Participants’ characteristics and follow-up
Figure 1 shows the participant selection procedures, and the baseline
characteristics of all study participants with follow-up information are
presented in Table 1. 105,199 cases started treatment at age 46 to 65
years in 1984−2017 and there were 224,643 matched controls. The mean (±
SD) age of all participants at first treatment was 53 (± 5.02) years.
The mean duration of HRT use was 6.0 (± 4.8) years. Among cases, 17,606
(17%) received oestrogen-only and 87,593 (83%) received combined
therapy. Around 75% of cases were prescribed first HRT between 46 and
55 years of age.
Controls have more missing records than the cases. Incomplete medical
records were more common in earlier birth cohorts, as expected from
previous research that showed great improvement in recording after the
initiation of the Quality and Outcomes Framework (QoF) in
2004.39 The prevalence of selected medical conditions
was nearly the same for cases and controls at baseline. However, there
were more oophorectomies and hysterectomies among cases (Table 1), and
more participants with osteoporosis among first HRT starters in the
older age group at baseline (Table S3). These conditions are more
prevalent among HRT users as these are often the cause of HRT
treatment.40 There were more obese controls, compared
to more healthy weight and overweight cases. The proportions of
ex-smokers and current smokers were slightly higher in cases. More than
half of the participants had high socio-economic status (Townsend index
1-2). In earlier birth cohorts more women took oestrogen-only HRT, and
in later birth cohorts more women took combined HRT.
The length of study follow-up was
32 years, and the mean follow-up of participants was 13.5 (SD ±7.0)
years.
The
interquartile range of participant follow-up was 10.8 years. During
follow-up 21,751 women died in total, of whom 6,329 (6%) were cases,
and 15,422 (7%) were controls. 44 cases died per 10,000
participant-years follow-up compared to 63 controls.
29,604 (28%) cases and 61,023
(27%) controls were transferred out during follow-up.