Methods Using a cross-sectional design, cost data of 3,124 participants aged 57–84 years in the 8-year-follow-up of the ESTHER cohort study were analyzed. Health care utilization in a 3-month period was assessed retrospectively through an interview conducted by trained study physicians at respondents’ homes.

Rar - Fsipanel Download - Free Ex4 To Mq4 Decompiler Software As A Service - Jigsaw Puzzle Platinum Edition Torrents. - Besm 3Rd Edition Monsters University - Bush Dab Radio Ne 3160 Manual Arts - Old 97S Too Far To Care Rarest.

Unit costs were applied to calculate health care costs from the societal perspective. Socio-demographic and health-related variables were categorized as predisposing, enabling, or need factors as defined by the Andersen model. Multimorbidity was measured by the Cumulative Illness Rating Scale for Geriatrics (CIRS-G). Mental health status was measured by the SF-12 mental component summary (MCS) score. Sector-specific costs were analyzed by means of multiple Tobit regression models. Background Due to demographic change, the proportion of elderly people in developed countries will increase substantially in the next decades [].

Germany is one of the countries most strongly affected by demographic change with the proportion of people aged ≥ 65 years expected to rise by about 50% until 2030 []. Due to the progressive increase in the proportion of elderly people, health care systems are faced with serious organizational and financial challenges [-]. For a better understanding of the future demand for health care services and health care costs, it is necessary to understand the specific mechanisms that determine the utilization of health care in the elderly.

As health care utilization is influenced by multiple individual and contextual factors, a reasonable starting point for analyzing health care utilization and costs is to define a theoretical framework. There are several explanatory frameworks identifying predictors of health care utilization []. One of the most comprehensive and widely used frameworks is the behavioral model developed by R. Andersen and J.F. Newman in 1973 [].

Therein the authors present a causal ordering of health care utilization within an integrated framework. In the model which has been discussed and continuously refined over the years [-], it is assumed that individuals’ use of services is a function of their predisposition to use services (predisposing factors), factors that support or impede use (enabling factors), as well as their need for health care (illness level). Predisposing variables pertain to socio-demographic (e.g. Age, sex, education, marital status) and belief characteristics (e.g. Values concerning health and illnesses measurable in consequence such as smoking behavior, alcohol consumption, or body mass index) while enabling factors are those that support or impede health care service use (e.g. Income, type of health insurance).

According to Andersen and Newman [], patients’ illness level (representing the need factor) is considered the major determinant of health care utilization. In the elderly, the illness level is often shaped by multimorbidity (MM), defined as the co-occurrence of two or more chronic conditions in one person without reference to an index disease []. Among the population aged 65+ the prevalence of MM has been reported to exceed 65% [-]. Approaches to measure MM have quantified the number of affected clinically relevant physiological systems weighted by severity [].

Thus, in the context of the Andersen model, patients’ illness level may be described by a measure of MM rather than individual conditions. Linking MM with health care service use can be a useful strategy for health services research in general populations where the focus is on care and costs of the patient as a whole rather than on the treatment of particular diseases [,]. A recent review of the literature asserted the positive association between MM and health care utilization or costs in the elderly and pointed out that in studies conducting multivariate analyses MM typically had a much stronger impact on health care utilization than variables operationalizing predisposing and enabling factors [].

However, in many of these analyses, predisposing and enabling factors such as age, gender, living arrangement, and health insurance status were still significantly associated with health service use independent of MM []. Another recent review of studies that specifically applied the Andersen model to analyse health service use in various populations pointed out inconsistencies in the strength and direction of associations which seemed to be strongly influenced by the study context and the characteristics of the study population []. Yet, of 16 studies included in this review, none focussed on the elderly population and only one was conducted in Germany []. The purpose of our study was therefore to analyze the association of health care costs with predisposing, enabling and need factors as defined by Andersen’s behavioral model of health care utilization in a random sample of the elderly population in Germany. As the German health care system aims at providing universal access to comprehensive health care services, we were interested in the relative impact of predisposing, enabling, and need factors in Germany as compared to findings from the international literature. We hypothesize that the need factor is much stronger associated with health care costs in the elderly in Germany than predisposing and enabling factors.

Sample This cross-sectional analysis was performed based on 8-year follow-up data from the ESTHER study, a large population-based prospective cohort study conducted in the German federal state of Saarland. The ethics committees of the Medical Faculty, University of Heidelberg, and of the Medical Association of Saarland approved the study which was conducted in accordance with the declaration of Helsinki. Written informed consent was obtained from each participant. Detailed information about the study design and the participants of ESTHER has been reported elsewhere [,]. Briefly, baseline recruitment was conducted between July 2000 and December 2002 and included 9,949 participants aged 50 to 74 years. Standardized questionnaires on socio-demographic, medical, and lifestyle factors provided in a postal survey were completed at the baseline assessment and three follow-ups (2, 5, and 8 years).

Until the 8-year follow-up 499 individuals deceased and 680 withdrew informed consent, 505 of them due to health reasons. 6,063 individuals participated in the 8-year follow-up (73.4% response rate) completing the standardized questionnaires. For 5,056 of these participants, additional information was collected with a questionnaire provided to their GPs. In addition, all 6,063 participants were asked to take part in an 8-hour geriatric assessment conducted at their homes by trained study physicians, which also included the assessment of health service use to be used for the present analysis (see below). 3,124 individuals participated in this geriatric assessment between July 2008 and December 2010.

The analysis presented here is based on these 3,124 individuals. Enabling factors Characterized by Andersen as a personal enabling factor, the participants’ income is reported as the square root equivalence scale (SRES), which divides the household income by the square root of the household size []. Since the income was assessed per month, the amount of the SRES was multiplied by 3 to obtain the income for a 3-month period (see below). Another personal enabling factor is the participants’ health insurance. In Germany health insurance is mandatory for the entire population.

