Untapped potential of multicenter studies: a review of cardiovascular risk prediction models revealed inappropriate analyses and wide variation in reporting

Wynants, L.; Kent, D. M.; Timmerman, D.; Lundquist, C. M.; Van Calster, B.

doi:10.1186/s41512-019-0046-9

Table 1 Key characteristics of 50 selected publications

From: Untapped potential of multicenter studies: a review of cardiovascular risk prediction models revealed inappropriate analyses and wide variation in reporting

1st author	Journal	Year	Data type	N development	N events development	No. of centers	Regression technique	Clinical setting	International	Externally validated	Generalizability statement
Agarwal	J Heart Lung Transplant	2012	Registry/cohort	339	111	3	Cox ignoring clustering	All tertiary/academic	No	Yes	Generalizability uncertain: high-risk population (but validated in low-risk population)
Alsonso	J Am Heart Assoc	2013	IPD-MA	18,556	1186	3	Stratified Cox	Community/population cohort	No	Yes	Generalizability uncertain: selection/referral/spectrum bias
Amin	J Am Heart Assoc	2013	Trial	19,121	988	674	Cox ignoring clustering		Yes	Yes	Generalizability uncertain: selection/referral/spectrum bias
Bekelis	J Neurointerv Surg	2014	Registry/cohort	4551	6	4573	Logistic ignoring clustering	Mix of community hospitals, diverse w.r.t. academic status	No	No	Validation needed
Carluccio	Eur J Heart Fail	2013	Registry/cohort	747	224		Cox ignoring clustering			No	Validation needed
Chung-Esaki	Am J Emerg Med	2014	Registry/cohort	249	18	2	Logistic ignoring clustering	All community hospitals		No	Validation needed
Cuschieri	J Dig Dis	2014	Registry/cohort	3218	107	10	Cox ignoring clustering	All veteran affairs hospitals	No	No	Generalizability uncertain: selection/referral/spectrum bias
de la Cámara	Cardiol J	2012	Registry/cohort	600	98	3	Logistic ignoring clustering	All tertiary/academic	No	No
de Man-van Ginkel	Stroke	2013	Registry/cohort	382	54	3	Logistic ignoring clustering		No	No	Validation needed
den Exter	Chest	2013	Registry/cohort	1048	225	288	Logistic ignoring clustering	Mix of teaching and non-teaching hospitals	Yes	Yes	Claim generalizability of model
Dodson	J Am Coll Cardiol	2014	Registry/cohort	168,442	3032	1428	Mixed effects logistic		No	No	Generalizability uncertain: selection/referral/spectrum bias
Eichinger	J Am Heart Assoc	2014	Registry/cohort	553	150	4	Fine and Gray ignoring clustering		No	No	Validation needed
Enajat	Plast Reconstr Surg	2013	Registry/cohort	430	17	2	Logistic ignoring clustering	All tertiary/academic	No	No	Validation needed
Farooq	Eur Heart J	2012	IPD-MA	6309	175	7	Stratified logistic		Yes	No	Validation needed
Felker	J Card Fail	2004	Trial	949	90	78	Cox ignoring clustering	Mix of academic and community hospitals	No	No	Validation needed
Goff	J Am Coll Cardiol	2014	IPD-MA	24,626	2689	4	Cox ignoring clustering	Community/population cohort	No	No	Validation needed
Grant	Br J Surg	2013	Registry/cohort	11,423	312	140	Logistic ignoring clustering		No	No	Validation needed
Gupta	J Vasc Surg	2012	Registry/cohort	9556	170		Logistic ignoring clustering	Mix of academic and community hospitals	No	No	Generalizability uncertain: selection/referral/spectrum bias
Hannan	JACC Cardiovasc Interv	2013	Registry/cohort	54,223	558	58	Logistic ignoring clustering	Community/population cohort	No	Yes	Validation needed
Hubert	J Am Coll Cardiol	2013	Registry/cohort	565	48	2	Fine and Gray ignoring clustering	All tertiary/academic	No	No	Generalizability uncertain: selection/referral/spectrum bias
Iannuzzi	J Vasc Surg	2013	Registry/cohort	562,791	2862		Logistic ignoring clustering		No	No	Generalizability uncertain: selection/referral/spectrum bias
Iida	Eur J Vasc Endovasc Surg	2012	Registry/cohort	1058	497	11	Cox ignoring clustering		No	No
Iung	Heart	2014	Registry/cohort	2552	253	34	Mixed effects logistic		No	No	Generalizability uncertain: high-risk population
