Skip to main content

Table 7 Modelling methods for characterising recurrent falls

From: Economic models of community-based falls prevention: a systematic review with subsequent commissioning and methodological recommendations

Study label (n = 46)a

Transition entityb

Cycle length

Main fall-related event

Possible to model recurrent falls

Binary decision model

 Beard (2006) [53]

Fall event

N/Ac

MA fall

Yes

 Carande-Kulis (2015) [55]

Individual

N/A

MA fall

No

 Comans (2009) [59]

Individual

N/A

Any fall

Yes

 Frick (2010) [66]

Individual

N/A

Hip fracture

No

 Hektoen (2009) [67]

Fall event

N/A

Any fall

Yes

 Howland (2015) [71]

Individual

N/A

MA fall

Yes: targeted recurrent fall

 Ippoliti (2018) [72]

Fall event

N/A

Hip fracture

Yes

 Ling (2008) [75]

Individual

N/A

Any fall

No

 Miller (2011) [77]

Individual

N/A

MA fall

No

 Poole (2014) [83]

Fall event

N/A

Hip fracture

Yes

 Sach (2007); (2010) [86, 87]

Fall event

N/A

Any fall

Yes

 Velde (2008) [91]

Fall event

N/A

Any fall

Yes

 Wu (2010) [93]

Individual

N/A

Any fall

Yes: targeted recurrent fall

Static modeld

 Agartioglu (2020) [50]

Individual

N/A

Any fall

No

 Albert (2016) [51]

Individual

N/A

Any fall

Yes

 CSP (2016) [56]

Individual

N/A

MA fall

Yes

 Day (2009); (2010) [60, 61]

Fall event

N/A

Any fall

Yes

 Hirst (2016) [69]

Individual

N/A

Fractures

No

 McLean (2015) [76]

Individual

N/A

Any fall

Yes: Adjusted risk

 PHE (2018) [85]

Fall event

N/A

Any fall

Yes

 Smith et al. (2016) [88]

Individual

N/A

MA fall

No

Cohort-level Markov model

 Alhambra-Borras (2019) [52]

Individual

1 year

Compositee

Yes: Compositee

 BODE3 models

Individual

1 year

MA fall

No

 Church (2011); (2012) [57, 58]

Individual

1 year

Any fall

No

 Eldridge (2005) [63]

Individual

1 year

Any fall

No

 Farag (2015) [64]

Individual

1 year

Any fall

No

 Franklin (2019) [65]

Individual

1 year

Any fall

Yes

 Honkanen (2006) [70]

Individual

1 year

Hip fracture

No

 Johansson (2008) [73]

Individual

1 year

Hip fracture

No

 Lee (2013) [74]

Individual

1 month

Any fall

Yes

 Moriarty (2019) [79]

Individual

1 year

MA fall/Hip fracture

No

 OMAS (2008) [81]

Individual

1 year

MA fall

No

 Poole (2015) [84]

Individual

1 year

MA fall

No

 RCN (2005) [34]

Individual

1 year

MA fall

No

 Tannenbaum (2015) [89]

Individual

6 months

Any fall

Yes

 Turner (2020) [90]

Individual

1 month

MA fall/Hip fracture

Yes

Individual-level Markov model (microsimulation)

 Hiligsmann (2014) [68]

Individual

6 months

Fractures

Yes

 Mori (2017) [78]

Individual

1 year

Fractures

No

 Nshimyumukiza (2013) [80]

Individual

1 year

Fractures

No

 Zarca (2014) [94]

Individual

3 months

Hip fracture

Yes

  1. Abbreviations: BODE3 Burden of Disease Epidemiology, Equity and Cost-Effectiveness Programme studies, including Boyd (2020) [54], Deverall (2018) [62], Pega (2016) [82] and Wilson (2017) [92], CSP Chartered Society of Physiotherapy, Int. Intervention, MA fall Fall requiring medical attention, N/A Not applicable, OMAS Ontario Medical Advisory Secretariat, PHE Public Health England, RCN Royal College of Nursing
  2. aSee Table 2 for study references; parenthesised number refers to the number of models included in the table
  3. bAll Markov models conceive individuals (or proportion of individuals within cohort) transitioning between model states. Some binary decision and static models have fall events transitioning through health and economic sequelae
  4. cCycle length was not relevant or applicable to non-cycle-based models such as the decision tree
  5. dAll studies under this category, except Smith (2016) [88], used a decision tree model
  6. eThis model used a composite measure of health consequences including recurrent falls, fear of falling and mobility and balance problems. Hence, recurrent falls were captured within the composite measure