Continuity of care and intermediate outcomes of type 2 diabetes mellitus

Abstract

Background. The relevance of continuity of care in chronic illness is uncertain.

Objective. We evaluated whether experienced continuity of care for type 2 diabetes is associated with HbA1c, blood pressure or body weight.

Methods. Cohort study in 19 family practices in London, UK. Participants were 209 type 2 diabetic patients with 156 (75%) followed-up at 10 months. Main measures were experienced continuity of care (ECC) by patient questionnaire (mean score 62.1, SD 16.0), satisfaction with care, health-related quality of life [short-form 12 (SF-12)], HbA1c, blood pressure and body weight. Analyses were adjusted for baseline values, age, sex, ethnicity, duration of diabetes, diabetes treatment, education, housing tenure and whether living alone.

Results. Experienced continuity scores were obtained for 193 (92%) of participants at baseline and 156 (75%) at follow-up with no difference in outcome measures between those followed-up and those not. Subjects with the highest satisfaction ratings, compared with the lowest, had higher experienced continuity (difference in experienced continuity 7.87, 95% confidence interval 3.22–12.5, P = 0.001). ECC was not associated with HbA1c (adjusted difference per 10-unit increase in ECC score, −0.09%, −0.29 to 0.12%, P = 0.402), systolic blood pressure (−0.41, −2.88 to 2.06 mm Hg, P = 0.746), body mass index (−0.08, −0.34 to 0.18 kg/m², P = 0.562) or SF-12 physical component score (0.73, −0.88 to 2.35, P = 0.375).

Conclusion. Experienced continuity of diabetes care is associated with greater patient satisfaction but not with improved intermediate outcomes during 10 months follow-up in this setting. Studies with more subjects will be required to determine whether continuity of care is associated with the frequency of adverse events.

Continuity of care, diabetes mellitus, patient experience, patient views, quality of care

Introduction

Improving the quality of care for patients with chronic conditions like diabetes mellitus is an important focus of health policy.1,2 ‘Continuity of care’ is concerned with the quality of health care in a longitudinal or temporal domain.3 Continuity, or lack of continuity, may be an important consequence of the organization and delivery care of patients with chronic illness but it is uncertain whether continuity of care is associated with patients’ health outcomes.4 This question is important because the answer may influence the priority given to sustaining interpersonal continuity as compared to ensuring rapid access, increasing choice or facilitating timely specialist intervention during the course of patients’ illnesses.5

Continuity of care can be viewed either from the perspective of the patient or from the perspective of the professional responsible for delivering care.6,7 There are two main elements of continuity of care.3,6,8 ‘Relational continuity’ describes the ‘continuous caring relationship’1 which is established between the patient and a named usual doctor or nurse, facilitating the delivery of personally tailored care. ‘Management continuity’ concerns the provision of a ‘seamless service’, with optimal coordination and communication between the different professionals and health care organizations that contribute to patients’ care.3 Relational continuity has traditionally been viewed as a key element of primary care8 but, with only 7 % of general practitioners in England now in single-handed practice, the notion of the ‘seamless service’ is increasingly important even in a family practice setting.

We recently developed and tested the reliability and validity of a new questionnaire measure of experienced continuity of care for type 2 diabetes (ECC-DM).9 The ECC-DM is grounded in patient interview data with four domains.10,11 Experienced longitudinal continuity refers to long-term interaction with the health service through regular review consultations; relational continuity is the experience of receiving care from a usual doctor or nurse who is well-known to the patient; flexible continuity requires that the patient can easily obtain consultations from a usual professional when this is required and team and cross-boundary continuity concern patients’ experiences of well-coordinated care from different members staff in different organizations.10 The development and testing of the 19-item measure have been described elsewhere. The measure includes four items for each of longitudinal and flexible continuity, six items for relational continuity and five for team and cross-boundary continuity. Responses to each item are coded using six-point Likert scales. In addition to the overall score, the ECC-DM provides subscale scores for the four domains of experienced continuity.9,11 The total scale score ranges from 0 to 100 with mean 62.1 (SD 16.0). Cronbach's alpha for the overall scale score was estimated to be 0.908 in the baseline data for the present study and a factor analysis generally confirmed the proposed factorial structure. Several observations supported the validity of the proposed measure: scores for patients treated at one practice were more similar to each other than those for patients treated at other practices, with mean scores ranging from 46 to 78 at different practices; experienced continuity was found to be higher if there was a named lead doctor for diabetes and experienced continuity was lower for patients who received diabetes care from hospital clinics than for patients who received diabetes care from the general practice.

