Multimorbidity in a cohort of patients with type 2 diabetes

ABSTRACT

Background: People with type 2 diabetes frequently have a variety of related and unrelated chronic conditions. These additional conditions have implications for patient education, treatment burden and disease management.

Objectives: The aim of this study was to examine the nature of multimorbidity, and its impact on GP visits, polypharmacy and glycaemic control as measured by HbA1c, in a cohort of patients with type 2 diabetes attending general practice in Ireland.

Methods: A cohort of 424 patients with type 2 diabetes enrolled in a cluster randomized controlled trial based in Irish general practice was examined. Patient data included: medical conditions, HbA1c, health service utilization, socio-economic status and number of prescribed medications.

Results: 90% of patients had at least one additional chronic condition and a quarter had four or more additional chronic conditions. 66% of patients had hypertension; 25% had heart disease; and 16% had arthritis. General practitioner visits and polypharmacy increased significantly with increasing numbers of chronic conditions. When comparing patient self-report with medical records, patients who reported a higher proportion of their conditions had better glycaemic control with a significantly lower HbA1c score.

Conclusion: There was a high prevalence of multimorbidity in these patients with type 2 diabetes and the results suggest that glycaemic control is related to patients’ awareness of their chronic conditions. The variety of conditions emphasizes the complexity of illness management in this group and the importance of maintaining a generalist and multidisciplinary approach to their clinical care.

Keywords::

• Diabetes
• patient involvement

Key message:

• The variety of conditions experienced by individuals with type 2 diabetes emphasizes the complexity of illness management and the importance of maintaining a generalist and multidisciplinary approach to clinical care

• Patients with a better awareness of the spectrum of conditions they have, appear to have a better management of their type 2 diabetes suggesting that patients need to be well-informed about their health generally and not just about type 2 diabetes.

INTRODUCTION

There is an increasing body of literature on the clinical importance of multimorbidity, which is defined as the co-existence of two or more chronic conditions (1). It has sometimes been described as co-morbidity, but the term co-morbidity implies an index disease to which coexistent diseases relate and may share an aetiology (2). Multimorbidity is common in primary care patients, with at least 50% of patients over 50 years of age having two or more chronic conditions (3). Multimorbidity is associated with increased GP visits and polypharmacy, both having implications for GP workload (3,4).Type 2 diabetes is one of the most common chronic conditions and there are well recognised comorbidities, particularly relating to diabetes and depression (5). However, many people with type 2 diabetes also have potentially unrelated chronic conditions. These additional conditions have implications for disease management, patient education and outcomes.

Previous studies examining chronic conditions in patients with type 2 diabetes have examined counts of chronic conditions though some have restricted these comorbidities to lists of more common chronic conditions. They have generally found an association between increasing comorbidities, higher health service utilization and impaired physical functioning (6–8). However, analysis of health service usage in the USA indicates that quality of care provision can improve with increasing levels of multimorbidity, perhaps because multimorbidity is associated with higher numbers of visits, leading to greater opportunity for preventive health care activities to be carried out (9).

The aim of this study was to examine the nature of multimorbidity, and its impact on GP visits, polypharmacy and glycaemic control as measured by HbA1c level, in a cohort of patients with type 2 diabetes attending general practice in Ireland. In addition, we sought to examine differences between patient self-report of chronic conditions and practice records.

METHODS

Study population

A cohort of 424 patients with type 2 diabetes enrolled in a cluster randomized controlled trial based in Irish general practice was examined. Eligible patients had to be aged over 18, have type 2 diabetes for at least a year, be able to participate in group sessions, and be attending one of the 20 practices participating in the study (10). Participating patients were randomly selected from those eligible within each practice. Non-participants were similar to participants in terms of sex, age and socioeconomic status. Baseline data was collected between November 2006 and April 2007.

