Quality of care in for-profit and not-for-profit nursing homes: systematic review and meta-analysis

BMJ 2009; 339 doi: http://dx.doi.org/10.1136/bmj.b2732 (Published 4 August 2009)
Cite this as: BMJ 2009;339:b2732
  1. Vikram R Comondore, resident1,
  2. P J Devereaux, associate professor2,
  3. Qi Zhou, statistician2,
  4. Samuel B Stone, resident3,
  5. Jason W Busse, research associate2, scientist4,
  6. Nikila C Ravindran, resident5,
  7. Karen E Burns, staff physician67,
  8. Ted Haines, associate professor2,
  9. Bernadette Stringer, assistant professor2,
  10. Deborah J Cook, professor2,
  11. Stephen D Walter, professor2,
  12. Terrence Sullivan, president and CEO8,
  13. Otavio Berwanger, professor9,
  14. Mohit Bhandari, associate professor2,
  15. Sarfaraz Banglawala, resident3,
  16. John N Lavis, associate professor2,
  17. Brad Petrisor, assistant professor3,
  18. Holger Schünemann, professor210,
  19. Katie Walsh, summer research assistant2,
  20. Neera Bhatnagar, reference librarian11,
  21. Gordon H Guyatt, professor2
  1. 1Department of Medicine, University of British Columbia, Vancouver, BC, Canada V5Z 1M9
  2. 2Department of Clinical Epidemiology and Biostatistics, McMaster University, Hamilton, ON, Canada L8N 3Z5
  3. 3Department of Surgery, McMaster University
  4. 4The Institute for Work and Health, Toronto, ON, Canada M5G 2E9
  5. 5Department of Medicine, Division of Gastroenterology, University of Toronto, Toronto, M5T 2S8
  6. 6St Michael’s Hospital, Toronto, M5B 1W8
  7. 7Keenan Research Centre and Li Ka Shing Knowledge Institute, Toronto, M5B 1W8
  8. 8Cancer Care Ontario, Toronto, M5G 2L7
  9. 9Department of Clinical Epidemiology, Federal University of Rio Grande do Sul, Porto Alegre-RS, Brazil
  10. 10Italian National Cancer Institute Regina Elena, Rome, 00144, Italy
  11. 11Health Sciences Library, McMaster University
  1. Correspondence to: P J Devereaux philipj{at}mcmaster.ca
  • Accepted 21 April 2009

Abstract

Objective To compare quality of care in for-profit and not-for-profit nursing homes.

Design Systematic review and meta-analysis of observational studies and randomised controlled trials investigating quality of care in for-profit versus not-for-profit nursing homes.

Results A comprehensive search yielded 8827 citations, of which 956 were judged appropriate for full text review. Study characteristics and results of 82 articles that met inclusion criteria were summarised, and results for the four most frequently reported quality measures were pooled. Included studies reported results dating from 1965 to 2003. In 40 studies, all statistically significant comparisons (P<0.05) favoured not-for-profit facilities; in three studies, all statistically significant comparisons favoured for-profit facilities, and the remaining studies had less consistent findings. Meta-analyses suggested that not-for-profit facilities delivered higher quality care than did for-profit facilities for two of the four most frequently reported quality measures: more or higher quality staffing (ratio of effect 1.11, 95% confidence interval 1.07 to 1.14, P<0.001) and lower pressure ulcer prevalence (odds ratio 0.91, 95% confidence interval 0.83 to 0.98, P=0.02). Non-significant results favouring not-for-profit homes were found for the two other most frequently used measures: physical restraint use (odds ratio 0.93, 0.82 to 1.05, P=0.25) and fewer deficiencies in governmental regulatory assessments (ratio of effect 0.90, 0.78 to 1.04, P=0.17).

Conclusions This systematic review and meta-analysis of the evidence suggests that, on average, not-for-profit nursing homes deliver higher quality care than do for-profit nursing homes. Many factors may, however, influence this relation in the case of individual institutions.

Introduction

Nursing homes provide long term housing, support, and 24 hour nursing care for people who are unable to function independently. Conservative forecasts from the European Union suggest that the need for nursing home care will double in the next 40 years as the population ages.1 Many nursing home residents are bound to the routines, diets, and treatments prescribed by the home where they reside. In addition, many of them are unable to advocate for themselves because of physical, medical, cognitive, or financial limitations.

