Table 4

Relative Association Between the Presence of Individual Conditions on Time to Developing Incident Multimorbidity Between 2002-2003 and 2012-2013: Complementary Log-Log Model

Condition	HR (95% CI)	P Value
Hypertension	1.00 [Reference]	…
Ischemic heart disease	1.27 (0.91-1.78)	.162
Arrhythmia	1.55 (1.06-2.26)^a	.024^a
Diabetes mellitus	1.06 (0.63-1.78)	.839
COPD	2.32 (1.55-3.46)^a	<.001^a
Asthma	1.33 (1.05-1.70)^a	<.019^a
Arthritis (any)	0.98 (0.83-1.16)	.819
Osteoporosis	1.32 (0.87-2.01)	.185
Cancer (any)	1.19 (0.86-1.63)	.295
Affective mental health condition (any)	0.97 (0.71-1.31)	.831
Other^b	1.21 (0.64-2.30)	.552

COPD = chronic obstructive pulmonary disease; HR = hazard ratio.
Note: From a sample of participants having 1 condition in 2002-2003 (n = 1,534), and an outcome of 1 or more incident conditions. Complementary log-log models are discrete time equivalent of Cox proportional hazards models for continuous time. This model adjusted for all baseline patient characteristics, and differential nonresponse using longitudinal sample weighting. HR >1.00 represents increased risk; an HR of <1.00 indicates reduced risk.
↵a Conditions significantly more more likely than hypertension to be associated with incident multimorbidity.
↵b Includes participants with conditions with low prevalence; congestive heart failure, stroke, psychotic mental health condition, and dementia/Alzheimer’s (Table 2). Combined, these individuals in the “other” category accounted for <1% of the sample.