Somatic health among heroin addicts before and during opioid maintenance treatment: a retrospective cohort study

Background The long-term impact of opioid maintenance treatment (OMT) on morbidity and health care utilization among heroin addicts has been insufficiently studied. The objective of this study was to investigate whether health care utilization due to somatic disease decreased during OMT, and if so, whether the reduction included all kinds of diseases and whether a reduction was related to abstinence from drug use. Methods Cohort study with retrospective registration of somatic disease incidents (health problems, acute or sub-acute, or acute problems related to chronic disease, resulting in a health care contact). Medical record data were collected from hospitals, Outpatients' Departments, emergency wards and from general practitioners (GPs) and prospective data on substance use during OMT were available from 2001 onwards. The observation period was five years before and up to five years during OMT. The cohort consisted of 35 out of 40 patients who received OMT between April 1999 and January 2005 in a Norwegian district town. Statistical significance concerning changes in number of incidents and inpatient and outpatient days during OMT compared with the pre OMT period was calculated according to Wilcoxon signed rank test. Significance concerning pre/during OMT changes in disease incidents by relation to the type of health service contacts, as well as the impact of ongoing substance use during OMT on the volume of contacts, was calculated according to Pearson chi-square and Fisher's exact tests. Results 278 disease incidents were registered. There was a reduction in all incidents by 35% (p = 0.004), in substance-related incidents by 62% (p < 0.001) and in injection-related incidents by 70% (p < 0.001). There was an insignificant reduction in non-fatal overdose incidents by 44% (p = 0.127) and an insignificant increase in non-substance-related incidents by 13% (p = 0.741). Inpatient and outpatient days were reduced by 76% (p = 0.003) and 46% (p = 0.060), respectively. The disease incidents were less often drug-related during OMT (p < 0.001). Patients experienced a reduction in substance-related disease incidents regardless of ongoing substance use, however there was a trend towards greater reductions in those without ongoing abuse. Conclusion Although as few as 35 patients were included, this study demonstrates a significant reduction in health care utilization due to somatic disease incidents during OMT. The reduction was most pronounced for incidents related to substance use and injection. Inpatient and outpatient days were reduced. Most probably these findings reflect somatic health improvement among heroin addicts during OMT.

In spite of considerable morbidity, drug users frequently neglect their health problems, and diseases may remain untreated. Several studies describe that patients with extensive drug use cause problems in hospitals [34] and are difficult to treat in ordinary general practice. Yet some studies based on central health registers show increased health care utilization, in particular due to intoxications/ overdoses, infections related to illicit drug use and injuries [35].
OMT leads to reduced illegal opioid use and injection [36][37][38][39] which probably reduces overdoses and infections. It is also likely that OMT improves nutritional status and general health. Moreover, OMT patients may become more motivated to seek medical help, and OMT may remove or at least reduce tension between patients and health service providers, thus leading to improved health care followup. It is therefore reasonable to assume that over time OMT will reduce morbidity and mortality. Reduction in mortality during OMT has been shown in observational studies [11], but in two recent meta-analyzes of randomised controlled trials (RCTs) mortality reduction could not be documented [36,38]. However, this might be due to problems applying RCT design in studies on OMT effects versus placebo or no treatment [40]. Some studies have shown improved psychological well-being, reduced frequency of self-reported physical health problems, and improved self-perceived HRQOL during OMT [37,41]. Except for this, documentation of OMT-induced health effects is poor [42]. Consequently observational studies with careful design might increase the understanding of OMT effects on health care utilization and also on morbidity.
With a national OMT program implemented during a short time span and a well documented and severe illicit drug problem [39], Norway is well suited for such studies. The number of IDUs in Norway is estimated to 8 200 -12 500 persons out of a population of 4.7 millions in 2005 [43]. The number of OMT patients December 31 th 2005 was 3 614 [44]. Although heroin is usually injected [45] the prevalence of HIV among heroin users is as low as 1-2% [46]. The cumulative number of IDUs infected with HIV from the early 1980s until 2006 is 528 [46]. The anti-HCV antibody prevalence among IDUs is 70-80% [47] and approximately 2/3 of these are Polymerase Chain Reaction (PCR) positive [48]. The death rate among drug users has been estimated to about 1-2% per year [45]. The number of registered overdose deaths has been high with a peak of 405 in 2001 falling to 231 in 2004 [45]. Severe psychiatric co-morbidity ("double-diagnosis") is documented in about 1/3 of IDUs [49].
The national OMT program keeps overall surveillance of patients entering and leaving. Entrance depends on specific criteria [39]: patients should be at least 25 years old, have been dependent on opioids for "several" years and have undergone abstinence-oriented treatment. Treatment is ended if patients fail to pick up the medication over time, and may be involuntary terminated if patients show continued illicit drug use, sell the OMT-medication or illegal drugs, act threatening or violent towards treatment personnel or show severe lack of willingness to fulfil the program regulations. Authorized regional centres cooperate with municipal social service and GPs. Only methadone and buprenorphine are accepted as substitution medication and the average dosage level is high: 114 mg and 18 mg respectively in 2005 [44]. Retention in treatment -which means the proportion of patients who stay in the program over time -is high, compared to most other countries [39].
The systematic collection of information on all participants in the OMT-program as well as computer-based record systems in primary health care and hospitals make Norway suitable for detailed studies of OMT related health effects. The objective of this study was to investigate health care utilization due to somatic disease before versus during OMT in a cohort of OMT-patients. The hypothesis was that such health care contact would decrease during OMT, mainly due to reduced health problems related to illicit drug use and injection. Further, we wanted to investigate whether such possible reduction would occur only in patients who stayed abstinent from illicit drug use or also among those with ongoing abuse.