Data

Medical conditions were recorded using two different approaches. Practice nurses recorded chronic conditions reported in medical records and patients were also asked to report chronic conditions. Patients were prompted with four conditions: heart disease, high blood pressure, chest/lung disease and mental illness, but were given space to record any additional conditions. Conditions reported were coded using the primary care specific ICPC-2 for coding illnesses (11). Coding was carried out independently by two academic general practitioners (SMS and TOD). Disagreements were resolved through discussion and consensus. Chronic conditions were defined as illnesses with a course of greater than three months. More generally, these are long-term conditions that are managed rather than cured. Acute conditions and those that were considered complications of another chronic condition were excluded. However, while some conditions such as ischaemic heart disease and diabetes have underlying linked pathology, we classified them as separate conditions if each appeared apart as ICPC-2 codes and could require separate specialist input.

Patients’ chronic conditions were determined as the list of unique conditions generated by combining the practice recorded and self-reported chronic conditions. The proportion of patients’ total chronic conditions that were self-reported was also examined and used as a measure of a patient's knowledge of their own chronic conditions.

Other patient details collected included: HbA1c, socio-economic status, body mass index (BMI), health service utilization and number of prescribed medications. Medications were recorded by name and converted into a count of unique medications. Socio-economic status was measured by whether the patient had a means-tested access to free medical care. Primary health care services in Ireland are free through the General Medical Services (GMS) scheme to the poorest 30% of the population. Eligibility is means-tested based on the total household income. At the time of the study, primary care services were also free for those aged 70 years or older. Those not eligible for the GMS scheme must pay out of pocket for GP consultations. Primary care service utilization was recorded by number of GP and practice nurse visits in the previous 12 months.

Analysis

In the study cohort, individual chronic diseases and chronic disease pairs were ranked by frequency of occurrence to examine the nature of multimorbidity. A network diagram was used to illustrate the most common disease pairs.

To assess the impact of multimorbidity, differences in the numbers of GP visits, polypharmacy and HbA1c by number of chronic conditions were compared using the Wilcoxon signed-rank test.

A linear model was used to predict HbA1c using patient characteristics including age; sex; medical card eligibility (as a measure of socio-economic status); BMI; cholesterol level; general practice visits; practice nurse visits; number of medications; number of chronic conditions; and proportion of patient's chronic conditions that were self-reported. Cholesterol level and the proportion of patient's chronic conditions that were self-reported were both log transformed due to the skewed nature of the variables.

The alpha level was set at 0.05 for all statistical tests. Calculations were carried out in R 2.12 (12).

RESULTS

Study population

The cohort contained 424 patients. The median age was 64 (IQR 57–72) and 54% were male. Of the 424 patients, 90% (n = 384) had at least one additional chronic condition and 25% (n = 104) had four or more additional chronic conditions. The number of chronic conditions increased with participant's age and socioeconomic deprivation. The difference in the median number of chronic conditions by sex (male = 3.4, female = 3.8) was not significant (P = 0.080), but for socio-economic status (high = 3.3, low = 3.8) the difference was statistically significant (P = 0.014). Patients tended to have good cholesterol control, with a median cholesterol level of 4.2 mmol/l, and 77% (n = 328) having cholesterol below 5 mmol/l (Table 1).

Table 1. Baseline characteristics of study cohort of patients with type 2 diabetes.

Males, n (%)	227	(53.5)
Age, n (%):
< 50	42	(9.9)
50–59	108	(25.5)
60–69	125	(29.5)
70–79	116	(27.4)
80 +	33	(7.8)
GMS card, n (%)	170	(40.1)
Chronic conditions^a, n (%):
1	40	(9.4)
2	94	(22.2)
3	104	(24.5)
4 +	186	(43.9)
Medications, n (%):
0	14	(3.3)
1	6	(1.4)
2	13	(3.1)
3	43	(10.1)
4	45	(10.6)
5	69	(16.3)
6 +	234	(55.2)
GP visits in 12 months, median (IQR)	5	(2, 7.5)
Practice nurse visits in 12 months, median (IQR)	3	(2, 4)
HbA1c (%), mean (SD)	7.1	(1.3)
BMI (kg/m^2), mean (SD)	31.1	(5.9)
Cholesterol (mmol/l), mean (SD)	4.3	(1.1)

IQR, interquartile range; SD, standard deviation.

