NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Vickrey BG, Shekelle P, Morton S, et al. Prevention and Management of Urinary Tract Infections in Paralyzed Persons. Rockville (MD): Agency for Health Care Policy and Research (US); 1999 Feb. (Evidence Reports/Technology Assessments, No. 6.)

  • This publication is provided for historical reference only and the information may be out of date.

This publication is provided for historical reference only and the information may be out of date.

Cover of Prevention and Management of Urinary Tract Infections in Paralyzed Persons

Prevention and Management of Urinary Tract Infections in Paralyzed Persons.

Show details

2Methodology

Meeting with Consortium for Spinal Cord Medicine

The Project Task Order Manager (Barbara G. Vickrey, MD) and Evidence-Based Practice Center (EPC) Director (Paul G. Shekelle, MD) met with the Steering Committee of the Consortium for Spinal Cord Medicine Clinical Practice Guidelines on October 24, 1997, to:

1.

Present the scope of work and key steps proposed for the EPC.

2.

Request the nominations of technical experts and affected persons to serve on an advisory panel and provide peer review for project documents.

3.

Identify key articles, including those on magnitude of the problem, cost, and evidence.

The Consortium for Spinal Cord Medicine is supported by the Paralyzed Veterans of America, who nominated this topic. The Steering Committee includes representatives from the consortium. The consortium member organizations are:

  • American Academy of Orthopedic Surgery
  • American Academy of Physical Medicine & Rehabilitation
  • American Association of Neurological Surgeons
  • American Association of Spinal Cord Injury Nurses
  • American Association of Spinal Cord Injury Psychologists and Social Workers
  • American Congress of Rehabilitation Medicine
  • American Occupational Therapy Association
  • American Paraplegia Society
  • American Physical Therapy Association
  • American Psychological Association
  • American Spinal Injury Association
  • Association of Academic Physiatrists
  • Association of Rehabilitation Nurses
  • Congress of Neurological Surgeons
  • Department of Veterans Affairs
  • Insurance Rehabilitation Study Group
  • Paralyzed Veterans of America

Scope of Work

Project staff described the scope of work as an evidence report and supplemental analysis on the prevention and management of urinary tract infections in paralyzed persons. They further outlined the key steps as follows:

1.

Identify technical experts to provide primary input and advice to the project.

2.

Refine the research topic.

3.

Perform a literature search and evaluation.

4.

Systematically synthesize the literature.

5.

Perform supplemental analyses.

6.

Produce and disseminate an evidence report.

Nominations of Technical Experts

Project staff requested the participation of technical experts to assist the project with topic refinement; develop the systematic literature search strategy, including suggestions for key words, journals, and databases; synthesize literature; provide supplemental analyses; and prepare the final evidence report. Project staff initially proposed to recruit technical experts with the following areas of expertise:

  • Urology
  • Physiatry rehabilitation (expertise in neurology/multiple sclerosis)
  • Infectious disease (expertise in issues of neurogenic bladder)
  • Rehabilitation/spinal cord injury nursing
  • Research psychology
  • Advocacy group representation

The Steering Committee suggested that the following five additional specialties be included on the technical expert panel:

  • Pharmacology
  • Bacteriology
  • Urology with urodynamics testing expertise
  • Surgical urology
  • Microbiology

The Steering Committee then provided the names of 24 suggested panel representatives and/or peer reviewers and further identified those perceived as "key."

Article Identification and Preliminary Assessment of the Literature

The Steering Committee provided project staff with citations for several seminal articles about spinal cord injury, in general, and urinary tract infection, in particular. Project staff also identified a Consensus Conference statement and obtained background papers from a National Institute on Disability and Rehabilitation Research Conference held in January 1992, on "The Prevention and Management of Urinary Tract Infections among People with Spinal Cord Injuries."

