Multimorbidity Patterns in Patients with Back Pain: A Study of Patient Records at a Primary Health Care Centre in Sweden

1.                   Abstract

In Primary Care, multimorbidity is the norm in most patients. A large part of these have pain disorders, very often related to the spine. Patients with back pain have a higher degree of multimorbidity than many other groups of patients. The aim of this epidemiological study was to elucidate various patterns of multimorbidity in terms of clusters of diseases among patients with low back pain (LBP).

1.1.              Methods: A retrospective cross-sectional study was performed containing all registered encounters with patients receiving a LBP related diagnosis at one Primary Health Care Centre (PHCC) in Stockholm area, Sweden. The period October 2011 to September 2014 was studied. The Johns Hopkins Case-mix System “Adjusted Clinical Groups” (ACG^©) was used for grouping and analysing data.

1.2.              Results: Out of 15,092 patients visiting the PHCC during the 3-year period exactly 10,000 got at least one diagnosis and 1,431 of those patients were diagnosed with LBP. Most common simultaneous groups of diagnoses were in order administrative concerns, hypertension, other musculoskeletal disorders and neurologic signs and symptoms. The proportion of patients with LBP disorders having five or more diagnoses was about 29%, and the equivalent proportion of patients without LBP was 9%. Different types of morbidity in terms of Aggregated Diagnosis Groups (ADGs) showed that about 55% of patients with LBP had three or more ADGs compared to 26% among patients who had no LBP. Patterns of multimorbidity in terms of the ACGs showed that patients with LBP were about twice as common in higher risk categories than patients without those diagnoses (52% vs 26%)

1.3.              Discussion: Our study showed that patients with LBP had a high degree of multimorbidity compared to those who did not have LBP and type of concurrent diseases differed between the two groups. The patterns of diagnosis clusters were analysed further and showed results that differed between various groups of patients with LBP, predominately depending on age. Further analysis is needed in order to understand what causes the various patterns of multimorbidity among patients with LBP.

2.                   Keywords: Low Back Pain; Multimorbidity; Adjusted Clinical Groups; Primary Care

1.                   Introduction

The concept of multimorbidity is defined as the simultaneous occurrence of several diseases where none of them is seen as an index disease [1,2]. Some international studies are focusing on the overall multimorbidity [3,4]. Studies in Sweden have shown patterns of multimorbidity, departing from some specific diseases [5,6]. However, there is a trend towards more patient oriented, or person-centred care, indicating a greater interest to deal with the consequences of co-morbidity and multimorbidity [7,8].

In Primary Care, as the first tier, all diseases might be relevant in the diagnostic process in order to decide what kind of treatment is the most suitable one in each case. Thus, in order to be able to deliver adequate care, the understanding of multimorbidity ought to play an important role.

Patients with back pain disorders often suffer from pain also in other parts of their body [9,10]. It is also known that low back pain (LBP) plays a central role in multimorbidity [11]. Much activity within Primary Care relates to LBP and its treatment; this group of patients is the single most frequent among patients with pain [6]. Most patients with LBP tend to become chronically ill, also leading to sick leave. Swedish studies describing LBP in general care have been published [12], but so far just two on Primary Care level [6,13].

The purpose of this study was to describe multimorbidity in patients with LBP and analyse various patterns of simultaneous diagnoses. Our study was approved by the regional ethical approval board in Stockholm, Sweden (Dnr: 2015/232-31/5).

2.                   Materials and Methods

This study is a retrospective cross-sectional approach based on patient’s record data from one Primary Health Care Centre (PHCC) in Stockholm Region in Sweden. All patients, enrolled at the PHCC at the end of September 2014, were included in our study. Information from the medical records regarding all those patients’ visits to physicians at the PHCC between 2011-10-01 and 2014-09-30 was retrieved. Data used were the age and gender of the patient, dates of the visits and all registered diagnoses. Identity numbers were decoded before the usage. Every diagnosis by each patient was retrieved. Patients with LBP were identified by the following three ICD-10 diagnoses: M54.4 (Lumbago with ischia), M54.5 (Lumbago) and M54.9 (Unspecified back pain). This group of patients was compared with all other patients, not given those diagnoses, who visited the PHCC each year during the defined 3-year period (Table 1). The three periods from October one year to September next year in our article further on are marked with “2012”, “2013” and “2014” respectively.