Concerns about quality of care in nursing homes are widespread among academic investigators,2 3 4 5 the lay press,6 7 8 9 10 11 and policy makers.1 12 Whether a facility is owned by a for-profit or a not-for-profit organisation may affect structure, process, and outcome determinants of quality of care. In the United States, for example, two thirds of nursing homes are investor owned, for-profit institutions; in the United Kingdom, more than half of healthcare beds belong to independent nursing homes for older people, most of which are operated by for-profit institutions.13 The type of ownership of nursing homes in Europe varies; countries with previously dominant public healthcare systems (such as Poland) now seek privatisation.14 In Canada, 52% of nursing homes are in for-profit ownership, and not-for-profit care is evenly split between charitable or privately owned not-for-profit facilities and government or publicly owned not-for-profit facilities.15 Both for-profit and not-for-profit nursing homes may have both public and private funding.

Several investigators have assessed the relation between for-profit/not-for-profit status and quality of care.16 If quality or appropriateness of care varies significantly by ownership, this should influence government policies related to regulatory assessments and the use of public funds for nursing homes. The objective of this systematic review and meta-analysis was to examine the quality of care in for-profit and not-for-profit (privately and publicly owned) nursing homes to enhance the evidence base for public policy. This work is part of our series of systematic reviews comparing health outcomes, quality and appropriateness of care, and payment for care in for-profit and not-for-profit care delivery institutions.17 18 19

Methods

Search strategy

We used a multimodal search strategy focused on 18 bibliographical databases, personal files, consultation with experts, reviews of references of eligible articles, and searches of PubMed (for related articles) and SciSearch (for articles citing key publications).

A medical librarian (NB) used medical subject heading terms and keywords from a preliminary search to develop database search strategies. In each database, the librarian used an iterative process to refine the search strategy through testing several search terms and incorporating new search terms as new relevant citations were identified. The search included the following databases from inception to April 2006: Medline, Embase, HealthSTAR, CINAHL, Cochrane Database of Systematic Reviews, Database of Abstracts of Reviews of Effects, Cochrane Central Database of Controlled Trials, NHS Economic Evaluation Database, AgeLine, Web of Science, Proquest Dissertations and Theses, ABI/INFORM Global, CBCA Reference, EconLit, Proquest European Business, PAIS International, and Worldwide Political Science Abstracts. Search terms included nursing home specific terms (such as nursing homes, homes for the aged, long-term care) combined with ownership terms (such as proprietary, investor, for-profit, and competition). The web appendix gives a complete description of our database search strategies.

Study selection

Eligibility criteria

Our inclusion criteria were as follows: patients—those residing in nursing homes in any jurisdiction; intervention—for-profit status of the institutions; comparator—not-for-profit status; and outcomes—measures of quality of care in for-profit and not-for-profit nursing homes.

Definition of quality of care

As described by the American Medical Association, quality of care is “care that consistently contributes to the improvement or maintenance of quality and/or duration of life.”20 Quality of care was conceptualised by Donebedian as having inter-related structure, process, and outcome components.21 Structure pertains to resources used in care (such as staffing). Process refers to action on the patient (such as use of restraint and urethral catheterisation). Outcome indicators assess the patient’s end result (such as pressure ulcers). Many quality of care instruments have been proposed, although none has been universally accepted.22 Consequently, we used measures that authors defined as representing “quality of care” or “appropriateness of care,” provided that they defined a priori what constituted “good” or “poor” quality of care. The most frequently used quality measures were as follows.

Number of staff per resident or level of training of staff—The US Medicare/Medicaid nursing home regulations emphasise the importance of this measure of structure.23 Studies have consistently shown a positive association between staffing and measures of both process and outcome quality.24 25 26

Physical restraints—Although use of physical restraints can prevent patients from injuring themselves, restraints diminish a patient’s self esteem and dignity. By restricting mobility, they lead to both physical deterioration and the formation of painful pressure ulcers.24 27 An Institute of Medicine report emphasised use of restraints as an important process measure.23

Pressure ulcers—The importance of this outcome quality measure was also stressed by the Institute of Medicine. Pressure ulcers are preventable and are associated with pain and infection risk.23

Regulatory (government survey) deficiencies—Deficiency citations by a regulatory body cover many aspects of nursing home care. Their strength lies in providing an overall measure of quality. Considerable work has gone into developing valid overall deficiency measures.4

Definition of nursing home

In keeping with other definitions,28 we defined a nursing home as a home for elderly people in which most residents require daily nursing care. We included all long term care facilities that met this definition, including those studies that specifically evaluated “skilled nursing facilities” and special care facilities such as those for patients with Alzheimer’s disease.