In addition, to obtain an overview of the scope of the available literature, the project reference librarian performed several preliminary literature searches of the MEDLINE and EMBASE databases prior to the December 4-5 technical experts' meeting. The preliminary search included multiple sclerosis, spinal cord injury, and cerebrovascular disorder (stroke). The key terms for the search included "bacteriuria," "neurogenic bladder," "urinary tract infection," and/or "bladder." This preliminary search, conducted solely to gauge the extent of the available literature, spanned 1985 to the present and included foreign language articles. A total of 1,278 articles were identified.

Panel Meeting

Recruitment of Panel

Project staff recruited the persons identified by the Steering Committee as key personnel for the expert panel by telephone and determined common dates for meeting attendance. Project staff identified areas of expertise not provided by recruited panel members, specified slots for additional panelists, and requested curriculum vitae from potential panelists for the additional slots. The panel included consumers and a managed care organization medical director, as well as physicians, nurses, and a psychologist.

Initial Meeting

Project staff scheduled an initial meeting for 13 technical experts, consumers, and a managed care organization representative, at RAND (Santa Monica, California) on December 4-5, 1997. (See Table 1 for the final list of 13 panelists, including name, expertise or perspective, and geographic location.)

Table 1. List of Technical Experts.

Table

Table 1. List of Technical Experts.

Purpose of the Meeting

The purpose of the meeting was to familiarize the technical experts with the project and to utilize their expertise to define key parameters for the project literature review and analysis. They addressed the following topics at the meeting:

1.

Definition of the clinical target population,

2.

Determination of the clinical outcomes of interest,

3.

Selection of the key research questions,

4.

Development of causal pathways for each of the selected key questions, to guide the review of the evidence, and

5.

Identification of potential keyword search terms and evidence outside the usual databases.

Causal pathways or evidence models were developed for key questions that were broad and could be conceptualized as having linkages between different sets of scientific evidence, for example, questions where there is a surrogate or intermediate outcome that may be related to a clinically relevant health outcome (Mulrow, Langhorne, and Grimshaw, 1997). These pathways are the framework for assessing the strength and amount of evidence linking different steps within a pathway.

To inform their discussion, the technical experts were provided with a handout (Table 2)containing summary data on spinal cord injury, multiple sclerosis, and stroke (although stroke was subsequently excluded by the expert panel from the scope of the evidence report). Prior to the panel meeting, these three conditions were deemed most likely to cause impairment that would lead to urinary tract dysfunction and to serve as the primary study populations for the report.

Table 2. Summary Data on Potential Target Conditions1.

Table

Table 2. Summary Data on Potential Target Conditions1.

Subject Definition

The technical experts agreed that management of urinary tract dysfunction is best directed by physiology rather than by specific cause. Thus, it would be most useful to consider UTIs in terms of the bladder physiology and urodynamic abnormalities, rather than in terms of the specific causes of the urinary tract problems (e.g., spinal cord injury, multiple sclerosis, etc.).

Because the type of bladder dysfunction caused by stroke is dissimilar to that due to spinal cord dysfunction, the panel recommended that the focus of the evidence report be on conditions resulting in spinal cord dysfunction and that the report exclude literature on stroke and disorders of the brain.

Identification of the Main Key Questions

The panel also agreed that prevention of recurrent infections is the clinician's most important responsibility and the area in which the biggest gap exists between best and usual care.

The panelists and project staff then reviewed and discussed the 11 proposed key questions originally developed by project staff and the onsite experts prior to the meeting (Table 3).

Table 3. Original List of Potential Key Questions1.

Table

Table 3. Original List of Potential Key Questions1.

As seen in Table 4 (Revised Key Questions), by discussion and consensus, the meeting participants agreed to do the following:

1.

Rewrite Question 3 to reflect the benefits, risks, and costs associated with drainage methods.

2.

Renumber Question 3 and combine it with Question 10.

3.

Fold Question 11 into Question 10.

4.

Divide Questions 4 and 5 into two questions each.

5.

Collapse Questions 6 and 7 into one question.

Table 4. Revised Key Questions1.

Table

Table 4. Revised Key Questions1.

The initial development of the original set of key questions and their subsequent revision were based on the panel members interpretation of the evidence known to them and their clinical judgment.

