PEDIATRICS Vol. 106 No. 6 December 2000, pp. 1520-1524

The Milliman-Robertson Length of Stay Debate

To the Editor.

The recent article by Sills et al¹ asks whether the recent Milliman and Robertson (M&R) Health Status Improvement and Management (HSIM) publication² presents realistic length-of-stay criteria. The article does not directly answer the question posed. Faculty members in the Department of Pediatrics at the University of Texas at Houston provided contributions to the M&R HSIM, with the intention as stated by the authors "of focusing on best practice rather than median performance." Even with best practice in mind, inspection of the histograms presented in Fig 1 demonstrates that the goal LOS was in fact modal in at least half of the 16 diagnoses. Reanalysis of the published data (Table 1) shows that the average of the "percentage of patients with LOS > M&R" for the 16 diagnoses equals 63%; thus, on average, 37% were discharged on or before the M&R goal LOS. Weighting the diagnoses by the number of patients with each diagnosis, 46% of actual patients were discharged on or before the M&R goal LOS. These published data demonstrate that even with retrospective application of 1999 guidelines to uncontrolled 1995 data in New York, which is characterized in the article as relatively "inefficient," about 40% of the discharges met or exceeded the goal LOS. Much has changed since 1995, generally in the direction of improving outpatient management of many pediatric diseases. However, if the question to be answered by the article is whether the goal LOSs are "realistic," as suggested by the title, then the data clearly indicate that the goal LOS was in fact achieved by patients with each illness, and that overall, almost half of the children went home by the goal LOS.

In the HSIM, LOS was consistently presented as "goal LOS," rather than "average LOS" or "maximum permissible LOS." The goal LOSs were intended as "ambitious," and to be applicable only to uncomplicated patients. Many contributors used the notion of goal LOS as analogous in part to the game of golf. If one considers pediatricians to be like golf pros providing professional services to the broad range of golfers, it should not be surprising that many or most receiving professional services (patients) do not achieve par. Less than 5% of the golfing population regularly shoots par on a given hole (G. Norman, personal communications, May 2000). However, I would submit that our problem in pediatrics is not with those who set par at a particular number of strokes for a golf hole or golf course, but rather with those who would set the greens fees (or reimbursement) too low, would ask the golfers to leave the course after 72 strokes, and would hold the golf pros responsible for the diversity in the performance of the golfing population and their routine failure to achieve par. Those actions would be unacceptable in golf, and by analogy, I would suggest that the pediatric community should offer strong and informed resistance to both episodic and systematic misapplications of the "goal of LOS" to their patients.

Pediatric HSIM was developed as a compendium of best practices within the continuum of care for pediatric patients, and was not intended to provide the detailed medical information necessary for diagnosis or for direct patient care. HSIM was not intended as guidelines in the sense of the Institute of Medicine or the Agency for Health Care Policy and Research usage of the term. While generally referred to in the article as "guidelines", HSIM itself uses the word guideline only in the introductions, and only in a lower-case, generic sense. However, HSIM does systematically cite and incorporate professional guidelines from the American Academy of Pediatrics (AAP) and other pediatric bodies. Although many physicians may feel powerless and frustrated by the current healthcare environment, the AAP and other pediatric organizations exert great influence in the managed care arena by publishing clear statements as professional guidelines, which then form the basis for the recommendations appearing in the HSIM and other similar publications. By working with the commercial consultants who advise those who administratively "manage" or pay for pediatric health care, I believe that the pediatric community is paradoxically empowered to reassert professional standards controlling the clinical practice of pediatrics and to serve and protect our patients in the changing health care environment.

John W. Sparks
Department of Pediatrics
University of Texas at Houston Medical School
Houston, TX 77030

REFERENCES

1. Sills MR, Huang ZJ, Shao C, Guagliardo MF, Chamberlain JM, Joseph JG Pediatric Milliman and Robertson length-of-stay criteria: are they realistic? Pediatrics. 2000; 105:733-737 [Abstract/Free Full Text]
2. Milliman & Robertson, Inc. Schibanoff JM, ed. Pediatric HSIM (Health Status Improvement and Management), 1998. (Note that the original article did not cite the printed publication, copyright 1998, but rather the access of the authors to the M&R Web Site).