Kay	J Am Coll Cardiol	2013	Registry/cohort	6037	169	3	Cox and fixed center effect			No	Generalizability uncertain: selection/referral/spectrum bias
Kramer	Heart Rhythm	2012	Registry/cohort	905	125	3	Logistic ignoring clustering	All tertiary/academic	No	No	Generalizability uncertain: selection/referral/spectrum bias
Ky	Circ Heart Fail	2012	Registry/cohort	1513	317	3	Cox ignoring clustering	All tertiary/academic	No	No	Validation needed
Liu	CNS Neurosci	2014		575	76	30	Logistic ignoring clustering	All tertiary/academic	No	No	Generalizability uncertain: selection/referral/spectrum bias
Myint	Int J Stroke	2014	Registry/cohort	12,355	2425	3	Logistic ignoring clustering	All tertiary/academic	No	No	Validation needed
Palmerini	JACC Cardiovasc Interv	2012	Trial	1692	173		Cox ignoring clustering		No	No	Validation needed
Pannucci	Chest	2014	Registry/cohort	6768	95	10	Logistic ignoring clustering		No	No	Validation needed
Park	Am J Cardiol	2012	Registry/cohort	63,118	217	126	Cox ignoring clustering		Yes	No	Claim generalizability of model
Perry	Stroke	2014	Registry/cohort	3906	86	8	Logistic ignoring clustering	All tertiary/academic	No	No	Validation needed
Piccini	Circulation	2013	Trial	14,155	575	1100	Cox and fixed region effect		Yes	Yes	Generalizability uncertain: selection/referral/spectrum bias
Pros	Eur J Vasc Endovasc Surg	2013	Registry/cohort	640	123	3	Reduce to single-center development set, then logistic	All tertiary/academic	No	Yes	Generalizability uncertain: high-risk population
Rao	JACC Cardiovasc Interv	2013	Registry/cohort	834,696	48,412	1142	Logistic ignoring clustering	Mix of teaching and non-teaching hospitals	No	No
Rizzi	BMC Infect Dis	2014	Registry/cohort	1456	499	25	Logistic ignoring clustering	Mix of secondary and tertiary	No	No	Generalizability uncertain: high-risk population
Schellinger	Stroke	2013	Trial	192		68	Logistic ignoring clustering		Yes	No
Singer	J Am Heart Assoc	2013	Registry/cohort	32,590	952	2	Reduce to single-center development set, then Cox	All community hospitals	No	Yes	Validation needed
Stiell	Acad Emerg Med	2013	Registry/cohort	559	65	6	Logistic ignoring clustering	All tertiary/academic	No	No	Validation needed
Takagi	J Am Coll Cardiol	2013	Registry/cohort	1286	76	47	Cox ignoring clustering		No	No	Validation needed
Than	Emerg Med Australas	2014	Registry/cohort	1974	305	2	Logistic ignoring clustering		Yes	Yes	Validation needed
Tolenaar	Circulation	2014	Registry/cohort	1034	110	12	Logistic ignoring clustering	All tertiary/academic	Yes	No	Validation needed
Trujillo-Santons	Am J Med	2015	Registry/cohort	15,280	173	125	Logistic ignoring clustering		Yes	No	Validation needed
Van De Bruaene	Int J Cardiol	2013	Registry/cohort	155	39	10	Cox ignoring clustering	All tertiary/academic	No	No	Validation needed
Van Hattum	Thromb Haemost	2012	Trial	482	287	6	Cox ignoring clustering		No	No	Validation needed
Wang	Cardiology	2013	Registry/cohort	1615	343	3	Logistic ignoring clustering		No	Yes	Validation needed
Wilson	Am J Med	2012	Registry/cohort	33,419	2394	5473	Cox and fixed region effect	Mix of public and private hospitals	Yes	No	Validation needed
Wimmer	Stroke	2012	Registry/cohort	10,186	177	364	Logistic and fixed effect for physician experience	Mix of academic and private hospitals	Yes	No	Generalizability uncertain: use of specific devices
Wimmer	J Am Heart Assoc	2013	Registry/cohort	14,387	366	5	Mixed effects logistic		No	No	Generalizability uncertain: high-risk population
Zheng	Front Med	2013	Registry/cohort	8602	215	43	Logistic ignoring clustering		No	No	Validation needed

Blank cells indicate unreported information

Back to article page

ISSN: 2397-7523

Contact us

Submission enquiries: Access here and click Contact Us
General enquiries: info@biomedcentral.com

Diagnostic and Prognostic Research

Contact us