Primary medical care continuity and patient mortality: a systematic review

Richard Baker; George K Freeman; Jeannie L Haggerty; M John Bankart; Keith H Nockels

doi:10.3399/bjgp20X712289

Article Figures & Data

Figures

Tables

Download figure
Open in new tab
Download powerpoint
Figure 1.
Flowchart of study selection process using PRISMA 2009. PRISMA = Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Tables

Figures

How this fits in

In 2018, a review of continuity of care was conducted with doctors in primary and secondary care; it concluded that mortality rates are lower with higher continuity. The study presented here not only confirms the association in the context of primary medical care, but also shows that it is variable and, indeed, not always present, possibly because the presumed benefits of continuity on mortality differ among different patient groups. The 13 studies reviewed say little about the mechanisms by which continuity may achieve lower mortality or why some patients may benefit more than others, and further research should focus on how, and when, continuity helps people, and how to achieve it in today’s challenging context. As there is an ongoing decline in continuity, despite evidence of its benefits on mortality and other outcomes, policy initiatives and resources must enable and incentivise services that help patients to achieve it.

View popup

Table 1.

Characteristics of included studies^a

Study	Country and setting	Population details	n	Study design	Follow-up	Follow-up sequence^b	Designed for CoC assessment? Y/N	Data source(s)	CoC measure	All-cause or disease-specific	Mortality measure mortality
*Selected study populations*
Wolinsky et al (2010)¹⁹	US, primary care	Aged >70 years	5457	Single retrospective cohort	12 years	CoC up to 12 years with mortality	N	Single interview with documentary follow-up	No more than 8 months between visits to the same primary care practitioner	All-cause	Medicare files
Worrall and Knight (2011)²⁰	Canada (Newfound land), family practice	Aged >65 years with diabetes	350	Single retrospective cohort	3 years	3-year CoC with 3-year mortality	N	Provincial administrative databases	UPC	All-cause	Mortality surveillance system
Leleu and Minvielle (2013)²¹	France, primary care	Salaried workers with ≥2 consultations, national sample	325 742	Single retrospective cohort	3 years	6-month CoC with 3-year mortality	N	National Health Insurance database records	COCI	All-cause	National Health Insurance database
McAlister et al (2013)²²	Canada (Alberta), primary care	Aged >20 years with acute admission with first-time diagnosis of heart failure	39 249	Single retrospective cohort	30 days	14-day + 1-year CoC then 30-day mortality	N	Alberta Health Administration databases	Seen by familiar physician <14 days of discharge	All-cause	Alberta Health Care Insurance Plan Registry
Bentler et al (2014)²³	US, primary care	Aged >65 years, Medicare patients	1219	Single cohort	5 years	1-year CoC with 5-year mortality	Y	Mailed questionnaire and record-based follow-up	Multiple measures	All cause	Medicare files
Nelson et al (2014)²⁴	US, primary care	Veterans with ≥2 consultations	4.3 million	Single retrospective cohort	1 year	1-year CoC then 1-year mortality	N	VHA records	UPC	All-cause	VHA files
Shin et al (2014)²⁵	South Korea, primary care	Hypertension, diabetes, or hypercholesterolaemia	47 433	Single retrospective cohort	5 years	2-year CoC then 5-year mortality	N	Korean National Health Insurance enrolees	UPC	All-cause and CVD	National death registry
Lustman et al (2016)²⁶	Israel, primary care	Aged 40–75 years, type 2 diabetes	23 679	Single retrospective cohort	2 years	1+1-year CoC with 1+1-year ^cmortality	N	HMO records database	UPC	All-cause	HMO records database
Maarsingh et al (2016)²⁷	The Netherlands, general practice	Aged ≥60 years	1712	Single retrospective cohort	17 years	7–17-year CoC then 1–14-year mortality	N	Triennial home interviews	Herfindahl–Hirschman Index	All-cause	Linked municipal registers
McAlister et al (2016)²⁸	Canada (Alberta), primary care	Aged >20 years, new diagnosis of heart failure made during an admission or ED attendance	24 373	Retrospective cohort	6 months	1-year + 1-month CoC then 6-month mortality	N	Alberta Health Administration databases	UPC	All-cause	Alberta Health Care Insurance Plan Registry
Geroldinger et al (2018)³¹	Austria, all medical disciplines and general practice	Aged ≥18 years with ≥2 diabetic medication records during index year	51 717	Single retrospective cohort	3.7 years	1-year CoC then mortality to study end	Y	Austrian social security database	Bice–Boxerman Continuity of Care Index	All-cause	Austrian social security database
*Entire primary care populations*
Levene et al (2012)²⁹	England, general practice	All patients	51.5 million	Cross-sectional	n/a	2-year CoC with 2-year mortality	N	NHS QOF and ONS data	Able to see preferred GP	All-cause, and CHD, cancer, stroke, COPD mortality	ONS
Honeyford et al (2013)³⁰	England, general practice	All patients, East Midlands	1.7 million	Cross-sectional	n/a	1-year CoC with 2-year mortality	N	NHS QOF and ONS data	Able to see preferred GP	CHD mortality	Primary Care Mortality Database