In the present study, we utilized the new measure of experienced continuity in a cohort study of type 2 diabetic patients. We aimed to determine whether experienced continuity was associated with control of blood glucose (HbA1c), blood pressure and body weight or with health-related quality of life and patient satisfaction.

Methods

Subjects and data collection

We invited all 52 family practices within the boundaries of two adjacent inner London primary care organizations to participate in the study; 19 practices, with 553 eligible registered patients with type 2 diabetes of whom 209 agreed to participate (see report for further details11). Our estimates suggested that we required a total sample of about 320 subjects from 16 practices to detect a difference in HbA1c of 0.7% between two groups, with higher or lower continuity, with SD 1.75% and 90% power and an intraclass correlation coefficient by practice of 0.01. Participating subjects were visited at home and gave written informed consent to participation. The interview-administered questionnaire included the experienced continuity of care (ECC) measure; age, sex, ethnicity and language; socioeconomic variables; diabetes treatment; type of care arrangement. The short-form 12 (SF-12) measure was included at follow-up.12 An item was included asking ‘How satisfied would you say you are with your current diabetes treatment?’ Responses were recorded on a six-point scale ranging from ‘very dissatisfied’ to ‘very satisfied’. After the interview, measurements were made of height, weight and blood pressure using standard measurement procedures.13 Height and weight were measured in subjects wearing light indoor clothes but without shoes. Height was measured using a Soehnle electronic height measuring unit; weight was measured using electronic weighing scales (Seca 884). Blood pressure was measured using an Omron 705IT blood pressure measuring device, using a large cuff if the patient was obese.14 Fieldworkers were trained in measurement procedures and a pilot study was conducted on non-study subjects. Finally, a blood sample was taken, or a test arranged, for HbA1c estimation. Data were extracted from each patient's family practice record. Information was obtained concerning the number of different individual professionals seen for consultations during the 12 months before interview.

Analysis

The main analysis estimated the association between HbA1c as primary outcome and ECC. Blood pressure, weight and body mass index or subjective health status were evaluated as secondary outcome measures. An analysis of covariance framework was employed, with HbA1c as the dependent variable and experienced continuity as the predictor, adjusting additionally for baseline values of HbA1c. This analysis is designed to determine whether changes in HbA1c over the time of study were associated with ECC. Experienced continuity was fitted as the mean of the values at baseline and follow-up, but the results were not sensitive to whether either baseline or follow-up value or both were fitted. Additional adjustment was made for age, sex, ethnicity, duration of diabetes, diabetes treatment, housing tenure, educational qualifications and whether living alone. Models were fitted, with the patient's family practice as a random effect, using maximum likelihood estimation.

Patients’ satisfaction with care was evaluated using a scale ranging from one to six in value but as there were only seven values rated lower than four, the lowest four categories were combined. The mean values for the total ECC scale and the four subscales were tabulated by satisfaction rating. P-values for tests were trend, and the estimated mean difference in score between patients with lowest and highest satisfaction ratings were obtained from a random effects model. Additional models were fitted to evaluate associations with the number of consultations in the last 12 months and the number of different professionals seen, as consultation-based measures of continuity of care.

Results

There were 553 eligible subjects and 209 (38%) agreed to participate in the research and were interviewed at baseline. Of these, 177 (85%) were reinterviewed at the follow-up visit which was at a mean 43 (range 24–76) weeks later. The characteristics of the subjects included in the study are shown in Table 1. The mean age was 65 years (range 32–90 years) and 97 (50%) were women. Consistent with the inner city location of the study, 52% of subjects were from ethnic minorities with African and African Caribbean groups being the most frequent; 24% did not have English as their first language; 65% were in rented accommodation; only 7% had university qualifications and 35% were living alone. There were 193 subjects whose responses provided a total experienced continuity score at baseline and 156 who gave scores both at baseline and follow-up. There were no differences in experienced continuity scale scores or health measures at baseline for patients who were lost to follow-up and those who were reinterviewed (Table 2). Correlation coefficients between experienced continuity and intermediate outcome measures were very small in the baseline data (Table 2).