Ranking of Conditions

Using the approach of a modified Delphi procedure, each technical expert then anonymously ranked the top five key questions that he or she believed the evidence report should address. Rankings were based on the importance and the perceived availability of data to answer the question. As seen in Table 5, Questions 1, 2, 8, 9, and 10 received the highest rankings.

Table 5. Technical Experts Ranking of Key Questions1.

Table

Table 5. Technical Experts Ranking of Key Questions1.

A second round of votes among the five key questions confirmed the order of the first round of balloting. Question 1 was viewed as the most important, followed by 9, 2, 8, and 10. Table 6 contains the causal pathways or tables for the five key questions determined by the technical experts to be most promising. These causal pathways were developed by the experts as a group during the meeting, guided by senior project staff (Drs. Shekelle and Vickrey). A subset of these five key questions would be the basis for the subsequent literature search.

Table 6. Causal Pathways for Top-Ranked Key Questions.

Table

Table 6. Causal Pathways for Top-Ranked Key Questions.

Revision of Key Questions and Causal Pathways

Several weeks following the initial meeting, project staff circulated the summary report of the panel meeting and the figures/tables for the five key questions selected to the technical experts for comments, corrections, and other input. Their comments were collated and divided into several categories:

  • Comments on issues that were raised during the meeting
  • Comments and questions regarding the original list of key questions
  • Comments on the five key questions selected at the meeting for indepth literature review (Table 7)

Table 7. Technical Experts' Comments on Report of Panel Meeting.

Table

Table 7. Technical Experts' Comments on Report of Panel Meeting.

Comments took the form of additions, corrections, and wording clarification. Project staff incorporated the technical experts' comments into the meeting summary and then sent them the revised summary with a letter indicating that their comments and changes were included in the literature review strategy.

A preliminary search on Key Question 10 indicated that synthesizing the volume of available literature would be beyond the size and scope of this project. Project staff continued with literature searches on the four remaining key questions.

Literature Search

Databases

MEDLINE and EMBASE were the two databases selected for the literature search for this evidence report.

MEDLINE

The MEDLINE database corresponds to three print indexes: Index Medicus, Index to Dental Literature, and the International Nursing Index. The MEDLINE database covers journal articles spanning from 1966 to the present. MEDLINE is indexed using the National Library of Medicine's Medical Subject Headings (MeSH). More than 59 percent of the records added to MEDLINE after 1974 include abstracts taken directly from the published articles. Approximately 69 percent of the records added after 1985 contain abstracts. (Records added prior to 1975 do not contain abstracts.) Approximately 400,000 records are added each year, of which more than 85 percent are in English. MEDLINE indexes articles from over 3,700 journals published in more than 70 countries. It includes citations to chapters or articles from selected monographs from May 1976 through 1981.

EMBASE

The EMBASE database provides access to periodical articles from more than 2,900 primary journals from over 110 countries. It screens an additional 600 journals for drug articles. The database covers articles from 1974 to the present. EMBASE is recognized as an important, comprehensive index of the world's literature and often contains literature from foreign language journals that is not contained in other databases. Each record in EMBASE is classified and indexed by medical research specialists who assign terms and codes in accordance with EMTREE, a highly developed classification schedule and controlled vocabulary consisting of over 37,000 terms and 150,000 synonyms. Approximately 370,000 records are added annually, over 75 percent of which contain abstracts.

(A third database, CINAHL, which focuses on journals pertinent to nursing and allied paramedical professionals, was searched after review and comment on a draft report by the panel and peer reviewers. Results of that search are presented in a later section of this report.)

