Types of studies

Randomized controlled stuides or quasi‐randomized controlled studies (both published and unpublished). Studies in which quasi‐randomized methods such as alternation were used wwould have been included if there was sufficient evidence that the treatment and control groups were similar at baseline.

Types of participants

Children (from 0 to 18 years old) with XLH defined by clinical, laboratory, or molecular criteria.

Types of interventions

Recombinant human growth hormone (given by subcutaneous injection, at any dose, any frequency and any duration) only or combined with conventional treatment (calcitriol and oral phosphate) compared with either placebo or conventional treatment alone.

Types of outcome measures

We planned to assess the following outcomes.

Electronic searches

We searched the Cochrane Cystic Fibrosis and Genetic Disorders Group's Inborn Errors of Metabolism trials Register using the term: hypophosphatemia.

The Coagulopathies Trials Register is compiled from electronic searches of the Cochrane Central Register of Controlled Trials (CENTRAL) (updated each new issue of the Cochrane Library) and weekly searches of MEDLINE and the prospective handsearching of one journal ‐ Haemophilia. Unpublished work is identified by searching the abstract books of major conferences: the European Haematology Association conference; the American Society of Hematology conference; the British Society for Haematology Annual Scientific Meeting; the Congress of the World Federation of Hemophilia; the European Association for Haemophilia and Allied Disorders, the American Society of Gene and Cell Therapy and the International Society on Thrombosis and Haemostasis. For full details of all searching activities for the register, please see the relevant section of the Cochrane Cystic Fibrosis and Genetic Disorders Group's website.

Date of most recent search: 12 January 2021

We searched for trials on the Cochrane Central Register of Controlled Trials (searched 2011 to 18th January 2021) and Ovid MEDLINE 2011 to 18th January 2021 (searched 18th January 2021). For the full search strategies, please refer to the appendices (Appendix 1; Appendix 2).

We also searched for trial registration entries on clinicaltrials.gov (https://www.clinicaltrials.gov/) and WHO ICTRP (https://www.who.int/clinical-trials-registry-platform). The search strategies used are given in an appendix (Appendix 3). Date of most recent search: 18th January 2021.

Searching other resources

The previous author team did not identify any further studies from the searches they carried out of the Journal of Bone and Mineral Research (1986 to 2003) and the proceedings of the American Society for Bone and Mineral Research Annual Meeting (1st to 24th). The original author team also contacted companies that market growth hormone for information on unpublished studies, however, they did not identify any studies. The current author team did not contact them again for this update.

Selection of studies

Two authors (SS and TR for this update; previously YH and WC ) independently selected the studies to be included in the review. There was no disagreement on the suitability of the studies for inclusion.

Data extraction and management

Each author used standard data acquisition forms to independently extract data. We used Covidence to independently extract the data and then used the compare function to reach an agreed data extraction form (Covidence 2019). This form was electronically uploaded into Revman 5.

We planned to group outcome data into those measured at baseline, 6, 12, 18, 24 months and annually thereafter. If outcome data were recorded at other time periods, we would have considered examining those as well.

Assessment of risk of bias in included studies

We assessed all randomized controlled studies using the Cochrane 'Risk of bias’ tool described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017). Two review authors (SS, TR) independently assessed each element of potential bias listed below as high, low, or unclear risk of bias:

• selection bias (random sequence generation and allocation concealment);

• performance bias (blinding of participants and personnel);

• detection bias (blinding of outcome assessment);

• attrition bias (incomplete outcome data);

• reporting bias (selective reporting);

• other bias.

We briefly described the rationale for the judgements of potential bias in the 'Characteristics of included studies' table. We ensured that a consensus on the degree of risk of bias was met through a comparison of the review authors' statements and where necessary, through consultation with a third person from the Cochrane Cystic Fibrosis and Genetic Disorders Group's editoral base.

Measures of treatment effect

For continuous outcomes, where possible we used the  mean change from baseline or post treatment values with standard deviations (SD) to calculate a summary statistic for the mean difference between groups with corresponding 95% confidence intervals (CIs); for the cross‐over trial, we describe the results in the text of the review. If future updates provide more studies presenting suitable data for analysis we will carry out a pooled estimate of effect by calculating the mean difference and 95 % confidence intervals (CIs).

