Prevalence of diagnosis and direct treatment costs of back disorders in 644,773 children and youths in Germany

Back pain is a cardinal symptom for various diseases of the back or spine, that is, for "dorsopathies" or back disorders. It is also acknowledged to be a common condition in childhood and youth [1‐4], with self-reported one-year prevalence rates of up to 83% [1], and might - even at this early age - be associated with difficulties and restrictions in everyday life [5, 6]. Although various authors report high prevalence rates of pain and especially back pain in children, adolescents, and youths in Germany [e. g. [7‐9]] with three-month prevalence rates between 30.2 - 49.0%, all of these reports focus on self-reported back pain. And while self-reported complaints of back pain seem to be relatively common in these age groups, the number of children seeking medical care because of back pain seems to be rather unclear and might be influenced by age or regional aspects and furthermore it might have increased over the years. Olsen et al. [10] for example reported in 1992 that 7% of the then examined U.S. adolescents with back pain sought medical attention. In 1999, 23% of Belgian school children (aged nine) with low back pain had sought medical help from a doctor or a physiotherapist [11], and in 2005, it was reported that 32% of Tunesian children and adolescents required medical help because of back pain [12]. With regard to German children, Roth-Isigkeit et al. [8] reported that 53% of German children with back pain visited a doctor. These immense differences in self-reported health care utilization when experiencing back pain led to the conclusion, that besides the analysis of self-reports on back pain, it might be worthwhile to consider back pain from a more population based or public health point of view and evaluate prevalence rates and costs of diagnosis in children, youths and adolescents in existing data sets of health insurances. As, to our knowledge, little is known about the prevalence rates of diagnosed back disorders in general in childhood and/or youth, we extended our research approach, notwithstanding the possible association of these disorders with back pain. Therefore this article addresses the question of prevalence rates of diagnosis, as well as direct costs of back disorders in German children and youths. An extensive, population based dataset focussing on all persons insured in statutory health insurance funds (approximately 90% of the German population) was used to answer this question.

In the scientific literature, there seems to be large differences in health care utilization of children and youths with back pain [8, 10‐12]. In particular, German children with back pain were reported to seek out medical care more often (in 53% of the cases) than children and adolescents elsewhere (7-32%). Apart from back disorders which are associated with the clinical feature back pain there are other back disorders, not necessarily associated with back pain, of which we know only little with regard to prevalence rates in children and youths. This situation gave us reason to believe, that it might be necessary to reappraise the number of diagnosis of back disorders for German children and the accordingly incurred costs from a public health point of view.

In the 3% random sample of all Germans insured with statutory health insurers, which was available for scientific analysis, we found that altogether 7.4% of the insured children between 0-14 years and 25.9% of the insured youths between 15-24 years were diagnosed with a back disorder. Assuming that all of those patients experienced back pain, and considering the reported health care utilization of 53% [8], we could estimate a one year prevalence rate of back pain of about 15% in children and a one year prevalence rate of about 52% in youths. These prevalence rates seem to be very high and of course it has to be taken into account that not all back disorders are associated with back pain, e. g. scoliosis (M41). As other authors (e. g. [9]) reported a three month prevalence rate for back pain of 38.6% (in German children and adolescents aged 10 - 18 years), and one year prevalence rates for lower back pain between 7-63% in adolescents [1], our results at least seem to cover a large proportion of children and youths seeking out a doctor because of back pain.