To the Editor.

Neither the article on Milliman and Robertson (M&R) criteria for hospital length of stay (LOS) by Sills et al¹ nor the dueling commentaries^2,3 in the same issue discuss some simple points that may improve pediatricians understanding of some of the issues at stake.

1. The M&R criteria are properly used and understood as targets or goals for hospitalizations. They cannot be used as a sole determination of appropriateness of hospitalization, if for no other reason than the fact that they are totally insensitive to secondary diagnoses and complications, including psychosocial issues! But they might be appropriately used as a benchmark, which can trigger an appropriate utilization review, and as a target for hospitals to use in the discharge planning process.
2. Given this understanding of how the M&R standards could be used, are they really so far off? Looking at the data in Fig 1, the discerning reader can easily see that data plots are almost all skewednot normaldistributions. Therefore, the mean is not the best measure of central tendency, and the authors have appropriately presented the median and mode for each data set in Table 1. Looking at Table 1, one can see that the M&R criteria differ significantly from the median and modal values in only 4 out of 16 data sets.
     The burn admission data shows a mode (1) that is less than M&R (3) and a median (5) slightly greater than M&R. This suggests a significant number of 1-day admissions, for unclear reasons. It is unlikely that all of these are attributable to observation for smoke inhalation, and the data on this admission pattern clearly need a deeper analysis to determine the appropriateness of very short admission.   The other areas of significant difference are for meningitis, osteomyelitis, and appendicitis.
3. For these 3 serious conditions, the mode is far above M&R targets. Why?
     First, these 3 conditions are likely to have very significant complication rates, may be associated with underlying immune disorders (even appendicitis in systemic lupus erythematosus, for example), and can be caused by organisms that are difficult to treat (tuberculosis, salmonella, etc).   But perhaps most important is the fact that for M&R criteria to make any senseand remember that they were developed by reputable physicians using a model for consensus development that is reasonable, if not perfect (3)one has to have aggressive home health services available for uncomplicated cases! With aggressive home health care, early discharge for these diagnoses can be done and is appropriate, but it is very unlikely that such services were readily available in New York in 1995! Even here in California, an "advanced" managed care market, good home health services for children, in particular, are relatively hard to come by. But this does not mean that we should not be striving to implement them.
4. The reason we need to try to minimize hospital admission time is not just to reduce costs. Hospital admissions are a major stress on children and families, and expose patients to the all-too-common complications of nasty, drug-resistant nosocomial infections and iatrogenic medication errors. One should not forget this in the midst of arguments about who is getting paid for what.

The authors have made a good effort, but with old data from a place (New York) and time (1995) which are not a fair comparison to reality in other parts of the country in the year 2000. In the future, investigators should be encouraged to examine comparison data from more "managed" markets, to look at the availability of home health care in relation to the data, and to present the rates of secondary diagnoses or complication diagnoses versus uncomplicated cases when analyzing their results.

Meanwhile, used properly as a set of targets for achieving reasonable, early hospital discharges, but not as sole criteria for determining the medical necessity of length of hospital stay, the M&R criteria are not as bad as some would assert,² though continued efforts to improve them, like all other benchmarks, also makes good sense.

John M. Goldenring
San Diego, CA 92131

REFERENCES

1. Sills MR, Zkihuan JH, Cheng S, Guagliardo MF, Chamberlain JM, Joseph JG Pediatric Milliman and Robertson length-of-stay criteria: are they realistic? Pediatrics. 2000; 105:733-737
2. Bauchner H, Vinci R, Chessare J Milliman and Robertsongoing in the wrong direction. Pediatrics. 2000; 105:858-859 [Free Full Text]
3. Yetman RJ Patient care guidelines in pediatrics. Pediatrics. 2000; 105:859-860 [Abstract/Free Full Text]

To the Editor.

