SELECT observation_year, 
       age_group, 
       gender, 
       count(*) AS num_people
  FROM -- person_id, observation_year
     ( SELECT DISTINCT person_id
            , EXTRACT( YEAR from observation_period_start_date ) AS observation_year
         FROM observation_period
     )
  JOIN -- person, gender, age_group
     ( SELECT person_id, gender
            , CAST( FLOOR( age / 10 ) * 10 AS VARCHAR ) ||' to '
              || CAST( ( FLOOR( age / 10 ) * 10 ) + 9 AS VARCHAR ) AS age_group
         FROM /* person, gender, age */
            ( SELECT person_id, NVL( concept_name, 'MISSING' ) AS gender, year_of_birth
                   , extract( YEAR FROM first_observation_date ) - year_of_birth AS age
                FROM -- person, gender_id, first observation date
                   ( SELECT person_id, gender_concept_id, year_of_birth
                          , min( observation_period_start_date ) AS first_observation_date
                       FROM observation_period
                       JOIN person USING( person_id )
                      GROUP BY person_id, gender_concept_id, year_of_birth
                   )
                LEFT OUTER JOIN vocabulary.concept ON concept_id = gender_concept_id
               WHERE year_of_birth IS NOT NULL
            )
       WHERE age >= 0
     ) USING( person_id )
 GROUP BY observation_year, age_group, gender
 ORDER BY observation_year, age_group, gender