↵a Studies have been grouped according to whether selected or entire populations were included, and ordered by year of publication.
↵b CoC with mortality means overlapping measurement periods (mortality may extend longer); CoC then mortality means sequential measurement periods.
↵c The 2 years’ data collected and analysed separately and later combined when differences found to be statistically insignificant. CHD = coronary heart disease. CoC = continuity of care. COCI = Continuity of Care Index. COPD = chronic obstructive pulmonary disease. CVD = cardiovascular disease. ED = emergency department. HMO = Health Maintenance Organization. ONS = Office for National Statistics. QOF = Quality and Outcomes Framework. UPC = Usual Provider Continuity Index. VHA = Veterans Health Administration.

View popup

Table 2.

Assessment of risk of bias using the MMAT¹⁷^{,^a}

	Is the sampling strategy relevant to address the quantitative research question?	Is the sample representative of the population under study?	Are measurements appropriate (clear origin, or validity known, or standard instrument)?	Is there an acceptable response rate/follow-up (≥60%)?	Overall score
*Selected study populations*

*Wolinsky et al* (2010)¹⁹**					4
Yes		X	X	X
No
Unclear	X
Comments

Worrall and Knight (2011)²⁰					4
Yes	X		X	X
No
Unclear		X
Comments		45 patients had to be excluded

Leleu and Minvielle (2013)²¹					4
Yes	X	X	X	X
No
Unclear
Comments

*McAlister et al* (2013)²²**					2
Yes	X			X
No
Unclear		X	X
Comments		16 357 patients having >1 admission were excluded	UPC is known, familiar physician less so; no separate analysis for UPC and deaths

*Bentler et al* (2014)²³**					2
Yes	X		X
No				X
Unclear		X
Comments		Limited to fee-for-service patients	Questionnaire items not validated for this study	The survey sample was 6060, but only 1219 were included in the analysis

*Nelson et al* (2014)²⁴**					3
Yes		X	X	X
No	X
Unclear
Comments	Older males	But not of general population

*Shin et al* (2014)²⁵**					4
Yes	X	X	X	X
No
Unclear
Comments

*Lustman et al* (2016)²⁶**					4
Yes	X	X	X	X
No
Unclear
Comments

*Maarsingh et al* (2016)²⁷**					2
Yes	X
No		X	X	X
Unclear
Comments		Disadvantaged were under-represented		n = 1712/3107 (55%)

*McAlister et al* (2016)²⁸**					4
Yes	X	X	X	X
No
Unclear
Comments

*Geroldinger et al* (2018)³¹**					4
Yes	X	X	X	X
No
Unclear
Comments

*Entire primary care populations*

*Levene et al* (2012)²⁹**					3
Yes	X	X		X
No			X
Unclear
Comments			Weak continuity measure

*Honeyford et al* (2013)³⁰**					3
Yes	X	X		X
No
Unclear			X
Comments

↵a All the studies were of quantitative descriptive design and were assessed against the MMAT question items for that design. MMAT = Mixed Methods Appraisal Tool. UPC = Usual Provider Continuity Index.

View popup

Table 3.

Summary of findings

Study	Mortality measure	Summary finding (95% CI)
*Selected populations*
Wolinsky et al (2010)¹⁹	All-cause	HR 0.84 (0.77 to 0.91) for high continuity
Worrall and Knight (2011)²⁰	All-cause	HR 0.50 for high continuity
Leleu and Minvielle (2013)²¹	All-cause	HR 0.96 (0.95 to 0.96) for high continuity
McAlister et al (2013)²²	All-cause	HR 1.00 for death within 12 months, no visits with familiar physician HR 0.77 (0.70 to 0.86)^a for all visits with familiar physician
Bentler et al (2014)²³	All-cause, time to death	Patient-reported (provider duration) measure: HR 0.54 (0.37 to 0.80) for highest tertile versus lowest tertile of continuity
Nelson et al (2014)²⁴	All-cause	OR 0.94 (0.91 to 0.96) for high continuity
Shin et al (2014)²⁵	All-cause, 5-year survival rate	HR 1.12 (1.04 to 1.21) for continuity below the median
Lustman et al (2016)²⁶	All-cause	OR 0.59 (0.50 to 0.70) for high continuity
Maarsingh et al (2016)²⁷	All-cause	Lowest continuity category showed 20% more mortality than the highest category, HR 1.20 (1.01 to 1.42)
McAlister et al (2016)²⁸	All-cause	HR 0.72 (0.63 to 0.81) with ≥1 follow-up visits with familiar physician HR 1.00 for no visits HR 0.98 (0.80 to 1.20) for visits with unfamiliar physician only^a
Geroldinger et al (2018)³¹	All-cause	Primary care continuity: comparison of COCI of 1.0 with COCI of 0.74, HR 0.95 (0.87 to 1.03)
*Entire primary care population*
Levene et al (2012)²⁹	All-cause	Patient-reported measure: IRR 0.999 (0.997 to 1.01) for high continuity
Honeyford et al (2013)³⁰	CHD mortality	Patient-reported measure: IRR 0.994 (0.989 to 1.000) for high continuity