^aCount of chronic conditions is inclusive of type 2 diabetes.

Nature of multimorbidity

We identified 189 different chronic conditions with unique ICPC-2 codes that were present in at least one patient. Almost 74% of patients had at least one circulatory condition with 36% having at least one metabolic/endocrine/nutritional condition in addition to diabetes (Table 2). Patients recorded co-morbidities from a median of two chapters (IQR 1–3) with 3% (n = 14) patients recording co-morbidities from five or more chapters. A total of 1064 additional chronic conditions were recorded for the 424 patients with 45% of these being circulatory conditions.

Table 2. Number of patients and conditions by ICPC-2 chapter in a cohort of patients with type 2 diabetes.

ICPC-2 chapter	Description reported	Prevalence^a	Conditions
		n (%)	n (%)
A	General	2 (0.5)	2 (0.2)
B	Blood/blood forming	17 (4.0)	19 (1.8)
D	Digestive	53 (12.5)	61 (5.7)
F	Eye	32 (7.5)	32 (3.0)
H	Ear	7 (1.7)	7 (0.7)
K	Circulatory	313 (73.8)	481 (45.2)
L	Musculoskeletal	79 (18.6)	87 (8.2)
N	Neurological	21 (5.0)	21 (2.0)
P	Psychological	49 (11.6)	51 (4.8)
R	Respiratory	71 (16.7)	73 (6.9)
S	Skin	12 (2.8)	12 (1.1)
T	Metabolic/endocrine/nutritional	152 (35.8)	178 (16.7)
U	Urinary	18 (4.2)	18 (1.7)
X	Female genital	8 (1.9)	8 (0.8)
Y	Male genital	14 (3.3)	14 (1.3)

^aIndividual patients reported multiple conditions, sometimes within the same chapter. Each chapter has only been recorded once per patient for the purposes of this table.

The most common additional condition was hypertension, which was present in two thirds of patients. The other commonly reported conditions were heart disease (23.1%), arthritis (including arthritis, osteoarthritis, rheumatoid arthritis, gout and spondylitis; 16.0%) and high cholesterol level (14.6%). 54% (n = 230) of patients had a BMI of 30 or higher, but obesity was only reported in 5% (n = 23) of patients by either the practice or the patient (Table 3).

Table 3. Prevalence of most common chronic conditions in a cohort of patients with type 2 diabetes.

Condition	Patients n (%)
Hypertension	281 (66.3)
Heart disease	107 (25.2)
Arthritis^a	69 (16.3)
High cholesterol	62 (14.6)
Hyperlipidaemia	39 (9.2)
Chest/lung disease	32 (7.5)
Asthma	26 (6.1)
Obesity	23 (5.4)
Hypothyroidism	18 (4.2)
Depression	17 (4.0)

^aArthritis includes arthritis, osteoarthritis, rheumatoid arthritis, gout and spondylitis.

In terms of disease pairs, all pairs observed in five or more patients included either hypertension or heart disease. A network diagram of the disease pairs observed in 10 or more patients is shown in Figure 1. Circles are proportional to the number of patients with that condition while the width of connecting nodes is proportional to the number of patients with that disease pairing. After hypertension with heart disease, the most common pairings are hypertension with arthritis and high cholesterol level, respectively.

Figure 1. Network diagram of most common chronic condition pairs in a cohort with type 2 diabetes. Circles are proportional to the number of patients with that condition while the width of connecting nodes are proportional to the number of patients with that disease pairing. For example, 53 patients have arthritis and hypertension. Figures in brackets refer to the total number of patients with a particular condition (e.g. 12 patients had cataracts). Only pairings observed in 10 or more patients are shown. Arthritis includes arthritis, osteoarthritis, rheumatoid arthritis, gout and spondylitis.

