Counting Events within Time Period: Seeking Code Assistance

[Fouziah Almouqati]

Hi all,

I am a very new Stata user and would like to seek your help with a code. I need to create a new variable that includes the count of presentations in the previous 4 hours of the current presentation. Below is a description of my data:

Each record represents a patient presentation. For each presentation, the presentation date and time are captured under the variable `pre_d_t`, already in the correct Stata date format and sorted. I need to count the number of presentations prior to the current record. To illustrate, below is an example where I manually did the calculation. Note: It does not matter if the presentation is for the same or different individuals.

fake_ID	pre	pre_d_t	previ_pres
34887	1	20/01/2015 14:53	0
26689	1	28/01/2015 10:47	0
13665	1	28/01/2015 10:47	1
2856	1	30/01/2015 10:53	0
86893	1	30/01/2015 10:57	1
35085	1	3/02/2015 7:04	0
7111	1	3/02/2015 7:08	1
32337	1	3/02/2015 7:13	2
55510	1	3/02/2015 7:21	3
87599	1	3/02/2015 7:50	4

Below is also an example generated by Dataex:

* Example generated by -dataex-. For more info, type help dataex
clear
input float(fake_ID pre pre_d_t)
34887 1 1.7373848e+12
26689 1 1.7380613e+12
13665 1 1.7380613e+12
 2856 1 1.7382344e+12
86893 1 1.7382347e+12
35085 1 1.7385663e+12
 7111 1 1.7385665e+12
32337 1 1.7385668e+12
55510 1 1.7385673e+12
87599 1  1.738569e+12
20471 1 1.7385695e+12
89276 1   1.73857e+12
58447 1 1.7385703e+12
36978 1 1.7385703e+12
85063 1 1.7385707e+12
39138 1  1.738571e+12
11966 1 1.7385714e+12
75424 1 1.7385716e+12
69502 1 1.7385727e+12
68662 1 1.7385727e+12
93193 1  1.738573e+12
45489 1  1.738573e+12
 6740 1 1.7385732e+12
33799 1 1.7385737e+12
97488 1 1.7385737e+12
72644 1  1.738574e+12
 4542 1  1.738574e+12
74597 1 1.7385757e+12
49613 1 1.7385765e+12
71672 1 1.7385768e+12
85974 1 1.7385773e+12
13408 1  1.738579e+12
48844 1 1.7385794e+12
87122 1 1.7385798e+12
76647 1   1.73858e+12
25126 1   1.73858e+12
16636 1 1.7385804e+12
74380 1  1.738581e+12
98051 1  1.738581e+12
72958 1  1.738581e+12
90110 1 1.7385815e+12
26436 1 1.7385815e+12
88565 1 1.7385817e+12
88211 1  1.738582e+12
74893 1 1.7385823e+12
91963 1 1.7385823e+12
69345 1 1.7385827e+12
21540 1  1.738583e+12
82859 1  1.738583e+12
 4422 1  1.738583e+12
86304 1 1.7385837e+12
35260 1 1.7385844e+12
77204 1 1.7385845e+12
58612 1 1.7385848e+12
32278 1  1.738585e+12
17293 1  1.738585e+12
80536 1 1.7385857e+12
30600 1 1.7385865e+12
21910 1  1.738587e+12
72473 1  1.738587e+12
69649 1 1.7385872e+12
91193 1 1.7385874e+12
67956 1  1.738588e+12
35494 1 1.7385887e+12
73897 1 1.7385887e+12
18740 1 1.7385892e+12
31461 1 1.7385895e+12
13757 1 1.7385897e+12
65377 1   1.73859e+12
27013 1   1.73859e+12
89984 1 1.7385908e+12
57342 1 1.7385908e+12
11147 1 1.7385908e+12
41452 1  1.738591e+12
  305 1  1.738591e+12
66600 1 1.7385914e+12
34629 1 1.7385916e+12
 7802 1  1.738592e+12
12758 1 1.7385922e+12
22970 1 1.7385927e+12
32955 1  1.738593e+12
41441 1 1.7385943e+12
 3608 1 1.7385947e+12
 8438 1  1.738595e+12
  988 1  1.738595e+12
32004 1  1.738595e+12
  520 1 1.7385952e+12
22754 1 1.7385956e+12
85147 1  1.738596e+12
98201 1 1.7385965e+12
 3248 1 1.7385965e+12
98748 1  1.738597e+12
89411 1 1.7385972e+12
96847 1 1.7385973e+12
23922 1 1.7385976e+12
69273 1  1.738598e+12
48844 1 1.7385983e+12
43765 1 1.7385988e+12
58580 1  1.738599e+12
37871 1 1.7385994e+12
end
format %tc pre_d_t

Any help, advice, or reply would be greatly appreciated.


Regards,
Fouziah

[Nick Cox]

Date-times should always be stored as doubles. In your case, results are just in minutes.

Consider rangestat (from SSC, and cited in many threads here.

.

rangestat (sum) pre, int(pre_d_t -14400000 -1)

[Fouziah Almouqati]

The code gave me exactly what I needed. Thanks, Nick, for introducing me to rangestat and for pointing out the date format issue.