10-Year Multimorbidity Trajectories in Older People Have Limited Benefit in Predicting Short-Term Health Outcomes in Comparison to Standard Multimorbidity Thresholds: A Population-Based Study

Figures & data

Figure 1 Illustration of the study periods used to identify multimorbidity trajectories and to perform outcome prediction.

Abbreviations: ≥2 CCs, ≥3 CCs, ≥4 CCs: Multimorbidity indicators of the presence of at least 2, 3 or 4 chronic diseases; CCs: Chronic conditions.

Table 1 Characteristic of Individuals in Each Multimorbidity Trajectory, and in the Total Study Population, Québec, Canada, n = 99,411

	Trajectories								Study population
	Stable 0 CC	Decrease 0 CC	Increase 1 CC	Stable 1 CC	Stable 2 CCs	Fast increase 4 CCs	Increase 4 CCs	Increase 7 CCs
Number of individuals	10,510	3205	6907	26,818	28,426	3269	16,398	3878	99,411
Number of CC at end of 10-year period, mean (SD)	0.1 (0.3)	0.1 (0.2)	1.5 (0.9)	1.3 (0.8)	2.5 (1.0)	4.1 (1.3)	4.5 (1.4)	6.9 (1.8)	2.3 (1.9)
Increase in number of CC over the 10-year period^a, mean (SD)	0.1 (0.3)	−0.9 (0.6)	1.4 (0.9)	0.3 (1.1)	0.5 (1.5)	3.7 (1.2)	1.3 (2.1)	1.8 (2.6)	0.7 (1.7)
Age at end of 10-year period, mean (SD)	71.2 (4.9)	71.9 (5.5)	72.3 (5.7)	73.5 (6.3)	75.9 (7.1)	75.2 (7.1)	78.1 (7.6)	79.7 (7.7)	74.8 (7.0)
Female (%)	47.0	53.0	48.4	52.9	57.1	46.7	58.0	60.7	54.1
Material deprivation quintile at end of 10-year period^b (%)
Q1: less deprived	21.5	22.3	20.1	20.5	18.9	18.2	17.2	14.5	19.4
Q2	19.5	18.0	19.1	19.6	18.2	19.3	16.9	15.9	18.5
Q3	19.2	20.1	19.8	20.2	19.9	18.8	20.6	20.9	20.0
Q4	20.1	19.5	20.0	20.1	21.3	20.7	21.5	22.7	20.7
Q5: most deprived	19.7	20.1	21.0	19.6	21.7	23.1	23.9	26.0	21.3
Mortality (%)	0.8	0.7	1.9	1.4	2.6	4.9	5.4	11.1	2.9
Hospitalisation (≥1/year)^c (%)	4.4	5.3	8.2	7.6	11.7	17.0	19.6	32.5	11.6
Emergency Unit (≥3/year)^c (%)	1.2	1.5	2.5	2.6	4.3	6.4	8.2	19.1	4.5
Specialiste (≥10/year)^c (%)	4.1	4.5	8.6	8.2	13.0	18.9	21.9	32.7	12.5
General practitioner (≥7/year)^c (%)	1.7	3.6	4.5	5.2	9.8	11.1	16.7	27.8	8.9
Polypharmacy (≥10/year)^c (%)	4.5	9.7	15.2	21.1	43.3	47.7	70.4	89.1	36.3
Number of drugs, mean (SD)^c	3.0 (3.2)	4.4 (3.9)	5.6 (4.2)	6.6 (4.4)	9.4 (5.0)	10.0 (5.5)	13.0 (6.0)	17.8 (7.3)	8.4 (6.0)

Notes: ^aProportion calculated among individuals without missing value on the number of chronic conditions in 2008 (Number of individuals included by trajectory: trajectory 1: n=10,106; 2: n=3193; 3: n=26,452; 4:n=6747; 5: n=28,426; 6: n=3221; 7: n=16,338; 8: n=3872). ^b Proportion calculated among individuals without missing value on material deprivation on April 1, 2019 (Number of individuals included by trajectory: trajectory 1: n=9801; 2: n=2989; 3: n=24,551; 4: n=6410; 5: n=25,252; 6: n=2901; 7: n=13,891; 8: n=3100). ^c Proportion calculated among individuals with complete drug insurance coverage during the one year follow-up (April 1, 2019 to March 31, 2020) (Number of individuals included by trajectory: trajectory 1: n=9319; 2: n=2805; 3: n=23,859; 4: n=6116; 5: n=25,433; 6: n=2853; 7: n=14,307; 8: n=3198).