Methods
Our study compares health care utilization due to somatic disease before versus during OMT using a retrospective cohort design.

Study population
The study was carried out in Gjoevik, a district town with 28 000 inhabitants and with considerable drug problems. OMT was started locally in 1999, according to the guidelines of the national program. However, over the years the GPs have come to play a more important role than is typical for OMT in Norway. Further, very few patients, even among those with ongoing substance use, have had their treatment involuntarily terminated, rather they have received increased follow-up by GPs and social workers. Outcome concerning social rehabilitation and continued substance use during OMT has been close to national average [50]. By the end of 2005, all 40 patients who had started OMT were still in treatment, and 36 consented to participate in the study. Data were not collected for one person, rendering 35 participants (87.5%). Key characteristics of the study population are summarized in Table 1.
The observation period was five years prior to and up to five years during OMT; the mean observation period during OMT was 35 months. Three patients temporarily terminated OMT and then restarted. Disease incidents and health care utilization that occurred while the three patients were between OMT periods (in total five years) were counted as pre-OMT.

Data sources
Thirty-two of the participants were interviewed about disease incidents during OMT and the years prior to OMT. One of the authors (IS, physician) performed all interviews, which took place in a primary care centre or in the patient's home. As no validated questionnaire suitable for collecting this information was available, a list of relevant diagnoses (Table 2) was used, as well as Time-line Followback procedures, in order to facilitate remembering disease incidents and treatment.
Based upon the information obtained in the interviews, records from hospitals, emergency wards and GPs were collected. For the three persons not interviewed, hospital records were collected based upon information in their GPs' records. All requested records concerning inpatient treatment, treatment in Outpatients' Departments (in Norway these are hospital units), emergency wards (in Norway these are part of the primary health care and staffed by GPs), and 75 out of 82 records from solo GPs and GP groups (in Norway most GPs work together in groups of 3-5 sharing a joint record system) were received and scrutinized. Data collection was concluded in June 2005. All data on diagnosis and health care utilization presented in the study originate from these records. Admissions and health care visits mentioned by patients which could not be verified from records were not included. Records from hospitals and GPs which had not been specified by the participants were not requested.