Approximately 88% of the German population – in particular employees below a certain income ceiling and their family members – are insured by the statutory health insurance (SHI) and keep this insurance after retiring. Self-employed persons and employees above the income ceiling can opt for private health insurance (PHI) and usually stay privately insured after retiring. Regardless of the type of health insurance, all beneficiaries have access to outpatient physician and non-physician services, hospital care, rehabilitation, dental care, prescription drugs, medical supplies and long-term nursing care.

For those insured by the SHI there are co-payments for most services but cost-sharing is limited to 2 per cent of household income per year or even 1 per cent for the chronically ill. Accordingly, health insurance was divided into the categories 'statutory’ and 'private’. To assess the degree of social isolation and thereby a community enabling factor, the LSNS-6 [], a 6-item short form of the Lubben Social Network Scale [] was administered to the respondents. The LSNS was specifically developed for older adult populations. Its single item categories are defined as “none”, ”one”, “two”, “three or four”, “five to eight”, “nine or more” persons from family or friends to whom respondents had contact during a one month period. Equally weighting the sum of the six items, the LSNS-6’s total score ranges from 0–30 while the subscale for family and friends ranges from 0–15. Need To assess illness level as an indicator of participants’ objective need of health service use, the Cumulative Illness Rating Scale for Geriatrics (CIRS-G) [] was used.

The CIRS-G is a modified version of the Cumulative Illness Rating Scale [], which is a well-established measure of MM in the elderly. On the CIRS-G, 13 categories referring to clinically relevant physiological systems and 1 category referring to psychiatric illness are rated on a five-point severity scale ranging from 0 (no problem) to 4 (extremely severe). The 13 physiological systems are 1. Hematopoetic, 4. Respiratory, 5. Eyes, ears, nose, throat and larynx, 6.

Upper gastrointestinal tract, 7. Lower gastrointestinal tract, 8. Genitourinary, 11. Musculoskeletal/integument, 12. Neurological, 13. Endocrine/metabolic and breast. Assuming that the impact of the categories is additive, a total score from the sum of each of the 14 single categories can be constructed.

This total score theoretically ranges from 0 to 56 and simultaneously accounts for the number of diseases and their severity []. The CIRS-G questionnaire was completed by respondents’ GPs.

Furthermore, since mental health is only reflected to a small extent in the CIRS-G, subjective mental health status was measured based on the SF-12-questionnaire, a widely used generic questionnaire that does not focus on specific disease groups. The SF-12 is a downsized version of the 36 short form health survey (SF-36), in which a subset of 12 items/questions (of the original 36 contained within the SF-36) is used to derive one summary score each for physical health (PCS score) and for mental health (MCS score) []. By covering the same dimensions as the SF-36, i.e. Physical functioning (2 questions), role-physical functioning (2 questions), bodily pain (1 question), general health (1 question), vitality (1 question), social functioning (1 question), role-emotional functioning (2 questions), and mental health (2 questions), while using only one-third of the items, the SF-12 is able to produce the two summary scores originally developed for the SF-36 with remarkable accuracy but far less respondent burden []. The SF-12 allows to calculate a mental component summary score (MCS) for mental health. The score ranges from 0 to 100 and is standardized to population norms (based on a US norm-sample), with the mean score set at 50 (SD = 10); lower scores indicate worse, and higher scores better mental health. The SF-12 has good psychometric properties [] and measures subjective Health-Related Quality of Life (HRQOL).

Health care use Based on a questionnaire on health service utilization developed by our working group on the specific demands of the German health care system and used in previous studies [,], a short health economic questionnaire, especially suited for the application in large epidemiologic studies, was developed within the project. The questionnaire is available from the authors upon request. It covers in-patient care, out-patient physician services, out-patient non-physician services (physical or occupational therapy), medical supplies and dental prostheses, formal and informal nursing care as well as out-of-pocket expenses for the corresponding categories (Table ). Assessment was retrospective for a period of 3 months for all resources and services.

In order to minimize recall bias, the questionnaire contains lists of common health services and goods used in old age. Pharmaceuticals were recorded during home visits by the study doctors by means of a barcode reader when respondents had packages available and by hand otherwise. Missing drug codes were searched using available information on trade name and pharmaceutical form. The health economic questionnaire was administered in personal interviews conducted by the study physicians during the geriatric assessment at respondents’ homes. Health care costs By recording all used resources and services, regardless of whether they were covered by health or nursing care insurance or paid for out-of-pocket, a societal perspective was adopted in this analysis. The cost categories analyzed in this study are direct costs of illness arising from the use of resources. Costs were calculated from resource use as recorded in the questionnaire by means of unit costs.

Resource categories and sources of unit costs are listed in Table. Pharmaceuticals were monetarily valued using German PZN-codes as recorded in the home visit in conjunction with the MMI Pharmindex database []. Medication taken occasionally was valued by means of the pharmacy retail price of one package per 3 months (using the package size as recorded).

For continuous medication, costs per unit of the drug were derived from the recorded package size and the corresponding pharmacy retail price, and 3-month costs were obtained by multiplication of the unit cost with the total dose for the 3-month period. Informal care was valued using the replacement cost approach, i.e.

It was assumed that the same amount of care by professional nursing services would have had to be paid for in the absence of an informal caregiver. Accordingly, hours of informal care were valued using the same hourly wage rate as for professional home care. Methods for the valuation of informal care are discussed by van den Berg et al. Costs were calculated in € at 2009 price levels.

Unit costs that were unavailable at year 2009 values were inflated or deflated to year 2009 price levels by means of the consumer price index []. For statistical analysis, we categorized cost data as follows: 1) Costs of inpatient care comprising inpatient treatment in general hospitals, specialized psychiatric and neurological hospitals or rehabilitation hospitals; 2) costs of outpatient care comprising outpatient physician treatment, other outpatient treatment, medical supplies and dental prostheses; 3) costs of medication comprising pharmaceuticals; 4) costs of nursing care comprising nursing home care, professional community nursing care and informal care. Missing values With the exception of items of the health economic questionnaire, missing values were imputed by means of multiple imputation by chained equations using the program ice [] in STATA Release 12 []. The last column in Table gives an overview of the percentages of missing values in the single variables.