Abbreviations: CC, chronic condition; Q1-Q5, First quintile - Fifth quintile; SD, standard deviation.

Table 2 Statistical Assessment of the Optimal Number of Trajectories from Latent Class Growth Analysis Models Based on Counts of Chronic Conditions Across the 10 Periods

Statistic	Number of trajectories
	2	3	4	5	6	7	8	9	10
BIC	3,021,335	2,843,389	2,783,447	2,767,115	2,749,090	2,740,220	2,735,442	2,732,890	2,731,395
LMRT p-value	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001	1
Entropy	91.2%	91.7%	87.8%	83.6%	83.6%	84.0%	82.7%	80.3%	79.8%
APP of each trajectory	0.973 0.977	0.979 0.960 0.957	0.962 0.909 0.935 0.934	0.863 0.904 0.968 0.860 0.909	0.846 0.896 0.898 0.859 0.974 0.884	0.856 0.876 0.840 0.974 0.896 0.870 0.904	0.720 0.810 0.856 0.892 0.865 0.970 0.859 0.898	0.870 0.758 0.721 0.890 0.861 0.970 0.848 0.791 0.838	0.881 0.746 0.699 0.824 0.906 0.660 0.967 0.873 0.781 0.834
OCC of each trajectory	34.5 44.6	221.3 21.5 52.4	149.3 15.8 99.8 27.2	12.8 175.2 193.8 23.1 26.3	11.7 174.7 104.1 25.8 303.1 22.3	29.2 207.0 11.9 315.9 112.7 17.6 232.0	75.5 10.6 30.3 111.4 17.3 272.6 184.2 216.9	551.0 9.5 75.7 110.9 193.6 272.7 16.9 16.2 74.4	103.1 14.8 90.6 65.2 311.6 90.5 247.1 618.0 10.1 15.2
Fraction of study population in each trajectory	0.512 0.488	0.174 0.528 0.298	0.145 0.387 0.126 0.342	0.329 0.051 0.134 0.210 0.275	0.319 0.047 0.078 0.191 0.110 0.255	0.169 0.033 0.307 0.106 0.071 0.275 0.039	0.033 0.286 0.164 0.069 0.270 0.106 0.032 0.039	0.012 0.248 0.033 0.068 0.031 0.106 0.248 0.189 0.065	0.067 0.166 0.025 0.067 0.030 0.021 0.106 0.011 0.261 0.249

Abbreviations: BIC, Bayesian information criteria; APP, Average posterior probability; LMRT, Lo-Mendell-Rubin test; OCC, Odds of correct classification.

Figure 2 Multimorbidity trajectories among individuals aged >65 years during a 10-year period, Québec, Canada, n = 99,411.

Notes: The year refers to the end of the fiscal year. For example, the count of prevalent chronic conditions (CCs) identified until March 31, 2010 was considered as the count for the year 2010.

Figure 3 Improvement in performance (c-statistic) for multimorbidity trajectories and each cross-sectional multimorbidity indicator (≥2, ≥3, ≥4 CCs, total count of chronic conditions) in comparison with the baseline model (which includes age, sex, deprivation) to predict 1-year mortality, polypharmacy, hospitalization, frequent ED admissions and frequent visits to the specialist or the GP, Québec, Canada, n = 99,411.

Abbreviations: ≥2 CCs, ≥3 CCs, ≥4 CCs: Multimorbidity indicator of the presence of at least 2, 3 or 4 chronic diseases; Count: Total count of chronic conditions; ED: Emergency department visits; GP: General practitioner.

Data Sharing Statement

The datasets generated and/or analysed during the current study are not publicly available due to data confidentiality requirements from the QICDSS. Code lists are available in supplementary data or in published papers.