METHODS

Identification and selection of the literature

Relevant publications were searched using the Cochrane library (1993-April 2000), Medline (1966-April 2000), and Embase (1980-April 2000) databases. The following keywords were used: (hip and (arthritis or arthrosis or osteoarthritis or osteoarthrosis) or coxarthrosis) and (risk factor or causative or determinants or predictor or aetiology) and (case-control or retrospective or prospective or longitudinal or follow up or cohort). (A detailed list can be obtained from the corresponding author.) We extended the search by screening the reference lists of all relevant articles identified.

A study was eligible for inclusion if it fulfilled all of the following criteria: (a) one of the aims of the study was to investigate an association between HD and hip OA; (b) the articles were written in English, Dutch, German, French, Danish, Norwegian, or Swedish; (c) the article was a full text article; (d) the patients in the studies had radiological and/or clinical hip OA, a (total) hip replacement, or were on the waiting list for one; and (e) the study design was a cohort or a case-control study.

A study was excluded if the studied population had a specific underlying pathology, such as trauma (fractures), infection, rheumatoid arthritis, ankylosing spondylitis, Perthes’ disease, tuberculosis, haemochromatosis, sickle cell disease, Cushing’s disease, or femoral head necrosis.

Methodological quality assessment

The variation of the methodological quality of observational studies may influence the results and conclusions of a systematic review. Therefore, the quality of each study included was assessed by the following procedure.

Two reviewers independently scored the quality of the selected papers according to a standardised set of criteria (table 1). These criteria have been used in previous reviews of observational studies in the field of musculoskeletal disorders and were modified to cover the topic of our review.^10–13 The criteria concern both the internal validity and the informativeness of the study. Only items reflecting the internal validity of the studies were used to assess the methodological quality (table 1 and appendix W1, available at http://www.annrheumdis.com/supplemental).

In cases of disagreement, both reviewers tried to achieve consensus; if disagreements were not resolved, a third reviewer was consulted to achieve a final judgment.

Several items are not applicable to a certain type of study design (for example, cohort study or case-control study), and therefore do not contribute to the total score of that particular study. This means that the maximum score of each study ( = 100%) was based only on the items applicable to that particular type of study design. Positive scores were summed to give an overall internal validity score.

Best evidence synthesis

Because the study population, methodological quality and determinants, and outcome measures for hip OA are heterogeneous in observational studies, we refrained from statistically pooling the data,¹⁴ and performed a “best evidence synthesis”.^{10,11,14–17}

The studies were stratified according to the type of study design. To clarify the question whether HD pre-dates hip OA, the only studies which can really answer this question are prospective cohort studies: that is they follow up patients without hip OA with the presence or absence of HD measured, forward in time, and identify who develops hip OA subsequently. Therefore these studies were judged to have the most valid design. Studies with cross sectional data on hip OA and HD were not considered in the best evidence synthesis.

Furthermore, the studies were ranked according to their methodological quality score^10,11,18:

• Strong evidence is provided by generally consistent findings in multiple high quality cohort studies

• Moderate evidence is provided by generally consistent findings

  • – In one high quality cohort study and two or more high quality case-control studies

  • – In three or more high quality case-control studies

• Limited evidence is provided by (generally consistent) findings

  • – In a single cohort study

  • – In ⩽2 case-control studies

• Conflicting evidence is provided by conflicting findings (that is, <75% of the studies reported consistent findings)

• No evidence is provided when no studies can be found.

A study was considered to be of high quality if the methodological quality score was >60%.

Data extraction

Two researchers (AML and SMABZ) independently collected the characteristics of the included studies. They collected items concerning the definition of the studied population, how the presence or absence of hip OA was assessed, the assessment of HD, if the study corrected for potentially confounding factors, and which results were reported.