For the imputation procedure a cycle length of 200 was chosen. A graphical analysis showed that this was sufficient for convergence of the imputations. Following the suggestions of van Buuren [,], we calculated 100 imputation steps as the recommendation of only 5 imputations by Rubin [] seemed to be too small for the task. Beyond the variables contained in Table, the following variables were entered into the multiple imputation model: SF-12 physical component score, Barthel Index, Mini Mental State Examination score, smoking status, alcohol intake and hand strength.

The single sectors of health care costs were part of the imputation model but were not imputed themselves. For the imputation the level of measurement of the particular variables was considered. Imputed values were restricted to the theoretical range of the original variable. Socio-demographic and health related sample characteristics (grouped by predisposing characteristics, enabling resources and need) Missing values in items of the health economic questionnaire and dosage of medication were not part of the multiple imputation model since these variables were too numerous and too varied to be imputed meaningfully. Therefore costs of medication with missing values for dosage were calculated using a conservative rule, whereby the pharmacy retail price of one package of the drug per 3 months was applied.

Missing values in the resource use questionnaire were set at zeros, resulting in a conservative estimation of the health care service use and costs. Statistical analysis “To assess the associations of covariates with health care costs, multiple Tobit regression models with marginal effects for the unconditional expected value E[y] of the dependent variable were estimated. Tobit regression models were used because resource use and cost data as used in our study often fall under the category of so-called corner solution outcomes. That is, the dependent variable “y takes on the value zero with positive probability but is a continuous random variable over strictly positive values” []. In contrast to censored data where the “real” values for the zeros are unknown, data observability is no issue in corner solution applications. Following Woolridge’s [] suggestions, we decided to present the marginal effects for the unconditional expected value E[y] of the dependent variable. These marginal effects are the most useful and informative measure since they provide information which is valid for the whole study sample.

For the estimation of the pooled results, necessary because of the multiple imputed data, the software MIM was used []. Pseudo-R 2s and Chi 2s were estimated with reference to [].

The level of significance was set at α = 0.05. All statistical analyses were performed using STATA Release 12 []. Sociodemographic characteristics and missing values The sample consisted of 3,124 respondents of whom 1,481 (47.4%) were male. The mean age was 69.6 years. Most of the respondents were married (71.8%), had less than 10 years of schooling (66.2%) and were covered by statutory health insurance (92.2%). The mean BMI was 28.7 and the mean SRES income for 3 months was 4,299 €. The mean CIRS-G score was 6.9 and the mean SF-12 MCS score was 47.8.

Compared to women, men were slightly older, more often married, had more years of schooling, had a slightly higher BMI, had a higher income, were privately insured more frequently and had a higher level of multimorbidity (CIRS-G score) as well as a slightly better mental health status (SF-12 MCS score) (all p. Health care utilization 98% of the respondents consumed at least one health care service or good during the 3 months preceding the interview (Table ). The highest rates of service utilization appeared for outpatient physician services (95%) and pharmaceuticals (85%). Inpatient care was used by 9%. Non-physician providers, medical supplies and dental prostheses were used by 20% and 24%, respectively. The utilization of nursing care was rather low with 0.9% for formal and 1.7% for informal care.

The utilization rates for services and goods tended to be slightly higher for women. Health care costs The mean total costs per respondent for the 3-month period were 889 € (Table ).

With 393 € per respondent, inpatient care was the sector with the highest mean costs followed by outpatient physician services (174 €) and pharmaceuticals (171 €). Mean costs of medical supplies and dental prostheses (69 €), informal nursing care (44 €) and non-physician providers (29 €) were comparatively low. With 8 € per respondent, mean costs of formal nursing care were lowest. Mean costs of non-physician providers were significantly higher for females (72 € vs. F.w. Bell Model 4048 Manual.

65 €), while mean costs of pharmaceuticals were significantly higher for males (183 € vs. Health care costs of users Looking at mean health care costs for the 3-month period of only those respondents who used respectively services or goods, inpatient care (4,518 €), followed by informal nursing care (2,515 €) and formal nursing care (909 €), were the three most costly sectors (Table ). Mean costs per user in all other sectors were comparatively low with 286 € for medical supplies and dental prostheses, 200 € for pharmaceuticals, 183 € for outpatient physician service and 145 € for non-physician providers. For pharmaceuticals and informal nursing care, the mean costs for users of health care services were significantly higher in males than in females.

Regression analyses The results of the regression analyses (Table ) revealed that the CIRS-G score and the SF-12 MCS score, representing the need factor in the Andersen model, were consistently associated with total, inpatient, outpatient and nursing costs: A one point increase in the CIRS-G score was associated with an increase of 18 € in 3-month inpatient costs, 17 € in outpatient costs and 3 € in nursing costs. This aggregates to an increase of the total costs of 41 € per score point of the CIRS-G. The SF-12 MCS score was inversely related to all cost sectors, resulting in a total cost decrease of 14 € per MCS score point. Multiple Tobit regression analyses with mean costs in 3-month period in € (year 2008 values) as dependent variable for total costs and by health care sector Of the socio-demographic variables representing predisposing factors in the Andersen model, only respondents’ age was significantly positively associated with inpatient costs (10 € per year of age), outpatient costs (3 € per year of age), nursing costs (3 €) and total costs (13 €). The BMI was associated significantly with outpatient costs, with a 6 € increase per kg/m 2 and total costs with a 11 € increase per kg/m 2.

12 or more years of schooling significantly decreased inpatient costs by an amount of 161 €. Neither income nor health insurance status or Lubbens social network scale, representing enabling factors of the Andersen model, were associated with costs in any sector. Denoted by the constants of the four regression models, the average 3-month cost per respondent, controlled for all covariates, was 843 € for total costs; 349 € for inpatient costs; 447 € for outpatient costs and 47 € for nursing costs. Discussion In our study we analyzed health care costs of the elderly population in Germany. In order to organize and categorize the multiple factors which may influence health care utilization, we applied the theoretical framework developed by Andersen and Newman which distinguishes predisposing, enabling and need factors. The main finding of our study is that the need factor of the Andersen model, operationalized through a measure of MM (CIRS-G) and complemented by a measure of mental health status (SF-12 MSC), was the dominant and most consistent predictor of health care costs. In our study a one point increase of the CIRS-G was associated with an increase in total costs of 41 € per 3 months.

