
+    )
     parser.add_argument(
         "--base-fontsize",
         help="Default text size",

analysis/study_definition_report.py

@@ -365,6 +365,13 @@ def generate_expectations_codes(codelist, incidence=0.5):
                     group_by=["population"],
                     small_number_suppression=True,
                 ),
+                Measure(
+                    id=f"event_code_{medication_key}_with_clinical_any_rate",
+                    numerator=f"{medication_key}_with_clinical_any",
+                    denominator="population",
+                    group_by=[f"event_code_{medication_key}"],
+                    small_number_suppression=False,
+                ),
             ]
         )
 
@@ -433,6 +440,13 @@ def generate_expectations_codes(codelist, incidence=0.5):
                     group_by=["population"],
                     small_number_suppression=True,
                 ),
+                Measure(
+                    id=f"event_code_{clinical_key}_with_medication_any_rate",
+                    numerator=f"{clinical_key}_with_medication_any",
+                    denominator="population",
+                    group_by=[f"event_code_{clinical_key}"],
+                    small_number_suppression=False,
+                ),
             ]
         )
 

project.yaml

@@ -65,7 +65,7 @@ actions:
     needs: [curation_weekly]
     outputs:
       moderately_sensitive:
-        measure_csv: output/curation/measure_event_*.csv
+        measure_csv: output/curation/measure_event_*_rate.csv
 
   curation_diagnostics:
     run: cohortextractor:latest generate_cohort
@@ -267,6 +267,7 @@ actions:
       --input-file output/report/results/measure_unrounded.csv
       --output-dir output/report/results
       --frequency month
+      --xtick-frequency 2
     needs: [join_measures_unrounded]
     outputs:
       moderately_sensitive:
@@ -637,15 +638,16 @@ actions:
       --output-name "phenoxymethylpenicillin_with_clinical_any_by_subgroup"
       --exclude-group "Missing"
       --scale "rate"
-      --xtick-frequency 3
+      --xtick-frequency 2
       --produce-season-table
       --base-fontsize 12
       --columns 1
     needs: [join_measures_rounded]
     outputs:
       moderately_sensitive:
         measure: output/report/results/paper/phenoxymethylpenicillin_with_clinical_any_by_subgroup.png
-        tables: output/report/results/paper/event_phenoxymethylpenicillin_with_clinical_any_age_band_rate_table.html
+        table_age: output/report/results/paper/event_phenoxymethylpenicillin_with_clinical_any_age_band_rate_table.html
+        table_region: output/report/results/paper/event_phenoxymethylpenicillin_with_clinical_any_region_rate_table.html
 
   supplemental_antibiotics_groups:
     run: >