Data was processed and analysed by the Johns Hopkins Case-mix System, Adjusted Clinical Groups (ACG^©), version 11.1 [14]. The resulting patterns of multimorbidity were represented in three dimensions; the 93 patient complexity categories (ACG), the 32 Aggregated Diagnosis Groups (ADG) and the 286 Expanded Diagnosis Clusters (EDC), bundled into 27 major clusters (MEDC), all built into the ACG system. When grouping into patient categories, the ACGs, the system uses ADGs that differentiates between types of morbidity for each diagnosis, meaning that the ADGs are the building blocks, which are combined, when constructing the ACGs. The intention of the EDCs is to describe, in a clinically meaningful way, what clusters of diagnoses are involved in a complex patient category.

ACGs were designed to represent categories for persons expected to require similar levels of healthcare resources. Because patients have different epidemiological patterns of morbidity they fall into different ACG categories. The full set of ACGs can be collapsed into six classes, Resource Utilization Bands (RUB), depending on the expected low or high use of healthcare resources.

3.       Results

The total number of patients involved in our study was 15,092 and 10,000 of them were registered with a diagnosis during the 3-year period. Among the latter, 53.4% were female and 46.6% were male patients.

The distribution of age groups is presented in Figure 1. Middle age groups contained a great proportion of patients with LBP.

The number of diagnoses per patient during each year differed between those patients with LBP and the others, as shown in Figure 2.

The multimorbidity pattern displayed by different types of morbidity, the ADGs, during 2014 for patients diagnosed with LBP is presented in Figure 3. Comparison was made with all other patients visiting the PHCC during the same time. There are similar numbers for all three years, with about the same proportion between the two populations compared.

The multimorbid patient categories in terms of ACG for patients with LBP at the end of the period studied is shown in a table attached to this article, Appendix A. Comparison is made with all patients without LBP for each of the three years in Figure 4, reduced to the ten most frequent groups of patients with vs without LBP the last year.

As shown in Figure 4, patients with LBP were represented in the more complex categories of the ACGs (21xx-49xx) to a higher degree.

The multimorbidity patterns are more obvious when the ACGs were collapsed into RUBs. This is shown below for all three years (Figure 5). RUB 0, containing patients with no diagnoses at all, is not displayed here.

The diagnosis clusters, the EDCs, differed between patients with LBP and the others. A total of all EDC in numbers, summarized in terms of MEDC, are shown for all three years as Appendix B.

Table 2 compares the distribution of all MEDCs during 2014 for the two populations and for the sum of them, meaning all patients registered with a diagnosis during 2014. The distribution is shown as a percentage within each population.

In Figure 6 the top-ten MEDCs during 2014 from the two patient groups are presented. The 30 most common EDCs among patients in our study, are shown in Table 3. Table 3a displays the population with LBP, while Table 3b contains the population without LBP; both regarding the year 2014 and in numbers of patients.

A comparison of the EDC distribution 2014 between the two populations is presented in Figure 7. The ten most frequent EDCs from each population are displayed (% of total EDC in each group).

4.       Discussion

Our study showed that patients with LBP had a higher degree of multimorbidity compared to patients without LBP. Patients with LBP often have combinations of diagnoses with other musculoskeletal disorders more often than patients without LBP. Abdominal pain tends to be part of the multimorbidity of patients with LBP [15] as are some neurological signs and symptoms.

One limitation of our study was that just one PHCC has been examined with a relatively limited number of patients with LBP involved, less than 1,000 patients each year. Thus, we made no efforts to study specific correlations.

A possible strength of our study was that we were able to elucidate the multimorbidity patterns over a 3-year period following the same population all years. We found a robustness in terms of ACG patterns, although a longer follow-up period maybe would provide a more detailed view of multimorbidity in terms of types of morbidity involved, the ADGs, and maybe some changing patterns among the diagnosis clusters, the EDCs.

5.       Further Studies

In an ongoing study we have identified more than 10,000 patients with LBP, enabling us to analyse various clusters of diagnoses (EDCs) to elucidate detailed patterns of multimorbidity. Variations between male and female patients might be of interest. Having data for four consecutive years, we will be able to study in what order the connecting diagnoses will appear.

