This  presentation  is  about  ecological  studies. It's  narrated  by  Dr.  Ruth  McClellan, drawing  on  material  created  by  Dr. An  Inca  housed  harbour  smart  and  Dr.  Ruth  McMillan. This  is  a  brief  summary  of  what  an  ecological  study  is, what  it  is  useful  for, what  its  strengths  are, and  what  are  its  limitations, what  to  look  out  for  when  you're  reading  or critically  appraising  an  ecological  study. We'll  also  look  at  some  examples  of  ecological  studies. An  ecological  study  is a  study  that  looks  for  a  correlation  between an  exposure  and  an  outcome  by comparing  aggregate  data  from  different  populations. There  are  some  examples  on  the  slide. These  populations  might  be geographic  or  they  might  be  occupational, or  they  might  compare  data  at  different  time  points. Unlike  all  the  other  epidemiological  study designs  that  we  will  look  at, the  units  of  comparison  is the  population,  not  the  individual. It  might  be  easier  to  understand this  by  looking  at  an  example. The  image  on  the  slide  shows national  colon  cancer  incidence and  per  capita  meat  consumption for  11  selected  countries. Each  of  the  11  countries  contributes two  data  points  to  this  analysis. Their  national  incidence  of  colon  cancer  and their  average  meat  consumption in  grammes  per  head  of  population. It  looks  as  if  there's  a  strong  correlation  between meat  consumption  and  the  incidence  of  colon  cancer. The  higher  a  country's  meat  consumption, The  higher  incidence  of  colon  cancer. But  it's  important  to  remember that  correlation  is  not  the  same  as  causation. We  cannot  conclude  from  this  graph that  high  meat  consumption  causes  colon  cancer. There  are  two  reasons  for  this. Firstly,  these  data  simply  show us  national  average  meat  consumption. We  don't  know  anything  about the  meat  consumption  of  individuals  in  these  countries. Just  because  average  meat  consumption in  a  given  country  is  high, it  doesn't  mean  that  everyone living  in  that  country  eats  a  lot  of  meat. It  could  be  that  most  of  the  meat  consumption  is concentrated  in  a  small  subset  of  the  population. Crucially,  we  don't  know  anything  about  the  levels of  meat  consumption  of  the  people  with  colon  cancer. Perhaps  they  don't  eat  meat  at  all. The  second  problem  with ecological  studies  is  that  they  are very  prone  to  confounding. It's  possible  that  there  is some  other  factor  that  we  don't  know about  that  causes  colon  cancer and  is  correlated  with  high  meat  consumption. E.g.  perhaps  countries  with  high  levels of  meat  consumption  also  consume high  levels  of  processed  foods  or have  low  average  levels  of  physical  activity. We  just  don't  know  from  looking  at  these  data. Ecological  studies  have  many  advantages because  they  use  routine  data. They're  much  quicker  and  cheaper  and  easier  to  conduct than  studies  that  involve collecting  data  from  individuals. They  are  really  good  as  a  starting  point  for an  emerging  area  of research  when  people  are  just  beginning  to  become  aware, that's  a  particular  issue  might  be  important. They're  good  for  generating  hypotheses  that  can  then  be tested  by  more  robust  methods based  on  individual  level  data. But  they  have  two  very  significant  drawbacks, which  mean  that  the  results  of an  ecological  study  should  always be  treated  with  great  caution. Firstly,  there's  a  major  risk  of  confounding, as  we  saw  in  the  previous  example. And  secondly,  we  cannot  draw  conclusions  about an  individual  based  solely  upon  the  group  they  belong  to. This  is  called  the  ecological  fallacy. Going  back  to  the  previous  example, we  cannot  conclude  on  the  basis  of  these  data  that the  people  with  colon  cancer  consume  a  high  meat  diet. Because  the  data  in  this  study  are  population  level  data, we  cannot  conclude  anything  at  all  about the  individuals  in  these  populations. John  Snow's  Cholera  study  is an  early  example  of  an  ecological  or  correlation  study. He  found  that  cholera  deaths tended  to  cluster  around  the  Broad  Street  water  pump. In  other  words,  he  found  a  correlation  between cholera  deaths  and  household  proximity to  this  particular  pump. This  was  an  ecological  study  because  he did  not  collect  data  at  the  level  of  individuals. Although  he  did  ascertain the  source  of  each  household  swatter, he  did  not  go  around  and interview  the  relatives  of  those  who  had died  to  ascertain  whether  they had  actually  drunk  water  from  the  pump. Instead,  he  showed  that  mortality  rates  from cholera  were  higher  in  the  population  of people  living  in  households which  used  the  Broad  Street  pump compared  with  the  population  of  people  living  in households  which  did  not  rely  on  the  pump. The  ecological  study  is  the  only  study  design  will  look at  that  collects  data  at  the  level  of  populations. All  of  the  other  study  designs  shown  in the  taxonomy  diagram  involve collecting  data  from  individuals. So  they  provide  much  stronger  evidence  of an  association  between  an  outcome  and  an  exposure. It's  possible  with  an  ecological  study. And  finally,  this  is  a  more  recent  example of  an  ecological  study. It's  a  review  that  was  published  in  2016. And  it's  interesting  because  it's a  good  illustration  of  how useful  ecological  studies  can  be. It's  about  the  link  between  dietary  factors  and Alzheimer's  disease  is  based  on  data  from  ten  countries. Average  meat  intake  for each  country  and  the  prevalence  of Alzheimer's  disease  in  each  country  were derived  from  routinely  available  data. The  researchers  found  a  strong  correlation. Countries  with  high  average  meat  intake, intake  also  had  a  high  prevalence  of  Alzheimer's  disease. So  this  ecological  study  provided an  early  indication  of  this  association, which  was  later  confirmed  by much  more  detailed  observational  studies, which  looked  at  individual  level  data.