The finding that higher levels of MM lead to higher health care costs is in line with most comparable studies conducted in elderly populations and emphasizes the relevance of elderly’s need of health care. In a systematic review of the international literature, Lehnert et al.

[] found ample evidence of a positive association between MM and health care costs. Similar to our study, hospital stays [,], physician visits [,] and pharmaceuticals [-] were reported to elevate health care costs with each additional chronic condition. Over and above the need factor, predisposing and enabling factors explained only little of the variance of costs.

Of the predisposing factors, only age, the BMI and 12 or more years of schooling were found to be significant predictors of costs. The association found between age and costs seems to contradict the finding of a review conducted by de Boer [], which found no definite influence of age on health care utilization in her review. However, de Boer’s findings apply to populations of chronically ill people, whereas our findings are based on a representative random sample of the elderly German general population.

Furthermore, in a recent review the direction of the association between age and the utilization of health care services was found to be dependent on specific characteristics of the study population []. Of all other variables characterized as predisposing factors in the Andersen model, only the BMI demonstrated a cost-increasing effect of potentially inadequate health behavior.

Yet, this only holds for outpatient and total costs. We could not detect a significant cost-increasing effect of female gender [,] or high education [,] as reported for health service use by other studies that applied the Andersen model. Instead, we found a cost-reducing effect of 12 or more years of schooling on inpatient costs with an amount of 161 €.

Nor did we find a significant association of marital status and health care costs; yet, associations of marital status and health service use have been reported in different directions by other studies applying the Andersen model [,]. Neither respondents’ income nor their health insurance status, representing enabling factors of the Andersen model, was significantly associated with health care costs.

By contrast, studies conducted in other countries, in particular the USA [,,], have repeatedly reported positive associations of insurance, income and costs. Since in our study the difference between statutory and private health insurance regarding total costs was rather high (227 €), the lack of significance of this effect might be caused by the relatively small number of privately insured respondents (7.8%). Yet, the lack of association of income and costs may be due to the specifics of the German health care system which is characterized by relatively low financial barriers to the utilization of comprehensive health care services (in particular by limiting co-payments to a maximum of 2% of income), despite undoubtedly existent income inequalities. Possible reasons why predisposing and enabling factors explained only little of the variance in costs may be insufficient operationalization and variable selection for the predisposing and enabling factors.

While there is no fixed set of predisposing and enabling variables defined for the Andersen model, our choice of variables was strongly oriented by Andersen’s suggestions, thereby trying to keep the regression models preferably parsimonious. However, as pointed out by Babitsch [], an unambiguous assignment of a variable to one single factor is not always possible. In fact, we used variables similar to those used by other studies that applied Andersen’s framework: a recent review of studies that applied the Andersen model found the most frequently used variables for the predisposing factor to be age, marital status, gender, education and ethnicity []. We included all of them except for ethnicity, which might not be as relevant in Germany as, for example, in. The USA [,], but still may have explained some additional variance in costs. For the predisposing factor, the most frequently used variables reported by the mentioned review were income, health insurance and having a usual family doctor.

We did not include the latter variable, as the study sample was recruited via their GPs. Yet we followed the suggestion to expand Andersen’s original model by the inclusion of the social network, following Pescosolido’s Network-Episode Model [] “larger, more supportive networks decrease the use of patterns of care”. Although not significant, the effect of the Lubben Social Network Scale characterizes the Social Network as a potential cost driver in our study with 5 € per scale point. This might be due to the fact that compared to network analysis as practiced by Pescosolido, the Lubben scale is a rather basic construct and therefore potentially unable to depict the expected associations. Furthermore, the impact of the Social Network may depend on cultural aspects. Mean total costs per respondent for a 3-month period were 889 € which - extrapolated to a 12 month period – corresponds to 3556 € per year. This figure is similar to annual total costs of 3315 € and 3730 €, respectively, recently reported by Nagl et al.

[] and Heinrich et al. [] for similar samples of the elderly population in Germany. Similar to a study conducted in Germany by Nagl et al. [], we found inpatient care, outpatient care and pharmaceuticals to be the three most costly health care sectors. However, amounting to 25% in the study by Nagl et al., their proportion of inpatient care was 19% lower than in our study. This might have been caused by the different time period of 12 months used by Nagl et al. The 3-month period in our study.

Since the utilization of inpatient care is a much rarer event than the use of outpatient care and pharmaceuticals, this in conjunction with the implementation of face to face interviews by trained study physicians in our study could have led to a reduction of recall bias and therefore more valid representation of inpatient costs as opposed to the study by Nagl et al., which used telephone interviews. Limitations Our study is based on a large and nearly representative sample of the German elderly population [,]. Yet the number of users of nursing care in our sample was very low. This is likely due to the inability or unwillingness of nursing care recipients to complete a 2-hour home interview.

A consequence of this selection process would be the underrepresentation of those with the highest level of MM and therefore an underestimation of the health care costs. This might have also influenced the prediction of the health care costs in the regression analysis. We collected comprehensive data on health service utilization from a societal perspective. Yet the period of 3 months for which data was collected was rather short, possibly increasing the variance of calculated health care costs, which may be a further reason for the variance explained by the models being only small.

On the other hand, this short period likely may have enhanced the accuracy of collected information about health care utilization because of less memory bias in participants’ responses. We used detailed information on morbidity from which we calculated a well-established and - due to the assessment by GPs - objective measure of MM.

How to best measure MM in this context is uncertain and needs further investigation. Research comparing the predictive ability of various MM measures on different health care related outcomes has produced inconclusive results [-], suggesting that no single measure of MM will completely capture the differences in the study subjects’ underlying illness level. The conclusions from the results of the regression models hold only under the assumption that all missing values were at random (MAR).