If binary outcome data, e.g. adverse event data, are included in future updates of the review we will analyse these using risk ratio (RR) and 95 % CIs.

Where data were reported narratively in the original papers or the analysis wasn't appropriate to include in our own analyses, we have reported treatment effect narratively.

Unit of analysis issues

One of our included studies used a cross‐over design. Ideally when conducting a meta‐analysis combining results from cross‐over studies we would have liked to use the methods that are recommended by Elbourne (Elbourne 2002). However, due to restrictions on the available data, the only method that we would have been able to use would be to treat the cross‐over studies as if they were parallel trials. Elbourne says that this approach will produce conservative results as it does not take into account within‐patient correlation (Elbourne 2002). Also each participant will appear in both the treatment and control group, so the two groups will not be independent. Since we believe the data will have been analysed more appropriately by the primary authors, we have reported the results that were obtained from the primary paper and not entered the data from the cross‐over study into Data and analyses. For future updates of this review, when suitable data are available for inclusion, we will aim to use the most appropriate method available to analyse these data.

Dealing with missing data

We have reported the numbers and reasons for dropout or withdrawal from the studies in both the intervention and control groups. In order to allow an intention‐to‐treat analysis, we sought data on the number of participants by allocated treatment group, irrespective of compliance and whether or not the participant was later thought to be ineligible or otherwise excluded from treatment or follow‐up.

Assessment of heterogeneity

We have not been able to combine any outcome data in a meta‐analysis, but if this becomes possible in a future update we plan to quantify the impact of statistical and clinical heterogeneity in the meta‐analysis using a measure of the degree of inconsistency in the studies' results (Higgins 2017). This measure (I²) describes the percentage of total variation across studies that isdue to heterogeneity rather than chance (Higgins 2003). The values of I² lie between 0% and 100%, and a simplified categorization of heterogeneity that we plan to use is of low (I² value of 25%), moderate (I² value of 50%), and high (I² value of 75%) (Higgins 2017).

Assessment of reporting biases

If we had been able to include sufficient studies in the review (at least 10), we planned to look for asymmetry in a funnel plot for the primary outcome measures as an indication for the presence of publication bias. We did however describe any selective reporting by comparing the study protocols or methods sections with the results presented in the papers (or both).

Data synthesis

We were unable to combine study data into a meta‐analysis as outcomes were reported differently and at different timepoints in the two included studies. If we are able to combine data in a future update we plan to carry out a meta‐analysis taking into account the amount of heterogeneity present. Where heterogeneity is low we will use a fixed‐effects model (I² less than 40 %), but if heterogeneity is greater than 40 % we will use a random‐effects model (Deeks 2021).

Subgroup analysis and investigation of heterogeneity

We were unable to carry out a subgroup analysis as we only had a small amount of data to add to the analyses. If in a future update we are able to carry out a meta‐analysis we will investigate possible causes of any heterogeneity by carrying out a subgroup analysis.

Sensitivity analysis

When possible, in future versions of this review, we plan to perform a sensitivity analysis based on the methodological quality of the studies by including and excluding studies that were at high risk of bias.

Summary of findings and assessment of the certainty of the evidence

In accordance with current Cochrane guidance we have included a summary of findings table for the comparison of rhGH versus control (summary of findings Table 1). We have selected the following outcomes to present in the summary of findings table based on their clinical importance.

1. Longitudinal growth: height SDS

2. Adverse effects

3. Measures of mineral metabolism: serum phosphate

4. Measures of endocrine function: alkaline phosphatase

5. Measures of renal function TmP/GFR

6. Body proportions: sitting height SDS

We used GRADE methodology to assess the quality of the evidence for each outcome based on the risk of bias within the studies, relevance to our population of interest (indirectness), unexplained heterogeneity or inconsistency (inconsistency), imprecision of the results or high risk of publication bias. The certainty of the evidence was deemed to be high before assessment. After assessment, the certainty of evidence was downgraded once for serious risk and twice for very serious risk.