RESULTS

Identification and selection of the literature

A total of 2921 references were identified. After an initial examination of the title and the abstract of the references, the full text of papers were screened to determine whether they met our selection criteria. Of these, 12 articles met the criteria.^6,9,19–28 After screening the reference lists of the selected studies, one other study was included.⁸ For three studies, more than one publication reported different aspects of the studies.^{6,9,19–22,28} All publications were used to extract data about the methods used and the results. Finally, nine studies were included in this review.

Description of the studies included

Table 2 gives the characteristics of the included studies.

Five studies had a cohort design.^{6,20,23,25,28} After screening the reference lists of the selected studies, one other study was included.⁸ For three studies, more than one publication reported different aspects of the studies^{6,9,19–22,28} and four studies had a case-control design.^8,24,26,27

Three of the five cohort studies collected the data in a prospective manner,^6,20,25 and only one study followed up patients without hip OA forward in time.⁶

One study included only men,²³ and one other only women.⁶ Six studies included patients aged 60–65 or older,^{6,23–25,27,28} and the other three included a population with younger patients as well.^8,20,26 Data on congenital HD were lacking. Three studies reported data on Asian,²⁰ or on Asian and white populations.^23,28 All others reported on white people.

Only one study was community based,⁶ all others were hospital based. Of these hospital based studies, two selected the patients not on the presence or absence of hip OA, but according to whether they had had an intravenous urography (IVU).^23,28

In all studies, HD was assessed by measuring the centre-edge (CE) angle—that is, the angle between (a) a vertical line drawn from the centre of the femoral head at right angles to the line joining the two femoral head centres, and (b) a line from the centre of the femoral head to the lateral edge of the acetabular roof. Seven of the nine studies also measured the acetabular depth (AD; the perpendicular distance from the deepest point of the acetabular roof to the line joining the lateral margin of the acetabular roof and the upper corner of the symphysis pubis on the same side), index, or ratio.^{6,8,20,23,25,27,28} The cut off point for the definition of HD according to the CE angle was ⩽25° in three studies^6,8,23,27 and ⩽30° in one study.⁶ One study used a CE angle ⩽20° as an exclusion criterion for the studied patients.²⁶

Two of the five cohort studies defined the presence of hip OA by a radiographic scoring system, described by Kellgren and Lawrence.^6,25 In the three other cohort studies the assessment of hip OA was based on the measurements of the joint space width. The four case-control studies based their case series on hospital patients with hip OA, so their definition of hip OA was a combination of clinical signs or complaints and radiographic abnormalities.^8,24,26,27

Results of the studies included

The included studies presented their results in various ways: one reported a correlation coefficient,²⁸ and two reported an odds ratio.^6,23 The six other studies described the outcomes in global terms only.^8,20,24–27

To provide a global overview of the reported outcomes we assessed whether the relationship between HD and the occurrence of hip OA was “positive” (the presence of HD is a risk factor for hip OA), “negative” (HD protects for hip OA), or “no” (there is no (significant) relation). Table 2 gives a more detailed description of the outcomes.

Six studies reported a positive association,^{6,8,20,24,25,27} five of them reported the outcomes in global terms, and the only prospective follow up study reported that patients with dysplasia had a 2.8 higher risk of developing hip OA.⁶ Two studies reported a negative association^23,28; one reported a correlation coefficient of approximately −0.38 implying that a smaller CE angle is correlated with a broader mean joint space.²⁸ The other study reported that people with dysplasia had a 0.5 to 0.4 smaller risk of having a hip OA.²³ Both studies reported partially on the same population. Only one study reported no relationship.²⁶

Methodological quality assessment

The two reviewers scored 514 items and agreed on 467 items (95%, κ = 0.80). The 47 disagreements were resolved at a single consensus meeting. Table 3 shows the studies in order of their methodological quality score, subdivided into the different types of study design (that is cohort and case-control studies).

The scores ranged from 75% to 23%, and the average rating was 45%.

Best evidence synthesis

Only one of the three prospective cohort studies had a design that really could answer the research question, and this study reached the level of high quality. The other two prospective cohort studies did not reach the level of high quality.