Panel Input

During the initial meeting, the panel members were invited to provide the staff research librarian with information on non-standard data sources and to suggest literature search strategies and key words relevant to the five selected key questions. Some of the terms suggested included the following:

  • Quality of life
  • Spastic bladder
  • Sphincterotomy
  • Hospitalization
  • Urodynamics
  • Bladder function
  • Bladder dysfunction
  • Dyssynergia or detrusor sphincter
  • Catheterization
  • Neurostimulation of bladder

Based on discussions at the technical experts' meeting, the research librarian performed MEDLINE and EMBASE searches for Key Question 1 (the symptoms, signs, and laboratory findings related to the risk of UTI in paralyzed people). "Urinary tract infection" was combined with "spinal cord injury" and "multiple sclerosis," and with selected symptoms (e.g., fever, malaise) and laboratory tests (e.g., leukocyte counts, microbiology techniques). This search yielded 178 journal articles (110, MEDLINE; 68 EMBASE, non-duplicated) (Table 8.)

Table 8. Literature Searches and Review of Titles and Abstracts.

Table

Table 8. Literature Searches and Review of Titles and Abstracts.

Based on a review of search results, "spinal cord injury" was broadened to include quadriplegia, paraplegia, neurogenic bladder, and neuropathic bladder (searched as a free- text term because it did not appear as a subject heading in either MEDLINE or EMBASE; yield=2,579: 1757 MEDLINE and 822 EMBASE). (A separate search using the term "UTI" alone yielded a total of 18,321 articles.) Subsequently, the term "bacteriuria" was added to the search strategy (yield=2,779; 1,881 MEDLINE and 898 EMBASE).

Final Search Strategy

Based on results of these literature searches, the research librarian was provided with the final strategy for all Key Questions (Tables 8 and 9). It was decided to: add the term "urinary tract" as well as "urinary tract infection," and exclude case reports and animal studies. The revised search was run for two time periods: 1966-84 and 1985-present. The yield for these final searches was a total of 4,998 articles (3,859, MEDLINE; 1,139, EMBASE).

Table 9. Literature Search Strategy.

Table

Table 9. Literature Search Strategy.

Literature Assessment

Literature Tracking

All reviewed articles were coded into EndNote, following the instructions shown in Table 10. Articles were coded by source (EMBASE, MEDLINE, technical expert, back search, bibliography), topic (spinal cord injury, multiple sclerosis), and RCT status. The status of article retrieval was also coded using the EndNote software.

Table 10. Instructions for Coding Articles in EndNote.

Table

Table 10. Instructions for Coding Articles in EndNote.

Title Reviews

The 4,998 titles identified by the final literature search were reviewed to identify articles for which full citations and abstracts should be requested. After systematic training by project staff, two physician reviewers independently reviewed the same 4,998 titles. One reviewer had expertise in rehabilitation medicine and the other had expertise in health services research. Agreement between them was checked and reconciled after an initial set of titles was reviewed. Any apparent systematic differences between the two reviewers' assessments were discussed in a meeting of the project team leaders until a consensus was reached on how to code titles in those kinds of situations. The reviewers then proceeded to complete their independent reviews of all 4,998 titles, then met to come to consensus on any titles for which they disagreed. Of the 4,998 titles, the two reviewers requested a total of 1,561 full citations and abstracts (1,072 of the 3,859 titles uniquely identified in the MEDLINE database; 489 of the 1,139 titles uniquely identified in the EMBASE database).

Abstract Reviews

The selected abstracts were reviewed to identify which full articles to pull for further review. Project staff developed abstract coding criteria for this purpose and then trained the physician reviewers in their use. Strict screening criteria and training were necessary to ensure the reliable (i.e., reproducible) review of abstracts. As seen in Table 11, articles were accepted if they potentially addressed at least one of the four highest-priority key questions (i.e., Key Questions 1, 2, 8, or 9). Only randomized controlled trials (RCTs) were accepted for Key Question #8 (Harms, Benefits, and Costs of the Long-term Use of Prophylactic Agents); non-RCTs were accepted for Key Questions 1, 2, and 9. If there was doubt as to whether the article represented an RCT, its RCT status was deferred until review of the full article.

Table 11. Key to Abstract Coding.

Table

Table 11. Key to Abstract Coding.

At the abstract review stage, project staff rejected articles for one of the seven following reasons:

1.

Non-relevance to a key question

2.

Study population less than 13 years of age

3.

Case report, editorial, or non-human population

4.

