Summary of the relationship between multimorbidity, costs, and utilisationa
| Utilisation/cost type | MM specification | Magnitude (95% CI) | Parameter estimate type | Reference | |
|---|---|---|---|---|---|
| Primary care visits (n = 3) | |||||
| Number of QOF LTCs |
 | 0.37 (0.36 to 0.38) | Marginal effect | 28 | |
| MM vs not |
 | 2.56 (2.48 to 2.64) | Odds ratio | 3 | |
| Dental visits (n = 1) | |||||
| MM vs not |
 | 1.23 (1.08 to 1.38) | Odds ratio | 36 | |
| A&E visits (n = 2) | |||||
| HADS score of 8 or more vs lower |
 | 1.58 (1.04 to 2.41) | Odds ratio | 35 | |
| 1 QOF LTC vs none |
 | 1.12 (1.10 to 1.13) | Odds ratio | 24 | |
| 2 QOF LTC vs none |
 | 1.28 (1.25 to 1.31) | Odds ratio | 24 | |
| 3 QOF LTC vs none |
 | 1.65 (1.59 to 1.71) | Odds ratio | 24 | |
| ≥4 QOF LTC vs none |
 | 2.55 (2.44 to 2.66) | Odds ratio | 24 | |
| Hospitalisations (n = 4) | |||||
| All | 1 LTC vs none |
 | 1.77 (1.59 to 1.98) | Odds ratio | 33 |
| 2 LTC vs none |
 | 2.41 (2.12 to 2.72) | Odds ratio | 33 | |
| 3 LTC vs none |
 | 3.53 (3.06 to 4.07) | Odds ratio | 33 | |
| ≥4 QOF LTC vs none |
 | 4.33 (3.63 to 5.17) | Odds ratio | 33 | |
| MM vs not |
 | 2.58 (2.48 to 2.69) | Yearly rate ratio | 3 | |
| Unplanned all | 1 PC vs none |
 | 1.70 (1.59 to 1.82) | Odds ratio | 27 |
| 2 PC vs none |
 | 2.69 (2.50 to 2.89) | Odds ratio | 27 | |
| 3 PC vs none |
 | 3.47 (3.21 to 3.76) | Odds ratio | 27 | |
| ≥4 PC vs none |
 | 5.87 (5.45 to 6.32) | Odds ratio | 27 | |
| Unplanned potentially preventable | 1 PC vs none |
 | 2.50 (2.07 to 3.03) | Odds ratio | 27 |
| 2 PC vs none |
 | 4.93 (4.06 to 5.99) | Odds ratio | 27 | |
| 3 PC vs none |
 | 6.82 (5.55 to 8.37) | Odds ratio | 27 | |
| ≥4 PC vs none |
 | 14.38 (11.87 to 17.43) | Odds ratio | 27 | |
| Prolonged length of stay | MM vs not (90+ population) |
 | 0.61 (0.32 to 1.13) | Risk ratio | 23 |
| Total costs (n = 3) | |||||
| 1–3 LTC vs none |
 | 1.62 (1.28 to 2.03) | Mean ratio | 31 | |
| 4–6 LTC vs none |
 | 2.53 (2.01 to 3.19) | Mean ratio | 31 | |
| 7–9 LTC vs none |
 | 3.82 (3.01 to 4.85) | Mean ratio | 31 | |
| 1 LTC vs none |
 | 1.99 (1.95 to 2.03) | Mean ratio | 26 | |
| 2 LTC vs none |
 | 2.53 (2.46 to 2.58) | Mean ratio | 26 | |
| 3 LTC vs none |
 | 2.86 (2.72 to 3.03) | Mean ratio | 26 | |
| Care transition costs (n = 1) | Comorbidity pairs vs index LTC |
 | P<0.001 | Increasing trend in association | 34 |
| Primary care costs (n = 2) | Costs of 1 patient with 2 |
 | Increasing or decreasing costs when co-occurring | Estimated prevalence- adjusted cost | 13 |
| LTC vs 2 separate | |||||
| patients with each LTC | |||||
| Hospital costs (n = 1) | Individual LTC |
 | P<0.01 for 90% of the estimated coefficients | Estimated coefficient | 25 |
| Time to death as a proxy for morbidity | |||||
↵a The number of articles is indicated in parentheses next to the cost or utilisation type (see Supplementary Appendix S3 for the complete 17 study references). Mean ratios can be obtained by exponentiating the parameter estimates from a generalised linear model with the log-link; they have an interpretation similar to an odds ratio. For example, individuals with 7–9 conditions have 3.82 times the mean expected total costs of individuals without comorbidities. A&E = accident and emergency. CI = confidence interval. HADS = Hospital Anxiety and Depression Scale. LTC = long-term condition. MM = multimorbidity. PC = physical condition. QOF = Quality and Outcomes Framework. Prolonged length of stay is defined as 7 days in the hospital. Care transitions are defined as healthcare changes from general practice to emergency department or hospital care.