We find the study done by Sills et al to be well-done and have no concerns regarding the methodologic approach. We have no concerns about the study's results that show New York average lengths of stay during 1995 were in excess of 1998 published inpatient goal Lengths of Stay (GLOS) in M&R Care Guidelines' Pediatric Health Status Improvement and Management (HSIM) publication. This result is the only possible result because the average length of stay (LOS) in any statewide hospital database must be, by definition, in excess of the GLOS. The GLOS represents the best achievable LOS; thus, the average of all stays, by definition, must be longer than the best achievable.

However, we do not agree with the Sills conclusions regarding the impact of GLOS on patient care. We believe the Sills conclusions are based on a misunderstanding of the definition of GLOS, which is an outcome measurement of each inpatient Guideline. We believe that Sills also misunderstands how the GLOS relates to a historical average LOS.

The study's conclusions should be reconsidered based on the following 5 points:

1. GLOS does not equal average LOS and can only be achieved in a percentage of patients admitted with each diagnosis.
2. The GLOS is only a small part of the inpatient Guidelines. The majority of the inpatient content is clinical content. The Pediatric HSIM publication also presents clinical content across the continuum of pediatric care. As pointed out in the response by Bauchner et al in their comments on page 858, Pediatric HSIM also includes Guidelines on anticipatory guidance, subspecialty care, chemotherapy, immunizations, neonatology, and comprehensive management of chronic disease.
3. All M&R Care Guidelines, including the Pediatric HSIM publication, are evidence-based. This point is included in Dr Yetman's commentary on page 859.
4. M&R Care Guidelines is committed to improving both the quality and efficiency of health care to children.
5. M&R Care Guidelines is scrupulous in how information is presented in its publications so that the content may be readily understood and used appropriately. M&R Care Guidelines is very diligent in presenting information and providing training on appropriate use to users of M&R Care Guidelines. We include explicit language in the Guidelines themselves, on our web page (www.mnr.com), in our training programs, and as a part of our licensing agreements.

	DEFINITION OF GOAL LENGTH OF STAY

We explicitly define the Goal Length of Stay (GLOS) in the Pediatric HSIM publication. The following 2 paragraphs are taken from the introduction to the inpatient Guidelines section:

"Goal length-of-stay: The expected length of inpatient hospitalization required to manage each condition is listed. This length-of-stay assumes that treatment and healing occur without significant complications. Should treatment and healing not occur in the timeframes outlined, the guidelines become appropriate when the patient's stage of recovery reaches a level of acute care similar to those listed."¹

"We reiterate that the purpose of the inpatient guidelines is to define care for patients who recover from their illness as well as can be expected and without complications. For some conditions, the presentation can be highly variable. For the more extreme presentation of these conditions, the optimal recovery timetable described herein is optimistic. The patient's clinical status on each day of recovery defines where on the guideline the patient falls. When the patient's clinical status does not allow progression to the guideline's next day of recovery, the ambitious goal length-of-stay listed will not be fulfilled."²

The Sills analysis used functional LOS: Functional LOS is equal to M&R's GLOS plus 1 day. The M&R GLOS is counted the way the health care industry counts hospital days; that is, overnight stays in the hospital. Each inpatient M&R Care Guideline includes clinical information for each functional day in the hospital. The clinical information includes both the day of admission or surgery and the day of discharge; thus, the GLOS is always 1 day less than the number of days the patient is functionally in the hospital. This definition is consistent with what Sills defines as the functional LOS.

	OVEREMPHASIS OF DIFFERENCE BETWEEN GLOS AND AVERAGE LENGTH OF STAY

We take exception to their statement in the discussion on page 736 where Sills describes the "discrepancy between M&R guidelines and observed LOS data." First of all, the GLOS is one measurement goal within the Guideline and is achieved only if the patient has optimal recovery, no complications, meets the discharge criteria, and goes home to a safe environment with adequate follow-up care. Second, Sills' analysis shows that a significant number of the patients in this study in New York in 1995 did go home at the M&R "functional LOS."

