Table 1 A comparison of study characteristics and definitions of Multimorbidity used in the 14 articles

Bahrmann et al. (2019) [20]

307 non-surgical patients ≥ 68 years presenting to the Emergency department. Germany.

Mixed

1-year CVD mortality (30)

CCI original version

1.16 (1.02–1.31) in fully adjusted model

Model fit (AIC).

Lowest AIC 317.7 for model 3 (assessing the addition of CCI to Barthel index, age, and sex).

Erickson et al. (2014) [21]

1,202 Patients ≥ 18 years hospitalised with unstable angina or acute myocardial infarction. 83.4% White. United States (US).

Secondary

Occurrence of secondary events within 6 months (317)

CCI D’Hoore version

adjOR 1.09 (1.02–1.16) in combined model

Calibration: Hosmer–Lemeshow test,

Discrimination: c-statistic,

Likelihood ratio test.

Likelihood ratio test indicated that adding the CCI to a model containing the GRPI improved model predictive ability (chi-square value 16.031).

Fauchier et al. (2019)* [22]

8,962 consecutive patients with atrial fibrillation. France.

Unspecified

CVD mortality. Mean follow-up of 929 days, SD 1082 days (1,294 patients died, and 54% were documented cardiovascular deaths)

CCI version unspecified in abstract

NA

Discrimination: AUC.

No

Fauchier et al. (2020)* [23]

2,641,626 patients with atrial fibrillation. France.

Unspecified

Cardiovascular death. Mean follow-up 2.0 years, SD 2.3 years (670,541 died, 30.3% cardiovascular deaths)

CCI version unspecified in abstract

NA

Discrimination: AUC.

No

Fauchier et al. (2022)* [24]

371,848 consecutive patients hospitalised with heart failure. France.

Secondary

CVD death. Mean follow-up 4.0 years, SD 2.8 years (220,774 patients died and 31.3% CVD deaths)

CCI unspecified version in abstract

NA

Discrimination: AUC.

No

Gelow et al. (2015) [25]

246 heart failure patients. Mean age 56.6 years, SD 13.1 years. 88.4% White. US.

Secondary

Fatal and non-fatal CVD events within 6 months (77)

CCI unspecified version

NA

Discrimination:

Harrell’s c-statistic.

No

Haji et al. (2022) [26]

334 coronary artery disease patients ≥ 45 years screened during cardiac admission. Australia.

Secondary

3-year heart failure-related hospitalisations (50)

CCI unspecified version

(Competing risk model)

SHR 1.64 (1.25–2.15) in fully adjusted model (“Model 5”)

Discrimination: AUC.

No

Jeong et al. (2022) [27]

8,666 newly diagnosed patients with colorectal cancer who survived at least 5 years (training cohort = 4,709, validation cohort = 3,957). Median age 60 years, IQR 53–67 years. South Korea.

Primary

1-to-5-year CVD hospitalisations and ischaemic heart disease hospitalisations (295 CVD events and 127 ischaemic heart disease events in validation cohort)

CCI unspecified version

1.035 (1.003–1.067) in total CVD model and 1.054 (1.007–1.102) IHD model for the training cohort

Discrimination:

AUC,

Calibration: plots assessed.

No

Kayama et al. (2018)* [28]

376 acute decompensated heart failure patients. Japan.

Secondary

Composite of cardiac death and heart failure hospitalisation. Mean follow-up period 2.0 years, SD 1.4 years (137)

Age-adjusted Charlson comorbidity index (ACCI).

ACCI 0–4: Ref

ACCI ≥ 7: 2.69 (1.65–4.36)

Discrimination: AUC.

No

Lip et al. (2022) [29]

3,435,224 people (67% in the training cohort and 33% in the validation cohort), from two health plans. Mean age 49.7 years, SD 15.3 years. US.

Mixed

The first incidence of stroke. Follow-up period not specified. (53,010 in total cohort)

Multimorbid index

NA

Discrimination: c-statistic,

Calibration: plots assessed,

Decision curve analysis for clinical utility.

No

Livingstone et al. (2022) [30]

2,904,773 primary care patients (derivation cohort and validation cohorts: n = 989,732 and 494,865 women, n = 946,784 and 473,392 men) who were free of CVD (age 25–84 years). 91.8% White. UK.

Primary

10-year fatal and non-fatal first CVD event (14,150 for women and 17,689 for men in derivation cohort)

modified CCI– using an adapted version for Read codes

(Competing risk model)

SHR in derivation cohort.

CCI 0: Ref

CCI ≥ 3: Women 1.18 (0.94–1.49), Men 1.17 (0.93–1.46)

Model fit (AIC),

Calibration: plots assessed,

Discrimination

Harrell’s c-statistic, patient reclassification and

estimated NNT.

Assessed incremental value within sex-stratified competing risks models with QRISK3 predictors, improving calibration (including by age and CCI subgroups) and patient reclassification.

van Bussel et al. (2019) [31]

1,811 primary care patients aged 70-78 years. 97.8% Caucasian. Netherlands.

Primary

5-year risk of combined CVD morbidity and mortality (277)

Polypharmacy (Binary)

(Competing risk model)

1.41 (1.08-1.83) in the Cox-PH model and SHR 1.40 (1.08–1.82) in the Fine-Gray models

Calibration plots assessed,

Discrimination IPCW c-statistic.

No

Walkey et al. (2022) [32]

39,590 KPNC and 16,388 IH sepsis survivors (two cohorts), median ages 70.0 years (IQR 60–81) years and 66 years (IQR 56–76), respectively. 62.7% White. US.

Mixed (although it excluded those with a CVD history in five years before baseline)

1-year atherosclerotic CVD outcomes (996 at KP and 192 at IH).

CCI unspecified version, as a presepsis risk factor.

(Competing risk model)

1.05 (0.96–1.14) for CCI in the KPNC model for combined CVD outcomes

Calibration plots assessed,

Discrimination c-statistic.

No

Zhan et al. (2020) [33]

428,251 COPD patients (training cohort n = 214,126 and validation cohort n = 214,125) mean age 67 years, SD 13 years.

Taiwan.

Mixed

1-year CVD hospitalisation (16,071 in the training dataset, 15,723 in the validation dataset) and cerebrovascular hospitalisation (7,337 in the training cohort, 7,306 in the validation cohort)

Charlson Index (Deyo), the updated CDS, and the PBDI.

NA

Calibration plots assessed,

Discrimination c-statistic,

Patient reclassification and NRI.

Compared to a reference group with age and sex only with different models for each index showing improved discrimination and reclassification ability of the comorbidity models.