Table 1

Characteristics of 193 subjects included in study

Variable	Category	Frequency (%)
Age (years)	<45	12 (6)
	45–54	24 (12)
	55–64	48 (25)
	65–75	73 (38)
	≥75	36 (19)
Sex	Women	97 (50)
Duration of diabetes (years)	≤5	63 (33)
	6–10	61 (32)
	11–15	33 (17)
	≥16	32 (17)
	Not known	4 (2)
Diabetes treatment	Diet	17 (9)
	Tablets	131 (68)
	Insulin	45 (23)
Type of care at baseline	Family practice	44 (23)
	Hospital	35 (18)
	Shared care	114 (59)
Ethnicity	White	92 (48)
	Black Caribbean/African	60 (31)
	Indian subcontinent	18 (9)
	Other	23 (12)
Language	English not first language	47 (24)
Housing tenure	Owner occupied	67 (35)
	Council rented	93 (48)
	Housing association/private rented	33 (17)
Education	No qualifications/O level	143 (74)
	A level or technical	36 (19)
	University	14 (7)
Living alone		68 (35)
Current smoker		28 (15)
Self-reported morbidity	High blood pressure	134 (69)
	Angina	48 (25)
	Heart attack	27 (14)
	Stroke	19 (10)
Self-rated health	Fair or poor	116 (60)

Table 2

Comparison of patients lost to follow-up with those interviewed at follow-up [figures are mean (SD) except where indicated]

	Baseline values		P-value	Correlation of outcome with ECC score (number of observations)
	Followed-up (156)	Lost to follow-up (37)
HbA1c (%)	7.62 (1.73)	7.61 (1.69)	0.999	−0.007 (188)
Systolic blood pressure (mm Hg)	148.9 (22.2)	149.2 (20.2)	0.932	0.033 (193)
Diastolic blood pressure (mm Hg)	79.3 (10.9)	79.0 (9.9)	0.840	0.074 (193)
Weight (kg)	84.2 (16.5)	80.3 (14.9)	0.193	0.092 (192)
Height (m)	165.7 (9.6)	163.8 (10.1)	0.277	0.138 (190)
Body mass index (kg/m2)	30.7 (5.7)	30.1 (6.1)	0.568	0.032 (189)
Total ECC	62.0 (15.3)	62.6 (18.8)	0.732	—
Number of consultations in preceding 12 months	4.2 (3.1)	3.5 (2.5)	0.187	0.314 (179)
Self-rated health	Frequency (%)
Excellent/very good	16 (10)	4 (11)	0.367	—
Good	48 (31)	9 (24)
Fair	69 (44)	14 (38)
Poor	23 (15)	10 (27)

The results from the cohort study are shown in Table 3. The difference in outcome for each 10-unit increase in ECC was estimated after adjusting for the baseline value of the outcome. Total ECC was not associated with HbA1c, systolic blood pressure, diastolic blood pressure, weight, body mass index or SF-12 physical or mental component scores. When ECC was modelled as an ordinal variable with four quartiles, there was no difference in interpretation. Adjusting for a wide range of confounders in model 2 gave similar estimates.

Table 3

Difference in outcome measure [95% confidence interval (CI)] associated with a 10-unit greater ECC score

Outcome	Cases analysed	Change in outcome per 10-unit increase in experienced continuity
		Model 1a		Model 2b
		Coefficientc (95% CI)	P-value	Coefficientc (95% CI)	P-value
HbA1c (%)	151	−0.07 (−0.26 to 0.13)	0.504	−0.09 (−0.29 to 0.12)	0.402
Systolic blood pressure (mm Hg)	155	−1.72 (−4.1 to 0.63)	0.150	−0.41 (−2.88 to 2.06)	0.746
Diastolic blood pressure (mm Hg)	155	0.09 (−1.13 to 1.31)	0.886	0.24 (−1.03 to 1.51)	0.713
Weight (kg)	155	0.31 (−0.31 to 0.93)	0.329	0.23 (−0.40 to 0.86)	0.473
Body mass index (kg/m2)	150	0.03 (−0.23 to 0.28)	0.850	−0.08 (−0.34 to 0.18)	0.562
SF-12d
Physical component score (range 0–100)	153	1.03 (−0.54 to 2.61)	0.201	0.73 (−0.88 to 2.35)	0.375
Mental component score (range 0–100)	153	0.73 (−0.61 to 2.07)	0.286	0.33 (−1.11 to 1.77)	0.655

a

Model 1: adjusted for baseline value of outcome only.

b

Model 2: adjusted for baseline value of outcome, age, sex, ethnicity, duration of diabetes, type of treatment, qualifications, housing tenure and living alone.

c

Coefficients represent the difference in outcome per 10-unit higher total ECC score.

d

Models were adjusted for self-rated health at baseline.