The high quality prospective cohort study reported a positive association between HD and the occurrence of hip OA with prospective data, implying that a patient with HD, has a higher chance of getting hip OA. This means that there is limited evidence for a positive association between HD and the development of hip OA.

DISCUSSION

In this systematic review we summarised the available reported evidence on the influence of HD on the development of hip OA. We may conclude, based on the evidence, that there is limited evidence for a positive association between HD and hip OA. However, most studies included older people (for example, 50–60 years or older). In younger age groups the relation between HD and OA or hip complaints seems to be much higher.²⁹ In these kind of populations, however, it is difficult to study the relationship, because the power of a prospective cohort study to estimate an effect in this rare disease would require implausibly large samples.

Five of nine studies had cross sectional data. One of the most important biases in these studies is that the measurements of HD can be modified by the OA process. For example, osteophyte formation might make it difficult to correctly locate the lateral acetabular margin, resulting in less reliable measurements of the CE angle and AD. Moreover, medial migration of the femoral head may increase the CE angle.

In fact, the data we found support the theory that changes in the hip joint geometry as a result of OA may be responsible for the (weak) negative association. All three prospective studies, including the only study with a population without hip OA at the beginning of the study, reported a positive relationship between HD and the occurrence of hip OA, and the single retrospective study reported no association.

Two of the five cohort studies included people who had had an IVU. Although these populations are hospital based, the requirement for an IVU implies no marked differences in acetabular dimensions, and therefore biases due to selection for treatment of hip OA, are avoided. However, the leg position during x ray examination is not standardised for an IVU, and patients with sustained hip OA tend to lie with the leg in external rotation. This rotation of the hip may influence the measurements of the CE angle. Although these differences are estimated to be small,³⁰ they may produce less reliable results. Both studies performed with data from IVU reported a negative association, whereas five of the seven other studies reported a positive association, and none reported a negative association. Also the difference in the x ray techniques used (for example, for trauma, IVU, and before hip replacement) might make it more difficult to compare the data.

Besides restrictions of the included studies, this review may also have several restrictions.

Identification and selection of the literature

Although we put much effort into identifying all relevant articles, our literature search may have some limitations, not only because some relevant articles might have been missed because they used other keywords or had unclear abstracts but also because not all published articles are indexed in databases. Additionally, we excluded articles written in languages other than English, Dutch, German, French, Danish, Norwegian, or Swedish.

To investigate the amount of potential publication bias for our study, we planned to make a funnel plot,¹⁰ but only three of the nine studies provided enough data to allow us to make such a plot. Unfortunately, this means that we are unable to visualise the possibility of publication bias in this review.

Quality assessment and best evidence synthesis

The quality assessment was challenging because no previously tested and validated criteria lists for observational studies in the field of OA have been published. In addition, limited data were found on performing a best evidence synthesis with observational studies (in contrast with randomised controlled trials).³¹ We therefore presented our methods in a reproducible manner, in order to give the reader insight into the way in which our level of evidence was reached.

In the best evidence synthesis we defined the criteria so that only prospective cohort studies are able to address the research question. In fact, only those with a disease-free population at the beginning of the study are able to do that, but because populations with early hip OA that were studied prospectively can support the evidence, we included them as well. We are however cautious about this kind of information because of the bias that can be introduced. (Acetabular deformity can pre-date hip OA.)

Explanations for the results

Despite the above mentioned biases and limitations, the evidence for an association between HD and the development of hip OA is limited. A hypothetical mechanism to explain the association is the increased joint stress due to decreased joint surface area (smaller CE angle) or joint incongruity (smaller AD).^32,33

In view of the limited evidence on the positive relationship between mild forms of HD and the occurrence of hip OA in the elderly, and the lack of information in younger patients, more well designed prospective population based studies are needed in order to provide more precise and valid information on this relation.