Assessment of study eligibility

Teams of two reviewers independently screened the titles and abstracts of all citations identified in our search, and if either reviewer thought that a citation might be eligible we retrieved the study for full text review. Research personnel who were not involved in the screening or data abstraction process masked the study results from the text and tables of potentially eligible articles by using a black marker. Teams of two reviewers independently evaluated each masked article to determine eligibility. All disagreements were resolved by consensus, with discussions with the project lead (VRC) about eligibility criteria as required. In the event of ambiguity about whether the outcome was a measure of quality of care, we erred on the side of being inclusive.

Data extraction and study quality evaluation

Multiple teams of two reviewers independently abstracted data from included articles. We collected data on geographical area, year, data source, unit of measurement (number of residents or nursing homes), and quality of care measure. We developed and applied a 0-5 scale for evaluating appropriate adjustments and a yes/no scale for inappropriate adjustments (box). We explored whether appropriate and inappropriate adjustment explained heterogeneity. Disagreements were resolved by consensus, with consultation of a third investigator when resolution could not be achieved.

Evaluation of quality of studies used in meta-analyses: appropriate and inappropriate adjustments

Appropriate adjustments (0-5)

One point for each of:

  • Having an adjusted analysis

  • Adjusting for age

  • Adjusting for severity of illness (comorbidities)

  • Adjusting for presence or absence or severity of dementia

  • Adjusting for payment status of residents (government funded v privately funded)

Inappropriate adjustments (yes/no)

Yes for adjusting for potential quality of care measures (that is, elements used to assess quality of care in a different study, such as pressure ulcer, restraint use, urinary catheterisation, staffing, or regulatory agency citations)

Statistical analysis

Many studies had for-profit versus not-for-profit comparisons including multiple measures of quality of care. When summarising results, we classified studies into three categories. (1) “All statistically significant differences favoured one ownership type”—studies fulfilled two requirements: at least one outcome with P<0.05 favoured either for-profit or not-for-profit and all outcomes with P<0.05 favoured the same funding structure (that is, all favour not-for-profit or all favour for-profit). (2) “Most but not all significant differences favoured one ownership type”—studies fulfilled two requirements: at least four quality measures had P<0.05 and three times as many outcomes with P<0.05 favour one ownership as favour the other. (3) “Mixed results”—all other results.

We pooled outcomes by using random effects models separately for the most frequently used quality of care measures: number of staff or level of training of staff, pressure ulcers, physical restraints, and regulatory (government survey) deficiencies. We considered P<0.05 to be statistically significant.

We used prevalence, rather than incidence, in reporting physical restraint use and pressure ulcers based on authors’ reporting of study outcomes. We report the odds ratios and 95% confidence intervals for these outcomes. When necessary, we converted other effect measures to odds ratios by using available data. For example, if the study reported a relative risk (RR) and the event proportion in for-profit nursing homes (Pfp), the odds ratios was calculated as (RR×(1− Pfp))/(1−Pfp×RR). Similarly, when the studies presented a β coefficient (an adjusted result representing difference in event proportions in for-profit and not-for-profit nursing homes, Pfp−Pnfp), if the event proportion (Pc) in the study population and sample sizes (Nfp and Nnfp) of the nursing homes in for-profit and not-for-profit were provided, solving the following two equations for Pnfp and Pfp, we computed the odds ratio: Pfp−Pnfp=β and (Pfp×Nfp+Pnfp×Nnfp)/(Nfp+Nnfp)=Pc. For the outcomes of deficiencies and staffing, we used the ratio of the effect from not-for-profit to for-profit nursing homes in pooling studies.

We avoided repetition of data on the same resident from different studies by preferentially using data from the larger dataset when necessary. One author (GHG) made these decisions by using blinded copies of articles while unaware of study outcomes. We requested supplemental data when available data was insufficient for analysis. We evaluated heterogeneity with both a χ2 test and the I2 statistic, interpreting a low I2 as 25% or lower and a high I2 as 75% or higher.29 We examined funnel plots for evidence of publication bias. We applied a univariate meta-regression random effects model to each pooled outcome to evaluate potential sources of heterogeneity.

