How does comorbidity affect cost of health care in patients with irritable bowel syndrome? A cohort study in general practice

Study design and material

This prospective cohort study was designed to identify and follow up a representative sample of patients with IBS (according to Rome II criteria) in Norwegian general practice. Of 3369 consecutive patients consulting their GP, 3092 (92%) answered a short questionnaire regarding abdominal complaints, prior to the consultation. Of 830 patients with abdominal complaints, for which they had consulted or wanted to consult their GP, 733 (89%) were assessed with the Rome II criteria for functional bowel disorders [11]. Of the 733 patients 278 patients had IBS according to the Rome II criteria. These 278 patients were invited to participate if the GP, based on all available information about the patient, agreed on the IBS diagnosis. A total number of 208 patients with IBS were included. Details about study design are given elsewhere [10]. The patients were invited by informed consent to further characterisation, by completing questionnaires at the first visit. After 6-9 months, included patients were invited to a follow up interview with repeated assessment of symptoms and detailed assessment of health resource use. No specific attention towards the abdominal complaints was to be given by the GP in the follow up period.

The study was performed during 2001in the county of Oppland in Norway, with 26 participating GPs working in nine health centres. In Norway, patients must seek health care through their locally assigned GP. Members of the practice staff (study nurses) performed the practical work related to the study. They were released from other duties, and trained to ensure satisfactory protocol adherence.

All patients were characterised at baseline for sociodemographic data and somatic and psychiatric comorbidity by self-completed questionnaires. Sociodemographic data included age, gender, smoking status, alcohol use, education (> 10 years), working status (employed, disability benefit, retired, students). IBS symptoms were assessed for severity and duration. Severity was measured as mild (with no interference of daily activities), moderate (with some interference, but no disruption of daily activities) and severe (with disruption of daily activities) with scores 1-3, and frequency as average number of days per week with abdominal pain/discomfort (0-1/2-3/4-5/> 5) with scores 1-4. An IBS symptom score was calculated by multiplying severity and frequency (range 1-12). Somatic diseases were measured by 12 questions about different organic diseases present within the 3 last months (yes/no) and presented as a sum-score (range 0-12). Somatic symptoms were measured by 17 questions about somatic symptoms from the Subjective Health Complaints (SHC) inventory [12]. The SHC inventory consists of a list of 29 common health complaints experienced the last 30 days, with the intensity of each complaint graded on a four point scale (not at all/little/some/severe). An SHC total score is created by adding the scores of each item. 12 questions about gastroenterological and psychiatric symptoms were excluded to avoid duplicate assessment, giving a SHC-17 score from 0 to 51. Psychiatric comorbidity was measured by Symptom Check List 10 (SCL-10, score range 1 - 4.0), which measures psychological distress, with a cut off point of > 1.85 as a valid predictor of mood disorder. Health anxiety was measured by Whitley index (score range 14-60) and neuroticism by EPQ-10 (score range 0-10) [13–15].

Measurement of health resource use and associated costs

Health resource use and associated costs were measured for investigations, consultations, hospitalisations, medications, and alternative health care products and number of sick leave days was noted. Assessment was primarily done by structured questions in a Case Report Form (CRF) completed by study nurses at the follow up interview. All questions in the CRF separated use of health resources related to symptoms of IBS (IBS-related) from use of health resources related to other symptoms and diseases (comorbidity-related). Nurses were instructed to examine the electronic medical records (EMR) to identify relevant information. If information was not available in the EMR, data were assessed by patients recall. Since follow up time varied between six and nine months, only the last six months of the follow up period was used to measure use of health resources. The cost analysis was restricted to direct medical costs for health resource use and did not include sick-leave days, which we considered an indirect cost.

Investigations were assessed only for IBS and included blood samples, stool tests, endoscopic procedures and medical imaging by x-ray, CT, MRI and ultrasound. Costs for consultations and investigations were calculated using National Tariffs for the Public Health Service for year 2001, in Norwegian kroner (NOK, €1 equals NOK 8). Consultations were measured as the number of visits to GPs, specialists, chiropractors, physiotherapists, psychotherapists and alternative health care providers. Hospitalisations were measured as numbers of hospital stays for each patient. Costs for hospitalisations were calculated using Diagnosis Related Groups (DRG) and national tariffs. Medications were measured as total use of drugs by physicians' prescription and drugs purchased over the counter. Medication costs were calculated by The Norwegian Pharmaceutical Product Compendium for year 2001. Costs for on demand medications were calculated using the lowest dose 3 times per week. Alternative health care products and their costs were measured and estimated by patients recall. Sick-leave days were measured directly by retrieving data from the EMR.

Data analysis and statistics

Data were analysed by descriptive statistical methods in SPSS v. 15. Use of health resources and its costs were analysed by univariate, bivariate and multivariate analyses. Non-parametric tests (eg. Wilcoxon, Spearman's rho and Mann Whitney U test) were performed in the cost analyses due to skewed distribution of costs as dependent variables. Variables with p < 0.2 in the bivariate analyses were entered in multivariate lineary regression analyses. Age and gender were kept as sociodemographic variables in the multivariate analysis, regardless of statistical significance in bivariate analysis, and the dependent variables (costs) were log 10 transformed.