Figure 1 clearly shows that the M&R functional LOS is identical to or 1 day less than the mode LOS in most of the diagnoses. This demonstrates that in this population a percentage of patients did achieve M&R's GLOS. Table 1 shows for all diagnoses except "Appendectomy, complicated" that many patients achieved either the GLOS or GLOS plus one day.

	QUALITY AND EFFICIENCY

We believe that the issue of efficiency must be looked at when developing Guidelines for inpatient care. References 11-19 in the Sills article are good examples of the kind of excellent work that supports our belief. We reject the notion that longer lengths of stay represent better patient care. We believe that the best quality is achieved when the patient is able to recover quickly. Potentially avoidable days of hospital care may not be safe for the patient. One cannot make the implicit assumption that better care is provided when the patients stay in hospitals longer. Safety issues recently presented in the Institute of Medicine study³ support efficient use of hospitalization. In addition, 2 studies in 1999 show that the rates of nosocomial infection are lower in the home than in the hospital.⁴ Rates of nosocomial infection in the hospital are 5% to 10% while only 2% in nonhospital settings.⁵ These do not support the notion that more inpatient care is better care.

	CHART REVIEWS AND DATA SUGGEST SIGNIFICANT OVERUSE OF INPATIENT RESOURCES

M&R's physician and nurse consultants have completed >75 000 inpatient chart reviews over the last 10 years. Potentially avoidable days for client organizations in these reviews ranged from 10% to >50%. The causes of these potentially avoidable days are many, including delays in hospital services, delays in consultations, delays in transfer, inability to arrange posthospital home care, or skilled nursing facilities. In >75 000 chart reviews that M&R has completed, we have found that most potentially avoidable days related to the last day of admission.

An analysis of all health maintenance organization (HMO) hospital discharges in the state of California in 1997 showed that, for diagnoses for which an M&R inpatient Guideline was available, 41.8% of pediatric discharges met the GLOS.⁶ In the state of Washington, the comparable percentage was 43.5%.⁷ These percentages represent performance as of several years ago in a mix of health plans with varying degrees of efficiency. Also, the percentages include patients with significant complications and comorbidities, for whom the Guidelines are not suitable for management to the GLOS. In this light, we believe the observed percentage of adherence to GLOSs to be consistent with our definition of the goal as a target attainable for a large percentage of patients receiving optimally managed care.

	EVIDENCE-BASIS FOR M&R CARE GUIDELINE DEVELOPMENT

M&R Care Guidelines are evidence-based and have been since our initial Guidelines were first published in 1990. We have always used current best evidence. As Dr Yetman said in his commentary:

"The HSIM guidelines are formulated on the principles of evidence-based medicine: `the conscientious, explicit, and judicious use of current best evidence in making decisions about the care of individual patients.' The evidence on which guidelines were based ranged from highly controlled, randomized trials to the applied pragmatic work associated with quality improvement projects. Practicing pediatricians and medical school faculty reviewed data for these guidelines from published and unpublished studies using three general levels of evidence. Level 1 evidence included recommendations based on scientifically rigorous, highly controlled studies (randomized, controlled trials). Ideally, randomized, controlled trials would be available to answer all questions in the pediatric care arena, but unfortunately for medicine (and for pediatrics in particular), this level of evidence is rare for many of the most basic questions. When Level I data were not available, we turned to Level II evidence (consisting of nonrandomized but published research) and Level III evidence (primarily unpublished evidence, quality improvement project, large databases, and expert opinion) as the basis of the recommendations found in the guidelines. Annotations for the sources of information are included with each guideline. Nothing in medicine, of course, stands still and the development and updating of guidelines is an ongoing process. It includes continuous research and verification of information from a wide variety of sourcesamong them clinicians, practicing physicians in managed care environments, medical school physicians, specialty societies, data from hospitals and health plans, and input from guideline users."

In 1999, our Inpatient and Surgical Care publication included annotation and a grading of the evidence used to develop the Guideline. We categorize evidence into three grade levels with the Guidelines. This annotation accompanies each Guideline. We are expanding the annotations of our evidence base to all of our publications to be released in 2000.