Hypotheses to explain heterogeneity

Our a priori hypotheses for sources of potential heterogeneity included analysis of privately owned and publicly owned nursing facilities in the same category, appropriate and inappropriate adjustments, the year of data collection, geography and political environment, and primary compared with secondary data collection. We did univariate meta-regression for each potential cause of heterogeneity. We present subgroup results if the likelihood of the differences between subgroups being due to chance was P<0.10. Our a priori hypotheses to explain heterogeneity are detailed below.

Analysing privately and publicly-owned not-for-profit facilities in the same category—We hypothesised that privately owned not-for-profit facilities may deliver superior care compared with publicly owned facilities, and thus comparisons between not-for-profit and for-profit facilities may yield different results if publicly owned facilities are included, as seen in previous studies.19 We decided, a priori, to present the result of a for-profit versus privately owned not-for-profit meta-analysis separately from a for-profit versus not-for-profit meta-analysis regardless of whether privately or publicly owned not-for-profit status explained heterogeneity of the pooled estimate.

Extent of appropriate and inappropriate adjustment—We have defined concepts of appropriate and inappropriate adjustment in the data extraction section above. We compared studies with above median scores against those with scores below the median for assessment of appropriateness. Similarly, we compared studies with inappropriate adjustment against those without inappropriate adjustment, excluding studies that did not have an adjusted analysis.

Year of data collection—Legislation on quality of care in nursing homes was introduced in the United States under the Federal Nursing Home Reform Act (part of Omnibus Budget Reconciliation, 1987). Most of the studies we reviewed were from the United States. As a result, we compared data collected before and during 1987 versus after 1987.

Geography—We compared data collected inside and outside the United States, as geography and political environment are potential sources of heterogeneity.

Primary versus secondary data collection—We compared data acquired by primary (direct) data collection with those acquired by secondary (administrative agency) data collection.

Results

Of the 8827 articles screened, we selected 956 for blinded full text review. Figure 1 details the steps in this review. Our agreement on the eligibility of studies was very good (κ=0.73 on the basis of two questions: does the study evaluate nursing homes, and does the study compare quality of care in for-profit and not-for-profit facilities?). Disagreements stemmed from implied but not stated definitions in the articles regarding good and poor quality and implied but not stated quality of care measures. We requested supplementary data from 36 authors; 25 responded, of whom three did new analyses in response to our queries.

Fig 1 Flow chart of steps in systematic review

We found 82 studies, spanning 1965 to 2003, comparing for-profit and not-for-profit nursing homes.w1-w82 We found 40 studies in which all statistically significant analyses (P<0.05) favoured not-for-profit homes and three in which all statistically significant analyses favoured for-profit homes. Similarly, 34 studies compared for-profit and privately owned not-for-profit nursing homes. In 16 of these, all statistically significant comparisons favoured higher quality in privately owned not-for-profit homes; none had all statistically significant analyses favouring higher quality in for-profit homes.

Tables 1 and 2 present a summary of the characteristics and outcomes of all studies included in this review and summarise the results of comparisons for quality measures evaluated by three or more studies. Tables 3 and 4 present the detailed study characteristics and outcomes of those studies that compared for-profit and privately owned not-for-profit facilities. Similarly, tables 5 and 6 present the detailed study characteristics and outcomes of studies that compared for-profit and not-for-profit (publicly and privately owned) facilities.

View this table:
Table 1

 Number of studies with quality of care comparisons favouring particular ownerships*: overall and staffing results

View this table:
Table 2

 Number of studies with quality of care comparisons favouring particular ownerships: other results*

View this table:
Table 3

 Characteristics of studies comparing private for-profit and private not-for-profit nursing home quality of care

View this table:
Table 4

 Quality of care measures and outcomes of studies comparing private for-profit and private not-for-profit nursing homes (favoured directions represent those with higher quality care)

View this table:
Table 5

 Characteristics of studies comparing for-profit and not-for-profit nursing home quality of care (public and private NFP homes)

View this table:
Table 6

 Quality of care measures and outcomes of studies comparing for-profit and not-for-profit nursing homes (public and private NFP homes): favoured directions represent those with higher quality care

We meta-analysed data for the four most commonly used quality measures. Table 7 presents a summary of the characteristics of studies meta-analysed, along with the results of sensitivity analyses to explain heterogeneity among studies in each meta-analysis. Two meta-analyses showed statistically significant results favouring higher quality care in not-for-profit nursing homes.

