Effects of continuity of care on health outcomes among patients with diabetes mellitus and/or hypertension: a systematic review

The global prevalence of chronic non-communicable diseases (NCDs), such as diabetes mellitus (DM) and hypertension (HT), is rising rapidly. By 2030, deaths attributable to NCDs are predicted to be 52 million annually, compared to 36 million in 2008 [1]. Population aging and earlier development of NCDs have exacerbated the burden of disease on healthcare systems [2]. As opposed to communicable diseases, management for NCDs is typically long term and requires ongoing healthcare interventions, such as asymptomatic screening for early diagnosis and addressing adherence to long-term medications [3]. Based on the framework for integrated people-centred health services, the World Health Organization recommends the practice of continuity of care in primary healthcare to optimise the management on NCDs such as diabetes or hypertension [4].

Continuity of care, or the quality of care between patients and providers extending over time and beyond illness episodes, is a key pillar in a good primary care system [5]. Specifically, relational continuity refers to “a therapeutic relationship between a patient and one or more providers that spans various healthcare events and results in accumulated knowledge of the patient and care consistent with the patient’s needs” [6]. As providers gain more knowledge about the patient [7], they can tailor medical advice in subsequent consultations. A high level of continuity of care has also been associated with reduced mortality, fewer hospitalisations, lower healthcare expenses, improved medication compliance, and higher patient satisfaction [8‐22].

Evaluating the effect of continuity of care on the health indicators of DM and/or HT patients is challenging due to high methodological heterogeneity amongst studies. Two previous reviews examining the effect of continuity of care on general populations highlighted the problems of inconsistent measurement of continuity of care [22, 23]. Both reviews stated that the number of high-quality studies focusing on continuity of care and health outcomes was insufficient. The effect of continuity of care on complication development or key disease monitoring indicators of DM and/or HT patients, including haemoglobin A1c (HbA1c) and blood pressure, remains difficult to interpret as there are conflicting results. To bridge the gap in evidence, this review aims to examine the association and effect of continuity on the management and outcomes for patients with DM and/or HT. Secondary objectives are to investigate the mechanisms behind the relationship, to assess the quality and strength of the existing evidence, and suggest possible considerations for the interpretation and application of these findings.

The initial search yielded 21,090 studies from the databases and 134 studies from the trial registry. After removing duplicates, screening of the remaining 19,130 studies identified 39 eligible studies. Citation searches of these 39 studies identified 66 potentially relevant articles. Of these, three studies met the eligibility criteria and were included in the review (Fig. 1, Table 1). Among the 42 studies included, seven focused on hypertension [7, 25‐30], 32 on diabetes [13, 18, 30‐60] and three examined both [10, 61, 62]. Most of them were retrospective cohort studies (31 studies), with a few being prospective cohort (five studies), cross-sectional (four studies), or case–control studies (two studies). These studies were conducted in United States (ten studies), Taiwan (nine studies), South Korea (nine studies), Canada (three studies), Australia (three studies), Malaysia (two studies), United Kingdom (one study), Portugal (one study), the Netherlands (one study), Finland (one study), Chile (one study) and Israel (one study). Most of the studies had a population age mean or median of between 50–70 years. The follow-up periods spanned from less than a year [32, 35, 49] to 17 years [53]; the most common length was 2 to 3 years (16 studies).

Meta-analysis was not performed due to the clear heterogeneity among studies, including differences in assessment methods and cut-off points of continuity of care, inclusion and exclusion criteria of participants and outcomes, as well as adjustment of confounders. There was also a lack of consistency in the reporting of effect size. Missing information such as the number of patients and cases in each group or the exact cut-off points of exposure in some studies made unification of effect size impossible.

This review identified studies from a wide range of settings representing a broad range of healthcare systems. The potential influence of different contexts and a diverse range of outcomes were assessed and categorised accordingly to allow for robust comparisons to be made between studies. These categories reflect a logical progression of stages in the management and control of NCDs allowing for a comprehensive synthesis of the current evidence on the effect of continuity of care for patients with diabetes and/or hypertension.

Beneficial effects of continuity of care were reported more consistently for service utilisation, mortality and disease-related complications than for health indicators such as HbA1c, blood pressure. To better understand the potential mechanisms behind the beneficial effect on service utilisation, the principle of how continuity of care comes into play in healthcare provision must be considered. Continuity of care has been described as a three-layered concept that encompasses informational continuity, longitudinal continuity and interpersonal continuity [20]. Interpersonal continuity is the most frequently examined aspect of continuity of care. It refers to the sustained and ongoing caring relationship, which catalyses the delivery of patient-centred healthcare, between the physician and the patient [35, 64]. It also reflects mutual trust and responsibility in this relationship. Healthcare providers with high levels of continuity of care are able to communicate better with their patients and thus have a better knowledge of their patients’ disease history and current situation that might not be included in the patient’s medical records. A study found that higher continuity of care was associated with better quality of care among DM patients, including more HbA1c testing and eye or foot examinations [65]. Therefore, deterioration in patients’ conditions could be less likely to go undetected or untreated. Moreover, continuity of care improves patient satisfaction [35], facilitates higher patient self-care behaviours, compliance and adherence to physicians’ recommendations and regime, which could be the reason for reduced preventable hospital admissions [18, 33, 36].