This our study was designed with LBP disorders as a point of departure. It might be of interest to investigate multimorbidity identifying patient groups with other diseases as point of departure, such as depression, sleeping or neurological disorders. Furthermore, back pain might be studied as a trigger for pain in other parts of the body, not just localized to the spine.

Correlations by causality were not examined in this study. With a data set with more than 1 million patients we will be able to stratify the population into groups with various combinations of diseases.

6.       Conclusions

Patients with LBP had more unique diagnoses and more various types of morbidity than patients without LBP. The degree of multimorbidity was higher among patients with LBP than in average, in terms of more complex combinations of diagnoses. The number of chronic diseases seemed not to be the most important factor. Instead, the variation of clusters of diagnoses had a great influence on the complexity, and the need for use of health care resources.

Figure 1: Distribution of age groups among patients with vs without low back pain vs all patients, year 2014.

Figure 2: Diagnoses per patient with vs without low back pain.

Figure 3: ADGs per patient with low back pain vs all other visiting patients, year 2014.

A: Sorted by patients with low back pain.

B: Sorted by patients without low back pain.

Figure 4: ACG distribution 2014 among patients with vs without low back pain.

Figure 5: RUB distribution among patients with low back pain vs all other patients.

Figure 6: Top-ten MEDCs among patients with vs without low back pain, year 2014.

Figure 7: EDC distribution 2014 – most common clusters among patient with vs without low back pain.

1. Starfield B (2006) Threads and yarns: weaving the tapestry of comorbidity. Ann Fam Med 4: 101-103.
2. Starfield B, Kinder K (2011) Multimorbidity and its measurement. Health Policy 103: 3-8.
3. Prados-Torres A, Calderón-Larrañaga A, Hancco-Saavedra J, Poblador-Plou B, van den Akker M (2014) Multimorbidity patterns: a systematic review. J Clin Epidemiol 67: 254-266.
4. Poblador-Plou B, van den Akker M, Vos R, Calderón-Larrañaga A, Metsemakers J, et al (2014) Similar multimorbidity patterns in primary care patients from two European regions: results of a factor analysis. PLoS One 9: e100375.
5. Wändell P, Carlsson AC, Wettermark B, Lord G, Cars T, et al (2013) Most common diseases diagnosed in primary care in Stockholm, Sweden, in 2011. Fam Pract 30: 506-513.
6. Olofsson H, Carlsson L, Bertilson BC (2018) Multimorbidity among Patients with Back Pain: A Study of Records at a Swedish Primary Health Care Centre, J Family Med Prim Care Open Acc: JFOA-118.
7. Smith SM, Soubhi H, Fortin M, Hudon C, O’Dowd T (2012) Managing patients with multimorbidity: systematic review of interventions in primary care and community settings. BMJ 345: e5205.
8. Valderas JM, Starfield B, Sibbald B, Salisbury C, Roland M (2009) Defining comorbidity: implications for understanding health and health services. Ann Fam Med 7: 357-363.
9. Hartvigsen J, Natvig B, Ferreira M (2013) Is it all about a pain in the back? Best Pract Res Clin Rheumatol 27: 613-623.
10. Nordeman L, Gunnarsson R, Mannerkorpi K (2012) Prevalence and characteristics of widespread pain in female primary health care patients with chronic low back pain. Clin J Pain 28: 65-72.
11. Schäfer I, Kaduszkiewicz H, Wagner HO, Schön G, Scherer M, et al Reducing complexity: a visualisation of multimorbidity by combining disease clusters and triads. BMC Public Health 14: 1285.
12. Jöud A, Petersson IF, Englund M (2012) Low back pain: epidemiology of consultations. Arthritis Care Res 64: 1084-1088.
13. Dong HJ, Wressle E, Marcusson J (2013) Multimorbidity patterns of and use of health services by Swedish 85-years-old: an exploratory study. BMC Geriatr 13: 120.
14. (2016) ACG System^® Version 11.1 Applications Guide.
15. Bertilson BC, Heidermark A, Stockhaus M (2014) Irritable Bowel Syndrome-a Neurological Spine Problem. Br J Med Med Res 4: 4154-4168.