	BACKGROUND ON M&R CARE GUIDELINES

M&R Care Guidelines are developed within a division of Milliman & Robertson, Inc. Milliman & Robertson, Inc. is an international consulting firm with 4 major disciplines serving the health care, insurance, employee benefits, and financial services industries. Milliman & Robertson is a privately held corporation that values its independence from special interests. Our Guideline development process is explicitly designed to avoid external special interest influence in the Guideline development process. This allows us to complete our research and development independently, which has been a challenge for many others trying to accomplish this Guideline development. The M&R Care Guidelines division consists of 50 employees with 12 clinical full-time clinicians including doctors and nurses with strong academic, clinical, managed care, delivery system, and information system backgrounds.

As described above, we use an evidence-based approach to our Guideline development. Our approach is uniquely different from most clinical groups that approach the Guideline development process, because our Guidelines look at both quality of care and the efficiency of that care. We believe very strongly that every clinical decision is a resource decision and every resource decision is a clinical decision, and thus we must include both perspectives in Guideline development.

Most of our Guidelines, including Pediatric HSIM and Primary and Pharmaceutical Care, include both clinical information and resource information to facilitate decision-making in care processes. Our inpatient care Guidelines include information pertinent to all aspects of recovery from inpatient care including case management information to assist care planners in preadmission and discharge planning. In addition, we present day-by-day expected care that follows the most common clinical progression of the patient. Each includes Guideline discharge criteria in the last day of the Guideline to assist providers in making the decision if the patient is ready to go home or to another level of care. The home health care Guidelines follow each inpatient care Guideline to assist providers if home health care is indicated.

	SUMMARY

Milliman & Robertson, Inc. goes to great lengths to provide information on the appropriate use of the Guidelines and the GLOS. Pediatric HSIM contains the statement, "Goal length of stay assumes no complications. Patients can be discharged to outpatient care when all the criteria for the last day of hospitalization have been met." Milliman & Robertson provides guideline training, consultation in medical management, publications such as the Questions & Answers brochure, information on its web site (www.mnr.com), and an annual Users Forum to educate doctors and nurses on the appropriate use of the guidelines. We respond directly to providers or others on any concerns they raise.

In response to Sills' last paragraph, we would agree that more research needs to be conducted in the area of LOS and its effect on patients and family. However, one must not make the assumption in this argument that long hospital stays equate to better outcomes and/or greater patient satisfaction. Sills et al refer to the publication of Hay et al.⁸ We agree that Hay et al is important research because it challenges the widely believed but not proven assumption that longer hospital stays produce better outcomes. In stable, low-risk patients with upper gastrointestinal tract hemorrhage, "the absolute number of preventable life-threatening iatrogenic events (catheter-associated sepsis) exceeded the number of major bleeding recurrences for the control group." The patients who were discharged earlier experienced better outcomes because they did not have hospital-related complications. We believe that many patients in the New York State sample were subject to similar unnecessary complications.

We are hopeful that further studies and guideline work conducted by others in addition to M&R Care Guidelines will also focus on the LOS as one of several outcome indicators. We will continue to include that as one of our indicators while at the same time doing our best to ensure proper understanding and use of the GLOS in providing quality and efficient care to children and other patient populations.

James Schibanoff
Editor-in-Chief
M&R Care Guidelines
Seattle, WA 98104

Richard L. Liliedahl
Chief Medical Officer
M&R Care Guidelines
Seattle, WA 98104

REFERENCES

1. Pediatric HSIM, Chapter VI, Inpatient Guidelines: December 1998:6.1
2. Pediatric HSIM, Chapter VI, Inpatient Guidelines, December 1998:6.2
3. Committee on Quality of Health Care in American, Institute of Medicine. To Err is Human, Building a Safer Health System. Washington, DC: National Academy Press; 1999
4. Bergogne-Berezin E Current guidelines for the treatment and prevention of nosocomial infections. Drugs. 1999; 58:51-67 [Medline]
5. Hoffman-Terry ML, Fraimow HS, Fox TR, Adverse effects of outpatient parenteral therapy. Am J Med. 1999; 106:44-49 [CrossRef][Medline]
6. 1997 Patient Discharge Data for California, Public Use Version A. Sacramento, CA: Office of Statewide Health Planning and Development, State of California; 1999
7. 1997 Comprehensive Hospital Abstract Reporting System (CHARS) Public Data File. Olympia, WA: Washington State Department of Health; 1999
8. Hay JA, Maldonado L, Weingarten SR, Ellrodt AG Prospective evaluation of a clinical guidelines recommending hospital length of stay in upper gastrointestinal tract hemorrhage. JAMA. 1997; 278:2151-2156 [Abstract]

To the Editor.