↵a Additional data provided by study authors. CHD = coronary heart disease. COCI = Continuity of Care Index. HR = hazard ratio. IRR = incidence rate ratio. OR = odds ratio.

View popup

Box 1.

Suggested mechanisms by which any type of continuity might influence mortality

Study	Suggested mechanisms
*Selected populations*
Wolinsky et al (2010)¹⁹	Continuity is defined as ‘”an ongoing relationship with a particular [primary care] physician in the outpatient setting with sufficient frequency for that physician to assume primary responsibility for both the patient’s basic health care needs and her overall disease and care management” […] Continuity is expected to result in “improved doctor–patient relationships, enhanced physician knowledge of the patient, greater rapport and disclosure, increased compliance, reduced hospitalization rates, increased patient and physician satisfaction, reductions in disability levels, costs, and missed appointments, and improved problem recognition and management”.’¹⁹
Worrall and Knight (2011)²⁰	None.
Leleu and Minvielle (2013)²¹	Consultations with the same primary care practitioner can lead to a better understanding of patients’ health needs, better management, and builds up a relationship of trust.
McAlister et al (2013)²²	None
Bentler et al (2014)²³	‘Longitudinal continuity … [provides] a chance for interpersonal continuity to develop … [which] means that knowledge, trust, and respect have developed … over time allowing for better interaction and communication. Within interpersonal continuity, there are both instrumental (provider knowledge about the patient) and affective (mode of provider behaviour toward the patient) [continuity] … that contribute to a good patient-provider relationship. […] establishing a caring, trusting bond as part of the patient-provider relationship helps both the patient and provider understand when outpatient and home care can substitute for hospitalization.’²³
Nelson et al (2014)²⁴	None. Continuity regarded as a feature of the patient-centred medical home.
Shin et al (2014)²⁵	‘A physician who attends the same patient regularly is likely to have better knowledge of him or her, to recognize problems earlier, and to provide higher quality of care. Furthermore, patients who have continuity with the same physician are more likely to adopt better self-management behaviours and to increase adherence to medication recommendations, probably because of greater trust and to have higher satisfaction with their physicians.’²⁵
Lustman et al (2016)²⁶	‘It is not possible to say if higher interpersonal continuity is causal in reducing mortality, this result is as likely due to very ill patients changing doctors or going to the most readily available doctor …’²⁶
Maarsingh et al (2016)²⁷	‘The assumed benefits of continuity of care include a better patient–provider relationship, increased patient satisfaction, improved uptake of preventive care, enhanced adherence to treatment, more accessible health care, and reduced healthcare use and costs. Especially vulnerable patients, such as older patients, are considered to benefit from continuity of care, as they are likely to have multiple chronic conditions.’ ²⁷
McAlister et al (2016)²⁸	‘It seems reasonable to hypothesize that healthcare providers (physicians or nurses/pharmacists) who have a longer-term relationship with a patient are likely to have a better sense of that patient’s unique situation and the numerous nonmedical issues that influence hospitalization risk.’ ²⁸
Geroldinger et al (2018)³¹	Patients who benefit from multidisciplinary care, which is reflected by low total continuity, may have a smaller risk of mortality. Measures of continuity are sensitive to the types of medical disciplines taken into account.

*Entire primary care population*
Levene et al (2012)²⁹	‘Starfield et al identified mechanisms potentially accounting for the beneficial impact of primary care on population health, including greater access to needed services, better quality of care, greater focus on prevention, earlier disease management, and the cumulative effect, with a holistic focus, of greater continuity and comprehensiveness.’ ²⁹
Honeyford et al (2013)³⁰	In a referenced conceptual model, the authors suggest that quality primary health care (access with sustained patient relationships and/or interventions) can modify the relationship between risk factors and probability of death.

Supplementary Data

SUPPLEMENTARY DATA

Supplementary material is not copyedited or typeset, and is published as supplied by the author(s). The author(s) retain(s) responsibility for its accuracy.