Measures
A "disease incident" was defined as a health problem, acute or sub-acute, resulting in a health care contact. Only somatic incidents were counted, psychiatric illness was only considered if it caused a somatic incident, e.g. an injury due to self harm. A disease incident could be an isolated case, for instance an overdose, an infection or an injury, or a new incident due to an underlying chronic disease, for instance an asthma attack. Even if a disease incident lead to more than one health care visit, e.g. followup visits for a fracture, it was registered as one incident. Routine hospital or GP check ups for chronic diseases or repeated treatment visits for a chronic disease, e.g. hepatitis C, were not included. Disease incidents documented in several records, e.g. from a hospital and a GP, were only counted once. We also counted number of inpatient treatment days (inpatient days) and treatment days in hospitals' Outpatients' Departments (outpatient days) due to the disease incidents we registered.
The full-text records were scrutinized by one of the authors (IS). ICD-10 [51] diagnoses from hospitals and ICPC [52] diagnoses from GPs were registered. Based on record information the disease incidents were categorized  Table  2). The list differentiates between drug related incidents and others. Drug related incidents were categorized as overdoses, injection related incidents and "others", like rhabdomyolysis and related neuro-muscular damage related to non-fatal overdoses, severe withdrawal reactions, inpatient treatment because of severe exhaustion, malnutrition and poor general condition due to drug use, severe sub-acute dental health problems and several others. The incidents not related to drug use were divided into infections, injuries and "others", the latter including all incidents not fitting into the specific categories.
Inter-rater agreement on relation to substance use and diagnostic categories was estimated for 22 disease incidents in six patients by two independent investigators (IS and another physician). Agreement regarding relation to substance use was perfect with a kappa value (κ) of 1. When diagnostic groups were considered, κ was 0.82.
Information about ongoing use of illicit drugs and alcohol during OMT, based on urinary testing and clinical assessment, was gathered from the annual reports made for each OMT patient in Norway since 2001 [39]. For four patients the treatment period was too short or provided insufficient information on substance use; thus rendering such information for 31 patients. The annual report scores overall drug use during the last four weeks on a five-point scale. In our study we simplified this to a dichotomized score for the entire treatment period, differentiating between "problematic" use with severe consequences for psychosocial function versus "abstinence or non-problematic use" without such consequences.

Statistics and ethics
Wilcoxon signed rank test was used to compare changes in rates of episodes before versus during OMT. Pearson chisquare and Fisher's exact test were used to evaluate the changes in the proportion of incidents related to substance use as well as assessment of health improvement versus ongoing use of illegal drugs and alcohol during OMT. Inter-rater agreement was estimated according to Cohen's kappa. All statistical calculations were performed in SPSS 14.0.  Table 1 gives a summary of basic demographic and treatment characteristics for the patient sample. The gender distribution is typical for IDUs and mean age at OMT start is 37 years. Treatment is high dosage, dominantly with methadone as agonist. Nearly all patients are HCV-antibody positive, reflecting the dominant injecting drug use pattern.

Results
Altogether, 278 disease incidents were registered during the observation period, 197 before and 81 during OMT. Table 2 presents findings on health care utilization before and during OMT. The overall reduction in disease incidents was 35% (p = 0,004). There was a reduction of 62% in substance-related incidents (p < 0.001), a 70% reduction in injection related incidents (p < 0.001), and an insignificant reduction of 44 and 59% respectively in overdoses and other substance-related incidents. There was an insignificant increase of 13% in non-substancerelated disease incidents, exclusively in the group "other", while infections and injuries showed minor change. Inpatient and outpatient days due to somatic disease incidents were reduced by 76% (p = 0.003) and 46% (p = 0.060) respectively. Table 3 shows the pre/during OMT shift in the distribution of disease incidents by relation to substance use. Before OMT 62% of the incidents were related to substance use, compared to 36% during OMT (p < 0.001). Table 4 displays health service contacts made during the 278 disease incidents. Forty per cent of all disease incidents during OMT were documented exclusively by GPs, compared with 25% before OMT (p = 0.02). Around 90% of all hospital treatment, before as well as during OMT, took place at the local hospital in Gjoevik. Table 5 shows changes in disease incidents in nine patients with and 22 patients without problematic substance use during OMT. Regarding injection-related inci-dents, there was no difference between the groups, both showing improvement. The reduction in all substancerelated incidents was greater for patients without problematic drug use, but the difference was not statistically significant (p = 0.06). The reduction in the total number of incidents was significantly greater for patients without problematic drug use (p = 0.007).

