Surveillance Problems
Yesterday, part of the surveillance team got together (Hasifa, Shereen, Allen, and I) with Heidi, and sat down to look at the surveillance data that we had. The reason why we sat down in the first place was because Heidi had been looking at some of the data that she had collected (her hired staff was collecting some duplicate data from patients) and the data that the surveillance team had gotten from the site (where the normal clerk was writing down data as part of his job, which we photocopied then entered into a computer here in Kampala). Heidi had noticed that there was a severe difference between what we had recorded and what she had recorded, and wanted to make sure that what she was seeing wasn't just a random mistake.
We sat down together in Hasifa's office, and started looking through the data- for the most part, it didn't seem too bad, but at one point, we had to go back to our hard copy records and Heidi had to pull out her digital camera photos of the data as well... and we found some very disturbing results. Our photocopies of the log books had been done relatively early in the month, and when compared to Heidi's photographs of the same log books (taken later in the month) we saw that a lot of data had been filled in on Heidi's photos that had not been on our photocopies. We verified this wasn't just an isolated incident, and had occurred on multiple days...
This type of thing wouldn't have been as bad if it were just one or two observations every page or so that was filled in, but we saw pages where almost 80% of the data had miraculously appeared later in the month... and if it was really made up, then the analysis that we're doing on trends in malaria burden in the area are probably bogus. It's bringing up lots of questions of what can be done, and what data is legit and what data is not legit.
Fortunately, a good number of the variables we collected are still valid- lab results, total number of visits, ages, all those variables are fine. But the main issue is what diagnosis and treatment these patients are getting. A lot of these patients go through the health facility carrying around little pieces of paper with them, and after getting seen by the medical officer and being diagnosed and prescribed a treatment, they're supposed to go back to the main clerk to get the diagnosis and treatment written down. It seems that either a lot of people are skipping this step or the clerk is just taking the afternoon off at times, or something else is going on.
We've speculated why this might be happening:
1) The guy doesn't care about his job, he's not getting paid too well, and if he leaves his desk and doesn't record a lot of the information, he doesn't really care. He might go back and fill it in to cover for being a slacker.
2) He cares a whole about his job and a lot of patients don't come back, and since he knows that we really really want complete data, he's trying not to let the surveillance project down, and is just making up stuff that sounds plausible, or looking at old books and copying down the diagnoses and treatments given on a different day.
3) Any number of other reasons.
One thing that Heidi said was that she could actually remember some of the patients that came in on those days, and that in particular one patient who was "diagnosed" with a broken clavicle very obviously was not in any pain in their shoulders/neck when Heidi saw them... the guy who's entering this data is pretty creative at least in his diagnoses and pretty competent at giving plausible treatments.
In any case, we have to decide what to do to fix or not fix this situation, but at the same time, we have to move to the southwestern part of the country to start surveillance there. It's especially critical that we don't get falsified data down there because another group is relying on us collecting good diagnosis and treatment data (they're doing a training program, and want to see if trained people do a better job of giving proper treatments for certain diagnoses, etc.).
We leave tomorrow morning at 6:30 to drive the 8 hours to Kabale and Kanungu, and will be back on Saturday (early morning departure from Kanungu, and hope to be back in Kampala by mid-afternoon. We had better be back, because Paula is arriving on Saturday night, and I have to be in Entebbe with Nuhu to pick her up).
In happier news, we have a new surveillance data entry person- his name is Elvis, and he wears cool looking alligator skin cowboy boots (only the boots go up to just above his ankles, so they're not quite knee high cowboy boots).
We sat down together in Hasifa's office, and started looking through the data- for the most part, it didn't seem too bad, but at one point, we had to go back to our hard copy records and Heidi had to pull out her digital camera photos of the data as well... and we found some very disturbing results. Our photocopies of the log books had been done relatively early in the month, and when compared to Heidi's photographs of the same log books (taken later in the month) we saw that a lot of data had been filled in on Heidi's photos that had not been on our photocopies. We verified this wasn't just an isolated incident, and had occurred on multiple days...
This type of thing wouldn't have been as bad if it were just one or two observations every page or so that was filled in, but we saw pages where almost 80% of the data had miraculously appeared later in the month... and if it was really made up, then the analysis that we're doing on trends in malaria burden in the area are probably bogus. It's bringing up lots of questions of what can be done, and what data is legit and what data is not legit.
Fortunately, a good number of the variables we collected are still valid- lab results, total number of visits, ages, all those variables are fine. But the main issue is what diagnosis and treatment these patients are getting. A lot of these patients go through the health facility carrying around little pieces of paper with them, and after getting seen by the medical officer and being diagnosed and prescribed a treatment, they're supposed to go back to the main clerk to get the diagnosis and treatment written down. It seems that either a lot of people are skipping this step or the clerk is just taking the afternoon off at times, or something else is going on.
We've speculated why this might be happening:
1) The guy doesn't care about his job, he's not getting paid too well, and if he leaves his desk and doesn't record a lot of the information, he doesn't really care. He might go back and fill it in to cover for being a slacker.
2) He cares a whole about his job and a lot of patients don't come back, and since he knows that we really really want complete data, he's trying not to let the surveillance project down, and is just making up stuff that sounds plausible, or looking at old books and copying down the diagnoses and treatments given on a different day.
3) Any number of other reasons.
One thing that Heidi said was that she could actually remember some of the patients that came in on those days, and that in particular one patient who was "diagnosed" with a broken clavicle very obviously was not in any pain in their shoulders/neck when Heidi saw them... the guy who's entering this data is pretty creative at least in his diagnoses and pretty competent at giving plausible treatments.
In any case, we have to decide what to do to fix or not fix this situation, but at the same time, we have to move to the southwestern part of the country to start surveillance there. It's especially critical that we don't get falsified data down there because another group is relying on us collecting good diagnosis and treatment data (they're doing a training program, and want to see if trained people do a better job of giving proper treatments for certain diagnoses, etc.).
We leave tomorrow morning at 6:30 to drive the 8 hours to Kabale and Kanungu, and will be back on Saturday (early morning departure from Kanungu, and hope to be back in Kampala by mid-afternoon. We had better be back, because Paula is arriving on Saturday night, and I have to be in Entebbe with Nuhu to pick her up).
In happier news, we have a new surveillance data entry person- his name is Elvis, and he wears cool looking alligator skin cowboy boots (only the boots go up to just above his ankles, so they're not quite knee high cowboy boots).
posted by Vince at 3:57 AM
0 Comments:
Post a Comment
<< Home