In 2007, Kamtsiuris et al. [14] questioned German children and adolescents (and their parents) and reported that scoliosis was diagnosed in 5.2% of all examined 0-17 years olds in Germany. Girls were more often affected (6.0%) than boys (4.4%), whereby the gender difference was especially pronounced in the age groups between 11-13 years and 14-17 years. The authors themselves discuss that the reliability of the diagnosis, reported by the "patients" themselves, is to be interpreted with caution, as it cannot be differed between a suspected case and a manifest case. In contrast to these results, in our study we found slight age and gender-effects within the ICD-10 diagnosis M41 for scoliosis. This might be on the one hand due to methodological issues, as we had to use larger age groups which might have disguised slight age- and/or gender effects. On the other hand, the prevalence rates of diagnosed scoliosis seem to be lower than reported by Kamtsiuris et al. [14] which might be due to the methodological differences or due to the fact that we report the one-year prevalence rates, while Kamtsiuris et al. reported lifetime prevalence [14]. Kim et al. [15] examined scoliosis in their review in more detail and reported the prevalence of scoliosis (defined as case of scoliosis with a scoliosis curve of 10° or more) in the age group of over 10 year olds to range between 0.5-3.0%. In addition, they stated that the gender distribution is fairly even when the scoliosis-curves are small; however, there is a clear female predominance as the curve magnitude increases, with some studies quoting a ratio of 1:8 (male to female) [15]. The differences between these reports and our results are probably due to the problems which occur when using ICD-10 codes as indicator of disease, which does not allow for differentiation between scoliosis with smaller angles (<10°) vs. scoliosis with larger angles (>20°). Another point which should be kept in mind when interpreting our results is that in Germany there are standardized physical examinations for children (which the statutory health insurer pays for) and scoliosis might be identified and be coded as an incidental finding during one of these examinations.

Quite a large amount of scientific literature considers back pain in children and adolescents [e. g. [1, 2, 5, 10, 16‐19]]. The one year prevalence rates in these other reports range from 3.5% - 83%, depending on localization (low back > neck > thoracic spine), age (older > younger), and gender (female > male), whereby Duggleby [20] and Olson [10] reported that only 8% and 7% of the children/youths respectively with back problems see a doctor. As described above, more recent German studies report a three month prevalence of back pain between 30.2 - 49% [7‐9] and according visits to the doctor in 53% of the back pain cases [8]. Though these numbers are quite impressive, the reports tell us nothing about the diagnosis the consulted doctors made. With our analysis we can close this gap of knowledge at least in part and state, that here, altogether 1.7% of children younger than 15 years were diagnosed with back pain as compared to 12.7% of youths between 15-24 years. This noticeable increase of prevalence with age goes along with findings of other authors who reported an increasing self-reported prevalence of back pain with age in children and adolescents [e.g. [1]]. The age dependent increase might at least partially be due to the growth period in puberty and problems that arise therein. Apart from the here presented study, other literature on diagnosis of back disorders in childhood and youth is scarce. But keeping in mind the numbers of affected children and youths in Germany alone, our results uncover a need for secondary and tertiary prevention strategies for musculoskeletal disorders, especially back pain, even at an early age.

Apart from that, we found a small number of cases with the medical conditions summarized under M42 to M51 (M42: "spinal osteochondrosis"; M43: "other deforming dorsopathies"; M45: "ankylosing spondylitis"; M46: "other inflammatory spondylopathies"; M47: "spondylosis"; M48: "other spondylopathies"; M49: "spondylopathies in diseases classified elsewhere"; M50: "cervical disc disorders"; M51: "other intervertebral disc disorders"). This result is not surprising as degenerative diseases are hidden behind these ICD-10 codes and are not expected in large numbers in childhood and youth. The fact that we actually found cases with degenerative disorders, even at an early age, might also be due to erroneous ICD-10 coding.

To our knowledge this is the first examination that focuses on the prevalence rates and costs of back disorders in childhood and youth, which were verified by a doctor's diagnosis by means of evaluating a huge population-based data set. The use of this large data set is also our study's biggest strength [21]. Nevertheless, using the International Classification of Diagnosis (ICD-10 code) implies some methodological problems. In general, the ICD-10 codes exist of up to five-digit codes. As only three-digit codes were systematically controlled and used for statistical analysis in this large data set, we were confined to a relatively general analytical approach, leaving no room for more detailed analysis of specific diseases. However, the use of terminal codes might not have provided more specific information, as many authors reported the reliability of coding to decrease with the increasing levels of the code used (e.g. [22]).