Discussion
The primary goal of this study was to investigate how OMT influences health service utilization in heroin addicts. The study demonstrates a significant reduction in health care contacts due to somatic disease incidents during the five first years of OMT compared to the five previous years. This is a significant finding. Even if several studies have shown severe morbidity among heroin addicts, and some have found health improvement during maintenance treatment [37,41], we have not been able to find any study systematically comparing somatic morbidity before OMT with morbidity during treatment, based on record information.
The key question regarding the interpretation of our findings is whether the observed reduction in health care utilization can be seen as an indicator of health improvement during OMT compared to the period before. Firstly, how complete was the registration of admissions and health care visits? The study cohort includes nearly all OMT-patients in a defined area; hence selection bias was not a problem. Recall bias could be a problem, greater the further back we go. The patients' information turned out to be chiefly correct, when controlled against the records, regarding type of disease or injury and where treatment had been received, but more imprecise regarding the point of time. Each patient had on average been treated at two GP centres, and approximately 90% of all hospital treatment had taken place at the local hospital which shows a high degree of stability in the relation between treatment services and the patient group in  Gjoevik. The study thus comprises the majority of health service contacts due to somatic disease incidents during the study period.
Secondly, there will be a gap between the volume of disease in any patient sample, and what results in health service contacts, and this is particularly so in a population of IDUs [34]. Due to the structure of the treatment program, contact between patients and the health services was close during OMT, probably leading to increased help-seeking and better medical follow-up and tending to reduce the proportion of disease incidents not resulting in a health service contact. Thirdly, the patients were five years older during the OMT period, leading to increased somatic morbidity. These factors all tend to increase the volume of registered health care contacts during OMT. Hence, when our study still shows a decline in utilization of health services, this most probably is a proxy for an improvement in somatic health status, and moreover, the OMT-induced improvement is probably more extensive than our findings indicate.
Even six out of nine patients with ongoing problematic substance use during OMT experienced a reduction in drug related disease incidents. The most likely explanation is that they stop or at least reduce injecting drugs. However, due to the increase in non-substance-related disease incidents, the majority of patients with problematic substance-use showed an increase in the total number of incidents during OMT. This could be a consequence of changed help-seeking behaviour and better medical follow-up during OMT. If so, this finding reflects improved follow-up and not a true rise in morbidity. On the other hand, it is conceivable that patients with ongoing drug abuse during OMT are more exposed to disease than those without. However, because of the small number of patients, and some uncertainty concerning the differentiation between patients with and without ongoing problematic substance abuse, these results and their significance should be interpreted with caution.
OMT is often evaluated primarily by its effect on social rehabilitation and continued substance use. According to our findings, this is not sufficient. Drug related disease incidents were reduced even among patients with ongoing abuse, though to a lesser degree. This might question involuntary termination of OMT in patients who still take illegal drugs.
The study has some weaknesses. The cohort is small and limited to one local community. The research instruments, especially the diagnosis categorisation system, have not been validated by other researchers. In addition, it is not always obvious whether a disease incident is related to substance use or not. However, the high level of inter-rater agreement on whether incidents were substance related or not (κ = 1) implies that this is possible to differentiate.
In spite of these weaknesses, our study of a small patient cohort showed a significant reduction in health care contacts caused by somatic disease incidents during OMT compared to the five years prior to treatment. These findings ought to be further investigated in an enlarged study. This could bring information about factors influencing somatic health status changes during OMT, like psychiatric co-morbidity or living in a larger city. The design chosen appears suitable for investigating OMT-related changes in somatic morbidity among heroin addicts in Norway.

Conclusion
Even with as few as 35 patients included, this study demonstrates a significant decrease in health care contacts due to somatic disease incidents during OMT compared to the five years before entering treatment. This reduction was most striking for incidents related to substance use, and drug injection in particular. Inpatient treatment days and treatment days in hospitals' Outpatients' Departments were reduced during OMT. These findings most probably reflect an improvement in somatic health status for drug abusers undergoing OMT compared to the period before entering treatment.