The cross-sectional study design provides only limited evidence of the causal associations between predictor and outcome variables. Conclusions MM and mental health status, both representing the need factor in the Andersen model, were the dominant predictors of health care costs. Predisposing and enabling factors had comparatively little impact on health care cost, possibly due to specific conditions of the German health care system. Overall, the variables used in the Andersen model explained only a small proportion of the total variance in health care costs.

Whether this indicates room for further improvement of the Andersen model or its limited ability to predict costs in general should be investigated in future studies. Different combinations of potential predictors of costs should be tested in various health care systems, as service use and costs in Germany are likely to differ from other countries. Nevertheless, in view of the predicted demographic change in developed countries, MM as a major cost driver has to be considered a key factor when planning future resource allocation. The development and implementation of integrated care models for patients with multiple chronic diseases and preventive programs aiming at modifiable risk factors could be options to reduce the financial strain of multimorbidity on health care systems.

The contents of articles or advertisements in The Clinical Biochemist – Reviews are not to be construed as official statements, evaluations or endorsements by the AACB, its official bodies or its agents. Statements of opinion in AACB publications are those of the contributors. Print Post Approved - PP25. © 2005 The Australasian Association of Clinical Biochemists Inc.

No literary matter in The Clinical Biochemist – Reviews is to be reproduced, stored in a retrieval system or transmitted in any form by electronic or mechanical means, photocopying or recording, without permission. Requests to do so should be addressed to the Editor. ISSN 0159 – 8090. Early detection of many disorders, mainly inherited, is feasible with population-wide analysis of newborn dried blood spot samples.

Phenylketonuria was the prototype disorder for newborn screening (NBS) and early dietary treatment has resulted in vastly improved outcomes for this disorder. Testing for primary hypothyroidism and cystic fibrosis (CF) was later added to NBS programs following the development of robust immunoassays and molecular testing. Current CF testing usually relies on a combined immunoreactive trypsin/mutation detection strategy. Multiplex testing for approximately 25 inborn errors of metabolism using tandem mass spectrometry is a relatively recent addition to NBS. The simultaneous introduction of many disorders has caused some re-evaluation of the traditional guidelines for NBS, because very rare disorders or disorders without good treatments can be included with minimal effort. NBS tests for many other disorders have been developed, but these are less uniformly applied or are currently considered developmental. This review focuses on Australasian NBS practices.

Introduction The goal of NBS is the pre-symptomatic detection of infants with congenital conditions so that treatment may be commenced as early as possible to prevent, or ameliorate, the long-term consequences of the condition. The foundations of NBS stem from the work of Robert Guthrie in the 1960s. Guthrie was a microbiologist with a disabled child, initially thought to have phenylketonuria (PKU).

This spurred Guthrie to develop testing procedures for this disorder. Guthrie’s child was ultimately shown not to have PKU but an intellectually impaired niece did. The testing procedures developed by Guthrie allowed rapid and large-scale testing of many children suspected to have this condition. Previous work had shown that a low phenylalanine diet could be used to treat PKU, and this was most effective in preventing mental retardation if the diet was commenced soon after birth and before any clinical symptoms became apparent. This raised the possibility of testing all infants soon after birth to detect PKU. Starting in the 1960s, many countries recognised the benefits of NBS for PKU and commenced programs, and today virtually all countries in which PKU is prevalent provide NBS for this disorder.

With the development of robust immunoassays for thyroxine and thyroid stimulating hormone (TSH) in the 1970s it became feasible to add congenital hypothyroidism (CH) to the NBS panel. Other disorders have subsequently been added to NBS panels and the composition of the NBS panel can vary between regions, depending on local prevalence, amongst other factors. While many conditions are potential candidates for NBS, this is not practical for all.

Guidelines for deciding whether a particular condition is a suitable candidate for screening were formulated by Wilson and Jungner in 1968 and are summarised in. Unfortunately, many aspects of these guidelines are subjective and there is not always agreement about which disorders should be part of the NBS panel. – There is almost universal consensus that NBS for PKU and CH results in greatly improved outcomes. There is also increasing evidence of improved pulmonary function, nutrition and long-term survival resulting from NBS for CF – and similar evidence is emerging for the disorders detected by tandem mass spectrometry (MSMS)., There is less evidence for other disorders. Comprehensive cost-benefit analyses are difficult to conduct for rare disorders and separate analyses often result in widely discrepant results. Furthermore, how does one place a monetary value on death if this is an outcome in unscreened babies?

The introduction of multiplex testing, exemplified by MSMS testing for several inborn errors of metabolism (IEMs), has further confounded these issues because disorders which, on their own, would not be strong candidates for NBS, can be included with minimal extra effort or cost. This has led some to propose revised NBS guidelines. As a result, NBS panels vary, even for regions with a similar population. For example, within Australasia, testing for CH, CF, PKU and other disorders detected by MSMS is universal, but only New Zealand currently tests for congenital adrenal hyperplasia (CAH) and biotinidase deficiency. Testing for galactosaemia is performed in all states except Victoria. Major NBS criteria. Summarised from Wilson and Jungner criteria.

This review considers NBS using biochemical testing and the overall strategies and practicalities for population-wide NBS testing, with an emphasis on Australasian practices. The interested reader is referred to recent reviews of PKU, – CF,, CH – and CAH, for more detail on the biochemistry, genetics and pathophysiology of these disorders. Non-biochemical NBS tests, such as newborn hearing testing, are outside the scope of the current review. Models of Service Provision NBS programs are well-established as the standard of care in most developed countries. In many regions, including Australasia, programs are state funded and testing is provided without charge to parents. Testing is voluntary in Australasia and programs report >97% coverage of newborns., Other countries offer mandated NBS programs or fee-for-service NBS. A typical service model comprises a central laboratory performing all tests and screening a sufficient number of babies to allow economies of scale and also to make adequate numbers of diagnoses to maintain laboratory expertise.

NBS cards are easily mailed and many NBS programs are heavily reliant on this mode of transport, however days on which there is no mail delivery may cause delayed turn around times and swings in laboratory workloads. Due to limitations in the way samples are collected and the inherent imprecision in dried blood spot (DBS) samples (below), it is important to emphasise that NBS is not diagnostic and any presumptive positive result therefore requires confirmation, preferably with an independent sample and test method. Referral and follow-up of abnormal results may be under the control of the NBS laboratory or through liaison with relevant clinical departments such as endocrinology, metabolic and respiratory medicine.