View this table:
Table 7

 Results of testing of a priori hypotheses to explain heterogeneity

We found more or higher quality staffing in not-for-profit homes (ratio of effect 1.11, 95% confidence interval 1.07 to 1.14, P<0.001, I2=91.6%) (fig 2). We found a similar result favouring not-for-profit homes when assessing staffing hours alone, with a ratio of effect of 1.11 (1.08 to 1.14, P<0.001, I2=70.3%), an absolute hours increase of 0.42 (0.31 to 0.53) hours/resident/bed/day, and a relative hours increase of 11% (8% to 14%). When the only non-US study was excluded, we arrived at a similar ratio of effect for more or higher quality staffing in not-for-profit homes of 1.11 (1.07 to 1.15, P<0.001, I2=92.4%).

Fig 2 Ratio of effect sizes for staffing quality in for-profit (FP) and not-for-profit (NFP) nursing homes. Ratios listed represent effect size in NFP homes compared with that in FP homes. Ratio >1 indicates that NFP homes had more, or higher quality, staffing (that is, favours NFP)

We found a lower prevalence of pressure ulcers in not-for-profit homes (odds ratio 0.91, 95% confidence interval 0.83 to 0.98, P=0.02, I2=52.1%), with an absolute risk reduction of 0.59% (0.13% to 1.12%) and a relative risk reduction of 8.4% (1.9% to 16%) (fig 3). When the only non-US study was excluded, we arrived at a similar odds ratio favouring lower pressure ulcer prevalence in not-for-profit homes of 0.89 (0.82 to 0.97, P=0.007, I2=50.2%).

Fig 3 Odds ratios (OR) comparing pressure ulcer prevalence in for-profit (FP) and not-for-profit (NFP) nursing homes. OR <1 indicates lower risk of pressure ulcers in NFP facilities than in FP facilities, suggesting that NFP facilities deliver higher quality care

The remaining two meta-analyses showed non-statistically significant differences. We found less use of physical restraints in not-for-profit homes (odds ratio 0.93, 0.82 to 1.05, P=0.25, I2=74.6%) (fig 4) and fewer deficiencies in governmental regulatory assessments in not-for-profit homes (ratio of effect 0.90, 0.78 to 1.04, P=0.17, I2=59.8) (fig 5).

Fig 4 Odds ratios (OR) comparing physical restraint prevalence in for-profit (FP) and not-for-profit (NFP) nursing homes. OR <1 represents less physical restraint use in NFP facilities than FP facilities, suggesting that NFP facilities deliver higher quality care

Fig 5 Ratio of effect sizes for regulatory deficiencies in for-profit (FP) and not-for-profit (NFP) nursing homes. Ratios listed represent effect size in NFP facilities compared with that in FP facilities. Ratio <1 represents fewer deficiencies in NFP homes, suggesting that NFP homes deliver higher quality care

Funnel plots for the four meta-analyses did not suggest publication bias. A priori hypotheses did not explain the observed heterogeneity (table 7).

Discussion

Our systematic review identified 82 studies comparing quality of care in for-profit and not-for-profit nursing homes. More studies had all statistically significant analyses showing higher quality in not-for-profit nursing homes than in for-profit nursing homes. Many studies, however, showed no significant differences in quality by ownership, and a small number showed statistically significant differences in favour of for-profit homes. This pattern held true when we compared for-profit homes with both privately owned and publicly owned not-for-profit facilities. Pooled analyses of the four most commonly used quality measures showed statistically significant results favouring higher quality care in not-for-profit homes for staffing and prevalence of pressure ulcers and non-statistically significant differences favouring not-for-profit homes in physical restraint use and regulatory agency deficiencies. The large observed heterogeneity was not explained by our a priori hypotheses.

Previous systematic reviews

Two previous systematic reviews have compared quality of care in for-profit and not-for-profit nursing homes. In 1991 Davis and colleagues found that many studies showed that higher quality of care was provided in not-for-profit nursing homes; however, weaknesses in the methodological design of the included studies limited the conclusions that could be drawn.30 In 2002 Hillmer and colleagues did a systematic review comparing for-profit and not-for-profit facilities (including publicly owned facilities), focusing on studies in North America completed after the previous review.31 This study also concluded that not-for-profit facilities provided better quality care than for-profit facilities.