Ethics

The study was performed according to the Declaration of Helsinki and approved by the Regional Committee for Medical Research Ethics and the Data Inspectorate, Oslo, Norway.

Characteristics of patients

Use of health resources and associated costs

In the 6 month follow up period, 143 patients (88%) consulted their GP of whom 31 (19%) consulted for IBS-complaints. The mean number of visits was 3.6 (SD 2.9). As for investigations for IBS-complaints in the study period, blood tests were performed in 24 patients (15%), endoscopy in 8 patients (5%) and barium enema in14 patients (9%). Medications were used by 146 patients (89%) of whom 14 (9%) used medications for symptoms of IBS. Sick-leave days related to IBS were significantly lower than related to comorbidity (mean 1.7 days (SD 16) versus 16.3 days (SD 43), p < 0.001).

Factors associated with and predicting costs for health resource use

Main findings

This study found that costs for health care generated by patients with IBS were mainly driven by comorbidity. Our finding that comorbid disorders generated approximately ten times the costs for IBS was consistent for consultations, hospitalisations, medications and use of alternative health care products. A similar pattern was found for sick leave days although we did not compute costs for this dimension of health resource use. We are not aware of other studies demonstrating a similar difference in IBS-related and comorbidity-related use of health resources. Another noteworthy finding was the limited use of health resources related to symptoms of IBS. Less than one third of patients reported consultations, investigations or the use of medications for IBS during the six-month follow-up period. In contrast, patients on average consulted their GP nearly four times and reported 17 sick leave days in the same period of time. As we have published earlier, our sample of patients were characterized by high levels of somatic and psychiatric comorbidity [10]. Importantly, the total direct costs for health care generated by our sample lie within the range of costs reported in the systematic review of costs for IBS in the US and UK [7]. Taken together the findings suggest that from a health economic perspective symptoms of IBS constitute a minor problem but patients with IBS constitute a major problem. Patients with IBS referred to specialists exhibit more severe IBS symptoms than patients handled in general practice [16]. Since these patients also exhibit higher levels of comorbidity than patients in general practice it would be useful to study the impact of comorbidity on health resource use in secondary health care [9]. This was partly investigated in a study of referred patients with severe IBS in the UK [17]. Interestingly, abdominal and psychological symptoms were independently associated with impaired health-related quality but the authors could not find predictors for the observed excess health care costs and loss of productivity. Given the findings in our study one likely predictor would be the somatic comorbidity, which was not included as an explanatory factor in the UK study.

Knowledge about predictors of health care costs in patients with IBS can help target interventions to reduce such costs. Our multivariate analyses revealed age, somatic comorbidity (ie. number of organic diseases and somatic symptoms) and psychiatric comorbidity as main predictors of different domains of health care costs. The individual predictive values of these factors are limited by the significant correlation between somatic and psychological comorbidity in this sample of patients (r² = 46%) [10]. Accordingly, bivariate analyses revealed several somatic and psychiatric comorbidity-variables to be associated with most domains of health care costs. In contrast to what has been reported by Longstreth et al, severity of IBS symptoms did not predict costs for IBS-related health care or total costs for health care in our study [18]. This discrepancy is most likely explained by our use of multivariate analyses to control for the confounding effects of comorbidity on health care costs. Other possible explanations include our crude measurement of IBS severity, the limited sample size and our sampling of patients with long standing symptoms who did not necessarily primarily consult for symptoms of IBS at the time they were included in this study. The latter possible explanation is reflected in the observed low number of investigations performed for IBS and the low median costs for medications for IBS. If we had included only patients with recently onset severe symptoms of IBS the costs for IBS-related health care would likely have been more impressive. The finding that increasing age independently predicted health care costs could be explained by sociodemographic factors or comorbidities (ie. more severe diseases and/or longer stays in hospitals) not captured by our analyses.

The considerable impact of comorbidites on health resource use observed in our study is unlikely to be unique to patients with IBS. A fairly similar pattern would probably emerge if we had included patients with a primary diagnosis of for example chronic low back pain, fibromyalgia or chronic fatigue syndrome instead of IBS as the qualifying diagnosis. Accordingly a systematic review of patients with chronic low back pain found an association between comorbid mood disorders and health care costs [19]. More recently a study of 16 567 patients with chronic low back pain reported that physical and mental health comorbidities were associated with healthcare utilization and costs [20]. The findings highlight the need to maintain a wide focus not restricted to specific diagnoses both in clinical research and practice for patients with a variety of somatic and psychiatric illness.

Strengths and limitations

Among the strengths of this study are the cohort study design, the high follow up rate and the meticulous retrieval of data from the EMR concerning consultations, hospitalisations, medications and sick-leave days. Another strength is that the IBS diagnosis was based on the internationally recognised Rome II-criteria designed for research purposes [11]. The Rome II criteria have now been superseded by the Rome III criteria but only minor changes have been made regarding the diagnosis of IBS. A weakness of our study and possible source of information error is that the use of medications over the counter, alternative health care and presence of organic diseases were assessed by patients recall. Another possible source of information error is that study nurses who were aware of the objectives in the study and responsible for collecting data at follow up could have influenced categorization of health resources as related to IBS or related to comorbidity. Another potential weakness is that we can not determine to what extent patients with IBS in our study used more health resources than patients in general practice without IBS. However, this has been firmly established by others [7].