Report of a surrogate outcome

5.

Study population only included subjects within 90 days of a spinal cord injury (unless article addressed Key Question 8 on prophylaxis for UTI)

6.

Study was not an RCT and it only addressed Key Question 8 on prophylaxis for UTI

7.

Miscellaneous (reason had to be provided)

Abstractors were instructed to use the lowest number rejection code (e.g., if an article was both not relevant to a key question and was a case report, the Reject Code would be R1). Overview and review articles deemed relevant to the evidence report were coded to indicate they should be pulled and their citation lists reviewed.

After the first 76 abstracts were reviewed independently by each reviewer, a meeting was held with project leaders to review agreement. Of the first 76 abstracts, 11 actually had no abstract in the MEDLINE or EMBASE database and 7 were review articles; for 47 abstracts there was agreement between reviewers, and for 11 articles there was disagreement. Reasons for disagreement were discussed, and the reviewers proceeded with their independent assessments. Subsequently, an abstract was only rejected if rejected by both reviewers.

A total of 708 citations were rejected at the abstract stage. Of these, 394 (56 percent) were rejected because a review of their abstracts indicated that the citation met one of the seven rejection criteria listed above. In addition, another 190 articles that had no abstract (and were published between 1966 and 1974) and 124 review articles were rejected on the basis of their titles.

The articles for the remaining 831 citations were pulled and photocopied for full screening. (Table 8 indicates that 884 articles were relevant to a key question. This number is larger than 831 because 53 articles were classified as potentially relevant to two key questions.) Of these, 428 were pulled on the basis of title only because no abstract was available. An additional 22 overview articles were pulled so that their references could be reviewed.

Article Screening

Each article was screened independently by at least two persons using a form designed to estimate study quality (Table 12). The form recorded the relevant outcome(s) measured in the study, study design, sample size, and key question(s) addressed. The results were entered into an Excel spreadsheet for analysis. The spreadsheet was programmed to generate a list of articles about which the two reviewers disagreed to accept or reject them. Dr. Vickrey or Dr. Shekelle reviewed each article for which there was a discrepancy and made a final decision as to its status.

Table 12. Screening Form.

Table

Table 12. Screening Form.

If possible, foreign language articles with an English abstract were screened. However, some of these articles could not be classified without additional review by someone fluent in that language. Most of these were in French, German, Italian, and Japanese (Table 13).

Table 13. Screening of Foreign Language Articles Requiring Translator Assistance.

Table

Table 13. Screening of Foreign Language Articles Requiring Translator Assistance.

Articles were rejected at this phase for one of the five following reasons:

1.

Study population was primarily less than 13 years of age or included subjects without neurogenic bladder due to spinal cord dysfunction.

2.

No relevant outcome measures.

3.

The publication was a case report, letter, review, or the article reported a non-human population.

4.

The study did not address a key question.

5.

The study addressed Key Questions 2 or 9 but the article was published in 1978 or earlier.

Rejection criterion #5 was added after review of the first 60 articles and discussion with the onsite technical experts (Drs. Bennett and Dobkin at UCLA) indicated that the introduction and dissemination of intermittent catheterization in the early- to mid-1970s greatly changed the management of neurogenic bladder and risks for UTI. Thus, it was agreed that studies performed prior to 1978 were not relevant to current risks for recurrent UTI or to the issue of management to prevent long-term complications of neurogenic bladder.

Of the 853 full-length articles selected for screening, 509 (60 percent) were rejected. All but 3 out of the 128 foreign language articles requiring translation assistance were either screened by a methodologist fluent in that language, or reviewed by a member of the project team in a face-to-face meeting with someone fluent in that language. These 3 articles represent 0.4 percent of the total number of 853 articles identified for review and screening.

Only 35 articles (4 percent of total) could not be secured through the UCLA Interlibrary Loan service. On review of the 29 journals in which these articles were published, none appear to be widely circulated (Table 14). In addition, none of these journals was on a list of relevant journals having to do with urinary tract complications in spinal cord dysfunction provided by one of the panelists early in the project. This list identified periodicals reviewed for "Publications of Interest" in the Journal of Rehabilitation Research and Development (1997;34(4):515-6).