Table 4 shows the association of ECC with patients’ global ratings of satisfaction with care received. The rating was made on a scale ranging from one to six in value but as there were only seven values rated lower than four, the lowest four categories were combined. Total ECC was strongly associated with patients’ global rating of satisfaction with their treatment. The estimated mean difference in experienced continuity between subjects with lowest and highest satisfaction ratings was 7.87, 95% confidence interval 3.22–12.5, P = 0.001. The subscales of flexible, relational and team and cross-boundary continuity were similarly associated with patient satisfaction. Longitudinal continuity, which depended on the number of consultations, was not associated with patient satisfaction. Relational continuity was only associated with satisfaction after additional adjustment for the type of setting in which care was provided.

Table 4

Association of continuity of care scores with global satisfaction ratings [figures are mean (SD) except where indicated]

	Satisfaction rating			P-valuea
	Lowest (43)	Middle (48)	Highest (65)
ECC
Total	60.9 (11.7)	62.8 (13.3)	68.9 (12.1)	0.001
Longitudinal	12.1 (5.4)	12.3 (6.0)	12.8 (5.5)	0.533
Flexible	16.9 (5.4)	18.5 (3.8)	20.6 (3.5)	<0.001
Relational	16.5 (4.6)	15.5 (6.6)	18.2 (5.1)	0.060b
Team and cross-boundary	15.4 (2.2)	16.5 (2.5)	17.4 (2.8)	<0.001

a

Test for trend across satisfaction ratings.

b

After adjusting for type of care setting (hospital, GP or shared) P = 0.028.

Table 5 shows the association of ECC with consultation-based measures. ECC increased with the number of consultations in the preceding 12 months, but decreased as the number of different individual professionals seen increased. Neither consultation measure was associated with the HbA1c value.

Table 5

Association of experienced continuity and HbA1c with consultation-based measures

	n	Median (range)	Experienced continuity score		HbA1c
			Coefficient (95% confidence interval)a	P-value	Coefficient (95% confidence interval)b	P-value
Number of consultations in last 12 months	135	5 (0–16)	2.21 (1.29 to 3.13)	<0.001	0.05 (−0.06 to 0.15)	0.398
Number of different professionals seen in last 12 months	135	2 (0–7)	−1.92 (−3.74 to −0.08)	0.041	0.05 (−0.16 to 0.27)	0.611

Discussion

In this health care setting, patient ECC is associated with overall ratings of satisfaction with care received. ECC also increases as the consultation frequency increases and the number of different professionals seen declines. However, over a period of approximately 10 months, ECC is not associated with intermediate outcomes of diabetes care including glycated haemoglobin, blood pressure and body weight or with health-related quality of life measured using the SF-12 questionnaire.

Interpretation

The view that establishing and maintaining ECC should result in better management decisions and improved personal care may be an oversimplification. This is illustrated by a number of hypothetical examples (Table 6). ECC may increase during the course of diabetic illness as patient and professional become better acquainted but, at the same time, the natural history of the diabetic illness is for blood glucose control to deteriorate.15 Health professionals will generally invest more effort in those patients whose health is deteriorating. Transitions in care are often associated with loss of continuity but may, under differing circumstances, be associated either with improving or deteriorating blood glucose control (Table 6). There may not be any direct link between patients’ experiences of continuity of care and the technical quality of care received. For example, Broom observed that familiarity may sometimes breed neglect.16 We suggest that blood glucose control and ECC may sometimes be positively or negatively associated, either in different patients or at different times in the same patient.

