NCD Combined analytical dataset

Main questionnaire data

For Karonga data ncdk_personal, ncdk_pershh and ncdk_persexam are merged on stid to ncdk_fieldlog, which provides the link to the ident, date of birth is merged in from gen_identity using ident. If a person appears in more than one household their first interview is kept. Marital status was not asked in Karonga so the nearest marital status report from css_seird1_8 or css_sei is merged into the NCD dataset using ident. The number of adults and children living in the house was also not asked in Karonga, so these are calculated using crs_memball by identifying people with an residency episode in the house which includes the NCD data, calculating their age (using the date of birth from gen_identity) and counting the people in different age groups: these summary variables are then merged into the NCD data on the house variable. For Lilongwe data ncdl_identity, ncdl_personal, ncdl_pershh and ncdl_persexam are merged on stid to ncdl_fieldlog, and ncdl_hhf is merge in on lhhno. If a person appears in more than one household their first interview is kept. The Karonga and Lilongwe data are then appended together.

NCD substudies

The sub-study tables (ncdk_hvf, ncdl_hvf, ncdk_cfb, ncdl_cfb, ncdk_cff, ncdl_cff, ncdk_cac, ncdl_cac, ncdk_opf, ncdl_opf, ncdk_bpf, ncdl_bpf, ncdl_dsf, ncdk_psf, ncdl_psf, ncdl_rsf) are summarised by stid to create variables indicating how many records each stid has in the study (if any). These summary variables are then merged into the main NCD dataset using stid.

Lab results

For Karonga basic information on all specimens taken, including date of collection, type of specimen etc., are held on ser_specimen. Results of the test carried out for NCD are held in the tables ser_glucose, ser_lipids, ser_fbc, ser_hba1c, ser_electrolytes, these also hold results from other studies so are (individually) merged to ser_specimen using labno (stid is held on ser_specimen). The glucose results are reshaped into separate variables for fasted, non-fasted and glucose tolerance test results. In Lilongwe the same process is undertaken (the tables have the prefix 'lab' rather than 'ser'), and Karonga and Lilongwe data are appended together, before merging into the main NCD dataset using stid.

HIV/ART data

In Karonga, only people who had not be tested recently were offered an HIV test, so the HIV and ART statuses are calculated using many other sources of data (actual HIV tests, self-reported data and data from clinics) which is summarised in the combined dataset hivart_kps_known_pop as a series of dates (first negative test, first positive test, first ART start etc.). These date of each person's NCD interview is merged into this dataset using ident, and their HIV & ART status at the time of the interview calculated. For Lilongwe, lab_rtr (rapid test results) is merged to lab_specimen using specno, and gp_rtfl (questionnaire data regarding HIV/ART status) merged in using stid. HIV and ART status are calculated using self-reported and test data. The Karonga and Lilongwe data are appended together and merged into the main NCD dataset using stid.