Table 14. Distribution by Journal Title of Articles That Could Not Be Obtained.

Table

Table 14. Distribution by Journal Title of Articles That Could Not Be Obtained.

Of the 306 articles definitely accepted for further study, the breakdown by key question is provided in Table 15.

Table 15. Results of Screening Full-Length Articles.

Table

Table 15. Results of Screening Full-Length Articles.

Case series study designs (63 articles) were subsequently excluded from further consideration for Key Question 2 because risk factors could not adequately be determined from a case series.

Data Extraction

Each study that passed the screening stage was reviewed by the senior project staff (Drs. Vickrey, Shekelle, and Morton). Senior project staff assessed clinical trials for quality using criteria developed by Jadad, which measures three study domains: randomization, double-blind method, and the handling of withdrawals. The Jadad assessment tool produces a 0 to 5 scale, which has been shown to discriminate between trials based on their effect size (Moher, Pham, Jones et al., 1998). Project staff assessed cohort studies for (1) the degree to which groups were comparable at baseline or to which adjustments were made in the statistical analysis, and (2) whether there was masking in the measurement of risk factors and outcomes.

Process for Review and Comment on Draft Evidence Report

In late June 1998 a draft of the evidence report was circulated to all of the technical experts who had met in December 1997. In addition, seven peer reviewers were identified. These included three technical experts who were nominated to the expert panel but were unable to attend the meeting, one managed care organization medical director, and three methodologists (Table 16). Information and questions for all reviewers to consider in reviewing the draft were provided (Table 17). Tabulation of reviewer comments and responses to questions and how they were addressed by the project team is provided in Appendix I of this report.

Table 16. Peer Reviewers.

Table

Table 16. Peer Reviewers.

Table 17. Instructions for Reviewing Draft Evidence Report.

Table

Table 17. Instructions for Reviewing Draft Evidence Report.

The section of the evidence report on the key question of signs, symptoms, and laboratory findings associated with risks to persons with spinal cord dysfunction was circulated to the panel and to peer reviewers in a later mailing. Comments and the center's responses to suggestions are in Appendix II.

In addition, two experts in meta-analysis (Vic Hasselblad, PhD, and Allan R. Sampson, PhD) reviewed and provided written comments on that section of the report.(Table 16; prophylaxis of UTI)

Search of CINAHL Database

After we received comments from reviewers, we conducted a literature search of CINAHL, a database for nursing and allied paramedical professions. The research strategy was:

URINARY TRACT INFECTION- or BACTERIURIA or PYURIA
and
SPINAL CORD in the same paragraph as INJUR- or PARAPLEGI- or
QUADRIPLEG- or MULTIPLE SCLEROSIS
or
NEUROGENIC BLADDER- or NEUROPATH- in the same sentence as BLADDER-
("-" means truncation)
The years of coverage were 1982 to the present.

This search yielded 255 titles. These were reviewed by a project investigator, who selected 31 titles for further consideration and review of abstracts. Eight studies were subsequently rejected after review of their abstracts: five were review articles, two had no relevant outcome, and one targeted a pediatric population.

Of the 23 remaining articles, 20 were on our original list of articles identified by searches of MEDLINE and EMBASE. Thus, a total of three articles were identified through the CINAHL search and not from other searches. (Note: one of those three articles was published in February 1998, after the MEDLINE and EMBASE searches were carried out.)

Of the three articles, one could not be obtained, one was a controlled trial of clean versus sterile intermittent catheterization and was entered into an evidence table and text for the key question on risk factors for recurrent UTI, and the third was accepted for that key question as well but excluded from evidence tables. (The reason for exclusion was that it compared spinal cord injury subjects having a UTI due to a certain bacterium with subjects having UTIs due to all other bacteria; however, there was no group without a UTI.)

PubReader format: click here to try

Views

  • PubReader
  • Print View
  • Cite this Page

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...