Table 6

Hypothetical scenarios giving differing associations between continuity of care and glycated haemoglobin values

Scenario	Events	Experienced continuity	HbA1c
Initial hypothesis	Patient and professional establish a relationship over time leading to improving treatment outcomes	↑	↓
Natural history of disease	Patient and professional establish a relationship over time; condition progresses over time with worsening blood glucose control	↑	↑
Transitions associated with lower continuity	Unsatisfactory blood glucose control; patient referred to hospital clinic for initiation of insulin	↓	↓
	Older practitioner retires, younger replacement adopts more intensive management approach	↓	↓
	Patient develops infected foot ulcer associated with worsening blood glucose control, requires admission to hospital and follow-up in hospital clinic	↓	↑

Strengths and limitations of study

The data collected for the study were obtained from face-to-face interviews and from specially organized measurements. We showed that the measure of ECC had satisfactory psychometric properties as well as good short-term repeatability.9 The measurement procedures were standardized and their reliability was demonstrated. A limitation of the study is the limited participation rate of approximately 38% of eligible subjects in the baseline survey. This participation rate is consistent with those generally reported from deprived inner city areas and low participation may limit the generalizability of the study findings. It may be difficult to judge the generalizability of any study and we caution that our results are most applicable to a health system such as the UK with stable patient registration in primary care. In other health systems, financial barriers to access may sometimes limit the development of continuity of care. The follow-up of subjects who participated at baseline was satisfactory and there was no difference in continuity of care or health measures between those who were successfully followed-up and those who were lost to follow-up. This is an important observation which provides evidence against a systematic non-response bias. While the achieved sample size was relatively modest, and smaller than planned owing to the low participation rate, the estimates were precise and it was clear to us that there was no evidence against the null hypothesis for any of the outcomes. Analysis of the sample as a single group with adjustment for baseline values gave a more precise analysis than the one initially envisaged in the sample size calculation. Adjustment for a wide range of confounders had negligible effect on the magnitude of the estimates and this argues against the possible influence of unmeasured confounders. Our findings are also consistent with previous work in other settings.

Comparison with previous studies

Two systematic reviews 4,17 have suggested that continuity of care is associated with more favourable patterns of health service utilization including fewer hospital admissions and greater uptake of preventive medical services. Previous empirical studies, including six cross-sectional studies18–23 and two cohort studies,24,25 investigated whether continuity of care is associated with glycated haemoglobin, using consultation-based measures of continuity, with inconsistent results. Some studies showed no association,21–23,25 others showed greater continuity associated with lower glycated haemoglobin,18,20,24 while one study found that greater continuity was associated with higher HbA1c.19 The present results are therefore compatible with those of previous studies.

Conclusion

We conclude that ECC is not an important influence on intermediate-term changes in glycated haemoglobin, blood pressure, body weight or health-related quality of life in subjects with type 2 diabetes in this health care setting. Our study contributes by showing that patients’ ratings of experienced continuity are strongly associated with overall satisfaction with care but with negligible estimated differences in HbA1c, blood pressure and body weight. We agree with Christakis26 therefore that experienced continuity should be valued as an indication of enhanced patient centeredness but our results suggest that experienced continuity is not directly associated with effectiveness of care. This is consistent with the results of the Diabetes from Diagnosis Trial27 in which the implementation of patient-centred consulting was not associated with better measures of disease control. Our data do not address questions of patient safety. It is plausible that continuity of provider may be associated with fewer adverse events such as prescribing errors, or avoidable metabolic emergencies including hypoglycaemic episodes, or late presentations with advanced diabetic complications. Larger studies would be required to determine whether such associations exist.

Declaration

Funding: This work was supported by the UK National Health Service Research and Development Programme in Service Delivery and Organization (SDO). The funding body commented on the research design but played no role in the collection, analysis and interpretation of data. The results reported here are included in our final report to the funders which can be accessed at http://www.sdo.lshtm.ac.uk/files/project/14-final-report.pdf. The results were also presented at the SDO Annual Conference, London, May 2006.

Ethical approval: The study was approved by the Research Ethics Committee at Guy's Hospital and subjects gave written informed consent to participation.

Conflicts of interest: None.

Contributors: All three authors contributed to the conceptual framework and the development of the continuity measure. SN led the fieldwork and completed the clinical data extraction. MCG analysed the quantitative data and drafted the paper. MM and SN contributed to the writing-up and all authors approved the final version. MCG is guarantor.

Author notes

Gulliford MC, Naithani S and Morgan M. Continuity of care and intermediate outcomes of type 2 diabetes mellitus. Family Practice 2007; 24: 245–251.