Sills et al's article comparing Milliman and Robertson (M&R) length-of-stay (LOS) criteria to New York state LOS observations for comparable diagnoses is a welcome addition to the debate on medical management guidelines. Unfortunately the analysis is flawed by the fact that a day to M&R is not the same as a day to the rest of us. As stated to me in private conversation with principles from M&R and loosely in the introduction to the new Pediatric HSIM Guidelines, an M&R day is a phase in the treatment of a specific condition and not necessarily 24 hours. This means that an M&R day is not complete until all the items in that phase have been accomplished. For instance, if a guideline says that a child is afebrile by day 2, that means that the child does not enter day 2 until he is afebrile. For many conditions this may take several days. Thus Sills et al are comparing a 24-hour day to an arbitrary phase in the course of treatment for the same illness.

This difference in the definition of a day has been brought to M&R's attention repeatedly. This difference has been the chief problem with the implementation of the guidelines by third-party payers who have not been educated in the proper use of the guidelines. I believe M&R has done little to remedy this problem. It is a defect in their product that is critical, and has caused pain and suffering to an unknown number of patients, families, and physicians. Such a defect in any other product, once discovered, would bring prompt modification or recall. Why hasn't M&R rectified this problem? I believe it is because this product was designed specifically to save money and not to improve the quality of care delivered to pediatric patients.

I commend Sills et al for their effort and hope it brings this topic to the attention of a broader number of people who will raise their voices for action to correct this problem.

Edward O. Cox
Pediatric Ambulatory Care
DeVos Children's Hospital
Grand Rapids, MI 49503

To the Editor.

The recent article by Sills et al [Pediatrics. 2000;105:733-737] and accompanying commentaries by Dr Bauchner et al and Dr Yetman deserve some clarifications and comments. The Milliman and Robertson (M&R) pediatric guidelines [Pediatric HSIM, December 1998] are seriously flawed with respect to the length-of-stay (LOS) guidelines, particularly for serious infections. Sills et al have correctly noted that "many length of stay guidelines are not rigorously evidence-based." They have assumed that the LOS guidelines were developed by the Department of Pediatrics at the University of Texas Medical School at Houston and reflect the local management practices within that institution. Such is not the case. Although we at the University of Texas are committed to providing high-quality care that is cost-effective, the M&R LOS guidelines do not reflect our practice. Indeed, many of us feel that the M&R LOS guidelines do not conform to standard of care and reflect poor practice rather than "best practice." M&R have stated that "the guidelines are a picture of what is possible andmost importantwhat is now being done" [M&R, June 1998]. They say the guidelines are based on actual practices of physicians. One wonders where they might find competent pediatricians whose practice conforms to these guidelines.

Sills et al demonstrate that the M&R guidelines frequently fail to reflect practice in New York state. However, the guidelines they discuss are not the only ones that are problematic. There are others that are even more outrageous: endocarditis, 3 days in hospital; brain abscess, 3 days in hospital postop; neonatal sepsis, 3 days in hospital; neonatal bacterial meningitis, 5 days in hospital; etc. M&R state that their guidelines are based on evidence including the medical literature. Where might one find randomized, controlled trials demonstrating the safety/efficacy of managing pediatric endocarditis with a 3-day hospital stay?

The commentary by Bauchner et al appropriately points out that these LOS guidelines fail to "meet most of the methodologic standards established for guidelines" and that the length of stay guidelines in many cases "might represent unethical care." In fact, even aspiring to these LOS goals might cause injury, as in the case of the 1-day recommendation for management of diabetic ketoacidosis in coma.