Strengths and weaknesses of this review

We did a comprehensive search, which identified 60 studies not included in previous reviews. We assessed studies spanning four decades and published in any language. We masked study results before determining eligibility and did duplicate citation screening, data abstraction, and quality assessment. We contacted authors for missing data and received responses from most of them. We compared quality of care in both for-profit versus not-for-profit nursing homes and for-profit versus privately owned not-for-profit nursing homes, did pooled analyses of quality of care measures, and found largely consistent results.

Our review has limitations resulting from the characteristics of the studies included. No randomised trials have compared quality of care across nursing home ownership, and no such trials are ever likely to be done. Furthermore, most studies are from the United States, which raises questions of generalisability to other jurisdictions.

Studies are also limited in that no standard definition of quality of care exists. The result is that studies used a very wide variety of alternative measures of quality. Even when the same measures were used, standardised approaches to the application of those measures were lacking. For example, meta-analysis for number and qualifications of staff fails to take into account staff turnover, the use of agency staff, and the professional mix of staff.25

Moreover, several eligible studies used administrative databases, which further limits the comprehensiveness and quality of the data. For example, the American Online Survey Certification and Reporting (OSCAR) database comprises self reported data from nursing home administrators; surveyors verify only a sample. Careful duplicate abstraction of data from patients’ charts with a priori definitions or, ideally, direct assessment of care provision would be preferable.

Our meta-analyses are limited in that many authors could not remove publicly owned facilities from their datasets for our for-profit versus privately owned not-for profit analysis. However, in our sensitivity analyses, results comparing for-profit and not-for-profit facilities were not significantly different from those in which we restricted the not-for-profit facilities to those for which we could confirm ownership.

Heterogeneity

On the one hand, one might see our results as compellingly favouring not-for-profit facilities. The gradient between studies in which all significant measures favoured not-for-profit (40 studies) and those in which all measures favoured for-profit (3) is large (table 1). All four meta-analyses favoured not-for-profit institutions, and two reached statistical significance.

On the other hand, 37 studies had mixed results (some measures favoured for-profit, some not-for-profit) and considerable heterogeneity was present in the results of the meta-analyses. This suggests that although the average effect is clear, that effect probably varies substantially across situations. The variability is probably explained, in part, by a variety of factors that vary within categories of for-profit and not-for-profit homes, including management styles, motivations, and organisational behaviour. For example, for-profit facilities owned and operated by investor owned corporations may have different motivations than facilities owned by small private businesses or single proprietors. Not-for-profit facilities run by charities might differ in structure and process from those run by municipalities; not-for-profit facilities that are managed by for-profit nursing home companies may function differently from those that are not.

We have partially mitigated this problem with our a priori hypotheses (extent of appropriate adjustments, year of data collection, geography and political environment, primary compared with secondary data collection, and, in particular, public versus private ownership of not-for-profit facilities). None of these variables, however, explained the substantial heterogeneity of our results. The studies failed to specify characteristics of individual nursing homes in sufficient detail to allow analyses exploring factors such as those listed above (ownership by corporation, small business, charitable organisation of municipality; management of not-for-profit homes by for-profit providers).

Significance of this study

Most of the studies in our systematic review showed lower quality of care in for-profit nursing homes than in not-for-profit nursing homes. However, a large proportion of studies showed no significant difference in quality of care by ownership. In the long term care market, in which funding is often provided by the government at fixed rates, both for-profit and not-for-profit facilities face an economic challenge that may affect staffing and other determinants of quality of care. In the for-profit context, however, shareholders expect 10-15% returns on their investments,32 taxes may account for 5-6% of expenses, and facilities tend to have higher executive salaries and bonuses, so for-profit facilities have a strong incentive to minimise expenditures.33 Minimising expenditures may lead to lower quality staffing and higher rates of adverse events (such as pressure ulcers), which may be reflected in citations for deficiency.

Proving causality by using observational studies is difficult. Furthermore, given their variability, the results do not imply a blanket judgment of all institutions. Some for-profit institutions may provide excellent quality care, whereas some not-for-profit institutions may provide inferior quality of care.