The reasons behind a lack of significant results among studies in other outcome categories could be due to methodological limitations such as insufficient sample sizes, and use of short-term dynamic outcome measurements. Firstly, studies without significant findings often contained smaller sample sizes [25, 30‐33, 36, 46, 51, 62]. Having a small or inadequately powered sample size can affect the validity of results and may mask potential associations between continuity of care and study outcomes. Secondly, biochemical measurements might not be the most reflective or accurate indicators for the effectiveness of continuity of care due to their dynamic nature [34, 36]. Mainous III et al., found a positive association between better continuity of care and improved HbA1c, but not with blood pressure or low-lipoprotein cholesterol among DM patients suggesting the result could be because providers placed a higher priority on achieving glycaemic control for DM patients over other outcomes [34]. As a result, significant associations between glycaemic control and relational continuity of care might be more observable. A similar conclusion was reported by Dearinger et al. [36]. The authors noted that blood pressure is a “dynamic measurement that can be influenced by many factors at any given time”, such as variation among operators which may have contributed to measurement bias [36]. Moreover, blood pressure measurement in the study was performed as part of the patient’s normal consultation, rather than using more meticulous methods, such as ambulatory or home-based measurements, which may impact the level of accuracy in the measurement. This may partially explain why blood pressure was not seen as a significant outcome in association with continuity of care. Both studies highlight the importance of best practice measurements when it comes to evaluating patient health. Careful consideration is needed to discern whether such dynamic outcomes when measured in a study accurately translates to the patient’s health and wellbeing.

Another issue identified in the reviewed studies was a lack of consistent cut-off points among studies that categorised levels of continuity of care into discrete groups. UPCI or CoCI at ≥ 0.75 was a relatively common definition for high continuity of care [26, 37, 38, 42, 44, 47, 54, 56]. However, a number of studies applied other definitions, for example, Robles et al.: CoCI ≥ 0.236 [7]; Dearinger et al.: UPCI ≥ 0.45 [36]; Hong et al.: CoCI ≥ 0.40 [10]; and Chen et al.: CoCI ≥ 0.44 [18]. As there is currently no consensus to what a high degree of continuity of care is, studies often defined their cut-offs based on the distribution of their study population, such as using the median and tertiles. While this might not have a great impact on the overall conclusion for individual studies, it reduces the comparability between studies and may have greater implications for policymaking to improve continuity in real-life settings, as there needs to be evidence-based aims for various health systems to strive towards.

In terms of the quality evaluation of included studies, consideration of confounding factors was one of the biggest challenges. Although most studies had controlled for confounding factors, it was unclear these adjustments were sufficiently comprehensive. Several factors were found to have a significant influence on continuity, such as patient’s perception on continuity, remoteness and characteristics of the patient, the physician or the healthcare organisation [66, 67], which could be potential confounders. Depending on the health system, socio-economic status, which is known to have a significant correlation with health [68], may affect an individual’s ability to continue seeing the same physician which would hence impact their continuity of care. Structure of the healthcare organisation can also have an impact, as continuity of care might be more difficult to sustain in larger group practices [69]. Therefore, it is important to adequately appraise the appropriate confounders to ensure a valid result.

The study context should also be considered in order to understand the nuances of the relationship between continuity of care and patient health outcomes. The studies included in this review took place across several countries, with different healthcare systems. The impact of culture on the physician–patient relationships and/or the actual patient outcomes may be an important confounding factor. For example, Taiwan’s healthcare system offers free consultations with lenient referral practices. This might increase the likelihood of hospital admissions compared to if the study was performed in a country with stricter referral policies [37]. And in many European countries, continuity of care is a well-established concept, emphasized by both healthcare providers and patients [70‐72]. These countries generally have strong primary care systems and have implemented various policies to promote continuity of care, such as mandating or providing financial incentives for patients to register with a primary care doctor [73, 74]. This might result in insignificant findings due to low levels of variability in continuity of care among patient populations [43]. While contextual factors do not necessary detract from the findings in these studies, they still need to be taken into consideration when applying the findings to policymaking in other settings.

There were several strengths to this review. Firstly, we studied a wide range of DM/HT-related outcomes. This enabled us to provide a more comprehensive narrative about the effect of relational continuity of care in DM and/or HT patients. Second, we summarised the common characteristics and limitations of existing literature to inform policy recommendations and future research. There were also several limitations. Firstly, no meta-analysis was performed. While we have provided an integrated summary on the associations of continuity of care with various DM/HT related outcomes, without meta-analysis, we are unable to examine the strength of such associations. However, given the heterogeneity, incomplete information and pre-existing biases among the observational studies, results from a meta-analysis would be questionable if not spurious [75]. Second, we did not review the non-English literature which introduces English-language bias. As the implementation and effects of continuity of care may differ across healthcare settings (such as the influence of local health-seeking behaviours), excluding non-English studies can cause selection bias. Third, while we have used a systematic approach to minimize biases, our review protocol lacks prior registration.

One of the challenges encountered in this review was overcoming discrepancies in the definitions of high and low levels continuity. While our review took such classifications as stated in the selected studies, further research should explore how differences in these classifications may affect the resulting conclusions and their statistical significance. Also, as research in this field is still relatively incomplete, there is no consensus on the best way to measure continuity. While this study examined various measures such as CoCI and UPCI, it may be worthwhile to further investigate into the strengths and weaknesses of each measure and how they might influence outcomes.

This review found strong associations between high continuity of care and reduced healthcare utilisation, mortality rate and complication risk in patients with diabetes and/ or hypertension, but hardly any impact on various health indicators. Heterogeneity in patient selection, confounding adjustment and assessments of continuity of care and outcome variables remain major constraints in current literatures. A more standardised measurement of continuity of care is needed in order to provide more reliable and comparable evidence for its implementation as a healthcare policy.