The commentary by Dr Yetman, who is employed by M&R as well as the University of Texas, says that his guidelines "apply to most patients in most situations." However, it is clear that many of the LOS guidelines would rarely apply to any patient with the conditions listed above.

It should further be pointed out that a number of individuals including myself, listed by M&R as "Contributing Authors," did not author sections and were not asked to be, nor agreed to be, "Contributing Authors." Legal action has been initiated against M&R and Dr Yetman to undo the damage associated with these dangerous, arbitrary, and invalid LOS guidelines and the unauthorized listing of individuals as "Contributing Authors" who, in fact, were vehemently opposed to their content.

Thomas G. Cleary
Department of Pediatrics
University of Texas at Houston Medical School
Houston, TX 77805-3354

In Reply.

We appreciate the comments of Drs Cleary, Cox, Goldenring, Sparks, and Schibanoff and Liliedahl in response to our manuscript. The debate reflected in these letters about whether the Milliman & Robertson (M&R) guidelines are reasonable is welcome and was, indeed, the point of the article. These letters raise 4 general issues that remain unresolved and require further investigation.

The first issue is how the M&R guidelines were developed. As the letters make clear, there is debate about the evidentiary basis for the length-of-stay (LOS) guidelines. Schibanoff and Liliedahl of M&R contend the guidelines were developed at the University of Texas at Houston based on "current best evidence." On the other hand, Dr Cleary, Professor of Pediatrics at that same institution, echoes Baucher et al's editorial opinion¹ that the techniques used fail to meet "methodologic standards established for guidelines," further contending that many "authors" listed did not contribute or agree to be listed as authors.

A second issue concerns how the guidelines are being applied in practice. Dr Cox's letter points out that an "M&R day" is not the traditional Copernican 24-hour day, but rather a "phase in the treatment of a specific condition," so that a "day" is not completed until the criteria for that phase are met. However, Dr Cox also suggests that many payers use the 24-hour day when implementing M&R's guidelines and that "M&R has done little to remedy this problem ... caus[ing] pain and suffering to an unknown number of patients, families, and physicians." Dr Goldenring recommends the guidelines be used as a "benchmark" for utilization review, not as a "sole determination of the appropriateness of hospitalization." Dr Sparks recommends that "the pediatric community should offer strong and informed resistance" to application of M&R criteria to their patients. The conflicting reports of how guidelines should be used and are being used underscore the importance of exploring how, in fact, they are being implemented in hospitals providing care for children.

The third issue, which naturally follows that of implementation, is the question of impact: what are the consequences of the guidelines, as implemented, for patients, families, physicians, and health care institutions? The letters by Drs Sparks, Goldenring, and Schibanoff and Liliedahl note that many patients did achieve M&R LOS guidelines. Even when some or even when a majority of patients meet criteria, uncompensated care still may be required by a large number of children. What is an acceptable burden of uncompensated care? And, just as importantly, who bears the financial, social, and health quality burdens of this uncompensated care? Although these questions are hardly new, they gain new urgency as payers gain prominence in the health care industry. The potential impact of LOS guidelines on hospitals caring for children is a particularly important topic for additional research, which we recently explored in an abstract presentation.²

Finally, given that efficient care can benefit all parties involved, what barriers exist to achieving that efficiency? Variations in illness severity challenge guidelines for all age groups. Pediatric guidelines face the additional challenges of variations in patient size and age, as well as family circumstances and social disadvantage. Finally, the issue of geographic variations in resources, such as home health services, makes uniform application of guidelines difficult.

In summary, the four issues raised by the letters represent important and open opportunities for further research.

REFERENCES

1. Bauchner H, Vinci R, Chessare J Milliman and Robertsongoing in the wrong direction. Pediatrics. 2000; 105:858-859
2. Joseph JG. Are some hospitals being disadvantaged by use of LOS guidelines? Relationship of selected hospital and patient characteristics to performance in achieving pediatric LOS guidelines. AHSR Annual Meeting, June 2000, Los Angeles, CA. Abstract 373