Our findings are, however, consistent with findings of higher risk adjusted death rates in for-profit hospitals and dialysis facilities as shown in previous reviews,18 19 as well as providing insight into average effects. Given the absolute risk reduction in pressure ulcers of 0.59%, we can estimate that pressure ulcers in 600 of 7000 residents with pressure ulcers in Canada and 7000 of 80 000 residents with pressure ulcers in the United States are attributable to for-profit ownership. Similarly, given an absolute increase in nursing hours of 0.42 hours per resident per bed per day, we can estimate that residents in Canada would receive roughly 42 000 more hours of nursing care a day and those in the United States would receive 500 000 more hours of nursing care a day if not-for-profit institutions provided all nursing home care. These estimates are based on the 2006 census from Canada showing that 100 740 of 252 561 nursing home residents resided in for-profit nursing homes and the 2000 census from the United States showing a total of 1 720 500 nursing home residents.34 35 These estimates assume that two thirds of US nursing home residents live in for-profit facilities.

Further research and conclusions

Although this review has fully assessed the data available comparing for-profit and not-for-profit nursing home care, additional work is needed to compare the costs between these types of facilities and to evaluate the consistency of these findings outside of the United States and Canada. Although we have extensively evaluated the literature comparing quality of care in for-profit, charitable organisation owned, and government owned nursing homes, the available studies did not allow comparison of the possible impact of factors such as subcategory of for-profit ownership (for example, chain v non-chain, investor v small business ownership, municipality v federal government ownership). Nursing home management companies further complicate the relation between ownership and quality of care. These are all important areas that warrant further research.

What is already known on this topic

  • The quality and appropriateness of care delivered in nursing homes is a major concern for the public, policy makers, and media

  • Controversy exists about whether for-profit compared with not-for-profit ownership affects quality of care

What this study adds

  • Most studies suggest a trend towards higher quality care in not-for-profit facilities than in for-profit homes, but a large proportion of studies show no significant trend

Notes

Cite this as: BMJ 2009;339:b2732

Footnotes

  • We acknowledge the outstanding work of Deborah Maddock, Denise Healey, Shelley Anderson, Michelle Murray, Monica Owen, and Laurel Grainger who coordinated this study. We thank our foreign article reviewers Janek Brozek, Matthias Briel, Toshi Furukawa, Marjuka Makela, Ben de Mol, Paola Muti, Patricia Smith, Kristian Thorlund, and David Wei. We appreciate the work of Dana Keilty, Navneet Binepal, Tony Soeyonggo, and Minji Kim, who blinded articles for us. We thank Christina Lacchetti, Michael Levy, and Rajesh Hiralal, who reviewed articles for us, and Diane Heels-Ansdell for her statistical help. We also thank the authors of included studies who did additional analyses for our systematic review: Chappin White, Robert Weech-Maldonado, and Ann L. Gruber-Baldini.

  • Contributors: VRC, PJD, GHG, and TS conceived and designed the study. OB, MB, NB, VRC, DJC, PJD, GHG, SB, JWB, KEB, TH, JNL, BP, NCR, HS, BS, SBS, QZ, and KW were involved in data acquisition. VRC, GHG, QZ, and PJD analysed and interpreted the data. VRC drafted the manuscript. GHG, PJD, and QZ critically revised the manuscript for important intellectual content. QZ and SDW provided statistical expertise. VRC, PJD, and GHG are the guarantors.

  • Funding: Atkinson Foundation Grant; the study sponsor did not contribute to the study design. JWB is funded by a Canadian Institutes of Health research fellowship award. DJC, MB, and JNL are supported, in part, by their respective Canada Research chairs. PJD is supported by a Canadian Institutes of Health Research new investigator award. HS is funded by a European Commission: The Human Factor, Mobility and Marie Curie Actions scientist reintegration grant (IGR 42192).

  • Competing interests: None declared.

  • Ethical approval: Not needed.

This is an open-access article distributed under the terms of the Creative Commons Attribution Non-commercial License, which permits use, distribution, and reproduction in any medium, provided the original work is properly cited, the use is non commercial and is otherwise in compliance with the license. See: http://creativecommons.org/licenses/by-nc/2.0/ and http://creativecommons.org/licenses/by-nc/2.0/legalcode.

References

THIS WEEK'S POLL