It is important that close communication exists between these departments and the NBS laboratory. False positive results are a concern for any NBS program because they cause parental anxiety and increased indirect costs associated with follow-up contact and testing. In some situations second-tier testing is used to improve the relatively poor positive predictive value (PPV) for some NBS tests.

Samples with an abnormal result on the primary screen test are subjected to a second, more specific test. This approach has been successfully used for CF NBS and several MSMS second-tier tests have been developed.

However, each of these second-tier tests adds to the complexity and expense of the NBS program and it remains to be seen how practical some of the more recently developed second-tier tests will be, particularly for disorders that require rapid confirmation. It is also recognised that some affected babies will be missed by NBS tests e.g. CF screening protocols will not detect ~5% of babies with CF. – It is important that false negative cases are referred back to the NBS laboratory for assessment and inclusion in audits of overall performance metrics. There are increasing regulatory, legal and ethical issues surrounding the collection, storage and use of NBS cards. Most NBS programs have established advisory committees with broad representation to assist with these issues.

The introduction of new NBS tests is another complex area with various lobby groups proposing new screening tests that then require evaluation by funding bodies. In the past this has sometimes happened in an ad hoc fashion.

In most countries there is a move to greater uniformity of NBS panels and protocols and clear guidelines for evaluating proposals for new tests. The US is probably the most advanced in this regard with comprehensive policies published by the American College of Medical Genetics. The Human Genetics Society of Australasia and the Australian Health Ministers Advisory Group on Human Gene Patents and Genetic Testing have also developed local guidelines for screening.

Sampling Cord blood sampling is attractive in regions with early discharge after delivery. Maternal contamination is a problem and, while cord blood can be used for some NBS tests such as CH testing,, it is recognised that better results are obtained from samples collected by heel prick at a later stage. In particular, cord blood has been shown to be of limited value in detecting disorders by MSMS. Collection of blood onto an absorbent paper card, often referred to as a ‘Guthrie card’, is the commonest type of NBS sample. A few drops of blood from a heel prick are collected onto a high quality cotton fibre-based paper and allowed to dry in air for a few hours before being sent to the central NBS laboratory.

Whatman TM 903 paper is widely used due to its well characterised properties and a few drops of blood are usually sufficient to complete most NBS panels. Some separation of blood components occurs during the spreading and drying of blood on the paper resulting in small concentration gradients across the blood spots.

These effects are influenced by haematocrit and drying conditions and limit the overall imprecision of any DBS test to approximately 10%. Timing of the collection is important because some metabolite and hormone levels vary markedly in the neonatal period both in normal and affected babies. Some markers decrease with age in affected babies while others increase. Consequently, sample timing is a compromise and most programs currently recommend sampling at 48 to 72 hours of age. Use of urine as a NBS sample stems from its early use for PKU testing using ferric chloride reagent for the detection of phenylketones.

Urine was collected onto an absorbent paper placed in the baby’s nappy. Amino acid screening could also detect several other IEMs. However, it was soon realised that urine testing was relatively insensitive for PKU and that blood phenylalanine levels were more effective in detecting it. The additional need for a blood sample for CH testing resulted in most programs dropping urine testing in favour of DBS testing. However, some NBS programs continued the practice of urine screening and this has recently been given renewed impetus by the application of multiplex MSMS testing to detect some IEMs that would be difficult to detect with blood testing.

Prematurity, birth weight, neonatal jaundice, parenteral nutrition, transfusions and type of feeds can all potentially influence NBS results and need to be taken into account when establishing cut-off values and interpreting results. Ideally, this information should be recorded on the NBS card to aid in the interpretation of results. Transport of samples is also important because some markers are relatively unstable and heat, humidity and delays in transport can cause degradation and potential false negatives. PKU and Hyperphenylalaninaemia PKU is caused by mutations in the phenylalanine hydroxylase ( PAH) gene and results in excessive levels of phenylalanine which are detrimental to brain development. Birth prevalence is 1:14,000. Treatment with a phenylalanine-restricted diet is effective in preventing the long-term consequences of PKU but needs to be commenced early in life.

The early NBS programs for PKU measured phenylalanine in DBS using a bacterial inhibition assay which monitored the growth of a mutant strain of Bacillus subtilis with a requirement for exogenous phenylalanine for growth. DBS samples were inoculated onto agar plates containing mutant bacteria and the size of the colonies assessed after incubation. The bacterial inhibition assay for phenylalanine has been largely replaced by other, more sensitive and specific assays such as enzymatic/colourimetric assays and most recently MSMS. MSMS testing also incorporates testing for many other metabolites (below). Babies with PKU typically have DBS phenylalanine levels >200 μmol/L (a typical NBS cut-off is ~150 μmol/L), and an increased phenylalanine/tyrosine ratio. Follow-up testing involves formal measurement of plasma amino acid levels. Some cases of classical PKU are responsive to tetrahydrobiopterin,, the co-factor for phenylalanine hydroxylase, and follow-up may also involve assessment of this response.

NBS also detects babies with mildly elevated levels of phenylalanine, termed hyperphenylalaninaemia, which may be caused by rarer defects in the biosynthesis or recycling of tetrahydrobiopterin. Assessment of urine pterin levels and response to phenylalanine and tetrahydrobiopterin loads are useful in classifying these babies. Congenital Hypothyroidism Primary CH has a birth prevalence of 1:2750.

Detection of CH by NBS relies on the immunoassay measurement of various combinations of TSH, thyroxine and thyroxine-binding globulin.,, Many NBS programs, including those in Australasia, perform a single TSH test due to its simplicity and relatively low false positive rate compared to combined strategies. This strategy will not detect central hypothyroidism and low birth weight and premature babies are a potential source of false negative screens due to hypothalamic immaturity and thus require a second sample., Iodine-containing disinfectants and contrast agents are a potential source of false positive results. Positive screen results require follow-up with formal thyroid functions tests and thyroid scans. Worldwide, iodine deficiency is an important cause of hypothyroidism. There is some evidence of mild iodine deficiency in Australasia, although the connection with hypothyroidism is uncertain. Galactosaemia Galactosaemia due to galactose-1-phosphate uridyl transferase (GALT) deficiency can result in on-going jaundice and E. Coli sepsis in newborns.

