Hi, this is my first time posting on this forum. I am new on stata and could really need some help.
The data below is from my database on dogs surgically treated for ortopedic diseases in two different clinics during a 5 y period. I want to compare the counts ("count") of dogs between different breeds in relation to the registered numbers of those particular breeds in the area (collected from the official breed registry database) ("relno")
breed dx relno count
1. Affenpinscher Fractures 21 3
2. Afghan Hound Fractures 8 0
3. Airedale Terrier Fractures 41 1
4. Akita Fractures 5 0
5. Alaskan Malamute Fractures 99 1
In total I have 183 different breeds in the database (all breeds with > 1 registered dog in the official breed registry during the 5 y period), and 99 different breeds presented with at least 1 case of ortopedic surgery, registered by total numbers and the 15 most common diagnoses ("dx"). There is also other variables (mean age, count of males and females, weight by breed/diagnosis) that is not shown here.
I have calculated relative risks manually by dividing the case count for each breed with a particular diagnosis on the total count registered of that breed and compared this to the total expected risk if there was no breed differences (total number of cases with that diagnosis divided by the total number of registered dogs). Using risk ratios breeds with no ortopedic cases, but high registered number of dogs in the area get the same RR as breeds with no orthopedic cases, but few registered number of dogs. I am therefore planning to present the risk differences, but can not figure out how to calculate this with confidence intervals in stata. In addition, I am unsure if I rather should use the registered numbers as a way of standardizing my breed data and if I have used the calculations above correctly (standardized mobility rate ratios).
Any inputs on commands/ how to best compare the relative propotions of the different breeds in the case registry with the proportions of the different breeds in the general dog population in the region will be highly appreciated.
Thanks in advance/Gudrun
The data below is from my database on dogs surgically treated for ortopedic diseases in two different clinics during a 5 y period. I want to compare the counts ("count") of dogs between different breeds in relation to the registered numbers of those particular breeds in the area (collected from the official breed registry database) ("relno")
breed dx relno count
1. Affenpinscher Fractures 21 3
2. Afghan Hound Fractures 8 0
3. Airedale Terrier Fractures 41 1
4. Akita Fractures 5 0
5. Alaskan Malamute Fractures 99 1
In total I have 183 different breeds in the database (all breeds with > 1 registered dog in the official breed registry during the 5 y period), and 99 different breeds presented with at least 1 case of ortopedic surgery, registered by total numbers and the 15 most common diagnoses ("dx"). There is also other variables (mean age, count of males and females, weight by breed/diagnosis) that is not shown here.
I have calculated relative risks manually by dividing the case count for each breed with a particular diagnosis on the total count registered of that breed and compared this to the total expected risk if there was no breed differences (total number of cases with that diagnosis divided by the total number of registered dogs). Using risk ratios breeds with no ortopedic cases, but high registered number of dogs in the area get the same RR as breeds with no orthopedic cases, but few registered number of dogs. I am therefore planning to present the risk differences, but can not figure out how to calculate this with confidence intervals in stata. In addition, I am unsure if I rather should use the registered numbers as a way of standardizing my breed data and if I have used the calculations above correctly (standardized mobility rate ratios).
Any inputs on commands/ how to best compare the relative propotions of the different breeds in the case registry with the proportions of the different breeds in the general dog population in the region will be highly appreciated.
Thanks in advance/Gudrun
Comment