Impact of multimorbidity

The median number of GP visits in the previous 12 months increased significantly with the number of chronic conditions (P = 0.039). The median numbers of visits were three and eight for those with one and nine or more chronic conditions, respectively.

Polypharmacy increased significantly with the number of chronic conditions (P < 0.001) (see Table 4). Just over half the patients (n = 234) were being prescribed six or more medications. Those with only one condition (i.e. type 2 diabetes) were on average prescribed 3.4 medications.

Table 4. Service utilization, polypharmacy and HbA1c by number of chronic conditions in a cohort of patients with type 2 diabetes.

Number of chronic conditions	Annual GP visits	Annual practice nurse visits	Annual primary care visits^a	Concurrent medications	HbA1c (%)
	Mean (SD)	Mean (SD)	Mean (SD)	Mean (SD)	Mean (SD)
1	4.18 (4.45)	2.45 (1.52)	6.30 (5.32)	3.40 (2.18)	7.35 (1.34)
2	5.30 (4.22)	3.09 (1.85)	7.85 (5.11)	5.00 (2.59)	7.12 (1.30)
3	6.17 (4.99)	3.15 (2.55)	8.96 (6.48)	5.81 (2.75)	7.19 (1.35)
4	5.99 (4.6)	3.51 (2.78)	9.04 (6.2)	7.28 (2.79)	7.03 (1.36)
5 +	6.05 (5.48)	3.40 (2.55)	9.00 (7.14)	8.04 (2.88)	7.11 (1.14)

SD, standard deviation.

^aPrimary care utilisation is the combination of GP and practice nurse visits.

The median HbA1c was 6.9 (IQR 6.3–7.6) across all patients and there was no statistically significant variation in HbA1c by the number of chronic conditions.

Differences in reporting of conditions between practices and patients

The median number of chronic conditions reported from practice records was three, while participants self-reported a median of two chronic conditions. Three of the five conditions most commonly reported by patients but not the practice were those prompted in the questionnaire (chest/lung disease, heart disease and hypertension). 33 participants had diagnoses of depression or mental illness with 10 of these not self-reported. The differences between practice reporting and self-report vary depending on the condition. For example, 3% (n = 13) patients reported arthritis where there was no practice record of the condition. Overall, patients tended to report fewer conditions than the practices. The most common practice-reported conditions not reported by patients included: hypertension, 9% (n = 39); high cholesterol, 8% (n = 34); hyperlipidaemia, 8% (n = 33); arthritis, 6% (n = 27); and obesity, 4% (n = 18). Of the 54% (n = 230) of patients with a BMI of 30 or higher, only 5 self-reported obesity as a chronic condition. Practice records reported 23 as being obese, which included the 5 patients who self-reported obesity.

From the linear regression model, three patient level characteristics were statistically significant predictors of HbA1c: patient age (P = 0.02), and the proportion of patient's chronic conditions that were self-reported (P = 0.002) had negative coefficients indicating that an increase in these predictors was associated with a decrease in HbA1c (Table 5). High cholesterol level (P = 0.003) and number of medications (P = 0.01) were associated with poorer glycaemic control.

Table 5. Regression coefficients (β) from a linear model predicting HbA1c percentage in a cohort of patients with type 2 diabetes.

Variable	Regression coefficient	P-value
(Intercept)	6.474	< 0.0001
Age	–0.016	0.018
Sex (female)	–0.038	0.769
GMS card (yes)	0.113	0.405
BMI	0.003	0.812
Log (cholesterol level)	0.860	0.003
Number of GP visits	0.001	0.926
Number of practice nurse visits	–0.013	0.654
Number of medications	0.068	0.010
Number of chronic conditions	–0.071	0.119
Log (proportion of conditions self-reported)	–0.600	0.002

