Table 3.
Multimorbidity patterns identified using EFA. Patterns highlighted in bold were included in the main analyses as the number of participants with multimorbidity (n=43,838, 28.4%) who were included in the pattern was greater than 1%
| Pattern | Eigenvalue | Long-term conditions | No. of participants (%) |
|---|---|---|---|
| Pattern 1 | 4.43 | CFS, CKD, epilepsy, Parkinson's disease | 23 (0.05) |
|
Pattern 2 Cardiometabolic |
3.19 | CHD, diabetes, hypertension | 5101 (11.6) |
| Pattern 3 | 2.46 | Hepatitis, liver disease | 27 (0.06) |
| Pattern 4 | 2.11 | Dementia, glaucoma, Meniere's disease, migraine | 77 (0.2) |
|
Pattern 5 Respiratory |
1.84 | Asthma, bronchiectasis, COPD | 727 (1.6) |
| Pattern 6 | 1.48 | Asthma, PCOS, PVD | 94 (0.2) |
|
Pattern 7 Cardio/cerebrovascular |
1.43 | Atrial fibrillation, CHD, stroke/TIA | 621 (1.4) |
|
Pattern 8* Reproductive |
1.35 | Endometriosis, prostate conditions | 3055 (7.0) |
| Pattern 9 | 1.23 | Heart failure, MS | 0 (0.0) |
|
Pattern 10 Pain/gastrointestinal |
1.18 | Dyspepsia, IBS, painful conditions | 4999 (11.4) |
| Pattern 11 | 1.12 | Pernicious anaemia, thyroid problems | 107 (0.2) |
| Pattern 12 | 1.05 | Cancer, osteoporosis | 330 (0.7) |
| Pattern 13 | 1.01 | Connective tissue disorders, psoriasis/eczema | 283 (0.6) |
CFS=chronic fatigue syndrome; CKD=chronic kidney disease; COPD=chronic obstructive pulmonary disorder; IBS=irritable bowel syndrome; MS=multiple sclerosis; PCOS=polycystic ovarian syndrome; PVD=peripheral vascular disease; TIA=transient ischaemic attack
Any male participant with multimorbidity which comprised prostate conditions was included in this pattern; any female with multimorbidity which comprised endometriosis was included in this pattern
Note: Although 12,665 participants fell into one or more of the multimorbidity patterns, the total here of 15,444 participants reflects a certain amount of overlap across patterns. In pattern 9, no participants experienced both conditions constituting the factor. This is likely an artefact of using tetrachoric correlations and an arbitrary factor loading of +/-0.4 to determine groups. It is probable that these two conditions are linked via a third condition that is either unmeasured or included in the analysis but is subthreshold.