A galactose-restricted diet is effective in minimising these symptoms. Birth prevalence is about 1:50,000. Most programs detect galactosaemia by measuring ‘galactose metabolites’ i.e. Galactose and galactose-1-phosphate using enzymatic assays. – This protocol may also detect deficiencies of galactokinase, which can result in juvenile cataracts, and galactose epimerase, usually a benign condition. Milder GALT variants, such as the Duarte variant with some residual activity, are also detected but these do not generally require treatment. Measurement of GALT activity is usually used as a second-tier test to help distinguish between the different forms of galactosaemia.

Its use as a first-tier test is sub-optimal because it can result in both false positives (due to instability of the enzyme activity in DBS or use of EDTA blood) and false negatives (due to exogenous addition of activity from transfusions). Early detection of galactosaemia due to GALT deficiency can prevent short-term morbidity and occasional infant deaths. However, many cases present with typical symptoms in the neonatal period and are clinically diagnosed regardless of NBS.

There is also uncertainty about the overall benefit of NBS for galactosaemia because the long-term benefits of earlier diagnosis and the need for dietary treatment are unclear. Despite early NBS diagnosis, many patients have intellectual and developmental deficits and many adult females are infertile due to ovarian failure. Cystic Fibrosis CF is caused by mutations in the CFTR gene and causes altered properties of secretions as a result of altered chloride transport. Lung and pancreatic function are affected and long term prognosis is significantly improved by early commencement of physiotherapy and antibiotic treatment.

Several blood biomarkers are available but immunoreactive trypsin (IRT) measurement with immunoassay is most widely used in NBS. – IRT is somewhat unstable so NBS testing is unreliable if there is a delay in analysis or DBS samples have been poorly stored. Elevated IRT has relatively poor PPV in the neonatal period and early programs relied on a second DBS sample to confirm the initial screening result. The identification of the CFTR gene and disease-causing mutations opened the way to second-tier mutation testing of the original blood spot, thus eliminating the need for the second sample and also providing superior performance. Most NBS programs have now adopted this IRT/ CFTR mutation protocol.

A typical protocol is outlined in the. Pancreatitis-associated protein is also increased in CF and an interesting alternative strategy is the measurement of both IRT and pancreatitis-associated protein. This strategy has similar performance to IRT followed by second-tier CFTR mutation analysis and avoids the problems of carrier detection and the need for a genetic test, viewed as controversial by some sections of the community. A typical screening protocol for CF. IRT, immunoreactive trypsin. Babies with a single detected CFTR mutation require follow-up sweat testing because they may carry a rare CFTR mutation that is not part of the NBS panel., A sweat chloride measurement >60 mmol/L is considered diagnostic for CF and remains the ‘gold standard’ for diagnosis. It should be noted that sweat chloride reference intervals vary somewhat with age which should be taken into account if older individuals are tested.

Babies with a single detected mutation and sweat chloride. Multiplex Testing for IEMs Using MSMS The early work of Millington et al. Demonstrated the potential of MSMS to simultaneously measure amino acids and acyl carnitines in a DBS sample and diagnose many IEMs.

Elimination of accumulating acyl-CoA esters in the form of acyl carnitines is a common detoxification mechanism and provides convenient markers for several IEMs involving fatty acid or organic acid metabolism. This early work relied on fast atom bombardment to produce ionisation but this process was difficult to automate in a robust fashion. This limitation was overcome with the development of electrospray ionisation in the late 1980s and high throughput testing suitable for NBS was later demonstrated. Metabolites are usually extracted from DBS samples using methanol and quantification is achieved by incorporating stable isotope internal standards. Formation of butyl derivatives is commonly employed to improve sensitivity but analysis of underivatised samples using current generation MSMS instruments with increased sensitivity is gaining in popularity because it simplifies the assay procedure and eliminates the toxic and corrosive reagent used for butylation. Samples are directly introduced into the electrospray ionisation source without chromatography with typical run times of less than two minutes. The MSMS is usually operated in multiple reaction monitoring mode in which specifically programmed metabolites are measured, but it can also be operated in scanning mode to produce full amino acid and acyl carnitine profiles.

In practice, not all metabolites detected in scanning mode are particularly useful NBS targets. For example, glycine has limited value in diagnosing non-ketotic hyperglycinaemia. Phenylalanine is easily measured using MSMS and the introduction of MSMS testing therefore replaces the need for a separate discrete test for PKU while at the same time introducing testing for many other disorders. The most important of these is medium chain acyl-CoA dehydrogenase deficiency, a disorder of fatty acid oxidation that can result in sudden, unexpected infant death if not diagnosed. Other IEMs detectable by MSMS are summarised in. Several other IEMs, such as deficiencies of ornithine transcarbamylase and carbamoyl phosphate synthetase, are not usually detectable by MSMS because the metabolic abnormalities are not reliably detected in DBS using current technology. Disorders and enzyme deficiencies detected by NBS using MSMS.

The ability of MSMS to simultaneously measure many metabolites changed the previous focus of NBS from ‘one test for one disorder’ to ‘one test for many disorders’ i.e. A multiplex test. Doing so has changed the previous views of the traditional Wilson and Jungner guidelines. Very rare disorders, or disorders without effective treatments, can be incorporated into MSMS testing with minimal extra effort for the laboratory, merely by including the MSMS parameters for the appropriate metabolite markers.

It can be argued that while detection of such disorders may not be of great benefit to the baby, early diagnosis can be helpful to the family in avoiding a long search for a diagnosis and in planning future pregnancies. On the negative side, every additional metabolite is associated with a small false positive rate. Collectively, these can potentially result in an overall significant false positive rate but, with appropriate planning and selection of cut-offs, overall false positive rates of.