DISCUSSION

The prevalence of multimorbidity and median number of conditions in the study cohort was higher than has been seen in some comparable studies of general populations (13). The prevalence of two or more chronic conditions, in addition to type 2 diabetes, was 68.4% equivalent to the figure of 66.2% reported for a cohort of Irish primary care patients (3). The types of co-morbid conditions present span the spectrum of ICPC-2 chapters and the variety of conditions emphasizes the complexity of illness management in this group and the importance of maintaining a generalist approach to their clinical care and education. It also emphasizes the challenges for patients’ managing their own conditions and highlights the potential burden of care when multiple treatments are indicated for different conditions (14). Clinicians and educators need to be aware of this potential treatment burden and support patients in prioritizing management strategies. Diabetes should not be treated in isolation, and we must consider how other conditions impact on health.

Nature of multimorbidity

Although patients with type 2 diabetes experience a wide range of other chronic conditions, the largest burden is generated by circulatory conditions. 74% of patients has at least one circulatory illness, which is far more than of the burden of additional metabolic/endocrine/nutritional conditions, which affect 36% of the cohort. The network diagram presented shows that there are a limited number of common disease pairs. This has implications for clinical guidelines as it would be possible to consider producing guidelines for the most common combinations, which would support clinicians managing these patients.

Impact of multimorbidity

The presence of additional chronic conditions does not appear to have a detrimental impact on an individual's glycaemic control, although the sample size is too small to be definitive. An increased number of conditions is associated with increased GP visits and might, therefore, lead to better management of HbA1c. This is consistent with the results found by Higashi et al. (9). Polypharmacy is associated with type 2 diabetes and the presence of additional chronic conditions was shown to exacerbate this problem as has been found in other studies (3,15).

Differences in reporting of conditions between practices and patients

Study participants tended to record fewer conditions than were reported in their medical records suggesting they may underestimate their burden of disease. Self-report can be an effective way of measuring multimorbidity in patients although the accuracy of self-report is condition specific (16,17). Consistent with other countries, the Irish population tend to underestimate BMI when self-reporting weight and height (18). Patients correctly reporting a higher proportion of their chronic conditions had significantly lower HbA1c levels, suggesting that patients who are knowledgeable about their conditions may manage their health status better. It may also suggest that investing time in educating patients about their illnesses could have benefits in terms of their engagement in treatment and illness management.

Strengths and limitations

This study analysed a RCT cohort generated by random selection from general practice diabetes registers in Ireland. However, it is a relatively small cohort of patients and generalizability may be reduced given that they were a group of patients who had consented to participate in a trial of per support for type 2 diabetes. However, while non-participants were demographically similar to participants we could not collect HbA1c levels from non-participants so it is possible that non-participants had poorer glycaemic control. However, the focus of this study is on multimorbidity and its impact, and we do not think these results would be significantly affected by selection bias. The comparison of medical records and patient self-report is limited by the possibility that some conditions could have been missed in the records as well and some patients did report chronic conditions that were not recorded in practice records. The conditions that were most commonly under-reported by patients included hypertension, high cholesterol and hyperlipidaemia, all conditions that are unlikely to have a big impact on day to day symptoms. It is possible that patients do not always classify chronic conditions as illnesses, but simply interpret them as part of their general condition. In general, this study highlights the variation in disease reporting depending on the source of information and that this is condition dependent.

Conclusion

This study highlights the high burden of multimorbidity in these patients with implications for health service utilization and polypharmacy. Multimorbidity needs to be considered when delivering patient education, giving risk reduction advice and making treatment recommendations. Patients with a better awareness of their conditions, appear to have better management of their type 2 diabetes, suggesting that patients need to be well-informed about their health generally and not just about type 2 diabetes.

FINANCIAL SUPPORT

CT was supported by the Health Research Board of Ireland through the HRB Centre for Primary Care Research under Grant HRC/2007/1. GP was supported by the Health Research Board of Ireland under Grant S/A 009.

ACKNOWLEDGEMENTS

The authors wish to thank the participating patients and practice staff for their time and effort.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.