Though the coding of medical entities with classifications is a hot topic in Germany, as the codes are used for reimbursement, the reliability of diagnosis coding based on the ICD-10 is still not clear [14]. A study of Stausberg et al. [22] found kappa-values between different coding groups on a three-digit ICD-10 level (full set of diagnosis) to range between 0.34 -0.58 (i.e. fair to moderate agreement [23]) and therefore the authors concluded, that the results concerning the reliability of diagnosis coding with ICD-10 are to be interpreted with caution. They also suggested that coding in daily practice might be worse than found in their study under "standardized" conditions. Another study reported ICD-coding to be reliable on chapter level only and found relevant coding uncertainties at three- and four-digit coding level, too (kappa-values of approximately 0.4 and 0.2 for three-digit and four-digit coding, respectively) [24]. Until now, there have been few validation studies for routine data of German statutory health insurances [25]. That is why we can only estimate the true underlying number of children with each ICD 10 diagnosis as well as the true underlying number of children with pain or activity limitations. A current study might at least allow a glimpse at the validity of coded diagnosis of the used data set with regard to the coding of General Practitioners in Berlin [26]. In this study, the authors reported a correctness of musculoskeletal disorder coding of 61% (one-digit code "M") and a correctness of M54 coding ("dorsalgia") of 71%. But while the reliability and validity of ICD-10 coding on a three-digit basis might still be under scrutiny, it is a fact that - especially with regard to population based studies in Germany - ICD-10 coding is often the only available information.

The existing data set allowed for evaluation of the mean expenditures for outpatient treatments. When looking at the 3% random sample, we found direct treatment costs of approximately € 1.9 million which shows, that the treatment costs per case are - on average - rather low. Age-effects were mainly due to increased prevalence rates and not to increases in mean costs per case. For extrapolating the costs caused by all insurants of statutory health insurers, we only considered costs accompanied by single diagnosis. Therefore, the relevant accounting case numbers were considerably smaller and do not necessarily reflect the cost profile of the "usual" case. Apart from that, we assumed that every prevalent case caused only the average direct treatment cost of outpatient treatment once per year. Inpatient treatments or outpatient treatments more than once were not included in this estimation. Therefore this analysis is only a minimum estimation of the incurring costs and gives only a first insight into the complex structure of medical accounting cases associated with back disorders. More extensive analyses are needed to provide more details. Nevertheless, the present analysis adds to the existing knowledge by the mere fact that it is based on real costs and money spent on musculoskeletal diseases in these age groups and does not rely on estimations only.

Until now, information on diagnosis of back disorders in children and youths was scarce. In our study, a large population-based data set of German insurants was used to elucidate this situation. Instead of focussing on self-reported disorder or disability, this approach allows for the evaluation of medical experts' opinion and therefore gives a more objective or public health oriented insight in the topics "back pain" and "back disorders". Some of our results supported the study results of other authors, other results, like the lacking gender differences in scoliosis, were rather surprising and might be due to methodological problems which arise from using ICD-10 coding. Apart from that, the data set allowed for analysis of direct treatment costs, and disclosed that the age-dependent increasing health care costs, are due to increasing prevalence rates and not due to increasing mean costs because of more problematic cases. Unfortunately, the structure of the here examined data set, as well as other available population-based data sets is not yet cut out for addressing all open questions with regard to back disorders of children and adolescents. Standardized random validity checks of these population based data sets should be mandatory in order to use them for scientific analyses. Furthermore it might be interesting to give researchers possibilities to contribute new ideas of data linkage. Finally, an overall similar construction of large population based data sets would be very useful to conduct comparisons or to link different data sets holding different information. A closer interaction between researchers and stakeholders would be desirable, preferably on a European or even worldwide level. Nevertheless the results presented in this paper, provide insight into the prevalence rates of back disorders as well as minimum cost estimations for these disorders. These results can be used for comparison with future analyses in order to detect changes.