Congenital Adrenal Hyperplasia CAH is caused by defects in steroid metabolism and has a birth prevalence of ~1:18,000., It results in deficient cortisol and mineralocorticoid production with concomitant increases in androgen production. Undiagnosed infants can present with potentially lethal salt-wasting crises and masculinisation of females, while growth failure and masculinisation are longer-term consequences of untreated cases. NBS testing for classical CAH relies on detection of increased levels of 17-hydroxyprogesterone (17-OHP) using immunoassay. Milder, non-classical forms of CAH usually have normal 17-OHP levels in the newborn period and are consequently not detected by NBS. NBS for classical CAH can prevent deaths from salt-wasting crises in neonates and older children.

However, the value of NBS is diminished somewhat by the fact that most females with classical CAH are clinically obvious at birth due to masculinisation of their genitalia. Most CAH cases are due to steroid 21-hydroxylase deficiency, but CAH can also be due to steroid 11β-hydroxylase deficiency which also causes an increased 17-OHP level. Confirming and distinguishing between these two disorders is readily achieved by profiling 17-OHP and other steroids in serum or with second-tier steroid profiling of the original DBS. Cases with equivocal results may require stimulation of the adrenal cortex with cosyntropin or synacthen to confirm the diagnosis. Existing immunoassays for 17-OHP are somewhat nonspecific and antibodies cross-react with a number of steroids of foetal origin. As a consequence of these interfering steroids, premature and low birth weight babies have apparently high values for 17-OHP by immunoassay, confounding the interpretation of results in these babies and resulting in a poor PPV for some NBS programs. Use of age- and weight-related reference intervals can improve this situation.

Removal of the interfering steroids by solvent extraction has also been used to improve testing. More specific testing for 17-OHP using LC-MSMS has also been developed and has been shown to significantly improve PPV when used as a second-tier test for samples with an increased 17-OHP immunoassay level. Other NBS Tests NBS programs for sickle cell anaemia and other haemoglobinopathies have been established using HPLC or isoelectric focusing. – NBS for glucose-6-phosphate dehydrogenase deficiency, is carried out in several Asian countries. There is currently insufficient justification to screen for these disorders in Australasia but this situation could change in the future with changes to the ethnic mix of the population. Biotinidase deficiency can cause seizures and developmental delay and NBS is carried out in several countries – including New Zealand. Testing for severe combined immunodeficiency has been developed and was recently added to the mandatory NBS panel in the US.

Many NBS tests for other disorders have been developed over the years, but large numbers of these have been disbanded as they were impractical or the long term benefits of incorporating them into NBS were not established. Several new NBS tests have recently been developed but are currently regarded as developmental (see below). Secondary Uses for NBS Samples DBS cards are usually stored for quality assurance purposes e.g. Follow-up of NBS false negative cases. Protocols vary between laboratories and storage periods typically range from two years (i.e. Until any NBS false negatives should be clinically diagnosed) to indefinitely.

Stored cards are a valuable resource as they represent a complete population and also may allow historical comparisons to be made when samples have been stored for long periods. Stored DBS cards have also been used for diagnostic purposes other than those for which they were originally collected e.g. Retrospective genetic diagnosis when the proband has died and there is no other DNA-containing material available. Detection of cytomegalovirus DNA in NBS DBS samples is also useful in retrospectively establishing congenital cytomegalovirus infection as a cause of deafness in older children, because the infection has typically cleared at the time of diagnosis. Stored DBS cards have been used for the forensic identification of human remains in accidents or natural disasters such as the Victorian bushfires of February 2009.

Researchers have also realised the potential of DBS cards e.g. Establishing reference intervals or carrier frequencies in the general population. Such research requires ethics approval and parental consent for projects which require identified DBS samples. De-identified DBS samples can be used for research without consent.

Some sections of the community have concerns regarding the long-term storage of DBS cards and the potential privacy issues and use of genetic information. These concerns have led to the destruction of stored DBS samples in some programs.

However, it is important to emphasise that the vast majority of stored DBS samples are never accessed once the initial NBS testing has been completed and NBS programs hold minimal genetic information on any individuals, which is usually limited to specific genes e.g. NBS programs have protocols in place to ensure privacy is maintained and that any requests for access to samples are carefully evaluated. It is to be hoped that these protocols and on-going education will maintain the public’s confidence in the NBS process and prevent a vocal minority dictating NBS policies. Future Trends Advances in computerisation, automation and sensitivity of analytical instruments have resulted in the proliferation of potential new NBS tests in recent years. Furthermore, improvements in treatments such as bone marrow transplantation and enzyme replacement and newer pharmacological approaches such as chaperone therapy and read-through of premature stop codons,, have raised the status of some disorders as candidates for NBS.

NBS tests have been proposed for lysosomal storage disorders, – Duchenne muscular dystrophy and Wilson’s disease, to name a few. While pilot projects have demonstrated the potential of these tests, it remains to be seen how effectively they can be applied in most NBS centres and how effective the newer treatments are.

Each new test adds an additional layer of complexity to the NBS program and some of the treatments are both expensive and long-term. Multiplexed protein assays may be one way to simplify some laboratory aspects of NBS. This approach has been demonstrated for lysosomal storage disorders using coded microbead immunoassay technology, but it can be applied to a wide range of proteins and could be used to combine existing tests such as IRT, TSH and 17-OHP. Increased use of molecular testing in NBS is attractive because it has the potential to increase the performance metrics of testing and target disorders that may not be amenable to biochemical testing.

An NBS chip with a large array of targeted mutations could detect a far greater range of disorders than is currently tested. Pilot studies have shown the feasibility of genomewide scans of DBS, and it has even been suggested that complete genome sequencing of newborns will occur in the near future. These ideas are clearly controversial. Apart from the issues of accuracy, the detection of carriers and sequence changes of unknown significance, it is unclear if such large-scale genetic testing will be acceptable to parents and the general public. However, well-targeted molecular testing has obvious technical benefits for many disorders and will become increasingly attractive as the cost of molecular testing falls in comparison to biochemical testing.